diff --git "a/data/part_4/01fd239204aa4c7903b2766b8eeba20b.json" "b/data/part_4/01fd239204aa4c7903b2766b8eeba20b.json"
new file mode 100644--- /dev/null
+++ "b/data/part_4/01fd239204aa4c7903b2766b8eeba20b.json"
@@ -0,0 +1 @@
+{"metadata":{"id":"01fd239204aa4c7903b2766b8eeba20b","source":"gardian_index","url":"https://digitalarchive.worldfishcenter.org/bitstream/handle/20.500.12348/3358/Pub%20SR76%2015.pdf"},"pageCount":386,"title":"On the Implementation of Management-Oriented Fishery Research: the Case of the Peruvian ~nchoveta*","keywords":[],"chapters":[{"head":"IMARPE Foreword","index":1,"paragraphs":[]},{"head":"R. VILLANUEVA Executive Director Institute del Mar del Peru","index":2,"paragraphs":[{"index":1,"size":23,"text":"The marine fisheries are vital to Peru and hence responsible management of our marine fish resources should be one of our national priorities."},{"index":2,"size":47,"text":"Much has been achieved as far as understanding the fishery biology of the various fish species inhabiting the Peru Current System is concerned; the life-histories of the major species have been largely elucidated, the fishery monitored and the gross effect of successive El Niiio events well documented."},{"index":3,"size":35,"text":"Our major impediment in this research effort and hence in formulating optimal management options has been, however, the piecemeal nature of much of our research largely due to lack of continued funding for sustained efforts."},{"index":4,"size":93,"text":"This has resulted in many of our valuable results remaining disconnected, delaying the emergence of a broad view of the upwelling ecosystem along our shores, both in the heads of some of our scientists and of some of our fishery managers. How else can we explain, over 30 years after the onset of the anchoveta fishery, the lack of a management plan which would simultaneously consider the exploitation of the anchoveta, of its major predators and competitors (bonitos, mackerels, horse mackerels, sardine) as well as the conservation of the guano birds and seals?"},{"index":5,"size":61,"text":"I was therefore very pleased when I first heard about the project initiated by Ms. Isabel Tsukayama, then Chief of IMARPE's Pelagic Resources Division, by Dr. Pauly of ICLARM and by PROCOPA staff and have since given it as much support as I could, both by encouraging IMARPE staff to participate in it and by encouraging other institutions to release data."},{"index":6,"size":48,"text":"I now feel satisfied that this project has produced the elements needed to reach a global view of the Peruvian ecosystem and I have no doubt that future generations of Peruvian researchers will find this book useful as inspiration and/or as source of background data for their research."},{"index":7,"size":103,"text":"With this volume, which includes in table or graphic form an extremely large amount of time series data, our Institute also provides the international scientific community with material that can be used to test various hypotheses on the dynamics of upwelling ecosystems and of their pelagic resources. We have convinced ourselves, during this exercise, that we held in our files, drawers and unpublished reports more and better data than we thought was the case. We hope others with social and economic constraints such as ours will also inspect their files and drawers and put together the data, as was done in this book."},{"index":8,"size":104,"text":"Finally, I wish to thank GTZ and ICLARM. Without their support and active involvement, this book could not have been written. This applies especially to Dr. Bilio, who approved the whole idea, and to Dr. Arntz and Dr. H. Salzwedel, the successive team leaders of PROCOPA, who so often acted as bridge between us and faraway ICLARM. This applies also to the latter organization and its Director General, Dr. I.R. Smith, for hosting a young Peruvian scientist involved in the project and especially for their support of Dr. D. Pauly, whose communicative enthusiasm certainly was a major element to the success of this project."},{"index":9,"size":82,"text":"Although some of the contributions included in this book are highly theoretical, I am confident that its publication represents a great step forward to the understanding of the dynamics of the principal Peruvian marine fishery resource and its management, if only because of the massive amount of data that has been assembled and standardized and of the sheer length of the time series that are now available for detailed study. GTZ is pleased to have been able to contribute to this effort."},{"index":10,"size":3,"text":"Eschborn, June 1987"}]},{"head":"ICLARM Foreword","index":3,"paragraphs":[]},{"head":"DR. IAN R. SMITH Director General, ICLARM","index":4,"paragraphs":[{"index":1,"size":27,"text":"It is with great pleasure that I have accepted to introduce, on behalf of the International Center for Living Aquatic Resources Management, this book to its readers."},{"index":2,"size":32,"text":"I believe this volume illustrates very well the many facets of ICLARM's concept of management-oriented fishery research conducted by cooperating institutions and more specifically of ICLARM's multidisciplinary emphasis when conducting such investigations."},{"index":3,"size":104,"text":"Three different institutions, based on three different continents, joined in this effort to which scientists from six different countries unselfishly contributed all of their data and much of their time. The disciplines these scientists represent range from physical oceanography and meteorology to fishery biology and economicswith one author actually having worked for almost a decade in the anchoveta fishery. However, it is not only this wide range of skills and talents which has made this book possible. Rather, the crucial factor seems to have been the support which the authors and the editors received from various institutions in Peru, notably from PESCAPERU and IMARPE."},{"index":4,"size":48,"text":"The scientific value of historic data held in the laboratories of developing countries is often unappreciated. This book illustrates the vision and foresight of those who collected these data as well as of those who released them for use by the authors of the various contributions included here."},{"index":5,"size":91,"text":"Three of the key papers included in this book are by staff of the National Marine Fisheries Service of the US Department of Commerce. All of the editors' effort at reaching a comprehensive coverage of the Peruvian ecosystem would have been vain without the massive data sets and in-depth analyses contributed by these authors. On behdf of ICLARM, an institution devoted to research on aquatic resources with institutions in developing countries, I would like to congratulate NMFS for this cost-efficient and well-focused method for transferring data back to where they originated."},{"index":6,"size":77,"text":"This book is the first ICLARM publication dealing exclusively with Latin American resources and I deeply regret that time constraints prevented the editors from preparing Spanish Abstracts and Table and Figure legends, as originally planned. We hope, however, that the opportunity will emerge in due time for a translation of the entire volume into Spanish, which incidentally would allow for an update of the time series in the various contributions, most of which stop in December 1982."},{"index":7,"size":30,"text":"ICLARM is pleased to have had the privilege of cooperating with IMARPE and GTZ, and hope that the volume resulting from this cooperation will be found useful by its readers."}]},{"head":"Manila, June 1987","index":5,"paragraphs":[]},{"head":"Introduction","index":6,"paragraphs":[{"index":1,"size":157,"text":"A huge amount of literature exists on the biology and population dynamics of the Peruvian anchoveta (Engraulis ringens) (see Table 1). However, this literature differs in an important way from the associated literature on the oceanography of the Peruvian ecosystem: the overwhelming majority of \"biological papers\" have a very short time scale, usually covering the period of a few months within, or the few years spanning major El Ni6o events (see e.g. contributions in Arntz et al. 1985). he reasons for this imbalance are numerous and include objective constraints (assembling and analyzing biological data, say on a monthly scale, over long time series is far more difficult than assembling say temperature measurements (see Tabata 1985) as well as subjective problems, such as those that emerge when biologists working in different laboratories, using different methods, have to share data and ideas. Table 1. Some major source of information o n the Peruvian upwelling ecosystem (as of early 1987)."}]},{"head":"Reference","index":7,"paragraphs":[{"index":1,"size":10,"text":"Type of document/remarks Garcilaso de la Vega (1609 and 1617)"},{"index":2,"size":23,"text":"Hutchinson (1950) Boerema et al. (1965) I Seminario Latino American0 sobre el Oceano Pacifico Oriental (1966), Univ. San Marcos, Lima, Peru, 218 p."},{"index":3,"size":92,"text":"One consequence of this imbalance is that long time series data, to which rigorous multivariate methods could be applied are lacking, and that various hypotheses, advanced decades ago to explain the dynamics of the anchoveta stock of Peru could not be tested and/or refined. Another consequence is that each author, while concentrating on the specific problems(s) she or he is investigating has to \"set constant\" those variables that are not specifically dealt with, leading to rather old estimates of important anchoveta population statistics being \"carried over\" from one study to the next."},{"index":4,"size":50,"text":"Underlying these problems, finally, is an enormous wasteor at least underutilizationof information which, while not necessarily being easy to access, does nevertheless exist and which, if properly standardized, could help interpret the behavior of the anchoveta stocks off Peru, and more generally, of stocks of small pelagics throughout the world."}]},{"head":"Genesis of the Project Leading to the Present Book","index":8,"paragraphs":[{"index":1,"size":72,"text":"We shall present, in the following paragraphs (based on Pauly et al. 1986) an outline of the genesis of the multidisciplinary project which led to the book presented here, meant to alleviate the situation described above. We shall elaborate on this item more than might appear necessary because we believe that this international effort illustrates a cooperative modus operandi that is used far too little in projects with aims similar to ours."},{"index":2,"size":102,"text":"The project emerged out of three, at first unrelated, developments. The first was that the staff of a GTZ project hosted by IMARPE, the Programa de Cooperacion Peruano-Aleman (PROCOPA) was given a mandate to develop, for purposes of fishery management, and in cooperation with their counterparts at IMARPE, a model of the Peruvian ecosystem that would be more realistic and versatile than those of Walsh (1981) or of Kremer and Sutinen (1975). Thus, an attempt was made to involve Dr. E. Ursin, who had earlier worked on a model of the North Sea (Andersen and Ursin 1977) in these efforts (Ursin 1980)."},{"index":3,"size":49,"text":"The second development was the 1980 release of the early version of the ELEFAN programs for the estimation of growth, mortality and related statistics from length-frequency data (Pauly andDavid 1980,1981;Pauly 1982;Pope et al. 1981), such as have been collected on the Peruvian anchoveta since the 1950s (Clark 1954;Jordan 1959)."},{"index":4,"size":49,"text":"The third development, finally, was the identification and refinement, by A. Bakun, R. Parrish and associates at the Pacific Fisheries Environmental Group of NOAA/NMFS (Monterey, California) of a methodology for the analysis of wind data in a context relevant to fisheries in upwelling systems (Bakun 1985;Parrish et al. 1983)."},{"index":5,"size":394,"text":"Two of these developments converged in 1981, during the fxst author's initial visit to, IMARPE. A \"test run\" was undertaken as a consequence of this visit, leading to an analysis of 20 years' worth of monthly anchoveta catch-at-length data using the ELEFAN I and I11 programs. This provided extremely encouraging results (Pauly and Tsukayama 1983), the three main conclusions from this preliminary analysis being: i> ii) iii) length-based methodsparticularly that described as \"VPA 111\" in Pauly and Tsukayama (1983) and Pauly, Palomares and Gayanilo (this vol.) appear eminently suited to study the dynamics of anchoveta, and reasonable estimates of monthly recruitment, biomass, fishing mortality and related information can be readily derived from an extremely limited amount of data in addition to catch-atlength data; the assumption of a constant natural mortality (\"Mu) underlying the preliminary analysis, as well as most fish population models is untenable in the case of the Peruvian anchoveta, and ways must be found to let M vary, e.g., with the biomass of major predators, or by adding anchovetas consumed by predators to those caught by the fishery (see Pauly, Palomares and Gayanilo, this vol.); the 20-year time series of catch-at-length data used in the preliminary analysis from 1961 to 1979 should be extended backwards to cover the initial phase of the fishery (when biomasses were high and recruitment variability low) and forward Fig. 14 shows anchoveta growth curves obtained for \"1950\" and \"1980\" conditions. Table 5 summarizes the results for two runs, while Table 6 shows the main results of the sensitivity analyses of the model. As might be seen, assimilation coefficient and caloric food content, both related to the food quality factor, have the strongest effects on model output. Table 6 also shows that these parameters, considered alone must be changed by 45% for anchoveta to change from its \"1950 pattern to the \" 1980\" pattern. On the other hand, if all parameters in Table 6 are changed simultaneourly, a change of only 15% is needed for the \" 1950\" to \" 1980'' transition (see also Fig. 14). Growth of Peruvian anchoveta as simulated using model described in text. Note that most of the difference between the \"1950\" and \"1980\" growth curves is due to growth rate differences of fish between 1 and 1 .5 years. a As compared with a length of 16 cm for the initial (\"1950\") conditions."}]},{"head":"Discussion","index":9,"paragraphs":[{"index":1,"size":89,"text":"Overall, our growth estimates match those obtained previously by other authors (Table 7, Fig. 15). However, the key finding presented here is obviously the strong evidence for densitydependent growth in the Peruvian anchoveta, a phenomenon for which only anecdotal evidence existed previously. Our ability to extract evidence for changes in the growth performance of anchoveta depended on two main factors not found elsewhere: i) the availability of an extremely long time series of length-frequency data, and ii) the availability of powerful software for the analysis of these time series."}]},{"head":"Introduction","index":10,"paragraphs":[{"index":1,"size":83,"text":"Numerous previous estimates of the biomass of Peruvian anchoveta exist and may be found in the form of time series in various pamphlets, reports and/or books on topics ranging from the economics of soya bean exports to oceanographic forecasting. Yet these time series not being structured by size (or age group) nor having a time scale sufficiently small to reflect the rapid changes of biomass experienced by the anchoveta, are largely useless for research, e.g., as element of models for predicting fisheries yields."},{"index":2,"size":72,"text":"these parameters caused a slight overlap of some \"cohorts\" (i.e., some of the catch data were used twice), and small gaps (i.e., some of the catch data were not included in any cohort). This source of error could have been avoided by using the same growth parameters throughout. This, however, would have caused a large bias, given the strong positive trend in anchoveta growth performance documented in Palomares et al. (this vol.)."}]},{"head":"Interpolation of Missing Sets of Monthly Size-Composition Data","index":11,"paragraphs":[{"index":1,"size":126,"text":"Although the monthly catch data (in weight) used for the present analyses are complete, the monthly size composition data needed to derive monthly catch-at-length data are not (see Tables 1 to 30 in Tsukayama and Pauly, this vol.). Since uninterrupted series of catch-at-length data are needed for VPA III, we have linearly interpolated size-frequency distributions where gaps appeared in the original data (except for January-October 1953, where we have used the corresponding values for 1954). We believe this approach had no major detrimental impact on our results for three reasons: i) linearly interpolated % length-frequency samples are very similar to \"real\" samples (see Fig. 2) if only because the overall shape of such samples is determined mainly by continuous, rather \"smooth\" processes (i.e., growth and mortality);"}]},{"head":"Month (1966)","index":12,"paragraphs":[{"index":1,"size":25,"text":"Fig. 2. Length-frequency data of Peruvian anchoveta (northern/central stock), including samples that were linearly interpolated. The reader is invited to guess which samples were interpolated."},{"index":2,"size":11,"text":"(.raqura~oN pun jsn8ny ' A ~I 'aunf are syluour pa$qod~avq aqL)"},{"index":3,"size":76,"text":"ii) VPA estimates of F and population size in a given time interval are not predominantly determined by the catch composition in that very interval, but by the catch in that interval and the catches and catch composition in previous intervals (VPA runs backwards!). Hence whatever zrror is introduced by interpolating w i l l be spread over several intervals, and its absolute impact )n mortality and population estimates pertaining to a given interval thus reduced."},{"index":4,"size":57,"text":"iii) catch-at-length data as used in VPA 111 are estimated from both length composition data wf catch data in weight. While the former were interpolated in some cases, the latter were ways \"real\" and hence the overall number of fish caught in a given month tended to be sonably approximated, even though their size composition was interpolated."}]},{"head":":mation of Monthly Recruitment and Biomass","index":13,"paragraphs":[{"index":1,"size":88,"text":"Monthly population estimates per length class were obtained by regrouping monthly values i obtained on a cohort basis into regular class intervals (the same intervals as those in which originally grouped). The population estimates in the smallest class 4.75 cm) are here defined as \"recruitment\" (of fish with mean length The population estimates (in numbers), by length class, were then multiplied, for each month and length class separately by the appropriate value mean weight values (see equation 4) to obtain monthly population biomass by length class (Fig. 3)."},{"index":2,"size":52,"text":"All analyses were run with Ft = 2(y-1). It is recalled that these values (a) pertain to Z-Mo (i.e., to the sum of true fishing mortality plus all sources of natural mortality represented by distinct predators and (b) that values of Ft have little impact on final results, especially on recruitment estimates."}]},{"head":"Estimation of Natural and Fishing Mortalities","index":14,"paragraphs":[{"index":1,"size":89,"text":"The natural mortality (M ) exerted by any given predator (p) with anchoveta consumption R (Cp(i)) was estimated, for eac month (i) from Mp(i) = mchoveta consumption (Cp(i))/anchoveta biomass (i) ... 12) Similarly, fishing mortality (F) was estimated from Fi = anchoveta fishery catch (i)/anchoveta biomass (i) ... 13) Thus, all estimates of mortality presented here pertain to weights, not numbers. [Note, however, that computations of number-based mortality values are possible, based on the data presented in this volume.] From these data, total mortality (Z) can be estimated from"},{"index":2,"size":15,"text":"where m is the number of predator groups considered here (i.e., birds, bonito and seals)."}]},{"head":"Results and Discussion","index":15,"paragraphs":[]},{"head":"Estimate of Mo and Their Implications","index":16,"paragraphs":[{"index":1,"size":161,"text":"Table 3 gives the values of Mo obtained iteratively, i.e., by changing values of & until biomass were obtained which closely matched the independent biomass estimates in that same table. The same values of Mo are also plotted as time series in Fig. 4. As might be seen, the estimates of biomass in Table 3 force us to assume that &, which took values of about 2 ~-1 in the 1960s, dropped to about O S y -l in 1975, rapidly increased to about 4y-1 in 1976 and stayed at this high level until the late 1970s. (We shall present below evidence suggesting that the natural mortality of anchoveta did decrease in the early 1970s, as illustrated on Fig. 4.) The value of Mo = 4y-1 for the late 1970s suggest that an-important predator, ignored in our analyses, managed to continue consuming a large amount of anchoveta in spite of their very reduced biomasses in the late 1970s (Pauly, this vol.)."}]},{"head":"Estimates of Anchoveta Biomass and Recruitment, 1953 to 1981","index":17,"paragraphs":[{"index":1,"size":113,"text":"Tables 4 to 33 present the key results of our analyses, i.e., the monthly recruitment estimates and the monthly biomasses, by length group, respectively, obtained through the VPA 111 routine of the ELEFAN III program.  Fig. 1 and text IMARPE (1974a) these parameters caused a slight overlap of some \"cohorts\" (i.e., some of the catch data were used twice), and small gaps (i.e., some of the catch data were not included in any cohort). This source of error could have been avoided by using the same growth parameters throughout. This, however, would have caused a large bias, given the strong positive trend in anchoveta growth performance documented in Palomares et al. (this vol.)."}]},{"head":"Interpolation of Missing Sets of Monthly Size-Composition Data","index":18,"paragraphs":[{"index":1,"size":126,"text":"Although the monthly catch data (in weight) used for the present analyses are complete, the monthly size composition data needed to derive monthly catch-at-length data are not (see Tables 1 to 30 in Tsukayama and Pauly, this vol.). Since uninterrupted series of catch-at-length data are needed for VPA 111, we have linearly interpolated size-frequency distributions where gaps appeared in the original data (except for January-October 1953, where we have used the corresponding values for 1954). We believe this approach had no major detrimental impact on our results for three reasons: i) linearly interpolated % length-frequency samples are very similar to \"real\" samples (see Fig. 2) if only because the overall shape of such samples is determined mainly by continuous, rather \"smooth\" processes (i.e., growth and mortality);"},{"index":2,"size":25,"text":"Fig. 2. Length-frequency data of Peruvian anchoveta (northernlcentral stock), including samples that were linearly interpolated. The reader is invited to guess which samples were interpolated."},{"index":3,"size":83,"text":"('IaquranoN prre 1sn8nv 'K[ny 'aunr a n sq~uour pa$qodra$q aqL) ii) VPA estimates of F and population size in a given time interval are not predominantly determined by the catch composition in that very interval, but by the catch in that interval and the catches and catch composition in previous intervals (VPA runs backwards!). Hence whatever error is introduced by interpolating will be spread over several intervals, and its absolute impact on mortality and population estimates pertaining to a given interval thus reduced."},{"index":4,"size":57,"text":"iii) catch-at-length data as used in VPA III are estimated from both length composition data and catch data in weight. While the former were interpolated in some cases, the latter were always \"real\" and hence the overall number of fish caught in a given month tended to be reasonably approximated, even though their size composition was interpolated."}]},{"head":"Estimption of Monthly Recruitment and Biomass","index":19,"paragraphs":[{"index":1,"size":61,"text":"Monthly population estimates per length class were obtained by regrouping monthly values q f Ni obtained on a cohort basis into regular class intervals (the same intervals as those in which thi3afch d E originally grouped). The population estimates in the smallest class considered here (3.75 to 4.75 cm) are here defined as \"recruitment\" (of fish with mean length 4.25 cm)."},{"index":2,"size":38,"text":"The population estimates (in numbers), by length class, were then multiplied, for each month and length class separately by the appropriate value mean weight values (see equation 4) to obtain monthly population biomass by length class (Fig. 3)."},{"index":3,"size":52,"text":"All analyses were run with Ft = 2(y-1). It is recalled that these values (a) pertain to Z-Mo (i.e:, to the sum of true fishing mortality plus all sources of natural mortality represented by distinct predators and (b) that values of Ft have little impact on final results, especially on recruitment estimates."}]},{"head":"Estimation of Natural and Fishing Mortalities","index":20,"paragraphs":[{"index":1,"size":104,"text":"The natural mortality (M ) exerted by any given predator (p) with anchoveta consumption (Cp(i)) was estimated, for eac ! month (i) from Mp(i) = anchoveta consumption (Cp(i))/anchoveta biomass (i) ... 12) Similarly, fishing mortality (F) was estimated from Fi = anchoveta fishery catch (i)/anchoveta biomass (i) ... 13) Thus, all estimates of mortality presented here pertain to weights, not numbers. [Note, however, that computations of number-based mortality values are possible, based on the data presented in this volume.] From these data, total mortality (Z) can be estimated from where m is the number of predator groups considered here (i.e., birds, bonito and seals)."}]},{"head":"Results and Discussion","index":21,"paragraphs":[]},{"head":"Estimate of Mo and Their Implications","index":22,"paragraphs":[{"index":1,"size":157,"text":"Table 3 gives the values of Mo obtained iteratively, i.e., by changing values of Mo until biomass were obtained which closely matched the independent biomass estimates in that same table. The same values of Mo are also plotted as time series in Fig. 4. As might be seen, the estimates of biomass in Table 3 force us to assume that &, which took values of about 2y-1 in the 1960s, dropped to about 0.5~-1 in 1975, rapidly increased to about 4y-1 in 1976 and stayed at this high level until the late 1970s. (We shall present below evidence suggesting that the natural mortality of anchoveta did decrease in the early 1970s, as illustrated on Fig. 4.) The value of & = 4y-1 for the late 1970s suggest that an-important predator, ignored in our analyses, managed to continue consuming a large amount of anchoveta in spite of their very reduced biomasses in the late 1970s (Pauly, this vol.)."}]},{"head":"Estimates of Anchoveta Biomass and Recruitment, 1953 to 1981","index":23,"paragraphs":[{"index":1,"size":73,"text":"Tables 4 to 33 present the kev results of our analvses. i.e.. the monthlv recruitment estimates and the monthly biGmasses, by length group, respectiiely; ob&ned through the VPA ID[ routine of the ELEFAN III program.  a ~t h e r inputs, used throughout, were C = 0.3, WP = 0.7 and Ft = 2 (see text). h o t a l length, in cm; from Fig. 7 in Palomares et al. (this voL)."},{"index":2,"size":48,"text":"-I ;from Fig. 7 in Palomares et al. (thisvol.). dLonthly means, as read off eye fitted line in Fig. 4; values used for VPA 111 linearly interpolated in steps of 3 months.  (1981) a ~l l values adjusted to pertain only to the region between 4 and 14's."},{"index":3,"size":28,"text":"b ~e e also Johannessun and Robles (1977). that seasonal oscillations and other within-year changes appear to be far more intense and rapid than had previously be assumed."},{"index":4,"size":80,"text":"Little comments are needed for the parts covering 1980 to 1982, except perhaps that the decline of the biomass to a very low level prior to the onset of the 1982-1983 El NGo is, in part, an artifact due to the absence of catches and consumption estimates to \"feed into\" the VPA in 1983 to estimate the 1982 biomasses. For this reason, we have omitted the year 1982 from the detailed results on biomasses given in Tables 5 to 33."},{"index":5,"size":202,"text":"The anchoveta biornasses estimated for the 10 years preceding the first available independent standing stock estimates, i.e., the years 1953 to 1962, are interesting in that they are lower than expected (see Table 3 in Muck and Pauly, this vol.). Interestingly, these biomass estimates suggests that the anchoveta suffered from the 1957 El Niiio almost as much as from that of 1971-1972. This indeed would explain the massive bird mortalities recorded from that period (see Tovar et al., this vol.). It should be noted, however, that this point is, in part at least, a circular argument, since we used, among other things, bird population data to estimate bird Year Fig. 6. Time series of recruitment (of fish ranging from 3.75 to 4.75 cm, slightly less than 3 months old) into the anchoveta stock, January 1953 to mid-1982. Above: monthly recruitment, showing increasing variability, from the late 1950s to 1970, probably due to increasing fishing pressure and leading to recruitment collapse in early 1971, prior to the onset of the 1971-1972 El NGO. Below: smoothed data (using a 12-month running average), showing that the 1960s, which saw the buildup of the fishery, may have been a period of exceptionally and steadily high recruitment."},{"index":6,"size":27,"text":"anchoveta consumption, and since such consumption, at a time when the fishery was just starting, should have had a great impact on the VPA estimates of biomass."},{"index":7,"size":51,"text":"Fig. 6, based on Table 4, shows two time series of anchoveta recruitment: one consists of monthly estimates of the number of young fish (of 3.75-4.75 cm, i.e., about 3 months old) entering the fishery, the other illustrating the same data, but as 12 months' running average to show interyear changes."},{"index":8,"size":190,"text":"As might be seen, these time series closely resemble the time series of biomass, which is not surprising in a fish with a short lifespan. Important aspects of the time series in Fig. 6 are: i) the increased within year variability of recruitment from the late 1950s to the late 1960s, presumably an effect of fishing, and ii) the fact that the major recruitment collapse of 1971 appears to have occurred before the onset of the [1971][1972] El N i k (see also Mendelsohn and Mendo, this vol.) These two aspects, already apparent in the earlier analysis of Pauly and Tsukayama (1983) imply that much of what has been written about the 11971-1972 El Nifio being the cause of the collapse of the fishery in 1972 is probably wrong. Fig. 7 shows time series of anchoveta fishing mortality, both of a monthly basis and smoothed (12 months' running average) to allow the interyear trend to become fully visible. As might be seen, fishing mortality increased steadily through the 1950s and 1960s, then fluctuated violently through the 1970s, mainly due to successive El Niiio events and associated closures of the fishery.  "}]},{"head":"Estimation of Anchoveta Mortality Caused by Various Predators","index":24,"paragraphs":[{"index":1,"size":169,"text":"Figs. 8A, 8B and 8C show that part of natural mortality is attributable to the birds, the bonito and the seals, respectively. As might be seen, birds and bonito had a roughly similar impact on the anchoveta stock with peaks of 0.3-0.4~-1 in the late 1950s and negligible impact thereafter. The impact of the seals on the anchoveta stock, on the other hand, was negligible throughout the whole period considered here. C. Mortality caused by sea lion (mainly) and fur seals. Note that the scale, ranging from zero to 0.005 implies a negligible pinniped impact on anchoveta natural mortality due to all causes. Note that shape of curve resembles that of Fig. 4, except for small \"bumps\" in the 1950s, corresponding to periods of bud and bonito abundances. Important here is that decline of M from 1971 to 1974 is confirmed by analysis of tagginglrecapture data performed by Malaga and Armstrong (MS). Note also that little support is available for low M estimates (B & S) in Schaefer (1967)."},{"index":2,"size":106,"text":"Fig. 8D, finally, shows a time series of natural mortality as a whole, i.e., combining the predators considered here and the estimates of Mo from Fig. 4. As might be seen from the overall shape of the curve, it is the estimates of Mo which, throughout determine the overall level of natural mortality, clearly illustrating that the predators explicitly considered here do not explain but a small fraction of overall natural mortality. Interestingly, the drop of M observed in 1973-1974 is closely matched, albeit at higher level by a drop of M-estimates based on tagging data and reported in Table 11 of Malaga and Armstrong (MS)."},{"index":3,"size":122,"text":"Fig. 8D also shows that the values of M reported in Schaefer (1967) and based on his simulations (\"S\") and on a personal communication of T. Burd (\"B\") appear low compared with our estimates. We note, finally that values of M computed from the empirical equation of Pauly (1980) ranged between 1.3 and 1.6, and thus were intermediate for the 1960s at least, between the estimates reported in Schaefer (1967) and the Mo values estimated by backcalibration of VPA estimates.      10,548,011 10,412,240 11,473,243 13,114,423 13,740,181 9,711,634 14,050,346 13,735,337 13,997,264 15,757,467 16,134,135 16,694,301   25 15.25 16.25 17.25 18.25 19.25 20.25 Sum in time to cover the period when the anchoveta was (partly) replaced by other small pelagic fishes, and its variability became high."},{"index":4,"size":37,"text":"Item (i) resulted in IMARPE, PROCOPA/GTZ and ICLARM formalizing an agreement to cooperate on a major program of data retrieval and standardization, such that subsequent studies using length-frequency methods, time-series analysis and other methods would become possible."},{"index":5,"size":30,"text":"Item (ii) implied a need to explicitly consider the major predators of anchoveta, of which the guano birds wereat the time (see below)the only ones we thought were really important."},{"index":6,"size":44,"text":"Estimating the population size and anchoveta consumed by guano birds along the stretch of the Peruvian coast between 4 and 140s (i.e., such that the \"southern stock\" of anchoveta is excluded, see Fig. 1) involved performing a planimetric analysis of over 10,000 maps showing"}]},{"head":"Engroulis ringens","index":25,"paragraphs":[{"index":1,"size":189,"text":"Northern /Central stock Southern stock the distribution of three species of guano birds on 40 guano islands and \"points\", completed by the guards of the Compania Adrninistradora del Guano and its various successors (see Tovar et al., this vol.). This also involved deriving a model of the predation on anchoveta by the guano birds (see Muck and Pauly, this vol.). Item (iii) above involved retrieving, from a number of scattered sources, information on the catch and catch size composition of anchoveta for the earlier years of the fishery. This effort brought a surprisingly large amount of material to light (see Tsukayama and Palomares, this vol.), matching previous experiences elsewhere (Ingles and Pauly 1984). These data, as well as data covering the late 1970s and early 1980s indeed allowed for the construction of numerous time series, and showing so few gaps that standard interpolation procedures could be used to obtain uninterupted series (see, e.g., Pauly, Palomares and Gayanilo, this vol.). This also applied to the time series of oceanographic and meteorological data compiled and analyzed by Brainard and McLain (this vol.), Bakun (this vol.) and Mendo et al. (this vol.)."}]},{"head":"Identity of the Anchoveta Stock","index":26,"paragraphs":[{"index":1,"size":81,"text":"The Peruvian anchoveta (Engraulis ringens Jenyns) belongs to the family Engraulidae (Pisces Clupeomorpha, Clupeoidei). It occurs exclusively along the eastern coast of South America, from 40303 off Peru (Jordan 1971) to 42030's off Chile (Brandhorst 1963;Mathisen 1979), with heaviest concentration along the coast of northern and central Peru, north (i.e., \"downstream\") of the strongest upwelling area. At the northern end of their range, anchoveta biomass drops off rapidly, while toward the south, this biomass only tapers off gradually (Fig. 1). 1)."},{"index":2,"size":76,"text":"Mapping of anchoveta distribution during \"EUREKA\" (Villanueva 1975) and other acoustic surveys and during egg surveys suggests the presence during the spawning season of isolated \"density centra\" (see maps in Santander, this vol.) which may or may not correspond to genetically distinct subgroups or populations. Mathisen (1979), after a thorough review of the then available literature, suggested these \"centra\" to be genetically distinct populations. However, electrophoretic studies of allelle distribution have not been conducted in anchoveta."},{"index":3,"size":89,"text":"At the gross level however, distinctions can be made straightforwardly between the anchoveta off northern/central Peru, and those from southern PeruIChile, with anchoveta from southern Peru and Chile having less, coarser gill rakers (Tsukayama 1966) and shorter guts (Rojas 1971), both items suggesting that these fish rely on zooplankton more than their northern counterparts. IMARPE (1973) wrote that \"the results of the recent tagging experiments support the hypothesis expressed in previous reports that the anchoveta resources in the southern area (Atico-110 area) are a more or less separate stock\"."},{"index":4,"size":117,"text":"For the purposes of this, and the other contributions included in this book, we have therefore used 140s as the limit between the northern/central stockhere reported uponand the southern PeruvianIChilean stock of anchoveta (Fig. 1). The strong interactions between the Peruvian and Chilean components of the southern stock of anchoveta would make studies based on isolated \"national\" data sets of limited usefulness. In fact a detailed investigation of this stock would require a high level of cooperation and data exchange between Peruvian and Chilean scientists and institutions. We hope that such cooperations will materialize in the future, and that the southern stock of anchoveta will become as well documented as the northedcentral stock covered in this book."}]},{"head":"The Peruvian Anchoveta and the Prediction of Its Recruitment","index":27,"paragraphs":[{"index":1,"size":32,"text":"A research project as comprehensive as the one reported upon in this book cannot be kept on course if a \"central hypothesis\" is kcking around which the various contributions can be structured."},{"index":2,"size":119,"text":"Our central hypothesis is that the recruitment of the Peruvian anchoveta, \"everything else being equal\", is determined by a short-frequency burst of wind-driven turbulence, i.e., our central hypothesis corresponds to Lasker's contention that \"storms\", by dissipating food-rich microlayers in which anchoveta larvae can feed, lead to their starvation and to recruitment failures (Lasker 1978). Given appropriate data,testing this hypothesis is rather straightforward, and indeed, it has been repeatedly and successfully tested off California both for average conditions (Husby and Nelson 1982) and based on time series data (Peterman and Bradford 1987). This hypothesis has also been tested, in the Peru Current area, for average conditions with somewhat equivocal results (see Bakun 1985 for a review of the relevant literature)."},{"index":3,"size":92,"text":"What has been lacking to date was a test of this hypothesis using time series data from Peru. For such a test, however, everything else must be at least approximately equal (see above), and the bulk of this book represents an attempt to collect data on those things that have varied, such that they can be explicitly accounted for, and the true effect of wind-induced turbulence isolated from the noise. Thus in a sense, this book is a test of Lasker's hypothesis, probably the most comprehensive test this hypothesis will ever get."},{"index":4,"size":26,"text":"Previous work dealing explicitly with the recruitment of anchoveta include the classic paper of Csirke (1980) who quite conclusively demonstrated that plotting a bivariate stock vs."},{"index":5,"size":164,"text":"recruitment relationship simply will not do for the Peruvian anchoveta (see Fig. 2), as is indeed also true for any other fish, notwithstanding suggestions to the contrary (e.g., by Shepherd 1982). Also, an attempt exists to deal with anchoveta recruitment in terms of bioenergetics (Ware and Tsukayama 198 1). An approach to deal with recruitment (R), suggested by Bakun et al. (1982), and following up on the work of Csirke (1980) is to use a model of the form log (R/S) = a+blS+b2 E l i b 3 E2+ .. . where S is the spawning stock and the Ei are environmental variables likely to affect the survival of prerecruits. Bakun et al. (1982) suggests that because data points for such approach are limited (they implied 1 point per year), \"the number of explanatory variables must be limited to a minimum\", and that \"this should be done on rational ground, based on the best available understanding of cause-effect relationships between recruitment and environmental factor\"."},{"index":6,"size":61,"text":"The problem of data point limitation alluded to by Bakun et al. (1982) has been resolved here at least in part by putting all time series included in this book on a monthly basis (this resolves the problem only in part because other problems, such as seasonal autocorrelation then crop up; see Mendelsohn and Mendo, this vol. and Pauly, this vol.)."}]},{"head":"On the Time Series and Graphs in This Volume","index":28,"paragraphs":[{"index":1,"size":55,"text":"The criteria applied to decide whether to include a given data set into the present volume were: i) do the raw data cover reasonably well the period January 1953 to December 1982?, or ii) does a given data set allow estimation of a \"constant\" or relationship useful for deriving time series covering 1953 to 1982?"},{"index":2,"size":36,"text":"Examples of data sets fulfilling criterion (i) or (ii) are the temperature data in Table 2 and in Bakun (this vol.) and the data on spawning of anchoveta in Pauly and Soriano (this vol.), respectively ."},{"index":3,"size":64,"text":"These criteria, on the other hand, led to the nonconsideration of some zooplankton and other time series reported in the literature, which were too short and could not be utilized as input to derived time series. This approach was neededat least as far as the major contributions included in this book are concernedto prevent a large numbers of nonoverlapping time series from being assembled."},{"index":4,"size":107,"text":"We feel vindicated in this approach in that: i) those who contributed to this book made a special effort to \"stretch\" their data,as far as possible, which now allows simultaneous analysis of a very large number of mutually compatible, uninrermpted time series covering, on a monthly basis, the whole 30-year period from 1953 to 1982; ii) some readers of this book will feel challenged to match the time series they encounter here with time series of their own; and finally, iii) a body of background data is now available allowing other authors working with data covering a shorter period to rigorously test whatever hypothesis they might have."},{"index":5,"size":30,"text":"To facilitate further analysis of the data presented in the various contributions included here, we have included throughout the book tables with unaggregated data which readers are welcome to use%"},{"index":6,"size":106,"text":"The astute reader will notice that this book, despite the restrictive inclusion criteria given above, incorporates more data on the Peruvian upwelling ecosystem than ever published in a single volume. In fact, an attempt was made to make each contribution included here cover the a'lhe bulk of the dau presented in this book is also available as Lotus 1-2-3 files on 5 114' diskettes for IBM PC and compatibles; please contact the f i author for details. bulk of the information available on a given topic. Thus for example virtually all bathytermograph casts hitherto taken off Peru have been analyzed by Brainard and McLain (this vol.)."},{"index":7,"size":39,"text":"The graphs included in this volume, whether original or redrawn from earlier graphs, have all be done at ICLARM, mainly by Messrs. Mark Anthony Go-Oco and Christopher Bunao, usually on the basis of drafts provided by the senior editor."},{"index":8,"size":96,"text":"As the reader will notice, these often include schematic representation of the animals or processes \"meant\" by the graphs. This was not done primarily to make the present volume more accessible to nonscientists (although this would be a nice side-effect). Rather, this style was chosen because we believe it is appropriate for scientists to develop, in the course of their research, what Keller (1983) calls \"a feeling for the organism\" they work on, i.e., to realize their investigations deal with living things and not disembodied entities that manifest themselves as numbers or dots on a graph."},{"index":9,"size":43,"text":"Following Keller (1985) we are thus suggesting \"that questions asked about objects with which one feels kinship are likely to be different from questions asked about objects one sees as unalterably alien\". The reader will decide whether we have asked the right question."}]},{"head":"The Mammals of the Peruvian Upwelling Ecosystem","index":29,"paragraphs":[{"index":1,"size":82,"text":"The most comprehensive account of the mammals of the Peru Currentat least as far as their interactions with fish stocks and fisheries are concernedis that of Northridge (1984). His list of marine mammals from Fishing Area 87 (Southeast Pacific) includes 38 species of cetaceans and pinnipeds. However, his reference to information other than occurrence records and population size estimates (e.g., Aguayo 1975, 1979; Vaz-Ferreira 1979a, 1979b, 198 1,  1982), are extremely sparse, almost vanishingly so when only the Peruvian coast is considered."},{"index":2,"size":109,"text":"This scarcity is, however, due to problems with accessing relevant sources since quite a few publications exist which discuss, at least in anecdotal form, actual or potential interactions between Peruvian mammals and fish stocks (Piazza 1959;Vinatera-Jaramillo 1965;Grimwood 1968;Majluf and Trillmich 1981;Trillmich and Majluf 1981;King 1983;Limberger et al. 1983;Majluf 1985;Rarnirez and Urquizo 1985;Rarnirez 1986 and see references in Muck and Fuentes, this vol.). Northridge (1984) concluded his review of Area 87 by stating that \"there are no documented examples of any effects of competition between marine mammals and fisheries in this area, although the collapse of the anchovy stock could well have affected some species, such as Bryde's whale. \""},{"index":3,"size":79,"text":"We have consulted Dr. P. Rarnirez Advincula, IMARPE's whale biologist, with regard to Bryde's whale (Balaenoptera edeni) as a potential anchoveta predator. He informed us that in all the stomach samples he collected at Paita land station over a period spanning 3 decades, only one (I!) ever contained anchoveta. He also asserted that the whales occurring off Peru actually tend to avoid waters in which anchoveta occur, concentrating instead on areas with abundant schools of sardines, mackerels and Vinciguerria."},{"index":4,"size":91,"text":"The sperm whale (Physeter macrocephalus = P. catodon), similarly, consumes no anchoveta, concentrating instead on squid (Vinatera-Jaramillo 1965), in line with Tornilin (1967) who states that \"the distribution of sperm whales is limited by the distribution of cephalopods, on which they feed, and which tend to prefer warmer, more salty waters\" (Northridge 1984). Burmeister's porpoise, Phocaena spinnipinis appears to be rather abundant off Chile and Peru, with rather high catches reported from the latter country. However, no data are available on its diet off Chile and Peru (Brownell and Praderi 1982)."},{"index":5,"size":56,"text":"This leaves only two species, the South American fur seal Arctocephalus australis (Zimmerman 1783) and the South American sea lion Otariaflavescens (Blainville 1820) as mammal species off Peru that are (a) sufficiently well documented and (b) that could have an impact on the anchoveta resources. The contribution of Muck and Fuentes (this vol.) examines this question."}]},{"head":"Coverage of the Fish Feeding on Anchoveta","index":30,"paragraphs":[{"index":1,"size":32,"text":"A crucial element of the time series of anchoveta biomass and derived series presented in this book is that they are based on an approach which explicitly considers some key anchoveta predators."},{"index":2,"size":52,"text":"Thus, large resources were devoted to estimating the guano bird populations in the Peru System (Tovar et al., this vol.) and their anchoveta consumption (Muck and Pauly, this vol.) as well as the population and anchoveta consumption of seals (Muck and Fuentes, this vol.) and bonito (Pauly, Vildoso et al., this vol.)."},{"index":3,"size":142,"text":"However, we overlooked, in the first phase of this project the potential impact of the mackerel and horse mackerel which we (erroneously) assumed to be largely limited to the anchoveta prerecruits (i.e., to fish of length under 4 cm). Dr. Peter Muck eventually convinced us that mackerel and horse mackerels most probably have a predatory impact on adult anchoveta far more important than that of birds, bonito and seals especially in later years. Unfortunately the contribution by Muck and Sanchez (this vol.) became available too late to be considered explicitly when deriving Virtual Population Analysis (VPA)-based estimates of anchoveta recruitment and biomass. This is probably the reason why Pauly, Palomares and Gayanilo (this vol.) found Mo (i.e., the part of natural mortality not explained by the redators explicitly P included in their VPA model) to take high values, ranging from 2-4 y-."},{"index":4,"size":48,"text":"dt is obvious from this that future estimates (or re-estimate) of anchoveta biomass should consider mackerels and horse mackerel predation explicitly. Information on the size composition of anchoveta in mackerels and horse mackerels' stomachs, along with other biological data are available which could be used for this purpose."},{"index":5,"size":168,"text":"The role of Peruvian hake Merluccius gayi peruanus as a potential anchoveta predator has not been investigated in any of the contribution included in this book, mainly because available time series of population estimates (Espino et al. 1984) do not reach sufficiently far back in time (i.e. do not fulfill criterion (i) above). However, a strong relationship between hake abundance and bottom oxygen concentration (i.e., temperature regime and occurrence of El Ni6o events) has recently been established (Espino et al. 1985(Espino et al. , 1986;;Espino and Urquizo 1986) possibly allowing, in combination with more recent population estimates, the construction of time series of inferred hake abundance covering the period from 1953 to the present. Such time series would provide the chronological \"backbone\" for the hake stomach content data held at IMARPE which suggest that hake preys heavily on anchoveta when its range, normally limited to the north of Peru, is extended southward by the well oxygenated waters typical of El Niiio events (M. Espino; H. Fuentes, pers. comm.)."}]},{"head":"Iteration of Anchoveta Biomasses and Derived Statistics","index":31,"paragraphs":[{"index":1,"size":18,"text":"The astute reader will notice that the interrelationships of the various contributions included here implies an iterative approach."},{"index":2,"size":133,"text":"Thus, in a first iteration, preliminary estimates of anchoveta biomass, available in the literature were used both to estimate the anchoveta consumption by guano birds and seals and to obtain reasonable values of Mo. Then, anchoveta consumption by bird and seals, the estimates of Mo and other data were used to re-estimate monthly anchoveta biomass for 1953 to 1982, which thus represent the results of a second iteration. The data presented in this book could be used quite straightforwardly for a third iteration, but we have abstained therefrom. We have done so because we believe that the results of the second iteration are good enough to be presented, and to allow others to perform (or to join us in performing) this third iteration, with better data and models than have been assembled here."},{"index":3,"size":30,"text":"We hope that the results presented here on the dynamics of the anchoveta stocks off Peru, and of their upwelling ecosystem wil be found useful for managing this valuable resource."}]},{"head":"Introduction","index":32,"paragraphs":[{"index":1,"size":213,"text":"The coastal waters off the west coast of South America, particularly off Peru, are among the most biologically productive regions of the world's oceans (Ryther et al. 197 1). The Peruvian anchoveta (Engraulis ringens) once supported the world's largest fishery. The high productivity of the area is a result of coastal upwelling which is an oceanic response to the southeasterly trade winds which cause offshore Ekman divergence, elevating the thermocline and bringing relatively cold, nutrient-rich water to the euphotic zone where the nutrients can be utilized by phytoplankton photosynthesis (Barber et al. 1985). The upwelling ecosystem off Peru is subject to considerable natural variability, with prominent time scales ranging from days to decades. This paper examines two temporal scales of oceanic variability which are likely to affect populations of anchoveta: seasonal (months) and interannual (years). The seasonal or intraannual variability, being strongly dependent upon the annual solar cycle, is relatively predictable, and therefore likely to promote evolutionary adaptation (Parrish et al. 1983;Bakun, this vol.). The interannual variability, by contrast, has an irregular period which would tend to promote population variations. The dominant form of interannual variability off Peru occurs when the normal seasonal upwelling of nutrients is interrupted by \"El Nifio\" intrusions of relatively warm, clear oceanic waters from the west and north."},{"index":2,"size":109,"text":"The coastal upwelling off Peru is imbedded within the Peru current system, which consists of several more or less independent currents interacting in a rather complicated manner (Wyrtki  1966). Gunther (1936) first distinguished a poleward countercurrent situated between the northwestward flowing Peru Coastal Current and the northwestward flowing Peru Oceanic Current farther offshore. This intermediate current, the Peru Countercurrent or Gunther Current, is a weak and irregular southward flow along 800W and is usually observed only as a subsurface current. At the surface it is usually concealed by the wind drift to the northwest and west. It is strongest near 100 m depth, but reaches to about 500 m."},{"index":3,"size":175,"text":"According to Wyrtki (1965Wyrtki ( , 1966)), the Peru Coastal Current flows northwestward along the coast with velocities of 10-15 c d s . At about 150S, much of this flow turns westward away from the coast and increases speed to 25-35 c d s as it joins the South Equatorial Current. Generally, the Peru Coastal Current is strongest from April to September. North of 1 50S, the wind drift remains northwestward, but it is shallow and the southward flow of the Peru Undercurrent lies immediately beneath the shallow surface layer. The combined system of the Peru Coastal Current, the westwardwind drift, and the subsurface Peru Countercurrent maintain the upwelling along the coast. North of 150S, the upwelling is supplied by equatorial subsurface water which is of high salinity and low oxygen content and flows southward in the Peru Countercurrent. The Peru Oceanic Current, which flows in a more westward direction and is slightly stronger than the Peru Coastal Current, seems to have little direct interaction with the more complicated processes closer to the coast."},{"index":4,"size":73,"text":"The mean topography of the thermal structure of the Eastern Tropical Pacific reflects the ocean currents and has been described by Wyrtki (1966). The thermocline is relatively shallow along the coast at depths of 40 to 60 m and slopes downward in the offshore direction to depths of over 200 m about 1,000 lun offshore. A region of shallow thermocline extends westward from the coast along the equator out to 1300W and beyond."},{"index":5,"size":180,"text":"The current system off Peru is related to the large-scale oceanic and atmospheric circulations over the entire tropical Pacific. The atmospheric circulation over the region is dominated by the Hadley circulation of rising air over the equatorial region and sinking air over mid-latitudes near 300N and 300s. The Hadley circulation creates the high-pressure systems observed over the oceans in these latitudes which are strongest in the summer and weakest in the winter of their respective hemispheres. The meridional Hadley circulation is modified by zonal Walker circulation of rising air over the w a m western tropical Pacific (WTP) and sinking air over the cold, upwelled water of the eastern tropical Pacific (ETP). The zonal Walker circulation normally causes heavy rainfall and low pressure over the WTP and sparse rainfall and high pressure over the ETP. The trade winds result from the combination of the Hadley and Wallcer circulations: the trades blow equatorward from the mid-latitude oceanic highs toward the lower pressure at the equator and westward from the higher pressure over the ETP to the lower presure over the WTP."},{"index":6,"size":97,"text":"The surface wind stress created by the northeast and southeast trade winds drive the warm surface water westward in the North and South Equatorial Currents, respectively. This westward transport of mass and heat depresses the thermal structure and raises the sea level in the WTP. By conservation of mass, the high sea level in the WTP requires a poleward flow of the western boundary currents of the North and South Pacific gyres and eastward flow in the North and South Equatorial Countercurrents and the equatorial Undercurrent or Cromwell Current (within a degree or so of the equator)."},{"index":7,"size":161,"text":"The zonal slope of the sea surface downward from the high sea levels in the WTP to the lower sea levels in the ETP establishes a reverse zonal slope of the thermocline upward from the WTP to the shallow thermocline of the ETP. Meyers (1979) showed that near the equator (between loN and loS), the 140C isotherm varies from depths of 200 to 250 m in the WTP to depths of 100 to 150 m in the ETP. Off the South American coast, local alongshore winds induce offshore Ekman divergence and the associated upwelling. This upwelling elevates the relatively shallow thermocline, bringing nutrient-rich deep water to the euphotic zone where it supports a high level of biological productivity. In addition, the trade winds cause oceanic divergence or surface transport away from the equator. This divergence forces local upwelling along the equator, which produces a region of shallow thermocline and above normal productivity that extends westward along the equator from the coast."},{"index":8,"size":210,"text":"Interannual variations of the stength of the trade winds cause changes in the ocean circulation and related changes in the upwelling of nutrients off Peru. According to the hypothesis of Wyrtki (1975), El Niiio occurs when a weakening or reversal of the trades occurs after a sustained period of anomalously strong trades. The period of stronger than normal winds forces an even greater than normal east to west slope of the sea surface. When the trade winds slacken or reverse, the forcing of the higher than normal sea levels in the WTP is removed. This imbalance generates equatorially-trapped baroclinic disturbances which propagate eastward along the equator in the form of equatorial Kelvin waves (Enfield and Allen 1980). The propagation of these long-period internal waves, and the associated energy, across the entire equatorial Pacific from Indonesia to South America has been observed using an extensive array of sea level monitoring stations (Wyrtki and Nakahara 1984). Upon encountering the South American coast, this energy is observed as a large intrusion of warm water which depresses the normally shallow thermocline and causes a rapid rise in sea level along the coast. As a result, normally arid regions of Peru and Ecuador receive inordinate amounts of rain, with severe flooding occurring during major events."},{"index":9,"size":208,"text":"Interannual variations in the strength of the trade winds are part of a global pattern of surface pressure variation called the Southern Oscillation. The Southern Oscillation is often measured by the difference of atmospheric pressure between weather stations in the ETP and WTP. Quim (1974) and Quinn and Neal (1983) have used the difference of atmospheric pressure between Easter Island (representative of the 1ndonesb.n low) as an index of the Southern Oscillation (SOI, see Table 1). Quinn (1974) demonstrated the strong relationship between anomalously low SO1 values and the occurrence of El N i h off the coasts of Peru and Ecuador. A time series of anomaly of the SO1 pressure difference shows the major El Niiio events of recent decades (Figs. 1 and 2). Note the strong positive SO1 presure differences (and implied strong trade winds) during 1954-1956 and 1970-1971. Subsequent sharp declines in the SO1 pressure difference in the winters of 1956-1957 and 1971-1972 were followed by El Niiio events, as evidenced by the increased SST and sea level at Talara and La Punta. Also, note that the strong 1982-1983 El Nizo was not preceded by a period of strong positive SOI, rather, it occurred during a period of ~redominantly negative SO1 which began in 1976."},{"index":10,"size":151,"text":"The formation of El Nino has been modelled numerically by McCreary (1976) who suggested that the anomalous deepening of the density structure observed during El Ni\"n events dissipates by reflection in the form of westward propagating baroclinic Rossby waves and transmission to the north and south along the coast as low-frequency coastally trapped waves and coastal Kelvin waves. Such bmlinic waves can be observed as anomalous deepenings of temperature and salinity surfaces adjacent to the coast and as anomalous rises of sea level at coastal tide stations. Poleward currents along the coast are created in geostrophic response to the anomalous deepening and change in slope of the density surfaces normal to the coast. The currents reverse to equatorward as the anomalous deepening dissipates. To some extent these processes occur each year and anomalous warm years are merely an extreme condition of the normal annual cycle of events (Chavez et al. 1984)."},{"index":11,"size":154,"text":"Although the interannual changes associated with El ~in\"o events are dominant, longer period fluctuations also occur. In their 34-year time series of temperature at 100 m along the west coast from British Columbia to Chile, Brainard and McLain (1985) showed a marked warming trend occurring between the early and late 1950s, cooling in the 1960s, and warming again in the mid-1970s and early 1980s (see also Tables 2 and 3). The causes of these long-term temperature trends are unknown, but like the interannual variations, they are related to changes in both the large-scale atmospheric and oceanic circulations. The period of below normal SO1 pressure differences during the years 1976-1983 (Fig. 2) is an example. El Nib-like conditions of above normal SST and sea level occurred during much of this period in the northeast Pacific (McLain 1983, and see Fig. 2 and Tables 4 and 5 for monthly sea level data from 1950 to 1974)."},{"index":12,"size":284,"text":"Development of historical time series of subsurface temperature conditions off Peru is important for modelling historical changes in fish populations of the area. This paper presents plots and tables of monthly mean values of four parameters computed from subsurface temperature observations for 1952 to 1984 for use in historical studies. These four parameters are SST, depth of the 140C isotherm, depth to the temperature that is 2.0oC less than the surface temperature (SST-2W), and heat content from the surface to the SST-20C isothenn The depth of the SST-20C isotherm indicates the thickness of the mixed layer which provides a measure of the depth of the thermocline and hence, relates to upwelling and availability of nutrients to the euphotic zone. This definition of mixed layer is similar to that used by Robinson and Bauer (1976), except that they chose the depth that is 20F (1.loC) less than the SST. The SST-20C depth was selected for defining the depth of the thermocline from smooth average temperatures because 20C is larger than the small positive and negative temperature changes near the surface that are present in both the raw data and the analyzed values. Also, a temperature change of 20C is large enough to reach the large gradients found in the thermocline. The 140C isotherm is at depths of 80 to 180 m off Peru and is below the strongest gradients of the thermocline. Variations in the depth of the 140C isotherm are indicators of large-scale vertical movements of the water column, such as upwelling. Also, Barilotti et al. (1984) related the depth of the 140C isotherm off San Diego, California, to the depth of the thermocline and hence to the supply of nutrients for kelp growth."},{"index":13,"size":61,"text":"Heat content down to SST-20C is an indicator of the overall environmental change in the euphotic zone. Combined with wind-derived Ekman transports (Bakun, this vol.; Mendo, this vol.), these vertical temperature parameters can be used to describe the offshore velocity structure which is critical to the reproductive success of the anchoveta (Panish et al. 1983, and other contributions in this vol.)."}]},{"head":"Monthly mean sea level (cm)","index":33,"paragraphs":[]},{"head":"Monthly mean sea level (cm)","index":34,"paragraphs":[]},{"head":"Monthly mean sea surface","index":35,"paragraphs":[]},{"head":"Monthly mean","index":36,"paragraphs":[{"index":1,"size":35,"text":"South Oscillation sea surface Index (SO11 temperature (OC)  temperature (OC)     time series of anomalies from 1941-1986 mean (left) are shown in standard deviation units for intercomparhn between data series. Based on data in Tables 1-5."}]},{"head":"Annual cycle Autocorrelation","index":37,"paragraphs":[{"index":1,"size":243,"text":"Mean values of the four parameters were computed for each month for the 33-year period from 1952 to 1984 for five areas along the coast from 1 to 170s (Fig. 3). Each of the given areas spanned 3 degrees of latitude except for the central-most area which spanned 4 degrees. The zonal extent of each area was chosen so that they would extend about the same distance offshore, roughly 300 km. In addition, all data prior to 1952 were combined to form a single , composite year. Thus, the resulting fields of monthly means for each parameter for the 5 areas covers 34 years (408 months), for a total of 2,040 cells. The average value for the region from 4 to 140s was obtained by averaging the values computed for the three central areas. Mean temperature at 25 m depth intervals fmm the surface to 350 m were also computed for the central area to show vertical variations of temperature with time. The data are plotted as contour isograms of latitude (area) or depth vs. month to show both seasonal and interannual variations. ----Fig. 3. Locations of five areas along the Peru coast for which subsurface temperature profiles w-ere extracted from the FNOC MOODS and used to perform analyses of surface and subsurface temperature. A total of 12,102 profiles were extracted from the MOODS Ties for all five areas for SST analyses. Area I is the northernmost area and Area V, the southernmost. "}]},{"head":"Data Acquisition and Processing","index":38,"paragraphs":[]},{"head":"Data Sources","index":39,"paragraphs":[{"index":1,"size":369,"text":"The profiles of subsurface temperature for the Peru coastal region were acquired from the US Navy Fleet Numerical Oceanography Center (FNOC) in Monterey, California. The profiles were obtained by merchant, naval and research vessels of many nations using a variety of sampling instruments, including bottle casts, mechanical bathythermographs (MBT), expendable bathythermographs (XBT) and electronic conductivity/temperature/depth profiles (CTD). The capability and accuracy of these instruments vary widely: MBTs, with typical accuracies of 0.3 to l.OoC, were used until the development of XBTs in the 1960s. Generally, MBTs reached depths of only 100-200 m, whereas the newer XBTs are capable of depths to 450,700, or even 1,500 m. The accuracy of XBTs are typically 0.1 to 0.40C. Bottle casts and CTD casts from research vessels are capable of any depth, with typical cast depths to 1,000 or 1,500 m and accuracies of 0.001 to 0. loC. Profiles from all of these sources are normally mailed to oceanographic data centers and assembled into common data sets. The time lag between observation and final assembly of the data by the data centers may be 5 to 10 years or longer. To reduce this time lag, many of the prof~les are manually digitized and transmitted by radio in near  With improved digital data acquisition and satellite data transmission systems, the time lags and transmission errors are being reduced. Subsurface temperature profiles from many available sources have been assembled by FNOC in the Master Oceanographic Observations Data Set (MOODS). MOODS is in a compact binary format and contains almost 5 million subsurface temperature profiles globally. The MOODS file is by no means a complete file of all ocean temperature profiles that have even been made; rather, it is only that subset which have been made available to FNOC and merged into the file. Probably many additional profiles exist and if obtainable, could be used to improve analyses of historical conditions. At present, MOODS occupies 12 reels of magnetic tape and is sorted in the sequence: month, 1 degree square of latitude and longitude, year, day and hour. For compactness, many of the temperature profiles are stored at significant or inflection points so that the original data can be recreated by linear interpolation between inflection points."},{"index":2,"size":151,"text":"The distribution of the profiles in time and space is critical for making consistent time series of subsurface temperature. A total for all years of only 12,102 profiles were available in the MOODS file for the five areas along the Peru coast pig. 3). Temperature profiles off Peru are almost nonexistent in the MOODS file for the years prior to 1955 but are more abundant for the years from the late 1950s to the early 1970s. Lags in data assimilation have reduced the amount of data in MOODS since the mid-to-late 1970s, with most of the recent data being acquired via BATHY messages. Profiles taken prior to 1952 have been included in the analysis by combination into the single composite year, labelled \" 195 1\" in the plots and tables. Inclusion of early profiles in the analysis is useful to help establish the edition scheme and to improve the long-term means."},{"index":3,"size":67,"text":"In addition to an insufficient quantity of profiles over the 33-year analysis period, many of the available profiles are very \"patchy\" in their distribution. A cell in the data fields having 100 or more observations is often surrounded by many cells with no observations. This inhomogeneity of the data fields reflects the fact that relatively large numbers of temperature profiles are taken during short, localized research expeditions."},{"index":4,"size":116,"text":"The subsurface temperature profiles in the MOODS file suffer from many types of errors. Teague et al. (1985) have described some of the errors based on samples of the data for the North Atlantic. Many of the profiles have erroneous spikes and tails which require editing, whereby the profiles are truncated to retain the portion of the profile above the erroneous data. About 1 to 5% of the profiles are from incorrect positions or times, as evidenced by reports from land areas. No aitempt was made in this analysis to correct for position and time errors of the profiles as this would have required resorting the profiles into original cruise sequences and tracking each ship individually."},{"index":5,"size":49,"text":"Because of the errors in profiles and more importantly insufficient distribution of observations in time and space, a complicated scheme was necessary to compute reasonable monthly mean values. Thus, the MOODS profiles were edited, monthly means were computed, and then the means were interpolated to fill gaps in coverage."}]},{"head":"Editing Scheme for Subsurjizce Temperature Profiles","index":40,"paragraphs":[{"index":1,"size":116,"text":"The first stage of editing the profiles was a gross error check requiring all reported temperatures to be in the range of -2.0 to +38.0oC and all depths to be nonnegative and increase sequentially. No two temperatures were allowed from the same reported depth, in such cases (which are rare), the depth of the second reported temperature was increased by an arbitrary value of 1 m. In order to eliminate gross ertor spikes, the size of allowable temperature changes between successive reported depth levels was limited between +2.0 and -12.0oC. When data were rejected by these edits, the profile was truncated at the depth of failure and the remaining upper portion of the profile was retained."},{"index":2,"size":99,"text":"--Thc second stage of profile editing checked for unusually strong positive and negative vertical temperature gradients to further reduce unreal spikes and vertical grachents. In the surface layer and thennocline, where the water temperatures were greater than 6.oOC, the vertical temperature gradients were required to be in the range of -2.0 to +0.50C per meter of depth. For reported temperatures less than 6.0oC, the allowable gradients were tightened to -0.5 to +O.loC/m. As for the first stage editing, when gradients exceeding these limits were encountered, the profiles were truncated and the remaining, upper portions of the profiles were retained."},{"index":3,"size":304,"text":"The third stage of editing was a check against the mean and standard deviations of a running series of 10 values of a temperature editing parameter. For SST, depths of the 140C and SST-20C isotherms and heat content, the computed parameter itself was used as the editing parameter. For the vertical temperature series, the temperature at 100 m was interpolated from each profile and used as the editing parameter. The running series was started with the first 10 profiles for each month and 1 degree square of latitude and longitude. (Because the data were sorted in the sequence: month, 1 degree square, year, day and hour, the first 10 profiles in a one degree square were often from years earlier than 1952 and thus errors in the first 10 profiles did not cause serious contamination of the 1952-1984 time series). After acceptance of the first 10 profiles, the mean and standard deviation of the running series of 10 editing values were computed and used to check the next profile. A new profile was accepted if the editing parameter computed from it was within a specified tolerance of the mean of the previous 10 values, where the tolerance was arbitrarily defined to be 1.3 times the standard deviation of the previous 10 values. Each new accepted value was then added to the series and the oldest value in the series deleted. Use of the running series of 10 values allowed the editing mean to move up or down with warm and cold periods defined by the data themselves. Similarly, the scheme allowed the editing tolerance range to widen as the data became more variable (in periods of climatic change or in areas near oceanic boundaries) and to narrow as the data became less variable (during more stable periods or in areas far from oceanic boundaries)."}]},{"head":"Computation of Individual Monthly Mean Values","index":41,"paragraphs":[{"index":1,"size":80,"text":"After editing the temperature profiles as described above, values of the four parameters (SST, depth of the 140C isotherm, depth of the SST-20C isotherm, and heat content down to the SST-20C isotherm were computed for each profile. Individual monthly means of the four parameters were computed for each cell (5 areas x 408 months) for the years 1951 to 1984. The resulting monthly mean fields were very sparse, having mean values in only about 37% of the 2,040 total cells."