Introduction

Populations of small pelagic fish such as sardines and anchovies are found in highly productive ecosystems, especially upwelling regions, around the world. These species account for nearly one-third of the global fish catches, mainly in the Humboldt, Benguela, California, and Canary current systems (^{FAO 2007}).

The northern Humboldt Current System off Peru is highly dynamic. Besides its high biological productivity, it exhibits high climate and oceanographic variability associated with seasonal, interannual (El Niño/La Niña), decadal, and long-term changes (^{Csirke and Gumy 1996}, ^{Bakun and Broad 2003}, ^{Chávez et al. 2003}).

The Peruvian sardine, Sardinops sagax (Jenyns 1842), inhabits this environment. This species used to be one of the main sources of raw material for Peru's canning industry and of great socio-economic importance for the country, particularly during the 1980s. It also plays an important ecological role in the transfer of energy from lower to higher trophic levels and as prey for a number of higher-level predators like large fish, birds, and marine mammals (^{Cury et al. 2000}). Over the last three decades, however, the sardine population in Peruvian waters has experienced notable changes in distribution and abundance levels as a result of overfishing and environmental factors, and has declined to very low levels as evidenced by the fall in sardine landings (^{Patterson et al. 1992}, ^{Csirke and Gumy 1996}, ^{Cárdenas et al. 1997}).

The objective of the present study is to analyze the historical development of the sardine fishery from 1978 to 2005 based on fishery indicators (fishing effort and landings), spatial and temporal distribution, and environmental indicators, factors that interact in complex and unexpected ways and affect the population. This information will allow a better assessment of the sardine fishery and thus help to develop appropriate management measures that will ensure its recovery.

Materials and methods

The study area included the coastal zone off Peru, from the northern border (Puerto Pizarro, 03°30'S, 80°24'W) to the southern border (La Concordia, 18°20.8'S, 70°22.6'W), and 200 nautical miles (nm) offshore. Three regions were considered for the analysis: northern, central, and southern (Fig. 1).

Figure 1: Study area along the Peruvian coast. 

The fishing effort and landing data (fishery indicators) were obtained from the project "Seguimiento de la pesquería de anchoveta y otros recursos pelágicos" (Monitoring of the anchovy fishery and other pelagic resources) carried out by the Instituto del Mar del Perú (IMARPE) from 1978 to 2005. Daily landing records per vessel were collected and examined based on subsamples of species composition. Fishing effort was determined considering the number and type of purse seine vessels and storage capacity (m³⁾.

The charts of the distribution and concentration of sardine catches were georeferenced using a geographic information system (ArcGIS v9.3). The years of the historical landing series were grouped by decades considering the mean cumulative catches per maritime area of 10 nm, parallel to the coastline every half-degree of latitude (isoparalittoral areas).

Data (1961-2005) on the following environmental indicators were used: the Pacific Decadal Oscillation (PDO) index, the Southern Oscillation Index (SOI), and Sea Surface Temperature Anomalies (SSTA) (1961-2005) from a fixed coastal station (Chimbote). The monthly sardine landing anomalies (SardAnom, dimensionless) were calculated as follows:

1

where D_SardMonth is the the original monthly sardine landing datum (tons), D_SardPattern is the sardine landing pattern (tons) from 1961 to 2005, and D_AvgTotal is the monthly average total sardine landing (tons). The physical and biological (sardine landing anomalies) time series were then linearly related based on a polynomial regression (better fit obtained with order 6), as follows:

2

where Y_SardAnom is the sardine landing anomaly (dimension-less), X are the years (1961-2005), and a₀ ... a_d are the coefficients of the equation.

Results

Fishing effort

Types of fishing vessels and gear

The Peruvian vessels targeting sardine, known locally as bolicheras, mainly use purse seine nets. Three types of fleet operated during the study period (Table 1):

Table 1: Type of fleet and characteristics of the vessels dedicated to sardine (Sardinops sagax) fishing. 

• Industrial fleet, consisting of vessels equipped with refrigeration systems that used nets with a mesh size of 38 mm (sardine net). The hold capacity (HC) of these vessels represented 27% of the total operative HC. The sardine catches were destined for direct human consumption (DHC), mainly for canning.

• Industrial fleet, consisting of vessels not equipped with refrigeration systems that used nets with mesh sizes of 13 mm (anchovy net) and 38 mm. This was the largest fleet, the HC representing more than 70% of the operative HC. The catches were destined for indirect human consumption (IHC), mainly for the production of fishmeal and fish oil.