},{"index":2,"size":82,"text":"Temperatures at 25 m depth intervals from the surface to 350 m were computed from each profile to display vertical variations of temperature versus time in each of the five areas. The data were processed as differences between the surface temperature and the temperature at each 25 m depth interval because of the effect of varying maximum depths of the profiles. Direct computation of mean temperatures from profiles of varying maximum depth can cause unrealistic subsurface temperature gradients (Robinson and Bauer 1976)."},{"index":3,"size":117,"text":"After computation of the individual monthly mean values, the 12 long-term monthly means and 12 between-year standard deviations were computed for each area (or depth for the vertical plots). Here, long-term mean is defined as the mean of all the individual monthly means, e.g., the January long-term mean is the mean of all individual January monthly means. The betweenyear standard deviation (bysd) is the standard deviation of the individual monthly means computed by month to show the interannual variability. The monthly anomalies were then computed as the differences between the individual monthly means and the appropriate longterm monthly mean, e.g., the January 1952 anomaly is the individual monthly mean for January 1952 minus the January long-term mean."},{"index":4,"size":96,"text":"In some cases, no profiles were available in a month during any year for an area (or depth). In these cases, it was not possible to compute a long-term monthly mean. Such gaps in the longterm mean field were filled using a 5 x 5 matrix interpolation which used information from surrounding long-term means. Empty cells were filled with averages of surrounding mean values, weighing proportionately to the square root of the number of years of data represented by the mean and inversely to the square of the distance (in grid lengths) away from the cell."},{"index":5,"size":97,"text":"The f ~l d s of individual monthly mean values were rather noisy, particularly those computed from small numbers of profiles which are considered less reliable than those based on relatively large numbers of profiles. To reduce the errors associated with limited numbers of profiles, the individual monthly means were adjusted toward the long-term mean for each month, i.e., means based on only a single profile were set to the average of the mean and the long-term mean for that month, while means based on two or three profiles were weighed proportionately less toward the long-term mean."},{"index":6,"size":188,"text":"Monthly anomalies were computed as the difference of the adjusted individual and longterm monthly means. To partially fill the gaps in the anomaly field between data values, the same 5 x 5 matrix interpolation scheme was used as for the long-term means, weighing proportionately with the square root of the number of profiles represented by the mean and inversely with the square of the distance away from the cell. In regions that were 3 or more cells away fmm mean values, no interpolation of the anomaly was made. Use of this interpolation scheme increased coverage of the field from about 37% to about 91%. Use of a 5 x 5 matrix interpolator was reasonable as autocorrelation functions (not shown) of the individual monthly means were computed and had magnitudes greater than 0.4 for lags of at least two months in time and at least three areas (9 degrees of latitude) along the coast in space. The correlations are in agreement with Enfield and Allen (1980) who showed similar strong coastwise coherence of sea level and SST along the coast of North and South America from Alaska to Chile."},{"index":7,"size":101,"text":"After adjustment and interpolation of the anomaly field, the fields of individual monthly means were recomputed. Gaps in the coverage were partially filled by the addition of the interpolated anomaly field and the long-term mean field. Use of the anomaly fields to interpolate the monthly mean fields for filling gaps in coverage is based on the assumption that the anomaly fields are smoother in time and space than the monthly means. This assumption is justified because of the large seasonal changes that are observed in the monthly means but relatively smoother changes in the anomalies (see, e.g., Fig. 2, autocorrelation functions)."}]},{"head":"Spatially Averaged Monthly Means for the 4 to 140s Region","index":42,"paragraphs":[{"index":1,"size":139,"text":"Monthly anomalies for the three central areas were further averaged to make time series of monthly mean anomalies for the entire region from 4 to 140s. The averaged monthly anomalies were then added to the appropriate averaged long-term means to obtain time series for the four computed parameters by month for the region 4-140s. Plots and tables of the time series values are presented along with the total combined number of observations for the three central areas. Values are only given if observed or interpolated mean values were available for all three areas. This requirement reduced the coverage of the time series to about 90% of the 408 possible months. For cases where anomaly values were not available for each of the three areas for any particular month, asterisks are printed in the tables and values are not plotted."}]},{"head":"Results and Discussion","index":43,"paragraphs":[{"index":1,"size":138,"text":"Each of the parameters (SOI, SST, sea level, depth of the 140C isotherm, depth of the SST-20C isotherm, vertical structure of subsurface temperature and heat content from the surface to the SST-20C isotherm) is presented separately. The data are displayed in a variety of formats to emphasize the seasonal and interannual scales of variability, both horizontally abng the coast and vertically through the water column. For each of the parameters, tables and plots of spatially-averaged (for the region from 4-140s) monthly means are presented. Plots of the longterm annual cycle, between-year standard deviation, monthly anomaly in standard deviation units, and autocorrelation functions of the anomalies for each of the parameters are presented to describe seasonal and interannual variability. Also, time-latitude and time-depth contour plots of profile data are presented to show horizontal and vertical variations of subsurface temperature."}]},{"head":"Southern Oscillation Index","index":44,"paragraphs":[{"index":1,"size":116,"text":"Time series of the monthly mean Southern Oscillation Index (Fig. lA, Table 1) and anomaly of SO1 (Fig. 2A) show the buildups and subsequent declines of pressure differences associated with the onset of El Niiio, as described earlier. Major buildups and declines occurred in 1949-1952,1954-1958, 1970-1972and 1975-1977. The most recent decline in 1982-1983, associated with that strong El Niiio, followed a long period (1976)(1977)(1978)(1979)(1980)(1981) of relatively weak negative pressure difference. There was also a sharp decline in 1979 associated with a weak coastal warming event in that year. Perhaps the 1979 event would have been more notable (more comparable to other moderate or weak El Niiios) if it had not occurred during an already warm period."},{"index":2,"size":146,"text":"The SO1 has a relatively strong annual cycle (Figs. 1 and 2) which varies from a peak difference of about 13 mbs in February, indicating strongest trade winds in late austral summer to a low of about 7 mbs in May, indicating weakest trade winds in late austral fall. The interannual variability of SO1 is relatively constant throughout the year as indicated by the similar values of between-year range and standard deviation. The SO1 is moderately persistent in time with an autocorrelation of anomaly of about 0.4 at one month lag. From 12 to 36 months lag, the autocorrelation of the anomaly remains very close to zero, then becomes weakly positive at lags of 36-48 months. This suggests that the period of important interannual changes in the SO1 is greater than 3 years, in agreement with the frequently reported period for El NiEo of 3-7 years."}]},{"head":"Sea Surface Temperature","index":45,"paragraphs":[{"index":1,"size":170,"text":"Time series of monthly mean SST (Figs. 1B and lC, Tables 2 and 3) and anomaly of SST (Figs. 2B and 2C) at Talara and La Punta (Callao) and spatially-averaged SST for the region from 4-140s (Table 6, Figs. 4A, 5A) show significant seasonal and interannual variability of SST. Both the shore station and spatially-averaged SST data show the major El Niiio and anti-El Niiio events. Positive anomalies occurred in the years 1953,1957-1958,1965,1972-1973,1976-1977, 1979 and 1982-1983. Each of the figures also show longer period interannual variations: cool conditions in the early 1950s, warm conditions in the late 1950s, moderately cool conditions throughout the 1960s and early 1970s (except the 1965and 1972-1973 El Niiios), and finally a long-term warming during 1976-1983. Comparison between the two shore stations, Talara in the north and La Punta in the south, shows the northerly station to have more low-amplitude, high frequency variability than the southerly station. This difference is assumed to be caused by the more complicated equatorial ocean dynamics occurring in the northern region."},{"index":2,"size":122,"text":"Along the coast, the El Niiio events of 1953,1957-1958,1965,1969,1972-1973,1976-1977and 1982-1983(Rasmusson 1984) ) are seen as tongues of warm SST, extending variable distances southward (Fig. 6). A moderate warming occurred in 1979-1980, in agreement with the below normal SO1 that year. The extreme magnitude (large region of SST > 280C), duration, and coastwise coherence of the 1982-1983 event distinguish it as the most significant warm feature of this series. The poorly documented 1953 El Niso shows a surprisingly strong surface manifestation of warm water. The 1954-1956 cold event is only weakly evident, probably due to sparse data. With the exception of the 1982-1983 warm event, each of the warm surface events are shown to be preceded by a period of anomalously cool SST."},{"index":3,"size":136,"text":"The annual cycles of SST at the two shore stations (Figs. 2B and 2C) and for the spatiallyaveraged region (Fig. 7A) vary from highs during the austral fall to lows during the austral spring. The amplitude of the annual variation of SST is greatest nearer to the equator and the complex dynamics associated with the interaction of equatorially-trapped waves with the eastern boundary (Fig. 6). The spatially-averaged long-term means (Fig. WA) show high SST (>23.50C) from January through March or April, when warm water intrudes from the north, followed by a rapid transition to lower temperatures in April with the onset of upwelling along the central and southern portions of the coast. SSTs of 17-190C occur during the upwelling regime from May to October along the coast, except for the northernmost area where upwelling is weak."},{"index":4,"size":105,"text":"Along the coast, the annual cycle is strong, varying between upwelling and nonupwelling regimes for the central and southern areas and the weak seasonal variation for the northern area.  . Timelatitude contour plots of annual cycles and between-year standard deviations of (A) SST, (B) depth of 14°C isotherm, (C) depth of SST-2OC isotherm, and (D) heat content down to SST-2OC isotherm for 5 areas along the Peru coast. The long-term mean annual cycles are shown in the left-hand plots. When no profiles were available for a month in any year, the long-term mean was interpolated. The between-year standard deviations are shown in the right-hand boxes."},{"index":5,"size":60,"text":"The time-latitude between-year standard deviation plot (Fig. 7A, left-hand plot) shows the highest interannual variability to occur during the fall transition from the warm current regime to the upwelling regime, particularly in area 2 which is between the strong upwelling to the south and the weak upwelling to the north. The interannual variability of SST is lowest during peak upwelling."},{"index":6,"size":95,"text":"The autocorrelation function of the anomaly of the spatially-averaged SST (Fig. 5A) is moderately persistent with a lag one value of about 0.6. After 18 months' lag, the autocorrelation function begins to rise, peaking again at a lag of about 44 months. Anomaly of SST at the two shore stations has autocorrelation functions with moderately strong persistence at lags one and two, negative correlation between lags of 12 and 36 months, and weakly positive correlation after a lag of about 40 months (see Figs. 2B, 2C). This pattern is similar to that described for SOL"}]},{"head":"Vertical Structure of Subsurface Temperature, 0-350 m","index":46,"paragraphs":[{"index":1,"size":108,"text":"The interannual variability of subsurface temperature for the central area (Area III) off Peru (Fig. 8) shows monthly variations of isotherm depths from the surface to 350 m for the period 1952 to 1984. A similar plot of the anomaly field was used for the analysis (but not shown because anomalies must be carefully analyzed to avoid misinterpretation of events caused by slight phase shifts). Interestingly, the 12, 14 and 160C isotherms show a general long-term depression of the thermal structure for the period 1976-1984, in agreement with the changes in SO1 and SST discussed previously and with other reports of a large-scale coastal warming during the period."},{"index":2,"size":280,"text":"Shorter duration depressions of the isotherms are observed for the 1957-1958,1965-1966, 1969,1972-1973,1976-1977,1979-1980 and 1982-1983 El Niiio warming events. The magnitude and vertical extent of these isotherm depressions varies noticeably between different events. Each of these El Niiio events is characterized by moderate to strong surface warming. The anomaly field (not shown) has double peaks for most of these warming events, as has been reported by others for many El Niiio events (e.g., Cane 1983;Reinecker and Mooers 1986). The 1957-1958 El Ni6o appears to be of shallower extent but of longer duration than most of the other events, lasting for about 3 years. The 1965The -1966 El Niiio had intense surface warming (down to 150 m) which began in January 1965 and lasted until about July, followed by a second, weaker warming which peaked in about December. There was also a weak isotherm depression between 275 and 350 m. The 1969 El Nib had a weak signal from the surface down to about 300 m. The 1972The -1973 El Niiio was similar in vertical extent and duration to the 1965 event, except the second peak was less defined. The 1976-1977 event was moderately strong at all depths from the surface down to 350 m. The 1979-1980 event had a weak depression at all depths. During the 1982-1983 El Nifio, a strong depression of 50 to 80 m was observed at all depths. For this event, it is interesting to note that the 12 and 140C isotherms were depressed 5-6 months prior to the depressions of the surface layer isotherms. It is yet to be determined whether this relates to the idea of downward and poleward propagating coastally trapped waves (McCreary 1976)."},{"index":3,"size":196,"text":"The annual cycles of subsurface temperatures for the five areas along the coast are shown by vertical contour plots of the long-term monthly means (Fig. 9). The areas are arranged from north to south from left to right across the page. The strongest vertical temperature gradients are in the upper 75 m, indicating a relatively shallow mean thermocline. The isotherms display a relatively linear slope upwards with increasing latitude (southward) along the coast, as would be expected. The 12 and 140C isotherms shoal from mean depths of about 285 m and 170 m for Area I, near the equator, to mean depths of 205 m and 90 m for Area V, in the south. Similarly, the SST varies between 21 and 240C for the northern area and between 16 to 21oC for the southern area. Each of these plots show a strong annual cycle having relatively warm temperatures during the austral summer, with annual highs occurring in February and March, and cooler temperature during the austral winter, with annual lows occurring in September. This pattern of the annual cycle becomes less apparent with increasing depth, where the 12,14 and 160C isotherms have an interesting double peak."},{"index":4,"size":28,"text":"The interannual variability of subsurface temperature off Peru is shown by vertical contour plots of the between-year standard deviation (bysd, see Fig. 10) for the long-term monthly means   "},{"index":5,"size":24,"text":"10.0-1 1.9 OC 14.0-1 5.9 OC 18.0-1 9.9 OC 22.0-23.9 OC Month 12.0-1 3.9 OC 16.0-1 7.9 OC 20.0-21.9 OC 224.0 OC Area II"},{"index":6,"size":26,"text":"Month Fig. 10. Between-year standard deviation of subsurface temperature vertically from 0 to 350 m for the 5 areas a l o q the Peru mast."},{"index":7,"size":2,"text":"1.0-1.9 \"C"},{"index":8,"size":21,"text":"The highest interannual variability occurs in the upper 100 m d u r h the relaxed upwelling period (December to June)."}]},{"head":"\"C","index":47,"paragraphs":[{"index":1,"size":64,"text":"just discussed. The highest variability is found in the upper 50 m, as would be expected for the thermocline. Seasonally, this upper layer variability appears to be highest (bysd = 2.7-3.0oC) from March through July and lowest (bysd = 0.7-1.20C) in September. With a few exceptions, the interannual variability is consistently low (bysd = 0.2-0.90C) below 100 m in each of the 5 areas."}]},{"head":"Depth of the 140C Isotherm","index":48,"paragraphs":[{"index":1,"size":132,"text":"Unlike the previous section which described time variations of the thermal structure . vertically for a single area and the long-term annual cycle of the slope of the thermal structure for the five areas along the coast, this section examines temporal and spatial variations of the depth of the 140C isotherm in greater detail. The time series of monthly mean depth of the 140C isotherm (Fig. 11, Table 7) shows considerable seasonal and interannual variability. The 140C isotherm deepened during each of the warming events, with the most striking example occurring during the 1982-1983 El NiZo. The anomaly of the depth of the 140C isotherm (Fig. 5B) shows persistent deep or shallow anomalies lasting several years. The 140C isotherm deepened in 1976 and remained anomalously deep throughout the rest of the record."},{"index":2,"size":77,"text":"The annual cycle of the depth of the 140C isotherm (Fig. 7B) differs from the annual cycles of each of the other parameters in that it displays a double peak. Seasonally, the depth of the 140C isotherm for the region from 4 to 140s has maximum depths in April and July and minimum depths in June and September. This double peak exists for each of the five areas (Fig. 9), although the months of occurrence differ slightly."},{"index":3,"size":100,"text":"The 140C isotherm is relatively deep in the north and shallow in the south (Figs. 9 and 11). The annual long-term mean depths of the 140C isotherm for Areas I through V are 152,135,109, 90, and 78 m, respectively. The transition from depth to shallow depths is usually rapid, typically occurring in 1-2 months. The interannual variability of the depth of the 140C isotherm is highest in December and January for the three northern areas (between-year standard deviations greater than 50 m), associated with the intrusion of the warm water. The interannual variability is lowest during peak upwelling in September."},{"index":4,"size":151,"text":"Along the coast, the depth of the 140C isotherm (Fig. 11) provides an indication of the coastwise interannual variability of the thermocline depth and the effect of coastal upwelling on the thermal structure. The 140C isotherm deepened moderately during the years 1957-1958, 1965,1969,1976-1977and 1979-1980and strongly during the 1972-1973and 1982-1983El Niiios. This plot shows the 1982-1983 El Niiio to be the largest event of the record, both in magnitude and duration. The 140C isotherm remained below 200 m for most of the period from February 1982 through July 1983 for the northern two areas. Likewise, the 140C isotherm remained significantly deeper than normal during this period for the southern three areas. Both plots show the 1957-1958,1965,1969,1972-1973 and 1976-1977 El Nifios were preceded by cold period having shallow depths of the 140C isotherm. The 1979The -1980The and 1982The -1983 warm events, by contrast, occurred during the long-term coastal warming from 1976 to 1984."},{"index":5,"size":240,"text":"The 1979-1980 warming, which was not included in Rasmusson's (1984) list of El Niiios, had moderate signals for each of the three parameters thus far discussed. Norton et al. (1984) described a strong surface warming in 1979-80 in the California current system which attenuated rapidly with depth, seemingly unrelated to tropical warming. Wyrtki (1975) pointed out that El Nifio conditions off the coast of Peru were not caused by a local weakening of the upwelling favorable winds, as had been previously hypothesized. He showed that not only did the upwelling favorable winds not weaken but also appeared to have strengthened during El Niiio events. Using Bakun's (this vol.) time series of wind stress, turbulent mixing index, and offshore Ekrnan velocity and transport, we now strengthen Wyrtlu's argument and show that each of the major El Niiio events of the past three decades occurred during periods of anomalously strong southeast trades and offshore Ekman transport. One would expect intense offshore transport to be associated with intense coastal upwelling and shallow thermocline. However, comparison of the time series of offshore transport with depth of the 140C isotherm indicates the opposite. Periods of strongest offshore transport correspond to periods of deepest depth of the 140C isotherm. The 1957-1958,1965,1972-1973 and 1982-1983  El Niiio periods of anomalously deep thermocline occurred during the four highest peaks of offshore Ekman transport. Thus, the thermocline deepens sharply during periods when local wind forcing should produce anomalously shallow thermocline."}]},{"head":"Depth of the SST-20C Isotherm","index":49,"paragraphs":[{"index":1,"size":143,"text":"The depth of the SST-20C isotherm is an indicator of the thickness of the surface mixed layer, i.e., the depth to the top of the thermocline. The time series of monthly mean depth of the SST-20C isotherm (Fig. 4C, Table 8) show that variations in the depth of this isotherm occur over both seasonal and interannual time scales. Although anomalies (Fig. 5C) occurred during the major El Nifio events, the interannual variations of the depth of the SST-20C isotherm are less closely correlated with El Niso events than were the other parameters discussed. Rather, the dominant interannual variations appear to occur over longer time scales. The SST-20C isotherm was anomalously shallow throughout most of the 1 9 0 0 ~~ anomalously deep from 1970 to 1976, near the long-term mean annual cycle from 1976 until 1982 and deep during the 1982-1983 El Niiio."},{"index":2,"size":140,"text":"The annual cycle of the depth of the SST-20C isotherm (Fig. 7C) is strong, varying from a minimum depth of 20-25 m from January through March (austral summer) to a maximum depth of 60-70 m from July through September (austral winter). This annual cycle corresponds well with Bakun's (this vol.) seasonal wind mixing index and surface wind stress calculations, further establishing the depth of the SST-20C isotherm as a reasonable measure of the mixed layer depth. Thus, the annual cycle of the depth of the SST-20C isotherm fluctuates with the upwelling cycle, being deep during the upwelling season and shallow during the nonupwelling season. Upwelling elevates the thermal structure in response to offshore transport of the surface water. The elevated thermal structure combined with increased turbulent mixing during the upwelling season apparently force the observed deepening of the SST-20C isotherm."},{"index":3,"size":107,"text":"The between-year standard deviation and range between extrema vary between lows of about 5 and 20 m, respectively, in May and June to highs of about 20 and 40 m, respectively, in August and September (Fig. 7C). There is an inverse correlation between SST and depth of the SST-20C isotherm. The autocorrelation function of the depth of the SST-20C isotherm indicates weak persistence (Fig. 5C). This suggests that the mixed layer changes due to local rather than large-scale processes, which makes sense because it is primarily an indicator of the thickness of the wind forced mixed layer (or local upper layer stratification due to heat budget consideraeons)."},{"index":4,"size":80,"text":"Along the coast, the depth of the SST-20C isotherm slopes from a shallow annual mean depth in the north (Area I) of 31 m to a relatively deep annual mean depth in the south (Area V) of 50 m. The maximum coastwise between-year standard deviation occurs in August in Area 11, which is located between the weak upwelling area to the north and the strong upwelling areas to the south, reflecting interannual variations in the northerly extent of the upwelling."}]},{"head":"Heat Content from the Surface to the SST-20C Isotherm","index":50,"paragraphs":[{"index":1,"size":141,"text":"The heat content from the surface down to the SST-20C isotherm (Fig. 4D) is computed as the vertically averaged mean temperature down to the SST-20C isotherm multiplied by that depth. Since the magnitudes of depth variations (in meters) are significantly greater than for temperature variations (in degrees Celsius), the computed heat contents are dominated more by depth than temperature. Thus, heat content is low (300 to 5000C x m) during nonupwelling periods when the surface layer is warm but the SST-20C isotherm is very shallow. Conversely, heat content is high(>l,OOOoC x m) during the upwelling regime when the surface layer is cool but the SST-20C isotherm is deep. Time series of monthly mean heat content (Fig. 4D, Table 9) and anomaly of heat content (Fig. 5D) reveal interannual variations almost identical to those described for the depth to the SST-20C isotherm."},{"index":2,"size":180,"text":"The annual cycle of heat content has characteristics similar to the annual cycle described for depth to the SST-20C isotherm (Fig. 7D). The long-term mean values vary from about 500 mC   from January through March to about 1,000 OCm from July through September. The betweenyear standard deviation varies from about 100 OCm in June to about 200 OCm in November and February. The range between extrema varies from 200 to 600 OCm. The autocorrelation function of heat content is 0.5 at lag one and decreases rapidly thereafter, indicating weak persistence with time (Fig. 5D). Along the coast, the long-term monthly means of heat content generally slope downward to the south. For instance, the long-term annual mean heat contents for Area I in the north and Area V in the south are 582 OCm and 843 OCm, respectively. This alongshore variation is a consequence of the deepening of the mixed layer in response to upwelling, which is stronger in the south. The interannual variability of the heat content is highest in areas and months around the edges of the upwelling regime."}]},{"head":"Sea Level","index":51,"paragraphs":[{"index":1,"size":249,"text":"Sea level represents a vertical integral of the thermohaline structure over the entire water column. Integrating over the water column has the effect of combining many subsurface processes into a single parameter. Frequently, this combining of factors provides an invaluable indication of large-scale oceanic change. Monthly means and anomalies of sea level at Talara and La Punta (Figs. ID,1E and 2D,2E,Tables  4 and 5) show seasonal and interannual variability similar to that described for SST. Sea level at both coastal stations was variable during the early 1950s, moderately low during the mid-1950s, moderately high for the 1957-1959 El Niiio episode, slightly below normal for most of the 1960s (except the 1965 and 1969 warm events when it was above normal), very high during the 1972-1973El N i k , and variable until 1974. . The annual cycle of sea level is characterized by relatively high levels from February through June and low levels from August through December. The interannual variability is generally high from December through June and low from August through September. Bigg and Gill (1986) showed that the long period response of sea level off Peru separates into a remotely forced component mainly due to zonal winds along the equator to the west, and a locally driven component where sea level slopes to balance the alongshore wind. Their examination of the annual component of sea level indicates that the locally forced component dominates, whereas the remotely forced component plays a major role at semiannual and interannual periods."}]},{"head":"Summary and Implications","index":52,"paragraphs":[{"index":1,"size":112,"text":"All of the time series presented show considerable seasonal and interannual variability. Each series had a strong annual cycle, dominated by seasonal shifts from an intensified upwelling regime from May to October to a relaxed upwelling regime in which warm water intrudes from the north from January through March. Contrary to this single peak pattern observed for SOI, SST, sea level, depth to the SST-20C isotherm, and heat content, deep isotherms, represented by the depth of the 140C isotherm, had double peaks in their annual cycles. Along the Peru coast, the seasonal variability increased from north to south, with lowest variability associated with weaker upwelling in the north than in the south."},{"index":2,"size":161,"text":"Interannual variations of most of the series were similar, being dominated by remotelyforced El NEo signals. Contrary to this pattern, however, interannual variations of the depth to the SST-20C isotherm and heat content down to this isotherm were only weakly correlated to El NCio signals. Interannual variations of these two parameters appear to occur at longer periods. Weaker persistence suggests that these two parameters may be dominated by local rather than large-scale processes, such as local wind or heating events. The parameters which were correlated to El Niiio signals showed that coastal waters off Peru (4-140s) were generally cool in the early 1950s, moderately warm during 1957-1959, near normal during 1960-1965, warm in  1965, near normal during 1966-1968, warm in 1969, cool in 1970 and 197 1, very warm in 1972,  cool during 1973-1975, warm during 1976-198 1, extremely warm in 1982-1983 and variable in  1984. La Punta SST was below normal in 1984-1985, but the 140C isotherm was still depressed."},{"index":3,"size":117,"text":"The warm events described by the W v e interannual variations represent a general deepening of the thermal structure along tht coast as indicated by a depression of the 140C isotherm. Such depressions cause a change in the slope of the thermal structure normal to the coast and a tendency to increase transport of warm water and associated organisms poleward along the coast. Furthermore, deepening the thennal structure would reduce the biological productivity of the surface waters by reducing the ability of upwelling favorable winds to upwell nutrient-rich water. These depressions of the thermal structure have been shown to occur even during periods of maximum offshore transport, which normally correspond to maximum upwelling and availability of nutrients."},{"index":4,"size":203,"text":"The effects of interannual variability of subsurface temperature on Peruvian anchoveta populations are difficult to access. If one assumed that some combination of factors associated with warming and depression of the thermal structure is detrimental to anchoveta recruitment and/or growth, a scenario of events based on the subsurface variability presented here may be as follows. Heavy fishing pressure in the late 1960s and early 1970s combined with the strong El Niiio of 1972, which had a deeply depressed thermal structure, caused a collapse of the anchoveta population. The stock then recovered slightly in [1974][1975][1976]  (Avaria 1985) following cool water conditions (upwelling) of 1973-1975 as described by the depression of the 14OC isotherm. However, this slight recovery was subject to continued strong fishing pressure. The moderate 1976-1977 El Niso and the warm, depressed conditions thereafter, especially during the 1982-1983 El Ni5o;have prevented good recruitment, resulting in very low anchoveta populations. A reversal to cooler conditions since 1983 may be associated with a modest recent recovery of the population. Clearly, the actual ecosystem of the Peruvian anchoveta is much more complicated than this simple scenario suggests. This scenario does, however, illustrate the potential value of multiple environmental time series such as presented in this volume."}]},{"head":"Introduction","index":53,"paragraphs":[{"index":1,"size":184,"text":"By international convention, weather observations are recorded routinely on a various types of ships operating at sea. These maritime reports remain the primary source of information on large-scale variability in the marine environment. Even with the increasing development of satellite observation systems, analysis of time series of decadal length and longer must continue to depend heavily on these maritime reports for some time to come. Observations of wind speed and direction, air and sea temperature, atmospheric pressure, humidity and cloud cover included in these reports provide a basis for estimating a number of environmental variables pertinent to the study of variqtions in ocean climate and of effects of these variations on the associated communities of Ifiarine organisms. In this paper, the historical files of these observations are summarized to monthly estimates of properties and processes at the sea surface within the extremely productive upwelling ecosystem off central and northern Peru. The 32-year period treated encompasses several dramatic El Ni6o events and the spectacular rise, collapse, and indications of a recent rebound, of the largest exploited fish population that has ever existed, the Peruvian anchoveta."},{"index":2,"size":236,"text":"Although remarkably rich both in biological productivity and in climatic scale ocean variability, the area off Peru is rather poor in maritime data density. Thus the region presents a particular challenge to the methodologies employed here. The area is very sparsely sampled in comparison to the corresponding eastern ocean boundary ecosystems of the northern hemisphere, with most of the reports coming from a narrow coastal shipping lane lying within about 200 krn of the coast (Panish et al. 1983). Maritime reports are subject to a variety of measurement and transmission errors, of which improper positioning is perhaps the most troublesome, sometimes introducing very large errors in all derived quantities (e.g., when a wrong hemisphere, etc., may be indicated). And it is difficult to establish effective procedures for rejecting erroneous reports without also suppressing indications of real variations, particularly in the area off Peru which is perhaps uniquely subject to drastic and abrupt natural environmental perturbations. For example, early indications of the 1982-1983 El Ni%.evmt went unnoticed by meteorological agencies in Europe and North America, because the reports which clearly indicated an event of unprecedented intensity were so far from the norm that they are rejected as erroneous by the automated data editing procedures (Siege1 1983). In addition, even when no actual errors are involved, irregular distribution of the reports in both time and space may introduce biases and nonhomogeneities into time series consmcted from these data."},{"index":3,"size":317,"text":"Tests of the precision of the methodology on interyear time scales, involving subsamplings of the much richer data distributions off the Iberian Peninsula in the northeast Atlantic Ocean, have indicated benefits to be gained by utilizing rather large areal samples, i.e., of the order of 10 degrees of latitude and longitude in extent, with the increase in report frequency overriding increases in sampling variance resulting from incorporation of additional spatial variability (Bakun, unpublished data). These same tests have indicated that the use of the ordinary 'standard error of the mean' provides a useful guide to the precision of monthly estimates, even though the underlying processes may be very highly variable on much smaller temporal and spatial scales than those used for data summarization. For the time series presented herein, reports available within an area extending some 10 degrees of latitude along the Peru coast and about 4 degrees of longitude offshore (Fig. I), between Talara and a point just to the south of Pisco, were ---a 9 .JI composited together. These composite samples are assumed to Characterize temporal variability, at least in the relative sense, in conditions affecting the neriticfkh habitat along that stretch of coastline which appears to have some degree of turd tmity both in terms of environmental processes and biological community (Santander \"% 1 8QI Pa&s&t al. 1983). The rather ragged offshore edge of the summary region was chosen to facilitaie initial extraction of the reports from the data archive files. Consistent features of spatial variability rend to be much less intense in offshore areas of coastal upwelling regions than in coastal areas; thus no substantial effect of the irregularity of shape of the offshore boundary is expected. Also all the monthly summaries are treated identically in terms of areal selection and so time series homogeneity is preserved. In any case, report density is extremely low at the outer edge of the summary area."}]},{"head":"Assembly of Data Series","index":54,"paragraphs":[{"index":1,"size":241,"text":"Impossible or highly improbable values occur occasionally in the maritime report files, due to keypunch errors, etc. In the data record format, temperature values between -99.9 and 99.90C are possible. Initial efforts to construct the data series resulted in rather large standard errors for certain of the monthly values due to incorporation of improbable data. For this reason, only values falling between the limits 11 to 3 1oC were accepted as valid observations of air temperature, sea surface temperature, or 'wet bulb' air temperature, for this region. (Note that the lower bound on the wet bulb temperature caused only 16 reports, no more than a single report in any one month, to be rejected). Wind speeds of up to 199 knots (102 rn/sec) are possible in the record format. Erroneously high wind speeds have a particularly serious effect since wind speed is squared in the stress computation and cubed in the wind mixing index formulation. Reports of wind greater than 45 knots (23 rn/sec) occurred within the summary region less than ten times in the entire 32-year record and were in no case corroborated by neighboring (in either space or time) data. Thus wind reports exceeding this value were excluded in preparing these time series. The data record format limited wind direction to values between 0 and 360 degrees, cloud cover observations to the range 0 to 100% of sky obscured, and barometric pressure to values between 890 and 1,070 millibars."},{"index":2,"size":311,"text":"In assembling the monthly data samples, if any one of the reported values of sea surface temperature, barometric pressure, wind speed, or wind direction, were missing or unacceptable the entire report was excluded from the summaries. These four observed properties are sufficient to produce time series of sea surface temperature (Table 2), atmospheric pressure (Table 4), wind stress components (Tables 5 and 6), and wind mixing index (Table 7). The numbers of reports having acceptable observations of these four items are entered as the first of the three numbers shown for each month in Table 1. In addition, if a valid cloud cover observation was available the report was also incorporated in the cloud cover series (Table 3); numbers of reports including acceptable observations of these five items are entered as the second number of each monthly set in Table 1. Finally, if acceptable values of both air (diy bulb) temperature and either wet bulb or dew point temperature were included, the report was also used for construction of time series of atmosphere-ocean heat exchange components (Tables 8 to 11). Numbers of available reports containing acceptable observations of all seven properties required to construct all the time series presented in this paper are shown as the third number under each month in Table 1. All computations of derived quantities were performed on each individual report prior to any summarization process. A simple mean was taken as an estimate of the central tendency of each monthly sample. Computed standard errors of these mean values are displayed within the parentheses following each monthly value presented in the various data tables. An approximate 95% confidence interval estimate can thus be generated by multiplying the indicated standard error by the factor 1.96, and adding and subtracting the result from the monthly mean value (point estimate) to yield the upper and lower limits of the interval."},{"index":3,"size":170,"text":"A small percentage of the reports contain wind observations in which the direction is noted as \"variable\"; i.e., no direction could be assigned. This properly occurs only when the wind speed is very low. In these cases the wind speed is used, as reported, in the calculations where it enters as a scalar quantity, i.e., in the calculations of wind mixing index, evaporative heat loss and conductive heat loss. In the computation of surface wind stress, wind enters as a vector quantity and directionality is crucial. Accordingly, for the surface wind stress calculations, variable winds are treated as calms. Because the wind speed enters the calculation as a \"square\", low wind values act essential as zeros in their effect on the monthly means, and so treating these weak variable wind observations as calms has no substantial effect. Also, the net effect of a stress from one direction is cancelled by an equal stress from the opposite direction, and so treating variable-directional stress as equivalent to calm conditions makes physical sense."}]},{"head":"Sea Surface Wind Stress","index":55,"paragraphs":[{"index":1,"size":7,"text":"Sea surface stress was estimated according to:"},{"index":2,"size":298,"text":"where TXand Ty are components of stress directed onshore and alongshore, respectively; a characteristic onshore direction of 62 degrees and an alongshore direction of 332 degrees (from true north) was assigned to the entire summary area. P, i s ~h e density of air, considered constant at 1.22 kg/m3. CD is a dimensionless drag coefficient. IWloI is the wind speed at 10 m height. U10 is the onshore-directed component of wind velocity; V10 is the alongshore-directed (positive equatorward) component. For the data series presented in the tables, CD was considered to be a constant equal to 0.0013. The use of this constant drag coefficient has been a somewhat standard practice in climatological studies of upwelling regions (Bakun et al. 1974;Nelson 1977;Parrish et al. 1983). However, it is recognized that the value of the drag coefficient is actually a variable which depends on the nature of atmospheric turbulence near the sea surface. Thus a dependence on both atmospheric stability and wind magnitude near the sea surface is indicated; the stability effect is particularly important in reducing air-sea transfers in upwelling regions due to the stable atmospheric boundary layer formed over cool upwelled surface water. No clear consensus as to the proper formulation of these dependencies is presently available. However, a reasonable variable drag coefficient formulation has been chosen and has been applied to these data for evaluation of possible differences from results based on the constant drag coefficient formulation. In this case we follow the method of Nelson (1977) for incorporation of the atmospheric stability effect, which is based on a bulk Richardson number parameterization (Deardorff 1968). We incorporate a dependence on wind speed according to the recommendations of Large and Pond (1981) who find a linear increase in the drag coefficient at wind speeds greater than 11 mlsec."}]},{"head":"Offshore Ekman Transport","index":56,"paragraphs":[{"index":1,"size":252,"text":"In their climatological study of seasonality and geography of anchovy and sardine reproductive habitats within eastern ocean boundary upwelling systems, Parrish et al. (1983) found a pattern of minimization of both wind-driven offshore surface flow (Ekman transport) and of wind-induced turbulence in the spawning habits of these fishes. They therefore suggest the likelihood of important effects of both processes on reproductive success. Offshore Ekman transport at a given latitude is proportional to the alongshore stress, being simply the product of the alongshore stress and the reciprocal of the local Coriolis parameter. Ekman transport (Ekman 1905) provides an acceptable description of ocean surface transport directly driven by surface wind stress at periods which are long compared to the half-pendulum day; the half-pendulum day is 2.9 actual days in length at 100 latitude but increases to infinity at the equator. Obviously, the Ekman transport description cannot be applied directly at the equator. Here we assume the Ekman transport description to be adequate for the effect of wind stress variations affecting the summary area as a whole on the monthly time scale; thus we simply divide the monthly alongshore wind stress by a characteristic value of the Coriolis parameter (we choose the local value at 10oS, i.e., 0.0000253/sec, to characterize offshore Ekman transport in response to largescale, long period wind variations over the anchoveta reproductive habitat; this choice will affect the average magnitude but not the time series properties of the resulting indicator series, which will be identical to those of the alongshore stress series)."}]},{"head":"Wind Mixing Index","index":57,"paragraphs":[{"index":1,"size":367,"text":"The rate at which the wind imparts mechanical energy to the ocean to produce turbulent mixing of the upper water column is roughly proportional to the third power, or \"cube\", of the wind speed (Elsberry and Garwood 1978). A \"wind mixing index\", which is simply the mean of the cube of the observed wind speeds in each monthly sample (Table 7) is presented as a guide to longer period variability in this particular process. However, it is to be noted that these series may not reflect energetic shorter-term variability which may be more crucial to reproductive success of anchovies (Husby and Nelson 1982). The hypothetical basis for interest in this process in relation to anchoveta reproductive success is Lasker's (1978) suggestion that first-feeding success of anchovy larvae may be dependent upon availability of fine scale food particle concentrations which may be dispersed by wind-driven turbulent mixing events. These occur at atmospheric stonn event scales which are much shorter than one month. Furthermore, it is not the exact magnitude of mixing that is crucial according to this hypothesis, but rather the existence of time-space \"survival windows\" within which the rate of addition of turbulence by the wind does not reach a level that homogenizes the food particle distributions (Bakun and Parrish 1980). The wind speed level at which this occurs and the minimum required duration of the window for substantial survival to result are unclear and undoubtedly are variable functions of other factors such as water column stability, the particular food particle organism's growth rate, behavior, motility, etc. In any case, the maritime reports occur irregularly in time and space and so are not amenable to indicating durations of periods characterized by specific conditions, even if we were able to specify the required nature of the conditions. This would require utilization of a time-and-space continuous meteorological analysis procedure (Bakun 1986) which might be ineffective due to the low maritime report density in the region and particularly seaward of the region. The use of shore station data, despite interference from local topographic influences, etc., might be the best available option for indicating short time scale wind variability over the ocean habitat off Peru (see Mendo et al., this vol.)."}]},{"head":"Solar Radiation","index":58,"paragraphs":[{"index":1,"size":15,"text":"Net incoming solar radiation, Qs, absorbed by the ocean was estimated according to the formula:"},{"index":2,"size":376,"text":"where a is the fraction of incoming radiation reflected from the sea surface, QO is the sum of the direct and diffuse radiation reaching the ground under a cloudless sky, C is the observed total cloud amount in tenths of sky covered and h is the noon solar altitude. For each maritime report, the total daily direct solar radiation reaching the ground under cloudless conditions was extracted from the Smithsonian Meteorological Tables (List 1949) as a function of the date and latitude of the report, using a 4 x 4 element curvilinear interpolation on the table entries via Bessel's central difference formula and assuming the atmospheric transmission coefficient of 0.7 recommended by Seckel and Beaudry (1973). The diffuse solar radiation was estimated according to List's recommendations as follows. The solar radiation reaching the top of the atmosphere was extracted from the appropriate table. This value was decreased by 9% to allow for water vapor absorption and 2% for ozone absorption. The result is subtracted from the value previously determined for the direct radiation reaching the ground to yield the energy scattered out of the solar beam. This is reduced by 50% (to reflect the fact that half is diffused upward and therefore only half is diffused downward) to yield the total diffuse solar radiation reaching the ground. The total daily direct and diffuse radiation values corresponding to each report are then summed to yield Qs. The remainder of the computation follows the procedures adopted by Nelson and Husby (1983). The linear cloud correction in Equation ( 2) is as suggested by Reed (1977), and Reed's recommendation that no correction be made for cloud amounts less than 0.25 of total sky was followed. Sea surface albedo was extracted from Payne's (1972) tables, following Nelson and Husby's (1983) algorithm which consists of entering the tables with the 0.7 atmospheric transmission coefficient reduced by a factor equal to the linear cloud correction applied in Equation ( 2) and the mean daily solar altitude. The possible error in the net radiation estimate introduced by using the mean daily solar altitude to indicate albedo, rather than an integration over the entire day of entries at short time intervals with instantaneous solar altitudes, is estimated to be of the order of 1%."}]},{"head":"Radiative Eeat Loss","index":59,"paragraphs":[{"index":1,"size":101,"text":"Effective back radiation is the difference between the outgoing long-wave radiation from the sea surface, which depends on the 4th power of the absolute temperature of the sea surface, and the incoming long-wave radiation from the sky, which depends on the water vapor content of the atmosphere and on the nature of the cloud cover. Here we follow exactly the computational scheme of Nelson and Husby (1983) who used the modified Brunt equation (Brunt 1932) with the empirical constants of Budyko (1956) and the linear cloud correction formula of Reed (1976) to compute the effective back radiation (radiative heat loss), QB:"},{"index":2,"size":76,"text":"The vapor pressure of the air, ea, was computed according to the formula provided in the Smithsonian Meteorological Tables (List 1949) using the observed barometric pressure, and \"dry bulb\" and \"wet bulb\" air temperatures. For reports that were without an acceptable wet bulb temperature but included an acceptable dew point temperature, the vapor pressure was computed as the saturation vapor pressure at the dew point temperature using an integrated form of the Clausius-Clapeyron equation (Murray 1967)."}]},{"head":"Evaporative and Conductive Heat Losses","index":60,"paragraphs":[{"index":1,"size":386,"text":"In estimating evaporative heat loss (latent heat transfer) and conductive heat loss (sensible heat transfer), the procedures of Nelson and Husby (1983) are again followed closely, except for a modification of the wind speed dependence in their variable transfer coefficient formulations as indicated below. The bulk aerodynamic formula for turbulent fluxes of latent and sensible heat across the air-sea interface in a neutrally stable atmospheric boundary layer (Kraus 1972) can be expressed as where P, and (q1, I are as in Equation (I), with q assigned the same constant value (1.22 kglm3) as in the stress computation. L is the latent heat of vaporization, assigned a constant value of 2.45 x 106 J/kg (585.3 cal/gm). cp is the specific heat of air, assigned a constant value of 1,000 J/kgbC (0.239 cal@X). The empirical exchange coefficients, CE and CH, were assigned constant values of 0.0013 in the construction of the time series presented in Tables 10 and 11. In addition, time series based on variable transfer coefficient formulations incorporating dependencies on atmospheric stability and on wind speed were also assembled for comparison. These formulations are again those chosen by Nelson and Husby (1983) which incorporate the atmospheric stability effect according to a bulk Richardson number parameterization (Deardorff 1968); however, Nelson and Husby's wind speed dependencies were in this case modified according to the recommendations of Large and Pond (1982) who suggest an increase in CE and CH which is proportional to the square root of the wind speed. The specific humidities of the air in contact with the sea surface, qo, and at 10 m or deck level, 910, were computed according to where E is the known ratio (a constant equal to 0.622) of-the molecular weight of water vapor to the net molecular weight of dry air, e is the vapor pressure and P is the barometic pressure. For this calculation the variation in P is negligible and so a constant value of 101,325 pascals (1,013.25 mb) was assigned. The calculation of e at 10 m, or deck level, is as indicated for the radiative heat loss calculation (Equation 3). To calculate e at the sea surface, the saturation vapor pressure over pure water was computed from a formula given by Murray (1967), and reduced by 2% to account for the effect of salinity (Miyake 1952).  "}]},{"head":"The Seasonal Cycles","index":61,"paragraphs":[{"index":1,"size":88,"text":"The idea of regular seasonal cycles for the coupled ocean-atmosphere system off Peru is to some degree illusory in view of the predominant influence of interyear variability in the region. However, the seasonal variation is the most cyclic and predictable of the large components of variability. It is therefore the component of variation which is most likely to be reflected in biological adaptations. Accordingly, a summary of the long-term mean monthly values of the various series (Figs. 2 and 3) serves as a useful starting point for discussion."},{"index":2,"size":68,"text":"Being situated within the tropical band, the region experiences two passages of the sun each year; the sun is directly overhead in October and again in February-March. Also, since the earth's meteorological equator is displaced to the north of the geographical equator, the region is dominated by southern hemisphere atmospheric dynamics; thus austral winter dominates the seasonality of transfers of momentum and mechanical energy from atmosphere to ocean."},{"index":3,"size":107,"text":"The 32-year mean monthly sea surface temperature (Fig. 2A) is at a maximum in March, coinciding with the second period of vertical sun which marks the culmination of the extended austral summer period of relatively high sun. The temperature falls to a minimum in August. The atmospheric pressure (Fig. 2B) tends to be directly out of phase, being at a minimum in the austral summer and at a maximum in the winter season. Cloudiness (Fig. 2C) lags the atmospheric pressure variation by about one month. On average less than 50% of the sky is obscured by clouds in April; this increases to greater than 85% in September."},{"index":4,"size":168,"text":"The strength of the wind exhibits a typical southern hemisphere seasonality, being strongest in austral winter and weakest in summer. Thus the 32-year mean monthly values of the index of rate of addition of turbulent mixing energy to the water column (Fig. 2D) reach a maximum in August-September and a minimum in January. The fact that the seasonal spawning peak of anchoveta is centered within this August-September turbulent mixing maximum would indicate non-adaptation of reproductive strategy for minimization of turbulent mixing effects. This is not in accordance with the general pattern suggested by Parrish et al. (1983) as generally characterizing seasonality and geography of spawning of eastern ocean boundary anchovy populations. Although no claim is made for conclusiveness, the inference would seem to be that Lasker's (1978) hypothesis is not, at least in most years, the major factor affecting anchoveta reproductive success off Peru. Note that the level of turbulent mixing index intensity off Peru is low compared to other anchovy reproductive habitats, even at its seasonal maximum."},{"index":5,"size":102,"text":"The alongshore component of wind stress on the sea surface is consistently equatorward; in no case in the 32-year series (Table 5) did any monthly wind mean stress value deviate from this predominently alongshore and equatorward tendency in the transfer of momentum from atmosphere to ocean. The long-term mean values of alongshore stress (Fig. 2E) follow the same seasonal pattern as the turbulent mixing index, reaching a maximum in September and a minimum in January. The 32-year mean monthly values of the onshore component of stress are small compared to those of the alongshore component, but are positive (onshore-directed) at all seasons."},{"index":6,"size":379,"text":"Surface Ekman transport, being proportional to the alongshore stress but directed perpendicularly to the left of the stress, is thus directed offshore, with a seasonal maximum again corresponding to the seasonal spawning peak of anchoveta. This \"anomaly\" to the pattern of apparent minimization of offshore transport in spawning strategies of engraulids puzzled Bakun and Parrish (1982). However, Parrish et al. (1983) showed that the seasonal variation in mixed layer depth off Peru proceeds in phase with that of transport, in response to the seasonalities in turbulent mixing (Fig. 2C) and surface cooling (Fig. 2A), but has greater relative amplitude. The result is that drifting organisms which are distributed through the upper mixed layer would experience a faster net offshore drift in the thinner surface mixed layer of austral summer than in the deeper mixed layer of winter, even though the winter transport (by volume) is much larger. This is illustrated in Fig. 2F, which shows calculations of mean monthly offshore Ekman velocity of the mixed layer performed in two different ways (based on the 32-year mean monthly values of the data presented in Table 5). Firstly, the monthly estimates of offshore Ekman transport are divided by the composite mean (20 m) of the mixed layer depth values given for 2month segments of the seasonal cycle by Parrish et al. (1983). Secondly, the same monthly estimates of offshore Ekman transport are divided by monthly mixed layer depth estimates produced by curvilinear interpolation of the Zmonth seasonal segments. The effect of variable mixed layer depth on the net offshore velocity is apparent, and suggestive of adaptation of spawning seasonality for avoidance of offshore loss of larvae (for additional discussion of this aspect, see Bakun 1985). The effect of the choice of a constant or variable drag coefficient formulation in the stress computation (Equation 1) on the seasonal signal is indicated in Fig. 3A. The 32-year mean monthly Ekrnan transport values based on the variable coefficient formulation follow a seasonal progression which is very similar to those based on the constant coefficient formulation (i.e., presented in Table 5); however they are slightly smaller in magnitude, reflecting the effect of stability in the atmospheric boundary layer which is stabilized as the onshore-directed airflow is cooled from below while traversing the coastal upwelling zone."},{"index":7,"size":115,"text":"Solar radiation entering the ocean (Qs) is at a maximum during the February overhead passage of the sun (Fig. 3B). This is due to substantially reduced cloud cover relative to the November solar passage. Solar radiation is at a minimum in July, when solar altitude has just passed its June minimum, and cloudiness is approaching its winter maximum. Heat loss from the sea surface via long-wave radiation (QB) is only a small fraction of the short-wave radiation absorbed reflecting the area's location within the tropical band (Fig. 3C). Radiative heat loss is at a seasonal maximum during April, corresponding to the minimum in cloudiness, and at a minimum in September, corresponding to the cloudiness maximum."},{"index":8,"size":69,"text":"Heat loss from the ocean via evaporation at the sea surface (QE) is at a maximum during austral winter and at a minimum during summer (Fig. 3D). The choice of constant or variable transfer coefficient has only a slight effect, with the results of the variable coefficient fortnulation appearing to increase very slightly in magnitude relative to those of the constant coefficient formulation toward the summer and fall seasons."},{"index":9,"size":147,"text":"Heat loss via conduction (Qc) is very small compared to the other heat exchange components (Fig. 3E). This is fortunate because the choice of transfer coefficient formulation completely changes the seasonal pattern. With the constant coefficient formulation, conductive heat loss is mostly negative, indicating heating of the ocean surface by contact with the atmosphere. This reflects the common situation of cool upwelling-affected surface waters being in contact with a generally warmer atmosphere. However, the strong stability of the atmosphere boundary layer inherent in this situation inhibits conductive heat transfer according to the variable transfer coefficient formulation. Thus the less common situation where the air is cooler than the water dominates the sensible heat transfer according to the variable coefficent formulation, with the result that conductive heat loss is indicated as being positive in all the 32year composite monthly means except the summer months of January and February."},{"index":10,"size":141,"text":"The 32-year monthly means of the time series of atmosphere-ocean heat exchange (QN), which represent the resultant differences between the amount of solar radiation absorbed by the ocean and the sum of the heat losses due to long-wave radiation, evaporation and conduction, indicate substantial heat gain by the ocean throughout the year (Fig. 3F). As expected, the average heat ain is greatest in austral summer, reaching values of the order of 200 wattdm2 I (413 cal cm-day-1) in January, and least in winter, falling to about 70 wattslm2 (144 cal cm-2 day-1) in July. The constant coefficient formulations yield slightly greater numerical values of net heat exchange than do the variable coefficient formulations, mainly due to the differences in the respective indications of the conductive heat loss component discussed in the previous paragraph; however the respective seasonal progressions are very similar."}]},{"head":"Interyear Variations","index":62,"paragraphs":[{"index":1,"size":126,"text":"If cyclical seasonal effects are those most likely to be adapted for and incorporated in life cycle strategies of organisms, major nonseasonal variations are those most likely to cause disruptions in life cycle processes and therefore to be reflected in population variations. Very short-scale nonseasonal variations are not well resolved in these monthly composites of irregularly distributed maritime reports. However, when shorter period variability is smoothed and the cyclic seasonal effects are suppressed, nonseasonal variations of longer than annual period, which represent substantial perturbations of the environmental \"normalcy\" to which reproductive strategies or other life cycle strategies should have become tuned, are clearly manifested. For the purposes of this discussion, a simple 12-month running mean filter is chosen to suppress seasonalities and smooth the higher frequencies."},{"index":2,"size":288,"text":"Problems (negative side lobes, wavelength-dependent phase shifts, etc.) with such equallyweighted moving average filters are well known (Anon. 1966). However, in this case the alternatives also present problems. We particularly wish to suppress the seasonal cycle, and so weighting the filter elements to suppress side lobes at other frequencies while increasing leakage of the seasonal frequency, is not desirable. Smoothed monthly series of anomalies from longterm monthly means (e.g., Quinn et al. 1978;McLain et al. 1985) have the property that the filtering is \"nonlocal\", i.e., that any value is dependent on other values in the same calendar month in temporally \"distant\" parts of the time series. Thus, for example, an intense warming (e.g., El NiEo) occurring within a generally cool climatic period appears as a much less intense anomaly than a warming of similar magnitude within a warm period; also, the degree of indicated intensity changes whenever the length of the series used for determination of the longterm mean changes. More importantly, if the amplitude (or shape, phase, etc.) of the seasonal variation is undergoing nonseasonal variation, taking anomalies introduces spurious seasonalscale variations into the filtered series. A \"local\" seasonal filter that avoids some of these problems can be based on 12th-differences, e.g., the result of subtracting from each monthly value the value for the same calendar month in the previous year, but the result is thereby transformed to annual rates of change of a property rather than the property itself, which complicates a descriptive discussion. However, the use of 12th-difference transforms is worth considering for empirical modelling efforts. For the purposes of this discussion, the simple 12month running mean provides a \"local\" seasonal filterlsmoother which will be familiar to many readers and adequate for a descriptive treatment."},{"index":3,"size":92,"text":"The filtered sea surface temperature series (Fig. 4A) illustrates well the major El Niso warm events of the period: 1957-1958,1965,1969,1972-1973,1976 and 1982-1983. Generally elevated temperatures in the period between the 1976 and 1983 events are also apparent. Also apparent is the extended cold period of the mid-1950s; the indication of rise in temperature from this cold period to the peak of the 1957-1983 El Niiio is comparable in total magnitude to that of the rise of the 1982-83 El NiZo from the much warmer climatic base temperature level of the late 1970s."},{"index":4,"size":113,"text":"Major features in the filtered cloud cover series (Fig. 4B) are visibly related to those in the temperature series, but not in any simple, consistent manner. Cloud cover minima often appear to coincide with the relaxation of El NiFio events. An extraordinarily low degree of cloudings appears to have coincided with the return to normal sea temperatures in 1984. Another sharp cloud cover minimum coincided with the leveling off of the temperature decline following the 1957-1958 event. Likewise cloud cover maxima often appear to coincide with rapid drops of temperature into cool periods. Atmospheric pressure variations (Fig. 4C) are obviously highly inversely correlated, at these low frequencies, with those of sea surface temperature."},{"index":5,"size":158,"text":"It is not surprising, in view of the dynamic linkage of wind to horizontal gradient of atmospheric pressure, that wind variations would be related to those of atmospheric pressure. The relation of the \"wind-cubed\" index of rate of addition of turbulent mixing energy to the ocean by the wind (Fig. 4D) to El NiKo periods is striking. El Niiio events are evidently strong wind-mixing events which, according to Lasker's (1978) scenario, would correspond to periods of high probability of starvation for first-feeding anchoveta larvae. The period during and immediately following the 1972 El Nifio appears to have been characterized by an extended period of highly turbulent upper water column conditions. The period during and following the 1982-1983 event appears to have been similarly turbulent, except for a 2-month \"window\" of relaxed turbulent mixing index during December 1983 and January 1984 (somewhat masked by the smoothing in Fig. 4C, but evident in the unsmoothed monthly values in Table 7)."},{"index":6,"size":97,"text":"The magnitude of alongshore (equatorward) wind stress also increases during El NiEo events (Fig. 4E), in agreement with Wyrtki's (1975) conclusions which were based on a summary area displaced somewhat southward along the coast (10-200S, 70-800W) from the one used here (Fig. 1). Thus in addition to potential increases in larval starvation due to increased destruction of food particle strata by turbulent mixing, an increase in potential offshore loss of larvae from the favorable coastal habitat is also indicated. The onshore component of surface wind stress is relatively small and consistently positive (onshore-directed) in the filtered series."},{"index":7,"size":176,"text":"In the previous section, the effect of seasonally-varying mixed layer depth on the offshore Ekman velocity of particles which are continually mixed through the upper mixed layer was discussed (i.e., in reference to Fig. 2F). To investigate the effect on interyear time scales, filtered time series of offshore Ekman velocity were calculated as in that section, i.e., (i) assuming a constant MLD of 20 m and (ii) assuming a seasonally varying MLD derived from the values given by Parrish et al. (1983). The result indicates that, at least for the MLD values chosen, the effect of seasonally-varying mixed layer depth is such as to substantially increase on average the rate of offshore movement of passive particles in the mixed layer. If the effective mixed layer depth is increased during El Nifio, as would be expected both from the effect of the propagating baroclinic wave in deepening the surface layer and also from the enhanced wind induced turbulent mixing, the effect would be to counteract the increased rate of offshore movement indicated from the Ekman transport calculations."},{"index":8,"size":501,"text":"The effect of the choice of constant or variable drag coefficient formulation in the stress computation (Equation 1) is illustrated in Fig. 4G, -where the alongshore stress variation is     T a b l e 7 .  6 5 )  2 6 1 ( 2 9 )  237  5 2 )  3 3 1 ( 4 5 )  2 6 1  5 6 )  2 1 4 ( 26)  228  2 3 )  3 0 3 ( 7 6 )  164  2 5 )  2 0 1 ( 1 9 )  198  1 7 )  2 1 1 ( 2 0 )  199  2 3 )  2 6 7 ( 32)  199  2 5 )  3 2 0 ( 31)  376  8 9 )  3 7 2 ( 58)  242  4 5 )  4 4 2 ( 52)  319  3 9 )  2 4 5 ( 1 8 )  243  4 1 )  3 5 7 ( 5 9 )  369  56)  3 2 6 ( 6 2 )  243  52)  2 7 2 ( 2 9 )  154  2 8 )  1 8 6 ( 2 2 )  2 5 1  7 1 )  2 5 8 ( 6 6 )  302  6 7 )  8 0 5 ( 1 3 9 )  220  8 1 )  6 7 5 ( 7 4 )  348  5 7 )  3 5 2 ( 4 5 )  295  5 8 )  3 1 6 ( 3   9 2 ( 1 6 )  152  2 3 )  1 0 8 ( 1 5 )  82  35)  9 9 ( 1 2 )  144  2 4 )  1 3 6 ( 1 2 )  100  2 0 )  2 1 2 ( 1 9 )  96  7 2 )  2 2 2 ( 33)  1 3 1  2 2 )  1 3 2 ( 2 0 )  164  4 4 )  2 4 0 ( 2 1 )  193  2 8 )  9 5 ( 1 2 )  8 3  5 7 )  1 8 0 ( 3 0 )  280  1 9 )  1 9 2 ( 1 6 )  283  2 6 )  1 0 4 ( 1 3 )  1 5 1  50)  1 9 4 ( 2 4 )  139  4 9 )  1 6 9 ( 2 3 )  199  2 7 )  1 5 6 ( 2 5 )  407  6 6 )  2 3 7 ( 2 8 )  179  4 2 )  2 7 4 ( 2 8 )  195  3 5 )  1 4 3 ( 2    uncommon in the figure with the notable exception of the early 1970s and again in the early 1980s. The main point is that features in Fig. 4L seem unrelated to any major features apparent in the other series plotted in Fig. 4. Thus, the major effect of uneven distribution of reports appears to be in increasing sampling variance rather than in introducing long-term nonhomogeneity in the various time series."}]},{"head":"\"Wind cubedI9 i n d e x of rate of a d d i t i o n t o t h e water column, by t h e wind, of t u r b u l e n t m i x i n g e n e r g y . The s t a n d a r d errors o f t h e mean appear i n p a r e n t h e s e s t o t h e r i g h t of e a c h mean","index":63,"paragraphs":[]},{"head":"Discussion","index":64,"paragraphs":[{"index":1,"size":142,"text":"Since the 12-month running mean filter used to highlight the long-term variations in Fig. 4 incorporates no data more than 6 months previous or following, the indicated multiyear features are certainly real, and in no way represent artifacts of filtering and smoothing procedures. Moreover, the interyear variations of annual mean values tend to be of similar magnitude to the cyclic seasonal components (Figs. 2 and 3), making the separation of these scales in analysis of effects (e.g., on the biota) a difficult problem. For example, the impact of long-term variability will generally depend on phase relationships with the seasonal variation. Additional complications involve the adaptations of the biota, not only the long-term evolutionary adaptations of life cycle processes to regular cyclic effects, but also lagged responses of community composition, etc., to events of the recent past (see Mendelsohn and Mendo, this vol.)."},{"index":2,"size":191,"text":"This area of the world's ocean may be uniquely troublesome in these respects. Because the Pacific Ocean is so large, it is much less subject to continental effects which amplify seasonalities due to the low heat storage capacities of continents relative to oceans. The apparent consequence is much less forcing of Pacific Ocean processes to follow a regular seasonal cycle than may be the case, for example, in the Atlantic; the result is the dominance of interyear variation in the Pacific (Picaut 1985). The location so near to the eastern terminus of the equatorial wave guide results in a focusing of variability initiated in various portions of the great Pacific ocean-atmosphere coupled system to particularly impact the ocean habitat off Peru. Indeed, this may constitute part of the explanation for the enormous biomasses of pelagic fishes that have inhabited the region; i.e., because of the intense irregular environmental variability, a single, rather unspecialized fish species with very rapid population responses may be able to dominate the system relatively free from predation and competition from more specialized, less responsive species that would be more subject to the inefficiencies of multiple food-chain steps."},{"index":3,"size":137,"text":"We have noted that the interyear variations tend to involve groups of years. In fact single features, e.g., the 1982-1983El Niiio, the 1954-1955 cold period, etc., so dominate the series that the entire series length becomes a dominant scale of variability. The result is that any sort of assumption of stationarity must be somewhat unrealistic; the real degrees of freedo'm useful for empirical analysis nearly vanish with respect to such features. It is also apparent from even casual inspection of Fig. 4 that the index series presented in this paper are all highly interrelated in terms of major interyear features, further exacerbating the problems of empirically sorting the various effects. Any available mechanistic constraints, provided by established physical or biological principles, that can be imposed on empirical analysis, would of course be very helpful in this situation."}]},{"head":"Variable vs. Constant Transfer Coefficient Formulations","index":65,"paragraphs":[{"index":1,"size":131,"text":"Some differences in the results of variable and constant transfer coefficient formulations in the wind stress estimates (Equation 1) and in the estimates of certain of the heat exchange components (Equations 4,5 and 6) have been noted in the discussions of both seasonal and interyear variability. The differences have mainly been in magnitude of the particular index, with temporal aspects of the variability appearing to be relatively unaffected. An exception was the conductive component of heat exchange, Qc, where the effect of stability in the atmospheric boundary layer introduced major discrepancies between the alternate formulations (Fig. 3E). Fortunately, QC is by far the smallest heat exchange component. Some summary information concerning gross effects of the differences in the various series with respect to time series properties is indicated in Fig. 5. "}]},{"head":"v A 12-month running means slope of regression","index":66,"paragraphs":[]},{"head":"Y","index":67,"paragraphs":[{"index":1,"size":156,"text":"In the case of alongshore component of wind stress, the raw monthly series computed according to the two types of drag coefficient formulation are well correlated, each accounting for more than 95% of the variance in the other. When 12th-differencing was employed to remove the cyclic seasonal variations, the correlation dropped only slightly; this lower than expected drop in r2 must be due to a strong seasonality in the differences resulting from the two formulations. Note that both the raw and 12th-differenced series pairs were much more highly correlated than the pair of 12-month running mean filtered series (which can be viewed in terms of directly proportional offshore Ekrnan transport in Fig. 4G). The slope of the regression of the variable coefficient alongshore stress series on the constant coefficient series is nearly one to one in the raw series; it drops to below 0.9 in the 12-differenced series, but is above 0.95 in the filtered series."},{"index":2,"size":167,"text":"Slightly lower degrees of relationship are seen for the onshore component which tends to be much the smaller of the two stress components. The respective formulations of evaporative heat loss (QE) were very highly correlated in raw and 12th-differenced series. The degree of relationship fell only slightly after the 12-month running mean was applied. In the case of the conductive heat loss term (Qc), the two formulations gave substantially greater differences, particularly after 12th-differencing or filtering. In the case of net heat exchange (QN) the differences between the results of the two formulations appear not to be appreciable. Note that in all cases the raw and 12-differenced series were more highly correlated between the constant and variable coefficient formulations than were the corresponding 12-month running mean filtered versions of the respective series; thus the degree of relationship is even higher in the case of the unsmoothed series, even after the seasonality is removed, than can be seen in the comparative examples of filtered series in Fig. 4."},{"index":3,"size":97,"text":"In view of the time series similarities, the constant transfer coefficient versions of the respective index series are the only ones presented herein in tabular form (Tables 5,6, 10 and  11). In view of the larger relative effect of the uncertainties as to proper formulation of the transfer coefficient for conductive heat loss, and because its very small magnitude makes it relatively unimportant in any case, no tabular series of QC is included. Of course, it would be possible to assemble the constant coefficient version of the QC series from the values in Tables 8  to 11."}]},{"head":"Offshore Ekman Velocity of the Mixed Layer","index":68,"paragraphs":[{"index":1,"size":133,"text":"In discussing the inference of Parrish et al. (1983) that the offshore Ekman transport should ideally be divided by the effective mixed layer depth, to yield the net rate of offshore transport of drifting larvae which are passively mixed through the layer, Bakun (1985) stressed the importance of the qualification \"ideally\". Ekman transport is estimated from relatively abundant surface wind reports, which reflect the fairly large spatial scales of atmospheric variation. Mixed layer depth may vary on much shorter oceanic length scales~and must be determined from generally much less abundant subsurface observations. In cases where the estimate of effective mixed layer depth may be very imprecise, the derived estimate of offshore Ekman velocity of the mixed layer could constitute a less reliable indicator of variability in this process than the Ekman transport alone."},{"index":2,"size":155,"text":"On long time scales, mixed layer depth and wind stress observations are likely to be substantially correlated. However, within any given month it is probably not too bad an assumption to regard observations used to estimate these quantities (surface wind and subsurface temperature structure) as largely independent samples of the respective monthly distributions, particularly since there will normally be many more surface than subsurface reports. In this case, combining the standard errors according to the rules for a quotient of independently observed quantities (e.g., Beers 1953) should provide a reasonable gauge of precision. Thus the ratio of the standard error to the monthly mean derived net offshore Ekman velocity of the mixed layer might be reasonably estimated as being equivalent to the square root of the sum of the squares of the respective ratios of the standard errors to the monthly mean values of the Ekman transport and mixed layer depth components of the calculation. "}]},{"head":"Monthly Turbulence and Ekman","index":69,"paragraphs":[]},{"head":"Introduction","index":70,"paragraphs":[{"index":1,"size":56,"text":"An important feature of the upwelling region off the Peruvian coast is the variability of both its physical characteristics and its living resources. This region supports large fish populations, such as the anchoveta (Engraulis ringens), upon which one of the largest fishery resources in the world was based (Zuta and Guillen 1970;Guillen and Calienes 1981;Cushing 1982)."},{"index":2,"size":48,"text":"The causes of coastal upwelling can be several, some of which are: transport caused by wind, currents, internal waves, mixing, etc. Of these, wind-caused transport is, off Peru, the most important factor, due to its persistence and direction both of which are favorable for upwelling (Smith 1968;Schaffer 1982)."},{"index":3,"size":55,"text":"The physical processes that regulate coastal circulation, important for the upwelling, vary both in time and space (Calienes et al. 1985) and can greatly affect the fishery resources. This applies especially to reproductive processes, transport and survivorship of larvae and, thus, to recruitment of such resources (Bakun 1973(Bakun ,1975(Bakun ,1985;;Lasker 1978Lasker ,1981;;Bakun et al. 1982)."},{"index":4,"size":134,"text":"Currently, recruitment is that aspect of fish population dynamics that is most difficult to model and predict. Nevertheless, powerful hypotheses exist that try to explain the mechanisms that regulate recruitment. One of them, proposed by Hjort (1914), suggests that the strength of a cohort of fish is established by mortality during the larval phase, as regulated by the food available when the larvae begin feeding. Another hypothesis, proposed by Hunter (1976), suggests that predation of larvae by vertebrates and invertebrates regulates recruitment. Lasker and Smith (1977), however, concluded that larvae survivorship is related more to the coincidence than to the abundance of food and predators. Lasker (1978Lasker ( , 1981)), finally, suggested that larval mortality is caused mainly by wind-generated turbulence, which leads to dispersion of particles necessary for first feeding of the larvae."},{"index":5,"size":85,"text":"Several contributions have been published pinpointing some biological-environmental relationships in anchovies, such as the ones presented by Parrish and MacCall(1978), Bakun and Parrish (1980), Csirke (1980), Bailey (1981), Collins and MacCall(1977) and others, most of them referring to the current systems off California. However, the possibility exists that the environmental processes which regulate recruitment and population fluctuations of a group of species, are similar in all four Eastern Boundary Currents regions, i.e., California, Peru, Canaries and Benguela (Bakun and Parrish 1980;Parrish et al. 1981;Bakun 1985)."},{"index":6,"size":52,"text":"The present study presents the monthly averages of the upwelling and turbulence indices calculated on the basis of wind data recorded at the Trujillo (08006's) and Callao (12000's) airports (Fig. 1), as a contribution to the identification of possible relationships between the environment and recruitment of fish stock in the Peruvian Current."}]},{"head":"Material and Methods","index":71,"paragraphs":[]},{"head":"Calculation of Upwelling and Turbulence Indexes","index":72,"paragraphs":[{"index":1,"size":111,"text":"An upwelling index is essentially a quantitative expression of water mass transport with reference to the coastline. The transport generated by superficial wind is based on Ekman's theory (Ekman 1905). Under the assumption of a stable uniform movement and an infinite homogeneous ocean, the transport of masses by area unit of ocean surface is directed 900 to the left (in the southern hemisphere) of the direction towards which the wind is blowing. Hence, wind blowing parallel to the coastline will generate a net water transport perpendicular to it and, thus, the upwelling index will be greater than that of a wind blowing with equal intensity but not parallel to the coastline."},{"index":2,"size":24,"text":"Ekrnan's transport is calculated according to Bakun (1973) using the wind-created stress (7) on the surface and the Coriolis parameter (f) of the area:"},{"index":3,"size":30,"text":"The wind stress is given by: where 'pa9' is the air density (= 1.2 kg m-3) CD is the empirical drag coefficient (= 0.0013) and V is the wind velocity."},{"index":4,"size":90,"text":"The direction of the stress is the direction from which the wind is blowing. The Coriolis parameter is dependent of the latitude of the area and is calculated by: where n is the earth's rotation angular velocity (=7.2921 x 10-5 rad sec-1) and Q, is the latitude in degrees (see Bakun, this vol.) The upwelling index is calculated by: where a is the angle representing the difference between the coastline angle and the direction of the wind. The coastline angles used were 1500 for Trujillo and 1450 for Call30 station."},{"index":5,"size":28,"text":"The turbulence index in the upper water column, following Elsberry and Garwood (1978), is here defined as equal to the third power of the wind velocity (\"wind cubed\")."}]},{"head":"Source and Analysis of Data","index":73,"paragraphs":[{"index":1,"size":42,"text":"A statistical analysis of the monthly average upwelling indices calculated on the basis of 24,12,8,6,3 and 1 (prevailing average) observations per day, taken during one year (1970) at Trujillo and Callao stations was performed and the results are shown on Table 1."},{"index":2,"size":106,"text":"All but one of the chi-squared values were lower than the critical value, the only exception being 1 observation per day (i.e., \"daily prevailing wind\") at Trujillo. Moreover, as we can see in Fig. 2, the trends of the daily means based on different numbers of observations per day are similar to those obtained using 24 observations per day. We performed the analysis of our data by computing the indices with 6 to 8 observations per day as suggested by A. Bakun (pers. comrn.). The data were entered into an HP-100 computer and processed using a routine based on the program of Navaluna et al. (1984). "},{"index":3,"size":66,"text":"The calculation of the daily upwelling and turbulence indices was obtained by computing these indices for each observation (i.e., for each period of 3-4 hours) with subsequent averaging. The monthly averages were obtained by taking the mean of the daily indices. This same procedure was used by Bakun (1973Bakun ( ,1975) ) to calculate daily, weekly, and monthly upwelling indices for the North American westerb coast."},{"index":4,"size":28,"text":"Positive values indicate upwelling and negative values indicate downwelling. A data volume was prepared (Mendo et al. 1987) which presents, on a daily basis, the data used here.  "}]},{"head":"Inshore-Offshore Diflerences","index":74,"paragraphs":[{"index":1,"size":150,"text":"The difference between the winds recorded at the airports and those from off the coast is one of the key points in this study. Enfield (198 la,198 lb) suggests that the differences between the seasonality of the easterlies, with a maximum in winter-spring and a minimum in summer (see Wyrt. and Meyers 1976;Barnett 1977), and the wind record at the Chimbote, Callao and Tacna airports (see Fig. 1) are due to land-sea thermal variations. Nevertheless, the Talara and San Juan winds, he points out, do have seasonal trends similar to those of the easterlies. Brink et al. (1980) also found a reasonable match between winds measured at the airports and those measured over the shelf near 150s. Based on these results, Schaffer (1982) used wind data recorded at the Trujillo, Chimbote and Callao airports to calculate mean stress vectors and to correlate these with the currents 50 km off Chimbote."},{"index":2,"size":112,"text":"During a cruise to the north and central part of the Peruvian coast in April and May 1986, wind data (direction and velocity) were recorded every half hour until approximately 113 km off the coast. Fig. 1 shows the vectors of the wind records. As may be seen, there is a considerable uniformity in the winds off Trujillo, Chimbote and Callao, at least up to 48 km off the coast. Likewise, the airport records for the abovementioned locdities resemble this neighboring seabased wind vectors. This suggests that the winds recorded at the airports, up to a not yet determined distance from the coast (possibly 32 to 48 km), tend to be similar."},{"index":3,"size":58,"text":"Considering that the coastal strip within 50 km is the area where upwelling occurs (Zuta and Guillen 1970;Guillen and Calienes 1981) and where the most important activities (e.g., spawning) of the different resources species take place, it appears that wind records from the two selected airports can be used for the calculation of upwelling and coastal turbulence indexes."}]},{"head":"Monthly Variation of the Upwellirig and Turbulence Indexes","index":75,"paragraphs":[{"index":1,"size":81,"text":"Tables 2,3,4 and 5 and Fig. 3 and 4 show that the monthly variations of the upwelling and turbulence indexes have similar tendencies at each of the two stations considered here . In Trujillo, minimum values occur more frequently in winter (July-August) and summer (January-February-March) months. Maximum values tend to occur, on the other hand, at the end of winter (September) and in spring (October-November-December). In Trujillo, minimum and maximum monthly values tend to vary from one year to the other."},{"index":2,"size":46,"text":"In Callao, the minimum and maximum values of upwelling and turbulence indices have a more regular seasonality than in Trujillo. The minimum values occur with a 70% incidence in Year the fall (May-June) months, and the maximum values in the spring (November-December) and summer (January-February) months."},{"index":3,"size":110,"text":"The monthly patterns obtained by averaging the monthly values of every year analyzed differ between both stations (Fig. 5). In Trujillo low values occur in summer and winter while in CaUao, low values are observed in fall. A greater seasonality in the indices can be observed for Callao than for Trujillo.  The marked irregularity of seasonal patterns, for the period covered by this study, suggests that the average monthly values and seasonal cycles presented here for both Trujillo and Callao should be viewed with caution. Likewise, the large differences found between both stations make it difficult to define a monthly pattern, say, for all of the northern zone of Peru."},{"index":4,"size":75,"text":"On the other hand, a preliminary calculation of the upwelling indices for Chimbote (90s) based on daily mean prevailing winds gave a monthly phttern similar to that of Trujillo, with the exception of a slightly more pronounced seasonality (Fig. 6). This allows us to assume that the wind-generated transport and turbulence could remain similar over coastal stretches of 20 and more. The extent of such coherence would depend, obviously, on interferences caused by coastal topography."}]},{"head":"Annual Variations of the Upwelling and Turbulence Indexes","index":76,"paragraphs":[{"index":1,"size":104,"text":"Tables 2,3,4 and 5 and Fig. 7 present the interannual transport and turbulence variations for the 1953-1985 period. Trujillo has higher values for both indices than does Callao, due mainly to a greater intensity of the winds. The interannual trends of both stations are almost opposite: while Trujillo experiences a rather regular decrease, Callao shows an increase in the values of both indices through the 33 years analyzed. Trujillo has its higher values during 1957, 1968 and 1983, and the lowest in 1953,1974, 1981 and 1982. On the other hand, in Callao, an increase can be observed for the years 1957,1966,1969,1972,1976-1977 and 1983, all  "}]},{"head":"Upwelling Index and the El NiGo Phenomenon","index":77,"paragraphs":[{"index":1,"size":161,"text":"The years characterized by the occurrence of El Nizo events show positive and negative anomalies in Trujillo (Fig. 8) i.e., positive anomalies in 1955-1977, 1967, 1969 and 1976, and negative anomalies in 1981-1985. On the other hand, in Callao only the years with positive anomalies correspond to El Ni6o events (e.g., 1957,1969,1972-1973,1976-1977 and 1982-1983). In addition, a positive anomaly occurred during 1984-1985. Parrish et al. (1983), after comparing the four eastern boundary systems, concluded that spawning rarely occurs in areas of strong turbulent mixing of the upper water column. Indeed, spawning grounds are characterized by weak to moderate values of the turbulent index. spawning peak of the Peruvian anchovy occurs during the austral winter when offshore Ekman transport is strong. Likewise, we know that the offshore transport off Chimbote is stronger than in the California Current System (Bakun 1985). This fact suggests that spawnkg succdss of the Peruvian anchoveta off Chimbote and Trujillo should be influenced more by transport than turbuleace."}]},{"head":"Ekman Transport and Turbulence Indexes Related to Spawning Seasons in Eastern Boundary Current Systems","index":78,"paragraphs":[]},{"head":"Introduction","index":79,"paragraphs":[{"index":1,"size":193,"text":"The anchoveta (Engraulis ringens) which, from the mid-1960s to the early 1970s formed the basis of the largest single-species fishery in the world, has been the focus of numerous studies both on its biology and population dynamics and on the environmental factors that affect these (Clark 1954;Jordan 1959;Boerema et al. 1965;Schaefer 1967; see also contributions in UNESCO 1980; Glantz and Thompson 1981 and references in Pauly and Tsukayarna, this vol.). The fishery has a well documented history dating back to the start of the fish meal and oil industries,which use anchoveta as raw materia1,in the 1950s (Freyre 1967),resulting in increased demand for anchoveta and a corresponding increase in the number of commercial vessels (Doucet and Einarsson 1967 and Aguero, this vol.). The rapid growth of the fishery led to various government interventions, notably the creation of fishery research institutions, i.e., the Consejo de Investigaciones Hidrobiologicas in 1954, the Instituto de Investigacion de 10s Recursos Marinos in 1960 and the Instituto del Mar del Peru (IMARPE) in 1964, whose aim was to study the fishery and to propose development and management schemes for the fishery (Schaefer 1967 and see Castillo and Mendo, this vol.)."},{"index":2,"size":187,"text":"In the early 1970s, IMARPE, based on a large volume of catch, length-frequency, environmental, economic and ancillary data, organized a series of meetings that were aimed at assessing the status of the anchoveta stock (IMARPE l970,1973(IMARPE l970, ,1974a(IMARPE l970, , 1974b)). In 1974, following the 1972 collapse of the fishery, a research scheme was proposed at an international workshop for the study of the effects of fishing on the anchoveta stock and the contributing effects of environmental factors, such as the \"El Ni5o\" phenomenon and predation pressure (see Anon. 1975 and contributions in UNESCO 1980). Pauly and Tsukayarna (1983) presented a preliminary analysis of a time series of catch-atlength data covering, on a monthly basis, the years 1961 to 1982 for the northern anchoveta stock. Their analysis brought to the fore the importance of a long time series to identify and quantify the factors which affect growth, mortality, recruitment, and thus, biomass and catches. Continuous and consistent series of catch and catch composition data are also needed to demonstrate monthly variability of stock sizes indicative of seasonal fluctuations in the physical environment and changes in biological processess."},{"index":3,"size":30,"text":"This paper presents a number of time series needed for these purposes. These have been derived in a two-phase process also used in some other contributions included in this volume:"},{"index":4,"size":19,"text":"a) collection and initial data st_andardization at IMARPE, and b) final data standardization and filling of gaps at ICLARM."}]},{"head":"Material and Methods","index":80,"paragraphs":[]},{"head":"The Catch Data","index":81,"paragraphs":[{"index":1,"size":91,"text":"Three data sources were used to construct the time series of monthly catch covering the years 1953 to 1982: i) for the years 1953 to 1958, we used annual catches in Table 2 of Murphy (1972), put on a monthly basis using monthly weighting factors based on the monthly catch data in Fig. 1 of Doucet and Einarsson (1967), which cover the years 1959 to 1964 (1963 and 1984 were not used due to the occurrence of a strike in January 1963 and to noncoverage of later months in 1964, respectively)."},{"index":2,"size":72,"text":"ii) for the period from January 1959 to February 1961 we used the catch data in Fig. 1 of Doucet and Einarsson (1967), which pertains to the whole of the Peruvian coast. These data were therefore adjusted to the narrower stretch from 4 to 140s using a factor of 96%, computed from the data in Fig. 2 in Doucet and Einarsson (1967); this factor was also applied to the data in (i)."},{"index":3,"size":34,"text":"iii) the catch data covering the period from March 1961 to December 1982 were collected by staff of the Instituto del Mar del Peru and its predecessor, the Instituto del Investigaciones de Recursos Marinos."},{"index":4,"size":75,"text":"The time series of catch presented here is uninterrupted in the sense that a catch figure is given for every month during which fishing activities occurred. Months with zero, or near zero catches refer either to a strike (huelga) or a closure of the fishery (veda), a measure which became increasingly necessary during the 1970s (the very low catch values during veda periods refer mainly to the small catches made by the survey ships themselves)."},{"index":5,"size":33,"text":"It must be realized that the catch figures presented here (see Tables 1 to 50) represent nominal catches, i.e., figures which may considerably underestimate the actual catches (see Castillo and Mendo, this vol.)."}]},{"head":"The Length-Frequency Data","index":82,"paragraphs":[{"index":1,"size":93,"text":"The length-frequency data presented in 1 cm classes in Tables 1 to 30 stem from four different sources: i) from the archives of the Instituto del Investigaciones de Recursos Marinos, and referring to standard length (SL) samples collected at Callao, Casma, Chimbote, Don Martin, Huacho and Samanco. These samples, representing individual fish measured to the nearest mm, were converted to total length (TL) using a regression established by Clark (1954), i.e., where L is expressed in mm. These data refer to the period ranging from October 1953 to February 1961 (see Tables 1-9)."},{"index":2,"size":141,"text":"ii) Jordan (1959, Figs. 5a-5d) presented detailed size-frequency data representing fishery catches from the Huacho area and anchoveta consumed by cormorants (the most important guano bird of Peru, see Tovar et al., this vol.), as reconstructed from regurgitated otoliths collected on Don Martin Island (see Tovar et al., this vol. for location). Jordan (1959) also showed that the size distribution of these two groups of samples largely overlap (see also Muck and Pauly, this vol.). These samples, originally presented as standard length in graphic forrns, were read off, coverted to total length and regrouped in 1 cm classes as was done in (i). They cover the period from June 1954 to June 1958; whenever they were taken in the same month as a sample from (i), an average sample was constructed for the month in question (see Tables 2 to 6)."},{"index":3,"size":105,"text":"iii) the bulk of the length-frequency data presented here for the period' from March 1961 to December 1982 have been collected by IMARPE staff generally following the sampling procedure described in Saetersdal and Valdivia (1964). These authors also presented data suggesting that for the stretch of coastline covering the northern and central part of Peru, within port variability of length-frequency samples was less than variability due to different sampling periods. Monthly samples representative of the stock as a whole were thus obtained by pooling, within each month, daily samples representing about 30% of the landing and most of the fishing areas covered by the fleet."},{"index":4,"size":117,"text":"Generally, one single sample was taken from each vessel sampled. The sample consisted of the content of a two-liter container, of which all anchoveta were measured and weighted. The data were originally in 0.5 cm classes, defined such that, upon regrouping into 1 cm classes, class medians (or \"midlengths\") of 4.25,5.25, ... 20.25 cm emerged. This is the reason why the data in (i) and (ii) were also regrouped within the somehow unconventional class limits that the above midlengths imply. iv) to complement the data in (iii) obtained from fishery catches, length-frequency data were obtained, from 1972 onward during scientific surveys such as the EUREKA surveys, conducted during closures of the fishery (Villanueva 1975  and Tables 20-27)."},{"index":5,"size":80,"text":"The length-frequency data obtained from (i) to (iv) were then expressed as percentages of the total monthly frequencies. These percent frequencies as presented in Tables 1-30 may, in some case, not add up exactly to 100% because of rounding off errors and cases where the original sample included fish smaller than 4.25 cm, excluded from the tables presented here. It should also be noted that on all these tables, dashes represent zero frequencies and \"0.000\" represents frequencies less than 0.0005%."},{"index":6,"size":91,"text":"Overall, the length-frequency data presented here cover 270 of 360 possible months, i.e., the time series is 75% complete. Standardization to 1 cm length classes and a single definition of length (here total length (TL)) led to certain features such as the consistency of the time of recruitment of young fish, the shift of modes reflective of growth phenomena and a trend toward larger sizes, from 1953 to 1982 to become immediately visible (see Tables 1 to 30). These themes are investigated in greater detail in Palomares et al. (this vol.)."}]},{"head":"The Condition Factors","index":83,"paragraphs":[{"index":1,"size":70,"text":"\"Condition factors\" (c.f.) are here defined as the multiplicative term in a length-weight relationship of the form where weight (W) is expressed in g live weight and length (L) as total length in cm. The exponent of 3 implies isometry which is assumed here for the sake of having all temporal changes in the length-weight relationship of anchoveta \"concentrated, as it were, in their monthly c.f. values (see Pauly 1984)."},{"index":2,"size":276,"text":"In earlier analyses of the length-frequency data from March 1961 to December 1982 presented in Tables 9-30, the weights of samples of fish that had been measured by IMARPE field staff had also been used to raise these samples to the total catch. Because these sample weights were not available to directly estimate c.f. values at the time this contribution was written, we have proceeded \"backward\", i.e., used the catch-at-length estimates to obtain approximations of the underlying condition factors. This was done on the basis of a lengthweight relationship of the form with (a 100) arbitrarily set equal to 1 (one). The \"pseudo-weight\" of the fish of a given class (j) was then estimated as where Lj2 is the upper limit of length class (j) and L-1 its lower limit (e.g., for say a midlength of t 10.25, Lj 1 = 9.75 and Li2 = 10.75 cm). The values o PWj were then multiplied, for each month separately, by the available catch-at-length data. This resulted in pseudo total weight of the monthly catch. Finally, the ratio of the real catch to the pseudo monthly catch was computed, corresponding to the ratio between real c.f. values and the values of a 100. The c.f. values so estimated are given in Tables 9-30. They were also plotted by Palomares et al. (this vol.) against the mean sea surface temperature (Ti, see Table 2 in Pauly and Tsukayaxna, this vol.) of the corresponding month (i). This resulted in which was used for all years (i.e., 1953 to 1960, see Tables 1-8) and months for which \"real\" c.f. could not be estimated from catch-at-length data using the method described above."},{"index":3,"size":20,"text":"Tables 1-30 present the monthly catch of Peruvian anchoveta (Engraulis ringens), northerrdcentral stock (4-140S), from January 1953 to December 1982."},{"index":4,"size":106,"text":"As might be seen from Fig. 1, this catch fluctuated enormously from month to month, this phenomenon being strengthened since 1972 by seasonal closures of the fishery. Palomares et al. (this vol.) present an analysis of the time series of catch composition data and condition factors compiled in Tables 1-30, while Castillo and Mendo (this vol.) discuss possible sources of bias in the nominal catch presented here. We leave it thus to the reader to consult these authors for a detailed discussion of the data in Tables 1-30.  Murphy (1972) and mean seasonaliiy of catch adapted from Fig% 1 and 2 in Doucet and Einarsson (1967)."},{"index":5,"size":80,"text":"C€ondition factors in brackets estimated from the linear relationship between temperature and condition factors m Palomares et aL (this vol.).   'Frequencies per length class adapted from Jordan (1959),~eferring either to fish eaten by birdsor fishery catch. ' ~e a n s of (a) and @). d ~o n t h l y catch in 1,080 t adapted from annual catch data in Murphy (1972) and mean seasonality of catch adapted from Figs. 1 and 2 in Doucet and Einarsson (1967)."},{"index":6,"size":95,"text":"e~ondition factors in brackets estimated from the linear relationship between temperature and condition factor in Palomares et al. (this vol.). b ~e a n s of (a) and of data collected by staff of the Instituto de Lnvestigacion de 10s Recursos Marinos. '~onthly catch in 1,000 t adapted from annual catch data in Murphy (1972) and mean seasonality of catch adapted from Figs.   Jordan (1959). 'Monthly catch in 1,000 t adapted from annual catch data in Murphy (1972) and mean seasonality of catch adapted from Figs. 1 and 2 in Doucet and E' sson (1967)."},{"index":7,"size":78,"text":"T20ndXion factors in brackets estimated from the linear relationship between temperature and condition kctor in Palomares et al. (this VOL). 'condition factors in brackets estimated from the linear relationship between temperature and condition factor in Palomares et aL (thisvol.).  'condition factors in brackets estimated fmm the linear relationship between temperature and condition factor in Palornares et at (this vol.).   d~ondition factors in brackets estimated from the linear relationship between temperature and condition factor in Palomares et aL (thisvol.). "}]},{"head":"0.643","index":84,"paragraphs":[{"index":1,"size":22,"text":"a~ercent catch-composition data obtained from EUREKA surveys (January, September and November), \"~rospeccfon Pesquera\" (June and July) and \"Exploration y Prospeccion Pesquera\" (August)."},{"index":2,"size":78,"text":"b ~o n t h s with closure of fishery (veda); catches mainly from surveys (August, September and October). C~ondition factors in brackets estimated from the linear relationship between temperature and condition factor in Palomares et al. (this voL). d''~eal\" value was 0.927, which is far too high to be correct and was therefore replaced wlth an interpolated value representing the mean of itself, the preceding and following values, and the values for April 1972 and April 1974.        "}]},{"head":"Introduction","index":85,"paragraphs":[{"index":1,"size":72,"text":"The Peruvian coast is characterized by a high primary production and by high fishery catches and indeed, during the 1960s and early 1970s, the Peruvian upwelling system supported the largest fishery in the world. At the peak of the fishery landings, anchoveta (Engraulis ringens) contributed one-sixth of all fish caught in the world and in 1970 Peru became, as far as bulk catches are concerned, the leading fishing nation in the world."},{"index":2,"size":41,"text":"Before, during and after these peak landings, numerous irregularities occurred which contributed to the catches and landings being underestimated (Fig. 1). IMARPE (1970) reported on this as follows: \"There are [...I several ways in which those reported statistics may be misleading."},{"index":3,"size":42,"text":"The reported landings are less than the true catches for several reasons, e.g.: a) losses at sea, including dumping of excess catch b) losses at unloading c) underreporting of actual quantities landed (especially during the peladilla season when meal yield is low)."},{"index":4,"size":170,"text":"Information presented suggested that occasionally these losses could be large, perhaps 40% of the reported landings. There are no data on how this proportion has changed during the history of the fishery, although [there are indications] that it has probably been rather constant, at least in  If the ratio of actual to reported catch has remained constant the analysis of sustainable yield, etc., will be little affected, except that all figures of catch, etc. are underestimated by a constant factor. The estimates of maximum sustained yield and permissible quota, etc., will then be correct guides for management provided that the future catches are also underestimated by the same amount. However, it is most desirable that records could be made of the true removals from the stocks by the fishery.\" Vasquez and Lam (1977) attempted to estimate the percentage of anchoveta caught that was not reported in official catch statistics and obtained a value of 12%. They also mentioned that \"some people\" believed a figure of 20-30% to be more realistic."},{"index":5,"size":29,"text":"The objective of this work is to provide a reassessment of these figures, and thus to contribute to the estimation of actual catches of anchoveta, for the period 1953-1982."}]},{"head":"Brief Historical Review of Anchoveta Fishery Management","index":86,"paragraphs":[{"index":1,"size":69,"text":"Although the anchoveta fishery began in the early 1950s, the anchoveta \"boom\" began in the mid-1960s. In 1968, the number of boats reached a total of 1,500, with an estimated fishing capacity of 26 million tonnes per annum (see Aguero, this vol.). These metal boats were more effective than their smaller wooden predecessor because of the use of new, superior purse seines, hydraulic and/or mechanized winches, power blocks, etc."},{"index":2,"size":39,"text":"The rapid growth of the fishery led, in the 1960s to the first management interventions by the Ministry of Agriculture and Fisheries (later Ministry of Fisheries), based on recommendations from IMARPE. Chronologically, these interventions may be summarized as follows:"},{"index":3,"size":21,"text":"-1950s: totally uncontrolled fishery, start of research, -1960s: intensification of research, first recommendations by IMARPE regarding the need to control effort."},{"index":4,"size":85,"text":"-1965: prohibition to land catches if 50% or more of the catch consists of fish of 12 cm or less. Also, boats are prohibited to utilize more than 70% of their hull capacity. Introduction of a fishing week of 5 days (Monday to Friday), with the intention of reducing catches by 8%, -1966: introduction of closed seasons (see Tsukayama and Palomares, this vol.), -1971: imposition of production guidelines for processing plants, -1973: creation of the state-owned firm PESCAPERU through nationalization of assets previously held privately."},{"index":5,"size":57,"text":"The creation of PESCAPERU followed the collapse of the fishery in 1972 (Tsukayama 1982 and see other contributions in this vol.) and a major decline of the world market for fish meal (see Aguero, this vol.). Also, large quantities of sardines (Sardinops sagax), mackerels (Scornber japonicus) and horse mackerels (Trachurus rnurphyi) appeared off Peru, partly replacing theanchoveta."},{"index":6,"size":71,"text":"At this time, a piece of legislation was passed which separated the fishing industry into two subsectors: the industrial fishery, based exclusively on the anchoveta, which was to continue producing fish meal, and the fishery for human consumption, based on sardine, mackerel and horse mackerel. The latter fishery, which was to produce fish for the canning industry, was to be conducted with assets (boats, nets, etc) originally used for catching anchoveta."},{"index":7,"size":35,"text":"What happened, however, is that the reduction plants processed large quantities of sardines into fish meal. Some reduction plants, on the other hand labelled themselves \"canning plants\", and continued to process anchoveta into fish meal."}]},{"head":"Material and Met hods","index":87,"paragraphs":[{"index":1,"size":109,"text":"The structured interviews which form the basis of the present contribution were conducted between March and April 1985. The persons interviewed were former crew members and captain of the purse seiners fleet, administrative personnel and employees of reduction plants (see Table 1). Overall, 40 persons were interviewed, all in the area between Puerto Chicama (70s) and Callao (120s). All interviewees had a good educational level and a long experience in the anchoveta fishery. The interviewees were assured they would remain anonymous. The first author, who conducted the interviews, has himself a long professional experience in the fishery, and hence felt he could elicit truthful answers even to \"difficult\" questions."},{"index":2,"size":68,"text":"The questions asked pertained to the following items: 1) Week-end fishing and respect of seasonal bans, 2) Excess fishing with regard to (actual and/or legal) boat capacity, 3) Weight loss of anchoveta as blood (sanguaza), 4) Anchoveta weighed without control in the containers of the factory; use of clandestine pipelines and/or illegal weights and measures, 5) Deals between boat owners and plant managers, 6) Irregular sales of anchoveta."},{"index":3,"size":40,"text":"The mean figures obtained from the interviews were used to correct nominal catches. These corrected catches were used to re-estimate reduction coefficients (landed catchtfish meal produced) and conversion efficiency (fish meal produced/total catch) for the anchoveta fishery as a whole."}]},{"head":"Results","index":88,"paragraphs":[{"index":1,"size":78,"text":"Table 1 summarizes basic information on the 40 interviewees. As might be seen, their ages ranged between 34 to 69 years, with a professional experience in the fishery sector of 19 to 40 years, of which 6 to 27 years were in the anchoveta fishery. A high percentage of the informants were crewmen, assistant of captains, or captain of purse seiners. The following summarizes the available information, by item as in the text above and in Table 2."},{"index":2,"size":42,"text":"1) Twenty-three (58%) informants stated that they had respected the ban on weekend fishing enforced since 1965. All informants appeared to have more or less respected seasonal fishing bans (vedas), whether long or short (this item is not included in Table 2)."},{"index":3,"size":34,"text":"2) Estimates of discard of fish at sea that had been caught in excess of hold capacity ranged from 5 to 15% with an average of 9.13%. Several boats sank because they were overloaded."},{"index":4,"size":33,"text":"3) The estimates of loss of fish as liquid (blood, liquified muscle protein, etc) during transport from fishing to processing plant ranged from 4 to 10% with a mean of 4.9 1 %."},{"index":5,"size":75,"text":"4) Underestimation through misreporting after weighing in processing plants, the use of fraudulent weights and measures and related irregularities were assessed as ranging from 5 to 20%, with a mean of 16.13%. We believe that our respondents overestimated this effect, which Vasquez and Lam (1977), on the other hand, considered to result in an underreporting of 5%. In this contribution, an intermediate value of 10% will be assumed for this step ilt the processing chain."},{"index":6,"size":69,"text":"5) Deals between boat owner and plant manager refer especially to the illegal landing of peladilla (i.e., fish under 12 cm) and/or catches in excess of 70% boat capacity (see above). Such deals also refer, however, to discounting of landed weight of fish because of their bad quality. The estimates of the effect of such deals on catch estimation ranged between 1 and 10% with a mean of 4.38%."},{"index":7,"size":97,"text":"6) Irregular sale of anchoveta refer according to Vasquez and Lam (1977) to sales made by the crew of a boat, without the boat owner being aware of the transaction. We consider here, however, the far more important aspect of irregular sale, called pesca negra, i.e., the reporting of anchoveta~atches under another species name (i.e., as sardines, mackerel or horse mackerel). Irregular sale as defined here is reported to have involved 2-10% of the catch, with a mean of 4.11% before and 6.23% since 1975. A mean of 5% is used for the period 1951 to 1982."},{"index":8,"size":22,"text":"The percentages presented above were then used to estimate actual from nominal catches, as shown on Table 3 (columns A and D)."}]},{"head":"Discussion","index":89,"paragraphs":[{"index":1,"size":111,"text":"The first question to be discussed here relates obviously to the trustworthiness of our informants and to the precision of their estimates. All we can suggest here is that these informants as a group appears to be broadly representative of the fishery, and that we have no reason to assume they would have lied to us. However, not being trained in interviewing techniques, we cannot exclude the possibility of having posed some \"leading\" questions, or unwittingly nudged our respondent toward values within a range we considered reasonable. More important however might be the possibility of personal biases affecting our respondents' recollections, as was probably the case with regard to item #4."},{"index":2,"size":57,"text":"Moreover, it is important to realize that the mean percentages presented here cannot really be generalized for all the years, seasons or months because they varied depending on factors only some of which were considered here (e.g., fishing regulations, see above) while some are not explicitly considered (e.g., the absolute and relative abundance of sardines and anchoveta)."},{"index":3,"size":51,"text":"Throughout the period 1951 to 1982, however, all irregularities discussed and quantified here did occur (as they presently do in the sardine fishery) although not all at the same time. 1) Answers to question \"did you respect the regulation stating that you should not fish on Saturdays and Fridays? (since 1965)."},{"index":4,"size":17,"text":"2) % catch in e&ss of storage ro& capacity (released dead-or dying after capture by purse seine)."},{"index":5,"size":86,"text":"3) % loss of catch due to losses of \"blood\" (icl. liquefaction of entire fish) during storage and transport. 4) % \"losses\" due to r n i s r e p h g after weighting in processing plants to illegal pipelines tapping h e plant fish transport system and to use of fraudulent weights and measures. 5) 40 \"losses\" due to illegal deals between plant managers and fleet owners. 6) % \"losses\" due to illegal and unregistered sales of fish (in part to clandestine processing plants)."},{"index":6,"size":35,"text":"Thus, excessive fishingin relation to storage capacityoccurred in the 1950s because the boats did not have echo sounders, hence the likely catch from a given purse seine set could not be estimated ahead of time."},{"index":7,"size":160,"text":"On the other hand, the abundance of anchoveta and the fact that the boats could land their catch without having to wait at the harbor allowed fishing to proceed systematically, without undue haste. We believe therefore that in the early 1950s, excessive fishing (item #2 in our list) may have been less prevalent than in the period from 1961 to 1973, because when the number of vessels increased, port waiting time also increased, leading to a reduced willingness to spend Table 3. Basic statistics on the Peruvian reduction fishery for anchoveta 1953-1982. Based on nominal landings and fish meal production figures in Tilic (1963a) in MIPE (1977)(1978)(1979) and unpublished data held at IMARPE. Corrected fish landings and estimated catch were computed using the raising factors given in the text. Also see text for discussion of reduction coefficients (columns F and G) and of gross conversion efficiency (column H). time transporting to other boats fish caught in excess of hold capacity."},{"index":8,"size":53,"text":"Similarly, when the decree to prevent the landing (and presumably the catch) of small fish came in force, vessels that caught such fish discarded them at sea (dead, obviously) in order to avoid payment of a fine. Later, when enforcement became lax, fishermen began again to land small fish, which however, remained unregistered."},{"index":9,"size":92,"text":"In the period 1974 to 1982, finally, restrictions of fishing areas, fishing time and catch quotas led to increased duration of fishing trips, and excess catch increased again, along with other irregularities in recording landings. These historical facts lead to consider the 5% reported by Vasquez and Lam (1977) as an estimate of item (#2) as too low. On the other hand, our estimate of the losses of anchoveta in the form of blood and other liquids is similar to those obtained by Vasquez and Lam (1977) and Sanchez and Icochea (1968)."},{"index":10,"size":132,"text":"With reference to deals between vessel owners and plant mazagers (item #5 in our list above), it must be noted that Vasquez and Lam (1977) included here only illegal sales of (illegally caught) small fish. For these, the plant managers would pay only 40-50% of the price of legally caught anchoveta. However, there were additional deals involving the sale of fish caught in excess of 70% boat storage capacity. Such fish were \"given\" to the factory, in return for preferential treatment upon landing regularly caught fish. This type of deal occurred from 1965 to 1975, when the pertinent decree was being enforced. Therefore, we believe that our estimate of losses in connection with item #5 (i.e., 4.38%) is more realisticfor 1965-1975 at leastthan the estimate of 0.5% in Vasquez and Lam (1977)."},{"index":11,"size":38,"text":"Our value of 5% for item #6, i.e., the irregular sale of anchoveta also appears more realistic than the 0.5% value of Vasquez and Lam (1977), who, here also, considered only a small aspect of the overall problem."},{"index":12,"size":136,"text":"This brings us, finally, to the question as to how realistic the combined effect of our various estimates are, when compared with independent, albeit indirect evidence. Such evidence is provided by \"reduction coefficients\" i.e., the quantity of raw material (anchoveta, wet weight) needed to produce one unit weight of fish meal. This coefficient was estimated on the basis of on-the-spot analyses by Tilic (1963aTilic ( , 1963bTilic ( ) in 1960Tilic ( -1961 for factories from Chimbote to Callao, and by Arnesen and Sanchez (1963) for 23 different factories along the Peru coast, as 5.1-6.5 and 5.85, respectively. Data pertaining to the fishery as a whole, for 1968-1977 lead to an overall mean reduction factor of 6.0 (MIPE 1979), a relatively high value which we consider realistic, and which is close to the values cited above."},{"index":13,"size":47,"text":"On the other hand, if we compute the reduction factor using columns D and E of Table 3, we obtain rather low values, ranging between 4.18 and 5.82, with a marked decrease in the 1970s, notwithstanding the increased use of low yielding small fish in that period."},{"index":14,"size":104,"text":"If, however, we assume, as indeed empirical investigations demonstrate, that the industry never operated with reduction coefficients below 5, and that a value of 6 is nearer to the mark, then we obtain corrected fish landings markedly higher than nominal catch: by 8.7% in the 1950s, 7.5% in the 1960s and 22% in the 1970s. Thus using realistic reduction factors, and reported fish meal production (which, incidentally, may also be underestimated) leads to catch corrections roughly similar to those derived previously, especially as far as the 1970s are concerned, i.e., the important period during which peak landings were achieved, and the fishery subsequently collapsed."}]},{"head":"Introduction","index":90,"paragraphs":[{"index":1,"size":39,"text":"Growth studies of the anchoveta (Engraulis ringens Jenyns) inclusive of growth parameter estimates such as needed for population modelling have been few, especially in view of the fact that this species once supported the largest fishery of the world."},{"index":2,"size":75,"text":"We shall first review, in this contribution, what is known of the growth of the Peruvian anchoveta, then proceed to present new information, based on daily rings, on the growth of larvae and growth parameter estimates for each year during the period 1954 to 1982, based on detailed analysis of length-frequency data. These growth parameters describe curves that oscillate seasonally, and we shall thus discuss seasonal oscillations as a feature of the biology of anchoveta."},{"index":3,"size":58,"text":"We shall also demonstrate that the overall growth performance of anchoveta increased-hm the early 1950s to the early 1980s and recent evidence of this being due to density-related effects will be presented. Also, we shall present a model developed to simulate the growth of anchoveta and use its results to identify and quantify some of these density-related effects."},{"index":4,"size":69,"text":"Overall, our goals shall be both to consolidate the present knowledge of the growth of the Peruvian anchoveta and to provide a basis for other investigations, notably population studies involving length-based Virtual Population Analyses (see, e.g., Pauly, Palomares and Gayanilo, this vol.). For the latter we shall use the seasonally oscillating growth equation of Pauly and Gaschiitz (1979) Lt = L, (1exp (-K (t-to) +GK/2 7 sin 27 ,(tts)))"},{"index":5,"size":110,"text":"... 1) where Lt is the length at age t, L, the asymptotic length, K a growth coefficient with dimension t-1, t, the (hypothetical) age at which length would be zero ifthe adult and subadult growth curve could be extrapolated back to the origin, ts the age at onset (with reference to t = 0) of a growth oscillation of period 1 year and C is a dimensionless constant expressing the amplitude of a growth oscillation, with dVdt reaching zero once a year when C = 1. Equation ( 1) reverts, when C = 0, to the standard von Bertalanffy growth function (VBGF), i.e., Lt = L, (1exp (-K (tto)))"},{"index":6,"size":8,"text":"... 2) (Bertalanffy 1938;Beverton and Holt 1957;Pauly 1984a)."}]},{"head":"Growth of Larvae and Early Juveniles","index":91,"paragraphs":[{"index":1,"size":71,"text":"Anchoveta growth, as reviewed here, relates only to posthatch growth (see Santander and Castillo 1973 for an account of embryonic growth). Little work has been done on the growth of anchoveta larvae, and even less was formally published. We have thus complemented this part of our review with some original data (see Table 1, Figs. 1 and 2) both to assess the validity of previous results and to expand on these."},{"index":2,"size":52,"text":"Table 1 summarizes some of the available information on growth and age of larvae anchoveta; these data suggest that anchoveta reach a length of about 1.5 cm at the end of their first month of life, and then continue growing at a rate of about 1.4 cm per month (see Fig. 2)."},{"index":3,"size":45,"text":"This inference confiis earlier results of Pastor and Malaga (1966), based on tetracycline marking of fish of 2.58 cm mean length, which reached 3.90 and 4.30 cm after 23 and 49 days, respectively, and also leading to a mean monthly growth rate of 1.4 cm."},{"index":4,"size":74,"text":"\"Recruits\" in the contributions included in the present volume are defined as fish of total length of 4.25 cm, corresponding to the median of the smallest length class frequently represented in catch samples (see Tsukayama and Palornares, this vol.). Estimates of the age of such recruits can be obtained using two independent approaches: i) forward projection of larval growth rate, or ii) backward projection of growth curves describing the growth of subadults and adults"},{"index":5,"size":30,"text":"The first of these approaches, using 1.4 crn/month and a length at hatching of 2 mm (Einarsson and Mendiola 1963) yields an age of 2.89 month for 4.25 cm fish."},{"index":6,"size":58,"text":"The second approach assumes that growth is described by the VBGF (without seasonal growth oscillations). Using the values of L, and K in Table 2 to estimate a t for fish of 1.5 to 4.25 cm from the equation  a Mean length of fish in study of Mendiola and Gomez (1981), with n = 8.7,10 and 10, respectively."},{"index":7,"size":19,"text":"Using agelength relationship (Gompertz curve) for E. rnordax and-assuming similar growth for E. ringens (from Mendiola and Gomez 1981)."},{"index":8,"size":27,"text":"Assuming that absolute age in days is equal to number of daily rings counted + 3 (note that addition of 3 days may not have been appropriate)."},{"index":9,"size":6,"text":"From linear regression in Fig. 2."},{"index":10,"size":35,"text":"- Fig. 1. Relationship between the total length and the otolith radius of anchoveta (Engraulis ringens) larvae and early juveniles. The drawings of anchoveta larvae and early juveniles are adapted from Einarsson and Mendiola (1963)."},{"index":11,"size":7,"text":"~t = (1K) loge ((L, -1.5)/(L, -4.25))"},{"index":12,"size":75,"text":"leads to ~tvalues ranging from 1.24 to 2.81 month, with a mean of 1.84 to which one month must be added to account for the time needed to grow to 1.5 cm. Thus, fish of 4.25 cm would have an age of 2.84 months, extremely close to the value obtained using the first approach, and in fact matching quite precisely the age at recruitment estimated on the basis of the original data in Fig. 2."}]},{"head":"Growth of Subadult and Adult Anchovies","index":92,"paragraphs":[{"index":1,"size":21,"text":"The studies on the growth of subadult and adult anchoveta conducted to date can be subdivided into two, very unequal groups:"},{"index":2,"size":67,"text":"(i) growth inferences based on the study of hard parts, i.e., scales (Barreda 1953;Simpson and Buzeta 1967) and otoliths (see Table 2), and Fig. 2. Relationship between the total length and the estimated age in days of anchoveta (Engraulis ringens) larvae and early juveniles. Note that \"recruits\", as defined in this volume (i.e., fish of 37.5-47.5 mm) have an estimated age of slightly less than 3 months."},{"index":3,"size":28,"text":"(ii) detailed analysis of length-frequency data using either Modal Progression Analysis (MPA, see Table 2) or the ELEFAN I program of Pauly and David (1981 and see below)."},{"index":4,"size":31,"text":"Overall, the results obtained by these two groups of methods tend to agree, although some of the \"annuli\" reported earlier (e.g. Barreda 1953) now quite clearly appear to have been artefacts."},{"index":5,"size":70,"text":"Thus, while studies of hard parts (especially otoliths) usually provide the crucial evidence in controversies about the growth of fish, it is apparent that the studies of hard parts of anchoveta have not (to date) progressed far enough to help settle controversies involving E. ringens. Therefore, all following considerations will have to be based on the results of length-frequency analyses and consequently on the problems occurring when performing such analyses."},{"index":6,"size":48,"text":"Three of the sets of growth parameters in Table 2 were estimated using MPA, while one set was obtained using the computer-based ELEFAN I method of Pauly and David (1981). A number of problems occur when performing MPA \"by eye\" (George and Banerji 1964). The most important are:"},{"index":7,"size":35,"text":"(i) the linking of modes thought to belong to the same cohort is entirely subjective (Pauly et al. 1984), (ii) seasonal growth oscillations can usually not be identified, let alone taken explicitly into account, and"},{"index":8,"size":39,"text":"(iii) incomplete selection to the gear and incomplete recruitment cannot be accounted for and hence modes pertaining to small fish, being shifted toward larger sizes, tend to lead to underestimation of the parameter K of the VBGF. (Pauly 1986b)."},{"index":9,"size":82,"text":"The first study providing estimates of growth parameters in E. ringens appears to be that of Saetersdal and Valdivia (1964). Although their estimate of L, is much lower and their value of K consequently higher than those of their successor (see Table 2), the value of the growth performance index @' = log loK + 2logloL, ... 4) (Pauly and Munro 1984;Moreau et al. 1986) is roughly similar to the other, later values, with differences being explainable through density-related changes (see below). "}]},{"head":"Changes in Anchoveta Growth Rates","index":93,"paragraphs":[{"index":1,"size":18,"text":"Three types of growth rate changes have been discussed by previous authors in relation to the Peruvian anchoveta:"},{"index":2,"size":32,"text":"(i) changes of growth rates related to t h ~ influence of El Niiio, (ii) density-dependent changes in anchoveta growth, due to the decline of stock biomass in the last decades, and"},{"index":3,"size":4,"text":"(iii) seasonal growth oscillations."},{"index":4,"size":99,"text":"With regard to the fnst of these three items, the only information available is by Tsukayama and Alvarez (1981), who stated (without presenting evidence) that the growth rate of anchoveta declined in 1976, an El NiEo year. However, it is doubtful, given the low overall reliability of length-frequency analyses \"by-eye\" that such effect could be convincingly demonstrated. Rather, indirect evidence should be sought, e.g., the reduction of the \"condition\" of the anchoveta during El N i k , a phenomenon pointed out previously by various authors (e.g., Valdivia 1978, Fig. 7) and which will be elaborated upon further below."},{"index":5,"size":140,"text":"At least three papers relate, albeit indirectly, to density-dependent changes of anchoveta growth: i) Jordan (1980) points out that \"recent information for the Peruvian coast [...I indicated an evident shift of Lmax up to 20.5 cm [... which] may be the result of a changing growth rate\", ii) Cushing (1981) assumed density-dependent growth (no evidence is given, however), and iii) Pauly and Tsukayama (1983, Table 1) presented a series of growth parameter estimates (L, and K) which lead to estimates of @' (see equation 4) that show a weak, positive correlation with time (r = 0.396, df = 17), a fact which they missed and which tends to contradict their statement that \"the growth parameters describing the growth in length of the northern stock of the Peruvian anchoveta have been more or less constant for the period 1961 to 1979\"."},{"index":6,"size":199,"text":"As will be shown below, we have now found strong evidence for density-dependent growth in the Peruvian anchoveta, confirming Jordan's observations. Seasonal growth oscillations have been previously discussed by Cushing (198 1) who suggested, with regard to newly recruited, small anchoveta that \"the first group arises from the spawning in August, September, and October and the second from that in January; the latter grows somewhat more slowly because its initial growth occurs outside the period of major upwelling\". Evidence for this statementwhich implies that (cold!) upwelled water accelerates growthwas not presented, however. Pauly and Tsukayama (1983), on the other hand, estimated for the period of 1961 to 1979 a mean value of 0.3 for the parameter C of equation (1). This implies that growth rate diminishes by 30% below normal during that part of the year when growth is most strongly reduced. They also showed that this 30% reduction occurs in September, i.e., in (southern) winter, when mean monthly water temperatures off Peru are lowest (Zuta and Urquizo 1972;Bakun, this vol.;Brainard and McLain, this vol.) With this, we conclude this brief review of anchoveta growth, since enough of a stage has been set for the new results which follow."}]},{"head":"Reassessing the Growth of Anchoveta","index":94,"paragraphs":[]},{"head":"Material","index":95,"paragraphs":[{"index":1,"size":28,"text":"The length-frequency data used for this analysis and pertaining to the northernlcentral stock are given in Tsukayama and Palomares (this vol.). The following important points should be noted:"},{"index":2,"size":67,"text":"i) interpolations of the length-frequency data, such as used in Pauly, Palomares and Gayanilo (this vol.) for the Virtual Population Analyses were not used for the growth analysis, ii) the available data were grouped in files covering the entire \"life\" of cohorts, from their entry into the fishery to their disappearance therefrom, or at least to the time when their modes merged with those of other cohorts."},{"index":3,"size":67,"text":"The second point implies that the procedure used by Pauly and Tsukayama (1983) for estimating growth parameters from data set covering at most the 12 months of a calendar year was avoided, and that the \"year\" used here to label a given cohort does not refer to the time when the data were collected, but to the time when the cohort in question had its peak biomass,"}]},{"head":"Methods","index":96,"paragraphs":[{"index":1,"size":129,"text":"The ELEFAN I method, used here to perform all growth analyses, essentially consists of three main routines: i) a routine to transform the available length-frequency data such that peaks on the data are expressed as positive points, and troughs as negative points, ii) a routine to \"trace\" growth curves through the transformed data, to score the positive and negative points \"hit\" by the curve (\"Explained Sum of Peak, or ESP, analogous to explained variance in parametric statistics) and to relate this ESP to the sum of positive peaks (\"Available Sum of Peaks\" or ASP, analogous to total variance) in the data set, iii) a routine that changes, in small steps, seeded values of the parameters of equation ( 1) until a curve is found which maximizes the ratio ESPIASP."},{"index":2,"size":27,"text":"This algorithm assumes that the single, consistent growth curve which passes through most peaks, while avoiding troughs as far as possible, is the mean population growth curve."},{"index":3,"size":27,"text":"An extensive literature now exists on the ELEFAN I program, its assumption and sources of bias (see Pauly 1985aPauly , 1985bPauly , 1986aPauly , 1986b, 1986~), 1986~)."},{"index":4,"size":393,"text":"For the actual fitting, we have used the graphics-oriented version of ELEFAN I developed by Saeger and Gayanilo (1986); this version incorporates several improvements suggested by early users of ELEFAN programs (see Pauly 1985a). Moreover, we have estimated growth parameters in a cohort-specific basis (as was not done in earlier applications of the programs) and largely compensated for the bias due to incomplete recruitment and gear selection (Paul-y 1986b), as follows: i) preliminminary estimate of growth parameters (L, and K) were obtained using the original length-frequency data; these parameters were then averaged over the period 1953 to 1984, ii) a file was created from the original data set in (i) which included the month(s) from each year (for the period 1953 to 1982) that contained the smallest fish, iii) using the values of L, and K obtained in (i) and the file derived in (ii), a lengthconverted catch curve (Pauly 1984~) was derived, from this, the number of fish that would have been caught, had it not been for incomplete selection and recruitment, was estimated using the method described in Pauly 1984a, Table 5.6), iv) the number of fish actually caught for each length class was divided by the number estimated in (iii) thus providing estimates of the probabilities of capture, and finally v) the probabilities of capture estimated in (iv) were used to correct all available lengthfrequency data files, and the corrected files were then used to re-estimate growth parameters (see Fig. 3). A) Estimation of a preliminary set of growth parameters for each cohort from 1954 to 1982 and averaging of the same; B) Derivation of a length-converted catch curve based on growth parameters estimated in (A) and an accumulated length-frequency file composed of data from the two months of each year containing the smallest fish (such as to obtain a correction for gear selection covering a size range as wide as possible); backward projection of catch curve to estimate number of fish that would have been caught, had it not been for gear selection and/or incomplete recruitment; C) Estimation of probabilities of capture from the ratio of fish caught to virtual fish, by length, and division of all original length-frequency data by the appropriate probabilities of capture, and D) Reestimation of growth parameters (solid line). The correction leads to much improved estimates of the von Bertalanffy parameter K (see text)."}]},{"head":"Growth Parameter Estimation of Anchoveta Cohorts, 1954 to 1982","index":97,"paragraphs":[{"index":1,"size":38,"text":"Table 3 gives the growth parameter estimates obtained using the ELEFAN I program. These estimates are, on the average, similar to those obtained, for the northern substock, for the years 1961 to 1982 by Pauly and Tsukayama (1983)."},{"index":2,"size":102,"text":"Fig. 4 gives a graphic representation of the growth of several cohorts, and of the transformed length-frequency data from which the growth curves were estimated. Faint seasonal growth oscillations will be noted: they would probably not have been picked up by anyone performing the same analysis by tracing growth curves \"by eye\" only. The mean value of C is 0.27, which is very near the 0.3 value estimated earlier by Pauly and Tsukayama (1983); the greatest reduction of growth rate occurred, on the average, in mid-August, i.e., the mean winter point value is 0.62, against 0.7 cm estimated earlier (see Table 3)."},{"index":3,"size":83,"text":"Note, finally, that the shaded histograms corresponding to very small fish tend to be to ;he right of the estimated growth curve: this is due to the fact that the correction for gear selection and/or incomplete recruitment did not suffice for entirely de-biasing the original lengthfrequency data. Still, we see that the growth curves so obtained start at the appropriate time after the spawning seasons and follow the cohorts throughout most of their passage through the fishery, up to their complete disappearance therefrom."}]},{"head":"Evidence for Density-Dependent Growth, 1953 to 1982","index":98,"paragraphs":[{"index":1,"size":77,"text":"The available evidences for density-dependent growth in anchoveta are: i) the strong positive correlation between time and the maximum length in catch samples, from 1953 to 1982 (Table 3, Fig. 5); ii) the extremely tight positive correlation between time and the values of the growth performance index @ ' from 1954 to 1982 (see Table 3, Fig. 5),and; iii) the significant (P = 0.05) negative correlation between @' and anchoveta biomass for 1954 to 1982 (Fig. 6)."},{"index":2,"size":28,"text":"These three lines of evidence, both separately and considered together, strongly confirm the earlier suggestions of a change in the growth patterns of anchoveta summarized in Fig. 7."},{"index":3,"size":94,"text":"A possible explanation for density-dependent growth in the Peruvian anchoveta might be provided by the fact that the surface area of fish gills (G) and hence fish metabolism increase in proportion to a power d <1 of body weight (W), i.e., Thus, relative gill areas and metabolism must decrease with increasing body weight, down to a metabolic level (at W,, the asymptotic weight) at which average oxygen supply meets average oxygen requirements (Fig. 8). Therefore, an increase in asymptotic size as demonstrated above impliesgiven no change in gill structurea decrease of average oxygen requirements."},{"index":4,"size":133,"text":"The magnitude of this decrease can be estimated, from predicted values of W, for 1950 and 1980 of 40 and 80 g, respectively (see Fig. 8) because in small fish such as anchoveta, the value of d (in equation 5) usually ranges between 0.7 and 0.8 (Winberg 1960;Pauly 1981). Thus, using the midrange of the two latter values we have which express the average oxygen consumption of anchoveta in 1980 as a fraction of their consumption in 1950. Or put differently the model in Fig. 8 and equation (6) suggest that the \"cost of living\" of an anchoveta may have dropped, for 1950 to 1980 by approximately 16%. We shall present further below a simulation model developed to test this estimate and explore some possible scenarios for density dependence in the Peruvian anchoveta."},{"index":5,"size":21,"text":"Before we turn to this model, however, three more aspects of anchoveta growth, not considered in this model, should be discussed. "}]},{"head":"Latitudinal Differences in the Growth of Anchoveta","index":99,"paragraphs":[{"index":1,"size":134,"text":"Fish stocks belonging to the same species, but occurring along a latitudinal gradient usually display trends in their growth parameters, with higher asymptotic sizes (and consequently lower K values) at the \"cold poleward ends of their distributions (see Pauly 1984a;Longhurst and Pauly 1987). The anchoveta is no exception to this rule as documented by Brandhorst (1966) who, based on a survey conducted in 1961, wrote that \"while [the anchoveta] has a maximal size of about 17 cm in the Arica-Iquique area [190S], fish of 18 cm were measured in Valparaiso [330S] and of up to 21 cm in Talcahuano [370SIu (our translation from German). Note that in 1961, the anchoveta off northern/central Peru reached a length of about 17 cm, similar to the one in northern Chile, but markedly less that off southern Chile."}]},{"head":"Temperature and the \"Condition Factor\" and Fat Content of Anchoveta","index":100,"paragraphs":[{"index":1,"size":161,"text":"The condition factor (c.f.) of fish is defined by in which c.f. is divided by 100, when length (L) is expressed in cm and weight (W) in g to have c.f. values near unity in fish with \"normal\" shapes. Changes in the length-weight relationship of fish can be directly related by comparing their c.f. values because the exponent of the lengthweight relationship (equation 7) is here fixed at a value of 3, corresponding to isometric growth. Tables 1 to 30 of Tsukayama and Palomares (this vol.) include actual values of c.f., computed from anchoveta samples that had been weighed, measured and counted, as well as values (in brackets) that have been estimated using the plot of actual c.f. values on the mean monthly sea surface temperatures in Fig. 9. As might be seen, Fig. 9 suggests a rather strong negative correlation between c.f. and temperature, confirming previous reports of anchoveta looking emaciated at the high temperature occurring during El Ni5o conditions."},{"index":2,"size":31,"text":"A plot of water temperature against the coefficient of variation of c.f. values (see Fig. 10) is presented here to show that c.f. values become more variable at high water temperatures."},{"index":3,"size":50,"text":"Fig. 11 shows plots of the monthly means of actual c.f. values (i.e., excluding the bracketed values in Tables 1 to 30 of Tsukayama and Palomares, this vol.) against month and sea surface temperature against month, showing in different form the strong inverse relationship between c.f. and sea surface temperature."},{"index":4,"size":50,"text":"Fig. 11 also shows the seasonal changes in growth rate (in length, i.e. dVdt) of anchoveta, as implied by a value of C = 0.3 and a \"winter point\" of 0.7 (WP = pe~iod of the year when growth is slowest; a parameter estimated by ELEFAN I, see Table 3)."},{"index":5,"size":36,"text":"As might be seen, dl/dt in anchoveta varies inversely with condition (and hence relative weight). This relationship can be understood, however, only when simultaneously considering the seasonal dynamics of anchoveta fat content, as briefly sketched below."},{"index":6,"size":52,"text":"In anchovies from temperate waters, as in most other fish from higher latitudes, the fat content fluctuates seasonally, being highest at the end of the \"growth season\" (summer/autumn) and lowest just after spawning. This cycle is here illustrated by Fig. 12A, which pertains to Engraulis encrasicholus and is adapted from Shul'man (1974)."},{"index":7,"size":32,"text":"In Engraulis ringens from northern Chile where strong seasonal temperature oscillations occur, the fat content also varies seasonally in sinusoidal fashion. This is illustrated here by Fig. 12B based on Brandhorst (1966)."},{"index":8,"size":48,"text":"Further toward the equator, however, the dynamics of fat content become more complex, with the available data suggesting that fat content peaks twice a year, i.e., before the August-October and January-February spawning seasons. This is illustrated here by Fig. 12C and 12D based on Einarsson et al. (1966)."},{"index":9,"size":236,"text":"In this context, it was also observed that anchoveta, during positive temperature anomalies (i.e., El NSo events), have a higher than average fat content. Thus, IMARPE (1972) wrote: Sea surface temperature (OC) Fig. 10. Relationship between the variability of anchoveta condition factor (c.f.) estimates and mean monthly sea surface temperature (based on data in Tables 1-30 in Tsukayama and Palomares, this vol.). At low temperatures, anchoveta have high c.f. values that vary little between different months. At high temperatures, wide fluctuations between adjacent months and between the same months of different years may be observed.  An indirect indication of a low level of spawning is given by the fat content of the fish, which is fairly high before spawning and normally decreases during the spawning period h m a level which is on the average around 12 or 13% in June to one of about 5% in September for fish of over 14 cm. In 1971 the fat content in March-June was in general somewhat below the average for these months, but in September-October the values were some 2% higher than average, and increased to very high levels in December (18.8% in Chimbote, 14.1% in Callao and 11% in Ilo). This behavior may be an adaptive response which enables anchoveta to store energy in a form which does not increase oxygen requirements (as an energy storage in the form of protein would (see Pauly 198 1)."},{"index":10,"size":35,"text":"The time-series data in the various contributions in this volume, the fat content data in Lam (1968) and other unpublished fat content data presently on file at IMARPE will allow a test of this hypothesis."},{"index":11,"size":48,"text":"In the meantime, it suffices to recall here that: i) anchoveta show seasonal oscillations of their growth rate (in length) which correlate with temperature, and ii) fat content and condition tend to be inversely related, with fat content usually being higher during periods when oxygen requirements are high."}]},{"head":"A Simulation Model of Anchoveta Growth","index":101,"paragraphs":[]},{"head":"Basic Concepts and Assumptions","index":102,"paragraphs":[{"index":1,"size":53,"text":"The remarkable increase of growth performance in the anchoveta, for the early 1950s (here \"1950\" for simplicity's sake) to the early 1980s (here \"1980\") could be the result of changes in availability and/or composition of anchoveta food resources as a consequence of the decline of the anchoveta's own biomass during the last decades."