• Artisanal fleet, consisting of vessels with a HC of <30 m³, operating within 5 nm from the shore. Between 1983 and 1995 this fleet landed 26,000 t of sardine per year, used for DHC (canning) and as bait.

By the end of the 1990s, commercial fishing vessels, with a HC of 50 to 800 m³, were using modern fish-finding instruments, and catching both anchovy (Engraulis ringens) and sardine for the production of fishmeal and fish oil.

Number of vessels and hold capacity

Fishing effort, translated into number of vessels and HC during the study period, varied significantly per type of industrial fleet (DHC, IHC) (Fig. 2). When commercial fishing began (1978-1979), an average of 241 vessels targeted fish for DHC, mainly for canning but also for the production of fishmeal and fish oil. In turn, the fleet targeting fish for IHC operated with 494 vessels, on average, with a mean HC equivalent to 139,160 m³.

Figure 2: Number of purse seine vessels (steel and wooden) and hold capacity (HC) of the Peruvian industrial fleet targeting anchovy and sardine for indirect human consumption (IHC) and direct human consumption (DHC) during the period from 1970 to 2008. 

During the 1980s, period when the sardine fishery saw greatest expansion, the average number of vessels fishing for DHC was 292, with an average HC of 36,473 m³. The fleet fishing for IHC consisted of 356 vessels, with an average HC of 96,376 m³.

From 1990 to 1993, the number of vessels fishing for DHC decreased to an average of 199. Conversely, the number of vessels fishing for IHC increased to an average of 435, with a HC of 113,044 m³. There is no information on fishing effort for the fleet targeting fish for DHC after 1994 because it has not been possible to establish differences between both fleets (DHC and IHC), since both the anchovy net and the sardine net were used indistinctly to capture both species.

Between 1994 and 1998, the industrial fleet fishing for IHC consisted of an average of 614 vessels, with a HC of 155,487 m³. These beach seiners indistinctly captured both anchovy and sardine; however, the greatest effort directed towards sardine was recorded in 1998, when the fleet consisted of 642 vessels.

During the period 1999-2005, when the sardine catches declined abruptly, the effort was almost exclusively directed towards anchovy.

Historical landings

The historical series of sardine landings (1970-2005) showed that catch levels were low in the early 1970s. When the anchovy fishery collapsed in 1972, sardine landings increased, reaching 132,000 and 72,0001 in 1973 and 1974, respectively. The industrial sardine fishery initiated in 1978; annual landings were ~1.5 million tons until 1982, decreasing in 1983 (El Niño year) to ~1 million tons.

The highest volumes of sardine landings were recorded during the 1980s (except during the 1982-1983 El Niño event), reaching a maximum of 3.5 million tons in 1988. After 1990, the catches decreased to average annual levels of ~1 million tons during the period 1992-1998. The levels subsequently decreased to 1321.

The highest sardine landings of all the time series analyzed were recorded in the northern and central regions of the Peruvian littoral, mainly in northern ports such as Chimbote (09°04'S, 78°36'W) and Paita (05°04'S, 81°06'W), from where between 40% and 75% of the landings originated, except during the 1982-1983 El Niño event. In the southern region, the landings were not very significant (Fig. 3).

Figure 3: Time series of annual landings (million tons) of the sardine Sardinops sagax (Jenyns 1842) by regions off the Peruvian coast (1970-2006). 

During the 1980s and 1990s, the sardine landings showed seasonality, the highest landings occurring during the austral summer and spring (Fig. 4).

Figure 4: Monthly comparison of sardine (Sardinops sagax) landings by decades. 

Alternation of sardine and anchovy landings

Changes in the dominance of sardine and anchovy indicate an alternation between the sardine and anchovy landings (Fig. 5). During the 1950s and 1960s the pelagic fishery was sustained by anchovy catches. The sardine landings were at that time limited to a few tens of thousands of tons but increased substantially after commercial sardine fishing began in 1978. During the 1980s, sardines dominated, the cumulative landings reaching 21 million tons, compared to 16 million tons for anchovy. After 1990, the anchovy landings began to increase and the sardine landings slowly decreased to 360 t in 2003, after which they were almost nil.