},{"index":2,"size":132,"text":"This hypothesis, suggesting that anchoveta growth is density-dependent:, cannot be tested using empirical data as the detailed time series data on file at IMARPE on anchoveta stomach contents are presently being processed, and will not be available early enough for consideration in the present volume. Therefore we have tested the hypothesis of density-dependent growth in anchoveta using a relatively simple deterministic simulation model based on the results of experiments and simulation conducted and reported upon earlier by Villavicencio (1981), and Villavicencio andMuck (1983a, 1983b). This model does not consider the seasonaly oscillating factors discussed in the preceding section, but does include a term for spawning (see where G = growth rate, R = ration, A = assimilation, MLr= losses associated with resting metabolism and MLa = losses associated with active (feeding) metabolism."},{"index":3,"size":67,"text":"Three different types of changes are here considered explicitly with regard to their potential impact on growth rate: i) competition forfood within the anchoveta stock: reduced biomasses of anchoveta imply, given constant production of anchoveta food, an increased availability of food per surviving anchoveta. Thus, if R*A in equation ( 9) remains constant, G should increase because MLa (expressing the metabolic losses associated with feeding) should decline."},{"index":4,"size":99,"text":"ii) food quality: here, it is assumed that the decline of anchoveta biomass has reduced the grazing pressure on phyto-and zooplankton, resulting in (a) reduced competition for filterfeeding zooplankters and (b) reduced zooplankton mortality through anchoveta predation. Both items (a) and (b) should result in a relative increase of zooplankton vis h vis phytoplankton and hence in an increase of zooplankton in the diet of the anchoveta, an opportunistic feeder. Such shift in diet composition would entail (a) an increase of caloric content per unit weight of anchoveta stomach content and/or (b) an increased assimilation of the ingested food."},{"index":5,"size":68,"text":"iii) feeding mode: a shift from feeding predominantly on phytoplankton to feeding predominantly on zooplankton would not only entail a shift in food quality, however. Rather, this would result in a shift of feeding mode, from filter to particulate feeding, and hence from a less to a more efficient (in terms of net energy gain) mode of feeding. This would result in G increasing because MLa would decrease."},{"index":6,"size":16,"text":"Fig. 13 summarizes the assumption listed above and the overall structure of the model presented here."}]},{"head":"Implementation of the Model","index":103,"paragraphs":[{"index":1,"size":54,"text":"Table 4 lists the equations used to implement the model in Fig. 13 along with their sources. Anchoveta growth was integrated in steps of one day, starting from an assumed length of 10 cm and a weight of 9 g at age 1 year up to an age of 4 years for all simulations."},{"index":2,"size":60,"text":"\"1950\" was simulated by calibrating the model such that 16 cm was reached at 4 years. Then, the sensitivity of the model was tested by changing one at a time the value of four parameters, related to the three factors discussed above. Finally, \"1980\" (when 20 cm is reached at an age of 4 years) was simulated, by assuming alternatively:"},{"index":3,"size":15,"text":"(a) only one of the four parameters changed, and (b) all four parameters changed simultaneously."},{"index":4,"size":7,"text":"The four parameters subjected to change were:"},{"index":5,"size":12,"text":"assimilation coefficient (A); caloric food content; costs per unit time feeding; and"},{"index":6,"size":5,"text":"duration of feeding period (ha)"},{"index":7,"size":4,"text":"(see also Table 4)."},{"index":8,"size":147,"text":"The duration of the feeding period'(O<ha&4h) was made to change by varying food density (in @liter). When food density is high, 'the time needed to reach R, the maximum ration is short; conversely, when food density is low, anchoveta may feed for up to 24 hours. [Daily ration (R) is usedhere as an upper limit which anchoveta try to reach either by increasing their feeding time or by changing their feeding mode.] Metabolic costs per unit feeding time were changed by multiplying the energy expenditure during filter feeding ( m a ) with a factor (f) and total metabolic losses (ML) computed from where ha and hr are as defined in Table 4, and where f was set = 1 when the fish are filter feeding and 0.3 when the fish had shifted toward particulate feeding. Table 4 shows the ranges within which each parameter were varied."},{"index":9,"size":17,"text":"Thus for each (daily) step, somatic growth in weight, length and gonad growth were computed as follows:"},{"index":10,"size":25,"text":"i) filtering rate (literlhour) is computed from the length-specific swimming speed, area of the open mouth and the frequency by which the mouth is opened;"},{"index":11,"size":241,"text":"ii) filtering rate was multiplied by particle density (gfliter) to give ingestion rate (ghour), iii) the computed ingestion rate is related to the upper limit of the ration, and the time is computed that is needed to reach this upper limit;   Villavicencio and Muck (1983a) adapted from Villavicencio and Muck (1983a, 1983b, 1985) Villavicencio and Muck (1983a) T was set at 1 8 ' ~ throughout 0.3 value from data in Winberg (1960) Villavicencio and Muck (1985) Villavicencio and Muck (1985) Villavicencio and Muck (1985) this contribution (see text) as defined in text assumed value (see text) assumed value (see text) Hunter et al. (1986); Hunter and Leong (1981) iv) the filtering time required to obtain the daily ration is subtracted from 24 hours to obtain the resting time; v) daily metabolic losses are computed as the sum of losses during feeding and losses during resting time; vi) estimated daily ration is converted into calories, and the fraction of these calories available for growth is computed by multiplication of the (caloric) ration by the assimilation coefficient; vii) daily total production is computed as the difference between the fraction of the food available and the sum of the metabolic losses; viii) daily production (i.e., the growth increment) is added to the previous size estimate, i.e., to length, body weight and gonad weight; ix) the body weight is decreased by the weight of the gonad when gonad weight reaches 10% of body weight;"},{"index":12,"size":12,"text":"x) return to (i) until an age of 4 years is reached."},{"index":13,"size":118,"text":"Fig. 14 shows anchoveta growth curves obtained for \" 1950\" and \" 1980\" conditions. Table 5 summarizes the results for two runs, while Table 6 shows the main results of the sensitivity analyses of the model. As might be seen, assimilation coefficient and caloric food content, both related to the food quality factor, have the strongest effects on model output. Table 6 also shows that these parameters, considered alone must be changed by 45% for anchoveta to change from its \"1950 pattern to the \" 1980\" pattem. On the other hand, if all parameters in Table 6 are changed simultaneously, a change of only 15% is needed for the \" 1950\" to \"1980\" transition (see also Fig. 14)."},{"index":14,"size":4,"text":"\" 1980 conditions I'"}]},{"head":"Age (years)","index":104,"paragraphs":[{"index":1,"size":53,"text":"Fig. 14. Growth of Peruvian anchoveta as simulated using model described in text. Note that most of the difference between the \"1950\" and \"1980\" growth curves is due to growth rate differences of fish between 1 and 1.5 years  a As compared with a length of 16 cm for the initial (\"1950\") conditions."}]},{"head":"Discussion","index":105,"paragraphs":[{"index":1,"size":155,"text":"Overall, our growth estimates match those obtained previously by other authors (Table 7, Fig. 15). However, the key finding presented here is obviously the strong evidence for densitydependent growth in the Peruvian anchoveta, a phenomenon for which only anecdotal evidence existed previously. Our ability to extract evidence for changes in the growth performance of anchoveta depended on two main factors not found elsewhere: i) the availability of an extremely long time series of length-frequency data, and ii) the availability of powerful software for the analysis of these time series.   (1978), Radovich and MacCall (1979), Volovik and Kozlitina (1983), Melo (1984) and Gallardo-Cabello (1985). B. Frequency distribution of growth performance indices in E. ringens showing that the species grows, as a whole, better than other Engraulis spp.; the arrows refer to values of @' for 1950 and 1980 on Fig. 5 and express the density-related change of anchoveta growth identified in the present study (see text)."},{"index":2,"size":87,"text":"Density dependence, as demonstrated here, is not confined to anchoveta and indeed numerous authors have shown that growth rates are density-dependent in a large number of fishes, especially as far as juveniles are concerned (Cushing 1973;Mathisen et al. 1978). This is confixmed by our simulation model, which generated mean growth rates of about 0.05 mrn/day for both \"1950\" and \"1980\" conditions in fish of more than about 1.5 years (see Fig. 14 and Tables 6 and 7), suggesting that density-dependent growth affects mainly juveniles and young adults."},{"index":3,"size":59,"text":"The existence of marked seasonal growth oscillations (not considered in the simulation model) in anchoveta also show this fish to behave, with regard to temperature, just as other fish do (see Fig. 16), i.e., length growth rate is reduced when temperature is low even if the cold period corresponds to a major upwelling and hence to increased food availability."},{"index":4,"size":18,"text":"The main problem thus remaining is the identification of the cause(s) for the densitydependent growth changes presented here."},{"index":5,"size":132,"text":"Clearly, the \"holistic\" model presented here as Fig. 8 and equation ( 5) cannot be used to distinguish potential causesa l l the model can do is roughly quantify their combined effect. A \"reductionistic\" model such as the one presented in Fig. 13 and in the preceding section, on the other hand, can be used to identify such factors and their effects, both singly and in the aggregate. However, translating the assumptions and results of a given model into those of another model is not straightforward, and is indeed rather akin to translating puns from one language to the other. Thus, to take the analogy one step further, one could provide a \"translation\" of the model in Fig. 8 in terms of the model in Fig. 13. This would read as follows:"},{"index":6,"size":61,"text":"Fish stop growing when the metabolic costs of all their activities invoke an oxygen demand which matches the average oxygen supply through their gills. Hence, factors such e.g. as reduced food competition, improved food quality or a shift from filter to particulate feeding, by reducing their overall metabolism, increases the body weight at which oxygen supply per unit weight becomes limiting."},{"index":7,"size":53,"text":"Thus, one should expect the two models to give similar results only in terms of combined effectsand here indeed the agreement is remarkable, as the holistic and reductionistic models give results (i.e., aggregate percent changes of growth-enhancing factors needed for the transition from \"1950\" to \" 1980\" conditions of 16 and 15%, respectively."},{"index":8,"size":44,"text":"Previous investigations by Tsukayama (1965Tsukayama ( ,1966) ) and Mendiola (1979) have shown that the fish of the southern stock of anchoveta which feed on zooplankton more than the mainly planktivorous fish of the northern/central stock, also have coarser gill rakers and shorter guts."},{"index":9,"size":132,"text":"If a major shift by anchovetas of the northern/central stock toward particulate feeding of zooplankton occurred, this should be reflected, therefore, in their now having coarser gill rakers and shorter guts than was previously estimated by the authors cited above, i.e.! the fish should have become \"southernized, as is also expressed by the increase of their mawnum size. It is suggested here that the analysis of the large database presently being standardized at IMARPE on anchoveta stomach contents, coverin the years 1954 to the present, be complemented by a f comparison of gill raker counts and gut ength analysis of fish taken at the sites previously sampled by the two authors cited above. Such comparison would indeed provide a clear-cut test of the postulated \"southernization\" of the northern/central stock of Peruvian anchoveta."}]},{"head":"Introduction","index":106,"paragraphs":[{"index":1,"size":83,"text":"Numerous previous estimates of the biomass of Peruvian anchoveta exist and may be found in the form of time series in various pamphlets, reports and/or books on topics ranging from the economics of soya bean exports to oceanographic forecasting. Yet these time series not being structured by size (or age group) nor having a time scale sufficiently small to reflect the rapid changes of biomass experienced by the anchoveta, are largely useless for research, e.g., as element of models for predicting fisheries yields."},{"index":2,"size":57,"text":"We shall present here biomass estimates that are structured by length (1 cm classes) and which have a monthly time scale. Also, the time span covered shall extend from 1953 to 1981, i.e., covering the Peruvian anchoveta fishery from its beginning to a period shortly preceeding its total demise at the height of the 1982-1983 El Niiio."},{"index":3,"size":84,"text":"The estimates we present do not call into question the overall magnitude of previous estimates of biomass as obtained previously, mainly through acoustic surveys. The reason for this is that the model we used -Virtual Population Analysis (VPA)and the specific fashion we implemented it, i.e., adding fish removed by major predators to fish taken by the fishery, necessitated estimates of residual mortality (i.e., that part of natural mortality not caused by major predators (here coded \"MoU) which we had no way of estimating independently."},{"index":4,"size":40,"text":"Therefore, we have used available estimates of biomass for calibration, i.e., to obtain estimates of Mo, with the result than our analysis, rather than challenging the existing anchoveta estimates, actually gives them additional coherence by integrating them with other information."}]},{"head":"Materials and Methods","index":107,"paragraphs":[]},{"head":"Computation of Total Withdrawals and Catch-at-Length Data","index":108,"paragraphs":[]},{"head":"#","index":109,"paragraphs":[{"index":1,"size":175,"text":"The bulk of the material used here is represented by the nominal catch and catch composition data in Tables 1 to 30 of Tsukayama and Palomares (this vol.). To account for unrecorded catches (Castillo and Mendo, this vol.), we have multiplied all monthly nominal catches by 1.2. Much could be saidpro or contra this value; however, it has little impact on VPA estimates of biomass given that the l & values estimated from the independent biomass estimates are inversely related to such multiplicative factor. The estimated quantities of anchoveta consumed monthly by cormorants, boobies and pelicans (from Muck and Pauly, this vol.) by bonito (from Pauly, Palomares and Gayanilo, this vol.), and by two species of seals (from Muck and Fuentes, this vol.) were then added to the corrected catches of the fishery to obtain total withdrawals on a monthly basis (Table 1). We shall refer to these total withdrawals as \"catches\" and use the terms \"catch-at-length\" when referring to the monthly \"number-of-fish-withdrawn-by-length-class, except in cases where fishery catches must be explicitly differentiated frompredatory losses."},{"index":2,"size":71,"text":"The detailed analyses on each of the major anchoveta predators considered here shows that they consumed anchoveta of sizes largely matching those of the fishery (see Jordan 1959;Figs. 1-3 in Muck and Pauly, this vol.;Fig. 7 in Pauly, Vildoso, et al., this vol.;Fig. 1 in Muck and Fuentes, this vol.), thus justifying the approach of pooling the withdrawals by the fishery with those of the birds, the bonito and the seals."},{"index":3,"size":239,"text":"The % catch composition data in Tables 1 to 30 of Tsukayama and Palomares (this vol.) were raised to this catch (i.e., total withdrawals) by means of raising factors (R.F.) computed, for each month (i) separately, from where Ws(i) is the weight of the % composition data in month (i). The values of Ws(i) were obtained from where Wij is the mean weight of fish in class j of sample (i.e., month) (i), n the total number of length classes in that sample, and fi' the % frequency of class (i) in sample (i). The values of Wj L were estimated, given a length weig t relationship of the form from where Ljl and Lj2 are the lower and upper limits of class (j), and which provides an unbiased estimate of the mean weight of fish in a given length class (Beyer 1987). All computations were performed with the value of b in equation ( 4) set equal to 3 (i.e., isometric growth and the appropriate values of a = c.f.1100, see Tsukayama and Palomares, Tables 1 to 30 and text). This procedure, implemented here through the ELEFAN III program was applied 360 times, i.e., to all % length-frequency spples in Tsukayama and Palomares (this vol.) as well as to the samples interpolated to fill in gaps (see below). The resulting matrix of monthly catch at length data, covering the years 1953 to 1982 was used for all VPAs."},{"index":4,"size":55,"text":"Brief Description of Length-Structured VPA Beverton and Holt (1957) showed that the catch (Ci) from population during a unit time period (i) is equal to the product of the population size at the beginning of the time period (Ni) times the fraction of the deaths caused by fishing, times the fraction of total deaths, or"},{"index":5,"size":27,"text":"where Fi is the fishing mortality in the ith period, M is the natural mortality, generally assumed constants for all period and Zi = Fi + M."},{"index":6,"size":169,"text":"The version of Beverton and Holt's catch equation which has become most widely used for stock assessment purposes, however, is also written which is the equation in Gulland's (1965) VPA and which can be derived from (5) by substituting for Ni the relationship Given values of Ci and an estimate of M (here: \"Mow) equation ( 7) can be used to estimate (retroactively) the size of past cohorts (i.e., of groups of fish born at the same time and exposed to the same mortalities throughout their lives), given an estimate of Ni + 1 from which to start the computation. Such estimate of Ni + 1 (expressing the last population size a cohort had before it went extinct) are usually called \"terminal populations\" (Nt).Values of Nt can be obtained from where Ct is the terminal catch (i.e., the last catch taken from a cohort before it went extinct) and Ft is the terminal fishing mortality, i.e., the fishing (here inclusive of predation) pressure that generated Ct (Mesnil1980; Pauly 1984)."},{"index":7,"size":182,"text":"The feature of VPA that is most important in the context of this contribution is that, given a high fishing pressure, estimates of population size obtained by repeated application of equations ( 6) or (7) tend to rapidly converge toward their true value, and hence usually provide, given reasonable estimates of M, very reliable estimates of recruitment (Pope 1972;Pauly 1984). Moreover, the speed of convergence from the guessed values of Nt (i.e., values of Nt based on guessed values of Ft) toward accurate values of Ni is a function of the ratio of F to M. That is, the higher the proportion of the cohort is which ends up being caught by the fishery (here: and eaten by birds, bonito or seals), the more reliable will the population estimates be. This is the reason why we have here, through the contributions of Muck and Pauly; Pauly, Palomares and Gayanilo; Muck and Fuentes (this vol.) accounted explicitly for the anchoveta consumed by birds, bonito and seals, respectively, and thus left Mo low, rather than replace predation by a higher constant estimate of M."},{"index":8,"size":85,"text":"Three forms of VPA, all included in the ELEFAN 111 program may be distinguished (Pauly and Tsukayama 1983): i) VPA I, which is the version originally proposed by Gulland (1965) and which Pope (1972)  VPA 111 is a version of VPA I performed on \"cohorts\" obtained by superimposing growth curves, drawn at monthly intervals, onto a set of catch-at-length data, the catch pertaining to each \"cohort\" and month being simply that part of the monthly catches contained between two adjacent growth curves (see Fig. 1)."},{"index":9,"size":85,"text":"For such cohorts to really consist of fish recruited at the same time, the growth curves used for \"slicing up\" a cohort must be obviously as close to the true growth curve of that cohort as possible. This, among other things, makes it imperative that a seasonally oscillating growth curve be used since, as shown in Pauly and Ingles (1981) and Pauly (1982), virtually all natural fish stocks, including those occurring in tropical waters, display seasonally oscillating growth (see also Palomares et al., this vol.)."},{"index":10,"size":70,"text":"In reality, not all fish of a given cohort have the same growth parameters, however, and it can be expected that some fish will \"leave their cohort\" because they grow either faster or slower than predicted by the mean cohort growth curve. Such differences in growth rate should here have the effect of artificially increasing the autocorrelation between estimates of recruitment (see Mendelsohn and Mendo, this vol.; Pauly, this vol.)."},{"index":11,"size":97,"text":"The growth curves used here to \"slice cohorts\" were based on the seasonally oscillating version of the von Bertalanffy Growth Function (VBGF) presented by Pauly and Gaschutz (1979) i.e., where Lt is the length at age t, L, the asymptotic length, K a growth constant, to the \"age\" at which length is zero if the fish always grew according to the equation, C is a dimensionless constant expressing the amplitude of the growth oscillations and ts is the time (with respect to t = 0) at the beginning of a sinusoidal growth oscillation of period one year."},{"index":12,"size":100,"text":"For practical purposes the estimation of ts was replaced by the estimation of a Winter Point (WP), defined as which expresses (as a fraction of the year) the time during which growth is slowest. It should be mentioned here that the ELEFAN programs, being based on length-frequency data (rather than length-at-age data) do not allow for the estimation, nor require estimates of to, hence of absolute ages (see Palomares et al., this vol.); all \"ages\" used internally by the programs are relative ages, expressed in relation to an arbitrary birthdate that is set internally and not output by the program."},{"index":13,"size":103,"text":"The VPA 111 routines of ELEFAN III were applied to the available catch-at-length data using the growth parameters given in Table 2. The small year-to-year differences in the values of showing monthly catch-at-length data (not to scale) and one of the many monthly cohorts which can be superimposed on these data. The upper two panels show the population and fishing mortality estimate pertaining to this cohort, along with its biomass (rightmost panel). The population estimate pertaining to month 1 is an expression of recruitment of fish of length Lr (here 3.75 cm, i.e., the lower limit of the class with midlength 4.25 cm)."}]},{"head":"VIRTUAL POPULATION ANALYSIS I11","index":110,"paragraphs":[{"index":1,"size":72,"text":"these parameters caused a slight overlap of some \"cohorts\" (i.e., some of the catch data were used twice), and small gaps (i.e., some of the catch data were not included in any cohort). This source of error could have been avoided by using the same growth parameters throughout. This, however, would have caused a large bias, given the strong positive trend in anchoveta growth performance documented in Palomares et al. (this vol.)."}]},{"head":"Interpolation of Missing Sets of Monthly Size-Composition Data","index":111,"paragraphs":[{"index":1,"size":201,"text":"Although the monthly catch data (in weight) used for the present analyses are complete, the monthly size composition data needed to derive monthly catch-at-length data are not (see Tables 1 to 30 in Tsukayama and Pauly, this vol.). Since uninterrupted series of catch-at-length data are needed for W A 111, we have linearly interpolated size-frequency distributions where gaps appeared in the original data (except for January-October 1953, where we have used the corresponding values for 1954). We believe this approach had no major detrimental impact on our results for three reasons: i) linearly interpolated % length-frequency samples are very similar to \"real\" samples (see Fig. 2) if only because the overall shape of such samples is determined mainly by continuous, rather \"smooth\" processes (i.e., growth and mortality); ii) W A estimates of F and population size in a given time interval are not predominantly determined by the catch composition in that very interval, but by the catch in that interval and the catches and catch composition in previous intervals (VPA runs backwards!). Hence whatever error is introduced by interpolating will be spread over several intervals, and its absolute impact on mortality and population estimates pertaining to a given interval thus reduced."},{"index":2,"size":58,"text":"iii) catch-at-length data as used in W A 111 are estimated from both length composition data and catch data in weight. While the former were interpolated in some cases, the latter were always \"real\" and hence the overall number of fish caught in a given month tended to be reasonably approximated, even though their size composition was interpolated."}]},{"head":"Estimation of_Monthly Recruitment and Biomass","index":112,"paragraphs":[{"index":1,"size":97,"text":"Monthly population estimates per length class were obtained by regrouping monthly values -~f Ni obtained on a cohort basis into regular class intervals (the same intervals as those in which &e?ziii:li &&+,+%re originally grouped). The population estimates in the smallest class considered here (3.75 to 4.75 cm) are here defined as \"recruitment\" (of fish with mean length 4.25 cm).  The population estimates (in numbers), by length class, were then multiplied, for each month and length class separately by the appropriate value mean weight values (see equation 4) to obtain monthly population biomass by length class (Fig. 3)."}]},{"head":"VIRTUAL POPULATION ANALYSIS I11","index":113,"paragraphs":[{"index":1,"size":52,"text":"All analyses were run with Ft = 2(y-1). It is recalled that these values (a) pertain to Z-& (i.e:, to the sum of true fishing mortality plus all sources of natural mortality represented by distinct predators and (b) that values of Ft have little impact on final results, especially on recruitment estimates."}]},{"head":"Estimation of Natural and Fishing Mortalities","index":114,"paragraphs":[{"index":1,"size":104,"text":"The natural mortality (M ) exerted by any given predator (p) with anchoveta consumption R (Cp(i)) was estimated, for eac month (i) from Mp(i) = anchoveta consumption (Cp(i))/anchoveta biomass (i) ... 12) Similarly, fishing mortality (F) was estimated from Fi = anchoveta fishery catch (i)/anchoveta biomass (i) ... 13) Thus, all estimates of mortality presented here pertain to weights, not numbers. mote, however, that computations of number-based mortality values are possible, based on the data presented in this volume.] From these data, total mortality (2) can be estimated from where m is the number of predator groups considered here (i.e., birds, bonito and seals)."}]},{"head":"Results and Discussion","index":115,"paragraphs":[]},{"head":"Estimate of Mo and Their Implications","index":116,"paragraphs":[{"index":1,"size":158,"text":"Table 3 gives the values of Mo obtained iteratively, i.e., by changing values of & until biomass were obtained which closely matched the independent biomass estimates in that same table. The same values of Mo are also plotted as time series in Fig. 4. As might be seen, the estimates of biomass in Table 3 force us to assume d.lat &, which took values of about 2y-1 in the 1960s, dropped to about OSy-1 in 1975, rapidly increased to about 4y-1 in 1976 and stayed at this high level until the late 1970s. (We shall present below evidence suggesting that the natural mortality of anchoveta did decrease in the early 1970s, as illustrated on Fig. 4.) The value of Mo . = 4y-1 for the late 1970s suggest that an-hportant predator, ignored in our analyses, managed to continue consuming a large amount of anchoveta in spite of their very reduced biomasses in the late 1970s (Pauly, this vol.)."}]},{"head":"Estimates of Anchoveta Biomass and Recruitment, 1953 to 1981","index":117,"paragraphs":[{"index":1,"size":35,"text":"Tables 4 to 33 present the key results of our analyses, i.e., the monthly recruitment estimates and the monthly biomasses, by length group, respectively, obtained through the VPA 111 routine of the ELEFAN III program."},{"index":2,"size":79,"text":"Fig. 5, based on Tables 5 to 33, shows the monthly sums of biomass over all length groups, for 1953 to 1982. One part of this time series, covering the years 1963 to 1979 illustrates the match of our biomass estimates to the independent biomass estimates used to calibrate the VPA (i.e., to esimate Mo). That part of the time series presents no surprise, except perhaps for the fact  -' ; fmm Fig. 7 in Palomares et al. (thisvol.)."},{"index":3,"size":41,"text":"dLonthly means, as read off eye fitted line in Fig. 4; values used for VPA 111 L i r l y interpolated in steps of 3 months.  (1981) a~~ values adjusted to pertain only to the region between 4 and 14's."},{"index":4,"size":28,"text":"b ~e e also Johannesson and Robles (1977).  that seasonal oscillations and other within-year changes appear to be far more intense and rapid than had previously be assumed."},{"index":5,"size":80,"text":"Little comments are needed for the parts covering 1980 to 1982, except perhaps that the decline of the biomass to a very low level prior to the onset of the 1982-1983 El Niiio is, in part, an artifact due to the absence of catches and consumption estimates to \"feed into\" the VPA in 1983 to estimate the 1982 biomasses. For this reason, we have omitted the year 1982 from the detailed results on biomasses given in Tables 5 to 33."},{"index":6,"size":204,"text":"The anchoveta biomasses estimated for the 10 years preceding the first available independent standing stock estimates, i.e., the years 1953 to 1962, are interesting in that they are lower than expected (see Table 3 in Muck and Pauly, this vol.). Interestingly, these biomass estimates suggests that the anchoveta suffered from the 1957 El Niiio almost as much as from that of 1971-1972. This indeed would explain the massive bird mortalities recorded from that period (see Tovar et al., this vol.). It should be noted, however, that this point is, in part at least, a circular argument, since we used, among other things, bird population data to estimate bird Year Fig. 6. Time series of recruitment (of fish ranging from 3.75 to 4.75 cm, slightly less than 3 months old) into the anchoveta stock, January 1953 to mid-1982. Above: monthly recruitment, showing increasing variability, from the late 1950s to 1970, probably due to increasing fishing pressure and leading to recruitment collapse in early 1971, prior to the onset of the 1971-1972 El N ~O .  Below: smoothed data (using a 12-month running average), showing that the 1960s, which saw the buildup of the fishery, may have been a period of exceptionally and steadily high recruitment."},{"index":7,"size":27,"text":"anchoveta consumption, and since such consumption, at a time when the fishery was just starting, should have had a great impact on the VPA estimates of biomass."},{"index":8,"size":51,"text":"Fig. 6, based on Table 4, shows two time series of anchoveta recruitment: one consists of monthly estimates of the number of young fish (of 3.75-4.75 cm, i.e., about 3 months old) entering the fishery, the other illustrating the same data, but as 12 months' running average to show interyear changes."},{"index":9,"size":190,"text":"As might be seen, these time series closely resemble the time series of biomass, which is not surprising in a fish with a short lifespan. Important aspects of the time series in Fig. 6 are: i) the increased within year variability of recruitment from the late 1950s to the late 1960s, presumably an effect of fishing, and ii) the fact that the major recruitment collapse of 1971 appears to have occurred before the onset of the I971 -I 972 El Niiio (see also Mendelsohn and Mendo, this vol.) These two aspects, already apparent in the earlier analysis of Pauly and Tsukayama (1983) imply that much of what has been written about the 2971-1972 El Niao being the cause of the collapse of the fishery in 1972 is probably wrong. Fig. 7 shows time series of anchoveta fishing mortality, both of a monthly basis and smoothed (12 months' running average) to allow the interyear trend to become fully visible. As might be seen, fishing mortality increased steadily through the 1950s and 1960s, then fluctuated violently through the 1970s, mainly due to successive El Niiio events and associated closures of the fishery. "}]},{"head":"Oct","index":118,"paragraphs":[{"index":1,"size":50,"text":"Nov Dec a ~o n t h beyond April 1982 not included due to their decreasing reliability; the values for early 1982 given here to allow computation of loge (Rec./Eggs) with egg production pertaining to 3-4 months prior to recruitment (see Mendelsohn and Mendo, this vol. and Pauly, this vol.). "}]},{"head":"Year","index":119,"paragraphs":[]},{"head":"Estimation of Anchoveta Mortality Caused by Various Predators","index":120,"paragraphs":[{"index":1,"size":185,"text":"Figs. 8A, 8B and 8C show that part of natural mortality is attributable to the birds, the bonito and the seals, respectively. As might be seen, birds and bonito had a roughly similar impact on the anchoveta stock with peaks of 0.3-0.4~-1 in the late 1950s and negligible impact thereafter. The impact of the seals on the anchoveta stock, on the other hand, was negligible throughout the whole period considered here. Mortality caused by bonito (Sarda chiliensis) (note similarity of estimates to those pertaining to the birds). C. Mortality caused by sea lion (mainly) and fur seals. Note that the scale, ranging from zero to 0.005 implies a negligible pinniped impact on anchoveta natural mortality due to all causes. Note that shape of curve resembles that of Fig. 4, except for small \"bumps\" in the 1950s, corresponding to periods of bud and bonito abundances. Important here is that decline of M from 1971 to 1974 is confirmed by analysis of tagging/recapture data performed by Malaga and Armstrong (MS). Note also that little support is available for low M estimates (B & S) in Schaefer (1967)."},{"index":2,"size":107,"text":"Fig. 8D, finally, shows a time series of natural mortality as a whole, i.e., combining the predators considered here and the estimates of Mo from Fig. 4. As might be seen from the overall shape of the curve, it is the estimates of Mo which, throughout determine the overall level of natural mortality, clearly illustrating that the predators explicitly considered here do not explain but a small fraction of overall natural mortality. Interestingly, the drop of M observed in 1973-1974 is closely matched, albeit at higher level by a drop of M-estimates based on tagging data and reported in Table 1 1 of Malaga and Armstrong (MS)."},{"index":3,"size":80,"text":"Fig. 8D also shows that the values of M reported in Schaefer (1967) and based on his simulations (\"S\") and on a personal communication of T. Burd (\"B\") appear low compared with our estimates. We note, finally that values of M computed from the empirical equation of Pauly (1980) ranged between 1.3 and 1.6, and thus were intermediate for the 1960s at least, between the estimates reported in Schaefer (1967) and the Mo values estimated by backcalibration of VPA estimates."}]},{"head":"Overall Assessment of Results","index":121,"paragraphs":[{"index":1,"size":34,"text":"Overall, our results both confirm and expend on previous results, and provide biomass and recruitment estimates for use in deriving further quantities (see, e.g., Pauly and Soriano, this vol.; Mendelsohn and Mendo, this vol.)."},{"index":2,"size":64,"text":"Both the data and the software we used have serious liabilities associated with them: the former had gaps which were \"filled\" using interpolation procedures which might not have all the required properties, while the latter is structured around assumptions about the growth of fish (i.e., that all fish in a given cohort have the same growth parameters) which are known not to be true."},{"index":3,"size":37,"text":"Improved estimates of biomass and recruitment may thus result from more sophisticated interpolation methods and by using a length-structured VPA model not based on the assumption that all fish of a cohort have the same growth parameters."},{"index":4,"size":47,"text":"However, such improvement might be minor compared with what we considered to be the main problem with our analysis, i.e., the fact that we did not account through the explicit inclusion of predators of as large a fraction of overall natural mortality as we would have liked."},{"index":5,"size":40,"text":"This is due to the fact that, following earlier authors, we believed the birds to be the key anchoveta predators in the Peru current system (see Pauly and Tsukayama, this vol.; Pauly this vol.). We have here been proven wrong.      "}]},{"head":"Introduction","index":122,"paragraphs":[{"index":1,"size":47,"text":"The reproductive output of the Peruvian anchoveta (Engraulis ringens) has been earlier reported either in the form of relative egg densities, as estimated from egg surveys, or computed through the \"egg-production method, applied once to data from the 198 1 peak reproductive season (Santander et al. 1984)."},{"index":2,"size":62,"text":"In this contribution, the literature on anchoveta reproduction is reviewed. Available information is used to estimate parameters used subsequently to compute monthly egg output by the stock of anchoveta from 4 to 140s off Peru, from January 1953 to December 1982, based on monthly biomass data derived by Pauly, Palomares and Gayanilo (this vol.) and ancillary data presented in detail further below."}]},{"head":"Materials and Methods","index":123,"paragraphs":[]},{"head":"Basic Model Structure","index":124,"paragraphs":[{"index":1,"size":32,"text":"Reproductive output (RO) in a given month (i = 1-360) can be viewed as the ultimate product of different factors summed up over all length classes, i.e., * ICLARM Contribution No. 382."},{"index":2,"size":96,"text":"whereBijis the biomass of male and female anchoveta in length class j, Pij is the fraction of mature fishes in length class j, Aj is a size-specific empirical factor relating anchoveta fecundity per unit weight to length j, Sk is the average number of times a female anchoveta spawns per month, with k = 1 (January) to k = 12 (December), PFj is the fraction of females of length j in the population and RF is the relative batch fecundity (eggslbody weight in g) of a female anchoveta whose value of A' is equal to unity."},{"index":3,"size":15,"text":"The following sections ocument each of the variables and constant used in implementing equation (1)."},{"index":4,"size":1,"text":"a"}]},{"head":"Biomass by Length Class","index":125,"paragraphs":[{"index":1,"size":54,"text":"The estimates of monthly biomass by length class used here were taken from Pauly, Palomares and Gayanilo (this vol.). The median lengths considered ranged from 4.25 to 20.25 cm. As will be seen below, only fish above 12 cm contributed significantly to egg production. However, all computations were performed with all length classes included."}]},{"head":"The Fraction of Mature Fish as Function of Length","index":126,"paragraphs":[{"index":1,"size":117,"text":"Fig. 1 presents maturation ogives for anchoveta, based on data gathered by various authors in the 1950s and 1960s. As might be seen, these data suggest that under average condition anchoveta have a mean total length at first maturity of L50 = 14 cm. The mean maturation range (i.e., the length where 75% of the anchoveta have reached first maturity minus the length where 25% have) is 1.8 cm. As will be noted from Fig. 1 and Table 1, the latter estimate is not very reliable, being based on three authors whose work suggest a mean range of about 2.9 cm, and another group of three authors whose work suggest a mean range of about 1.1 cm."},{"index":2,"size":86,"text":"Tsilkayama and Alvarez (1981) showed that in \"warmer years\" (i.e., in years during El Niiro events), mature fish were usually smaller than during cold years. Their figure has been redrawn here as Fig. 2, and temperatures added, together with estimates of L50 based on the median of the class immediately preceding the model classin analogy with length-converted catch curves, where the length class immediately to the left of the sample mode usually provides reasonable estimate of mean size at first capture (see Ingles and Pauly 1984)."},{"index":3,"size":65,"text":"The mean of these rough estimates of L50 is, in colder years indeed higher than the mean of warmer years (Table 1). Moreover, the mean of the 10 values of L50 in Fig. 1, corresponding to an intermediate mean temperature, fits neatly between the points for the \"cold\" and \"warm\" years (Fig. 3), confirming that the estimates of L50 in Fig. 2 are indeed reasonable."},{"index":4,"size":174,"text":"These 3 sets of averages, representing a total of 25 pairs of L50 and temperature values, along with the mean maturation range of 1.8 cm provide the key elements of our model for estimation of the factor Pij in equation ( 1). This model is structured around the logistic equation, which generates curves resembling the graphs in the lower panels of Fig. 1, and has the form Given a knowledge of L50 and L25 (or L75) as obtained from the relationships in Fig. 3, the parameters rm of equation ( 2 Table 1. Summary of data on anchoveta maturation as a function of length and temperature (see also Figs. 1  and 2). Pig. 2. Relationship between two groups of estlmates of mean length at fist maturity (L as approximated by position of arrows, see texpand environmental temperature. Note that L50 values during \"warm years\" are lower than durlng cold years. The letters K to Y refer to the rows of Table 1 (adapted from Tsukayama and Alvarez 1981 and I. Tsukayarna, IMARPE, pers. wmm.)."}]},{"head":"I *C range used for modelling","index":127,"paragraphs":[{"index":1,"size":43,"text":"Sea surface temperature PC) Fig. 3. Model used to estimate values of L25, L50 and L75 in anchoveta for temperature between 14 and 24OC. The points refer to the means L50 values in Figs. 1 and 2 (see also Table 1 and text)."}]},{"head":"Relative Fecundity and Spawning Frequency as a Function of Anchoveta Size","index":128,"paragraphs":[{"index":1,"size":115,"text":"Not all anchoveta are equal. Major differences occur in the relative fecundity of mature anchoveta as well as frequency of spawning (Parrish et al. 1986). As no data on these sizespecific differences are available for Peruvian anchoveta, data presented by the abovementioned authors have been assembled and analyzed (see Table 2). These data yielded values of the factor Aj in equation ( 1) ranging fiom near zero for very small fish (which are likely to be immature in any case), 1 in fish of 26 g (the mean weight of Peruvian anchoveta used by Santander et al. 1984 for estimating batch fecundity and spawning frequency) and about 6 in anchoveta of 50 g (Table 3)."},{"index":2,"size":63,"text":"As presented and used here, these values of the factor Aj account for the fact that larger anchoveta spawn more frequently, have a longer spawning season and produce more eggs per spawning than smaller ones, a feature also reported for numerous fishes other than engraulids (see Parrish et al. 1986 and references therein). a Adapted from Table 3 in Parrish et al. (1986)."},{"index":3,"size":20,"text":"Estimated relationship between relative fecundity (R.F. and body weight is R.F. = 0.582 w ~. ~~ (r2 = 0.981,2 d.f.). "}]},{"head":"Table 3. Multiplication factor (\"","index":129,"paragraphs":[]},{"head":"The Seasonality of Anchoveta Spawning","index":130,"paragraphs":[{"index":1,"size":176,"text":"Numerous authors give accounts of the seasonal_pattern of maturation of anchoveta and its link to spawning (see e.g., Simpson and Gil 1967;Vildoso and Alegre 1969;Santander and Castillo 1969). Here, graphs showing seasonal patterns in the percentage of mature females of anchoveta caught off Chimbote and Callao and presented by Jordan (1980) were used to relate the monthly change in the fraction of mature fish in the stock as a whole to the fraction mature in September, i.e., of the only month for which an estimate of spawning frequency is available (see Table 4). This procedure allows scaling of every month of the year in terms of the September spawning frequency, leading to an estimate of annual number of spawnings (for females of about 26 g) of 24 per year. This value is sufficiently close to the estimate of 20 times per year in northern anchovy (Hunter and Leong 1981) to be acceptable. The estimated mean number of spawnings per month in Table 4 shall thus be used here as estimates of Sk in equation ( 1)."}]},{"head":"The Fraction of Female Anchoveta in the Population","index":131,"paragraphs":[{"index":1,"size":156,"text":"Data are available from three different authors showing that the proportion of females in anchoveta catch samples increases rapidly from about 0.5 at 12 cm ('I'L) to about unity near 17-18 cm (Fig. 4). Moreover tho data from two of these authors (Clark 1954;Miirano 1958) contradict Jordan's contention that \"the sex ratio is moderately in favor of females in the small fish\" (Jordan 1980, based on Jordan 1959 and see Fig. 4). Santander et al. (1984) show that the proportion of females in anchoveta of mean weight 26 g may range from 0.1 to 0.9. Thus, small samples such as will occur when fish of very small sizes are considered separately, may suggest trends that are difficult to explain. A gradual increase of the proportion of females such as suggested by the line superimposed on the data points of Fig. 4 would be, on the other hand, very As read off Fig. 3 in Jordan (1980)."},{"index":2,"size":53,"text":"As computed with reference to September value. d ~a s e d on Santander et al. (1984) who report that 16.04% of the females sampled in September 1981 had been spaw* ing during the previous 24 hours. (The values in this column correspond to the values of Sk in equation 1, see text)."},{"index":3,"size":31,"text":"Note that 24.2 spawnings per year per female (of 26 g, see text) is close to the value of 20 estimated by Hunter and Leong (1981) for northern anchovy (E. mordax)."},{"index":4,"size":92,"text":"A Clark (1954,Table  III) Miiiano  easy to explain, as the result of a slightly lower mortality among the females, ultimately linked to slight differences in growth parameters. The eye-fitted curve in Fig. 4 assumes that the sex ratio of anchoveta embryos is 1: 1, as should be assumed when no information is available suggesting otherwise (Conover and Heins 1987). The curve was used here to provide estimates of the mean proportion of females in the anchoveta population by length class, i.e., of the parameter PFj in equation (I), (see also Table 3)."}]},{"head":"The Batch Fecundity of Female Anchoveta","index":132,"paragraphs":[{"index":1,"size":132,"text":"As mentioned above, only one estimate of batch fecundity in the Peruvian anchoveta is presently available (Santander et al. 1984). It pertains to fish with a mean weight of 25.84 g and is of 15,401 eggs per female. Expressed as relative fecundity, this amounts to 596 eggdg female, similar to values for the northern anchovy, where relative batch fecundity values ranging between 556 and 720 have been reported (Hunter and Goldberg 1980;Lyczkowsky-Laroche and Richardson 1980) The reader will notice that the form of equation ( 1) makes monthly egg output of anchoveta directly proportional to the relative fecundity (RF) value used here. Hence, use of another value of RF would imply a rather straightforward rescaling of final results but no change in the appearance of any of the graphs presenting these results."}]},{"head":"Computation of Reproductive Output","index":133,"paragraphs":[{"index":1,"size":60,"text":"A BASIC program available on request from the authors was used to compute, for each ' length class (from ML = 4.25 to ML = 20.25 cm) and every month (from January 1953 to December 1982) multiplicators which were subsequently applied to the biomasses by length classes in Pauly, Palornares and Gayanilo (this vol.) to give estimates of egg production."}]},{"head":"Results and Discussion","index":134,"paragraphs":[{"index":1,"size":80,"text":"Fig. 5 shows the total biomass (from Pauly, Palomares and Gayanilo, this vol.), spawning biomass (females and males) and mature female biomass, for every month from January 1953 to December 1982 (see also Tables 5 and 6). All of these display strong seasonal and interannual oscillations, reflective of spawningfrecruitrnent seasonality and of major events such as the 197 l/l972 collapse of the fishery.  On the other hand, our estimate of spawning stock size for Au st1September 1981 of 1.14 t"}]},{"head":"P","index":135,"paragraphs":[{"index":1,"size":250,"text":"x 106, is extremely close to the independent estimate of 1.2 t x 10 (with 95% cod. interval = 56.8%) obtained by Santander et al. (1984) for the area between 4 and 140s (see Table 5, footnote [a] and Fig. 5). This extremely good match between two estimates that were obtained based on completely independent data sets, assumptions and models is extremely encouraging, as it suggests that our approach for estimating past spawning stock size may be realistic. Tsukayama and Alvarez (198 1) reported a range of 0.067 to 0.244 and an overall mean of 0.133 for the ratio of spawning stock to total anchoveta biomass, for the period 1964 to 1978. Our results (Fig. 6), expressed on a finer temporal scale than those of Tsukayama and Alvarez (1981) not only have a greater range, but a higher overall mean of 0.254, due to our different definition of mature biomass (Tsukayama and Alvarez 1981 defined this as \"the biomass of all fish >14 cm\"; our definition is \"Bij.PijW, see equation 1). Fig. 7 presents our estimates of monthly anchoveta egg production, from January 1953 to December 1981 (1982 was excluded because the VPA III estimates of anchoveta biomasses for the last year of the series are questionable, see Pauly, Palomares and Gayanilo, this vol.). As might be seen, monthly egg production had, during the period covered, several peaks, notably in 1968,1970,1971 and 1974. However, these peaks of egg production did not lead to peak intermediate and an advanced stage, respectively."},{"index":2,"size":52,"text":"Year recruitment. In fact, the annual season of peak egg production (September-November) is usually not followed, a few months later, by a recruitment peak. Rather, it is the minor egg production peak occurring in February to April which usually leads to massive recruitment (see Mendelsohn and Mendo, this vol.; Pauly, this vol.)."},{"index":3,"size":36,"text":"Assessing the overall reliability of the egg production estimates in Fig. 7 and Table 7 is difficult. Assuming that the biomass estimates used here were reasonable, we think that the weakest point in our analysis are:"},{"index":4,"size":165,"text":"i) lack of data on size-specific egg production in E. ringens, and ii) lack of readily quantifiable evidence on temporal changes in the fecundity of anchoveta.  Alheit et al. (1983, Table 1, Figs. 1-4) presented data on the size-dependence of batch fecundity of E. ringens, based on fish samples from January 1970 to September 1981. Although the analysis presented by these authors is somewhat confused, the key result does emerge that larger anchoveta have higher batch fecundities than small ones. However, we could not use any of the regression lines expressing this relationship, because large engraulid females not only have higher batch fecundities, but also spawn more often than smaller ones. This is the reason why we have used a relationship between size and egg production based on data for E. mordax (see Table 2); Fig. 4 in Alheit et al. (1983) shows that E. ringens and E. mordax are similar enough, at least as far as their batch fecundity is concerned, to justify our_appro;ich."},{"index":5,"size":187,"text":"Temporal changes in anchoveta fecundity may be expected, given the density-dependent changes in growth and related parameters discussed in Palomares et al., this vol.). However, concrete evidence for such changes is scanty. Ware and Tsukayama (1981) write that \"in the winter of 1962, the anchovy population was of the order of 24 t x 106. According to Mifiano (1968, Fig. 5), the gonad wei ht of a 21-g % anchovy at that time averaged 1.1 g. By 1974, the stock declined to 4.5 t x 10 and the corresponding gonad weight of a 21-g anchovy was 1.7 g. This finding clearly contradicts the long standing assumption of classical theory that fecundity of marine fish is not affected by fluctuations in population size (Harris 1975)\" Alheit et al. (1983), commenting on their fecundity estimates suggested similarly that \"the only remarkable feature is the difference between the relatively low values for the seventies and the high values from 1981 \". They also noted that \"obviously, the fecundity values fiom the southern Peruvian anchovy stock are higher than in the central and northern stock batch fecundity increases with increasing latitude\"."},{"index":6,"size":44,"text":"Combining the temporal and the latitudinal trends mentioned above leads to the conclusion that the Peruvian anchoveta may be undergoing, with respect to its fecundity, the same process of \"southernization\" that is apparent with regard to its growth (see Palornares et al., this vol.)."},{"index":7,"size":42,"text":"We have not accounted for \"southernization\" when deriving our estimates of egg production for the period January 1953 to December 1981. Explicit consideration of this process would obviously have improved our estimates and this should be taken into account in future analyses."}]},{"head":"Introduction","index":136,"paragraphs":[{"index":1,"size":81,"text":"Numerous anchoveta egg surveys have been conducted since 1964 off Peru, and this paper presents ninety maps summarizing the results of all these surveys. A brief, preliminary investigation of the egg standing stock estimates based on these maps is presented in which egg standing stock, adjusted for the different sea surface temperatures during the different surveys, is related to anchoveta parent stock estimates (i.e., biomass of mature female and male anchoveta, as defined and computed by Pauly and Soriano, this vol.)."}]},{"head":"Materials and Methods","index":137,"paragraphs":[{"index":1,"size":43,"text":"This contribution is based on ninety surveys carried out between 1964and 1985. Between 1964and 1971, four surveys were performed every year, with usually about 120 samples taken on each cruise, covering much of the Peruvian coast, up to 120 nautical miles (nm) offshore."},{"index":2,"size":44,"text":"From July 1972 to 1985, two to four EUREKA surveys were performed per year, with about 350 samples taken during the 2-4 days of these surveys which are performed by commercial vessels operating all along the Peruvian coast (6-180S), up to 80-100 nm offshore."},{"index":3,"size":81,"text":"A Hensen net (175 cm long, 70 cm aperture diameter and 300 ,urn was used for all samplings, which consisted of raising the net from a depth of 50 m to the surface. Samples were fixed with 10% formaldehyde immediately after collection; separation and counting of anchoveta eggs were done at IMARPE. aThe author passed away on 25 March 1987. This contribution is based on an analysis of egg distribution maps she compiled and on her extensive notes m e Editors)."}]},{"head":"Estimation of Egg Standing Stock and Adjusted Egg Standing Stock","index":138,"paragraphs":[{"index":1,"size":56,"text":"The maps presented here were standardized and analyzed in a series of steps implemented such as to ensure greatest possible comparability between maps, as follows: i) the areas with positive egg records (i.e., 0-500 to >2,000 or >4,000) were estimated by planimetry, ignoring at first the fact that not all surveys covered the whole Peruvian coast;"},{"index":2,"size":150,"text":"ii) the surface areas estimated in (i) were multiplied with the midrange of the egg density classes to obtain egg standing stocks, e.g., with 250 eggs/m2 in the case of the class with limits 1 to 500 eggdm2. Ideally, the mean density of samples with counts >4,000 (or >2,000) should have been used to obtain proper factors for these two open-ended classes. Unfortunately, original data allowing computation of such means could be retrieved only for seven maps, all referring to the \">4,000\" density class (Table 1). The mean density estimated for this class from-the data in Table 1 was 5,013 eggdm2 and this value was used with all maps for which a density value for the \"A 000\" class was missing. For maps with \"2,000\" as highest class, a value of 3,000 eggs/m$ was used (i.e., loweq class limit + 1,000, similar to what was used for the \">2,000M class);"},{"index":3,"size":16,"text":"iii) standardization of maps for variable coverage of the Peruvian coast was performed in two steps:"},{"index":4,"size":302,"text":"a) within each 20 stretch of the coast (i.e., 2-4, 4-6, ..., 12-140S), simple extrapolations were used. Thus, if a survey had covered only half of the stretch from 10 to 120S, the estimated number of eggs within that stretch was multiplied with 2; b) then, using 49 maps representing surveys that had covered all that part of the Peruvian coast of interest here (4-140S), the mean percentage of total eggs in each 20 stretch of coast was estimated (see Table 2). These percentages were then used to convert estimates of egg standing stock for a given set of 20 stretches (ranging from 1 to 4) to an estimate pertaining to the whole coast. The assumption here is that any given map, with say few eggs between 4 and 100S, would also have had few eggs between 10 and 140s had the area also been surveyed. (Readers who disagree with this approach will be able to identify the maps for which this approach was used and delete those in their reanalysis.) The estimates of total egg numbers for the whole Peruvian coast obtained in this fashion are given in Table 3 where each row corresponds to a given month and in Table 4 where egg standing stock estimates for a few surveys and maps with the same reference month are given separately. iv) \"adjusting\" egg standing stock estimates refers here to dividing the standing stock estimate obtained in step (iii) by the estimated egg development time (itself a function of sea surface temperature (SST); see Table 3 and Pauly , this vol.). This procedure allows comparison of standing stock estimates fiom periods with different SST and hence different duration of the egg stage itself. Standing stock estimates adjusted in this fashion and expressed on a daily basis are given in Table 3."}]},{"head":"Results and Discussion","index":139,"paragraphs":[{"index":1,"size":140,"text":"Tables 3 and 4 summarize the key data derived from the maps. It might be noted that estimates of egg standing stock were generally higher in the 1960s than in the 1970s (means 78 x 1012 and 39 x 1012, respectively), as might be expected. Yet adjusted egg standing stocks and estimates of parent stock size in the corresponding month (see Table 3) did not directly c o d a t e with each other. Rather it is log (adjusted standing stocklparent stock) which appears to be linearly related to parent stock (Fig. 1). This suggests a rather strong, density-dependent effect of parent stock on egg survival (see also Fig. 2). Some implications of this finding are discussed in Pauly (this vol.).  4 for separate statistics on these tables); maps with numbers in boxes are those used in Table 2."},{"index":2,"size":29,"text":"b ~n cases where two maps refer to the same month(s), the dates given here refer to the start of the first survey and the end of the second."}]},{"head":"C","index":140,"paragraphs":[{"index":1,"size":34,"text":"Refers to Pemvian coast between 4 and 14's; see text for inter-and extrapolation methods. Where two map numbers are given, egg standing stock is the mean of these two maps (see also Table 4)."},{"index":2,"size":57,"text":"d~omputed using empirical equation in Pauly (this vol.) and sea surface temperatures in Pauly and Tsukayama (this vol.). e From Pauly and Soriano (this vol.).  Other findings based on anchoveta egg surveys conducted off Peru are presented in Santander (1981), Santander and Castillo (1981), Santander and Flores (1983), Santander and Tsukayama (1983), Santander andZuzunaga (1984) andIMARPE (1986)."},{"index":3,"size":87,"text":"However, the present contribution is the first in which a comparative analysis of the many egg surveys conducted off Peru has been performed. That these surveys, related to an independent data set (i.e., parent stock estimates), should lead to insights about cannibalism in anchoveta is rather gratifying. It is hoped that future analyses will lead to even more information being extracted from the maps and related data presented here, thus justifying post hoc the immense effort and resources that went in obtaining them.   I Map no. 17          "}]},{"head":"Introduction","index":141,"paragraphs":[{"index":1,"size":121,"text":"Of the many species of seabirds living along the Peruvian coast, on islands and \"points\" (i.e., small peninsulas), three, the cormorant (Phalacrocorax bougainvillii L., Family Phalacrocoracidae), the booby (Sula variegata Tschudi, Family Sulidae) and the pelican (Pelecanus thagus Molina, Family Pelecanidae) are the most conspicuous (Fig. 1). They are also the three species known as \"guano birds\" whose feces, known as \"guano\" (i.e., \"guano de islas\") were used in early, pre-Inca times, as fertilizer in agriculture. Garcilaso de la Vega (1539-1616) in his \"Royal Commentaries\" mentions that the Incas, following their conquest of the people along the Peru coast, explicitly protected the guano birds, and imposed the death sentence on those reported to have disturbed them, particularly during their reproductive season."},{"index":2,"size":74,"text":"Following the Spanish conquest, the agricultural infrastructure built andlor improved by the Incas was left to decay, and gold and silver mining developed. According to Maisch (1946) the extraction of accumulated guano started again only in the mid-1840s. This extraction was totally uncontrolled, and led to rapid depletion of the guano accumulations. Thus, for example, on the Chincha Islands, 95% of the accumulated 12 million tonnes of guano were removed between 1853 and 1913."},{"index":3,"size":62,"text":"The Compania Administradora del Guano (CAG) took over the administration of the guano islands and points in 1909, and later implemented the recommendation of Forbes (1914) who, after examination of all guano deposits along the Peruvian coast, suggested a series of protective measures such as the deployment of guards, the protection of nests and the rotation of the crews exploiting the deposits."},{"index":4,"size":304,"text":"Following Gamarra (1964aGamarra ( , 1964b)), six periods may be identified with regard to the guano resources and their exploitation: i) first period -Inca times, when the birds reached their climax, being totally protected; third periodimmediately following the Spanish conquest during which the knowledge of the usefulness of guano receded, and hence the birds ceased to be protected; fourth periodstarting about 1844 when, the importance of guano (for export markets) being apparent, the exploitation of guano expanded rapidly and included the use of dynamite to loosen up hardened layers of deposits without consideration to the birds. This started a regressive process which considerably reduced the bird populations which were at the time only viewed as interfering with the extraction activities; fifth periodstarting in 1909 when the CAG was created, with the specific purpose of regulating the guano industry. The CAG initiated measures based on scientific studies and its work has been described as the greatest effort ever made for the protection of any (wild) animal species in the world (see e.g., CAG 1946CAG , 1955CAG , 1958)); sixth periodstarting in the mid-1950s, this period saw the development of the anchoveta fishery and the decline of the anchoveta stocks which provide the bulk of the food of guano birds (Gamarra 1941). This period is characterized by a collapse of the bird populations and a large change in the relative abundance of the three species discussed here (Tovar 1983 and see below). This paper presents monthly population estimates for the years 1953 to 1982 from Isla Terra Lobos de Tierra in the north to Isla Ballestras in the south (Table 1, Fig. 2). Because of our limitations to islands and points not farther south than 140S, population estimates are not directly comparable with those in earlier papers (e.g., Tovar 1983), although the general trends certainly are."}]},{"head":"Materials and Methods","index":142,"paragraphs":[{"index":1,"size":181,"text":"Forms (maps) of islands and points, onto which guards had, for each month separately, graphed the distribution of each bird colony were used (e.g., Fig. 3). Managing the guano islands and points, and the bird colonies thereon from 1909 onwards was the duty of the CAG (see above). From 1964 to 1968, the role of the CAG was perfomed by the Corporation Nacional de   Fertilizantes (CONAFER). From 1970 to 1974, the management of the guano resources became the perogative of the Servicio Nacional de Fertilizantes (SENAFER); from 1976 to date, the guano resources have been managed by Pesca Peru Fertilizantes. All maps used here were obtained from the latter, which inherited the archives of its predecessors. Altogether, 10,080 maps were analyzed, covering the 360 months from January 1953 to December 1982. Of these, only 336 maps corresponding to the 1957 El Niiio, had been previously reported upon by Tovar and Garcia (1982). A few more maps covering August 1971 to April 1976 were analyzed by Tovar and Galarza (1983). Thus, the overwhelming part of the materials reported upon here is new."},{"index":2,"size":39,"text":"A planimeter was used to estimate for each month and island (or point) the surface area covered by the different stages (nonreproducing adult, reproducing adult, chick) of the three species in question (see Fig. 3 for a first example)."},{"index":3,"size":98,"text":"The empirically estimated densities in Table 2 were used to obtain, for each estimate of an area covered by birds, the corresponding population estimate. Note that the use of a mean density is justified due t~ the opposite tendencies of guano birds to both crowd themselves (to reduce predation), and to keep a minimum distance from each other (as determined by the birds' ability to peck at each other).  (see Table 1 for location). The color codes, i.e., black for cormorants (guanay), red for boobies (piquero) and blue for pelicans (alcatraz) were here replaced by different shading patterns."}]},{"head":"Data Interpolation","index":143,"paragraphs":[{"index":1,"size":45,"text":"Interpolations were necessary for various islands (or points) and month, mainly because original graphs had been lost. Thus e.g., for the case: month January February March April population 5,000 no data no data 2,000 we used the linear interpolation: 0 5,000 -2,000 = 3,000 ii)"},{"index":2,"size":62,"text":"3,000 / 3 = 1,000 iii) 2,000 + 1,000 = 3,000 (March) iv) 3,000 + 1,000 = 4,000 (February) As it turned out, interpolating missing data was never necessary for all islands and points in the same month and hence months with interpolated values have not been distinguished from months without such values when computing overall population sizes from 6 to 14%."}]},{"head":"Computational Example","index":144,"paragraphs":[{"index":1,"size":18,"text":"The Chincha Sur bird colonies (Fig. 4) demonstrate the planimetric method used for estimation of bird population sizes."},{"index":2,"size":98,"text":"In September 1974, the birds did not reproduce, hence the maps showing the distribution of reproducing adults with nests and chicks are empty. The planimeter conversion factor for Chincha Norte Island was computed as 81.5. Hence we have: number of cormorants = 0.019 8 1.5 15 = 232,275 number of boobies = 0.088 8 1.5 5 = 358,600 number of pelicans = 0.050 8 1.5 3 = 122,250 where 0.019,0088 and 0.05 are relative surface areas estimated by planimetry (see Fig. 4) and 15,5 and 3 are the empirical estimates of adult bird density taken from Table 2.  "}]},{"head":"Results and Discussion","index":145,"paragraphs":[]},{"head":"Areas Inhabited by Seabirds","index":146,"paragraphs":[{"index":1,"size":23,"text":"From 1909 to 1945, there were 36 guardposts (guardianas) on islands and points along the Peruvian coast, 23 of which were on islands."},{"index":2,"size":73,"text":"From 1945 on, the protection of the birds resulted in the expansion of their populations, requiring the establishment of new posts, especially on points which represent second-choice breeding locations compared to islands. As these colonies were frequently disturbed by people and animals (especially by the fox Dusycyon sechurae), the sites were isolated by walls of up to 3 m high. This intervention, which caused a further increase in bird population led to a"}]},{"head":"Changes in Relative Abundance","index":147,"paragraphs":[{"index":1,"size":80,"text":"Fig. 6 shows on an annual basis the changes in the relative abundance of cormorants, boobies and pelicans from 1953 to 1982 (see Table 6 for data pertaining to 1983). As might be seen, the major change is that the cormorants, being far more specialized on anchoveta than both boobies and pelicans (see Tovar and Galarza 1984;Muck and Pauly, this vol.) lost their overwhelming dominance over the latter two species, a trend which was intensified by the 1982-1983 El Niiio."}]},{"head":"Relationship Between Environmental and Guano Population Fluctuations","index":148,"paragraphs":[{"index":1,"size":49,"text":"Earlier authors commenting o the causes of mass mortalities in guano birds disagreed as to the relative impact of epizootic &dies and simple starvation due to a decline of fish biomass (Lavalle 1912(Lavalle , 1917;;Ballen 1f24,1925;Murphy 1931;Stiglich 1931;Torrico 1933;Vogt 1940; La Rosa Llosa and Ramos Sake 1949; Avila 1953)."},{"index":2,"size":124,"text":"More recent work (Jordan 1964;Tovar and Garcia 1982;Tovar and Galarza 1983;Tovar and Cabrera 1985) have vindicated those who suggested that lack of food is the ultimate cause maximum in 1955 of 52 guardposts along the Peruvian coast. However the expansion of the anchoveta purse seine fishery, which started in the early 1950s and the 1957 El Nifio led to a decrease in the number of points used for breeding by the birds and hence to a decline of the number of guardposts to 45 in 1964. Urban sprawl and the further decline of the bird population led to more points and islands becoming free of seabirds. In 1982, only 26 islands and 12 points from 6 to 180s were still used by the birds."}]},{"head":"Absolute Population Sizes of Guano Birds","index":149,"paragraphs":[{"index":1,"size":51,"text":"Previous reports of the population of guano birds along the coast of Peru have been presented for various, relatively short periods by Vogt (1942), Jordan (1959Jordan ( , 1961Jordan ( ,1963Jordan ( ,1964)), Gamarra (1964aGamarra ( , 1964b)), Jordan and Fuentes (1966), Galarza (1968), Fuentes (1969) and Tovar (1978Tovar ( ,1983))."},{"index":2,"size":22,"text":"Tables 3,4 and 5 present population estimates for cormorants, boobies and pelicans respectively. Fig. 5 presents the same data as time series.   "}]},{"head":"Cormorants","index":150,"paragraphs":[{"index":1,"size":1,"text":"-7"},{"index":2,"size":34,"text":"Year Fig. 6. Relative abundance of the three guano birds considered in this study, 1953 to 1982. Note gradual decline of the importance of cormorants and the small increase of pelicans (see also text)."},{"index":3,"size":97,"text":"for the mass mortalities of guano birds observed so far, and this is also confirmed by the present study, which shows that there is, for every El Niilo event off Peru, a corresponding decline of guano bird numbers. The fishery thus affects the guano bird populations, by reducing their food base before, during and after an El Nifio event, and this indeed has been assumed by Muck and Pauly (this vol.) who, based on the population estimates presented in this contribution, proceeded to estimate the amount of anchoveta eaten, from 1953 to 1982 by Peruvian guano birds."}]},{"head":"Introduction","index":151,"paragraphs":[{"index":1,"size":30,"text":"In the Peruvian upwelling ecosystem guano birds are probably the species most dependent on anchoveta (Engraulis ringens) as a food resource and the most conspicuous competitors for the anchoveta fishery."},{"index":2,"size":113,"text":"This competition is particularly visible when one compares the range of anchoveta sizes consumed by cormorants and boobies with the size distribution of the Peruvian purse seine fishery (see Figs. 1 and 2). That this size overlap is neither confined to Peru, nor to these two guano bird species is illustrated by Fig. 3, showing an even more pronounced overlap between the anchovy (E. mordax) size distribution in the California purse seine fishery and in regurgigated food bolus of the brown pelican (Pelecanus occidentalis californicus), a close relative of the pelican occurring off Peru (earlier taxonomies in fact saw P. thagus as a subspecies of P. occidentalis, just as P. occidentalis californicus is)."},{"index":3,"size":40,"text":"Periods of high anchoveta biomass have thus usually also been periods of high abundance of cormorants (Phalacrocorax bougainvillii), gannets (Sula variegata) and pelicans (Pelecanus thagus) which are the principal species of so-called guano birds (see Tovar et al., this vol.)."},{"index":4,"size":100,"text":"Their excrements (Spanish guano) are of economical importance as fertilizer and, in the past, for gunpowder production. Following the 1972 anchoveta collapse due to overexploitation and unfavorable oceanographic conditions, the guano birds became reduced to around 20% of their former population size (of 24 million in 1954-1955). Furness (1982) using data of Schaefer (1970) suggests that, between 1961 and 1965, guano birds were consuming annually 17% of the total Peruvian anchoveta biomass. Jordan (1967) estimated that the annual anchoveta consumption between 1961 and 1963 was around 45% of the total annual anchoveta catch; similar figures are given in Murphy (1972)."},{"index":5,"size":39,"text":"It is the aim of this paper to present monthly estimates of anchoveta consumption by guano birds for the period 1953 to 1982 and to relate this consumption to the fishery and environmental conditions such as El Ni5o events."}]},{"head":"Materials and Methods","index":152,"paragraphs":[{"index":1,"size":53,"text":"Anchoveta consumption was calculated using (i) monthly biomass data for the three guano bird species (based on the abundance data for juveniles and adults given in Tovar et al., this vol.), (ii) \"possible\" consumption rates (based on the guano birds' body weights and on metabolic considerations) and (iii) anchoveta availability to guano birds."},{"index":2,"size":18,"text":"Abundance data were converted to biomass by multiplying with the birds' average body weight, given in Table 1."},{"index":3,"size":61,"text":"A large body of literature on daily consumption rates based on metabolic equations, stomach contents analysis, feeding experiments and guano production exists for guano birds (Table 2). These data, plotted against body weight, suggest that specific differences can be neglected, and that the daily food consumption is a logarithmic function of body weight and of the analytic method (see Fig. 4)."},{"index":4,"size":25,"text":"The metabolic equationswhich result in notable lower valueswere used to calculate consumption rates corresponding to the daily energy expenditures (DEE, see Wiens and Scott    2)."