Figure 5: Alternation of sardine (Sardinops sagax) and anchovy (Engraulis ringens) landings on the Peruvian coast. Time series: 1950-2005. 

Spatial and temporal distribution of sardine catches

The distribution of sardine catches showed significant spatial and temporal variations off the coast of Peru (Fig. 6).

Figure 6: Spatiotemporal distribution of sardine (Sardinops sagax) by fishing areas during the 1980s, 1990s, and 2000-2005. 

During the 1980s, S. sagax showed a wide latitudinal distribution, from Paita in the north to Ilo in the south, and some main centers of concentration: off Chimbote and Paita in the northern region, off Callao (12°00'S, 77°04'W) in the central region, and off Ilo (17°38'S; 71°20'W) in the southern region. These fishing areas were the most important during the period under study, yielding catches of >700,0001, especially between 10 and 70 nm and up to 100 nm offshore. During this period, the concentrations of sardine were higher in areas close to shore.

In the early 1990s, the main areas of concentration were similar to those of the previous decade, but the catches were smaller. By the late 1990s, the fishing areas were more scattered, extending to 200 nm offshore. This wide distribution continued from 2000 to 2005, with yields of ~5000 t.

Comparison of Peruvian, Ecuadorian, and Chilean sardine landings

The comparative analysis of the landings of the northern-central Peru and southern Ecuador stock showed that Peru contributed more than 90% of the annual landings during the period from 1982 to 1996 (Fig. 7a). The largest landing recorded in Peru was 3.5 million tons in 1988 and the largest landing recorded in Ecuador was 1.2 million tons in 1985. The landings decreased significantly in subsecuent years, but Peru's contribution was always higher, except from 2000 to 2005.

Figure 7: Sardine (Sardinops sagax) landings (1980-2005): (a) northern-central Peru and southern Ecuador stock, and (b) southern Peru and northern Chile stock. 

Conversely, in the case of the southern Peru and northern Chile stock, during the period from 1980 to 2005 Peru contributed only 10% of the total landings and Chile the rest (Fig. 7b). Total landings (southern Peru and northern Chile) and landings per country decreased after 1985 and notably so after 1994, except in 1992, when there was a slight increase. The largest landing in southern Peru (~458,000 t) was recorded in 1980, and in Chile (2.6 million tons), in 1985.

Variability of sardine abundance in relation to environmental factors

There was a significant correlation between the sardine landing anomalies and the environmental indicators (PDO index, SOI, and SSTA). The landing anomalies showed a positive correlation with the PDO index. The very low landings from 1961 to 1975 were associated with negative values; between 1979 and 1990, when the landings increased, the PDO index values were positive; and after 1999, the trend reverted to that observed in the earlier years. This indicates long-term fluctuations of sardine landings, that is, periods of high and low yields (Fig. 8a).

Figure 8: Relationship of sardine (Sardinops sagax) landing anomalies (SardAnom): (a) Pacific Decadal Oscillation Index (PDOI), (b) Southern Oscillation Index (SOI), and (c) Sea Surface Temperature Anomaly (SSTA, data from the Chimbote station). The polynomial fit for SardAnom is also shown. Time series: 1961-2005. 

The SOI also confirmed this long-term fluctuation of sardine landings during the cold and warm periods of the historical series from 1961 to 2005 associated with positive and negative values of this index (Fig. 8b). Likewise, the landings increased and decreased when the SSTA values were positive or negative, respectively (Fig. 8c). Analysis of the Pearson correlation coefficient values showed that the PDO index had a significant positive correlation with the sardine landing anomalies (r_Pearson = 0.50; P = 0.0005; n = 45) compared with the other indices (SOI and SSTA) (Table 2).

Table 2: Correlation of sardine landing anomalies (SARD) vs the environmental indices (SOI, Southern Oscillation Index; PDOI, Pacific Decadal Oscillation Index; SSTA, Sea Surface Temperature Anomaly) based on the Pearson correlation coefficient. Time series: 1961-2005. 

Discussion

Our results indicate that overcapacity of the industrial fleet targeting anchovy in the early 1970s (~1500 vessels with a HC of 260,000 m³⁾ led to the collapse of the fishery and the onset of the commercial sardine fishery in the late 1970s. During the 1980s when the sardine catch levels were highest, the fleet HC had decreased to 89,946 m³, but proved to be sufficient for the development of the fishery.