},{"index":5,"size":63,"text":"1975; Schneider and Hunt 1982;Furness and Cooper 1982;Laugksch and Duffy 1984) sufficient for individual growth and egg production and for balancing, at the population level, mortality by reproduction. The higher consumption rates derived by the other methods can be interpreted as providing rough estimates of \"possible daily consumption\" (PDC). Increase of population would thus occur when actual consumption ranges between DEE and PDC."},{"index":6,"size":14,"text":"Two curves corresponding to these two consumption levels were established by regression analysis, i.e.:"},{"index":7,"size":57,"text":"Possible daily consumption PDC % = 29.78 -9.77 loge Weight Daily energy expenditures DEE % = 17.20 -3.73 loge Weight Dividing the PDC %-value by 100 and multiplying by 30 allows computation of total possible consumption for each of the three species \"j\" at month \"i\" (Cij) from Cij = Bij PDC%j and anchoveta consumption (Caij) as:"},{"index":8,"size":29,"text":"where \"Bij\" is the bird biomass for species '9\" at month \"i\" and \"Aaij\" is an index (>0.1) quantifying the availability of anchoveta during month \"i\" to predator \"j\"."},{"index":9,"size":36,"text":"The availability index is calculated as the product of (i) relative anchoveta biomass (Bai, >O, <I), (ii) anchoveta vulnerability (Vulij, >O, <I) and (iii) an index expressing competition for anchoveta as food resource (Compi, >O, 4):"},{"index":10,"size":75,"text":"This allows taking into account the suggestions of Furness (1982) and MacCall(1982) for modelling of the Peruvian anchovy-seabird system that (i) El Niiio related changes in sea temperature should affect anchoveta vulnerability to guano birds, (ii) that predation pressure from the guano birds themselves should reduce the availability of their food and that (iii) in periods following extremely high bird mortality the food supply per bird should be much greater than during the preceding period."},{"index":11,"size":14,"text":"The estimation of anchoveta availability to the three guano species was performed as follows:"},{"index":12,"size":38,"text":"Relative anchoveta biomass. Relative anchoveta biomass at month \"i\" (Bai) has been calculated using the preliminary anchoveta biomass data in Table 3, divided by the highest annual mean of 20.8 t x 106, in 1967 (see Fig. 5)."},{"index":13,"size":172,"text":"Anchoveta vulnerability index. Vulnerability (Vuli') is treated as a function of the overlap between predator and prey which might be expressed t Id ough an index of vertical (i.e., depth) overlap (V) and another index for horizontal overlap (H). Also, dispersion (D) was considered, leading to: a ~a s e d on independent biomass estimates cited in Pauly, Palomares and Gayanilo (this vol.), with linear interpolation to obtain \"monthly\" values and backward extrapolation using standardized catch/effort data in Boerema et al. (1965) and Doucet and Einarsson (1967). Cuution: Data in this table represent no more than educated guesses, presented here t~ allow replication of our results. See contribution cited above for improved estimates of monthly anchoveta biomass for 1953 to 1981. Spatial overlapping and anchoveta dispersion are related to sea surface temperature. Increases in water temperature (e.g., through El Nifio events) result in the dispersion of anchoveta (Furness 1982), their southward migration (Vogt 1942) and/or migration to deeper, cooler water (Schweigger 1940;Fiedler et al. 1943;Jordan and Fuentes 1966;Valdivia 1978;Johanneson and Vilchez 1980)."},{"index":14,"size":92,"text":"The data given in Table 1 on maximum foraging depth of the three bird species considered here shows that they are unable to forage at depths exceeding 15 cm. This diving capacity has evolved to correspond to the vertical distribution pattern of anchoveta school under normal conditions, for which estimates of 6-20 m, 10-30 m and 7-18 m have been presented by Clark (1954), Jordan (1967) and Guillen et al. (1969), respectively. However, during periods of high temperatures (i.e., El Nfio events), anchoveta schools occur at depths of 100 m and more."},{"index":15,"size":25,"text":"Therefore the vertical component \"Vi\" in equation ( 6) should be the most sensitive to temperature changes and therefore the key factor for vulnerability estimates."},{"index":16,"size":53,"text":"Acoustic surveys of pelagic fishes (sardine, mackerel, horse mackerel) conducted off Peru in 1983-1984 by IMARPE were used here to establish a relationship between fish depth distribution, time of the day and temperature. The range of temperatures covered was 16-300C, while the range of depth where fish concentrations were encountered was 5-120 m."},{"index":17,"size":91,"text":"It was found that the three species mentioned above are concentrated, under normal temperature conditions (<230C), at a mean depth of 15 m, slightly more than the anchoveta for which a value of 10 may be more appropriate (R. Vilchez, IMARPE, pen. comm.). From this information, we have derived an empirical equation relating the depth of main anchoveta biomass (DMBi) to sea surface temperature, i.e., which provides estimates of DMBi pertaining to the period from 6 a.m. to 10 p.m., i.e., during the main period of cormorant foraging activity (Jordan 1959)."},{"index":18,"size":28,"text":"The vertical overlapping index (Vi) for guano birds to anchoveta schools was calculated as dependent on the birds' maximum foraging depth (MFD) and DMB-values using the sigmoid function:"},{"index":19,"size":62,"text":"For all three guano bird species, the same maximum foraging depth of 15 m has been used, ignoring the 2 m value reported for pelicans (Table 1) because one important foraging mode of pelican is piracy on the other two species (Duffy 1980) which enables pelicans to \"extend\" their low foraging depth by indirectly using the diving capacity of cormorants and boobies."},{"index":20,"size":41,"text":"Using -18.5 for the constant \"p\" and 14.1 for \"q\" in equation ( 8), we obtained values of Vi close to 1 for a depth of 15 m and a rapid decline to zero for depths near and beyond 30 m."},{"index":21,"size":34,"text":"The latter depth value was chosen because the average total extension (in the vertical plane) of pelagic schools is around 30 m, representing an approximately normal distribution with the density maximum in the center."},{"index":22,"size":34,"text":"Thus, when the main concentration passes the 15 m threshold, a \"tail\" of around 15 m still remains within the birds' diving range, at least until the density maximum exceeds the 30 m depth."},{"index":23,"size":24,"text":"With respect to temperature-related horizontal displacement (offshore and/or southward migration) and fish dispersion, a quantification is difficult because reliable data are not yet available."},{"index":24,"size":50,"text":"In agreement with the literature cited above we assume that Hi and Di of equation ( 6) decrease with increasing temperature. As a first attempt we set Hi x Di = Vi resulting in: and using the same maximum foraging depth of 15 m for all the three seabird species:"},{"index":25,"size":52,"text":"Thus for example, an increase of surface temperature from 18 to 240C causes a change of the preferred depth of anchoveta from 15 to 17.3 m which results in a decrease of Vi from 1 to 0.9 and, finally in a change of anchoveta vulnerability for guano birds from 1 to 0.8."},{"index":26,"size":49,"text":"The use of mean monthly temperatures (see Table 2 in Pauly and Tsukayama, this vol.) is problematic because of the lack of information on within-month variability. This becomes important for the vulnerability estimates because a small change of 2 or 3 degrees centigrade causes drastic changes in the results."},{"index":27,"size":23,"text":"Therefore, we have chosen to use two temperature values for each month, expressing the temperature range within month \"i\", for the vulnerability estimates."},{"index":28,"size":55,"text":"Using an average within-month temperature deviation of + 15% we thus re-expressed, for example a mean value of 220C as a range of 18.7-25.30C and a mean of 21oC as a range of 17.9-24.20C. Table 4 shows that vulnerability estimates differ considerably depending upon whether a single mean temperature or a temperature range is used."},{"index":29,"size":17,"text":"All the vulnerability estimates cited have been calculated using the vulnerability means of the monthly temperature ranges."},{"index":30,"size":50,"text":"Competition for anchoveta. Competition (Cornpi) is split into an intraspecific (Icompi) and a fishery-related component (Fcompi): The product of both components varies between 1 (competition does not notably affect anchoveta consumption) and values close to zero (competition by fishery and/or the guano birds themselves drastically reduce anchoveta consumption per bird)."},{"index":31,"size":42,"text":"The simplest way to calculate \"Fcomp\" is to set it to zero when there is no anchoveta fishery (exploitation rate = 0) and to one respectively when the catch by the fishery is equal to the standing stock (exploitation rate = 1):"},{"index":32,"size":53,"text":"We have used a multiplication factor of 0.01 1 instead of 0.010 resulting in Fcompi becoming close to zero when the exploitation rate by the fishery is less than 1. This takes into account the competition pressure of the other anchoveta predators which might become important when the anchoveta biomass is very low."},{"index":33,"size":68,"text":"In view of our lack of knowledge of the mechanisms regulating seabird intraspecific competition, the concept applied here is a very crude one: we assume that competition between guano birds is high (Icomp = 0.7 in periods when seabird biomasses are high compared to that of anchoveta and becomes neligible (Icomp = 1) when the seabird biomass is very low compared to that of the anchoveta (Table 5)."},{"index":34,"size":76,"text":"Values of Icomp = 0.7 were thus used for the period between 1953 and 1956 when (i) the bird population sizes were highest and (ii) there was almost no fishery-induced disturbance and (iii) there were no temperature anomalies. During this period, the annual means of the ratio of the biomasses of the anchoveta and the birds (BAIBB) ranged between 340 and 550. The highest values occurred in 1965 (BAIBB = 3,200) and 1971 (BA/BB = 2,460)."},{"index":35,"size":154,"text":"Table 5. Assumed relationship between the ratio of anchoveta biomass to bird biomass (BAIBB) and the index of competition between individual guano birds (Icomp.); see text. A minimum value of Icomp = 0.7 was chosen because this corresponds to approximately zero population growth in an undisturbed system. Equation (2) gives the total consumption necessary for an approximately zero population growth. For the period 1954, 1955, 1956 the calculated annual means are: 2.5,2.5 and 2.1 t x 106. For the same period (using: Icomp = 0.7) anchoveta consumption values of 2.1,2.0 and 1.7 t x 106 were calculated. According to Jordan (1967) the non-anchoveta food in total seabird diet in 1955 ranged from 5 to 20%. Thus, adding a value of 20% to the annual anchoveta consumption we calculate for 1954,1955,1956 an annual total consumption of 2.5,2.4 and 2.0 t x 106 and this is almost identical to the amounts needed for zero population growth."},{"index":36,"size":48,"text":"Fig. 6 shows the flowchart of the computer program used for the anchoveta consumption estimates with emphasis on the dynamics of the variables controlling anchoveta consumption by guano birds between 1953 1982, notably relative anchoveta biomass, anchoveta exploitation rate, anchoveta biomasslbird biomass and the vulnerability and availabillity indices. "}]},{"head":"Results and Discussion","index":153,"paragraphs":[]},{"head":"Exi","index":154,"paragraphs":[{"index":1,"size":67,"text":"Monthly anchoveta consumption estimates are given in Tables 6,7 and 8 for cormorants, boobies and pelicans respectively. Fig. 7 shows total annual anchoveta consumption for all three guano bird species, as well as their total annual possible consumption. Anchoveta consumption by guano birds was maximum between 1953 and 1956 (1.3-2.1 t x 106) and declined continuously to less than 20,000 t at the beginning of the 1970s."},{"index":2,"size":70,"text":"Comparing annual anchoveta catch by the fishery with the relative anchoveta consumption by guano birds, we note that these values vary beween 1953 and 1982 by a factor of 50 (Fig. 8):    before 1960 the annual guano consumption declined from 5,000% to 45% of the annual anchoveta catch. Between 1960 and 1970 the percentage dropped to 0.1% then reached an average of around 3% between 1970 and 1982 (Fig. 8)."},{"index":3,"size":23,"text":"The dynamics of seabird biomass between 1953 and 1964 correspond widely to the anchoveta availability curve (Fig. 5) and suggests a direct dependency."},{"index":4,"size":47,"text":"The sudden population breakdown in 1957-1958 and the following recovery is accompanied by similar changes in anchoveta availability caused by extremely low anchoveta vulnerability during the 1956-1957El Niao (Fig. 5), while between 1965and 1967and from 1977to 1982 the trends of bird biomass and anchoveta availability were opposite."},{"index":5,"size":56,"text":"Comparing species specific seabird consumption to total anchoveta consumption between 1953 and 1983 the following dynamics are established (Fig. 8): between 1953 and 1972 the dominant species are cormorants (Phalacrocorax bougainvilli), with between 50 and 90% of the total anchoveta consumption, while boobies (Sula variegata) consumed 10 to 40% and pelicans (Pelecanus occidentalis) 5 to 20%."},{"index":6,"size":70,"text":"Between 1972 and 1979 a shift in dominance occurred from cormorants to boobies: during this period around half of the total anchoveta consumption was taken by boobies, around 40% by cormorants and around 10% by pelicans. As might be seen in Fig. 9, from 1979 on, the dominance pattern prevailing before 1972 was re-established, with cormorants taking 65%, boobies 30% and pelicans 5% of all anchoveta eaten by guano birds."}]},{"head":"Introduction","index":155,"paragraphs":[{"index":1,"size":28,"text":"A key element for understanding the fluctuations of resource species observed in the Peruvian upwelling system is the analysis of predation and its dynamics during the last decades."},{"index":2,"size":49,"text":"In addition to fishes and guano birds, marine mammals especially pinnipeds, may be considered to have an impact on pelagic fish, particularly anchoveta, as suggested by Laevastu and Favorite (1980), who calculated that the herring consumption by mammals is about ten times the commercial catch in the Bering Sea."},{"index":3,"size":64,"text":"The objective of this paper is to give an estimate of fish consumption by the Peruvian sea lion (Otariaflavescens) and fur seal (Arctocephalus australis) with special emphasis to their consumption of anchoveta (Engraulis ringens). Throughout this paper, it is assumed that the anchoveta consumed by seals off Peru have a size distribution roughly similar to those caught by the purse seine fishery (Fig. 1)."}]},{"head":"Materials and Methods","index":156,"paragraphs":[]},{"head":"The Abundance Model","index":157,"paragraphs":[{"index":1,"size":48,"text":"The sea lion and fur seal populations are estimated using a population model which calculates in monthly intervals the abundance of 5 age groups (0-1 year, 1-2 years, 2-3 years, 3-4 years, >4 years), and in which population growth is assumed to be influenced by El Niiio events  "}]},{"head":"!la","index":158,"paragraphs":[{"index":1,"size":175,"text":"Body length (LT,cm) . because these warm water periods seem to be critical for (i) the survival of the newborn, (ii) the mortality of juveniles and adults and (iii) the reproductive success of the adult females (Table 1, Fig. 2). Majluf (1984Majluf ( , 1985) ) and Tovar and Fuentes (1984) registered extremely high pup mortality during the 1982-1983 El Niiio with near zero survival of the generation born between October 1982 and March 1983 and with juvenile and adult mortalities higher than in normal years. Studies of fur seal behavior (Majluf 1985) have shown that the high mortality of the sucklin'g pups is probably caused by starvation because the females are longer absent (searching for food), during El Niiio events, than under normal environmental conditions and/or produce less milk. P. Majluf (pers. comm.) suggests that under such conditions, the average number of births per female fur seal is less than one per year and that this lack of reproductive success is probably related to the low level of food available to the adult females."},{"index":2,"size":24,"text":"This agrees with results of Tovar and Fuentes (1984) suggesting depressed reproduction of fur seal in the season immediately succeeding the 1982-1983 El Niiio."},{"index":3,"size":28,"text":"Census data collected before and after the 1982-1983 El NCio suggest strong differences in the reproduction of sea lion and fur seals to El Niiio events (Table 2)."},{"index":4,"size":36,"text":"However, it is probably the combined effect of (i) low food availability, (ii) long foraging trips and (iii) metabolic stress because of abnormal, high environmental temperatures which affects pinniped mortality and reproduction during El Niiio periods."},{"index":5,"size":61,"text":"Food availability to the seal and thus duration of their foraging trips are dependent on sea temperature because rises in water temperature (El Niiio) result in the dispersion of their pelagic fish food (Furness 1982), in their southward migration (Vogt 1942) and/or their migration to deeper, cooler water (Schweigger 1940;Fiedler et al. 1943;Jordan and Fuentes 1966;Valdivia 1978Valdivia ,1980;;Johanneson and Vilchez 1980)."},{"index":6,"size":15,"text":"Table 1. Initial data set and constants used in pinniped population and anchoveta consumption model."}]},{"head":"Characteristic","index":159,"paragraphs":[{"index":1,"size":4,"text":"Fur seal Sea lion"}]},{"head":"---","index":160,"paragraphs":[{"index":1,"size":62,"text":"Initial population size (Jan. 1953) 0-0.9 year 1-1.9 year 2-2.9 year 3-3.9 year  Table 1 and text).  Although the use of sea temperature seems to be an adequate concept for the modification of biological rates by El Niiio conditions, its application to the population growth of Peruvian fur seals and sea lions is difficult because of the complex character of temperature-mediated mechanisms."},{"index":2,"size":5,"text":"Modelling was done as follows:"},{"index":3,"size":29,"text":"Mortality and reproductive success are described empirically as functions of surface temperature. Thus mortality has been split into a basic term (MB) and a temperature related one (MT), i.e.,"},{"index":4,"size":21,"text":"Similarly, annual reproductive success (RS) is set to 1 under normal conditions and becomes reduced by a temperature related term (XT):"},{"index":5,"size":44,"text":"MT and XT are zero for temperatures less than 200C (= mean maximum sea surface temperature in March, 30-year average. For the mathematical description of MT and XT, we have used a sigmoid function, in agreement with the general theory of biological response curves:"},{"index":6,"size":41,"text":"The constants a, b and c were estimated empirically by fitting the calculated results of population growth to field data (see below). The temperature related mortality effect on pups was assumed to be three times higher than on juveniles and adults."},{"index":7,"size":12,"text":"The number of newborn pups for the month \"in is calculated by:"},{"index":8,"size":124,"text":"where \"N4+ i\" is the total number of adults (i.e., animals of 4 years and more (Piazza 1959)) for month \"i\", \"SR\" the sex ratio, \"RS\" the annual reproductive success and \"Wi\" is a weighting factor for the reproduction intensity in the month \"i\". Wi is set at zero for all months without reproduction and reaches a cumulative value of 1 between December and March in sea lion and October and December in fur seal (see Table 1). Fig. 2 shows the flowchart of the program used for the monthly calculation of the five fur seal and sea lion age groups. Table 1 gives the constants and initial values used for the computation and Fig. 3 shows a comparison between field and model data."},{"index":9,"size":57,"text":"The field data are based on census conducted between 1961 and 1984 by different private and state organizations authorized by the Peruvian Government. Most of the census were done during the reproductive season when the main part of the population is terrestrial. The fraction of juveniles and adults in the sea was estimated visually by seaborn teams."},{"index":10,"size":27,"text":"The authors participated in the last two of these population census which, while not being well standardized, tend to reflect the trend of pinniped populations off Peru."},{"index":11,"size":145,"text":"There are no direct estimates of populhtion size before 1961. Piazza (1959) mentioned that one single company killed 36,500 individuals (88% pups) between December 1941 and March  I). 1942, andWalker (1975) gives annual values of total catches of up to 75,000 individuals for the years before 1942. Majluf (1980) assumes a total population size of both species of between 400,000 and 500,000 individuals for 1940. When compared with the 12,500 individuals estimated in 1961 or the 49,200 estimated in the 1984 census, these figures, however crude, nevertheless give vivid impression of the extent of the population collapse due to overexploitation during the first half of this century. In 1946 the Peruvian Government initiated first preventive measures by restricting the harvest season to 4 months per year and since 1950 (fur seal) and 1956 (sea lion), except for the period 197 1-1975, harvesting is total1 prohibited."},{"index":12,"size":21,"text":"For the present calculations we assume that bo species had their lowest population size during the first half of the 1950s."}]},{"head":"X","index":161,"paragraphs":[{"index":1,"size":58,"text":"We do not believe in zero catches during the years with harvest prohibition, because there is information that even the persons responsible for the protection of the fur seal and sea lion colonies traded in pelts, mainly those of pups. For our model, we assumed that an average of 10% of each pup generation dies of unnatural causes."},{"index":2,"size":115,"text":"The causes of the decrease (-40%) in population size of fur seals between the census of March 1968 (1 1,800 individuals) and May 197 1 (7,250) are difficult to identify. A possible reason could be the combined effect of (i) increasing mortality due to food shortage (intensified by abnormal sea temperatures) and (ii) high pup mortality by harvest: positive temperature anomalies occurred in 1969 (May: +2.50C). The harvest prohibition for pups was suspended in 1971, and the anchoveta biomass as an important food resource for pinnipeds was reduced from 21 t x 106 in 1967 to only 12 t x 106 in 1971 (annual means, see Table 3 in Muck and Pauly, this vol.) ."},{"index":3,"size":42,"text":"Assuming that population growth is regulated mainly by survival of the newborns, we have adjusted the population curve to the field results of this period by gradually increasing the monthly pup mortality rate of fur seals between April 1968 and February 1971."},{"index":4,"size":51,"text":"Table 3. Mean length, mean weight and mean daily ration of sea lion and fur seal age groups (see text for sources). Males 0-0.9 1-1.9 2-2.9 168 3-3.9 To convert sea lion abundance into biomass the following equations based on length vs. age data in Vaz-Ferreira (1982), have been used: males"},{"index":5,"size":114,"text":": length = 68.18 x age0.27 females : length = 70.43 x age0.21 males : weight = 0.000079 x length2.74 females : weight = 0.000209 x length2.54 where the weight is given in kg, the total body length in cm and the age in months. For fur seals, the lengtwweight equation given in Ximenez et al. (1984) was used. Since these authors did not find any difference between males and females, a common equation has been applied to both. For the lengtwage relationship, we have used the same equation as for sea lion females because no data for fur seal were available: males & females:length = 70.43 aae0s2l males & females:weight = 0.000034 length2.88"}]},{"head":"The Consumption Model","index":162,"paragraphs":[{"index":1,"size":34,"text":"This model follows the approach described in Muck and Pauly (this vol.) for the consumption of anchoveta by guano birds. Monthly total possible consumption (Ci) is defined by: Ci = Bi PMC% ... 11)"},{"index":2,"size":55,"text":"where \"Bi\" is the predator biomass for the month \"i\" and \"PMC%\" the possible monthly consumption in percent of body weight. For the estimation of \"PMC%\" we have used the formula given in Perez and Mooney (1984) calculated from data reported by Bigg et al. (1978) on feeding rates of captive adult fur seal females:"},{"index":3,"size":81,"text":"where \"MJ\" is the daily energy consumption in megajoules and \"W\" the seal's body weight in kg (Table 3). To convert energy consumption in % of body weight, the estimates given in Antonelis and Perez (1984) for the relationship mjkg of northern fur seal diet (75% fish, 25% squid) have been used. The average of 28 monthly samples is 1 kg = 6.31 mj, s.d. = 0.85. Therefore the total possible monthly consumption in % of body weight is given by:"},{"index":4,"size":9,"text":"The monthly anchoveta consumption (Cai) can be described as:"},{"index":5,"size":30,"text":"where \"Ai\" is the relative anchoveta availability to pinnipeds at month \"i\" (>O, <I) equivalent to the product of relative anchoveta biomass (Bi, >O, <I), anchoveta vulnerability (Vuli, >O, <I)"},{"index":6,"size":12,"text":"and an element (Cornpi) expressing competition for anchoveta by other predators, i.e.:"},{"index":7,"size":85,"text":"The relative anchoveta biomass at month \"in has been calculated using the preliminary anchovy biomass values in Muck and Pauly (this vol.) divided by the highest annual mean (20.8 t x 106 in 1967). Anchoveta vulnerability has been treated as a distribution problem between predator and prey which is most probably controlled by the temperature-related migratory behavior of pelagic fish schools. Thus, vulnerability can be split into a spatial overlapping index with a vertical (Vi) and a horizontal (Hi) component, and a dispersion index (Di):"},{"index":8,"size":38,"text":"The diving capacity of pinnipeds is much higher than that of seabirds (Kooyman et al. 198 1, 1982). Majluf (1985) reports a mean foraging depth of 20-40 m and maximum values of around 100 m for fur seals."},{"index":9,"size":43,"text":"Thus, using an average of 50 m as a first approximation for the \"critical depth\" for both species, equation ( 8) in Muck and Pauly (this vol.), can be used to calculate the vertical overlapping index \"Vi\" for predator (pinniped) and prey (anchoveta):"},{"index":10,"size":15,"text":"where \"DMBi\" gives the (temperature-dependent) mean depth at which the main anchoveta biomass is concentrated."},{"index":11,"size":130,"text":"Horizontal displacements (offshore and/or southward migrations) and fish dispersion, in addition to sea temperature-related changes in vertical overlapping between predator and prey, will also reduce anchoveta vulnerability for pinnipeds (Majluf 1985). We use as a first approximation Hi x Di = Vi, resulting in: Table 4 gives calculated estimates of V and Vul for different DMB-values. With respect to the temperature used for the computation of DMB, the same technique as described in Muck and Pauly (this vol.) has been used: a mean DMB-value for month \"i\" was calculated using the minimum and the maximum temperature values which result from expressing the monthly mean temperatures as a +/-15% range (see Table 4 in Muck and Pauly, this vol.); an application example of this approach is given here as Table 5."},{"index":12,"size":24,"text":"Between 1959 and 1982, anchoveta exploitation by the fishery was probably the dominant component of \"competition\" affecting the anchoveta availability for all anchoveta-feeding species."},{"index":13,"size":35,"text":"Thus, as a fxst attempt to estimate the influence of competition on anchoveta availability for Peruvian pinnipeds we use the anchoveta exploitation rate by the fishery (Expli) as defined in Muck and Pauly (this vol.):"},{"index":14,"size":47,"text":"No attempt was made to model intraspecific competition as described for guano birds (Muck and Pauly, this vol.) because, in contrast to the guano birds, the pinnipeds, between 1953 and 1982, had very low biomasses in relation to those of the anchoveta, and less erratic population fluctuations."},{"index":15,"size":58,"text":"The biomass and consumption estimates calculated by the model refer to the whole of the Peruvian littoral (3-18.50s) but in agreement with the reference area used in this volume for biological and physical parameters, we give corrected estimates for this smaller area (4-140S), using the mean fraction of population within the restricted range, as derived in Table 6. "}]},{"head":"Results","index":163,"paragraphs":[{"index":1,"size":27,"text":"Tables 7 and 8 give monthly data for biomass and anchoveta consumption for sea lion and fur seal, respectively between 1953 and 1982 for the area 4-140s."},{"index":2,"size":67,"text":"Fig. 4 shows the dynamics of total annual consumption and annual anchoveta consumption between 1953 and 1982. Total consumption increases continuously from around 10,000 t in 1953 to 60,000 t in 1982, directly reflecting the population growth of both pinniped species. Anchoveta consumption had its maximum of approximately 11,000 t in 1965-1967 and an average of approximately 3,700 t (s.d. = 2,000 t) between 1968 and 1982."},{"index":3,"size":52,"text":"The mean anchoveta fraction in total pinniped diet is 40% and varies from 81% in 1953 to 2% in 1982. Because of their dominant biomass in the area considered here (4-140S), sea lions account for about 88% of total pinniped consumption while anchoveta consumption by fur seals accounts for only around 12%."},{"index":4,"size":78,"text":"The population growth of sea lions does not seem to be affected by the continuous decline of anchoveta availability (Fig. 5), which decreased from 0.8 to less than 0.1 between 1953 and 1982 while the sea lion biomass increased from approximately 300 t in 1953 to 1,800 t in 1982. A similar relationship was found for fur seals although there is a short period (1968)(1969)(1970)(1971) for which anchoveta availability and fur seal biomass show the same declining trend."}]},{"head":"Discussion","index":164,"paragraphs":[{"index":1,"size":94,"text":"The consumption estimates presented were derived to obtain an idea on the importance of pinniped predation for the pelagic ecosystem off Peru with particular reference to anchoveta (Engraulis ringens J.). ( 87  The estimated anchoveta fraction in pinniped diet varies from 81 to 2% and reflects the dynamics of anchoveta availability between 1953 and 1982. Antonelis and Perez (1984) calculated that the percentage of northern anchovy (Engraulis mordax) in the diet of northern fur seal changed from 43% off California to only 11% off the Washington-Oregon coast, depending on the relative availability of anchovy."},{"index":2,"size":68,"text":"This corresponds to the findings of Antonelis and Ficus (1980), Bonner (1982), Kajimura (1982) and Perez and Bigg (1981) who suggested that pinnipeds are not specialized predators but opportunistic ones, feeding on a wide range of seasonally abundant prey species. Changes in availability might result from a variety of factors such as depletion of resource by commercial fishery, interspecific competition and climatic changes (MacCall 1983;Moyle and Cech 1982)."},{"index":3,"size":109,"text":"The consumption model presented here suggests that anchoveta availability for pinnipeds is principally controlled by changes in anchoveta biomass and by competition of the anchoveta fishery. In contrast to the Peruvian sea birds (Muck and Pauly, this vol.) the influence of El NiZo periods on anchoveta vulnerability and therefore anchoveta consumption seems to be of little importance before 1983. The situation becomes different for the 1982-1983 El Niiio which was probably the strongest in this century. Table 5 gives vulnerability estimates (Vuli) and estimates of the mean depth were the main anchoveta biomass is concentrated (DM&) related to the (surface) temperature regime during the first 8 months of 1983."},{"index":4,"size":202,"text":"Total consumption by pinnipeds was calculated using feeding rates of captive fur seal females (Perez and Mooney 1984) which most probably are equal or at least close to the possible upper limits of food consumption when food is abundant. These rates were used for the entire 1953-1982 period because the steady increase of pinniped biomass from around 600 t in 1953 to approximately 50,000 t in 1982 suggests no limitations by food resource. The period 1968 to 1971 is an exception in the case of fur seals, when the census results show a 40% decrease in population size. Comparing anchoveta availability for the same period a similar tendency has been calculated (Fig. 5): as a consequence of (i) the anchoveta fishery as competitor to pinniped anchoveta predation and (ii) reduced total anchoveta biomass, anchoveta availability dropped from 0.19 in 1968 to 0.05 in 1971. Therefore we cannot exclude the possibility that a shortage of anchoveta between 1968 and 1972 affected fur seal population growth. However, in general, population growth of Peruvian pinnipeds seems to be independent of anchoveta availability and biomass, as confirmed by the fact that both seal populations had highest growth rates when the anchoveta stock was much reduced (1978)(1979)(1980)(1981)(1982)."},{"index":5,"size":121,"text":"This suggests a change of food by the predator and agrees with results of Majluf (1985) for the Peruvian fur seals. Analyzing the otoliths from fur seal scats, she found a shift from almost 100% anchoveta to a diet of sardine and horse mackerel in times of low anchoveta abundance. Therefore it is doubtful whether pinnipeds still feed on anchoveta when its availability drops below a threshold value. Ecological models, e.g., that of Murdoch et al. (1975) and field data on pinnipeds (Bailey and Ainley 198111982) suggest a \"switch\" in feeding behavior to alternative prey when the traditional resource becomes scarce. In this case the calculated anchoveta consumption for periods with very low anchoveta availability (1969 to 1982) could be overestimated."}]},{"head":"Introduction","index":165,"paragraphs":[{"index":1,"size":6,"text":"This account is (Engraulis ringens) 1982."},{"index":2,"size":164,"text":"an attempt to estimate, on a monthly basis, the consumption of anchoveta by bonitos (Sarda chiliensis chiliensis) off Peru (4 to 140s) from 1953 to Peruvian data on the biology and fishery of bonitos are available which make such an estimate possible, given that one dares to fiu in gaps through inferences from comparable stocks and/or related species. The assumptions made here to estimate the anchoveta consumption of bonitos are justified by the fact that any account of the population dynamics of the Peruvian anchoveta will be incomplete if it ignores the role of bonitos as highly specialized predators of anchoveta. The value of the present exercise, however, may not be solely in the fact that it provides estimates of anchoveta consumption. Rather, it is also meant to illustrate how far historic data collected on a given fish stockoff Peru or elsewherecan be \"stretched when one includes in an analysis as much as possible the relevant information on a given species and allied forms."},{"index":3,"size":49,"text":"This exercise, therefore may be seen as a complement to Yoshida's (1980) excellent synopsis of data on the genus Sarda, emphasis being given here to the Peruvian stock of Sarda chiliensis chiliensis, and to the numerous Peruvian contribution on bonitos (most in manuscript form) not incorporated in said synopsis."},{"index":4,"size":151,"text":"This contribution may also be seen, finally as an attempt to refine two previous \"educated guesses\" of the annual consumption of anchoveta by bonito. The first of these educated guesses was by Saetersdal et al. (1965) who wrote that the estimated annual consumption of anchoveta by guano birds is of about 3 t x 106. If we assume that bonitos (with an estimated biomass of 300,000 t) and other fishes consume the same quantity (i.e., 1.6 t x 106) of anchoveta, then ... Boerema et al. (1965) wrote, on the other hand that ... Another predator of the anchovy is the bonito. Nearly 100,000 t of this species are caught annually in Peruvian waters. It it is assumed that these fish have needed for their growth about 10 times their weight in food, and that this food mainly consists of anchovy, this adds another million tons to the figure of predation."}]},{"head":"The Catch and the Species Caught, 1953 to 1982","index":166,"paragraphs":[{"index":1,"size":87,"text":"Although there have been uncertainties concerning the taxonomic status of the Peruvian bonitos, it is now agreed that all bonitos occurring in the eastern Pacific can be attributed to two species, Sarda orientalis and S. chiliensis. The latter has an anti-equatorial distribution (Randall 198 1) and is separated in two units now given subspecies rank -S. chiliensis lineolata north and S. chiliensis chiliensis south of the equator while the distribution of S. orientalis straddles the zero latitude line (Collette and Chao 1975;Yoshida 1980; and see Fig. 1)."},{"index":2,"size":105,"text":"Because of its narrow distributionbarely touching the north coast of Peru and rarely overlapping with the distribution of anchoveta (Vildoso, pers. obs.) -S. orientalis usually contributes negligible amounts to the Peruvian catch of bonito, for which reason this catch will be treated below as if consisting only of S. chiliensis chiliensis. This might introduce an error in periods where El Nifio events, by warming up the surface waters off Peru, necessarily extend southward the distribution of S. orientalis. However, this occasional (and partial) replacement of one bonito species by another shall be treated here as if it had no impact on fishermen or anchoveta consumption."},{"index":3,"size":60,"text":"We should note here that three different, size-dependent names are given to specimens of S. chiliensis chiliensis in Peru, namely chauchilla for fish up to 30 cm, cerrajon for fish between 30 and 50 cm, and bonito (sensu stricto) for fish above 50 cm (Vildoso 1961); the latter group shall be called \"large bonitos\" in the rest of this contribution."},{"index":4,"size":62,"text":"Table 1 gives estimates of the annual bonito catch off Peru from 1951 to 1983. This catch, which had gradually increased, from near zero in 1940 to 50.8 thousand t in 1951, peaked in the early 1960s with a maximum of over 100,000 t in the early 1960s, declined to negligible values in the late 1970s then increased again in the 1980s."},{"index":5,"size":76,"text":"At this peak, from the mid-1950s to the mid-1960s, the Peruvian bonito fishery represented a very significant part of the overall fishery sector, both in terms of supplies to the Peruvian (fresh and frozen bonitos) as well as the export (i.e. US and western Europe) markets. For example, the value of canned bonitos exported in 1958-1959 amounted to 25% of the value of Peru's fishmeal exports for the same period (computed from data in Anon. 1960)."},{"index":6,"size":29,"text":"The bonito fishery for various reasons, however, never attracted an amount of attention proportional to its value, and hence the impact of bonitos on the anchoveta stock remained unappreciated."}]},{"head":"Growth and Natural Mortality of Sarda chiliensis chiliensis","index":167,"paragraphs":[{"index":1,"size":73,"text":"Several models presented further below require a concise description of the growth parameters of S. chiliensis. This applies particularly to our attempt to estimate biomasses through catch data and fishing mortality estimates. However, the growth curves and growth parameter Prom Mejia (1967) andIMARPE (1971). Units are tonnes of bonitos caught per trip. See text and Fig. 4. Obtained by multiplication of standardized fishing effort by q = 0.0897 (see text and Fig. 6)."},{"index":2,"size":46,"text":"Correction by 0.9 to account for catches outside of 4 to 14's. obtained by smoothing (through running average over 3 years) the biomass estimates obtained in (e). Catch values in excess of \"MSY\" were assumed to have been obtained by an effort to equal to fopt."},{"index":3,"size":170,"text":"estimates available in the literature are contradictory and confusing. Kuo (1970), based on otolith readings in S. chiliensis lineolata and S. chiliensis chiliensis, presented growth \"curves\" that are almost linear over the interesting span of ages (1 to 6 years), and which have values of Loo much larger than the maximum sizes recorded in bonitos (Fig. 2) while the growth parameters reported by Campbell and Collins (1975), based on analysis of otoliths and length-frequency data, appear also to be questionable (Table 2). The maximum length reported for S. chiliensis lineolata is about 102 cm (Yoshida 1980). The southern subspecies, on the other hand, tends to remain smaller, with a definitive record of 79 cm (De Buen 1958). Mann (1954) and Vildoso (1961) state that Chilean bonitos may exceed 80 cm and reach 90 cm, respectively. Vildoso (1955) reported never to have encountered a bonito in excess of 75 cm in Peruvian waters from which however, one exceptionally large specimen of 93 cm and 94 kg has been reported (Anon. 1958)."},{"index":4,"size":85,"text":"Length-frequency data from the early phase of the fishery are available (see Table 3) which allow estimation of asymptotic length (L, ) in Peruvian bonito using the method of Wetherall  4). (1986), in which a length-frequency sample representative of the steady-state population is used in conjunction with a regression of the form (where is mean length above L'i, the lower limit of length class (i)) to estimate L, through Fig. 3 illustrates this method, which also allows estimation of the ratio Z/K (see below) via"},{"index":5,"size":163,"text":"The value of L, = 76 cm, estimated using equations ( 1) and ( 2) and length-frequency data for the period 1951-1953 (Fig. 3) corresponds rather well to the maximum lengths discussed above, and shall be used here as best estimate of asymptotic length in Peruvian bonito. Pauly and Munro (1984) suggested that the parameters 4' could be used to express the average growth performance in a given taxon when growth can be described in the von Bertalanffy Growth Function (VBGF) of the form where Lt is the length at age t, and K and to are the constants which, besides L, , must be estimated empirically. In such case, 4' is defined by (See Pauly 1979 for derivation of this equation). As might be seen in Table 4, the three sets of L, and K values available on S. chiliensis lead to a mean value of B' = 3.40, very close to the value of 4' = 3.43 estimated for Sarda sarda."},{"index":6,"size":31,"text":"We have used here equation ( 5), the overall mean value of @' = 3.42 (see Table 4) and L, = 76 cm to estimate K = 0.455 for Peruvian bonito."},{"index":7,"size":86,"text":"Table 2. Length-at-age data in Fig. 1 of Bartoo and Parker (1983) based on Campbell and Collins (1975)   The value of the parameter to is estimated finally from an empirical equation of Pauly (1979), i.e. which, given L, (in cm) and K (llyear) values, provides crude estimates of to when these cannot be obtained by any other method. Thus, we have to describe the growth in length of Peruvian bonitos, where Lt (fork length, in cm) is the length at age t (in years) (Fig. 2)."},{"index":8,"size":94,"text":"A number of allometric length-weight relationship have been estimated for Peruvian bonitos by various authors (Table 5). We shall use here, however, a simple isometric length-weight relationship of the form with a mean c.f. (condition factor) estimated from the data in Table 4 of 1.348, and leading to weights in g when (fork) length is expressed in cm. Thus we shall express the growth in weight of Peruvian bonito as which is reasonable in view of the fact that none of us ever observed a Peruvian bonito exceeding 5 kg (see also Vildoso 1955)."}]},{"head":"Catch per Effort and Effort in the Peruvian Bonito Fishery, 1953 to 1982","index":168,"paragraphs":[{"index":1,"size":66,"text":"Effort and catch-per-effort data are not available which cover the whole period considered here. In fact, the only previous contributions dealing in some detail with the bonito fishery off Peru is the unpublished thesis of Mejia (1967) and an IMARPE internal report (1971), in which standardized effort and catch-per-effort of the Chimbote fleet of purse seiners for the period 1967 to 1970 were derived and analyzed."},{"index":2,"size":167,"text":"Since Chimbote was the main base of the Peruvian bonito fishery (Mejia 1967) we have divided the catch per effort of the Chimbote fleet for this period into the total Peruvian catch to obtain estimates of total Peruvian effort (see Table 1). Then, we have fitted a simple parabolic surplus yield model to the total catch and overall effort data (Fig. 4). The yield curve was then used to generate, from the catch data prior to 1962 and beyond 1970, estimates of effort, based on the assumption that the low yields of the earlier period were associated with effort levels below fopt (the effort level associated with MSY), and that the low yields of the recent period were associated with effort levels beyond fo t. We believe these two assumptions to be reasonable given what is presently known o !' the early development of the Peruvian fishing fleet in general, and of the bonito stock and fishery in particular (see below and other contributions in this volume)."}]},{"head":"Standardized effort units","index":169,"paragraphs":[{"index":1,"size":46,"text":"Fig. 4. \"Schaefer\" model of the Peruvian bonito fishery, based on data for the period 1962 to 1970 (Table 1) and used to estimate fishing effort from available catch data for the periods 1951 to 1961 and 1971 to 1983 (see also Table 1 and text)."}]},{"head":"Total, Natural and Fishing Mortality of Bonito","index":170,"paragraphs":[{"index":1,"size":71,"text":"Size-frequency distributions of fish caught by the Peruvian bonito fishery off Callao and Paita are available for the years 1951-1953,1962,1964to 1971and 1976 (Table 3) (Table 3). From these, total mortality (Z) was estimated using the equation wherez is the mean length in the catch, computed from the first length (L') fully represented in samples (Beverton and Holt 1956). The estimate of Z obtained in this fashion are given in Table 3."},{"index":2,"size":109,"text":"A first estimate of natural mortality was obtained through the model of Csirke and Caddy (1983), in which annual catch is plotted against total mortality, and in which the Z-intercept of the fitted parabola provides an estimate of M (see Fig. 5). The estimate of M = 0.878 obtained in this fashion corresponds well with the value of M obtained through the empirical equation of Pauly (1980) in which and which leads for a mean water temperature off Peru (4 to 140S, 1953Peru (4 to 140S, -1982) ) of T = 180C and the values of L , and K obtained above, to an estimate of M = 0.675."},{"index":3,"size":24,"text":"Averaging these two relatively independent estimates of M yields a mean value of M = 0.775 which will be used for all further computations."},{"index":4,"size":43,"text":"The estimates of effort and of Z derived above were then combined with the estimate of M to estimate a mean catchability coefficient (q), defined as where f is fishing effort and F is fishing mortality, defined for any given year (i) by"},{"index":5,"size":77,"text":"The value of q estimated from the data at hand (see Fig. 6) is a = 0.0897, which was used to compute, using equation ( 12), values of F given the corres~onding*values of effort for 1951 to 1983 (see Table 1). to 1976 and used to obtain preliminary estimate of natural mortality. Note that estimated MSY corresponds well with value estimated using Schaefer model (see Fig. 4). (P refers to Paita, C to Callao; see Table 3)."}]},{"head":"Standardized effort units","index":171,"paragraphs":[{"index":1,"size":63,"text":"Fig. 6. Estimation of a mean catchability coefficient (q) for the Peruvian bonito fishery, 1951 to 1983. Note that the line is not a linear regression used to estimate q and M, but rather was forced through an assumed value of M (see text) and the mean of the available f and Z values (Prefers to Paita, C to Callao; see Table 3). "}]},{"head":"Food, Feeding Habits and Food Consumption of Bonitos","index":172,"paragraphs":[{"index":1,"size":77,"text":"Several studies on the food habits of bonitos have been conducted in Peru (Table 7). All of them emphasize the importance of anchoveta in the diet of this predator. Fig. 7 summarizes the available data on the size distribution of anchoveta found in bonito stomachs which resembles that of anchoveta in the commercial catch (see also Tsukayama and Palomares, this vol.). Thus, the estimates of anchoveta consumption obtained below can be straightforwardly added to the commercial catch."},{"index":2,"size":67,"text":"Several authors have presented anecdotal information on the weight of stomach contents in Peruvian bonito (Table 7), but only the data of Canal (1974) were presented in a fashion whichallowed further analysis (see Table 8). Table 9 suggests that, on the average, only about half of the bonito of a given stock have food in their stomach at any given time, as is commonly reported from scombrids."},{"index":3,"size":243,"text":"Data on stomach evacuation rate, from which food consumption estimates could be obtained using the model of Elliott and Persson (1978), are not available for any Sarda species, although aquarium observations on the feeding behavior of S. chiliensis lineolata are available (Magnuson and Prescott 1966). We shall therefore use here data obtained by Magnuson (1969) from captive skipjack (Katsuwonus pelamis). This approach appears justified in that bonito and skipjackones account is taken of differences in internal and external temperatureare quite similar, particularly with regard to the shape and length of their guts (Godsil 1954;Collette 1983). Another justification for our use of data obtained from an experiment with skipjack is that these were fed small pelagic fish, the main food of Peruvian bonito, and that food type has in experiments, along with temperature, a stronger effect than the taxonomic affinities of the investigated fish (Durbin et al. 1983). a These results are reported again in Vildoso (1962), along with the remarks that (i) the mean stomach contents of bonitos are usually proportional to their weinht, although there is much variability; (ii) females of a given length have smaller stomach wntents than males of the same length; (iii) stomach content weights of more than 50 g always consist of anchoveta; (iv) the size of anchoveta in bonito stomach is rarely less than 8 cm and that (v) the highest stomach content observed was 500 g, and the bulk of it consisted of 52 identifiible anchoveta."},{"index":4,"size":34,"text":"See Table 8 for more detail on this study, notably for stomach contents expressed in % of body weight. Also note that field data were wllected from April 1967to September 1972(without samples in 1970)."},{"index":5,"size":193,"text":"C Note that the miscellaneous fishes reported by Mayo (1976) from bonito stomachs all tended to be larger than anchoveta, and ranged between 13 and 31 cm. Durbin et al. (1983), based on Fig. 2 in Magnuson (1969), estimated a value of 8.5 (per day) for the instantaneous rate of stomach evacuation (R) in K. pelamis, held at a temperature of about 250C. They also estimated from data on a number of fish species a mean value of b = 0.1 15 for the exponent of a generalized relationship linking, in fishes, stomach evacuation rate (R) and temperature (T) of the form Solving for the value of R = 8.5 gives a = 0.48. Magnuson (1969) used fish between 39 and 50 cm, with a mean weight of 1.6 kg. We shall assume that the relationship between R and body weight is, as most other physiological processes controlled by metabolic level, i.e., by the relationship between body weight and gill surface area (Pauly 1981). This relationship in skipjack is Gill surface area = w0ag5 a Adapted from Table VII in Canal (1974); note that fish with stomachs considered \"empty\" are not included here."},{"index":6,"size":28,"text":"Original data were grouped in weight classes; presented here are the geometric means of the lower and upper class limit (except No. 20, which is the actual weight)."},{"index":7,"size":56,"text":"Weighted mean content of stomachs that were not empty. Note that this value refers to anchoveta plus a number of other organisms, notably the fishes Odontestes (= Austramenidia) regia, Trachurus murphyi, Nonnanichthys crokeri and Prionotus (quiescens?). Other organisms identified were Loligo sp. (but probably not L. opalescens, as report*), Octopus sp., Munida cokeri and unidentified euphausids."},{"index":8,"size":100,"text":"The value of m (mean stomach content, in % BWD) in the text considers the stomachs reported to be empty (see Table 9, footnote b) and is thus m = 1.6045 (1 -0.514) = 0.78 i.e., m = 0.8%. (Muir 1969;Muir and Hughes 1969) which would lead to a decline of R with weight such that Gooding et al. (1981) report a positive exponent for the respiration-weight relationship of skipjack, but also note that their \"weight coefficient is opposite in sign from that typical of fishes (and of organisms, generally)\" for which reason their results will not be considered further."},{"index":9,"size":38,"text":"Combining equation ( 17) with equation ( 15), and solving for the mean fish weight in Magnuson's experiments leads to as generalized equation relating evacuation rate, body weight and temperature in skipjack and, by inference, also in bonito."},{"index":10,"size":211,"text":"% of empty f40. of fish Species Area stomachs sampled Source  X) Oliphant (1971) Ancieta (1945, as  cited in Ancieta 1964) Canal (1974) this studyb a None of these studies provides data on seasonal changes of the % of empty stomachs. Estimated by taking the mean % (weighted by sample size) of the four values reported from Peru. Elliott and Persson (1978) derived a model in which food consumption (Q) is computed as the product of mean stomach content (m) and instantaneous stomach evacuation rate (R), or which applies to a given fish size (age). To estimate food consumption per unit biomass (QIB) of an age-structured fish population, equation ( 19) can be extended to a model of the form tr where Rt is the stomach evacuation rate expressed as a function of age (obtained by insertion of equation (15) into equation (9)), Wt the age at age t, see equation ( 9), tr and tmax are the age at recruitment and the maximum age reached by bonitos respectively, mt the mean stomach content expressed as a function of age (obtained by multiplying Wt by the mean stomach content expressed as a fraction of body weight, here 0.008 (see Table 8), and Nt is the number of fishes in the population."}]},{"head":"Sarda orientalis","index":173,"paragraphs":[]},{"head":"Sarda orientalis","index":174,"paragraphs":[]},{"head":"Sarda chiliensis lineolata","index":175,"paragraphs":[]},{"head":"Sarda chiliensis chiliensis","index":176,"paragraphs":[]},{"head":"Sarda chiliensis chiliensis","index":177,"paragraphs":[]},{"head":"Sarda chiliensis chiliensis","index":178,"paragraphs":[]},{"head":"Sarda chiliensis chiliensis","index":179,"paragraphs":[]},{"head":"Sarda chiliensis chiliensis","index":180,"paragraphs":[{"index":1,"size":17,"text":"The simplest fashion by which Nt can be modelled realistically is by using Nt = e -Z(t-tr)"},{"index":2,"size":8,"text":"with N = 1 when t = tr."},{"index":3,"size":34,"text":"Age at \"recruitment\" will be set here at tr = 2, nearly corresponding to the age at which cerrajon (i.e., middle-sized bonito) reach 50 cm, and hence recruit into the stock of \"large bonitos\"."},{"index":4,"size":99,"text":"All terms of equation ( 20) are available as continuous functions oft, and can be numerically integrated. The integrations were performed using a BASIC program (available from the first author) on a monthly basis, i.e., using the temperature value for the month in question and monthly estimates of \"F\", the latter values being computed, as defined in equation ( 14) from the monthly biomass of a given year in Table 6, and the annual catch for the same year in Table 1. (Note that the monthly values of F so obtained are not necessarily proportional to monthly fishing effort)."},{"index":5,"size":43,"text":"The estimates of Q/B obtained through equation ( 20) (see Table 10) are well within the range of rations estimated for tunas (Table 1 I), while being, as should be expected, above values typical of less active fishes (see Durbin et al. 1983). "}]},{"head":"Monthly Anchoveta Consumption by Bonito, 1953 to 1982","index":181,"paragraphs":[{"index":1,"size":31,"text":"There are two basic approaches by which the anchoveta consumption by bonitos in the Peruvian upwelling system could be computed for the period 1953 to 1982, using the information presented above."},{"index":2,"size":89,"text":"One, rather straightforward, could consist of multiplying, for each month, the biomasses in Table 6 by the Q/B estimates in Table 10, then multiplying these estimates of absolute food consumption by some constant factor expressing the average proportion of anchoveta in the diet of bonito.  The other, more elaborate alternative, could consist of deriving a density-dependent model of anchoveta predation by bonito similar to the ones developed for the mammals and the birds of the Peruvian upwelling ecosystem (see Muck and Fuentes, and Muck and Pauly, respectively, this vol.)."},{"index":3,"size":71,"text":"We have opted for an intermediate approach which acknowledges the feature that bonito, as opposed to air-breathing birds and mammals can pursue their prey into any depth and location along the coast, and hence are able to maintain a relatively constant proportion of anchoveta in their diet, yet takes into account the major reduction of anchoveta biomass which occurred in 197211973 (see Table 12). The results are presented in Table 13. "}]},{"head":"Discussion","index":182,"paragraphs":[{"index":1,"size":42,"text":"Although the data available for this investigation are somewhat spotty, they provided results that are comparable with those obtained elsewhere with comparable fishes and which match what might have been expected, given the previous knowledge on the Peruvian bonito and its fishery."},{"index":2,"size":24,"text":"Growth parameters were obtained which match those in other Sarda stocks, while similar estimates of natural mortality were derived, using two widely different methods."},{"index":3,"size":47,"text":"Also, two similar estimates of MSY (82.4 x 103 and 73.6 x 103 t/year) were obtained, along with corresponding estimates of optimum fishing mortality, both surplus production models suggesting that the bonito stock off Peru became seriously overfished in the 1960s, and virtually, collapsed in the 1970s."},{"index":4,"size":76,"text":"This decline through overfishing is consistent with the observed decline of mean size (length and weight) occurring during the period considered here (Table 1). A similar decline, in fact seems to have also occurred in the Chilean bonito fishery (Serra et al. 1980). Thus, we see no reason to agree with earlier statements to the effect that the reduction of the mean size of bonito was caused by the decline of the anchoveta, their favorite prey."},{"index":5,"size":49,"text":"It can be expected, obviously, that the availability of anchoveta has had an effect on the abundance of bonito. However, such effect might be indirect, e.g., through the production of eggs by the bonito spawning stock and the survival of juveniles (on which absolutely no quantitative data are available)."},{"index":6,"size":112,"text":"The estimates of food consumption per unit biomass (QJB) appear reasonable when compared with QB estimates in other scombrids. These estimates, when multiplied with the estimated biomass of large bonito indicate that, at their peak, the large bonitos off Peru devoured approximately 500,000-700,000 t/year of anchoveta. These values make bonito an anchoveta predator whose impact is between that of the fish-eating birds and that of-he seals (see Muck and Pauly this vol. and Muck and Fuentes, this vol.). Interestingly, it also turns out that the \"educated guesses\" of anchoveta consumption by bonito, cited in the introduction, were of the right order of magnitude, even though they were based on rather tenuous assumptions."}]},{"head":"Introduction","index":183,"paragraphs":[{"index":1,"size":52,"text":"Sardine (Sardinops saga.), mackerel (Scomber japonicus) and horse mackerel (Trachurus murphyi) are important predators of Peruvian anchoveta (Engraulis ringens). Sardine has been shown to prey on anchoveta eggs (Santander et al. 1983) while the two other species have been found to ingest anchoveta of all sizes (see Muck and Sanchez, this vol.)."},{"index":2,"size":95,"text":"Information on the biomass of sardine, mackerel and horse mackerel off Peru are available only since 1977 (see Muck and Sanchez, this vol.). Annual catch statistics are available for earlier years but complementary information (such, e.g., as detailed effort data) are missing which, after calibration, could be used to turn these earlier catch figures into absolute biomass estimates for each of these three species. On the other hand, egg and larval surveys have been conducted since 1964 whose results could be used to infer the relative abundance of sardine, mackerel and horse mackerel off Peru."},{"index":3,"size":23,"text":"The present contribution presents a preliminary analysis of data obtained from the surveys, with emphasis on two questions: i) Were earlier abundances (1964)(1965)(1966)(1967)(1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975)(1976)  "}]},{"head":"Introduction","index":184,"paragraphs":[{"index":1,"size":73,"text":"The present contribution is an attempt to estimate the consumption of anchoveta (Engraulis ringens) by mackerel (Scomber japonicus or \"caballa\", Fam. Scombridae) and horse mackerel (Trachurus murphyi or \"jurel\", Farn. Carangidae), two important pelagic fishes of the Peruvian upwelling ecosystem. Beyond catches, ichthyoplankton samples and some scattered biological information, very little is available on the stocks of mackerel and horse mackerel before 1977, when echo-acoustic surveys aimed at stocks other than anchoveta began."},{"index":2,"size":78,"text":"Thus, a model was constructed which simulate biomasses and overlap with anchoveta as a function of sea surface temperature (SST), and anchoveta consumption as a function of predator biomass, a temperature and size-dependent ration and estimates of anchoveta vulnerability and availability. The presentation of the data used for parameterization of the model also provides an opportunity to review some major aspects of the biology of mackerel and horse mackerel off Peru, and to present some previously unpublished data."}]},{"head":"Food and Feeding Habits of Mackerel and Horse Mackerel, with Emphasis on the Role of Anchoveta","index":185,"paragraphs":[{"index":1,"size":45,"text":"Stomach contents of mackerel off Peru have been studied by Miiiano and Castillo (1971), Chavez (1976), Ojeda and Jaksio (1979) and Mendo (1984), among others; the same has been done for horse mackerel by Rosario (1970), Videla (1976), Konchina (1981) and Sanchez andMuck (1983, 1984)."},{"index":2,"size":209,"text":"From these studies, it clearly emerges that both species are opportunistic predators, feeding exclusively on anchoveta when these are abundant and vulnerable and switching to a diet dominated by zooplankton (mainly copepods and euphausids) when anchoveta are less abundant and/or vulnerable (Fig. 1, Tables 1 and 2). Predator size is also important; mackerel switches from planktivory to piscivory (given that anchoveta are both abundant and vulnerable) at a length of about 28 cm (i.e., 300 g and 3 years), horse mackerel at a length of about 30 cm (i.e., 400 g and 2.7 years). Between these sizes and their maximum sizes of about 40 and 70 cm, respectivey, mackerel and horse mackerel consume anchoveta with sizes such as also caught by the purse seine fishery (Fig 2), as is also the case for the other predators in the Peruvian upwelling ecosystem (see Muck and Pauly, this vol., Muck and Fuentes, this vol. and Pauly, Vildoso et al., this vol.) Anchoveta egg predation by these two species, although not quantified here may be substantial. Thus, off Chimbote, where such predation appears to be highest, 40% of the stomachs sampled between September and December 1967 were positive for anchoveta eggs, with a mean of 517 f 427 (s.d.) eggs per stomach."},{"index":3,"size":41,"text":"Anchoveta predation by mackerel and horse mackerel appears to occur with about equal intensity along the coast of Peru, except for the area off Paita, i.e.,north of the area where anchoveta concentrate (Table 3 and see Pauly and Tsukayama, this vol.)."},{"index":4,"size":68,"text":"Note, however, that during the 1982-1983 El Nifio, the diet of mackerel and horse mackerel along the whole Peruvian coast resembled that off Paita (at any time) in that it consisted overwhelmingly of macrozooplankton (Fig. 1). Overall, from south of Paita to the Chilean border, anchoveta provided in 1979, a non-El Ni5o year, about 40% of the food of mackerel and 75% of the food of horse mackerel."},{"index":5,"size":42,"text":"Table 1 shows that the value of 40% for mackerel in 1979 is intermediate between the high value for 1976 (79%) and the low value for 1982 (2%), and that the trend itself is a function of relative anchoveta biomass (see below)."}]},{"head":"Daily Food Consumption of Mackerel and Horse Mackerel","index":186,"paragraphs":[{"index":1,"size":108,"text":"Table 4 presents data for 1976-1982 on the weight of anchoveta in \"anchoveta positive\" mackerel stomachs from which a mean anchoveta food bolus of 23.5 g can be derived,  3 of Muck and Pauly (this vol.). d ~n 1978, off Callao, 22% of the \"anchoveta positive stomachs\" contained 4 or more anchoveta, with a maximum of 7, weighing a total of 86.4 g in a 827 g mackerel. The maximum % of anchoveta weight VS. predator weight was 15.4 and occurred in a mackerel of 543 g.   24 in Tsukayarna and Palomares, this vol.).   The coefficient of variation (C.V., here assumed constant) is thus (23 100)/70 = 33%."},{"index":2,"size":94,"text":"Seasonal changes between low winter @ = 70, s.d. = 23) and high summer biomass within 40 miles from the shore are modelled through consideration of the fact that summer biomasses are about 2.5 times higher than in winter (see above). Thus, 1.1 x 2.5 = 2.8 t x 106, corresponding to 25.5% of the total biomass of 11 t x 106. The value of 25.5% corresponds to 125.5% = 0.65. The density maximum (i.e., x) thus occurs 51 miles offshore because and the s.d. value is 16.8, because C.V. = 33% (see above)."},{"index":3,"size":80,"text":"As mentioned previously, mackerel and horse mackerel move, during El Nifio events, very close to the coast and catch statistics for 1972 and 1976 (Fig. 3) suggest that their inshore biomass increased 3.5-fold when temperature was highest. Thus 1.1 x 3.5 = 3.9 t x 106 occurred within the -MAR, comsponding to 35% of the 11 t x 106 stock. This corresponds to 135% = 0.36, with X and s.d. estimated, as shown above, as 45 and 14.8 miles, respectively."},{"index":4,"size":119,"text":"Similarly, 4 t x 106 of mackerel and horse mackerel occurred within the MA& in March-May 1983, i.e., during the 1982-1983 El Nifio. This corresponds to 140% = 0.22, X = 40 and s.d.  Table 9 gives values of z, Dmax and the corresponding % of mackerel and horse mackerel within 40 miles (B%), for temperatures ranging from 16 to 250C. The regression equations expressing the relationship between temperature and B% is B% = 0.948 exp 0.158T ... 9) which can be used to estimate B% from T without using the z-transformation. B% can be turned into absolute biomass for the area ranging from 4 to 140S, i.e., within the MAR, by using the appropriate factor in Table 7."},{"index":5,"size":94,"text":"As an alternative to this model, which assumes a constant overall biomass of mackerel and horse mackerel, and a highly variable overlap with the MAR, we have also estimated the anchoveta predation that would have been generated by these two fishes, had their biomass within the MAR been equal to their catch, which is largely taken within the MAR. These estimates shall be called here \"least predator biomass estimates\" to emphasize the fact that they provide lower limits to the biomass that mackerel and horse mackerel may have had during the period considered here."}]},{"head":"Food and Anchoveta Consumption By Mackerel and Horse Mackerel","index":187,"paragraphs":[{"index":1,"size":26,"text":"The total daily food intake of a population (Rtot) can be expressed as the sum of the intake by each age (or weight) group (Rt), i.e."}]},{"head":"tmax max","index":188,"paragraphs":[{"index":1,"size":9,"text":"Rtot= Z Rt = z N t 0 CW?"},{"index":2,"size":39,"text":"... 10) t = l t = l Table 9. Relationship between coastal SST and distance offshore of main mackerel biomass (Dm,), z-value and the derived % of overall mackerel biomass occurring within 40 miles of the coast (B%)."}]},{"head":"SST (\"c)","index":189,"paragraphs":[{"index":1,"size":12,"text":"Dm,, za B% a ~r o m Table 13 in Sachs (1978)."},{"index":2,"size":105,"text":"where Nt is the number of individuals with age t, Wt their mean body weight, and c and d constants of the relation linking ration and body weight (Table 10); the equation can be solved in relative terms, by setting No = 1 for t = 0 in the equation where Z is the total mortality (Table 10). For a given biomass B, relative abundance of fish of age t can be turned into absolute abundance (Nt) using Weight (Wt) at age (t) was estimated using the von Bertalanffy growth equation  c, factor (-)  d, exponent (-) total mortality, Z (y-l) p m Mendo (1984)."},{"index":3,"size":6,"text":"From Dioses (1986). ;~rorn Tsukayama (1983)."},{"index":4,"size":38,"text":"T. Dioses (1983 and pers. comm.). ;~rorn Mendo (1983), mean of values for Callao andChimbote for 1976-1978. From Sanchez andMuck (unpublished data). p m Hatanaka and Takahashi (1956) and Hatanaka et al. (1957), Conover (1978) and see text."},{"index":5,"size":3,"text":"From Tsukayama (1986)."},{"index":6,"size":23,"text":"and the length-weight relationship whose parameters (L, , K, to, a and b) are given, for mackerel and horse mackerel, in Table 10."},{"index":7,"size":252,"text":"Thus, our model for esmating total daily consumption of a predator population with tmax = 12, and a total biomass B is Table 11 gives a computation example for mackerel, based on a biomass of 2 t x 106; note that for actual computations, the values of B used referred only to the MAR, as discussed above. Monthly anchoveta consumption by mackerel and horse mackerel within the MAR may now be viewed as a fraction (Ra(i)) of their total ration (Rtot(i)) in month (i). Ra(i) can be further viewed as depending on anchoveta availability (Ai), i.e., Anchoveta availability can be viewed, finally, as the product of relative anchoveta biomass (Br(i ) and an anchoveta vulnerability index (Vd(i)), as in Muck and Fuentes (this vol.) and k Muc and Pauly (this vol.), i.e., Thus, decreasing anchoveta biomasses and/or decreasing vulnerability reduces anchoveta availability and hence anchoveta consumption by mackerel and horse mackerel. Unfortunately, information is insufficient to treat vulnerability as a true variable; thus, the parameter Vul in Equation ( 17) has been set constant, assuming that, on the average, the maximum percentage of anchoveta in the diet of the two predators considered here is 80% when anchoveta are very abundant (Br(i) = 1). The estimation of A(i) for values of Br(i < 1 was done using the data of 2 Table 2. From these, daily anchoveta consumption (Ra(i)) an its relationship (Ra%i) to daily total consumption (Rtot i ) can be computed using Equation (1) with th = 17 and Equation (2)."}]},{"head":"V","index":190,"paragraphs":[{"index":1,"size":125,"text":"This leads to the data in able 12, which show Ra%(i) values decreasing from 72% in 1976 to 1.5% in 1982, in relation with the decline of anchoveta biomass, and as also observed in Table 1. The data of Table 12 have been used to estimate, for absolute anchoveta biomasses Ba(i) ranging from zero to 9 t x 106 the empirical equation with Br(i) set equal to 1 when Ba(i) > 9 t x 106. Anchoveta consumption by mackerel and horse mackerel of 3 years or more was estimated on a monthly basis, using this and preceding equations, as well as the preliminary anchoveta biomass estimate in Muck and Pauly (this vol.) and the SST values in Table 2 of Pauly and Tsukayama (this vol.)."},{"index":2,"size":47,"text":"Additionally, estimates of anchoveta consumption based on the \"least biomass estimates\" mentioned above were obtained, considering that about 75% and near 100% of the fishery catches of mackerel and horse mackerel, respectively, consisted of fish of < 30 cm, i.e., of potential anchoveta predators (Menz 1983;Tsukayama 1986)."},{"index":3,"size":56,"text":"Least annual anchoveta consumption by mackerel and horse mackerel (LC) was estimated from their respective annual catches (Ymack and Yhone through where the value of 0.05 is the mean daily ration as a fraction of body weight and A(i mean) is the mean anchoveta availability computed, for each year, as the mean of monthly A(i) values. "}]},{"head":"Results and Discussion","index":191,"paragraphs":[{"index":1,"size":151,"text":"Tables 13 and 14 give our estimates of monthly anchoveta consumption by mackerel and horse mackerel, respectively, for the years 1953 to 19823 These data are also plotted for both species combined, on an annual basis in Fig. 5, which also shows, for comparison, annual fishery catches of anchoveta and anchoveta consumption by the birds. As might be seen, our estimates of anchoveta consumption by mackerel and horse mackerel are, throughout, well above the anchoveta consumption estimates for the birds, and ipso facto well above the corresponding estimates for bonito (see Pauly, Vildoso et al., this vol.), and dwarf those for the seals (Muck and Fuentes, this vol.). This is also true for the \"least consumption estimates\" based on catches instead of biomasses (Table 15), and which have been also plotted onto Fig. 5, for the years 1974-1982, i.e., for the period with an aimed fishery for mackerel and horse mackerel."},{"index":2,"size":116,"text":"It thus appears that mackerel and horse mackerel jointly create an enormous predatory pressure on anchovetaan interesting finding in view of the fact that anchoveta predation by the guano birds of Peru has generated to date a considerable literature, to which there is no equivalent as far as Scornber japonicus and Trachurus muphyi are concerned. Some details on the seasonal and interannual variability of anchoveta predation by these two species are provided by Fig. 6, which shows species-specific anchoveta consumption on a monthly basis. As might be seen, anchoveta predation by mackerel is far below that for horse mackerel, but both species react similarly to changes of SST, thus reflect El Nifio events in similar fashion."},{"index":3,"size":59,"text":"We also ran our model for the period from January 1983 to July 1986, using the SST data in Table 4 of Muck et al. (this vol.). There was a reasonable agreement between the model estimates of Dm= (50 and 70 miles for the summer and winter seasons, respectively) and the maxima of the observed density distribution (Fig. 7)."},{"index":4,"size":52,"text":"On the other hand, our model, although it suggested a biomass reduction, failed to predict the extremely low mackerel and horse mackerel biomass of 57 x 103 observed during a survey conducted by IMARPE in March-May 1985. This is probably due to the extremely cold water (<I 60C) observed during this period."},{"index":5,"size":97,"text":"Overall, we interpret these findings as supporting our concept of a temperature-mediated inshore-offshore migration and its implication concerning changes in anchoveta predation. Ursin (1980) suggested anchoveta predation on the eggs of their potential predators as a possible key regulatory mechanism in the Peruvian upwelling ecosystem, and Muck et al. (this vol.) have confirmed this for sardine (Sardinops sagax). Sardine spawn from 5 to 120S, within 40 miles off the coast (Santander and Flores 1983), i.e., well within what we have termed here the Main Anchoveta Area (MAR). This is not so for mackerel and horse mackerel (see   "}]},{"head":"Introduction","index":192,"paragraphs":[{"index":1,"size":125,"text":"Recruitment is one of the most important but least understood components of fish population dynamics. Several authors have tried to empirically relate larval survivorship with environmental parameters (Bailey 198 1;Lasker 198 1). These efforts to relate recruitment variability with environmental variability have not been notably successful due to the very short data series generally available for empirical analysis (Bakun 1986). Major recent studies have indicated environmental-biological linkages for pelagic and demersal species of the California Current, Canary Current and Guinea Current (see Bakun 1986). Wind induced Ekman transport, a turbulent mixing index (wind speed cubed) and temperature also have been used in several studies of environmental effects on fish stocks (e.g., Collins and MacCall 1977; MacCall 1980;  Bakun and Parrish 1980; Mendelsohn and Cury 1987)."},{"index":2,"size":93,"text":"Similar studies have yet to be done for the Peruvian anchoveta (Engraulis ringens) which in the early 1970s generated the highest catch of any single-species fishery in the world (Tsukayama and Palomares, this vol.; Castillo and Mendo, this vol.), and whose biomass was well over 20 t x 106 by the end of the 1960s (Pauly, Palomares and Gayanilo, this vol.). The collapse of the anchoveta fishery has been attributed both to overfishing and to the influence of El Nifio which may have changed the usual spawning patterns of the anchoveta (Santander 1980)."},{"index":3,"size":104,"text":"The biology of the anchoveta is rather well documented (see relevant contributions in this vol.). Spawning in anchoveta tends to have two peaks per yearone in the austral winter-spring and the other in austral summer. The major spawning concentrations usually occur near the coast up to around 105 krn offshore. Since the 1960s, the Institute del Mar del Peru (IMARPE) has carried out anchoveta egg and larval surveys (see Santander, this vol.), but these data do not necessarily indicate variations in anchoveta recruitment due to density-dependent egg cannibalism, other sources of egg mortality and fluctuations in survival of the larvae (see Pauly, this vol.)."},{"index":4,"size":51,"text":"Several authors (Santander 1980;Mendiola and Ochoa 1980;Tsukayama and Alvarez 1980;Santander and Flores 1983) have presented evidence showing that during El NGo years, primary productivity decreased and the periods of peak spawning changed resulting in lower recruitment. These two factors have been thought to be the major environmental influences on recruitment variability."},{"index":5,"size":76,"text":"In this paper we perform an exploratory data analysis of the estimated recruitment and biomass series of Pauly, Palomares and Gayanilo (this vol.), of the egg production series in Pauly and Soriano (this vol.) and of several of the various environmental series gathered in this volume. Our interest is in determining if it is possible to forecast anchoveta recruitment at least three months ahead of time, a forecast that would be useful in managing the fishery."},{"index":6,"size":106,"text":"Exploratory data analysis has the advantage in this situation where little is known in that we do not try to force the data into models that may not agree with the structure of the observed data. Despite the variety of fisheries models available, there is no \"physics of fisheries\" in the sense of known laws that have been experimentally verified. Letting the data lead us to a model structure leaves us open to alternative causal explanations as well as providing a check on the realism of the estimated series, identifying outliers in the data and flagging data points that may be overly influential in the analysis."}]},{"head":"The Data","index":193,"paragraphs":[{"index":1,"size":78,"text":"The anchoveta population data used are the estimates presented by Pauly, Palomares and Gayanilo (this vol.). The environmental series consisted of sea surface temperature (SST) and oceanic wind speed cubed and offshore transport derived by Bakun (this vol.), as well as hourly wind direction and intensity recorded at Trujillo and Callao airports by CORPAC (Corporacion Peruana de Aviacion Comercial). The Trujillo and Callao data are described further in Mendo et al. (1987) and Mendo et al. (this vol.)."},{"index":2,"size":89,"text":"Offshore transport at both Trujillo and Callao were calculated as in Bakun (1973 and this vol.), as was wind speed cubed for both locations. Peterman and Bradford (1987) report using an index reflecting Lasker's hypothesis (1978) to predict the survival of anchovy Engraulis mordax larvae off California. A similar index was computed for both Trujillo and Callao; it measures the number of 4-day periods during which the wind speed did not exceed 5 rn/sec. These 4-day periods are here called \"Lasker events\", as suggested by D. Pauly (pers. comm.)."},{"index":3,"size":92,"text":"Spectral analysis showed that the Trujillo and Callao wind series were not significantly coherent at all frequencies. The Trujillo wind series were only coherent with the oceanic wind series at a frequency of 6 months (though out of phase) and at frequencies longer than a year. The oceanic series show a pronounced seasonal cycle, the Callao series a smaller seasonal cycle and the Trujillo series showed essentially no seasonal cycle. The general impression is of strong local differences suggesting the need for care in choosing what variables to use in the analysis."},{"index":4,"size":220,"text":"The Lasker events were problematic at best. At Callao, the wind always satisfied the criteria, while at Trujillo most months had a zero incidence of the necessary periods of calm, especially during 1955-1970, when recruitment was at high levels (see below and Table 1). The estimated recruitment following the periods of zero counts vary greatly, making Lasker events of little use for these periods. For the periods where numerous Lasker events occurred, examination of the data revealed that a high number of events tended to precede low rather than high recruitment. Lasker's hypothesis is based on experience in California where turbulence disperses the food necessary for larval survival; in Peru the food concentration in the mixed layer may be high enough for turbulence not to have that great an effect. All that can be said with certainty from this is that the situation in Peru appears to differ from that in California. Based on this preliminary analysis of the environmental series, we restricted our attention to the wind variables at Trujillo. The Trujillo station has no nearby topographic interference, such as mountains, and the coastline from one degree north of Trujillo to one degree south of Trujillo is almost straight. The area off Trujillo also is one of the major anchoveta spawning areas (Santander and Castillo 1973;Santander 1980, this vol.)."},{"index":5,"size":113,"text":"Localized trends were calculated using the \"lowess\" algorithm of Cleveland (1979) for recruitment, egg and total biomass as well as for offshore transport and wind speed cubed at Trujillo. Both the recruitment and total biomass series exhibit three periods with distinct mean levels. A pre-1960 period with low levels of biomass and recruitment, a period from 1960 to 1972 when both biomass and recruitment were at high levels, and a post-1972 period following the collapse of the fishery (see Pauly, Palomares and Gayanilo, this vol.). On a log scale, for the 1960-1972 period, the trend line for both recruitment and total biomass is almost flat, suggesting random variation around a fixed mean level."}]},{"head":"Recruitment Series","index":194,"paragraphs":[{"index":1,"size":188,"text":"For our initial look at anchoveta recruitment we examined the dynamics of the recruitment series by decomposing it into three componentslocalized trend, seasonal and autoregressive (AR) components (Fig. I), using an algorithm of Gersch and Kitigawa (1983). The variation in recruitment is dominated by longer-term behavior captured in the estimated trend, as the AR component is an order of magnitude smaller than the recruitment series and the seasonal component is almost two orders of magnitude smaller. The estimated local trend (Fig. 1B) suggests three different periods: a low level of recruitment from 1953 to 1958, a higher level from 1959 to 1969 or 1970, followed by a sharp drop in recruitment which preceded: (i) the El Ni5o even to 1972-1973, (ii) the collapse of the fishery, (iii) and by a year, the drop in total biomass. This suggests that the effects of El Nifios on the population dynamics of anchoveta may be more complex than previously thought. The 197 1 recruitment collapse must have been due to overfishing or environmental effects the previous years since the collapse of the fishery appears to have begun before the El Niiio."}]},{"head":"Year Year","index":195,"paragraphs":[{"index":1,"size":102,"text":"The seasonal component (Fig. ID) is very regular but of such small magnitude as to be almost unnoticeable in the overall variation in recruitment. The autoregressive component (Fig. 1C) also suggests three periods: a period from 1953 to 1961 of relatively small variations, a period from 1961 to 1972 of large variations coinciding with the growth of a large-scale fishery and a period from 1972 or so onwards following the collapse of the fishery which is also a period of smaller variations in recruitment. The period of high variability from 1961 to 1972 also is a period of high estimated total biomass."},{"index":2,"size":94,"text":"The autocorrelation function of recruitment both on the normal and the log scale is highly nonstationary and dominated by a value of .95 at a lag of one month. A simple model that includes an autmegressive term of recruitment on itself at lags 1,2,6 and 12 months explains almost 96% of the variance in the monthly series. This agrees with the decomposition of the time series: monthly estimates of recruitment have not increased our degrees of freedom as the series is not independent at monthly time spans; rather, the important variation is between years."},{"index":3,"size":124,"text":"The high degree of autocorrelation in the monthly recruitment series is probably caused by (i) the fact that the sampled length-frequency data may not be accurate enough to resolve monthly differences in population structure so that the size information tends to be \"smeared\" across months, (ii) the fact that Virtual Population Analysis (VPA) uses a series of recursive equations that estimate the present biomass levels from the data from successive periods; these equations will tend to smooth the observed data towards low-order autoregressive behavior, reducing the independence of the monthly data and (iii) the fact that the version of lengthstructured VPA used by Pauly, Palomares and Gayanilo (this vol.) increases the inherent tendency of VPA to smooth data across months @. Pauly, pers. comm.)."},{"index":4,"size":106,"text":"There are several approaches for analyzing the recruitment series given this high degree of autocorrelation. We can low-pass filter the series and then decimate the results to one or two representative values a year and use these data points as the new time series. Alternatively, we can form separate annual series of seasonal segments of the annual cycle and fit separate models across years for each of the aggregate series, comparing the results of the different models. This second approach has an added advantage that it allows us to check for time-dependent relationships in the data. In what follows we restrict ourselves to this second approach."}]},{"head":"Bimonthly Recruitment Series","index":196,"paragraphs":[{"index":1,"size":127,"text":"As an initial examination of the data by months across years, we constructed box plots (Velleman and Hoaglin 1981) for each of the series (Fig. 2). A plus marks the median of the data, the 1's mark the hinges plus or minus 1.5 times the H-spread, where the hinges are the medians of the two halves of the series defined by the median and the H-spread is the difference between the two hinges. The lines beyond the boxes show the outer normal range of the data which is defined as the hinges plus or minus 3 times the H-spread. Values somewhat outside this range are marked with dots and values far outside this range with open dots. Finally, the parentheses denote a confidence limit on the medians."},{"index":2,"size":290,"text":"Recruitment (Fig. 2A) starts out in January at a relatively lower level, increases gradually to a recruitment peak around May-July and then slowly decreases to the end of the year. Egg production, however, has a relative peak around February-March and a much larger peak around September-October. Parent biomass (Fig. 2C) has a large peak in February-March concurrent with the relative peak in egg production but usually has relatively low value during the September-October period. Thus assuming the recruits to be 3-4 months old (see Palomares et al., this vol.), the larger recruitment in June-July would stem from the lesser spawning of February-March, which is based on a large parent biomass. Conversely, the lesser recruitment in the beginning of the year would stem from the major spawning in September-October, which is based on a relatively small parent biomass. The wind-related environmental series (Figs. 2E and  2F) are at a relative minimum during the February-March period, but are at their maxima during the September-October period and thus generate unfavorable environmental conditions. Parrish et al. (1984) also comment on the large austral spring spawning during unfavorable conditions. One possible explanation for a large spawning during unfavorable conditions is that the conditions during this period are favorable for the adults (it is the upwelling period). A. Bakun (pers. comm.) has suggested that such a low degree of apparent adaptation of spawning effort to the normal seasonality of larval survival prospects might be in some way due to the fact that detrimental El Niiio effects may exert strongest impact during the warm austral summer period, thus the effects of intermittent reproductive failures during summer, associated with El Niilo anomalies may have tended to counter adaptive tendencies to concentrate spawning in the normally favorable summer season."},{"index":3,"size":261,"text":"After examination of the recruitment series and the environmental series we decided on bimonthly aggregation of the recruitment series as follows: February-March, April-May, June-July, August-September, October-November and December-January. Note that the December-January aggregate crosses years: December 1953 is combined with January 1954, etc. These pairs of months vary across years in a similar manner and at similar levels (Fig. 3). For [ ! I -predictors we considered parent biomass and egg production lagged 3 and 4 months (the spawning period) as well as the monthly offshore transport and wind speed cubed at Trujillo. Spectral analysis of the transport and wind speed cubed at Trujillo shows that these series are highly coherent (> .9) at all frequencies and thus have essentially the same dynamics due to limited variability in the wind direction (Mendo et al., this vol.). We therefore arbitrarily restrict our attention to offshore transport, though similar results can be found using wind speed cubed, i.e., turbulence. Cross-correlation matrices as well as generalized partial correlations were calculated as in Tiao and Box (1982). Model and variable selection were also examined using the multivariate subset autoregression procedure of Akaike et al. (1979). As it is likely that the relationships between recruitment and the environment is nonlinear, final model selection and identification was done using the AVAS procedure (Tibshirani 1987). This procedure is a modification of the optimal transformation algorithm ACE (Breiman and Friedman 1985) which appears to correct period. The June-July and August-September models, when recruitment appears to be at or near a maximum, appear to anticipate changes in recruitment quite well."},{"index":4,"size":108,"text":"The biomasses and the recruitment estimates for the period 1953 to 1959 were obtained through VPA using values of residual natural mortality (Mo) which could not be calibrated against independent biomass estimates; moreover, during this period, a large proportion of the withdrawals used for the VPA were estimates of anchoveta consumption by birds and bonito (see Pauly, Palomares and Gayanilo, this vol.). These appear to be sufficient reasons for differences between this period and the succeeding ones and thus explain why, in terms of our models, recruitment during this time period behaved in a manner different from the rest of the model period (see also Pauly, this vol.)."}]},{"head":"Discussion and Conclusions","index":197,"paragraphs":[{"index":1,"size":62,"text":"We have shown that while the high degree of autocorrelation in the monthly recruitment data makes it difficult to identify causal models of anchoveta recruitment, we could, however, identify yearly models for bimonthly recruitment series. These models tend to have a similar structure, predict recruitment with approximately equal success and appear promising as a means to forecast future trends in anchoveta recruitment."},{"index":2,"size":121,"text":"The pattern of the cross and partial correlation matrices, as well as several different modelling approaches, have all suggested the same basic models. Thus we feel confident that the relationships described in our models reflect the basic structure of the data. However, there is no clear biological interpretation of this structure. As we have done an extensive amount of searching through different sets of variables and estimating transformations to increase model fit, it is likely that our estimates of the goodness of fit of our models and of the ability of the models to predict future data is somewhat inflated. We would therefore recommend to implement the following steps before attempting to implement a model similar to the one analyzed here:"},{"index":3,"size":38,"text":"(i) Attempt to calibrate the estimated recruitment series with independent data (i.e., data not used in obtaining the estimates) to further verify that the estimated recruitment reflects the actual changes in recruitment, at least on a yearly basis;"},{"index":4,"size":24,"text":"(ii) Calculate offshore transport and wind speed cubed from other stations near Trujillo to check on the accuracy and consistency of this data set;"},{"index":5,"size":40,"text":"(iii) Develop a better mechanistic understanding of the underlying models. We are distrustful of forecasting models that do not have a clear biological interpretation and for which there is no independent evidence for the relationships developed in the exploratory analysis;"},{"index":6,"size":77,"text":"(iv) For log recruitment, use techniques that estimate transformations without transforming the dependent variable, and that allow for greater testing of model parameters, such as GAIM (Hastie and Tibshirani 1986); and finally (v) Use generalized cross-validation or related techniques to test the stability of both the transformations and the degree of fit of the models, in order to get a better idea of the predictive capability of the models for data to be obtained in the future."},{"index":7,"size":80,"text":"Despite these reservations, we feel we have shown that the series of recruitment estimates produced by Pauly, Palomares and Gayanilo (this vol.) have properties which enable them to forecast far enough in advance for consideration in the formulation of management actions. Further, we have indicated the importance of including variables that reflect environmental processes. However, the fact that the resulting models do not appear to conform to any of the conventional hypotheses concerning major influences on recruitment success remains unsettling."}]},{"head":"Introduction","index":198,"paragraphs":[]},{"head":"Evolution of Peruvian Fisheries","index":199,"paragraphs":[{"index":1,"size":38,"text":"Archeological findings in the Illescas peninsula have shown that a number of fishing communities existed in the pre-ceramic era which already made use of important fishing techniques for preserving and designing fishing materials (dying, netting, etc.) (Kostritsky 1955)."},{"index":2,"size":65,"text":"However, the present fishing industry dates back onIy to the early 1950s when the purseseine caught anchoveta began to be processed into fish meal (Schaefer 1967). Fig. 1A shows the evolution of landings since that time and the spectacular growth of the industry until 1970 when total catch of anchoveta reached the world record level for a single species of over 12 t x 106."},{"index":3,"size":55,"text":"Trends in the following decade were equally spectacular, but in the opposite direction as also seen in the 5 ear smoothed curve of Fig. 1B. By 1974, total landings had dropped to 3.8 t x d 106, to 2.5 t x 10 in 1980 and to slightly above 2.8 t x 106'in 1984 (Table 1)."},{"index":4,"size":119,"text":"Graphs illustrating aspects of the Peruvian fisheries are usually very \"spiky\", which is a reflection of the instability and dramatic changes that have taken place following the occurrence of various El Nifio events (see other contributions in this vol.). Intermediate oscillations due to lagged responses of other elements of the altered system also contribute to the observed instabilities in curves describing biomass, catches, effort, capacity, production, prices, etc. Several institutional changes in public administration have also influenced the development of the fishery (Borgo 1966;Malpica 1976;Chaparro 1983;Peralta 1983) generally through interventions aiming to counterbalance the short-run effects of environmental fluctuations but inconsistent with a long-run perspective of renewable resource management (see also Castillo and Mendo, this vol.; Pauly, this vol.)."},{"index":5,"size":29,"text":"Technological developments and improvements in harvesting and processing activities have thus produced negative as well as positive results generating forces for subsequent changes (see Castillo and Mendo, this vol.)."}]},{"head":"The Management Problem","index":200,"paragraphs":[{"index":1,"size":81,"text":"If management of fishery resources is to provide the highest possible benefits for society, economic evaluation of alternative management interventions as well as changes of the basic characteristics of the fishery system are needed. This is due to the fact that the renewable, stochastic and complex nature of the fishery resource system is extremely difficult to apprehend with simple and elementary conceptualizations. Added feedback loops, such as those created by regulatory measures, makes understanding of a complex system even more difficult."},{"index":2,"size":38,"text":"Nonetheless, methods exist that allow for efficient handling of large, interrelated systems. Programming techniques, simulation and impact analyses &e among the available tools which are becoming increasingly applicable, especially through the wide availability of microcomputer hard and software."},{"index":3,"size":108,"text":"This paper discusses the evolution of the Peruvian purse seine fishery from an economic perspective, in order to represent the basic structure of the fisheries as an economic activity and to evaluate the relative costs and benefits of alternative interventions. Following the implicit suggestion of the Second Panel of Experts on the Economic Effects of Alternative Regulatory Measures in the Peruvian Anchoveta Fishery in 1974 (IMARPE 1974), a conditional mathematical model structyred in terms of an optimization problem is presented as a tool for modelling the Peruvian purse seine fishery and for assessing possible outcomes of specific changes of key factors.  From Pauly, Palomares and Gayanilo (this vol.)."}]},{"head":"An Overview of the Peruvian Fishing Industry","index":201,"paragraphs":[]},{"head":"Fishery Resources","index":202,"paragraphs":[{"index":1,"size":145,"text":"The Peruvian fisheries are based mainly on the exploitation of pelagic resources occurrring on 2,800 lun of coastline. The resource base of these fisheries is very diverse and consists of about 500 species (Chirichigno 1974). Of these, the most important are anchoveta (Engraulis ringens), sardine (Sardinops sagax), horse mackerel (Trachurusmurphyi) and mackerel (Scomber japonicus). Fishing became an important economic activity in Peru in the early 1950s when anchoveta began to be processed into fish meal and oil. Since then, fishing activities centered on anchoveta, grew at spectacular rates until overfishing and unfavorable environmental conditions (i.e., the El NiEo of 1971-1972) drove the fishery to collapse (Walsh 1981). The period that followed was characterized by conditions of instability, vulnerability, the occurrence of more El NiEo events and the buildup of large biomasses of sardines, mackerels and horse mackerels (Jordan et al. 1978 and see Table 1)."},{"index":2,"size":47,"text":"Although the compensatory phenomena reported by Jordan et al. (1978) imply a shift from what was essentially a monospecies to a multispecies fishery, total catches have been fluctuating around a much lower level than those reached during the anchoveta peak of the late 1960s and early 1970s."}]},{"head":"Harvesting and Processing Capacity","index":203,"paragraphs":[{"index":1,"size":82,"text":"The characteristics and behavior of fish resources define the technology required for their exploitation. Peruvian anchoveta are harvested by encircling schools with a purse seine of relatively small mesh sizes (10-30 rnrn) of a length of approximately 300 fathoms and a depth of 35 fathoms, and which is set bypurse seiners of 25-30 t LOA driven by diesel motors ranging from 300 to 800 hp. Fishing consists of daily trips with 2 or 3 purse seine sets (Engstrom et al. 1974;IMARPE 1975)."},{"index":2,"size":64,"text":"Important changes in the fleet composition and characteristics have taken place since the beginning of the anchoveta fishery. Fig. 1B shows the rapid increase in the number of vessels which took place since the early 1950s, in response to favorable market and biological conditions (Segura 1973), and which began to be reversed after 1964, after a total of 1,623 vessels was reached (Table 2)."},{"index":3,"size":137,"text":"Simultaneously, technological improvements such as nylon nets, echo sounder, hydraulic systems, steel vessels and better information on stock distribution (fishing strategies supported by planes, radio communication, etc.) have taken place along with changes of fleet size. Estimates of excess fleet capacit show that the fleet in 1970-1971 was able, under average X conditions, to harvest 9.5 t x 10 of fish in 130-133 days, which would have amounted to a fishing season of 6.5 months (IMARPE 1974). IMARPE (1974) estimated that the excess capacity of the fleet during that time was about 30% for an expected catch of 9.5 t x 106. Fig. 1C shows, along with the annual catch per vessel, the trends of mean hold capacity in the period 1970-1980, illustrating further the increase in potential harvest (and excess) capacity of the fleet (Berrios 1983)."},{"index":4,"size":71,"text":"Catches are used primarily as raw material for fish meal production. Efforts to rationalize the industry performance, to increase aggregated value and make better use of fish for direct human consumption have led in the last few years to increased alternative uses of catches, such as canning and landing of frozen fish, especially with regard to the \"new\" pelagic speciessardine, mackerel and horse mackerel (Blondet 1986, and see also Table 3)."},{"index":5,"size":55,"text":"Processing capacity for fish meal also grew at a disproportionate rate during the \"boom\" period of the anchoveta fishery. Estimates of excess processing capacity for 1970-1971 are of similar magnitude than those for the harvesting capacity, creating a feedback force to increase effort in order to obtain larger shares of the total catch (IMARPE 1970)."},{"index":6,"size":53,"text":"The number of canning plants grew from 16 in 1947 to 69 in 1956, that is, more than 300% growth in less than ten years. During the 1976-1982 period, another large increase in the number of plants occurred, from 34 to 82 plants. Plant use capacity was low, however, near 26% (Peralta 1983)."}]},{"head":"Products and Market Conditions","index":204,"paragraphs":[{"index":1,"size":43,"text":"Fish meal is a high protein animal feedstuff obtained by cooking, pressing, drying and grinding fish or shellfish. In this process, fish oil and soluble are obtained as byproducts (Fig. 2) which are used in the production of shortening, compound oil, margarine, etc."},{"index":2,"size":45,"text":"Raw material used in manufacturing fish meal and its byproducts is obtained from fish (usually small pelagics) harvested for this purpose. Also, incidental catch and offal obtained during processing of other fishery products (canning, filleting, etc.) are also used for fish meal and related byproducts."},{"index":3,"size":90,"text":"The value of fish meal as feedstuff component is related to several factors of which the content of amino acids, minerals and range of B-complex vitamins are among the most  a~eriod 1953a~eriod -1958a~eriod (Mills 1969););period 1959period -1973period (IMARPE 1974););period 1974period -1980period (Berrios 1983)). Berrios (1983). Sum of first four columns in Table 1.  Monogastric animals such as pigs and poultry obtain essential amino acids from the breakdown of protein during digestion. In this, fish meal is very superior to vegetable proteins and is thus an excellent feedstuff for these animals."},{"index":4,"size":58,"text":"The demand for fish meal therefore, is a derived demand based on raising animals such as swine and poultry (Capurro 1983). The world feed market is an extremely competitive and complex one and because of its importance on the final consumer demand for animal protein, the outlook for fish meal is favorable over the long run (ADB 1983)."},{"index":5,"size":45,"text":"In the pigfeed market, fish meal competes with soymeal as a source of protein; in the poultry industry fish meal is considered a superior product because of its wide spectrum. Finally, in other minor markets such as commercial fish farming, fish meal has little competition."},{"index":6,"size":23,"text":"The demand for fish meal then depends both on the production of these animals and the elasticity of substitution with its competitive products."},{"index":7,"size":80,"text":"Since 1972, the world fish market has undergone dramatic changes. Although the export market is concentrated in a relatively small number of producing countries with sophisticated fisheries industries accounting for a large percentage of the total market, the heterogeneity of the suppliers prevents this oligopolistic industry to behave as expected. The major exporters, for example, have not succeeded in establishing cartels or other types of mechanism in their favor with the exceptions of some bilateral arrangements between Chile and Peru."},{"index":8,"size":87,"text":"Imports, on the other hand, although concentrated in a small number of countries with a large share of the total, do not show the kind of concentration observed in exports and production. Thus, trading patterns depend greatly on transport costs, supply and demand levels as well as quality and traditional buyerlseller relationships (ADB 1983). Fish meal production in Peru is almost entirely for export. Although a major supplier to the international fish meal market, Peru has been a \"price taker\" when world market supplies have been normal."},{"index":9,"size":103,"text":"Historically, the prices for Peruvian fish meal have been rather erratic. A dramatic increase in 1973, as a result of the fall in the world market conditions, created in part from the Peruvian shortage of supply (Vondruska 1981), was followed by an equally dramatic fall in 1975, followed by a recovery and fluctuations around a higher average until 1985. The fluctuations between 1963 and 1983 ranged from US$l45 to 542/t, which, in constant prices of 1980, corresponds to a range between US$527 and 1,204, the latter value being the peak of 1973. Fig 1D shows the relative instability of Peruvian fish meal prices."}]},{"head":"The Mathematical Conditional Model","index":205,"paragraphs":[{"index":1,"size":95,"text":"The analyses and evaluation of the performance of a particular fishery require examination of the key variables expressing dynamic interactions of a set of complex elements of varying importance. Taken as whole, these elements are like links of a chain, with the performance of the chain depending on the strength of each of the components. Information for management purposes must therefore not only identify, describe and quantify these elements (links) but also establish the functional relationships among them. Only in this way is it possible to describe, explain and predict the outcome of specific interventions."},{"index":2,"size":60,"text":"A bioeconomic model, structured in terms of a constrained optimization linear programming problem is presented below. Constraints are of biological, technological and economic nature. Fundamental relationships between biological characteristics of the resource and the technology of capture are incorporated along with relationships between market conditions and fishing effort, processing techniques and harvest levels, product type and demand levels (Fig. 3)."}]},{"head":"Model Implementation","index":206,"paragraphs":[{"index":1,"size":49,"text":"A Fisheries Net Benefit Function (FNBF) has been established. Each activity of the fishery process impacts the FNBF in a negative or positive way according to whether that activity generates costs or revenues. The absolute difference between total costs and revenues is the net benefit that the fishery generates."},{"index":2,"size":28,"text":"The management problem then is to make FNBF, as great as possible (i.e., to maximize FNBF) without violating the restrictive conditions imposed by the system (i.e., the constraints)."},{"index":3,"size":102,"text":"The model is structured into six sequential blocks (Table 4) defining the different activities involved in the fishery within a time period consistent with the validity of the various parameters, functional relationships and constraints. The objective function summarizes all relevant activities into a single value. Thus we have: Variables in this block represent catches (in t) as segments of the hypothetical sustained yield function (SY) of the exploitable stock. To each level of sustained yield there is an associated yield per unit of effort (Y/f) which defines a cost per unit of catch ( C K ) coefficient in the objective function."},{"index":4,"size":94,"text":"This block contains the effort capacity constraint defined in terms of available vesseVdays (VDA).    The functional relationship between sustained yield, effort and cost per unit of effort is consistent with standard bionomic theoretical concepts, as represented in Fig. 3A and 3B. Note that throughout this contribution, the terms \"catch\" and \"catch per effort\" are replaced by \"yield\" (Y) and \"yield per effort\" (Ylf) such as to avoid confusion with costs (C) and costs per effort (Clf). Note also that using \"yield for catch in weight corresponds to the standard notation compiled by Holt (1960)."}]},{"head":"Processing Block","index":207,"paragraphs":[{"index":1,"size":66,"text":"Variables in this block represent tonnes of fish processed into different product lines with associated yield coefficients and processing unit costs (PUC) in the objective function. This block also contains processing capacity constraints expressed in terms of total available processing capacity (APC, in hours). PUC do not include costs of raw material since these have been incorporated into the harvesting block coefficients of the objective function."}]},{"head":"Transport Block","index":208,"paragraphs":[{"index":1,"size":49,"text":"Variables in this block represent tonnes of final products transported from processing plants to storage facilities. Coefficients in the objective function are the transport unit cost per tonne of final product (TUC). Available total transport facilities (ATT) set the constraint for the maximum transport of products per unit time."}]},{"head":"Storage Block","index":209,"paragraphs":[{"index":1,"size":44,"text":"Variables in this block represent tonnes of final products stored. Coefficients in the objective function are cost per unit storage (CUS) per tonne of product stored. This block also contains restrictions on available total storage (ATS) in terms of storage per unit of time."}]},{"head":"Marketing Block","index":210,"paragraphs":[{"index":1,"size":42,"text":"Coefficients in the objective function associated with this block represent the average cost per unit marketed (CUM) per tonne of final product sold. Balance constraints force all stored products to be sold in order to avoid modelling inventory behavior and accidental losses."}]},{"head":"Sales Block","index":211,"paragraphs":[{"index":1,"size":31,"text":"Variables in this block represent total tonnes of products sold into different markets. Coefficients in the objective function associated with this block are the unit prices (UP) per tonne of product."}]},{"head":"Balance Equations, Auxiliary Activities and Convex Sets","index":212,"paragraphs":[{"index":1,"size":22,"text":"Balance equations are included in the block structure in order to assure flow of product throughout the matrix while avoiding unaccounted losses."},{"index":2,"size":45,"text":"Auxiliary activities are also used to allow transit and distribution of products throughout the matrix without impacting the objective function. Convex set equations are included as means of assuring compliance with segmentation and grids defined by the piecewise linearization technique used in incorporating nonlinear functions."}]},{"head":"Optimal Value","index":213,"paragraphs":[{"index":1,"size":52,"text":"The structure of the problem in terms of activities performed allows for the computation of total expenditures and revenues. All activities implying costs have negative signs in the objective function while activities generating revenues have positive signs. The summation of negative and positive values is the net benefit derived from the fishery."}]},{"head":"Implementation of the Model","index":214,"paragraphs":[{"index":1,"size":58,"text":"Data for the conditional model has been obtained from different secondary sources (IMARPE, unpublished data; various FA0 statistical yearbooks and extrapolation from similar fisheries, such as the Chilean northern pelagic fisheries (Aguero and Adriasola 1983). The base model was specified to represent the situation prevailing in 1982, which is here used as the reference year (see Table 5)."},{"index":2,"size":13,"text":"Several alternative data configurations are presented to represent alternative possible scenarios (Table 6)."},{"index":3,"size":25,"text":"The linear programming algorithm in the SOFPES program of Aguero and Lampe (1986), implemented on an IBM PCXT microcomputer was used to run the model."}]},{"head":"Theoretical Issues and Devices","index":215,"paragraphs":[]},{"head":"Harvesting Block","index":216,"paragraphs":[{"index":1,"size":29,"text":"A hypothetical yield curve was used to simulate alternative levels of sustained yield for the stock of Peruvian pelagics. Linearly decreasing yield per effort coefficients with increasing  -Fishmeal (US$/t):"},{"index":2,"size":8,"text":"-Fish oil (US$/t) -Canned fish (US$/t) Storage sector:"},{"index":3,"size":2,"text":"-Fishmeal (US$/t):"},{"index":4,"size":7,"text":"-Fish oil (US$/t) -No. of fishing dayslyr:"},{"index":5,"size":2,"text":"-Total boatdayslyr:"},{"index":6,"size":10,"text":"-Processing, transport and storage capacity arbitrarily 4 t x lo6"},{"index":7,"size":90,"text":"set at unbinding levels -Catch composition: levels of effort account for the renewable but exhaustible nature of the fishery (Schaefer 1957). The relevant portion of the curve was segmented into discrete steps (grids) of 250,000 t each of which was associated with yield per effort and effort levels as shown in Fig. 3C. Since cost per unit of effort was assumed constant, cost per unit of yield will be an inverse function of yield per unit of effort (Fig. 3D), which in turn will be an increasing function of output."},{"index":8,"size":26,"text":"This technique, derived from piece-wise linearization methods (Duloy and Norton 1975), has been successfully used in integrating nonlinear sustained yield functions using linear programming (Aguero 1983)."}]},{"head":"Processing Block","index":217,"paragraphs":[{"index":1,"size":41,"text":"Harvested fish were transferred into the processing block through auxiliary activities (which have zero impact on the objective function). Within this block, yield leads to alternative product lines (fish meal and/or canning), each one bearing different processing yields and unit costs."},{"index":2,"size":37,"text":"Balance equations assured that all harvested fish were transferred and processed. Loss factors such as those identified by Castillo and Mendo (this vol.) could easily be added in this block to account for leakage in the system."},{"index":3,"size":45,"text":"Estimates of PUC obtained from a detailed analysis conducted for the northern pelagic Chilean fishery and fish meal industry (A uero and Adriasola 1983) were used whenever data from the Peruvian fisheries were unavaila le assuming that reduction plants operated at half their normal capacity."}]},{"head":"%","index":218,"paragraphs":[{"index":1,"size":49,"text":"Two levels of yield coefficients for fish meal ranging from 19 to 24% were used to allow for variations between the different species groups used as raw material; more species groups could be easily incorporated. Fish oil yield level was estimated at 4%, i.e., the average throughout the industry."}]},{"head":"Transport Block","index":219,"paragraphs":[{"index":1,"size":38,"text":"Transport cost coefficients have been determined based on cost estimates of the various means of transport operating between plant and storagelport facilities. Coefficient in the objective function reflect the average cost of transporting 1 t of final product."}]},{"head":"Storage and Marketing Block","index":220,"paragraphs":[{"index":1,"size":49,"text":"Both blocks have similar structures and the corresponding coefficient in the objective function represent estimates of average costs incurred in storing and selling 1 t of product. The marketing block does not need to have constraints since its level can be assumed to be indirectly proportional to total production."}]},{"head":"Sales Block","index":221,"paragraphs":[{"index":1,"size":44,"text":"This block contains the unit price (free on board or FOB) estimated as an average for a 10year period (1973)(1974)(1975)(1976)(1977)(1978)(1979)(1980)(1981)(1982)(1983) for each product line. It is the only coefficient bearing positive sign in the objective function. It represents gross returns (revenues) to the country."},{"index":2,"size":86,"text":"Although Peru is generally a \"price taker\" at the international market, the possibility exists that, at very low or very hiqh levels of demand, Peru's share in the world market might have an impact on prices. To reflectithis possibility, a hypothetical downward sloping demand curve faced by Peru was assumed and linearized piece-wise. Thus, prices are assumed highly elastic throughout a large portion of the hypothetical demand curve but relatively inelastic at very low and very high levels of supply, specially for fish meal and oil. "}]},{"head":"Results and Discussion","index":222,"paragraphs":[]},{"head":"Base Model","index":223,"paragraphs":[{"index":1,"size":129,"text":"Conditions assumed in the base model for the Peruvian pelagic fisheries are described in Table 5. Prices and cost structure resemble those of the early 1980s. A summary of the results (LP solution) for the base model is presented in Table 6 under the column heading BASE MODEL. It shows that (given the conditions in Table 5) the Peruvian pelagic fishery is capable of generating a net benefit of approximatly US$173 million per year from a total catch of 3.5 t x 106. Gross annual revenues are about US$371 million of which US$300 million would be export revenues. However, it costs about US$200 million to harvest the corresponding yield (3.5 t x 106) and to send it to the market as final product through processing, transporting and storage intermediaries."},{"index":2,"size":31,"text":"The fleet capacity required is 87,000 boat-days of fishing. This is equivalent to 364 vessels, operating full time for 240 days per year, i.e., reflecting an excess capacity of about 37%."}]},{"head":"Sensitivity Analysis","index":224,"paragraphs":[{"index":1,"size":16,"text":"In order to foresee possible outcomes of alternative conditions, five different scenarios were modeled (Table 6)."},{"index":2,"size":122,"text":"Two types of changes were assumed: increaseldecrease in vessel technology with related changes in cost structure. increaseldecrease in product prices. Results obtained are presented in Table 6 under the column heading Model 1 to Model 5. The results of Model 1 indicate that a 25% improvement in vessel technology (accompanied by a decrease in cost per unit of yield) would bring an increase in net benefit from the fishery of about 18%, while harvesting the same amount of fish as in the base model. The entire increase of the net benefit in this case would come from the decrease in total cost of aproximatly 16%. However, this would create more idle fleet capacity (50%) as compared to 37% in the base model."},{"index":3,"size":58,"text":"On the other hand, results of Model 2 (25% decrease in vessel technology, accompanied by increase in C/Y) would reduce total catch to 2.5 t x 106 (29% decrease) per year. This would reduce net benefit to US$152 million (a decrease of 12%) and fleet size to 278 boats. Total costs would also be 25% lower (US$120 million)."},{"index":4,"size":45,"text":"Models 3 and 4 show the effects of a change in prices by 25% (increase and decrease, respectively). Price increase has no effect on total catch, fleet size or costs. The only effect observed (for Model 3) is an increase in net benefit of 56%."},{"index":5,"size":75,"text":"The fact that vessels and catch do not respond to increase in price is probably due to the range of optimal values for prices in the objective function. This suggests that the model should incorporate a finer segmentation in the harvesting and processing blocks to allow the algorithm to find a cost/price combination consistent with the proposed price change. Otherwise, a larger increase would be necessary to generate a response of fleet size and yields."},{"index":6,"size":44,"text":"Price decrease, on the other hand, shows an impact on number of vessels, yield, revenues and costs, as shown in Table 6. This shows that the model will respond if the price change is large enough to be located outside the range of optimality."},{"index":7,"size":48,"text":"In Model 5, the results of a substantial increase in vessel technology (300% over the base model) are shown. This change would enable an increase of catch by 57% from the level in the base model, using a total of 382 vessels. Net revenues would increase by 66%."}]},{"head":"Conclusion","index":225,"paragraphs":[{"index":1,"size":59,"text":"The model presented reflects rather closely the basic behavior of the Peruvian fishing industry. Total catch, total revenues, number of vessels, catch rates, final products, yield coefficients, prices and export revenues are similar to the values of these variables for the 1982 year as obtained from the sources consulted (Blondet 1986, FA0 1986 Catch andLandings Statistics Vol. 58, 59)."},{"index":2,"size":29,"text":"Simulated changes in the input values showed movements in expected directions, reflecting consistency and accuracy of the model structure as a tool to predict outcome under alternative simulated scenarios."}]},{"head":"Introduction","index":226,"paragraphs":[{"index":1,"size":75,"text":"The Oxford English Dictionary provides, as one of its definitions for the word \"synthesis\", the \"action of proceeding in thought from causes to effects, from laws or principles to their consequences\". The Dictionary points out, however, that \"different logicians and philosophers, though severally applying the term only in a single sense, are still at cross purposes with each other. One calls Synthesis what another calls Analysis; and this both in ancient and modern times. \""},{"index":2,"size":191,"text":"And so also here: while attempting to present a synthesis of the preceding contributions in this volume, I shall have to perform some further analyses and in fact constantly switch between these two forms of reasoning. However, one limiting factortimeshall overall limit the quantity and quality of the analyses and syntheses presented here. The book of which this contribution represents the final chapter is to be presented at the 2nd Latin American Marine Science Congress to be held from 17 to 21 August in Lima, Peru. It is also to provide a background to a workshop on \"Models for Yield Predictions in the Peruvian Upwelling Ecosystem\", to be held immediately following the Congress. Some of the contributions included in the present volume had been available to the editors in early 1986; the bulk of them became available in early 1987, however, with a few coming in as late as mid-May. The present contribution, written under considerable time pressure could thus not consider more than a few obvious items; I hope the workshop itself and interested readers will fill the gaps using the time series and other data included in this book."}]},{"head":"Review of Some Important Questions","index":227,"paragraphs":[]},{"head":"On Major Trends in the Data Sets","index":228,"paragraphs":[{"index":1,"size":126,"text":"When referring to the Peruvian upwelling ecosystem, the authors of both popular and scientific papers usually use the 1972 collapse of the anchoveta fishery and some related phenomena (e.g., the collapse of the bird populations) as illustrations of the key changes that occurred in that system (see Fig. 1 and Walsh 1981). However, as demonstrated by the contributions of Bakun (this vol.) and Mendo et al. (this vol.), the winds off Peru have intensified since the 1950s, resulting in increased turbulence (\"bad\" for first-feeding larvae?) and increased upwelling (\"good\" for adult anchoveta?). These trends, and their possible consequences (reduced anchoveta recruitment and biomass, improved conditions for the growth of adults) are matched, overall by our findings regarding the biology of anchovetabut the mechanisms are not obvious."},{"index":2,"size":182,"text":"Thus, mean annual anchoveta recruitment does not correlate directly with any of the windderived indices presented in this volume (see Mendelsohn and Mendo, this vol.). There is on the other hand a correlation of r = 0.359 between log upwelling index (mean annual values, from Table 5 in Bakun, this vol.) and the annual values of the anchoveta growth performance index $' (from Table 3 in Palomares et al., this vol.), which, with 27 d.f. is close to being significant (critical value is r = 0.367 for 5% level). This is much less than the value of r = 0.433 for the correlation of #' against anchoveta biomass discussed in Palomares et al. (this vol.), but indicates the possibilitywhich should be followed upof interactions between various trends, and that decadal increase in anchoveta growth performance may be due to more than strict densitydependence. Further investigations along these lines should consider, moreover, competition with sardine, whose egg abundance is closely related to anchoveta abundance (Fig. 2), more so than suggested by Fig. 1 in Muck et al. (this vol.) which shows untransformed variables."}]},{"head":"On Equilibria and Multiple Steady States","index":229,"paragraphs":[{"index":1,"size":237,"text":"None of the contributions included in this volume is structured around the assumption that the Peruvian ecosystem is in \"equilibrium\", or had reached at some point a \"steady state\" (although this assumption has been used in some cases to estimate ancillary variables, e.g., the gear selection curve for anchoveta, see Palomares et al. (this vol.) or fishing effort on bonito in Pauly, Vildoso et al., this vol.). It is apparent, on the other hand, that the period from 1972 to the present differs from the period say from 1958 to 1971: not only were the anchoveta egg standing stocks, the survival of the prerecruits (Fig. 3) .and the biomasses different, but the structure of the ecosystem itself appears to have changed (Walsh 1981). Thus, a system overwhelmingly dominated by anchoveta was replaced by a system in which anchoveta and sardine compete for dominance as regulated by small temperature shifts (Zuzunaga 1985;Villavicencio and Muck 1985) in a fashion reminiscent of the description of Skud (1982). This suggests the existence of at least two possible \"states\" for the Peruvian ecosystem, each with its own \"ascendency\" and related properties (Ulanowicz 1986 and see Table 1). Santander and Castillo (1973) reported that, at temperature between 14.9 and 16.90C, fertilized anchoveta eggs need 50 hours to develop and hatch; they also noted that off the Peruvian coast, anchoveta eggs occurred from 13.5 to 18.50C, with a pronounced maximum from 15 to 170C."}]},{"head":"Temperature Dependence of Development Time in Anchoveta Eggs","index":230,"paragraphs":[{"index":1,"size":156,"text":"In fishes, egg development time depends on temperature (see Hempel 1979) and egg size, among other things. Pauly and Pullin (1987), based on a data set covering 84 species of teleost fishes, temperatures ranging from 2.8 to 29.50C and egg diameters ranging from 0.6 to 3.4 rnrn established, for spherical marine fish eggs, the empirical relationship  Total withdrawals of onchoveta  , 1953 to 1982 (see text for various data sources used'). Above: populations of cormorants (\"guanay\", Phalacrocorax bougainvillii -), boobies (\"piquero\", Sula variegata ----) and pelicans (\"alcatraz\", Pelecanus thagus -. -. -). Note that change in the relations of the three species have gone along with the large decline affecting these birds. Center: estimates of anchoveta consumption by cormorants, boobies and pelicans (aggregated from the results of three species-specific models -),by bonitos (Sarda chiliensis ---) and two species of sea mammals ( -. -. -), the fur seal (Arctocephalus australis) and the sea lion (Otaria flavescens)."},{"index":2,"size":102,"text":"The models upon which the consumption estimates are based used among other temperature (and anchoveta biomass for the buds and seals) to regulate food reqwuements and anchoveta accessibility. Hence, they reflect the occurrence of El Niiro events. Below: Withdrawals of anchoveta (Engraulis ringens) by the fishery ( -. -. -), and by the fishery plus key predators (seabirds, bonitos and seals -). Note that the fishery, which at f i s t took an insignificant proportion of total anchoveta production ended up taking the overwhelming part of that production except, obviously in months with no fishing (\"veda\"). Log lo anchoveta biomass (t)"},{"index":3,"size":186,"text":"Fig. 2. Relationship between sardine egg density and anchoveta biomass of Peru (based on data in Table 1 of Muck et al., this vol. and Pauly, Palomares and Gayanilo, this vol.). The correlation is much higher than that obtained by Muck et al. (this vol.) due both to the use of improved biomass estimates and especially to the logarithmic transformation applied to both variables, which generates normally distributed residuals. where D is the development time, in days, E$ the egg diameter in mm and T the water temperature in degrees centigrade. Anchoveta eggs are not spherical, however, and hence this equation cannot be used directly. Rather, the equation can be solved for 50h = 2.083 days and 15.90C (=midrange of the temperatures given by Santander and Castillo 1973, see above), i.e., which gives 0.574 mrn as the diameter of a sphere equivalentin terms of equation ( 1) and of the data available hereto an oblong anchoveta egg. Solving equation (1) for 0.574 rnrn and simplifying gives which can be used to predict development time in anchoveta eggs at any temperature likely to be occurring off Peru."}]},{"head":"On the Cannibalization of Anchoveta Eggs","index":231,"paragraphs":[{"index":1,"size":50,"text":"That Engraulis ringens cannibalize their eggs has been demonstrated by a number of authors (e.g., in Sharp 1980). There is too some evidence that anchoveta also cannibalize their larvae, the reason for the scarcity of evidence concerning the latter probably being that they are digested faster than eggs (MacCall 1980)."},{"index":2,"size":32,"text":"The data in Table 3 of Santander (this vol.) allows quantification of anchoveta egg cannibalism, via the definition (from Gulland 1969) mean age in a stock = 1 1 2 ... 4)"},{"index":3,"size":36,"text":"where Z is the instantaneous rate of mortality (t-1) and the mean \"age\" in a stock of eggs is the mean time eggs have from spawning until they either hatch or die through predation (including cannibalism)."},{"index":4,"size":170,"text":"Using the data in Table 3 of Santander (this vol.) the mean age of anchoveta eggs can be estimated, for different ranges of parent stock sizes, as the slope linking the estimated number of eggs produced by the anchoveta stock (as estimated by Pauly and Soriano, this vol.) and the number of eggs observed (i.e:, as recorded on H. Santander's maps) and corrected for the temperature-dependent hatching time. Results obtained through this approach are given in Table 2. As might be seen, the Z estimates obtained in this fashion range from less than 1 d-1 to over 4 d-1, and, in fact, allow separation of anchoveta egg mortality into density-dependent and densityindependent components. Fig. 4 suggests an extremely strong impact of parent stock size on anchoveta egg survival, and may thus provide a mechanism for the two-year cycles of anchoveta recruitment detected by Mendelsohn and Mendo (this vol.) and further discussed below.  Pauly and Soriano (this vol.). The ratio of the two production estimates expresses mean egg longevity, in days."},{"index":5,"size":16,"text":"' ~o t a l egg mortality (Z, dK1) is the inverse of mean egg age."},{"index":6,"size":4,"text":"Estimates for E rnordax "}]},{"head":"On Biannual Cycles of Anchoveta Recruitment","index":232,"paragraphs":[{"index":1,"size":144,"text":"Although Ricker's (1954) theory of recruitment would imply the occurrence of biannual cycles of anchoveta recruitment (i.e., two times the mean generation time), the first evidence for a two-year lag in the egg-to-recruit relationship was provided, in an entirely non-Rickerian context, by Mendelsohn and Mendo (this vol.). This correspondence of independent evidence (see also Fig. 5), along with the mechanism provided by cannibalism (see above) would be very gratifying, were it not for the fact that sea surface temperature (SST) fluctuations off Peru also have recently been shown to include a strong biannual component (Fig. 6). Two possibilities, not mutually exclusive, come here to mind: i) biannual cyclicity has been artificially introduced into the recruitment data because these were computed based (in part) on anchoveta consumption estimates by predators that are themselves affected by SST, ii) there are \"real\" biannual oscillations in anchoveta recruitment."},{"index":2,"size":92,"text":"Obviously, option (ii) can itself be subdivided, i.e., a) the biannual oscillations of anchoveta recruitment are directly linked to SST or to another physical phenomenon reflected by SST, or b) cannibalism (i.e., a Rickerian density-dependent process) generates biannual fluctuations of anchoveta recruitment. Which of these options (or combination of options) is more realistic cannot be investigated here and now, and in fact should not before a new time series of anchoveta recruitment, explicitly considering mackerel and horse mackerel predation and including missing years (1982 to 1986) has been (re)computed, as discussed below."}]},{"head":"On Biases in the Recruitment Time Series","index":233,"paragraphs":[{"index":1,"size":87,"text":"As explained in Pauly and Tsukayama (this vol.) the monthly time series of anchoveta biomass derived in this book may be viewed as a \"second iteration\", i.e., part of a process which started with a first iteration (represented by the rough biomass estimates in Table 3 of Muck and   -3 3 -4 1958596061 6 2 6 3 6 4 6 5 8 6 6 7 6 8 6 9 7 0 7 1 7 2 7 3 7 4 7 5 7 6 7 7 Year -2"},{"index":2,"size":108,"text":"\" \" \" \" \" \" \" \" \" ' Pauly, this vol.) and which will eventually lead to reliable series. In this second iteration, the predation by mackerel and horse mackerel was not explicitly considered and its effects were assumed to be part of a residual natural mortality (Mo), set constant for the period 1953 to 1970. As it turned out, mackerel and horse mackerel consumption of anchoveta during this period was not only substantial (much higher than that of the predators that were explicitly considered) but also drastically declined from the 1950s to the 1980s (see Fig. 5 and 6 in Muck and Sanchez, this vol.)."},{"index":3,"size":144,"text":"This implies that Virtual Population Analyses for 1953 .to the 1980s that would account for horse mackerel and horse mackerel predation would lead to estimates of biomass (and of recruitment!) higher, for the period 1953 to 1963 than those of Pauly, Palomares and Gayanilo (this vol.). [There would be little change for 1964 to 1982, on the other hand, because of the availability of independent acoustic of biomass estimates for this period]. Moreover since mackerel and horse mackerel tend to invade the nearshore areas and to feed on anchoveta when SST are high (i.e., during El Niiio events, see Muck et al., this vol.), there would be a mechanism to (partly) compensate (or overcompensate, as the case might be) for the reduced consumption of anchoveta by birds and seals during warm anomalies (see Muck and Pauly, this vol., and Muck and Fuentes, this vol.)."},{"index":4,"size":167,"text":"It can thus be hypothesized that the recruitment and biomass time series that would be obtained would be somewhat smoother, and that the relatively low biomasses and recruitment estimated for the mid-to late 1950s would be closer to the values estimated for the 1 9 6 0 ~~ This would bring the present outlying point for 1957 in Fig. 6 of Palomares et al. (which illustrates the strong negative relationship between anchoveta growth performance and biomass) in line with the rest of the points. Even more interestingly, it would improve the fit of the'models for predicting anchoveta recruitment developed by Mendelsohn and Mendo (this vol.), of which as they write, \"none do a very good job for the years 1955-1959\". This is quite encouraging; indeed the specific structure of the bias in our recruitment time series suggests that a third iteration, incorporating mackerel and horse mackerel predation will markedly improve our descriptions and models, although the questions raised above in conjunction with biannual cycles will probably remain."}]},{"head":"On Lasker Events, Anchoveta Food and the Plankton off Peru","index":234,"paragraphs":[{"index":1,"size":194,"text":"Fig. 7 shows that, during the years 1953 to 1981, on the average, the anchoveta spawning peak of February to March resulted, three months later, in more recruits than the larger spawning peak of September-November. In terms of Lasker's hypothesis, this would imply that the February-March spawning usually occurs during a period of low turbulence, while September-November should be a period with very few of the well defined periods of calm (see Peterman and Bradford 1987 and Mendelsohn and Mendo, this vol.) which I suggest should be called \"Lasker events\". As might be seen from Table 1 in Mendelsohn and Mendo (this vol.) this is indeed the case on the average. However, as the latter authors point out, the relationship between Lasker events and recruitment success off Peru evanesces upon closer examination. None of the major turning points in the available recruitment time series can be predicted from the available time series of Lasker events. Mendelsohn and Mendo (this vol.) suggest that this is due to plankton concentration being, off Peru, generally so high that anchoveta larvae may always have enough food, whether plankton-rich microlayers can establish themselves or not, i.e., independently of turbulence."},{"index":2,"size":118,"text":"Their statement clearly implies the need for a re-examination of the available, published information on Peruvian plankton, the re-sorting and re-analysis of the long time series of plankton samples held at IMARPE (as recommended by the IOcPAO Guiding Group of Experts on Ocean Science and Living Resources in its 2nd session in Rome, 6-12 June 1987), as well as new data, sampled to specifically address this issue. Haydee Santander's bibliography @. xi to xii) provides an excellent entry into the Peruvian literature on Peruvian zooplankton, while additional references on this topic, as well as thoughtful analyses may be found in Calienes et al. (1985), Barber and Chavez (1983), Walsh (1975Walsh ( ,1981) ) and Muck et al. (1984)."},{"index":3,"size":169,"text":"Plankton densities affect recruitment not only through their impact on the survival of larvae, however, but also through their effect on the somatic and gonadal growth of the adults, and hence on quantity and quality of spawning products (Nikolskii 1969). Reports on anchoveta stomach contents are available, (see, e.g., Rojas 1953;Mendiola 1966Mendiola , 197 1, 1980;;Mendiola and Ochoa 1973;Mendiola et al. 1969;Sanchez et al. 1985) documenting that anchoveta stomachs have been sampled off Peru since the early 1950s. In fact, at the time this is written, the raw data used for the papers cited above are being entered into computer files such that they, too, can be turned into standardized time series similar to those presented in this volume. Hopefully, it will then be possible to address some of the questions left open by our analyses, notably in relation to growth changes, to parental egg cannibalism, as well as to establish links between observed plankton densities, adult anchoveta stomach and fat contents and the survival of anchoveta early stages."},{"index":4,"size":141,"text":"These studies should be complemented, however, by an analysis of the link between phytoplankton standing crop (i.e., the main food of adult anchoveta) and \"new primary production\". Information on the former is available in the form of numerous maps, both quasisynoptic and for \"average conditions\", (see, e.g.; Mendiola 1966 andCalienes et al. 1985). The latter can be computed from SST, an upwelling Index and a few well-established empirical constants (R. Barber, Duke University, North Carolina, pers. comm.). Bakun (1985) has recently reviewed the comparative approach as a framework for the interpretation of time series and other data on the recruitment of fish, particularly in Eastern Boundary current systems. Such systems indeed resemble each other both in the physical as in the taxonomic sense, as most dominant species in eastern boundary currents belong to the same genera or even species (Table 3)."}]},{"head":"On Comparative Studies between Areas, Geological Periods andlor Anchovy Species","index":235,"paragraphs":[{"index":1,"size":143,"text":"For the comparative method to continue to be useful here. however, the subtle differences between these species must be acknowledged, e.g.,that Engraulis encrasicholus which range from Norway to West Africa (200s and beyond if E. capensis is a synonym) may not be as welladapted to upwelling conditions as the more advanced E. ringenslE. rnordax species pair (Fig. 8). It is in this context that phylogenetic studies, encompassing taxonomy, biogeography (of both recent and fossil forms) and paleontology (i.e., analysis of fish scale abundance and sizefrequency distribution in sediment cores) might be most useful. These studies might lead to  further generalizations needed for an understanding of the evolutionary biology of Engraulis ringens Jenyns (1842) (see Soutar and Isaac 1969;De Vries and Pearcy 1982;Lasker and MacCall 1983;Nelson 1984Nelson , 1986;;Grande and Nelson 1985;Whitehead, in press and references therein for an entry into the relevant literature)."}]},{"head":"Some Concluding Remarks on Anchoveta Predators and Other Elements of the Peru Current System","index":236,"paragraphs":[{"index":1,"size":109,"text":"Our analysis of the predation of the guano birds on the anchoveta stocks off Peru suggests that these birds may have less of an impact on the fish stock than'proposed by earlier authors (e.g., Furness 1982 andsee MacCall 1982). This is in line with the reassessment performed by Bailey (1986) who showed that seabirds may take 5-8%, rather than 20-30% of North Sea fish production. This theme will, in any case, need some re-analyses possibly based on a different model of bird energetics and feeding dynamics than the one used by Muck and Pauly (this vol.), e.g., using some of the new data compiled in Furness and Monaghan (1987)."},{"index":2,"size":112,"text":"Sea mammals off Peruat least as far as sea lions and fur seals are concernedappear to have a negligible impact on Peruvian pelagic stock (see Muck and Fuentes, this vol., Pauly, Palomares and Gayanilo, this vol.). Thus, it is not the sea mammals which threaten the Peruvian pelagic fisheries, but rather the converse and indeed more will have to be done toward the conservation of the lesser species of sea mammals. Studies presently conducted in Peru toward this aim are listed in Table 4. Optimal results will be obtained here if those conducting such studies continue to interact with \"modellers\", thus guaranteeing that their inputs will be considered in future management plans."},{"index":3,"size":41,"text":"The impact of mackerel and horse mackerel on anchoveta, and the error we initially committed of not considering them to be important was discussed above. This error now leads to the question as to which other species may have been neglected."},{"index":4,"size":239,"text":"Hake have been discussed in Pauly and Tsukayarna (this vol.) and should be included in future iterations. It is difficult to imagine other fishes (i.e., fishes other than mackerel, horse mackerel, bonito and hake) as having a major impact on anchoveta. This leaves squids (especially Dosidicus gigas) as the only fish predators with a potential impact on anchoveta. As mentioned in Pauly and Tsukayama (this vol.) this impact should be limited, however, because the squid in question tend to occur in offshore waters, outside of the range of anchoveta (there are indications that anchoveta in the 1960s ranged further offshore than they presently do, and hence may have been accessible to squid, see Muck and Sanchez, this vol.). A recently available contribution by Benites and Valdivieso (1986), confirms this, both in terms of the distribution of the w s e that events are described which are all unique, however much we want to generalize. In our field, this makes the availability of historical information a far more crucial factor than say in chemistry. Indeed, numerous insights, e.g., on the variability of fish stocks or of the intensity of El Niiio events, could be gained only because of the availability of old written records (see Cushing 1982 andWoodman 1985, respectively). For this reason, an explicit policy to encourage its staff to publish is crucial for any fishery research institution, and some suggestions to this effect are given in Table 5."},{"index":5,"size":69,"text":"Important also are bibliographies, which, while remaining well focused, should be as complete as possible. While having a clear focus, the recently produced bibliography of Mariategui et al. (1985) misses a large fraction of the literature cited in this volume, including classics such as, e.g., Bini (1952), which, however, one finds cited in works such as De Buen (1960) and listed in the bibliography of Stephenson and Hicks (1975)."},{"index":6,"size":130,"text":"Equally problematic is the fact that a significant fraction of the literature on the Peru Current was originally published in Russian, and that only a small part thereof was translated into Spanish or English (see, e.g., Aivazyan and Krapivin, 1984 and references therein). Clearly, an effort should be made to have the bulk of this literature translated, e.g., through some cooperative agreement between IMARPE and its sister institution in the USSR, or as part of the \"package\" enabling Soviet vessels to operate in Peruvian waters. This would both help Peruvian fishery science and management and provide a larger audience for the important work that Soviet fishery scientists and modellers do.  * Produce an Annual Report, with brief accounts of the work in each research group, their findings and published output;"}]},{"head":"On Managing the Peruvian Pelagic Fishery","index":237,"paragraphs":[{"index":1,"size":57,"text":"Fig. 10 gives an integrated summary of what was covered by the studies presented in this book, while areas not dealt with are indicated as such (shaded box). As might be seen, we have covered rather comprehensively the oceanographic/biological side of things, while our coverage of the social and economic aspects of fishery management was very limited."},{"index":2,"size":52,"text":"Various contributions in Glantz and Thompson (1981) have highlighted the problems associated with formulating and carrying through a fishery policy with a fixed target (e.g., \"MSY\") in the face of environmental variability. Clearly, any plan for the management of the Peruvian upwelling ecosystem will have to take environmental variability explicitly into account."},{"index":3,"size":62,"text":"I have termed \"Fourth Level of Integration\" the level of research and management sophistication required to do this (see Fig. 10). However, Dr. Max Aguero (ICLARM, pers. comrn) feels that few studies have been conducted to date which would correspond to this \"Fourth Level\", one of the few exceptions being the \"World Model\" of the Club of Rome (Meadows et al. 1972)."},{"index":4,"size":101,"text":"Species Indexa  , vi, vii, 4,9,87,143,145,149,154,155,248-253,255-267,289,305,327,334,335,338 see also Sarda chiliensis booby, 143 155,[212][213][214][215][216]219,222,223,226,228,230,231,232,327 Brevoortia,4 caballa,xi,276,292,293,324 Callorhinus ursinus, see seal, northern fur Carangidae, 276 Cephalopoda, 9 cerrajon, 249,263, see also Sarda chiliensis cetacean, 9 Cetengraulis, 335,342 chauchilla, 249, see also Sarda chiliensis Cichlidae,139 Clupeoidei,5,341,342 Clupeomorpha,5 cod,266,280,292,293 copepod,[277][278][279]91,143,155,[212][213][214][215][216][217]219,220,223,226,228,[231][232][233]327, see also guano bird Crustacea, 260,278 Delphinus delphi, see dolphin dolphin, 336 Dosidicus gigas 335 see also squid Dusycyon sechurae, see fox Engraulidae,5,139,141,171,342 Engraulis,136,137 172,178,306,335,  mordax, xii, 12,87, 119, 136,139,140, 171, 173,177,178,219,245,295,306,330,334, 335,342 mysticetus, 335 euphausid, xi, 261,277,278,279 apage numbers for 'Engraulir ringens,\" \"Anchoveta\" and \"anchovy\" are omitted due to their occurrence throughout the book."}]},{"head":"fox, 213","index":238,"paragraphs":[{"index":1,"size":170,"text":"Gadus morhua, see cod gannet, 222, see also booby guanay, 209,212,213,218,233,327, see also  cormorant  guano bird, vi, vii, 2,4,5,9, 10,89,91,208-211,213-217,219,221-233,249,291,335,339, see also seabird haddock, 293  hake, 10, 11,87,245,246,305,323,335,   mackerel, vi, vii, 4,9,87, 11 1, 113,139,225,245,266, 268-293,309,310,315,320,332,334,335, 338 horse, 4 , 9 , l l l , 113,225,245,268-272,274jack, 338,   Otariaflavescens,9,11,12,234,245,247,327,336  thagus, 208,219,222,231,327 pelican, 143,155,212-217,219,221,223,226,228, 23 l-233,327,   pilchard, 334,342  pinniped, 9,234-237,238,240,241,244-246  piquero, 209,212,213,327, see also booby  plankton, xi, xii, 5,7, 14,132, 133, 138,269,273,276,  277,283,333,334,336,337  Pneumatophorus japonicus, 292,293  porpoise, 9, 10,336  Prionotus quiescens, 260,261   Rastrelliger kanagurta, 139   salmon, 292  Sarda, 254,259,265-267  chiliensis, vi, 155,248,249,252,254,262,  264,266,267,334  chiliensis chiliensis, 249-255,258,260-262  chiliensis lineolata, 249-252,259,262  orientalis, 249,250,262,267  sarda, 252,254,334  sarda chiliensis, 266,293  sarda peruanus, 293  sardine, vi, vii, xii, 9,49, 111,113, 141,184,225,245,  268-275,289,308,310,315,320,338,342  Sardinia pilchardus, 334  Sardinops, 4  occelatus, 334   sagax, xii, 11 1, 141,260,268,275,289,309,  334,341,342 Scomber japonicus, 1 1 1,113,268,276,278,281,289,290,309,334 japonicus peruanus,xi,292,293 scombrus,264 Scombridae,248,259,264,266,267,276 sea lion,vi,9,11,12,155,[234][235][236][237][238][239][241][242][243][244][245][246][247]335,336, see also p i ~i p e d sea lion, California, see Zolophus californianus seabird, 9,10,95,143,145,149,151,154,155,217, 224,226-228,232,233,240,246,264,265, 289,272,327,332,335,338,341 seal, 9,10,143,145,149,154,265,289,332,338, see also pinniped fur, vi, vii, 9 11, 12,155,234-239,241-246, 327,335 "}]}],"figures":[{"text":" Fig.14. Growth of Peruvian anchoveta as simulated using model described in text. Note that most of the difference between the \"1950\" and \"1980\" growth curves is due to growth rate differences of fish between 1 and 1 .5 years. "},{"text":"Fig. 5 , Fig. 5, based on Tables 5 to 33, shows the monthly sums of biomass over all length groups, for 1953 to 1982. One part of this time series, covering the years 1963 to 1979 illustrates the match of our biomass estimates to the independent biomass estimates used to calibrate the VPA "},{"text":"Fig. 5 , based on ~a b l e s 5 to 33, shows the monthly sums of biomass over all length groups, for 1953 to 1982. One part of this time series, covering the years 1963 to 1979 illustrates the match of our biomass estimates to the independent biomass estimates used to calibrate the VPA (i.e., to esimate &). That part of the time series presents no surprise, except perhaps for the fact "},{"text":"Fig. 4 .Fig. 5 . Fig. 4. Estimates of baseline natural mortality (Mo) as required to reproduce independent estimates of anchoveta biomass using the VPA 111 routine of the ELEFAN 111 program. Line is eye fitted. Note forward and backward extrapolations of Mo = 4 y-l and Mo = 2 y -l , respectively (see also Table2 and text). "},{"text":"Year  "},{"text":"Fig. 7 . Fig. 7. Fishing mortality (ie., catch in weight/biomass) exerted on Peruvian anchoveta (northern/central stock, 4-14OS) from January 1953 to December 1981. Above: actual values, by month. Note sharp spikes, partly due (from 1972 on) to closures of the fishery. Below: running average (over 12 months) of monthly values, to show major, increasing trend from the 1950s to the early 1970s. "},{"text":"Fig. 8 . Fig. 8. Monthly time series of natural mortality in juvenile and adult anchoveta, 1953 to 1981. A. Mortality caused by cormorants (mainly), boobies and pelicans. B. Mortality caused by bonito (Sarda chiliensis) (note similarity of estimates to those pertaining to the birds). C. Mortality caused by sea lion (mainly) and fur seals. Note that the scale, ranging from zero to 0.005 implies a negligible pinniped impact on anchoveta natural mortality due to all causes. Note that shape of curve resembles that of Fig.4, except for small \"bumps\" in the 1950s, corresponding to periods of bud and bonito abundances. Important here is that decline of M from 1971 to 1974 is confirmed by analysis of tagginglrecapture data performed byMalaga and Armstrong (MS). Note also that little support is available for low M estimates (B & S) inSchaefer (1967). "},{"text":"Fig. 1 . Fig. 1. Distribution of anchoveta stocks along the Eastern Coast of South America. Based on FA0 (1981), Jordan (1971), Chirichigno (1974), Brandhorst (1963) and IMARPE (1973). "},{"text":"x lo6 1 Fig. 2 . Fig. 2. Stock/recruitment relationships of Peruvian anchovy (Engmulis ringens) showing effect of taking an additional variable into consideration. (A) simple Ricker plot, showing rather bad fit and correspondingly low correlation of observed to expected recruitment (GM tine, r = 0.494). (B) Plot of the residuals of a multivariate relationship involving r~ruitment, parent stock and concentration index, Q, related to occurrence of El Nino events onto stock recruitment relationship, drawn for an average value of Q. This shows an improved fit with a correlation of observed to expected recruitment of r = 0.893 (based on data in Csirke 1980). "},{"text":"Fig. 2 . Fig. 2. Annual cycles, autocorrelation functions, and anomalies of monthly means of daily observations at coastal and island stations. Parameters are (A) Southern Oscillation Index, (B and C) SST, and (D and E) sea level at Talara and La Punta, Peru. The annual cycle plot (center) shows the long term (1943-1986) monthly means, between-year standard deviations (bars), and range of interannual variability (dots). The autowrrelation function plot (right) shows the autocorrelation of each original data series (dotted) and the autowrrelation of each anomaly series(solid). The time series of anomalies from 1941-1986 mean (left) are shown in standard deviation units for intercomparhn between data series. Based on data in Tables "},{"text":"Fig. 4 .YearFig. 6 . Fig. 4. Time series of spatially-averaged monthly means for the region 4 to 14's as computed from profiles of subsurface temperature. Parameters are (A) SST, (B) depth to 14OC isotherm, (C) depth to SST-2OC isotherm, and (D) heat content from the surface down to SST-2OC. "},{"text":"Year  "},{"text":"Fig. 8 . Fig. 8. Time-depth contour plot of monthly mean subsurface temperature off Peru from 1952 to 1984. Data are computed at 25 m depth intervals from 0 to 350 m from subsurface temperature profiles for central area (Area Ill). Values shown for \"1951\" are a composite of data for all years prior to 1952. Note the depression of the lZ°C and 14OC isotherms from the early to mid-1970s to a maximum depth in 1982. "},{"text":"Area  "},{"text":"Fig. 9 . Fig. 9. Annual cycles of subsurface temperature vertically from 0 to 350 m for 5 areas along the Peru coast. Note the progressive rise of the isotherms from north (Area I on the left) to south (Area V on the right). For example, the 14°C isotherm rises from depths of 250-325 m in the north to depths of 75-125 m in the south. "},{"text":"Fig. 1 . Fig. 1. Summary area. Maritime reports from within the area indicated by diagonal hatching were used for assembling monthly samples. "},{"text":"Fig. 2 . Fig. 2. Seasonal cycles. 32-yr mean monthly values. "},{"text":"Fig. 3 . Fig. 3. Seasonal cycles. 32-yr mean monthly values. (Note that the SI-standard heat flux units, watts per square meter, may be converted to calories per square centimeter per day by multiplying by the factor 2.604) "},{"text":"Fig. 4 . Fig. 4. Low-frequency nonseasonal variations, 12-month running means of monthly time series values. (Note that the SI-standard heat flux units, watts per square meter, may be converted to calories per square centimeter per day by multiplying by the factor 2.064.) "},{"text":" i n d e x value. Nominal u n i t s are r n = ~e c -~. "},{"text":"Fig. 5 . Fig. 5. Graphs of r2 (= coefficient of determination, i.e., proportion of variation in one series 'explained' by other series) and slope of the regression of the variable transfer coefficient formulation versus the constant transfer coefficient formulation of alongshore stress (T,), onshore stress (T ), evaporative heat loss (QE), conductive heat loss (QC) and net atmosphere-ocean heat exchange (QN). "},{"text":"Fig. 1 . Fig. 1. Insert: Locations of Peruvian towns mentioned in this contribution. Larger Map: Wind strength and direction during a cruise on 20 April-14 May 1966 showing correspondence between wind records at TrujiUo, Chimbote and Callao airports (thick arrows) and nearshore wind records (thin arrows). "},{"text":"Fig. 2 . Fig. 2. Effect of using less and less subdaily observations when computing daily mean upwelling indices. Note that basic features of time series are maintained when 6-8 observations per day are used. (Data refer to Callao, 1970; see alsoTable 1 and text). "},{"text":"Fig. 3 . Fig. 3. Monthly variation of the upwelling index, Trujillo and Callao, 1953 to 1985. "},{"text":"Fig. 4 . Fig. 4. Monthly variation of the turbulence index, Trujillo and Callao, 1953 to 1985. "},{"text":"Fig. 5 .Fig. 7 . Fig. 5. Monthly cycles of turbulence and upwelling at Trujillo and Callao, based on 33-year means of monthly values (1953 to 1985). "},{"text":"Fig. 9 . Fig.9. Seasonal cycles of turbulence and upwelling at various locations in the Peru Current. Each numbered symbol represents n 2-month climatological sample, with the number corresponding to the fvst of the pair of months. Seasons andlor area with low turbulence and offshore transport may be assumed to be appropriate for survival of fry (redrawn fromParrish et al. 1983, with data for Callao and Trujillo added). "},{"text":"\"Fig. 1 . Fig. 1. Monthly nominal catch of Peruvian anchoveta (Engraulis ringens) (northern/central stock, 4-14's) from January 1953 to December 1982. "},{"text":" Condition factors in brackets estimated h m the linear relationship between temperature and condition factor inPalomares et al. (thisvoL). "},{"text":" 1 and 2 in Doucet and   E' sson (1967).TCondition factors in brackets estimated from the Unen relationship between temperature and condition factor inPalomares et al. (this voL). "},{"text":" catchcomposition data obtained during EUREKA (August and September) and CATEO (February) surveys. b ~o n t h s with closure qf fishery (veda); catches mainly from surveys (February and September). C~ondition factors in brackets estimated from the linear relationship between temperature and condition factor inPalomares et al. (this voL). "},{"text":" : % frequencies refer to northern region only. Condition factors in brackets estimated from the Smear relationship between temperature and condition factor in Palomares et aL (this voL). "},{"text":"Fig. 1 . Fig. 1. Schematic representation of major causes of anchoveta catch underreporting in the Peruvian reduction fishery, with emphasis o n thepesca negra (arrows not to scale; see text for quantitative information). "},{"text":" recent years. Type (a) losses may likely have decreased, but types (b) and (c) most probably have increased as more peladilla are now being taken. "},{"text":"Fig. 3 . Fig.3. Steps used to correct ELEFAN I estimates of growth parameters in Peruvian anchoveta, using the method ofPauly (198613). "},{"text":"Fig. 4 . Fig. 4. Example of cohort-specific growth curves identified using the ELEFAN I program. Black histograms refer to what were peaks in the original length-frequency data, open histograms refer to what were the troughs separating peaks. Note faint seasonal growth oscillations, and the fact that growth curve originate at spawning seasons (Sept.-Nov. and Feb.-April). See text for details on the ELEFAN I program. "},{"text":"FigFig. 6 .YearFig. 7 .Fig. 8 . Fig. 5. A: time series of maximum recorded length in Peruvian anchoveta catch samples "},{"text":"Fig. 9 . Fig. 9. Relationship between condition factor and sea surface temperature in anchoveta Engrnulis ringens off Peru (4-14's). Based on original c.f. data in Tsukayama and Palomares (this vol.) and temperature data in Pauly and Tsukayama (this vol.). See text for discussion. "},{"text":"Fig. 11 . Fig. 11. Seasonal changes of some growth-related Sea-; ; A ; ; A s , , , ; tures of the Peruvian anchoveta and its ecosystem. A) changes in length growth rate; B) changes in condition; Month C) changes in sea surface temperature (see text). "},{"text":"Fig. 12 . Fig. 12. Time series of fat content in Black Sea anchovy (A) and Peruvian anchoveta (B-D): note tendency for fat content to peak before spawning season(s). "},{"text":"Fig. 13 . Fig. 13. Basic elements of a model for indentifying possible causes for a change in the growth performance o f anchoveta. The \"Problem\" level shows the question to be answered, while the \"Model\" level shows some of the mechanisms hypothesized to affect growth rate (see text). "},{"text":" food = 1,000-1,450 cal V = Vi 0.05 T (V in body length/s; T = OC) Vl = 0 . 5 8 2 2 ~O . ~~~~ (for L < 12 cm) V2 = ( 0 5 8 2 2 ~~~~~~~) + (6.522-2.645 logeL) (for L >= 12 cm M 4 = 0.0887 e0.0959T + 0.036V L W O d (M1, = active metabolish; callh) ML, = 0.3MLa (ML, = resting metabolism) FR = A.0.78.V.L.3.6 (FR = filtering rate, in literlhr; S = frontal area (surface) of open mouth, in cm2, 0.78 = mouth openings per second, 3.6 = factor for convertmg cm2 & body lengthla to literlhr) S = 15/(l + e a.44L) R = FR-D.0.8 (R = g/h; D = food density, in g/liter and 0.8 is an empirical retention factor) R = 7.57 T . L -~-~ (R = % of individual body weight per day) Ptot = (R.ha-A) -((ML;$) + (ML;$)) (Ptot = callday; h, = hours of activity, i.e., feeding; hr = hours resting) A = fraction of ingested food available after fecal and excretory losses, as well as Specific Dynamic Action (Jobling 1983) are accounted for. h, = R/I (for ha > 24, ha = 24; 1 = food ingestion rate expressed as % of body weight, per day; when Go = 0.1 of gonad-free body weight Eggs = 1,104 + 614 W(W = ovary free female wet weight, in g) Batch = Eggs 0.0301 5.8/1,000 (g wet weight) 0.0301 = egg dry weight (mg); 5.8 = dry to wet weight conversionVillavicencio and Muck (1983a)    "},{"text":"Fig. 15 . Fig. 15. A. Frequency distribution of growth performance indices in Engraulis spp. (excl. E. ringens), showing a wide range covered by the genus. Based on data inPauly (1978),Radovich and MacCall (1979),Volovik and Kozlitina (1983),Melo (1984)  andGallardo-Cabello (1985). B. Frequency distribution of growth performance indices in E. ringens showing that the species grows, as a whole, better than other Engraulis "},{"text":"Fig. 16 . Fig. 16. Relationship between the amplitude of seasonal oscillations in the growth in length of temperate and tropical fishes (= C in equation l) and the difference between highest and lowest mean monthly water temperature. The shaded area, referring to E. ringens off Peru (4-14OS) shows that anchoveta growth is reduced, during the cold season just as one would predict from other fishes. Modified from Pauly and Tsukayama (1983). "},{"text":"Fig. 1 . Fig. 1. Facsimile of an ELEFAN 111 output (via plotter) showing monthly catch-at-length data (not to scale) and one of the many monthly cohorts which can be superimposed on "},{"text":"Fig. 2 . Fig. 2. Length-frequency data of Peruvian anchoveta (northern/central stock), including samples that were linearly interpolated. The reader is invited to guess which samples were interpolated. (.~aquraaoN pue ~sn%ny ' L ~I ' a u n ~ a n sqluour palqod~alu! a y ~) "},{"text":" Fig\\ 3. Facsimile of an ELEFAN 111 output (via plotter) showing summary of result of a VPA 111 pertaining to a given month (here May 1969), and showing the catch data (lower pan91, right, histograms) the cohorts \"passing through\" that month (box with narrowing lines), the size structure of the population (upper boxes, center and left, as well as summary stat. stics (lower panel, left). Details are given in the software documentation. "},{"text":" t "},{"text":"Fig. 4 . Fig. 4. Estimates of baseline natural mortality (Mo) as req%red to reproduce independent estimates of anchoveta biomass using the VPA I11 routine of the ELEFAN 111 program. Line is eye fitted. Note forward and backward extrapolations of Mo = 4 y-l and Mo = 2 y-l, respectively (see alsoTable 2 and text). "},{"text":"Fig. 5 . Fig. 5. Biomass estimates of Peruvian anchoveta, 4-14OS, obtained through the VPA I11 routine of the ELEFAN I11 program (see text), and showing independent biomass estimates used to calibrate VPA. "},{"text":"Fig. 7 . Fig. 7. Fishing mortality (ie., catch in weight/biomass) exerted on Peruvian anchoveta (northern/central stock, 4-14OS) from January 1953 to December 1981. Above: actual values, by month. Note sharp spikes, partly due (from 1972 on) to closures of the fishery. Below: running average (over 12 months) of monthly values, to show major, increasing trend from the 1950s to the early 1970s. "},{"text":"Fig. 8 . Fig. 8. Monthly time series of natural mortality in juvenile and adult anchoveta, 1953 to 1981. A. Mortality caused by cormorants (mainly), boobies and pelicans. B. Mortality caused by bonito (Sarda chiliensis) (note similarity of estimates to those pertaining to the birds). C. Mortality caused by sea lion (mainly) and fur seals. Note that the scale, ranging from zero to 0.005 implies a negligible pinniped impact on anchoveta natural mortality due to all causes. Note that shape of curve resembles that ofFig. 4, except for small \"bumps\" in the 1950s, corresponding to periods of bud and bonito abundances. Important here is that decline of M from 1971 to 1974 "},{"text":"Fig. 1 . Fig. 1. Relationship between the fraction of mature anchoveta (E. ringens) and their total length off Peru (1953-1964). A: based on Clark (1954, Fig. 5, both sexes combined) B: based on Clark (1954, Fig. 5, males only) C: based on Jogan (1959, Fig. 9, erroneous values < 130 mm SL omitted) . D: based on Minano (1958, Table VIII, females, n = 6,142) E-J: based on Einarsson et al. (1966, Fig. 8, both sexes combined) n = 726,686,565,508,929 and 614 for D-J, respectively. "},{"text":" A*) accounting for the sizerelated variability of reproductive output in Engraulis mordax female^.^ Fraction of female anchoveta in catch samplese a Based on R.F. -W relationship in Table 2, footnote [b] . Median of length classes in the anchoveta catch (see Tsukayama and Palomares, this vol.). Estimated through the mean relationship W = 0 "},{"text":"Fig. 4 . Fig. 4. Relationship between length of anchoveta catch samples and the fraction of females. With the exception of four less credible data points (dots) between 7 and 10 cm, all available data suggest an increase of females toward larger sizes. "},{"text":"-Fig. 5 . Fig. 5. Total biomass, parent (mature) stock and biomass of mature anchoveta females off Peru, 4-14'S, 1953 to 1982. Note strong, regular seasonal oscillations and also very good match with independent spawning stock estimate in August-September 1981 by Santander et at (1984). "},{"text":"4 Fig. 7 . Fig. 6. Relationship between spawning stock and total "},{"text":"Fig. 1 .- Fig. 1. Relationship between adjusted anchoveta egg standing stock/anchoveta parent biomass and parent biomass, showing strong correlation. "},{"text":"Fig. 4 . Fig. 4. Distribution of anchoveta eggs off Pau during four surveys conducted in 1964 and 1965 (seeTable 3 for egg standing stock estimates based on maps 5 to 8). "},{"text":"Fig. 8 .I Fig. 7. Distribution of anchoveta eggs off Peru during four surveys conducted in 1968 and 1969 (seeTable 3 for egg standing stock estimates based on maps 17 to 20). "},{"text":"Fig. 12 .Fig. 13 .Fig. 14 . Fig. 12. Distribution of anchoveta eggs off Peru during f ~u r surveys conducted in 1972 and 1973 (see Tables3 and 4for e g standing stock estimates based on maps 37 to 40). "},{"text":"Fig. 15 .Fig. 16 . Fig. 15. Distribution of anchoveta eggs off Peru during four surveys conducted in 1974 and 1975 (seeTable 3 for egg standing stock estimates based on maps 49 to 52). "},{"text":"Fig. 18 .Fig. 19 .Fig. 20 .Fig. 22 . Fig. 18. Distribution of anchoveta eggs off Peru during four surveys conducted in 1977 (see Tables3 and 4for egg standing stock estimates based on maps 61 to 64). "},{"text":"Fig. 1 . Fig.1. The three main species of fish-eating birds of the Peruvian upwelling ecosystem (Spanish names in brackets). "},{"text":"Fig. 2 . Fig. 2. Location o f guano islands and points along the Peruvian coast (islands and points outside o f 4-14' s not considered). "},{"text":" .... de ................... a ........ dc .................... de 19 C A N T I D A D D E A V E S .......................... GUANAYES (en negm) OBSERVACIONES .......................................... ....................................... .......................................... "},{"text":"Fig. 3 . Fig. 3. Example of map partly completed by the guard of Chincha Sur Island on the 10th of January 1962 at 3 p.m. (see Table1for location). The color codes, i.e., black for "},{"text":"Fig. 4 . Fig. 4. Distribution of cormorants (guanay), boobies (piquero) and pelicans (alcatraz) on Chincha Norte Island (see Table1) in September 1974, with relative areas estimated by planimetry. The sums of the relative area are 0.019,0.088 and 0.050 for cormorants, boobies and pelicans, respectively (see text). "},{"text":"Fig. 5 . Fig. 5. Monthly fluctuations of the population of the three major seabird species off Peru (4-14\"S), 1953 to 1982. "},{"text":"* 1 .Fig. 3 . Fig. 2. Relationship between the reconstructed % size composition in booby food bolus and the sizefrequency composition of the northernlcentral Peruvian anchoveta fishery (based on H. Tovar, IMARPE, pers. comm.; data for Chinca Sur Island are from Fuentes 1985). "},{"text":"Fig. 4 . Fig. 4. Relationship between estimated food consumption and body weight in three genera of fish-eating guano birds. Note difference between line derived from stomach content analyses and line based on metabolic studies (based on data in Table2). "},{"text":"YearFig. 5 . Fig. 5. Time series (annualmeans) of variables affecting the consumption of anchoveta by guano birds off Peru (4-14\"S), 1953 to 1982 (see text for definitions and hses to which these variables were put). "},{"text":" i s value corresponds to a bird population growth rate of about zero in undisturbed system. "},{"text":"Fig. 6 . Fig. 6. Flowchart o f the FORTRAN program u s e d to estimate anchoveta consumption by c o r m o r a n t s , boobies and pelicans off Peru; month = i; bird species = j (see also text). "},{"text":"-Fig. 7 . Fig. 7. Annual anchoveta and total possible food consumption by guano birds off Peru (4-14'S), 1953 to 1982. Note that anchoveta represented, in the mid-1950s about half of total possible consumption, a fraction much reduced in the following period. "},{"text":"Fig. 9 . Fig. 9. Changes in the relative importance of anchoveta consumption by the three major bird species in the Peruvian upwelling ecosystem, 1953 to 1982. "},{"text":"Fig. 1 . Fig.1. Size-frequency distribution of anchoveta eaten by fur seal (Actocephalus australis) (shaded histograms) as reconstructed from otoliths recovered from scats(Majluf 1986 and pers. comm.), compared with the fishery catch size distribution in a comparable period (solid line, unpublished IMARPE data). Note rough correspondence in 1983 and late 1985 and shift toward smaller sizes in 1984, early 1985. "},{"text":"Fig. 2 . Fig.2. Flowchart of model used to estimate population sizes of fur seals and sea lions off Peru (see also Table1 and text). "},{"text":"YearFig. 3 . Fig. 3. Population changes of fur seal and sea lion off Peru (4-14's); dots and squares are census estimates, used to caliirate parameters of population growth model (see TableI). "},{"text":"Fig. 4 .Fig. 5 . Fig. 4. Estimated food consumption and anchoveta consumption by pmnipeds (sea lion and fur seal) off Peru (4-14OS), 1953 to 1982. "},{"text":"Fig. 1 . Fig. 1. Distribution of Sarda orientalis, S. chiliensis chiliensis and S. chiliensis lineolata in the Eastern Pacific. Note that S. orientalis also occurs in other parts of the Pacific and the Indian Ocean, while S. chiliensis occurs only in the Eastern Pacific. Note also that occasional specimens of S. chiliensis lineolata have been reported from Alaska (adapted from Yoshida 1982). "},{"text":"Fig. 2 . Fig. 2. Available growth curves for Sarda chiliensis: A. Campbell and Collins (1975), for S. chiliensis linedata; B . This study, for S. chiliensis chiliensis off Peru (see also Fig. 3); C. Kuo (1970) for S. chiliensis lineolata; D. Kuo (1970) for S. chiliensis chiliensis and E. This study, growth in weight of S. chiliensis off Peru (see also Table4). "},{"text":"ZFig Fig. 3. Estimation of L , and Z/K using the method of Wetherall (1986); based on length-frequency data in Table 5 (average of two samples) and equations (1) to (3). "},{"text":"Fig. 5 . Fig. 5. \"Csirke and Caddy\" model of the Peruvian bonito fishery, based on data for the period 1951to 1976 and used to obtain preliminary estimate of natural mortality. Note that estimated MSY corresponds well with value estimated using Schaefer model (see Fig.4). (P refers to Paita, C to Callao; see Table3). "},{"text":"\"Fig. 1 . Fig. 1. Diet composition of mackerel (Scornber japonicus) during a \"normal year\" and an El NZo "},{"text":"Fig. 2 . Fig. 2. Sizefrequency distribution of anchoveta (Engraulis ringens) ingested by mackerel (Scomber japonicus) (28-64 cm) off Chimbote (19761, compared with size-frequency distribution of catch samples from the purse seine fishery (JuneJuly, adapted from Table24in Tsukayarna and Palomares, this vol.). "},{"text":"Fig. 4 . Fig. 4. Example of a model output, showing the density distribution of mackerel and horse mackerel biomass as a function of distance offshore and SST (El Nice > Summer > Winter); see also text. "},{"text":"= 14.2 miles. The four estimates of derived are recalled in Table 8, along with the corresponding temperatures. These data allowed derivation of the equation where Dm= is the maximum of the biomass density function (i.e., W as used previously) and T is the SST, in OC. Thus we have which allows estimation of the percentage of the total biomass of mackerel and horse mackerel within 40 miles as a function of SST. "},{"text":" in length-weight rel. (-1 Rationweight r el. (R = c . wd) "},{"text":" * Nt(rel))/( Z Ntbel) Wt)). ( a * L,(l -e x p -~( t -t ~) ) ~) ) "},{"text":" references above) and indeed the abundance of their larvae from 1964 to 1982 has been shown aIntermediate results such as monthly mackerel and horse mackael biomass within the MAR, their estimated total food consumption and dated statistics an: available on 5.25' microcomputer diskettes. Picase amtact the smior editor d this volume for details. "},{"text":"Fig. 5 . Fig. 5. Annual estimated consumption of anchoveta by mackerel and horse mackerel, compared with the bird consumption, the fishery catches and minimum estimates of mackerel and horse "},{"text":"Fig. 1 . Fig. 1. Decomposition of the anchoveta recruitment time series into its component parts: (A) observed data; (B) estimated trend; (C) autoregressive component; (D) seasonal component (see text for details). "},{"text":"Fig. 2 . Fig. 2. Boxplots of the monthly anchoveta and wind-based environmental series: (A) recruitment; (B) log of egg production; (C) parent biomass; (D) total biomass; (E) Trujillo offshore transport; (F) Trujillo wind speed cubed (see text for definition of ordinate and data sources). "},{"text":"Fig. 6 . Fig. 4. Estimated transformations for predicting anchoveta recruitment for successive 2-month periods (see text). "},{"text":"*Fig. 1 . Fig. 1. Basic trends of biological and economic characteristics of the Peruvian purseseihe fishery for the period 1950-1983. A: Annual catch of anchoveta, of sardine and total annual catch, 1950 to 1983. B: Total annual catch and number of vessels, 1950-1980, in 5-year segments such as used in linear programming model. C: Annual catch per vessel and mean vessel hold capacity, 1970 to 1980. D: Current price of fishmeal in adjusted US dollars, 1963 to 1983. "},{"text":"Fig. 2 .Processing 1 rans sport 1 Fig. 2. Schematic representation of major material and money flows in the processing and marketing of Peruvian pelagic fisheries (arrows not to scale, see Tables 1-6 for actual numbers). "},{"text":"FNBF= (Pucdi * Qai) -C (TUCi * Q:) -(cuSi * Q,') -C (CUM * Q :) of species group d going into product type i (0.5). k = reduction coefficient of species group d at product type i (4.166).where :i = type of product il = fishmeal iZ = fish oil i3 = canned fish j = demand segment b = yield segments bl = first segment < 250 t b2 = second segment < 500 t b3 = third segment < 1,500 t b36 = 36th segment < 9 x lo6 t d = species group composition dl = mix of anchoveta and sardine d2 = mix of mackerel and horse mackerel S = stock size of species mix S1 = first segment S2 = second segment S3 = third segment S9 = ninth segment Harvesting lock = Unit price of product type i at demand segment j = Total quantity of product type i at demand segment j Cost per unit of yield of species group d at yield segment b = Total quantity harvested of species group d at yield segment b = Total quantity of species group d transferred into processing = (adi Xdb) where adi = is reduction coefficient = Processing unit cost of fish of species group d into product type i = Quantity of species group d processed into product type i = Transport unit cost of product type i to market place = Storage unit cost of product type i = Quantity of product type i transported = Quantity of product type i stored = Quantity of product type i marketed = Marketing unit cost of product type i "},{"text":"Fig. 3 . Fig. 3. Fundamental relationships between biological characteristics of fish resources and technology of capture and various cost factors. A: Relationships between yield and the underlying fish biomass (note that B1 > B2 > B3). B: Relationships between yield, effort and the quotients cost per unit of yield and yield per unit of effort. C: Relationships between yield per effort, effort and cost per unit of yield. "},{"text":"- Fish oil (US$/t) -Canned fish (US$/t) Transport sector: "},{"text":"Fig. 1 . Fig. 1. Selected monthly time series on the Peruvian upwelling ecosystem, 4 to 14OS, 1953 to 1982 (see text for various data sources used'). Above: populations of cormorants (\"guanay\", "},{"text":"w This graph, taken from Pauly et al.(1986)  does not consider predation by mackerel and horse mackerel; the absolute values (note log scales!) would be higher, but the trends similar, . I were these two fishes considered, becauseMuck et al. (this vol.)  found them to have consumed more anchoveta in the 1950e1960s than in the 1970s-1980s. "},{"text":"Fig. 3 . Fig. 3. Time series of an anchoveta prerecruit survival index (i.e., loge monthly recruitment in month (i)/egg production in month (i-3)), 1953-1981. Above: original monthly series, based on data in Pauly and Soriano (this vol.) and Pauly, Palomares and Gayanilo (this vol.). Below: Smoothed series (12 months running means), showing major difference between 1953 and 1969 (= high values showing little fluctuations) and 1970 to 1981 (highly variable values). "},{"text":"Fig. 4 . Fig. 4. Relationship between anchoveta egg mortality and parent stock off Peru (based on data in Table2), with values for the northern anchovy added for comparison. Note extremely high impact of parent stock, and simlarity, for low parent stock values, of estimates for Peru and California. "},{"text":" Fig. 5 . Spectral analysis of the monthly recruitment time series of Pauly, Palomares and Gayanilo (this vol.) showing (expected) peak at 12 months and (unexpected) peak near 24 months. The analysis was performed using the SPECTRA Procedure in Helwig and Council (1979). "},{"text":"Fig. 6 . Fig. 6. Left: Five-month running mean sea surface temperature anomalies (OC) from ship observations between 4 and 8's along the ship track parallel for the South American coast. Right: Same timeseries after band-pass filtering to reveal the biennial component of variability. Filter has full response at 24 months, with onehalf power points at 18 and 30 months (E. Rasmussen, University of Maryland, Dept. of Meteorology and C. Ropelewski, Climate Analysis Center, NOAA, pers. comm.). "},{"text":"Fig. 7 . Fig. 7. Demonstrating the existence of an anchoveta \"recruitment window\": the smaller spaw@ng peak in February-March produces the bulk of anchoveta recruitment (ie., f i h of 3.75-4.75 an) while the huge spawning peak of September-J F M A M J J A s 0 N D J F M November corresponds, three months later to a Month "},{"text":"Fig. 8 . Fig. 8. Interrelationships of species assigned to thegenera Engraulis and Cetengmulis; 0-3, groups for which there is evidence of relationships in the form of shared advanced character (from Nelson 1984). "},{"text":"Fig. 9 . Fig. 9. Flow diagram of the major elements of a simple model of the Peruvian upwelling ecosystem (from Kremer and Sutinen 1975). Note that even in this single model, plankton dynamics are simulated with far more details than the fish, here represented by the \"anchovy\". "},{"text":"FimFig. 10 . Fig. 10. Flow chart showing integration of studies included in this volume, along with areas not covered (shaded box), which include fishery-related government policies and \"Fourth Level of Integration\", not achieved here. Note that arrows (not shown here) should connect the two sides (pages) of this graph. "},{"text":"1 Expon Fig. 10. Continued "},{"text":"Table 5 . Summary of two runs of the anchoveta growth simulation model, for high (  "},{"text":"\"1950\") and low (\"1980\")  anchoveta biomass.     anchoveta biomass.  1950 conditions  1980 conditionsb 1950 conditions1980 conditionsb Parameter (units) mean range mean range Parameter (units)meanrangemeanrange Weight increment 0.27 0.17-0.41 0.65 0.32-1.49 Weight increment0.270.17-0.410.650.32-1.49 (% B W D ) ~     (% B W D ) ~ Length increment 0.056 0.002-0.1 0.09 0.03-0.4 Length increment0.0560.002-0.10.090.03-0.4 (mm .d-l)     (mm .d-l) Metabolic losses 1.5 1.27-2.02 1.16 1.07-1.69 Metabolic losses1.51.27-2.021.161.07-1.69 (% BWD)     (% BWD) Ration 3.5 2.8-4.6 2.60 2.1-4.6 Ration3.52.8-4.62.602.1-4.6 (% BWD)     (% BWD) Growth efficiency 9.3 5.0-12.2 23.76 10.7-22.4 Growth efficiency9.35.0-12.223.7610.7-22.4 (%I     (%I Duration of feeding 15.3 11.3-24.0 12 10.7-22.4 Duration of feeding15.311.3-24.01210.7-22.4 (h.d-1)     (h.d-1)  "},{"text":"Table 6 . Sensitivity analysis of anchoveta growth simulation model.  Small change (25%) Resulting Changes needed to Small change (25%)ResultingChanges needed to  applied to \"1950\" lengtha at move from \"1950\" to applied to \"1950\"lengtha atmove from \"1950\" to Parameter conditions 4 Yr \"1980\" conditions Parameterconditions4 Yr\"1980\" conditions  from  to (in all1 % change ie., to fromto(in all1% changeie., to Duration of feeding 15.3h  11.4h 17.2 -7 6 3.6 h Duration of feeding15.3h11.4h17.2-7 63.6 h period      period Metabolic cost per unit 100%  75% 17.8 -7 0 30\" Metabolic cost per unit100%75%17.8-7 030\" time feeding      time feeding Assimilation 0.60  0.75 18.7 +45 0.87 Assimilation0.600.7518.7+450.87 coefficient      coefficient Energy content of 1,000  1,250 18.7 +45 1,45 0 Energy content of1,0001,25018.7+451,45 0 1 g of food (cal)      1 g of food (cal)  - - -   ---  "},{"text":"Table 3 . Independent estimates of anchoveta biomass off Peru (4-14's) and estimated values of baseline natural mortality (Mo) using VPA 111.  Independent   Independent  biomass Source of VPA 111 biomassSource ofVPA 111  estimate biomass estimates estimatebiomassestimates Date (t x lo6 )a estimates of Mo Date(t x lo6 )aestimatesof Mo  "},{"text":"Table 1 . Total monthly withdrawalsof anchoveta, 1953Total monthly withdrawalsof anchoveta,  -1982 (4-14's)  (4-14's)  in tonne.s.8 A d A d  "},{"text":"Table 3 . Independent estimates of anchoveta biomass off Peru (4-14's) and estimated values of baseline natural mortality (Mo) using VPA 111.  Independent   Independent  biomass Source of VPA 111 biomassSource ofVPA 111  estimate biomass estimates estimatebiomassestimates Date (t x 1 0 ' ) ~ estimates of Mo Date(t x 1 0 ' ) ~estimatesof Mo Jan 64 14.20   Jan 6414.20 Jan 65 11.20   Jan 6511.20 Jan 66 13.30   Jan 6613.30 Jan 67 13.80   Jan 6713.80 Jan 68 13.30 Fig. 1 and text  Jan 6813.30Fig. 1 and text Jan 69 12.50 IMARPE (1974a)  Jan 6912.50IMARPE (1974a) Jan 70 18.30   Jan 7018.30 Jan71 15.40   Jan7115.40 Mar 72 3 . OO   Mar 723 . OO Sep 72 2.00   Sep 722.00 Jan 73 3.16   Jan 733.16 Feb 73 3.30~   Feb 733.30~ Sep 73 3.09   Sep 733.09 Nov 73 4.48 IMARPE  Nov 734.48IMARPE Feb 74 3.39 IM-168 (1974b)  Feb 743.39IM-168 (1974b) May 74 2.19   May 742.19 Aug 74 3.92   Aug 743.92 Sep 74 3.09   Sep 743.09 Nov 74 3.25   Nov 743.25 Feb 75 4.32   Feb 754.32 Aug 75 3.39   Aug 753.39 Sep 75 4.27   Sep 754.27 Jan 76 7.41   Jan 767.41 Aug 76 4.62 Johannessun  Aug 764.62Johannessun Feb 77 1.89 and Vilchez  Feb 771.89and Vilchez Ju177 1.39   Ju1771.39 Jun 78 3.78   Jun 783.78 Nov 78 2.02   Nov 782.02 Apr 79 2.15   Apr 792.15  "},{"text":"Table 5 . VPA III estimates of biomass per length class and month for Pemvian anchoveta (Engradis ringens, northem/central stock, 4-140s) for 1953. in tonnes. Midlength             Midlength flL, cm) Jan Feb Mar API May Jun Jul Aug S ~P Oct Nov Dec flL, cm)JanFebMarAPIMayJunJulAugS ~POctNovDec 4.25             4.25 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 Midlength Jan Feb Mar API May Jun Jul Aug Sep Oct Nov Dec MidlengthJanFebMarAPIMayJunJulAugSepOctNovDec 4.25             4.25 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 Sum             Sum  "},{"text":"Table 7 . Biomass per length class and month of Peruvian anchoveta (Engradis ringem, northedcentral stock, 4-140s) for 1955, in tonnes. Midlength Jan F eb Mar A P ~ May Jun Jul Aug SeP Oct Nov Dec MidlengthJanF ebMarA P ~MayJunJulAugSePOctNovDec 4.25             4.25 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 Sum             Sum  "},{"text":"Table 8 . Biomass per length class and month of Peruvian anchoveta (Engradis ringens, northemtcentral stock, 4-140s) for 1956, in tonnes.  Midlength Jan ~e b ~a r APT May lun ~u l A w Ssp oct NOV Dsc MidlengthJan~e b~a rAPTMaylun~u lA wSspoctNOVDsc Sum 6,456,791            Sum6,456,791  "},{"text":"Table 9 . Biomass per length class and manth of Peluvian anchoveta (Engradis ringens, northem/central stock, 4-140s) for 1957, in tonnes.  Jan Feb Mar Jun Jul Oct Nov JanFebMarJunJulOctNov  5 1,445 59,132 96,045 100,776  57,800 5 1,44559,13296,045100,77657,800  85,900 91,929 149,510 162,046  102,445 85,90091,929149,510162,046102,445  131,439 135,577 184,811 208.418  290,433 131,439135,577184,811208.418290,433  181,382 189,677 219,453 236,646  297.306 181,382189,677219,453236,646297.306  221,965 245,622 259,658 263,421  352,865 221,965245,622259,658263,421352,865  262,357 285298 316,228 307,543  318,924 262,357285298316,228307,543318,924  279,135 310,373 375,648 371,254  288,521 279,135310,373375,648371,254288,521  319,972 330,636 376,696 376.726  273,247 319,972330,636376,696376.726273,247  265,055 303217 309,136 329,090  275073 265,055303217309,136329,090275073  208,863 216,456 248,321 201,547  200.382 208,863216,456248,321201,547200.382  163,666 131.592 137,352 106,892  99,688 163,666131.592137,352106,89299,688  239,157 206,790 107,248 86,077  56,413 239,157206,790107,24886,07756,413  331,538 328,069 194,099 172.755  80,603 331,538328,069194,099172.75580,603  19,419 48,799 95,355 106,001  13,989 19,41948,79995,355106,00113,989  "},{"text":"Sum 2,990,330 2,777.737 2,902,278 3,035,479 2,994,961 3,089,527 3,049,796 3,050,043 2,876,937 2,791,409 2,720,782 2.573984 Table 10. Biomass per length class and month of Peruvian anchoveta (Engradis ringem, northemlcentral stock, 4-140s) for 1958, in tonnes.   "},{"text":"Table 11 . Biomass per length class and month of Peruvian anchoveta (Engradis ringem, northem/central stock, 4-140s) for 1959, in tomes.    -- -- --  ------ Midlmth Jan Feb Mar May Jun Jul MidlmthJanFebMarMayJunJul 4.25 150,504 196,129 237,778 232,422 252,386 297,846 4.25150,504196,129237,778232,422252,386297,846 5.25 227,025 262,409 376,148 427,597 397,556 406,160 5.25227,025262,409376,148427,597397,556406,160 6.25 293,705 352,587 422,852 678,350 626,023 603,670 6.25293,705352,587422,852678,350626,023603,670 7.25 397,204 408,859 520,486 858,243 829,776 843,758 7.25397,204408,859520,486858,243829,776843,758 8.25 488,146 489,764 556,874 796,306 918,033 987,639 8.25488,146489,764556,874796,306918,033987,639 9.25 536,017 516,180 598,910 756,133 883,114 930,761 9.25536,017516,180598,910756,133883,114930,761 10.25 479,395 516,510 561,437 672,607 767,295 799,917 10.25479,395516,510561,437672,607767,295799,917 11.25 910,908 657,181 463,134 551,168 661,808 678,541 11.25910,908657,181463,134551,168661,808678,541 12.25 342,941 607,350 900,832 474,977 536,207 548,368 12.25342,941607,350900,832474,977536,207548,368 13.25 341,984 321,134 308,831 829,816 601,669 413,908 13.25341,984321,134308,831829,816601,669413,908 14.25 382,587 347,169 330,997 294,361 525,294 746,125 14.25382,587347,169330,997294,361525,294746,125 15.25 468,826 417,927 389,709 340,185 313,885 286,354 15.25468,826417,927389,709340,185313,885286,354 16.25 543,580 505,585 471,452 408,370 374,398 339,826 16.25543,580505,585471,452408,370374,398339,826 17.25 10,240 106,493 206,721 169,256 215,736 239,836 17.2510,240106,493206,721169,256215,736239,836 18.25       18.25 19.23       19.23 20.25       20.25 Sum      8,184,362 Sum8,184,362 Midlength Jan     Jul oct Nov Dec MidlengthJanJuloctNovDec 4.25 234,496      309,562 302,184 289,814 4.25234,496309,562302,184289,814 5.25 408,252      515,341 514,060 507,388 5.25408,252515,341514,060507,388 6.25 636,390      768,936 758,373 765,793 6.25636,390768,936758,373765,793 7.25 915,652      1,046,281 1,048,687 1,046,488 7.25915,6521,046,2811,048,6871,046,488 8.25 1,087,594      1,287,515 1,301,497 1,326,802 8.251,087,5941,287,5151,301,4971,326,802 9.25 942,163      1,498,220 1,496,626 1,555,648 9.25942,1631,498,2201,496,6261,555,648 10.25 1,158,115      1,546,983 1,634,375 1,717,459 10.251,158,1151,546,9831,634,3751,717,459 11.25 960,075      1,622,502 1,624,768 1,684,182 11.25960,0751,622,5021,624,7681,684,182 12.25 524,347      1,722,405 1,659,723 1,551,754 12.25524,3471,722,4051,659,7231,551,754 13.25 227,550      1,468,154 1,570,733 1,461,177 13.25227,5501,468,1541,570,7331,461,177 14.25 158,017      570,183 691,633 874,931 14.25158,017570,183691,633874,931 15.25 408,404      216,298 170,208 197,647 15.25408,404216,298170,208197,647 16.25 171,094      107,502 79,035 77,823 16.25171,094107,50279,03577,823 17.25 74,145      123,093 115,946 110,534 17.2574,145123,093115,946110,534 18.25       18.25 19.25       19.25 20.25       20.25 Sum 7,980,528      12,869,625 Sum7,980,52812,869,625  "},{"text":"Table 13 . Biomass per length class and month of Peruvian anchoveta (Engradis ringens, northemlcentral stock, 4-140s) for 1961, in tonnes. Mdlength Jan Feb Mar A P ~ May Jun Jul Aug S ~P Oct Nov Dec MdlengthJanFebMarA P ~MayJunJulAugS ~POctNovDec 4.25 282,817 293,158 339,722 353,165 341,723 318,174 297,757 293,344 282,810 267,862 238,669 212,535 4.25282,817293,158339,722353,165341,723318,174297,757293,344282,810267,862238,669212,535 5.25 482,130 493,639 553,267 625,568 607,726 562,264 511,679 505,061 512,590 470,433 433,848 390,066 5.25482,130493,639553,267625,568607,726562,264511,679505,061512,590470,433433,848390,066 6.25 748,970 734,811 818,210 881,140 959,017 892,032 807,776 784,628 771,316 750,808 681,687 632,341 6.25748,970734,811818,210881,140959,017892,032807,776784,628771,316750,808681,687632,341 7.25 1,017,978 1,064,097 1,123,721 1,213,668 1,238,174 1,271,996 1,180,474 1,143,004 1,120,477 1,056,291 989,698 925,508 7.251,017,9781,064,0971,123,7211,213,6681,238,1741,271,9961,180,4741,143,0041,120,4771,056,291989,698925,508 8.25 1,281,975 1,349,015 1,524,657 1,562,445 1,588,371 1,555,359 1,518,713 1,568,984 1,541,656 1,443,912 1,314,868 1,234.181 8.251,281,9751,349,0151,524,6571,562,4451,588,3711,555,3591,518,7131,568,9841,541,6561,443,9121,314,8681,234.181 9.25 1,590,841 1,614,818 1,822,848 1,992,946 1,944,910 1,876,318 1,793,339 1,785,247 1,906,117 1,897,335 1,711,919 1,578.254 9.251,590,8411,614,8181,822,8481,992,9461,944,9101,876,3181,793,3391,785,2471,906,1171,897,3351,711,9191,578.254 10.25 1,746,525 1,889,215 2,099,263 2,252,671 2,344,848 2,266,295 2,061,108 2,039,918 2,130,651 2,056,177 2,050,932 1,986,571 10.251,746,5251,889,2152,099,2632,252,6712,344,8482,266,2952,061,1082,039,9182,130,6512,056,1772,050,9321,986,571 11.25 1,804,715 1,954,542 2,292,622 2,514,161 2,501,221 2,426,368 2,331,082 2,322,205 2,314,130 2,242,123 2,186,522 2,155,969 11.251,804,7151,954,5422,292,6222,514,1612,501,2212,426,3682,331,0822,322,2052,314,1302,242,1232,186,5222,155,969 12.25 1,561,342 1,840,135 2,239,165 2,445,055 2,596,824 2,492,108 2,306,642 2,326,872 2,445,067 2,404,963 2,252,020 2,164,741 12.251,561,3421,840,1352,239,1652,445,0552,596,8242,492,1082,306,6422,326,8722,445,0672,404,9632,252,0202,164,741 13.25 1,337,263 1,374,389 1,708,746 2,154,749 2,324,475 2,239,041 2,126,363 2,148,048 2,222,138 2,179,442 2,096,726 2,103,808 13.251,337,2631,374,3891,708,7462,154,7492,324,4752,239,0412,126,3632,148,0482,222,1382,179,4422,096,7262,103,808 14.25 924,170 872,013 979,861 1,250,361 1,510,564 1,625,588 1,624,007 1,670,501 1,670,730 1,677,500 1,607,136 1,496,940 14.25924,170872,013979,8611,250,3611,510,5641,625,5881,624,0071,670,5011,670,7301,677,5001,607,1361,496,940 1525 187,258 238,259 391,339 468,718 514,256 576,607 639,145 702,199 839,847 879,548 815,084 674,194 1525187,258238,259391,339468,718514,256576,607639,145702,199839,847879,548815,084674,194 16.25 70,019 76,933 59,413 73,876 80,602 106,721 119,316 136,272 148,688 164,388 129,500 89,198 16.2570,01976,93359,41373,87680,602106,721119,316136,272148,688164,388129,50089,198 17.25 30,063 38,113 3,808 5,597 5,420 5,760 6,562 7,604 4,687 6,535 6.696 4,818 17.2530,06338,1133,8085,5975,4205,7606,5627,6044,6876,5356.6964,818 18.25             18.25 19.25             19.25 20.25             20.25  "},{"text":"Table 14 . Biomass per length class and month of Peruvian anchoveta (Engradis ringens, northemlcentral stock, 4-140s) for 1962, in tomes. Midlength Jan Feb Mar Jun Jul Oct Nov Dec MidlengthJanFebMarJunJulOctNovDec 4.25 194,713 182,437 184,924 205,287 220,945    4.25194,713182,437184,924205,287220,945 5.25 359,441 326,395 338,693 319,854 339,259    5.25359,441326,395338,693319,854339,259 6.25 585,714 539,376 533,856 458,185 468,101    6.25585,714539,376533,856458,185468,101 7.25 869,114 815,784 818,282 643,711 625,177    7.25869,114815,784818,282643,711625,177 8.25 1,221,443 1,137,835 1,166,697 905,646 845,447    8.251,221,4431,137,8351,166,697905,646845,447 9.25 1,526,809 1,522962 1,548,684 1,194,717 1,113,319    9.251,526,8091,5229621,548,6841,194,7171,113,319 10.25 1,921,621 1,813,884 1,963,416 1,558,261 1,406,386    10.251,921,6211,813,8841,963,4161,558,2611,406,386 11.25 2,228,015 2,211,722 2286,742 1,867,485 1,704,019    11.252,228,0152,211,7222286,7421,867,4851,704,019 12.25 2,182,356 2,322,334 2,599,352 2,161,376 1,988,211    12.252,182,3562,322,3342,599,3522,161,3761,988,211 13.25 2,089,439 2,185,871 2,497,450 2,335,910 2,150,553    13.252,089,4392,185,8712,497,4502,335,9102,150,553 14.25 1,442,881 1,649,937 1,979,780 2,102,246 2,064,318    14.251,442,8811,649,9371,979,7802,102,2462,064,318 15.25 604,311  832,240 1,131,984 1,196,071    15.25604,311832,2401,131,9841,196,071 16.25 106,416  160,532 215,396 270,841    16.25106,416160,532215,396270,841 17.25 4,826  5,850 25,542 27,779    17.254,8265,85025,54227,779 18.25     70    18.2570 19.25         19.25 20.25         20.25 Sum 15,398,889        Sum15,398,889  "},{"text":"Table 15 . Biomass per length class and month of Peruvian anchoveta (Engraulir ringens, northemlcentral stock, 4-140s) for 1963, in tomes. . . . .  "},{"text":"Table 16 . Biomass per length class and month of Peruvian anchoveta (Engradis ringens, northemlcentral stock, 4-140s) for 1964, in tomes. Midlength Jan Feb Mar Aer May Juu Jul Aug SeP Oct Nov Dec MidlengthJanFebMarAerMayJuuJulAugSePOctNovDec 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.             14.  "},{"text":"Table 2 . Sea surface temperature off Peru in OC.a  Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec YearJanFebMarAprMayJunJulAugSepOctNovDec  "},{"text":"a Mean of values off Talara, Chimbote and Callao and thus referring to the entire Peru coast between 4 and 14OS. Based on data provided by P. Lagos (Institute Geofisico del Peru, pers. comm.) complemented by data from Zuta and Urquizo (1972).   "},{"text":"Table 1 . Southern Oscillation Index (SOI). Monthly mean pressure difference in millibars between Easter Island and Darwin, Australia. Data courtesy of Dr. W. Quinn, Oregon State University.  Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec YearJanFebMarAprMayJunJulAugSepOctNovDec  "},{"text":"Table 2 . Monthly mean sea surface temperature in degrees Celsius at Talara, Peru. Data courtesy of Dr. D. Enfield, Oregon State University.   "},{"text":"Table 3 . Monthly mean sea surface temperature in degrees Celsius at La Punta, Peru. Data courtesy of Dr. D. Enfield, Orqon State University.  Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec YearJanFebMarAprMayJunJulAugSepOctNovDec  "},{"text":"Table 4 . Monthly mean sea level (cm) at Talara, Peru. Data courtesy of Dr. D. Enfield, Oregon State University. Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec YearJanFebMarAprMayJunJulAugSepOctNovDec  "},{"text":"Table 5   "},{"text":" -time as BATHY messages for support of real-time ocean analyses. BATHY messages, although more timely, require additional editing to correct digitizing and transmission errors.      -- - - - -    ------ Year Jan Feb Mar Apr May Jun  Jul Aug Sep Oct Nov Dec YearJanFebMarAprMayJunJulAugSepOctNovDec  "},{"text":"Table 6 . Monthly mean surface temperature (C) for the region 4-14O off Peru. These means are averages of the monthly means for the three Central areas, computed from subsurface temperature profiles The corresponding number of profiles used in computing each mean is printed (in brackets) to the right of the mean. Means based on zero profiles are computed from interpolated values. Asterisks indicate months in which neither observed nor interpolated means were available in all three areas.   -           - Year Jan Feb Mar A P ~ May Jun Jul Aug S ~P Oct Nov Dec YearJanFebMarA P ~MayJunJulAugS ~POctNovDec  "},{"text":"Spatially -averaged SST Depth of 14°C isotherm off Peru Depth of SST -2°C Heat content to depth of SST -2°C D Y e a r   "},{"text":"Table 7 . Monthly mean depth of the 14OC isotherm (m) for the region 4-14OS off Peru. These means are averages of the monthly means for the three central areas, computed from subsurface temperature profiles. The corresponding number of profiles used in computing each mean is printed (in brackets) to the right of the mean. Means based on zero profiles are computed from interpolated values. Asterisks indicate months in which neither observed nor interpolated means were available in all three areas. Year Jan Feb Mar A P ~ May Jun Jul Aug S ~P Oct Nov Dec YearJanFebMarA P ~MayJunJulAugS ~POctNovDec  "},{"text":"Table 8 . Monthly mean depth of the SST-2OC isotherm (m) for the region 4-14's off Peru. These means are averages of the monthly means for the three central areas, computed from subsurface temperature profiies. The corresponding number of profiles used in computing each mean (in brackets) is printed to the right of the mean. Means based on zero profiles are computed from interpolated values. Asterisks indicate months in which neither observed nor interpolated means were available in all three areas. Year Jan Feb Mar A P ~ May Jun Jul A% S ~P Oct Nov Dec YearJanFebMarA P ~MayJunJulA%S ~POctNovDec    (195)          (195)    (0)          (0)    (62)          (62)   ** ** (0) (0)          ** **(0) (0)    (0)          (0)    (SO)          (SO)    (90)          (90)    (36)          (36)    (47)          (47)    (97)          (97)    (35)          (35)    (55)          (55)    (47)          (47)    (1)          (1)    (65)          (65)    (16)          (16)    (1)          (1)    (8)          (8)    (0)          (0)    (0)          (0)    (96)          (96)    (114)          (114)    (45)          (45)    (56)          (56)    (6)          (6)    (0)          (0)   ** (5) (0)          **(5) (0)    (17)          (17)    (21)          (21)    (36)          (36)    (6)          (6)    (0)          (0)  "},{"text":"Table 9 . Monthly mean heat content from the surface down to the SST-20°C isotherm (\"C x m) for the region 4-14's off Peru. These means are averages of the monthly means for the three central areas, com- puted from subsurface temperature pmfies. The corresponding number of profies used in computing each mean (in brackets) is printed to the right of the mean. Meansbased on zero profie are computed fmm interpolated values. Asterisks indicate months in which neither observed nor interpolated means were available in all three areas.  Year Jan Feb  Mar A P ~  May Jun Jul Aug S ~P Oct  Nov  Dec  YearJanFebMarA P ~MayJunJulAugS ~POctNovDec 491 270 458 (1) (37) (0) 524 401 465 (1) (0) (0) 506 (188) 440 (0) 419 (59) 609 *** 558 (137) (0) (0) 748 789 613 (0) (0) (24) 772 895 748 (0) (0) 1,154 (122) 1,069 801 (31) 1,063 (0) 984 (0) *** (3) (0) 1,036 984 (0) (0) 837 *** (0) (0) 596 *** (0) (0) 553 *** (0) (0) 491 270 458(1) (37) (0)524 401 465(1) (0) (0)506 (188) 440 (0) 419 (59)609 *** 558 (137) (0) (0)748 789 613(0) (0) (24)772 895 748(0) (0) 1,154 (122) 1,069 801 (31) 1,063 (0) 984 (0) ***(3) (0) 1,036 984(0) (0)837 ***(0) (0)596 ***(0) (0)553 ***(0) (0)  "},{"text":"Table 1 . Numbers of observations in monthly samples used toconstructtime series. For each month the first number refers to observations used in constructing the values in Tables 2,4,5,6 & 7 (sea temperature, atmospheric pressure, wind stress components, and \"wind cubed\" index; the 2nd number refers to observations used in constructing the values in Table 4 (cloud cover); the 3rd number refers to observations used for values in Tables 8,9,10 & 11 (heat exchange components).  Jan 21 19 19 F e b 1 29 29 291 Mar 22 22 19 Awr 23 23 10 Mav 1 33 33 331 16 16 6 Jun Jan 21 19 19F e b 1 29 29 291Mar 22 22 19Awr 23 23 10Mav 1 33 33 331 16 16 6 Jun 61 57 40  21 21 21 59 59 53 61 57 4021 21 2159 59 53 24 22 20  16 15 12 9 9 5 24 22 2016 15 12995 -41 38 15  16 16 3 39 39 21 -41 38 1516 16339 39 21 32 32 32  98 98 47 38 38 29 32 32 3298 98 4738 38 29 153 153 106  69 69 34 84 84 28 153 153 10669 69 3484 84 28 128 125 51  143 138 37 115 112 37 128 125 51143 138 37115 112 37 98 95 42  166 166 105 151 146 101 98 95 42166 166 105151 146 101 148 147 127  115 111 94 183 180 150 148 147 127115 111 94183 180 150 185 183 127  163 162 142 157 155 122 185 183 127163 162 142157 155 122 78 76 71  182 181 167 180 177 170 78 76 71182 181 167180 177 170 129 129 98  82 81 81 76 76 76 129 129 9882 81 8176 76 76 59 59 58  137 137 137 99 99 96 59 59 58137 137 13799 99 96 107 107 107  113 112 111 106 106 106 107 107 107113 112 111106 106 106 111 109 102  80 80 78 147 144 144 111 109 10280 80 78147 144 144 76 76 76  108 108 108 139 137 137 76 76 76108 108 108139 137 137 67 67 67  134 131 131 91 91 91 67 67 67134 131 13191 91 91 90 87 82  103 100 90 51 51 48 90 87 82103 100 9051 51 48 103 101 93  89 89 89 53 53 52 103 101 9389 89 8953 53 52 109 108 104  49 49 47 85 85 85 109 108 10449 49 4785 85 85 55 54 40  89 89 89 50 50 48 55 54 4089 89 8950 50 48 84 78 78  113 109 100 57 54 54 84 78 78113 109 10057 54 54 59 57 56 70 69 68 51 51 51  140 140 140 76 73 71 79 79 73 99 92 90 59 59 55 101 99 94 ' I 59 57 56 70 69 68 51 51 51140 140 140 76 73 71 79 79 7399 92 90 59 59 55 101 99 94' I 57 56 56  71 69 67 92 92 88 57 56 5671 69 6792 92 88 97 96 89  105 104 98 85 85 77 97 96 89105 104 9885 85 77 96 93 80  125 123 102 75 70 54 96 93 80125 123 10275 70 54 116 113 85  139 137 124 108 103 95 116 113 85139 137 124108 103 95 91 89 82  152 146 139 100 98 88 91 89 82152 146 139100 98 88 68 64 54  94 91 86 76 72 61 68 64 5494 91 8676 72 61 90 75 26  90 59 14 75 41 16 90 75 2690 59 1475 41 16 J u l     oct J u loct 10 10 9    18 11 10 10918 11 40 39 38    43 37 40 39 3843 37      44 34 44 34      40 29 40 29      110 65 110 65      108 47 108 47      74 21 74 21      75 46 75 46      127 80 127 80      125 125 125 125      70 70 70 70      127 121 127 121      67 67 67 67      125 125 125 125      71 71 71 71      109 108 109 108      102 97 102 97      50 50 50 50      51 47 51 47      74 74 74 74      54 50 54 50      52 50 52 50      65 60 65 60      76 73 76 73      104 104 104 104      49 49 49 49      77 53 77 53      118 89 118 89      115 97 115 97      102 81 102 81      150 90 150 90      54 25 54 25  "},{"text":"58 39 61 Nov Dec 24 24 181 19 18 11 30 30 6 ) 18 18 9 27 27 201 13 13 5 63 61 201 24 24 6 100 99 531115 115 86 115 110 631 58 43 26 110 110 55 (114 114 85 63 63 231 82 81 59 197 197 1921135 133 118 173 172 1521 91 88 79 61 61 581 56 56 53 62 61 581 105 104 102 177 176 1761111 111 106 89 89 881123 123 123 73 73 731 99 97 99 118 118 1181 66 65 65 137 137 1271 101 101 96 127 125 124) 68 68 67 78 78 781 34 31 30 31 28 261 51 51 48 100 98 981 71 71 70 78 78 771 44 44 44 68 68 67) 88 88 88 46 44 441 58 57 57 183 166 163) 67 59 59 99 79 761 87 87 85 77 75 591 39 35 26 80 74 611 70 66 47 174 155 1421154 147 127 132 124 1131 93 89 51 73 73 241 96 90 32 62 40 141 77 61 35 Table 2. Sea surface temperature in degrees Celsius. The standard error of the mean appears within parentheses to the right of the mean temperature value. Feb Ma .50) 23.081.64) 24.63 Aur May Jun 24.78(.49)   ,  , 1 1 27.72i.15j 25.93i.30j 1 127.72i.15j25.93i.30j 20.95(.331 19.171.311 19.491.28) 20.95(.33119.171.31119.491.28) Oct Nov Dec OctNovDec .34) 18.24(.35) 18.72(.40) 18.81(.41) .34) 18.24(.35)18.72(.40)18.81(.41)  17.11(.26) 19.24(.54) 17.11(.26)19.24(.54)  "},{"text":"Table 3 . Total cloud amount. Values indicate mean fraction of sky obscured. The standard error of the mean appears in parentheses to the right of mean value.        each each  Jan Feb Mar A m Mav J JanFebMarA mMavJ 1953 .661.06) ,581.06) .431.06) .431.08) . 5 4 ( . 0 6 ) 1953.661.06),581.06).431.06).431.08). 5 4 ( . 0 6 )  "},{"text":"Table 4 . Atmospheric pressure at sea level. Add 1000.0 to the mean values in the table to yield pressure in millibars. The standard error of the mean appears in parentheses to the right of each mean value; the standard errors are given (directly) in millibars.  Jan Fa JanFa ll.2l .31 10.4 ll.2l .3110.4  "},{"text":"Table 5 . Alongshore component (positive equatorward) of wind stress on the sea surface. Units are dynes per square centimeter. The standard error of the mean appears in parentheses to the right of the mean alongshore stress value. Values in this table multiplied by the factor 3.95 (see text) yield offshore Ekman transport in cubic meters per second across each meter width.  -a Ma -Ja e Jun -a Ma -Ja e Jun 1953 .36(.06) .32(.05) ( 0 7 ) .99(.20) .83(.09) -37f.101 1953.36(.06).32(.05)( 0 7 ).99(.20).83(.09)-37f.101  .24(.03) .41(.14) .10(.02) .46(.05) .56(.08)  .24(.03).41(.14).10(.02).46(.05).56(.08)  "},{"text":"Table 6 . Onshore component of wind stress on the sea surface. Units are dyn'es per square centimeter. The standard error of the mean appears in parentheses to the right of each mean onshore stress value. Jan Feb Mar   "},{"text":"051 Jul Auu SeD Oct Nov        Dee Dee 1953 .15(.04) -06(.02) .12(.03) .07(.07) .08(.03) .04(.02) 1953.15(.04)-06(.02).12(.03).07(.07).08(.03).04(.02)  .06(.03) -.02(.11) -191.08) 0 4 0 1 .21(.05) .08(.02) .06(.03)-.02(.11)-191.08)0 4 0 1.21(.05).08(.02)  "},{"text":"Table 8 . Daily total (both direct and diffuse) solar radiation absorbed by the ocean, Qs. The standard error of the mean appears in parentheses to the right of each mean value. Units are watts@. (Values may be converted to units of cal ci2 daf' by multiplying by the factor 2.064 .) 7 1 260 1 101 223 (101 I   4 ) 147 4 )147 37) 2 7 0 ( 4 9 ) 235 37)2 7 0 ( 4 9 )235 2 1 ) 2 4 1 ( 1 7 ) 220 2 1 )2 4 1 ( 1 7 )220 2 9 ) 2 3 7 ( 49) 203 2 9 )2 3 7 ( 49)203 3 3 ) 2 3 3 ( 2 8 ) 226 3 3 )2 3 3 ( 2 8 )226 2 8 ) 1 8 1 ( 2 2 ) 188 2 8 )1 8 1 ( 2 2 )188 1 8 ) 2 4 9 ( 2 2 ) 146 1 8 )2 4 9 ( 2 2 )146 5 7 ) 3 1 6 ( 2 2 ) 275 5 7 )3 1 6 ( 2 2 )275 33) 2 2 7 ( 1 1 1 ) 122 33)2 2 7 ( 1 1 1 )122 301 3 5 0 t 271 229 3013 5 0 t 271229  "},{"text":"Table 9 . Radiative heat loss, QB. The standard error of the mean appears in parentheses to the right of each mean value. Units are watts/m2. (Values may be converted to units of cal cni2 day'' by multiplying by the factor 2.064.) Jan Feb Mar.  Awr Mav Jun AwrMavJun  "},{"text":"Table 11 . Net atmosphere . ocean heat exchange, QN. The standard error Of the mean appears in parentheses to the right of each mean value. Units are watts/m2. (Values may be ~onverted to units of cal cni2 day-' by multiplying by the factor 2.064.) an Feb Mar AD^  Mav Jun MavJun  "},{"text":"Transport Indexes, 1953 to 1985, Based on Local Wind Records from Trujillo and Callao, Peru JAIME MENDO Programa Cooperativo Peruano-Aleman ale Investigation Pesquera (PROCOPA) c/o Instituto ale1 Mar del Peru P.O. Box 22, Callao, Peru LUIS PIZARRO Instituto del Mar del Peru P.O. Box 22, Callao, Peru SADY CASTILLO Pedro Muiioz 323 Trujillo, Peru   "},{"text":"Table 1 . Chi-square test for the monthly upwelling indices (IA) computed using24,12,8,6, 3 and 1 (prevailing wind, PV) observations per day as recorded by CORPAC at Trujillo and Callao during 1970. The chi-squared values for 12, 8, 6.3 observations and for PV were computed with reference to the IA values for 24 observations per day. The critical value of the chi-squared statistics for P = 0.05 and 0.01 and 11 d.f. are19.68 and 24.7, respectively.         T~jiU0      c a b 0  T~jiU0c a b 0    Number of observations     Number of observations  Number of observationsNumber of observations 24  12 8 6 3 1 24 12 8 6 3 1 2412863124128631 IA I A CHM IA CHI-2 IA CHI-2 IA CHI-2 IA CHI-2      IAI ACHM IA CHI-2 IA CHI-2 IA CHI-2 IA CHI-2 Jan            Jan Feb            Feb Mar            Mar Apr            Apr May            May Jun            Jun Jul            Jul Aug            Aug %P            %P Oct            Oct Nov            Nov Dec            Dec Chi-square (a 0.49 0.99 1.28 6.13 34.18  1.1 1.112 2.45 8.76 15.86 Chi-square (a 0.490.991.286.1334.181.11.1122.458.7615.86  "},{"text":"Table 1   "},{"text":"Table 2 . Monthly upwelling indices for Tmjillo (m3/aec/100 m coastline). TrujiUo ( 0 8 ~0 6 ' ~ 7 9 °0 3 1 ~ Yanr Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dee Mean  Mean 213 205 207 229 233 213 192 219 247 242 230 217 220 Mean213205207229233213192219247242230217220  "},{"text":"Table 3 . Monthly upwelling indices for Calbo (m3/aec/100 m coastline). calla0 ( 1 2 °~~' ~ 77O07k)  -- -- - - ------  "},{"text":"Table 4 . Monthly turbulence indices for TrujUb (m3/sec3). TrujiUo (08~06's 79003') Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Mean Table 5. Monthly turbulence indices for Callao (m3/w3). calla0 ( ~Z ~O O ' S 77007'~)    -  ------  --       --------- Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Mean YearJanFebMarAprMayJunJulAugSepOctNovDecMean  "},{"text":"Table 1 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northernlcentral stock, 4-14'S), 1953. Midlength             Midlength VL, cm) Jan Feb Mar Apr May Jun Jul Aug Sep Octa N O V ~ ~e c ~ VL, cm)JanFebMarAprMayJunJulAugSepOctaN O V ~~e c ~  "},{"text":"Table 2 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northun/central stock, 4-14*S), 1954. Midlength             Midlength VL,cm) Jan Feb Mar ~p r ~ ~a y ~ Jun Jul ~u g ~ sepb Octa NO? ~e c ~ VL,cm)JanFebMar~p r ~ ~a y ~ JunJul~u g ~ sepbOctaNO?~e c ~ Cat8hc 8.60 7.17 6.79 6.72 6.44 6.52 4.44 3.78 4.58 7.43 12.10 12.40 Cat8hc8.607.176.796.726.446.524.443.784.587.4312.1012.40 c.f. (0.671) (0.666) (0.672) (0.685) (0.693) (0.702) (0.697) (0.710) (0.706) (0.705) (0.698) (0.674) c.f.(0.671)(0.666)(0.672)(0.685)(0.693)(0.702)(0.697)(0.710)(0.706)(0.705) (0.698)(0.674)  "},{"text":"Table 3 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northernlcentral stock, 4-14\"S), 1955. frequencies based on data collected at Callao, Casma, Chimbote, Don Martin, Huacho, Pisco and Samanco by staff of the Instituto de Investigacion de ecursos Marinos. Midlength (TL, cm) Janb Febb ~a r ~ A ~P Mayb Jun Julb Augb Sepa Octa Novb ~e c ~ Midlength (TL, cm)JanbFebb~a r ~ A ~P Mayb JunJulbAugbSepaOctaNovb~e c ~  "},{"text":"Table 5 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northernlcentral stock, 4-14'S), 1957. Mialength (TL, cm) Janb ~e b ~ Marb A ~P Mayb Junb Jutb Auga sepb Octb N O V ~ Mialength (TL, cm)Janb~e b ~ MarbA ~P MaybJunbJutbAugasepbOctbN O V ~  "},{"text":"Table 6 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northern/central stock, 4-14'S), 1958. Midlength            Midlength (TL, cm) Janb ~e b ~ M g Apra Mayb h n b Jula Auga sepa 0 c t NO? ~e c ~ (TL, cm)Janb~e b ~ M gApraMaybh n bJulaAugasepa0 c tNO?~e c ~  "},{"text":"Table 7 . Monthly catcha and percent catch compositionb of Peruvian anchoveta (E. ringens, northernlcentral stock, 4-14'S), 1959.  -- -- -- -- -- - - - --    --------------- Midlength             Midlength (TL, cm) Jan Feb Mar Apr  May Jun Jul Aug S ~P Oct Nov Dec (TL, cm)JanFebMarAprMayJunJulAugS ~POctNovDec 4.25             4.25 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 Catch             Catch c.tC             c.tC    -- -   -- --    ------- Midlength             Midlength (TL, cm) Jan  Feb Mar  Apr  May Jun Jul Aug SeP Oct Nov Dec (TL, cm)JanFebMarAprMayJunJulAugSePOctNovDec Catch 303            Catch303 c. f? (0.666)            c. f?(0.666) a ~o n t h l y catch in 1,000 t adapted from annual catch data in Murphy (1972) and mean seasonality of catch adapted from Figs. 1 and 2 in Doucet and a ~o n t h l y catch in 1,000 t adapted from annual catch data in Murphy (1972) and mean seasonality of catch adapted from Figs. 1 and 2 in Doucet and Ein sson (1967).            Ein sson (1967).  "},{"text":"Table 9 . Monthly catcha and percent catch compositionb of Peruvian anchoveta (E. ringens, northernlcentral stock, 4-14'S), 1961. Midlength             Midlength (TL, cm) Jan Feb Mar Apr May Jun Jul Aug SeP Oct Nov Dec (TL, cm)JanFebMarAprMayJunJulAugSePOctNovDec 4.25             4.25 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 Catch             Catch c.f?             c.f?    - - - -- -      ------ Midlength             Midlength (TL, cm) Jan Feb Mar Apr May Jun Jul Aug S ~P Oct Nov Dec (TL, cm)JanFebMarAprMayJunJulAugS ~POctNovDec 4.25             4.25 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 Catch             Catch c.f.             c.f.  "},{"text":"Table 11 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northernlcentral stock, 4-14's) based on data collected by IMARPE staff in 1963. Midlength             Midlength (TL, cm) Jan Feb Mar Apr May Jun Jul Aug S ~P Oct Nov Dec (TL, cm)JanFebMarAprMayJunJulAugS ~POctNovDec 4.25             4.25 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 Catch             Catch c.f.             c.f.  "},{"text":"Table 12 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northernlcentral stock, 4-14OS) based on data collected by IMARPE staff in 1964. Midlengt h            Midlengt h (TL,cm) Jan Feb Mar Apr May Jun Jul Aug S ~P Oct Nov Dec (TL,cm)JanFebMarAprMayJunJulAugS ~POctNovDec Catch 1,026            Catch1,026 c.f.             c.f.  "},{"text":"Table 13 . Monthly catch and percent catch composition of Peruvian anchoveta (E. Midlength            Midlength (TL, cm) Jan Feb Mar Apr May Jun Jul ~u g ~ S ~P Oct Nov Dec (TL, cm)JanFebMarAprMayJunJul~u g ~ S ~POctNovDec  "},{"text":"Table 15 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northern/central stock, 4-14\"s) based on data collected by IMARPE staff in 1967. Midlength           Midlength (TL,cm) Jan Feb Mar Apr May Jun Jula ~u g ~ S ~P Oct Nov Dec (TL,cm)JanFebMarAprMayJunJula~u g ~ S ~POctNovDec Catch 1,506           Catch1,506 c.f. 0.691           c.f.0.691 a ~o n t h s with closure of fishery (veda).         a ~o n t h s with closure of fishery (veda).  Jan Feb Mar Apr May Juna J U I ~ ~u g ~ S ~P Oct Nov Dec JanFebMarAprMayJunaJ U I ~~u g ~ S ~POctNovDec Catch 1,469           Catch1,469 c.fb 0.681           c.fb0.681  "},{"text":"Table 17 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northern/central stock, 4-14's) based on data collected by UdARPE staff in 1969. Midlength           Midlength VL, cm) Jan ~e b ~ Mar Apr May ~u n ~ J U I ~ ~u g ~ Sep Oct Nov Dec VL, cm)Jan~e b ~ MarAprMay~u n ~ J U I ~~u g ~ SepOctNovDec Catgh 1,226 0.00 1,762 1,252       Catgh1,2260.00 1,7621,252 c.f.  0.659 (0.666) 0.633 0.686       c.f.0.659(0.666)0.6330.686 a ~o n t h s with dosure of f i e r y (veda).        a ~o n t h s with dosure of f i e r y (veda). Midlength           Midlength VL, cm) Jan Feb Mar A P ~ May Jun ~u l ~ ~u~~ S ~P Oct Nov Dec VL, cm)JanFebMarA P ~MayJun~u l ~ ~u~~S ~POctNovDec Catgh 1,998          Catgh1,998 c.f.  0.627          c.f.0.627  "},{"text":"Table 19 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northern/central stock, 4-14% based on data collected by IMARPE staff in 1971. b~ondition factors in brackets estimated from the linear relationship between temperature and condition factor in Palomares et al. (thisvot). Midlength             Midlength (TL, cm) Jana ~e b ~ Mar Apr May Jun Jula ~u g ~ Sep Oct Nov Dec (TL, cm)Jana~e b ~ MarAprMayJunJula~u g ~ SepOctNovDec Catch 0.00 0.00 2,366 1,453 636 92.0 0.00 0.00 1,118 1,229 1,066 1,224 Catch0.000.00 2,3661,45363692.00.000.00 1,1181,2291,0661,224 c.f. (0.674) (0.665) 0.646 0.609 0.626 0.615 (0.676) (0.676) 0.669 0.713 0.761 0.726 c.f.(0.674)(0.665)0.6460.6090.6260.615(0.676)(0.676)0.6690.7130.7610.726 a ~o n t h s with closure of fishery (veda).          a ~o n t h s with closure of fishery (veda).  "},{"text":"Table 20 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northern/central stock, 4-14's) based on data collected by IMARPE staff in 1972. Midlength    Midlength (TL, cm) Jana Mar Jun (TL, cm)JanaMarJun  "},{"text":"Table 21 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northernlcentral stock, 4-14QS) based on data collected by IMARPE staff in 1973. Midlength (TL, cm) an^' ~e b ~ Mar Apr tdayb ~u n ~. ~ Jul A U ~~' ~e ~~. 02 Nova> ~ ~e c ~ Midlength (TL, cm)an^'~e b ~ MarAprtdayb~u n ~. ~ JulA U ~~' ~e~~. 02 Nova>~~e c ~ 4.25          4.25 5.25          5.25 6.25          6.25 7.25          7.25 8.25          8.25 9.25          9.25 10.25          10.25 11.25          11.25 12.25          12.25 13.25          13.25 14.25          14.25 15.25          15.25 16.25          16.25 17.25          17.25 18.25          18.25 19.25          19.25 20.25          20.25 Catch          Catch  "},{"text":"Table 22 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northernlcentral stock, 4-14\"s) based on data collected by IMARPE staff in 1974.  "},{"text":"Table 23 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northernlcentral stock, 4-14\"s) based on data collected by IMARPE staff in 1975. Midlength           Midlength (TL, cm) Jan Feb Mar Apr May ~u n ~ h l a ~u g ~ sepa Octb N O V ~ ~e . ' (TL, cm)JanFebMarAprMay~u n ~ h l a~u g ~ sepaOctbN O V ~~e . '  "},{"text":"Table 24 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northernlcentral stock, 4-14\"s) based on data collected by IMARPE staff in 1976. Midlength           Midlength (TL, cm) Jan Feb Mar Apr May Jun JUI ~u g ~' sepb ~c t NOV Dec (TL, cm)JanFebMarAprMayJunJUI~u g ~' sepb~c t NOVDec 4.25           4.25 5.25           5.25 6.25           6.25 7.25           7.25 8.25           8.25 9.25           9.25 10.25           10.25 11.25           11.25 12.25           12.25 13.25           13.25 14.25           14.25 15.25           15.25 16.25           16.25 17.25           17.25 18.25           18.25 19.25           19.25 20.25           20.25 Catch           Catch c.f?           c.f? a~ercent catchcomposition data obtained from EUREKA surveys.      a~ercent catchcomposition data obtained from EUREKA surveys.  "},{"text":"Table 25 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northern/central stock, 4-14% based on data collected by IMARPE staff in 1977. Midlength            Midlength (TL, cm) Jan           (TL, cm)Jan Catch 184           Catch184 C.E.~ 0.531           C.E.~0.531 a~ercent catch-composition data obtained fmm EUREKA (October), CRUCERO 7703 (March) and CRUCERO SNP-1-ICANE (November).  a~ercent catch-composition data obtained fmm EUREKA (October), CRUCERO 7703 (March) and CRUCERO SNP-1-ICANE (November). b ~o n t h s with closure of f ~e r y (vedu); catch for March mainly from surveys.      b ~o n t h s with closure of f ~e r y (vedu); catch for March mainly from surveys. Midlength (TL, cm) an^ Feb hfara Apr May Jun Jul ~u g ~ Sep Octb NOV Dec Midlength (TL, cm)an^FebhfaraAprMayJunJul~u g ~ SepOctbNOVDec 4.25            4.25 5.25            5.25 6.25            6.25 7.25            7.25 8.25            8.25 9.25            9.25 10.25            10.25 11.25            11.25 12.25            12.25 13.25            13.25 14.25            14.25 15.25            15.25 16.25            16.25 17.25            17.25 18.25            18.25 19.25            19.25 20.25            20.25 Catch            Catch c.f.C            c.f.C a ~o n t h s with closure of fishery (veda).         a ~o n t h s with closure of fishery (veda).  "},{"text":"Table 27 . Monthly catch and percent catch composition of Peruvian anchoveta (E. -composition data obtained fmm EUREKA surveys for northern region only (veda in central region). Months with closure of F l e r y (veda). 'condition factors in brackets estimated from the linear relationship between temperature and condition factor in Palomares et al. (this vol.). Midlength            Midlength (TL, cm) Janb ~e b ~ Mar Apr Mayb JU$ Julb ~u~~ ~e ~~. Oct Nov ~ ~e c ~ (TL, cm)Janb~e b ~ MarAprMaybJU$Julb~u~~~e~~. Oct Nov~~e c ~ 4.25            4.25 5.25            5.25 6.25            6.25 7.25            7.25 8.25            8.25 9.25            9.25 10.25            10.25 11.25            11.25 12.25            12.25 13.25            13.25 14.25            14.25 15.25            15.25 16.25            16.25 17.25            17.25 18.25            18.25 19.25            19.25 20.25            20.25 Catch            Catch ~. f . ~            ~. f . ~ :percent catch           :percent catch  "},{"text":"Table 28 . Monthly catch and percent catch composition of Peruvian anchoveta (E. ringens, northernlcentral stock, 4-14OS) based on data collected by IMARPE staff in 1980. C~ondition factors in Rackets estimated from the linear relationship between temperature and condition factor in Palomares et al. (this voL). Milength           Milength (TL, cm) Jana ~e b ~ hfara A P P May Jun Jula ~u g ~ sepa Octa N O V ~ Dec (TL, cm)Jana~e b ~ hfaraA P PMayJunJula~u g ~ sepaOctaN O V ~Dec 4.25           4.25 5.25           5.25 6.25           6.25 7.25           7.25 8.25           8.25 9.25           9.25 10.25           10.25 11.25           11.25 12.25           12.25 13.25           13.25 14.25           14.25 15.25           15.25 16.25           16.25 17.25           17.25 18.25           18.25 19.25           19.25 20.25           20.25 Catch           Catch c.f.C           c.f.C a ~o n t h s with closure of fishery (veda).         a ~o n t h s with closure of fishery (veda).  "},{"text":"Table 29 . Monthly catch and percent catch wmposition of Peruvian anchoveta (E. ringens, northern/central stock, 4-14's) based on data collected by IMARPE staff in 1981. catch in October mainly fmm Jurveys. b~ondition factors in brackets estimated from the linear relationship between temperature and condition factor in Palomares et al. (thisvol.). Midlength (TL, cm) an^ ~e b ~ ~a r ~ Apr May Jun h l a ~u g ~ sepa Octa Nov Dec Midlength (TL, cm)an^~e b ~ ~a r ~ AprMayJunh l a~u g ~ sepaOctaNovDec 4.25          4.25 5.25          5.25 6.25          6.25 7.25          7.25 8.25          8.25 9.25          9.25 10.25          10.25 11.25          11.25 12.25          12.25 13.25          13.25 14.25          14.25 15.25 , 16.25          15.25 , 16.25 17.25          17.25 18.25          18.25 19.25          19.25 20.25          20.25 Catch          Catch c.f.\"          c.f.\" a ~o n t h s with closure of f i e r y (veda);        a ~o n t h s with closure of f i e r y (veda);  "},{"text":"Table 30 . Monthly catch and percent catch wmposition of Peruvian anchoveta (E. ringens, northern/central stock, 4-14's) based on data collected by IMARPE staff in 1982. Midlength (TL, cm) an^ Feb Mar Apr May Sun Jul A U . ~~ sepb OctC Nov ~e c ~ Midlength (TL, cm)an^FebMarAprMaySunJulA U . ~~ sepbOctCNov~e c ~    -   -- --     -----  "},{"text":"Table 1 . Basic information on age and career development of 40 informants on the Peruvian anchoveta fishery. 1953 to 1982 INFORMANT A B C D E F G H I J K L M INFORMANTABCDEFGHIJKLM #              #  0 0 P           00P  0 0 0  P         000P  O O P           OOP  0 0 P           00P  0   P          0P  0 0 P           00P  O O P           OOP  0   P          0P  P   0          P0     P          P  0 0 0  P         000P  O P O   P        OPOP  0 0 0           000  O O P           OOP  O O P           OOP  0 0 P           00P  0 0 0           000  P             P  P   0          P0  0 0 0  O P P       000OPP       P 0       P0  O O P           OOP  0 0  0          000  O O P           OOP  P O 0           PO0  0 R            0R  P 0            P0  P             P  0 P  0          0P0  0 0 0  P         000P  O O P           OOP  O O R           OOR  P          50   P50  P          45   P45  0 P         57   0P57           P 44   P44          0 P 53   0P53       P P O O  48   PPOO48         O O P 5 1   O O P 5 1           P 48   P48 Legend:              Legend:  "},{"text":"Table 2 . Estimate by our informants of 'losses\" in the Peruvian anchoveta reduction fihery, by source of \"loss\". The corresponding estimates presented byVasquez and Lam (1977) are given for comparison. Mean 9.13 4.91 16.13 4.38 4.1 Mean9.134.9116.134.384.1 95% confidence 8.1- 4.5- 15.0- 3.9- 3.5- 95% confidence8.1-4.5-15.0-3.9-3.5- interval 10.1 5.5 17.2 4.9 4.7 interval10.15.517.24.94.7 Estimates of 5 4-5 5 0.5 0.5 Estimates of54-550.50.5 Vasquez and Lam (1977)      Vasquez and Lam (1977)  "},{"text":" Column A to E are t x 103/year. Year Estimated Corrected Corrected Nominal Fish meal Reduction Coefficients Conversion YearEstimatedCorrectedCorrectedNominalFish mealReductionCoefficientsConversion  catch landings f ~h landing production @B) (cffi) efficiency catchlandingsf ~hlandingproduction@B)(cffi)efficiency   (= landings + \"blood\" landings     @/A) (= landings + \"blood\"landings@/A)   losses)       losses)  A B C D Ea Fa G H ABCDEaFaGH  "},{"text":"Table   "},{"text":" 1. Age in days in anchoveta larvae of different lengths, based on different authors. Total Kramer and Mendiola and Herrera This TotalKramer andMendiola andHerreraThis length (mm)a Zweifei (1970)~ Gomez (1981)~ et al. (1985) studyd length (mm)aZweifei (1970)~Gomez (1981)~ et al. (1985)studyd  "},{"text":"Table 2 . Growth parameters of subadult and adult anchoveta Engraulis ringens off Peru and Chile, as reported from various authors, Sampling area (and period) Method L,(TL cm) K (y-l) Source  Chimbote-Callao (196 1-6 3) M P A ~ Saetersdal and Valdivia (1964) Chimbote-Callao (196 1-6 3)M P A ~Saetersdal and Valdivia (1964) Chimbote, Callao, 110 (1961 6 4 ) MPA Robles (MS) Chimbote, Callao, 110 (1961 6 4 )MPARobles (MS) Arica, Iquique (1970-72) Otoliths Simpson and Buzeta (1967) Arica, Iquique (1970-72)OtolithsSimpson and Buzeta (1967) Arica, Iquique (1970-72) Otoliths Simpson and Buzeta (1967) Arica, Iquique (1970-72)OtolithsSimpson and Buzeta (1967) Chimbote, Callao (196266) Otoliths Vildoso and Chuman (MSS), Chuman (MSS) Chimbote, Callao (196266)OtolithsVildoso and Chuman (MSS), Chuman (MSS) Peruvian Coast MPA Tsukayama and Zuzunaga (MS) Peruvian CoastMPATsukayama and Zuzunaga (MS) Arica, Iquique (1970-72) Otoliths Aguayo (1976) Arica, Iquique (1970-72)OtolithsAguayo (1976) Northern Peru (1961-1979) ELEFAN I Pauly and Tsukayama (1983)' Northern Peru (1961-1979)ELEFAN IPauly and Tsukayama (1983)' a r (b = loglO K + 2 loglO TL,.   a r (b = loglO K + 2 loglO TL,.  "},{"text":"Progression Analysis, see text. 'parameters presented here are means of 19 values in original paper, each of which pertained to a different year, from 1961 to 1979 (see text).   "},{"text":" Table 4). The model is based on the terms Growth = [food ingested]-[food used for purposes other than growth] ... 8 ... 8 or put differently or put differently  "},{"text":"Table 4 . Formulas and constants used in anchoveta growth simulation model. Relationship Equation(@ Sources and/or remarks RelationshipEquation(@Sources and/or remarks Length vs. weight   Length vs. weight Wet weight vs. caloric content   Wet weight vs. caloric content Swimming speed during feeding vs.   Swimming speed during feeding vs.  "},{"text":"Table 5 . Summary of two runs of the anchovetagrowth simulation model, for high (  "},{"text":"\"1950\") and low (\"1980\")  anchoveta biomass.     anchoveta biomass.  1950 conditions  1980 condition& 1950 conditions1980 condition& Parameter (units) mean range mean range Parameter (units)meanrangemeanrange Weight increment 0.27 0.174.41 0.65 0.32-1.49 Weight increment0.270.174.410.650.32-1.49 (% B W D ) ~     (% B W D ) ~ Length increment 0.056 0.002-0.1 0.09 0.03-0.4 Length increment0.0560.002-0.10.090.03-0.4 (mm .d-l)     (mm .d-l) Metabolic losses 1 5 1.27-2.02 1.16 1.07-1.69 Metabolic losses1 51.27-2.021.161.07-1.69 (% BWD)     (% BWD) Ration 3.5 2.8-4.6 2.60 2.1-4.6 Ration3.52.8-4.62.602.1-4.6 (% BWD)     (% BWD) Growth efficiency 9.3 5.0-12.2 23.76 10.7-22.4 Growth efficiency9.35.0-12.223.7610.7-22.4 (%I     (%I Duration of feeding 15.3 11.3-24.0 12 10.7-22.4 Duration of feeding15.311.3-24.01210.7-22.4 (h.d-1)     (h.d-1)  "},{"text":"Table 6 . Sensitivity analysis of anchoveta growth simulation model.  Small change (25%) Resulting Changes needed to Small change (25%)ResultingChanges needed to  applied to \"1 950\" lengtha at move from \"1950\" to applied to \"1 950\"lengtha atmove from \"1950\" to Parameter conditions 4 Yr \"1980\" conditions Parameterconditions4 Yr\"1980\" conditions  from to (in a) % change ie., to fromto(in a)% changeie., to Duration of feeding 15.3h 11.4h 17.2 -76 3.6 h Duration of feeding15.3h11.4h17.2-763.6 h period      period Metabolic cost per unit 100% 75% 17.8 -7 0 30\" Metabolic cost per unit100%75%17.8-7 030\" time feeding      time feeding Assimilation 0.60 0.75 18.7 +45 0.87 Assimilation0.600.7518.7+450.87 coefficient      coefficient Energy content of 1,000 1,250 18.7 +45 1,45 0 Energy content of1,0001,25018.7+451,45 0 1 g of food (cal)      1 g of food (cal)  "},{"text":"Table 7 . Review of some growth-related statistics in adults of Engruulis spp. Parameter Mean Range Species Source ParameterMeanRangeSpeciesSource Daily ration 4.5 4-6 E. mordax Hunter and Leong (1 981) Daily ration4.54-6E. mordaxHunter and Leong (1 981) (% body weight) 2.6 1.5-3.7 E. encrusicholus Sirotenko and Danilevsky (1977) (% body weight)2.61.5-3.7E. encrusicholusSirotenko and Danilevsky (1977)  2.4 1.4-3.4 E. encmsicholus Mikhman and Tomanovich (1977) 2.41.4-3.4E. encmsicholusMikhman and Tomanovich (1977)  3Sa 2.8-4.5 E. ringens this 3Sa2.8-4.5E. ringensthis  2.6b 2.1-4.6 E. ringens study 2.6b2.1-4.6E. ringensstudy Growth efficiency 12.8 - E. mordax Hunter and Leong (1981) Growth efficiency12.8-E. mordaxHunter and Leong (1981) (%I 9.36 6-12.2 E. ringens this (%I9.366-12.2E. ringensthis  23.9 15-32 E. ringem] study 23.915-32E. ringem]study Daily length 0.020~ 0.015-0.025 E. mordax Parrish et al. (1985) Daily length0.020~0.015-0.025E. mordaxParrish et al. (1985) increment (mm) 0.053~' 0.020-0.106  thig increment (mm)0.053~'0.020-0.106thig  0.056~' 0.020-0.016 E. ringens study 0.056~'0.020-0.016E. ringensstudy Daily weight 0.74~ - E. ringens Walsh (1975) Daily weight0.74~-E. ringensWalsh (1975) increment (% BWD) 0 . 2 8 ~ 0.17-0.41  this increment (% BWD)0 . 2 8 ~0.17-0.41this  0 . 6 5 ~ 0.32-1 S O E. ringens study 0 . 6 5 ~0.32-1 S OE. ringensstudy events 15.1 20.0 5.3-23.5 - E. mordax E. mordax Parrish et al. (1986) Hunter and Leong (1981) events15.1 20.05.3-23.5 -E. mordax E. mordaxParrish et al. (1986) Hunter and Leong (1981)  19.61 - E. ringens Alheit (1986) 19.61-E. ringensAlheit (1986)  6 . O 4.0-9.5  this 6 . O4.0-9.5this  1 7 . 3 ~ 14.0-22.0 E. ringens study 1 7 . 3 ~14.0-22.0E. ringensstudy  24.2 -   24.2-  "},{"text":" reformulated as \"Cohort Analysis\", ii) VPA 11, the VPA equivalent of Jones' (1981) \"Length Cohort Analysis\" (see also Jones and van Zalinge 1981; Pauly 1984), and iii) VPA 111, the model used here as originally proposed by Pope et al. (MS).  "},{"text":"Table 1 . TOM monthly withdrawals of anchoveta, 1953-1982 (4-14's) in tomeba  AnmrpI AnmrpI  "},{"text":"Table 2  . Inputs used for VPA 111 estimates of monthly anchoveta . Inputs used for VPA 111 estimates of monthly anchoveta biomass off Peru, 1953 to 1982.~    biomass off Peru, 1953 to 1982.~ Year L(-) b K'C M~~ Year L- K M, Year L(-)b K'CM~~YearL-KM, 1953 18.4 0.78 2.00 1968 20.4 0.87 2.00 1953 18.40.782.00196820.4 0.87 2.00 1954 18.6  0.78 2.00 1969 205 0.88 2.00 1954 18.60.782.001969205 0.88 2.00 1955 18.7 0.79 2,OO 1970 20.7 0.88 2.00 1955 18.70.792,OO197020.70.88 2.00 1956 18.8 0.79 2.00 1971 20.8 0.89 2.00 1956 18.80.792.00197120.8 0.89 2.00 1957 19.0 0.80 2.00 1972 20.9 0.90 1.95 1957 19.00.802.00197220.9 0.90 1.95 1958 19.1 0.81 2.00 1973 21.1 0.90 1.75 1958 19.10.812.00197321.1 0.90 1.75 1959 19.2 0.81 2.00 1974 21.2 0.91 1.15 1959 19.20.812.00197421.2 0.91 1.15 1960 19.4  0.82 2.00 1975 21.3 0.92 0.80 1960 19.40.822.00197521.3 0.92 0.80 1961 19.5  0.82 2.00 1976 21.5 0.93 1.70 1961 19.50.822.00197621.5 0.93 1.70 1962 19.6 0.83 2.00 1977 21.6 0.94 3.90 1962 19.60.832.00197721.6 0.94 3.90 1963 19.8 0.84 2.00 1978 21.7 0.94 4.00 1963 19.80.842.00197821.7 0.94 4.00 1964 19.9 0.84 2.00 1979 21.9 0.95 4.00 1964 19.90.842.00197921.9 0.95 4.00 1965 20.0  0.85 2.00 1980 22.0 0.96 4.00 1965 20.00.852.00198022.0 0.96 4.00 1966 20.2  0.85 2.00 1981 22.1 0.97 4.00 1966 20.20.852.00198122.1 0.97 4.00 1967 20.3 0.86 2.00 1982 22.2 0.98 4.00 1967 20.30.862.00198222.2 0.98 4.00 a ~t h e ~ inputs, used throughout, were C = 0.3, WP = 0.7 a ~t h e ~ inputs, used throughout, were C = 0.3, WP = 0.7 and Ft = 2 (see text).     and Ft = 2 (see text). h o t a l length, in cm; from Fig. 7 in Palomares et al. (this vol.). h o t a l length, in cm; from Fig. 7 in Palomares et al. (this vol.).  "},{"text":"Table 3   . lndependent estimates of anchoveta biomass off Peru . lndependent estimates of anchoveta biomass off Peru (444%) and estimated values of baseline natural mortality (444%) and estimated values of baseline natural mortality (Mo) using VPA Ill.   (Mo) using VPA Ill.  Independent   Independent  biomass Source of VPA I11 biomassSource ofVPA I11  estimate biomass estimates estimatebiomassestimates Date (t x lo6 la estimates of Mo Date(t x lo6 laestimatesof Mo Jan 64 14.20 '   Jan 6414.20 ' Jan 65 11.20   Jan 6511.20 Jan 66 13.30   Jan 6613.30 Jan 67 13.80   Jan 6713.80 Jan 68 13.30 Fig. 1 and text  Jan 6813.30Fig. 1 and text Jan 69 12.50 IMARPE (1974a)  Jan 6912.50IMARPE (1974a) Jan 70 18.30   Jan 7018.30 Jan 71 15.40   Jan 7115.40 Mar 72 3 .OO   Mar 723 .OO Sep 72 2.00   Sep 722.00 Jan 73 3.16   Jan 733.16 Feb 73 3.30~   Feb 733.30~ Sep 73 3.09   Sep 733.09 Nov 73 4.48 IMARPE  Nov 734.48IMARPE Feb 74 3.39 IM-168 (1974b)  Feb 743.39IM-168 (1974b) May 74 2.19   May 742.19 Aug 74 3.92   Aug 743.92 Sep 74 3.09   Sep 743.09 Nov 74 3.25 IMARPE  Nov 743.25IMARPE Feb 75 4.32 (1975)  Feb 754.32(1975) Aug 75 3.39   Aug 753.39 Sep 75 4.27   Sep 754.27 Jan 76 7.41   Jan 767.41 Aug 76 4.62 Johannesson  Aug 764.62Johannesson Feb 77 1.89 and Vilchez  Feb 771.89and Vilchez Ju177 1.39   Ju1771.39 Jun 78 3.78   Jun 783.78 Nov 78 2.02   Nov 782.02 Apr 79 2.15   Apr 792.15  "},{"text":"Table 4 . Number (x lo6) of $month old fish of 3.75-4.75 cm entering (i.e., recruiting into) the Peruvian anchoveta stock, 4-14\"S,January 1953 to April 1982.\"     Year Jan Feb Mar A P ~ May Jun Jul '4% S ~P YearJanFebMarA P ~MayJunJul'4%S ~P  "},{"text":"Table 5 . VPA III estimates of biomass per length class and month for Peruvian anchoveta (Engraulis ringens, northemlcentral stock, 4-140s) for 1953, in tonnes. Midlength             Midlength CTL, cm) Jan Feb Mar A P ~ May Jun Jul Aug S ~P Oct Nov Dec CTL, cm)JanFebMarA P ~MayJunJulAugS ~POctNovDec Sum 4,807,883            Sum4,807,883 Midlength Jan Feb Mar A P ~ May Jun Jul Aug SeP Oct Nov Dec MidlengthJanFebMarA P ~MayJunJulAugSePOctNovDec 4.25             4.25 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 Sum             Sum  "},{"text":"Table 7 . Biomass per length class and month of Peruvian anchoveta (Engraulis ringens, northedcentral stock, 4-140s) for 1955, in tonnes. Midlength Jan Feb Mar A P ~ May Jun Jul Aug sep Oct Nov Dec MidlengthJanFebMarA P ~MayJunJulAugsepOctNovDec Sum 7,814,363            Sum7,814,363  "},{"text":"Table 8 . Biomass per length class and month c $ Pemvian anchoveta (Engradis ringens, northedcentral stock, 4-140s) for 1 956. in tonnes. . Midlemgth Jan Fsb Mar Apl May Jun Jul A w sep Oct Nov Da: . MidlemgthJanFsbMarAplMayJunJulA wsepOctNovDa: Sum 6,456,791 6,310,710 6,182,133 6,112664 5937209 5,583,914 5,078,527 4,605,280     Sum6,456,7916,310,7106,182,1336,11266459372095,583,9145,078,5274,605,280  "},{"text":"Table 9 . Biomass per length class and manth of Peruvian anchoveta (Engradis ringens, northernlcentral stock, 4-140s) for 1957, in tonnes. Midlength Jan Feb Mar ' A P ~ May Jun Jul Aug SeP Oct Nov Da: MidlengthJanFebMar' A P ~MayJunJulAugSePOctNovDa: 4.25             4.25 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 Sum             Sum  "},{"text":"Table 10 . Biomass per length class and month of Pemvian anchoveta (Engradis ringens, northedcentral stock. 4-140s) for 1958, in tonnes. Midlength Jan Feb M ~I APr May Jun Jul AW sep Oct Nov Da: MidlengthJanFebM ~IAPrMayJunJulAWsepOctNovDa: 4.25 48,813 45,392 41,117 48,676 76,014 100,862 107,186 115,318 124,878 118,706 116,842 138,401 4.2548,81345,39241,11748,67676,014100,862107,186115,318124,878118,706116,842138,401 5.25 87,235 84,608 78,896 72,690 83,000 125,934 162,883 174,326 198,834 198,694 195,995 199,691 5.2587,23584,60878,89672,69083,000125,934162,883174,326198,834198,694195,995199,691 6.25 133,441 135,582 132,670 125,895 110,117 123,130 159,265 218,424 278,347 285,717 287,631 295,275 6.25133,441135,582132,670125,895110,117123,130159,265218,424278,347285,717287,631295,275 7.25 216,872 191,106 196,524 196,088 156,885 153,042 157,640 168,450 298.222 349,454 379.606 387,386 7.25216,872191,106196,524196,088156,885153,042157,640168,450298.222349,454379.606387,386 8.25 391,100 304,142 260,391 271,646 252,486 216,488 192,894 186,451 307,910 312,492 381,655 466,022 8.25391,100304,142260,391271,646252,486216,488192,894186,451307,910312,492381,655466,022 9.25 430.660 467,509 436,064 330,813 329,500 314,109 281,349 242.751 719,162 420,629 336,883 428,615 9.25430.660467,509436,064330,813329,500314,109281,349242.751719,162420,629336,883428,615 10.25 317,291 491,300 528,292 576,324 355,090 373,150 348,326 338,071 604,019 782,109 714289 403,238 10.25317,291491,300528,292576,324355,090373,150348,326338,071604,019782,109714289403,238 11.25 250,869 271,656 480,030 584,951 695,879 559,741 342,284 369,837 598,279 531,872 515.080 787.355 11.25250,869271,656480,030584,951695,879559,741342,284369,837598,279531,872515.080787.355 12.25 171,967 208,772 229,546 231,641 472,701 503,902 611,353 493,385 621.151 576.348 498,393 384,005 12.25171,967208,772229,546231,641472,701503,902611,353493,385621.151576.348498,393384,005 13.25 81,216 88,390 104,535 0 185,821 218,670 299,216 323,349 569,959 519.178 479634 455,083 13.2581,21688,390104,5350185,821218,670299,216323,349569,959519.178479634455,083 14.25 61.616 64,581 41,394 0 80,620 77,463 73,672 21,438 604,779 549,680 502,499 470,696 14.2561.61664,58141,394080,62077,46373,67221,438604,779549,680502,499470,696 15.25 41,225 35,700 16,606 5,715 2,328 3,291 8.226 0 728,079 647,420 577599 530,216 15.2541,22535,70016,6065,7152,3283,2918.2260728,079647,420577599530,216 16.25 50,409 43,638 41,603 39.141 33.438 23,528 16,107 10,761 759,727 753,409 698,842 638,088 16.2550,40943,63841,60339.14133.43823,52816,10710,761759,727753,409698,842638,088 17.25 4,700 15,818 21.766 15.171 11,350 15,711 17,965 19,110 321 22,375 100,159 203,603 17.254,70015,81821.76615.17111,35015,71117,96519,11032122,375100,159203,603 18.25             18.25 19.25             19.25 20.25             20.25 Sum 2,287,413 2,448,192 2,609,434 2,498,751 2,845,229 2,809,021 2,778,367 2,681,671 6,413,668 6,068.086 5,785.107 5,787,677 Sum2,287,4132,448,1922,609,4342,498,7512,845,2292,809,0212,778,3672,681,6716,413,6686,068.0865,785.1075,787,677  "},{"text":"Table 11 . Biomass per length class and month of Peruvian anchoveta (Engraulir ringens, northemlcentral stock, 4-140s) for 1959, in tonnes. Midlength Jan Feb Mar API May Jun Jul Aug S ~P Oct Nov Dec MidlengthJanFebMarAPIMayJunJulAugS ~POctNovDec Sum 5,621,667            Sum5,621,667  "},{"text":"Table 12 . Biomass per length class and month of Peruvian anchoveta (Engradis ringens, northemlcentral stock, 4-140s) for 1960, in tonnes. Midlenpth Jan Feb Mar Apr May Jun Jul Aug sep Oct Nov Dec MidlenpthJanFebMarAprMayJunJulAugsepOctNovDec 4.25             4.25 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 Sum             Sum  "},{"text":"Table 13 . Biomass per length class and month of Peruvian anchoveta (Engraulir ringens, northemlcentral stock. 4-140s) for 1961, in tonnes. Midlength Jan Feb Mar Apr May Jun Jul Aug S ~P Oct Nov Dec MidlengthJanFebMarAprMayJunJulAugS ~POctNovDec 4.25             4.25 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 Sum             Sum  "},{"text":"Table 14 . Biomass per length class and month of Peruvian anchoveta (Engradis ringens, northemlcentral stock, 4-140s) for 1962, in tonnes. Midlength Jan Feb Ma Aer May Jun Jul Aug S ~P Oct Nov Dec MidlengthJanFebMaAerMayJunJulAugS ~POctNovDec  "},{"text":"Table 16 . Biomass per length class and month of Peruvian anchoveta (Engradis ringens, northemlcentral stock, 4-140s) for 1964, in tonnes. Midlength Jan Feb Mar A P ~ May Jun Jul Aug SeP Oct Nov Dec MidlengthJanFebMarA P ~MayJunJulAugSePOctNovDec 4.25             4.25 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 Sum             Sum  "},{"text":"Table 17 . Biomass per length class and month of Pemvian anchoveta (Engraulis ringens. northern/cenual stock. 4-140s) for 1965, m tomes.  Midlength Jan Feb Mar A P ~ May Jun Jul Aug Sep 0.3 Nov Dec MidlengthJanFebMarA P ~MayJunJulAugSep0.3NovDec 4.25             4.25 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 Sum             Sum  "},{"text":"Table 18 . Biomass per length class and month of Pemvian anchoveta (Engraufk ringens, northern/central stock, 4-140s) for 1966, m tomes.  Midlength Jan Feb Mar APT May Jun Jul AW Sep Oct Nov Dec MidlengthJanFebMarAPTMayJunJulAWSepOctNovDec 4.25             4.25 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 1125             1125 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 Sum             Sum  "},{"text":"Table 19 . Biomass per length class and month of Peruvian anchoveta (Engraulk ringem, northemlcentral stock. 4-140s) for 1967, in tomes.  Midlength Jan Feb Mar Aer May Jun Jul Aug S ~P Oct Nov Dec MidlengthJanFebMarAerMayJunJulAugS ~POctNovDec 4.25             4.25 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 sum             sum  "},{"text":"Table 20 . Biomass per length class and month of Peruvian anchoveta (Engraulis ringens, northemlcentral stock, 4-140s) for 1968, in tonnes.    - - ---         ----- Midlength Ian Feb Mar A P ~ May Jun Jul Aug sep Oct N O ~ Dee MidlengthIanFebMarA P ~MayJunJulAugsepOctN O ~Dee 425             425 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11 .25             11 .25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 1825             1825 19.25             19.25 20.25             20.25 Sum -             Sum -  "},{"text":"Table 21 . Biomass per length class and month of Peruvian anchoveta (Engraulis ringens, northem/central stock, 4-140s) for 1969, in tomes.  Midlength Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec MidlengthJanFebMarAprMayJunJulAugSepOctNovDec 4.25 127,654            4.25127,654 5.25 195.411            5.25195.411 6.25 418,882            6.25418,882 7.25 791,795            7.25791,795 8.25 1,175,855            8.251,175,855 9.25 1,475,645            9.251,475,645 10.25 1,572,086            10.251,572,086 11.25 1,274,695            11.251,274,695 12.25 586,757            12.25586,757 13.25 274,192            13.25274,192 14.25 278,892            14.25278,892 15.25 465,961            15.25465,961 16.25 266,603            16.25266,603 17.25 60,483            17.2560,483 18.25 5,727            18.255,727 19.25             19.25 20.25             20.25 Sum 9,019,245            Sum9,019,245  "},{"text":"Table 22 . Birmass per length class and manth of Peruvian anchoveta (Engraulis ringens, northem/central stock. 4-140s) for 1970, in tonnes.  Midlength Jan Feb Mar A P ~ May lun Jul Aug sep Oct Nov Dec MidlengthJanFebMarA P ~MaylunJulAugsepOctNovDec 4.25             4.25 5.25             5.25 6.25             6.25 725             725 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 sum             sum  "},{"text":"Table 23 . Bignass per length class and month of Peruvian anchoveta (Engradis ringens, northedcentral stock. 4-140s) for 1971, in tonnes.  Millength Jan Feb Mar A P ~ May Jun Jul Aug sep Oct Nov Dec MillengthJanFebMarA P ~MayJunJulAugsepOctNovDec 4.25             4.25 5 1 5             5 1 5 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 Sum             Sum  "},{"text":"Table 24 . Biomass per length class and month of Pemvian anchoveta (Engradis ringens, northedcentral stock. 4-140s) for 1972, in tomes.  Millength Jan Feb Mar Aer Jun oct Nov Dec MillengthJanFebMarAerJunoctNovDec 4.25     87,068 45,011 35,597 31,857 4.2587,06845,01135,59731,857 5.25     110616 94,655 78,872 72.667 5.2511061694,65578,87272.667 6.25     171,739 166,673 149,980 145,257 6.25171,739166,673149,980145,257 7.25     231,444 246,081 244,961 260,399 7.25231,444246,081244,961260,399 8.25     248,537 295,568 337,718 403,751 8.25248,537295,568337,718403,751 9.25     220,359 334,318 370,985 526,023 9.25220,359334,318370,985526,023 10.25     163,299 349,366 418,900 545.808 10.25163,299349,366418,900545.808 11.25     174,827 288,561 371,140 575,524 11.25174,827288,561371,140575,524 12.25     151,400 190.649 261,995 446,800 12.25151,400190.649261,995446,800 13.25     92,424 101,407 146,066 286,522 13.2592,424101,407146,066286,522 14.25     48,793 64,627 81,035 136,282 14.2548,79364,62781,035136,282 15.25     35,312 33,366 43,646 72,025 15.2535,31233,36643,64672,025 16.25     43,544 21,396 21,787 29,318 16.2543,54421,39621,78729,318 17.25     19,039 23,944 18,038 15,848 17.2519,03923,94418,03815,848 18.25     9,509 17,493 14,453 16,592 18.259,50917,49314,45316,592 19.25     2,085 377 2,401 5,261 19.252,0853772,4015,261 20.25         20.25 Sum      2,300,883 2,611,186 3,584,466 Sum2,300,8832,611,1863,584,466  "},{"text":"Table 25 . Biomass per length class and month of Peruvian anchoveta (Engradis ringem, northemlcentral stock, 4-140s) for 1973, in tonnes.  Millength Jan Feb Mar Apt May Jun Jul Aug sep Oct Nov Dec MillengthJanFebMarAptMayJunJulAugsepOctNovDec 4.25 21,233 22,951 31,997 37,312 29,765 28,549 23,900 21,774 15,543 11,860 9,706 9,457 4.2521,23322,95131,99737,31229,76528,54923,90021,77415,54311,8609,7069,457 5.25 42,449 40,549 55,062 59,847 48,696 55,269 47,458 43,735 32,884 28,554 21,165 18,474 5.2542,44940,54955,06259,84748,69655,26947,45843,73532,88428,55421,16518,474 6.25 83,397 67,386 83,833 86,305 72,516 83,426 83,999 77,925 60,004 54,193 46.610 33,810 6.2583,39767,38683,83386,30572,51683,42683,99977,92560,00454,19346.61033,810 7.25 164,630 118,567 119,635 114,860 90,573 113,076 121,260 126,551 98,960 91,892 82,379 71,778 7.25164,630118,567119,635114,86090,573113,076121,260126,55198,96091,89282,37971,778 8.25 279,968 221,721 191,521 155,495 113,937 131,941 154,687 167,515 146,024 142,869 132,525 121,200 8.25279,968221,721191,521155,495113,937131,941154,687167,515146,024142,869132,525121,200 9.25 373,890 374,618 331,578 226,405 147,225 157,163 174,906 189,247 184,273 192,564 197,222 187,629 9.25373,890374,618331,578226,405147,225157,163174,906189,247184,273192,564197,222187,629 10.25 397,915 499.629 537,593 365,939 210.002 195,698 203,672 211,932 186,899 217,680 247,167 270,691 10.25397,915499.629537,593365,939210.002195,698203,672211,932186,899217,680247,167270,691 11.25 446,338 508,889 712,735 560,054 340,587 287,351 257,561 248,008 208,212 223,052 251.704 317,377 11.25446,338508,889712,735560,054340,587287,351257,561248,008208,212223,052251.704317,377 12.25 495,323 579,222 698,103 624,529 504,729 462,337 393,687 342,954 258,668- 247,945 259,928 294.385 12.25495,323579,222698,103624,529504,729462,337393,687342,954258,668-247,945259,928294.385 13.25 322,214 550,952 768,833 435,284 426,514 588,100 610,034 551,400 406,170 338,797 300,701 310,207 13.25322,214550,952768,833435,284426,514588,100610,034551,400406,170338,797300,701310,207 14.25 149,281 301,924 574,961 406,496 237,431 351,603 516,983 655,681 596,928 558,000 476,173 403,841 14.25149,281301,924574,961406,496237,431351,603516,983655,681596,928558,000476,173403,841 15.25 61,489 108,490 224,419 220,017 152,351 197,878 248,810 315,602 385,074 514,438 630.986 662,947 15.2561,489108,490224,419220,017152,351197,878248,810315,602385,074514,438630.986662,947 16.25 21,129 42,252 68,451 38,610 42,661 89,684 133,155 166,954 180,579 217,241 278,696 431,630 16.2521,12942,25268,45138,61042,66189,684133,155166,954180,579217,241278,696431,630 17.25 7,992 10,896 16,554 10,957 5,738 10,293 19,911 35,705 51,844 80,261 117,422 159,029 17.257,99210,89616,55410,9575,73810,29319,91135,70551,84480,261117,422159,029 18.25 7,320 7,915 7,974 6,429 3,823 4,342 4,698 4,992 4,192 4,843 9,586 22,607 18.257,3207,9157,9746,4293,8234,3424,6984,9924,1924,8439,58622,607 19.25 478 1,275 2549 4,015 1,651 2,114 2,438 2,691 248 668 1,154 1,826 19.254781,27525494,0151,6512,1142,4382,6912486681,1541,826 20.25             20.25 Sum 2,905,451 3,469,293 4,444,552 3,372,050 2,441,329 2,770,415 3,007,292 3,171,552 2,821,991 2,929,347 3,067,438 3,321,788 Sum2,905,4513,469,2934,444,5523,372,0502,441,3292,770,4153,007,2923,171,5522,821,9912,929,3473,067,4383,321,788  "},{"text":"Table 26 . Biomass uer length class and month of Perwian anchoveta (Enwadis rinnem. northernlcentral stock. 4-140s) for 1974, in tonnes.  Midlength Jan F& Mar A P ~ May Jun Jul Aug Sap Oct Nov Dec MidlengthJanF&MarA P ~MayJunJulAugSapOctNovDec 435             435 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11.25             11.25 12.25             12.25 13.25             13.25 14.25             14.25 1535             1535 16.25             16.25 17.25             17.25 1825             1825 19.25             19.25 20.25             20.25 Sum             Sum  "},{"text":"Table 27 . Biomass per length class and month of P m i a n auchoveta (Engradis ringer, northemlcentral stock, 4-140s) for 1975, in tomes.  Midlsngth Jan Fab Mar A P ~ May Jun lul A% sap Oct Nov Dec MidlsngthJanFabMarA P ~MayJunlulA%sapOctNovDec 4.25 3,119 2,894 4,297 8,486 20,533 32,586 40,828 47,299 156.015 169,410 149,757 123,271 4.253,1192,8944,2978,48620,53332,58640,82847,299156.015169,410149,757123,271 5.25 6,622 5,624 6,751 12,083 28,598 37,040 56,748 69,052 210.908 255,462 257,598 243,230 5.256,6225,6246,75112,08328,59837,04056,74869,052210.908255,462257,598243,230 6.25 13,894 10,165 8,560 13,625 37,045 48,502 58,925 85.988 256,495 304,783 350,795 571,094 6.2513,89410,1658,56013,62537,04548,50258,92585.988256,495304,783350,795571,094 7.25 25,404 19,115 14,222 13,545 32,408 55,162 68,060 80.125 270,736 331.041 379,830 483,540 7.2525,40419,11514,22213,54532,40855,16268,06080.125270,736331.041379,830483,540 8.25 41,638 36.044 24,021 20,605 34,663 44.005 69,666 82,573 241,610 310,936 377,287 483,139 8.2541,63836.04424,02120,60534,66344.00569,66682,573241,610310,936377,287483,139 9.25 72,077 56,049 45,473 32,809 45,336 44,329 56.872 76,176 203,877 262,852 318,779 448,470 9.2572,07756,04945,47332,80945,33644,32956.87276,176203,877262,852318,779448,470 10.25 167,978 93,751 69,553 59,547 69,765 57,226 50,429 62.242 165,510 217,090 254,213 346,422 10.25167,97893,75169,55359,54769,76557,22650,42962.242165,510217,090254,213346,422 11.25 369,169 211,369 110,551 91,801 120,588 88.107 73,314 63,708 146,983 164,861 195,365 257,820 11.25369,169211,369110,55191,801120,58888.10773,31463,708146,983164,861195,365257,820 12.25 587.017 466,108 237,599 141,405 169,167 145,934 119,137 93,567 191,087 159,269 148,413 182,383 12.25587.017466,108237,599141,405169,167145,934119,13793,567191,087159,269148,413182,383 13.25 610,835 690.811 496,520 320,506 227,168 187.256 175.109 157,983 294,892 233,108 170.930 140,006 13.25610,835690.811496,520320,506227,168187.256175.109157,983294,892233,108170.930140,006 14.25 457,410 643,984 651,832 564,228 374,655 243,989 213.073 198,161 365.909 339,473 262956 194,527 14.25457,410643,984651,832564,228374,655243,989213.073198,161365.909339,473262956194,527 15.25 247,240 397,752 514,943 591,295 418,713 248,614 266,700 252.998 424,428 370.844 311,764 282.112 15.25247,240397,752514,943591,295418,713248,614266,700252.998424,428370.844311,764282.112 16.25 214,087 220,844 271,687 354,078 278,280 126,365 158,196 196.214 393,245 408,294 356,446 306,849 16.25214,087220,844271,687354,078278,280126,365158,196196.214393,245408,294356,446306,849 17.25 189,631 171,031 179,281 170,838 125,582 70.036 77,232 85,154 138,870 178,970 200,733 247,432 17.25189,631171,031179,281170,838125,58270.03677,23285,154138,870178,970200,733247,432 18.25 102,551 99,461 113,250 106,198 59,584 15.784 20,468 27,748 59,676 67,500 65,556 71,884 18.25102,55199,461113,250106,19859,58415.78420,46827,74859,67667,50065,55671,884 19.25 24,667 29,957 35,538 35,644 15,163 2,846 2.839 3,075 2,604 3.591 7,038 10,128 19.2524,66729,95735,53835,64415,1632,8462.8393,0752,6043.5917,03810,128 20.25             20.25  "},{"text":"Table 28 . Biomass per length class and month of Peruvian andioveta (Engradk ringem. narthemlcmtral stock. 4-140s) for 1976, in tomes.  Milltmgth Jan Feb Mar Apr May lun Jul AM Ssp Oct Nov Dec MilltmgthJanFebMarAprMaylunJulAMSspOctNovDec  "},{"text":"Table 29 . Biomass per length class and month of Peruvian anchoveta (Engradis ringem, northedcentral stock, 4-140s) for 1977, in tonnes. Midleogth Jan Feb Mar API May Jun Jul AM Sep Oct Nov Dsc MidleogthJanFebMarAPIMayJunJulAMSepOctNovDsc 4.25 16,549 6,986 8,866 46,026 61,408 84,405 143,124 148,203 148,596 182,564 182,450 161,882 4.2516,5496,9868,86646,02661,40884,405143,124148,203148,596182,564182,450161,882 5.25 14,582 20,632 14,469 57,173 77,620 99,186 122,248 199,289 206,944 211,316 278,243 270,257 5.2514,58220,63214,46957,17377,62099,186122,248199,289206,944211,316278,243270,257 6.25 9,777 46,334 17,466 22,117 85,039 120,498 134,407 157,518 231,698 259,124 276,072 402,069 6.259,77746,33417,46622,11785,039120,498134,407157,518231,698259,124276,072402,069 7.25 16,847 15,784 57,920 23,621 27,739 104,892 141,662 147,299 176,166 247,787 307,734 344,997 7.2516,84715,78457,92023,62127,739104,892141,662147,299176,166247,787307,734344,997 8.25 37,473 20,490 43,303 52,096 31,796 42,487 115,550 136,637 156,326 178,925 263,411 357,379 8.2537,47320,49043,30352,09631,79642,487115,550136,637156,326178,925263,411357,379 9.25 81,844 39,586 22,976 69,397 52,671 34,885 34,663 96.239 112,514 153,125 178,409 289,929 9.2581,84439,58622,97669,39752,67134,88534,66396.239112,514153,125178,409289,929 10.25 150,019 81,212 39,911 25,498 78,890 60,138 32,900 34,088 62,980 100,687 137.784 180.214 10.25150,01981,21239,91125,49878,89060,13832,90034,08862,980100,687137.784180.214 11.25 208,730 162,661 80,212 40,321 22,989 63,698 75,284 35,167 28,682 27,609 77,881 125,957 11.25208,730162,66180,21240,32122,98963,69875,28435,16728,68227,60977,881125,957 12.25 199,876 228,369 166,093 83,769 27,199 15.112 29,577 63,489 51,848 24,600 23,818 59,295 12.25199,876228,369166,09383,76927,19915.11229,57763,48951,84824,60023,81859,295 13.25 149,719 205,461 225,686 169,451 50,458 12,293 10,596 12,174 24,153 45,868 34,487 20,191 13.25149,719205,461225,686169,45150,45812,29310,59612,17424,15345,86834,48720,191 14.25 95,013 116,099 172,501 203,185 89,761 18,284 11,999 8,162 6,962 7,896 18,838 34,036 14.2595,013116,099172,501203,18589,76118,28411,9998,1626,9627,89618,83834,036 15.25 74,363 42,153 78,802 128,783 67,455 19,777 16,083 11,798 7,600 5,044 4,084 5,614 15.2574,36342,15378,802128,78367,45519,77716,08311,7987,6005,0444,0845,614 16.25 34,465 8.215 22,089 41,504 20,835 12,517 12,088 10,214 7,807 5,193 2,668 2,380 16.2534,4658.21522,08941,50420,83512,51712,08810,2147,8075,1932,6682,380 17.25 8,411 1.985 4,125 6.818 4.917 2.954 3.956 3.945 3,260 2,346 805 845 17.258,4111.9854,1256.8184.9172.9543.9563.9453,2602,346805845  "},{"text":"Table 30  Midlength Jan Feb Mar A P ~ May Jun Jul Aug Sep Oct Nov Dsc MidlengthJanFebMarA P ~MayJunJulAugSepOctNovDsc 4.25             4.25 5.25             5.25 6.25             6.25 7.25             7.25 8.25             8.25 9.25             9.25 10.25             10.25 11 .25             11 .25 12.25             12.25 13.25             13.25 14.25             14.25 15.25             15.25 16.25             16.25 17.25             17.25 18.25             18.25 19.25             19.25 20.25             20.25 Sum             Sum  "},{"text":"Table 3 1 . Biomass per length class and month of Peruvian anchoveta (Engraulis ringem, northedcentral stock, 4-140s) for 1979, in tonnes. Midlength Jan Feb Mar A P ~ May Jun Jul Aug sap Oct Nov Dsc MidlengthJanFebMarA P ~MayJunJulAugsapOctNovDsc   28,182 30,540 28,742 41956 42,086 44,032 45,841 44,612 28.967 22,811 30,840 28,18230,54028,7424195642,08644,03245,84144,61228.96722,81130,840   39,471 52,165 50,179 63,014 65,132 60,371 62,298 64,260 54,091 41,722 40,049 39,47152,16550,17963,01465,13260,37162,29864,26054,09141,72240,049   29,090 65,269 71,712 82.106 93,263 85,759 77,155 77,019 68,453 72,985 60,221 29,09065,26971,71282.10693,26385,75977,15577,01968,45372,98560,221   59,597 49,321 85,361 103,394 100,584 110,681 98,847 91,407 74,593 84,875 103,465 59,59749,32185,361103,394100,584110,68198,84791,40774,59384,875103,465   114,641 57,050 84,180 127,053 123,660 113,198 114,065 105,687 83,226 85,113 115,043 114,64157,05084,180127,053123,660113,198114,065105,68783,22685,113115,043   197,712 106,354 54,409 121,868 141,734 129,657 117,207 107,602 95,869 89,652 106,930 197,712106,35454,409121,868141,734129,657117,207107,60295,86989,652106,930   257,319 188,907 99,807 56,601 122,360 136,724 130,145 112,695 81,540 95,306 106,155 257,319188,90799,80756,601122,360136,724130,145112,69581,54095,306106,155   261,019 252,445 176,781 75,261 58,294 94,840 118,646 115,951 88,578 81,685 104,953 261,019252,445176,78175,26158,29494,840118,646115,95188,57881,685104,953   315,091 252,818 233,306 117,788 78,036 56,489 50,839 82,994 82.491 78,120 87,684 315,091252,818233,306117,78878,03656,48950,83982,99482.49178,12087,684   407,357 290228 219,503 143,527 122,179 83,849 56,029 41,185 37,790 50,468 76.399 407,357290228219,503143,527122,17983,84956,02941,18537,79050,46876.399   424,516 381,651 222,372 116,185 124,516 110,884 86,213 60,666 34,134 15,988 35,898 424,516381,651222,372116,185124,516110,88486,21360,66634,13415,98835,898   251,127 312,844 239,516 46,231 65,225 85,765 81,851 73,156 50,891 19,131 9,760 251,127312,844239,51646,23165,22585,76581,85173,15650,89119,1319,760   92,875 150,302 126,060 33,203 40,806 32,210 35,374 40,861 39,789 19,080 3,373 92,875150,302126,06033,20340,80632,21035,37440,86139,78919,0803,373   14,309 37,812 35,567 7,970 9,774 13,503 19,227 18,274 12,801 6,959 1,066 14,30937,81235,5677,9709,77413,50319,22718,27412,8016,9591,066   2,504 2,804 2,291 367 1,246 2,038 3,022 3,412 2,950 4,143 82 2,5042,8042,2913671,2462,0383,0223,4122,9504,14382   6 408 120 2 0 2 2 9 113 159  640812020229113159          0    0 Sum 2,594,252        1,063,974 852,059 786,337 911,478 Sum2,594,2521,063,974852,059786,337911,478  "},{"text":"Table 32 . Biomass per length class and month of Peruvian anchoveta (Engradis ringem, northemlcentral stock, 4-140s) for 1980, in tonnes.  Midltngth Jan Feb Mar  A P ~ MnY Jun Jul Aug S ~P Oct Nov Dac MidltngthJanFebMarA P ~MnYJunJulAugS ~POctNovDac     - - -- ---    ------- 4.25 29,722 40,628 63,517  49,782 122,687 182,009 272.259 284,277    4.2529,72240,62863,51749,782122,687182,009272.259284,277 5.25 42,927 65,399 105,187  85,422 165,046 203,778 317.619 390.674    5.2542,92765,399105,18785,422165,046203,778317.619390.674 6.25 52,529 74.167 101.030  111,047 162,471 240,923 347,999 410,350    6.2552,52974.167101.030111,047162,471240,923347,999410,350 7.25 74,952 71,041 110,768  115,609 189,599 180,515 350,066 386,489    7.2574,95271,041110,768115,609189,599180,515350,066386,489 8.25 110,471 86,913 109,400  113,110 214,375 210.206 247,141 350,136    8.25110,47186,913109,400113,110214,375210.206247,141350,136 9.25 143,398 126,313 103,613  98,170 178,402 219,456 265,761 262,571    9.25143,398126,313103,61398,170178,402219,456265,761262,571 10.25 94,723 156,965 134.667  80,935 142,530 168,066 247,227 256,786    10.2594,723156,965134.66780,935142,530168,066247,227256,786 11.25 103,610 117,241 164,587  97,547 110,536 125,622 182,224 195,812    11.25103,610117,241164,58797,547110,536125,622182,224195,812 12.25 92,769 105,948 128,939  118,397 130,204 90,070 126,975 144,168    12.2592,769105,948128,939118,397130,20490,070126,975144,168 13.25 78,510 89,994 103,309  85,532 152,061 91,513 89,518 83,768    13.2578,51089,994103,30985,532152,06191,51389,51883,768 14.25 45,302 72,574 83,107  66,645 86,469 75,144 86,296 73,995    14.2545,30272,57483,10766,64586,46975,14486,29673,995 15.25 12,182 27,759 60,794  49,161 70.369 39.067 41,750 52,120    15.2512,18227,75960,79449,16170.36939.06741,75052,120 16.25 4,481 7,097 9,924  23.153 46,120 19,646 17,695 20,701    16.254,4817,0979,92423.15346,12019,64617,69520,701 17.25 1,315 1,495 2,158  3.073 6.035 4,140 4,128 4,296    17.251,3151,4952,1583.0736.0354,1404,1284,296 18.25 123 6 289  394 819 531 470 487    18.251236289394819531470487 19.25      31 46 31 55    19.2531463155 20.25             20.25 Sum 916,992 1,091,711 1,335,348 1,165,950 1,914287 2,003,593 2,797,802 3.132562    Sum916,9921,091,7111,335,3481,165,9501,9142872,003,5932,797,8023.132562  "},{"text":"Table 33 . Bicanass perlength class ahd month of Peruvian ancboveta (Engradis ringens, northem/central stcck, 4-140s) for 1981, in tonnes.  Midlsngth Jan Feb Mar Apt May Jun Jul Aug sap MidlsngthJanFebMarAptMayJunJulAugsap  "},{"text":"Table 2 . Data on the size dependence of fecundity in northern anchovy (E. r n ~r d a r ) . ~ Spawning season 1 st 2nd 3 rd 4th plus Spawning season1 st2nd3 rd4th plus Range of weights (g) 11 .O-15.4 15.5-18.3 18.3-22.7 20.9-26.6 Range of weights (g)11 .O-15.415.5-18.318.3-22.720.9-26.6 Mean weight (g) 12.38 16.71 19.76 23.23 Mean weight (g)12.3816.7119.7623.23 Spawnings per season 5.3 11.9 19.2 23.5 Spawnings per season5.311.919.223.5 Eggs/g female per seasonb 2,803 6,550 11,434 13,386 Eggs/g female per seasonb2,8036,55011,43413,386  "},{"text":"Table 4 . Selected information on the seasonality of spawning in Peruvian anchoveta (  "},{"text":"Engmulis ringens).    % mature females in catch  % mature as Absolute no. % mature females in catch% mature asAbsolute no.   samples  fraction of of spawning samplesfraction ofof spawning h40ntha callaob chirnboteb Mean September value per monthC h40nthacallaobchirnbotebMeanSeptember valueper monthC Jan      Jan Feb      Feb Mar      Mar A P ~      A P ~ May      May Jun      Jun Jul      Jul A w      A w S ~P      S ~P Oct      Oct Nov      Nov Dec      Dec  "},{"text":"Table 5 . Estimated biomass of mature female and male anchoveta (Engraulis ringens) off Peru (4-14\"s). 1953 to 1981 (in tonne^). Date Jan Feb Mar A P ~ May Jun Jul Aug SeP Oct Nov Dec DateJanFebMarA P ~MayJunJulAugSePOctNovDec  "},{"text":"Table 7 . Estimated monthly egg output by the anchoveta stock off Peru (4-14\"N), 1953 to 1982 in billions ( l o 9 )  Year Jan Feb Mar May Jun Jul Dec YearJanFebMarMayJunJulDec  "},{"text":"Table 1 . Actual egg denties in the density class \"> 4,000\" in 7 maps for which such information is available.    Latitude  Latitude  Date south Eggslm2 DatesouthEggslm2 1 25-02-64 to 18-03-64 10-12O 4,244 125-02-64 to 18-03-6410-12O4,244 5 11-11-64 to 22-12-64 6-8' 4,581 511-11-64 to 22-12-646-8'4,581   8-10' 4 3 94 8-10'4 3 94 6 03-03-65 to 22-03-65 6-8' 5,382 603-03-65 to 22-03-656-8'5,382   10-12O 6,060 10-12O6,060 9 22-10-65 to 10-12-65 10-12O 4,151 922-10-65 to 10-12-6510-12O4,151 13 17-11-66 to 13-12-66 6-8' 4,046 1317-11-66 to 13-12-666-8'4,046   8-10' 5,267 8-10'5,267 16 08-11-67 to 21-12-67 6-8' 5,864 1608-11-67 to 21-12-676-8'5,864   8-10' 5,864 8-10'5,864   10-12O 5,504 10-12O5,504 28 16-08-71 to 02-09-71 8-10' 4,604 2816-08-71 to 02-09-718-10'4,604  Mean  5,013 Mean5,013  "},{"text":"Table 2 . Mean % of total eggs off Peru (4-14OS) in each 2O stretch of coastline?  O~at, south Mean % O~at, southMean % 4-6 3.3 4-63.3 6-8 22.2 6-822.2 8-10 31.9 8-1031.9 10-12 21.9 10-1221.9 12-14 20.7 12-1420.7  "},{"text":"a~ased on 49 surveys which co- vered the area comprised between 4 and 14OS, i.e., maps with boxed num- bers in Table 3.   "},{"text":"Table 3 . Key statistics on 90 egg surveys considered in this contribution (see also Table4).   Map eggs Development Adjusted egg  Map eggsDevelopmentAdjusted egg Map no.a b Time of survey standing stockC (billions) timed (days) standing stocke (billions/day) Parent stock (t)e Map no.ab Time of surveystanding stockC (billions)timed (days)standing stocke (billions/day)Parent stock (t)e  "},{"text":"Table 4 . Egg standing stocks for maps that are combined with another map in Table3.   Egg standing Egg standing  Time of survey stock Time of surveystock Map # (daylmonth-day/month) (X 1015) Map #(daylmonth-day/month)(X 1015)  "},{"text":"Table 3 for egg standing stock estimates based on maps 5 to 8). Map no. 9 Map no. 12      1-500 1-500     M I -1000 M I -1000     1001-4000 1001-4000     ,4000 ,4000 Map no. 13    Map no. 14 Map no. 13Map no. 14 . . .     . . . Map no. Map no. 15    Map no. 16 Map no. Map no. 15Map no. 16 84.W 82.W 80.W 7a.W 76.W 84.W82.W80.W7a.W76.W Fig. 6.     Fig. 6.  "},{"text":"Distribution of anchoveta eggs off Peru during four surveys conducted in 1966 and 1967 (see Table 3 for egg standing stock estimates based on maps 13 to 16).   "},{"text":"Table 3 for egg standing stock estimates based on maps 25 to 28). Map no. 29 Map no. 31  14.5      14.5 e8W 84.W 82.W 80.W 78-44 78-44 e8W84.W82.W80.W78-4478-44  "},{"text":"Map no. 30 Crulu EunU XXII 18-20 July 1972 Map no. 32 Fig. 10. Distribution of anchoveta eggs off Peru during four surveys conducted in 1971 and 1972 (see Tables 3 and 4 for egg standing stock estimates based on maps 29 to 32).       1-54. 1-54.      541 -woo 541 -woo      1001-4WO 1001-4WO      >40W >40W      Map no. 34 Map no. Map no. 34 Map no.      14.5 14.5 B B W 8 4 W 8ZW 80.W 78.W 74.W B B W8 4 W8ZW80.W78.W74.W  Map no. 35 Map no. 39    Map no. 40 Map no. 35 Map no. 39Map no. 40  "},{"text":"Table 3 for egg standing stock estimates based on maps 45 to 48). Map no. 49 Map no. 50 Map no. 52  14.5    149    14.5149 86.W W W 82.W 80.W 14.W 86.W 84.W 78.W 86.WW W82.W80.W14.W86.W84.W78.W  "},{"text":"Table 3 for egg standing stock estimates based on maps 53 to 56).  Map no. 58 . . . . Map no. 58. . . .  Crulu 7704 -0s I8 A m Crulu 7704 -0s I8 A m  "},{"text":"-3 May 1977 Fig. 17. Distribution of anchoveta eggs off Peru during four surveysconducted in 1976 and 1977 (see Table 3 for egg standing stock estimates based on maps 57 to 60).   "},{"text":"Table 3 for egg standing stock estimates based on maps 77 to 80).  ...-~.. ...-~..  "},{"text":"82 Map no. 84 Fig. 23. Distribution of anchoveta eggs off Peru during four surveys conducted in 1982 and 1984 (see Tables 3 and 4 for egg standing stock estimates based on maps 81 to 84). Map no. 85 Map no. 87 Map no. 88 Fig. 24. Distribution of anchoveta eggs off Peru during four surveys conducted in 1984 and 1985 (see   "},{"text":"Table 3 for egg standing stock estimates based on maps 85 to 88). Map no. 90 Fig. 25. Distribution 89 and 90). 82.W anchoveta off 78.W Peru during 76.W 74.W four surveys conducted  4.S      4.S 8.S      8.S I0.S      I0.S 1Z.S      1Z.S 14.5      14.5 86-W 84.W 82.W BG' W 7O.W 76.W 86-W84.W82.WBG' W7O.W76.W  "},{"text":"7WW 1985 and 1986 (see Table 3 for egg standing stock estimates based on maps   "},{"text":"Table 1 . Summary of information on the guano islands (I) and capes (P) of the Peruvian coast considered in this contribution.   Latitude Longitude  LatitudeLongitude No. Naine (south) west) Area (km2) No.Naine(south)west)Area (km2)  "},{"text":"Table 2 . Mean densities of guano birds in colonies.  Number of birds per m2 Number of birds per m2  Cormorant Booby Pelican CormorantBoobyPelican Reproducing adults 7 5 3 Reproducing adults753 (with nests)    (with nests) Nonreproducing adults    Nonreproducing adults  "},{"text":"Table 1 ) in September 1974, with relative areas estimated by planimetry. The sums of the relative area are 0.019,0.088 and 0.050 for cormorants, boobies and pelicans, respectively (see text).   "},{"text":"Table 3 . Population (in millions) of adult cormorants off Peru (6 to14OS, 195314OS,   to 1982)). Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec YearJanFebMarAprMayJunJulAugSepOctNovDec  "},{"text":"Table 5 . Population (in millions) of adult pelicans off Peru (6 to 1 4 O ~, 195 3 to 1982). Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec YearJanFebMarAprMayJunJulAugSepOctNovDec  "},{"text":"Table 6 . Changes in guano bud populations between 6 and 14OS off Peru following El Niiio 1982182, (adapted from data inTovar and  Cabrera 1984).   Bird population (millions)  Bird population (millions) Month Cormorants Boobies Pelicans MonthCormorantsBoobiesPelicans Ju11982 4.68 2.78 0.16 Ju119824.682.780.16 Dec 1982 1.34 1.19 0.13 Dec 19821.341.190.13 Mar 1983 0.91 0.28 0.16 Mar 19830.910.280.16 May 1983 0.23 0.15 0.23 May 19830.230.150.23 Jun 1983 0.1 1 0.05 0.08 Jun 19830.1 10.050.08 Jul1983 0.19 0.08 0.08 Jul19830.190.080.08  "},{"text":"Table 1 . Feeding behavior, body weight and daily ration for the main Peruvian guano bud species.   Maximum  Adult Daily  MaximumAdultDaily Species Feeding methoda depth of Foraging weight ration % anchovy SpeciesFeeding methodadepth ofForagingweightration% anchovy   foraging (m) period (kg)d (%body weight) in diet foraging (m)period(kg)d(%body weight)in diet Pelecanus thagus piracy, surface 2.0a day and nightb9~ 5.9-7.8 11.2~ > 80\" Pelecanus thaguspiracy, surface2.0aday and nightb9~ 5.9-7.811.2~> 80\"  seizing, surface   mean 6.68 10.lf  seizing, surfacemean 6.6810.lf Sula variegata plunging deep plunging 5 .Oa dawn-sunsetb 1.2-1.4 27 .4e > 8ob Sula variegataplunging deep plunging5 .Oadawn-sunsetb1.2-1.427 .4e> 8ob   15.0\"  mean 1.28 1 6 . 3 ~  15.0\"mean 1.281 6 . 3 ~ Phalacrocomx bougainvillii pursuit diving 1 2 . 0 ~ dawn-sunsetb 1.7-2.2 23.3e 9gbg Phalacrocomx bougainvillii pursuit diving1 2 . 0 ~dawn-sunsetb1.7-2.223.3e9gbg   1 5 . 0 ~  mean 1.94 14.7~  1 5 . 0 ~mean 1.9414.7~ a In Duffy (1980) after Ashmole (1971)      a In Duffy (1980) after Ashmole (1971) Jordan (1967)       Jordan (1967) Vogt (1942)       Vogt (1942) H. Tovar (IMARPE, pen. comm.)      H. Tovar (IMARPE, pen. comm.) from equation (1)       from equation (1) from equation (2)       from equation (2) g Avila (1954)       g Avila (1954)  "},{"text":"Table 2 . Estimates of fish consumption in three genera of fii%athg birds (  "},{"text":"Pkhcmconx, Suh and Peleeamrs).    Adult Daily consumption AdultDaily consumption Species Area weight (Irg) SpeciesAreaweight (Irg)  "},{"text":"Table 3 . Preliminary estimates of anchoveta biomass for the area 4-14OS (in t x lo6), as used to compute the availability of anchoveta to the guano birds.a Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec YearJanFebMarAprMayJunJulAugSepOctNovDec  "},{"text":"Table 4 . Relationships between sea surface temperature, depth of main anchoveta biomass (DMB), vertical overlap index between anchoveta and birds (V) and anchoveta vulnerability to the birds (vulha   Depth of Vertical Vulnerability Midrange of Depth ofVerticalVulnerabilityMidrange of  Temperature anchoveta overlap bird of anchoveta vulnerability Temperatureanchovetaoverlap birdof anchovetavulnerability  (OC) biomass (m) vs anchoveta (vul) estimates (OC)biomass (m)vs anchoveta(vul)estimates    (V)   (V) Monthly mean 22.0 15 .O 0.99 0.98  Monthly mean22.015 .O0.990.98 % range 18.7 25.3 15 .O 24.8 1 .OO 0.01 0.00 0.50 % range18.7 25.315 .O 24.81 .OO 0.010.000.50 mean 21 . O 15 .O 1 .OO 1 .OO  mean21 . O15 .O1 .OO1 .OO % range 17.9 24.2 15 .O 18.7 1 . OO 0.72 0.52 0.76 % range17.9 24.215 .O 18.71 . OO 0.720.520.76 a Note effect of using range instead of mean.    a Note effect of using range instead of mean.  "},{"text":"Table 6 . Monthly anchoveta consumption by cormorants (Phalacrocorax bougainvillii) off Peru (4-14OS), 1953 to 1982 (in t x 103). Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec YearJanFebMarAprMayJunJulAugSepOctNovDec  "},{"text":"Table 7 . Monthly anchoveta consumption by boobies (Sub variegata) off Peru (4-14'~), 1953 to 1982 (in t x lo3). Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec YearJanFebMarAprMayJunJulAugSepOctNovDec  "},{"text":"Table 2 . Differences in the impact of the 1982-1983 El N&O on fur seal and sea lion populations.   Population estimate Population estimate Census data Fur seal Sea lion Census dataFur sealSea lion April-May 1979 20,455 25,05 5 April-May 197920,45525,05 5 March 1984 difference 15,369 -25% 33,861 + 35% March 1984 difference15,369 -25%33,861 + 35%  "},{"text":"Table 4 . Estimates of vertical overlap between anchoveta and seals (V) and anchoveta vulnerability to seals (Vul) as a function of the depth of main anchoveta biomass (DMB) (see text for details). DMB (m) V Vul DMB (m)VVul  "},{"text":"Table 5 . Example of the estimated relationships between mean sea surface temperature in month (i), depth of main anchoveta biomass (DMBi) and vulnerability of anchoveta to seals (Vuli).  Mean temp. DMBi  Mean temp.DMBi Month (i) (OC) (m) Vuli Month (i)(OC)(m)Vuli 1 24.9 60 0.6 124.9600.6 2 25.2 6 3 0.4 225.26 30.4 3 25.5 65 0.3 325.5650.3 4 25.5 65 0.3 425.5650.3 5 26.3 67 0.1 526.3670.1 6 26.1 66 0.2 626.1660.2 7 21.1 15 1 .O 721.1151 .O 8 18.1 15 1 .O 818.1151 .O  "},{"text":"Table 6 . Estimation of mean fraction of Peruvian pinniped occurring between 4 and 14Os.  Fraction of total  Fraction of total  population occurring  population occurring  within 4-14's  within 4-14's Year Fur seal Sea lion Source YearFur sealSea lionSource 1969-1979 - 0.52 Majluf and Trillmich (1981) 1969-1979-0.52Majluf and Trillmich (1981) 1971-1979 0.20 - Majluf and Trillmich (1981) 1971-19790.20-Majluf and Trillmich (1981) 1978 0.15 0.60 Tovar (1979); Tovar and Fuentes (1984) 19780.150.60Tovar (1979); Tovar and Fuentes (1984) 1982 0.29 0.64 Tovar (1979); Tovar and Fuentes (1984) 19820.290.64Tovar (1979); Tovar and Fuentes (1984) Means 0.21 0.59 Used in this study Means0.210.59Used in this study  "},{"text":"Table 1 . Annual catch, fishing effort, fishing mortality and estimated biomass of large bonitos offPeru, 1951Peru,   to 1983.   .      Annual catch Standardized Standardized Estimated 0.9 catch/ Biomass of Annual catchStandardizedStandardizedEstimated0.9 catch/Biomass of  whole of Peru catch per effort fishing fishing fishing large bonitos whole of Perucatch per effortfishingfishingfishinglarge bonitos Year (x lo3 t)a of chimbote fleetb effortC mortality mortalitye (4-14'~, x lo3 tlf Year(x lo3 t)aof chimbote fleetbeffortCmortalitymortalitye(4-14'~, x lo3 tlf  "},{"text":"Table 3 . . Note bimodal distribution in age groups I and 11, possibly due to erroneous age reading; L = fork length. % length composition data on bonito (Soldo chiliensis chilien@ caught offP a , 1951-1976.a     L Age L Age L Age L Age L Age LAgeLAgeLAgeLAgeLAge (cm) I (cm) I1 (cm) I11 (cm) IV (cm) V (cm)I(cm)I1(cm)I11(cm)IV(cm)V  "},{"text":"Table 4 . Available growth parameter estimates in Sarda ~pecies.~ Species FL_ (cm)  K (Y @' Mean K (Y@'Mean Sarda sarda   Sarda sarda Sarda sarda   Sarda sarda Sarda sarda   Sarda sarda Sarda sarda   Sarda sarda Sarda sarda   Sarda sarda Sarda chiliensis   Sarda chiliensis Sarda chiliensis   Sarda chiliensis Sarda chiliensis   Sarda chiliensis a   a  "},{"text":"Growth parameters for S. sarda from Pauly (1978); growth pwa- meters for S. chiliensis from Yoshida (1980).   "},{"text":"Table 5 . Le yth-w eight relationships in Peruvian bonito (Sarda chiliensis chi- lien&).  Case L-W relationships Range  Computed Source and/or CaseL-W relationshipsRangeComputed Source and/or No. L' 1, \"b\" (in cm) n weight at remarks No.L'1,\"b\"(in cm)nweight atremarks       60 cm  60 cm 1 0.006311 3.19 20-71 513 2,968 Vildoso (l962), 10.0063113.1920-715132,968Vildoso (l962), 2 0.006491 3.19 22-71 565 3,052 Vildoso (1962), 20.0064913.1922-715653,052Vildoso (1962), 3 0.01169 3.03 30-71 930 2,855 Canal (1974) 30.011693.0330-719302,855Canal (1974) 4 0.02011 2.89 30-58 204 2,769 Mayo (1976) 40.020112.8930-582042,769Mayo (1976) 5 0.01348 3 (60) - 2,911a used in this study 50.013483(60)-2,911aused in this study a ~e a       a ~e a  "},{"text":"n of cases No. 1-4, as used to estimate,based on b = 3, a value of a = 0.01348, corresponding to c.f. = 1.348.   "},{"text":"Table 6 . Estimated monthly biomass of large bonitos off Peru (4 to 14'~), 1953 to 1983 (in t x lo3) derived by linear interpolation of the smoothed annual biomass estimate in Table 1, last column). Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec YearJanFebMarAprMayJunJulAugSepOctNovDec  "},{"text":"Table 7 . Summary of information on m j o r diet components of Peruvian bonito (  "},{"text":"Sarda chiliensis chiliensis).  No. of  % i n   No. of% i n bonitos  total Details on Source and/or bonitostotalDetails onSource and/or examined Major food items food stomach contents remarks examinedMajor food itemsfoodstomach contentsremarks  a) Engraulis ringens   del Solar (1942) a) Engraulis ringensdel Solar (1942)  b) Crustaceans (Munida    b) Crustaceans (Munida  cokeri)    cokeri)  a) Engraulis ringens  anchoveta occurring in Ancieta (1945, 1964), based on a) Engraulis ringensanchoveta occurring inAncieta (1945, 1964), based on    60.6% of stomachs; data wllected from November 60.6% of stomachs;data wllected from November    remaining stomachs 1944 to January 1945 remaining stomachs1944 to January 1945    were empty  were empty    anchoveta occurring in De Buen (1958); data collected anchoveta occurring inDe Buen (1958); data collected    all stomachs examined off northern Chile all stomachs examinedoff northern Chile  a) Engraulis ringens  usually 5-6 anchoveta Some other pelagic fishes, not- a) Engraulis ringensusually 5-6 anchovetaSome other pelagic fishes, not-  b) Crustaceans (mainly  (60-80 g, max. of ably Sardinops mgax, Austrame- b) Crustaceans (mainly(60-80 g, max. ofably Sardinops mgax, Austrame-  Munida cokeni  330 g) nida re& and Prinotus q u i e ~ Munida cokeni330 g)nida re& and Prinotus q u i e ~  C) Unidentified items   cens also reported from bonito C) Unidentified itemscens also reported from bonito     stomachs by earlier authors (VB stomachs by earlier authors (VB     doso 1955)~ doso 1955)~  a) Engraulis ringens  size of anchoveta  a) Engraulis ringenssize of anchoveta  (plus some other fishes)  12-14 cm, max. of 28  (plus some other fishes)12-14 cm, max. of 28    cm, max, weight of  cm, max, weight of    stomach content is 404 g  stomach content is 404 g  a) Engraulis ringens  see Fig. 7 for size Mayo (1976)' based on data col- a) Engraulis ringenssee Fig. 7 for sizeMayo (1976)' based on data col-    distribution of lected from January to August distribution oflected from January to August    anchoveta 1976 anchoveta1976  b) Miscellaneous fishes and    b) Miscellaneous fishes and  invertebrates    invertebrates  "},{"text":"Table 8 . Relationship between body weight and mean stomach content weight in Sarda chiliensis chiliensis. a  Body weightb Mean stomach  Stomach content Body weightbMean stomachStomach content No (in g) content (ing) n (as % of body weight) No(in g)content (ing)n(as % of body weight)  "},{"text":"Table 10 . Estimated values of Q/B (daily ration in % of body weight) for large bonito off Peru (4 to 14's). See text for variables and constants used.- Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec YearJanFebMarAprMayJunJulAugSepOctNovDec  "},{"text":"Table 11 . Estimates of daily ration in four scombrid species. Daily Daily  "},{"text":"Table 12 . Values of the proportion of anchoveta in the diet of large bonitos, used for estimating their consumption of E. ringens for the years 1953 to 1982.  Assumed proportion of  Assumed proportion of  anchoveta in the diet  anchoveta in the diet Period of large bonito Rationale Periodof large bonitoRationale January 1953 to  During this period of high ancho- January 1953 toDuring this period of high ancho- December 1971  veta abundance about 0.75 of the December 1971veta abundance about 0.75 of the   food consumed by large bonitos food consumed by large bonitos   consisted oft?: ringens (see Table 7) consisted oft?: ringens (see Table 7) January 1972 to Proportion declining 1972 and 1973 are the years when January 1972 toProportion declining1972 and 1973 are the years when December 1973 linearly from 0.75 to 0.25 the anchoveta stock collapsxl (see December 1973linearly from 0.75 to 0.25the anchoveta stock collapsxl (see   other contributions in this vol.) other contributions in this vol.) January 1973 to 0.25 The only available estimate for the January 1973 to0.25The only available estimate for the December 1982  period of low anchoveta abundance December 1982period of low anchoveta abundance   suggests about 0.25 of the food of suggests about 0.25 of the food of   large bonito consists of E. ringens large bonito consists of E. ringens   (see Table 7) (see Table 7)  "},{"text":"Table 13 . Estimated anchoveta consumption in t x lo3 by large bonitosoff Peru (4 to 14OS), 1953 to 1983 (based on Tables 6, 10 and 12 and on constants presented in the text).  Year Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec Annual sum YearJanFebMarAprMayJunJulAugSeptOctNovDecAnnual sum  "},{"text":"The Importance of Mackerel and Horse Mackerel Predation for the Peruvian Anchoveta Stock (A Population and Feeding Model)* PETER MUCK C/O Instituto &l Mar del Peru P.O. Box 22, Callao, Peru GUADALUPE SANCHEZ Instituto del Mar del Peru P.O. Box 22, Callao, Peru  in these three species markedly different from more in these three species markedly different from more recent ones (1977-l986)? recent ones (1977-l986)? ii) If so, what are the possible reasons for the change? ii) If so, what are the possible reasons for the change?  "},{"text":"Table 1 . Percent anchoveta in the diet ofmackerel, 1976-1982.      Number of % mackerel stomachs % anchoveta Relative Number of% mackerel stomachs% anchovetaRelative Year mackerel sampled with anchoveta remains in mackerel stomachsb anchoveta biomassC Yearmackerel sampledwith anchoveta remainsin mackerel stomachsbanchoveta biomassC 1976 1977d 105 - 7 ga - 16.4 - 32.4 - 1976 1977d105 -7 ga -16.4 -32.4 - 1978 340 44 10.1 17.1 19783404410.117.1 1979 501 40 7.7 11.9 1979501407.711.9 1980 230 26 8.9 10.0 1980230268.910.0 1981 388 13 4.2 8.6 1981388134.28.6 1982 230 2 0.3 7.1 198223020.37.1 %his value would be 100% if anchoveta eggs were counted.   %his value would be 100% if anchoveta eggs were counted. b% of stomach content weight.    b% of stomach content weight. '     '  "},{"text":"Table 2 . Summary of data on feeding and diet composition of mackerel/horse mackerel sampled in 1979 along the Peruvian mast. Note that % diet composition refers to weight and that the % wntribution of \"other\" items can be obtained by subtraction. Location          Location Parameter Jan Feb Mar Apr May Jun Jul Oct Nov ParameterJanFebMarAprMayJunJulOctNov Paita (5'S)          Paita (5'S) Sampled fish (n)       75 / 3  Sampled fish (n)75 / 3 Mean individual          Mean individual weight (g)       412 1494  weight (g)412 1494 Mean stomach          Mean stomach content (g)       15.01 1.2  content (g)15.01 1.2 %diet wmposition          %diet wmposition Anchoveta       0 1 0  Anchoveta0 1 0 Euphausiacea       0.31 5.3  Euphausiacea0.31 5.3 Copepods       66.21 94  Copepods66.21 94 Chimbote (gOS)          Chimbote (gOS) Sampled fish (n)       57 1 -  Sampled fish (n)57 1 - Mean individual          Mean individual weight       461 1 -  weight461 1 - Mean stomach content (g)       2.51 -  Mean stomach content (g)2.51 - %diet wmposition          %diet wmposition Anchoveta       46.21 -  Anchoveta46.21 - Euphausiacea       0 1 0  Euphausiacea0 1 0 Copepods       39 I -  Copepods39 I - Cauao (12%)          Cauao (12%) Sampled fish (n)       7 I110  Sampled fish (n)7 I110 Mean individual          Mean individual weight (g)       560 1519  weight (g)560 1519 Mean stomach weight (g)       17 1 12.5  Mean stomach weight (g)17 1 12.5 %diet composition Anchoveta       92 1 76.7  %diet composition Anchoveta92 1 76.7 Euphausiacea       0 1 0  Euphausiacea0 1 0 Copepods       0 1 0  Copepods0 1 0 Pisco (14's)          Pisco (14's) Sampled fish (n)       -1 8  Sampled fish (n)-1 8 Mean individual weight (g)       -I789  Mean individual weight (g)-I789 Mean stomach weight (g)       -1 23.6  Mean stomach weight (g)-1 23.6 % diet composition          % diet composition Anchoveta Euphausiacea       -1 97.4 -1 2.6  Anchoveta Euphausiacea-1 97.4 -1 2.6 Copepods       -I 0  Copepods-I 0 110 (17\"s)          110 (17\"s) Sampled fish (n)       -I 10  Sampled fish (n)-I 10 Mean individual          Mean individual w e w t 6)       -1459  w e w t 6)-1459 Mean stomach          Mean stomach weight (g)       -1 13.2  weight (g)-1 13.2 %diet composition Anchoveta       -1 100  %diet composition Anchoveta-1 100 Euphausiacea       -1 0  Euphausiacea-1 0 Copepods       -I 0  Copepods-I0  "},{"text":"Table 8 . Relationship between estimated distance offshore of mackerel concentrations and SST within 40 miles off the coast.a Distance   Distance offshore (nm) SST (OC) Reference period offshore (nm)SST (OC)Reference period   July-August, 30 year mean July-August, 30 year mean   January -February, 30 year mean January -February, 30 year mean   November-December (1972 and 1976) November-December (1972 and 1976)   March-April 1983 March-April 1983 a ~a s e d on unpublished IMARPE survey data.  a ~a s e d on unpublished IMARPE survey data.  "},{"text":"Table 10 . Constants used in model for estimation of food consumption by mackerel and horse mackerel off Peru (equations 9-14). Parameter (units) Mackerel Horse mackerel Parameter (units)MackerelHorse mackerel Asymptotic length, L_ (cm)   Asymptotic length, L_ (cm) von Bertalanffy constant K (y-l)   von Bertalanffy constant K (y-l) Theoretical age at zero length, to (y)   Theoretical age at zero length, to (y) Maximum observed age (y)   Maximum observed age (y) Maximum observed length (cm)   Maximum observed length (cm) a, factor in length-weight rel.   a, factor in length-weight rel.  "},{"text":"Table 11 . Steps in estimating the food consumption of a stock of mackerel with biomass = 2 x 10 t.  Daily food consumption Daily food consumption  "},{"text":"Table 12 . Model estimates of anchoveta consumption by mackerel in relation to anchoveta biomass.    Weight of Daily  Relative Weight ofDailyRelative Mackerel anchoveta in anchoveta % anchoveta anchoveta Mackerelanchoveta inanchoveta% anchovetaanchoveta weight mackerel stomachs consumption in mackerel biomass weightmackerel stomachsconsumptionin mackerelbiomass Year a ) (9) Ma diet (%lb Yeara )(9)Madiet(%lb a ~a ( i ) in text.      a ~a ( i ) in text. b ~r ( i ) in text.      b ~r ( i ) in text.  "},{"text":"Table 13 . Estimated monthly anchoveta consumption of adult mackerel (Scornberjaponicus) off Peru (444% < 40 nautical miles) in t x lo6, 1953 to 1982. Month Month  "},{"text":"Table 15 . Minimum estimates of mackerel and horse mackerel capable of feeding on anchoveta, and their esti- mated anchoveta consumption, as used to pmvkle a loww limit to estimatesof anchoveta withdrawals by these mated anchoveta consumption, as used to pmvkle a loww limit to estimatesof anchoveta withdrawals by these two f i i w , from 1974 to 1982.   two f i i w , from 1974 to 1982.   Mean annual Least annual Mean annualLeast annual  Least predator bmmass anchoveta a p b i l i t y anchoveta consu ption Least predator bmmassanchoveta a p b i l i t yanchoveta consu ption yea^ (t lo31a index (t x lo6 x y3) yea^(t lo31aindex(t x lo6 x y3)  "},{"text":"Table 1 . Monthly occurrence of \"Lasker events*' near Trujillo, 1953-1985.~  Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec YearJanFebMarAprMayJunJulAugSepOctNovDec  "},{"text":"aBased on data in Mendo et al. (1987); a \"Lasker event\" is a period of calm (wind speed below 5 m/s) lasting 4 days; periods of 5 days are viewed as two partli overlapping 4-day events, etc. (see Peterman and Bradford 1987, note 16).   "},{"text":"Table 1 . Catches (in t) of the Peruvian purse seinefisherya, 1950-1983.     Horse Horse  "},{"text":"Table 2 . Basic data on the Peruvian purse seinesfleet, 1953-1980.~         Estimated of Estimated of  Vessels Average Total pelagic catch per vessel VesselsAverageTotal pelagiccatch per vessel Year (no. units) capacityb catch (tlC (tlyear) Year(no. units)capacitybcatch (tlC(tlyear)  52 na. 46,700  52na.46,700  137 n.a. 56,300  137n.a.56,300  192 n.a. 77,700  192n.a.77,700  238 n.a. 142,900  238n.a.142,900  296 n.a. 344,800  296n.a.344,800  354 n.a. 75,190  354n.a.75,190  414 n.a. 1,963,400  414n.a.1,963,400  667 n.a. 3,333,100  667n.a.3,333,100  756 n.a. 5,025,800  756n.a.5,025,800  1,069 n.a. 6,707,700  1,069n.a.6,707,700  1,655 n.a. 6,654,900  1,655n.a.6,654,900  1,744 n.a. 8,874,500  1,744n.a.8,874,500  1,623 na. 7,250,700  1,623na.7,250,700  1,650 na. 8,543,800  1,650na.8,543,800  1,569 n.a. 9,842,900  1,569n.a.9,842,900  1,490 na. 10,273,800  1,490na.10,273,800  l,45 5 n.a. 8,972,400  l,45 5n.a.8,972,400  1,499 161 12,296,600  1,49916112,296,600  1,47 3 175 10,302,300  1,47 317510,302,300  1,399 181 4,476,500  1,3991814,476,500  1,256 188 1,880,405  1,2561881,880,405  795 217 3,838,808  7952173,838,808  785 218 3,314,998  7852183,314,998  556 256 4,282,280  5562564,282,280  514 260 2,601,117  5142602,601,117  504 26 1 3,402,540  50426 13,402,540  484 262 3,112,711  4842623,112,711  403 254 2,522,795  4032542,522,795  "},{"text":"Table 3 . Peruvian fish meal and oil exports, revenues and prices, 1963-1983.a        Fish meal   Oil  Fish mealOil  Exports Revenues Price CIF price Exports Revenues price ExportsRevenuesPriceCIF priceExportsRevenuesprice Year (t) (US$) (US$/t) (US$/t) ( 0 (US$) ( U W ) Year(t)(US$)(US$/t)(US$/t)( 0(US$)( U W )  "},{"text":"Table 4 . LP. tableau for Peruvian fishing industry.   2 -Z 36 Z (CPUCdb * Xdb)  2 -z ( o * X d )  2 -Z n Z: (PUCd, * Qd,) 2 -Z36 Z (CPUCdb * Xdb)2 -z ( o * X d )2 -Zn Z: (PUCd, * Qd,)   d . 1 b . 1    d -1   6 -1 , = 1 d . 1 b . 1d -16 -1 , = 1 AC1mtl~s   llnrvestlng block   h t   Ro- block AC1mtl~sllnrvestlng blockh tRo-block  cpYl cPY2 CPY3 CPY4 --- CPY36 X1 X2 PUCll PUC12  PUC21/22 PUCj2 cpYlcPY2CPY3CPY4---CPY36X1X2PUCllPUC12PUC21/22PUCj2  + 1           + 1 stock  + 1 + I         stock+ 1+ I     + 1 ---       + 1---       + 1      + 1 Catch wmpoml~n s l a2 a1 a2 a1 12 a1 a2 ------ a1 12 + I 0 0 + I    Catch wmpoml~ns l a2a1 a2a1 12a1 a2------a1 12+ I 00 + I Vesxl l/bl/ll l/b2/f2 I/b3/f3 llb4/f4 --- 'b36If36      Vesxll/bl/lll/b2/f2I/b3/f3llb4/f4---'b36If36 Convu M h libl llb2 1 lib, --- lP36      Convu M hliblllb21lib,---lP36 Balance eq 1 ~s o e q 2 BaLPncc eq 3 Balance eq 4       + I + I 0 5 0 5 05 -k11 -k1z  -k21/22 -k32 Balance eq 1 ~s o e q 2 BaLPncc eq 3 Balance eq 4+ I + I0 5 0 5 05-k11-k1z-k21/22-k32 Plant capapaerty         + I + I  + I Plant capapaerty+ I+ I+ I Balance eq 1            Balance eq 1 Balance eq 2            Balance eq 2 Balance eq 3            Balance eq 3 Tnnaport espacny            Tnnaport espacny Balameq 1            Balameq 1 Balance eq 2            Balance eq 2 Balance eq 3            Balance eq 3 Storage capacdy            Storage capacdy malance eq mkl            malance eq mkl Bahce eq 1            Bahce eq 1 Balance eq 2            Balance eq 2 Balance eq 3            Balance eq 3 cornex s t 1            cornex s t 1 Coma set 2            Coma set 2 Convex s t 3            Convex s t 3 Balance eq 1            Balance eq 1 Balance gl 2            Balance gl 2 ~a l a m c q 3            ~a l a m c q 3 Balance eq 1            Balance eq 1 Balance eq 2            Balance eq 2 Bllpnceeq 3            Bllpnceeq 3  "},{"text":"Table 4 . (Continued) LP. tableau for Peruvian fishing industry.    3 n     3n    + 2 (P,, * Q,,)    +2(P,, * Q,,)    1 -1 1 -1     1 -1 1 -1     Sder block   Consfrauns Sder blockConsfrauns UPll \"12 \"1\" W21 \"22 W2n UP31 \"32 \"3\" UPll\"12\"1\"W21\"22W2nUP31\"32\"3\"  "},{"text":"Table 5 . Summary of data used for base conditional programming model for Peruvian pelagic fisheries (see text for sources). Variables Values Variables Values VariablesValuesVariablesValues Prices:    Prices: Assumed demand functions:    Assumed demand functions:  "},{"text":"Table 6 . '~ensitivity of Base Programming Model of the Peruvian pelagic fishay to changes of key inputs. LP results BASE MODEL Model 1 Model 2 Model 3 Model 4 Model 5 LP resultsBASE MODELModel 1Model 2Model 3Model 4Model 5   25% increase in 25% decrease in   300% increase in 25% increase in25% decrease in300% increase in  Results of Base vessel technology vessel technology 25% increase 25% decrease vessel technology Results of Basevessel technologyvessel technology25% increase 25% decrease vessel technology Remarks Rogramming accompanied by accompanied by in product in product accompanied by RemarksRogrammingaccompanied byaccompanied byin productin product accompanied by  Model decrease in Y/f increase in Y/f prices prices decrease in C/Y Modeldecrease in Y/fincrease in Y/fpricespricesdecrease in C/Y - - -     --- -Net benefit (US$ x lo6): -Total revenues (US$ x lo6): * Fish meal: * Fish oil -Total export revenues: * Canned fish: 172.90 371.40 201.62 29.33 301.18 140.45 204.40 371.40 201.62 29.33 140.45 152.36 301.00 150.23 31.58 119.19    -Net benefit (US$ x lo6): -Total revenues (US$ x lo6): * Fish meal: * Fish oil -Total export revenues: * Canned fish:172.90 371.40 201.62 29.33 301.18 140.45204.40 371.40 201.62 29.33 140.45152.36 301.00 150.23 31.58 119.19 (assuming 50% local s$e of canned f i i )      (assuming 50% local s$e of canned f i i ) -Total costs (US$ x lo6): 19850      -Total costs (US$ x lo6):19850 -Total catch (t x lo6): -Total products (t x lo6): * Fish meal: * Fish oil: * Canned fitah: -Boat capacity used: * No. of boats: * Total boatdays (x lo3): -Product processing capacity 3.50 0.9515 0.6157 0.1259 0.2099 364.00 (63%) 87.42     -Total catch (t x lo6): -Total products (t x lo6): * Fish meal: * Fish oil: * Canned fitah: -Boat capacity used: * No. of boats: * Total boatdays (x lo3): -Product processing capacity3.50 0.9515 0.6157 0.1259 0.2099 364.00 (63%) 87.42 used (t x lo6): -Market prices (USSlt): * Fish meal: * Fish oil: * Canned fish: 0.95 (35%) 327.00 267 .OO 670.00     used (t x lo6): -Market prices (USSlt): * Fish meal: * Fish oil: * Canned fish:0.95 (35%) 327.00 267 .OO 670.00  "},{"text":"Table 1 . Whole-syste properties derived from box models inWalsh (1981; Fig. 1:  a budget for the flux of i\" carbon (g C m2 yr-) through the Peru food web before(% 1966-1969)  and after (m 1976-1979) overfishing of anchoveta), using the BASIC program inUlanowicz (1986). propertya Before 1 9 7 2 ~ After 1972' propertyaBefore 1 9 7 2 ~After 1972' Total system throughput   Total system throughput Full development capacity   Full development capacity Full ascendency   Full ascendency Overhead on inputs   Overhead on inputs Overhead on exports   Overhead on exports Overhead on respiration   Overhead on respiration Internal capacity   Internal capacity Internal ascendency   Internal ascendency Tribute to other systems   Tribute to other systems Dissipation   Dissipation System redundancy   System redundancy a ~e e Ulanowicz (1986) for definitions.   a ~e e Ulanowicz (1986) for definitions. b ~h e s e numbers imply that the Peruvian upwelling ecosystem was ,before 1972 \"better organized\" than b ~h e s e numbers imply that the Peruvian upwelling ecosystem was ,before 1972 \"better organized\" than thereafter; see footnote a.   thereafter; see footnote a.  "},{"text":"Table 2 . Estimates of apparent mean longevity (A) and daily mortality of anchoveta eggs (B) in comparison to their daily egg production estimates based on egg surveys maps and an egg production model.a e d on data in Table3ofSantander (this vol.).b ~a p prod. = egg standing stocklegg development time; theor. prod. as estimated by A B AB  "},{"text":"Table 3 . Dominant anchovy, pilchard, mackerel, horse mackerel, bonito and hake in the four major eastern boundary c ~r r e n t s . ~ Benguela Current Canary Current Peru Current California Current  Engraulis capensis b E. encrasicholus E. ringens E. rnordax Engraulis capensis bE. encrasicholusE. ringensE. rnordax Sardinops occelatus Sardina pilchardus Sardinops sagax Sardinops sagax Sardinops occelatusSardina pilchardusSardinops sagaxSardinops sagax Scomber japonicus S. japonicus S. japonicus S. japonicus Scomber japonicusS. japonicusS. japonicusS. japonicus Trachurus trachurus T. trachurus T. murphyiC T. symmetricus Trachurus trachurusT. trachurusT. murphyiCT. symmetricus Sarda sarda S. sarda S. chiliensis S. chiliensis Sarda sardaS. sardaS. chiliensisS. chiliensis Merluccius capensis M. merluccius M. gayi, M. productus Merluccius capensisM. merlucciusM. gayi,M. productus  "},{"text":"e r Bakun and Parrish (1980). b~ossibly a synonym of E. encrasicholus (see Whitehead 1981). %he horse mackerel occurring off Peru was earlier seen as a subspecies, i.e., T. symmetricus murphyi (see eg., Chirichigno 1974).   "},{"text":" Odontestes regia, see Austramenida regia Oncorhynchus nerka, see salmon Otaria byronia, 247,336, see also sea lion see also Trachurus see also Trachurus Macrocystis pyrifera, 44 Macrocystis pyrifera, 44 Melanogrammus aeglefinus, see haddock Melanogrammus aeglefinus, see haddock Merlangius merlangus, see whiting Merlangius merlangus, see whiting Merluccius bilinearis, 266 Merluccius bilinearis, 266 capensis, 334 capensis, 334 gayi, x i gayi, x i gayi peruanus, 10,11,334 gayi peruanus, 10,11,334 merluccius, 334 merluccius, 334 productus, 87,305,334 productus, 87,305,334 mollusk, 278 mollusk, 278 Munida cokeri, 260,261 Munida cokeri, 260,261 Normanichthys crokeri, 261 Normanichthys crokeri, 261 Octopus, 261 Octopus, 261  "}],"sieverID":"b4fcd7e5-6abc-44a1-b00f-ceb9fd74b422","abstract":"Peruvian-German cooperation in fisheries research is getting close to the completion of its first decade of existence, while cooperation in the fisheries sector in general is even older. Research cooperation is being conducted through the Programa Cooperativo Peruano-Alema5 de Investigacio'n Pesquera (PROCOPA) at the Instituto del Mar del Peni(1MAWE) in Callao. The main purpose of this project is to provide assistance in research areas that could not be covered sufficiently by Peruvian scientists alone, due to the restricted means available from national resources.PROCOPA, dealing with fishery resources largely dependent on the productive conditions connected with the Peruvian upwelling system, has from the start aimed to include species other than anchoveta in its research program. In this context, emphasis was put on the stock assessment of fish species in demand, or at least highly suitable, for human consumption. The final objective of the German contribution was the integration of all available and newly acquired knowledge into a model of the ecosystem. Such a model should enable us to understand the essential interactions among the major components of the system and, hence, to predict changes due either to natural causes, such as \"El NGo\" events, or to the exploitation of parts of the system by a fishery.Collaboration of Peruvian and German scientists has produced a respectable series of reports and publications culminating in the publication of \"El Ngo, Su Impacto en la Fauna Marina\", edited by W. Arntz, A. Landa and J. Tarazona (Bole& del Instituto del Mar del Penf, special volume, Callao 1985). However, PROCOPA has not only involved Peruvian and German researchers. It has, indeed, included a considerable number of internationally renowned scientists from other countries, in particular the USA and the UK, and derived benefit from their participation, which is also documented in PROCOPA's list of publications. One of the foreign scientists who participated in PROCOPA's activities already at an early stage was Dr. Daniel Pauly of ICLARM who was invited in 1981 by Dr. W. Arntz, then PROCOPA team-leader, for what was planned to be a brief consulting exercise on single-species population dynamics and a series of lectures to IMAWE staff. The present book documents into what Dr. Pauly's collaboration has grown!The involvement of ICLARM in PROCOPA is in line with the close cooperation between GTZ and this leading international research institution in the field of fishery and aquaculture development in the tropics and subtropics. It is also in agreement with the principle of making the best possible use of scientific data by analyzing them according to the highest methodological standards and to render the results available to the largest possible scientific audience and to potential users in fisheries management and politics.The present book appears to me as an excellent example of the fruits that trustful international cooperation among scientists can bring forth, even when half of the world has to be bridged in the process. I would therefore, like to congratulate first of all the editors and contributors on their success and thank them for their commitment. I would-further like to express my gratitude to Dr. H. Salzwedel, present team-leader of PROCOPA, who wholeheartedly supported this cooperative effort and successfully steered it through some of its crises, and to Dr. W. Schmidt, the project officer of PROCOPA, who helped with essential arrangements from GTZ Headquarters. Particular thanks are due to the management of IMARPE and its Executive Director, Dr. R. Villanueva, for their enthusiastic support and to ICLARM for letting Dr. Pauly contribute a major part of his \"research time \" through 1986 and 1987.   viii   "}
\ No newline at end of file