During the 1990s, however, the number of vessels increased to 642, and between 1994 and 1998 the HC of the sardine fleet was 155,000 m³. Even though the number of vessels comprising the fleet was lower than during the 1970s, the fishing power was greater due to improved fish-finding technology, renewal of fishing gear, net modifications, and increased HC of the individual vessels. These improvements allowed greater mobility to areas of concentration and increased efficiency in view of the changes in sardine distribution, concentration, and behavior. ^{Csirke and Gumy (1996)} reported the existence of a fundamental problem related to the overcapacity of the fishing fleet and fish plants in Peru in the early 1990s.

During the 1997-1998 El Niño event, there was a notable decrease in anchovy and greater availability of sardine, mainly juveniles (0, 1, 2, and 3 years of age). The sardine population, which was already showing signs of deterioration, was subjected to high fishing pressure, especially in 1998, and thus negatively affected (^{Cárdenas 2009}).

A spatial and temporal alternation between sardine and anchovy stocks has occurred over the past three decades. During the 1980s, the main fishing grounds were occupied by sardine and the higher catch levels reflected the increase in biomass; this period is known as the "sardine decade". ^{Santander (1981)} attributed the dominance of sardine to more extensive spawning grounds, an increase in the number of eggs, and an expansion of the distribution area. The opposite occurred during the 1990s, when anchovy dominated, and during the period 2000-2005, sardine disappeared from the traditional fishing grounds.

Sardine/anchovy alternations have been reported for other upwelling areas, for example, between Sardinops caeruleus and Engraulis mordax off California and between Sardinops melanostictus and Engraulis japonicus off Japan (^{Lluch-Belda et al. 1989}, ^{Klyashtorin 1998}, ^{Bakun and Cury 1999}, ^{Bakun and Broad 2003}), possibly related to large-scale climate changes (^{Kawasaki 1983}, ^{Kawasaki and Omori 1988}, ^{Lluch-Belda et al. 1992}). ^{Van der Lingen et al. (2006)}, based on morphological evidence in the feeding apparatus of anchovy and sardine, hypothesized that alternations between these species are likely motivated by the competition for and adaptation to specific food sources. This trophodynamic hypothesis relates changes in physical and oceanological regimes and the plankton community with changes in anchovy and sardine dominance under bottom-up processes.

^{Ayón et al. (2008)} related time series of anchovy and sardine landings to zooplankton biovolumes. The zooplankton biovolumes and anchovy landings are globally in phase, whereas the zooplankton biovolumes and sardine landings are out of phase, especially when the nocturnal biovolumes are considered. This provides clear evidence of the bottom-up control of zooplankton on the anchovy population but not on the sardine population and requires further investigation.

The spatial and temporal distribution of S. sagax showed greater alterations during ENSO events (El Niño and La Niña). During El Niño 1982-1983, one of the most intense events on record, this species migrated southwards from its traditional fishing grounds off northern Peru (^{Tsukayama and Santander 1983}, ^{Vela and Rubio 1985}, ^{Cañón 1985}).

Three ENSO events occurred during the 1990s: La Niña 1995-1996, La Niña 1999, and El Niño 1997-1998, the latter of great intensity. During the two La Niñas, when upwelling intensified, the latitudinal and longitudinal distribution of S. sagax increased (<200 nm offshore). During El Niño 1997-1998, the marine environment experienced important changes due to the advance or input of Equatorial Surface Water and Subtropical Surface Water, which conditioned the distribution of the sardine fishing grounds, but not as strongly as during El Niño 1982-1983.

The changes observed in the distribution of S. sagax reveal considerable interannual variations in migration, mainly during intense El Niño events, when its distribution deepened and it migrated southwards in search of better survival conditions (food). These alterations, however, have not been observed in other ecosystems. Under normal conditions, the California sardine (S. caeruleus) migrates northwards from its spawning grounds in the south to feed (^{Bailey et al. 1982}, ^{Schwartzlose et al. 1999}). ^{Agostini (2005)} reported that the pelagic habitat is dynamic and its limits present spatial and temporal changes related to physical oceanographic variables. Nonetheless, the mechanisms causing these changes are not clear and the migratory behavior of sardines is not yet fully understood.

For the past 50 years, a significant correlation was observed between the environmental indicators (PDO index, SOI, and SSTA) and the sardine landing anomalies, which showed long-term fluctuations with periods of high and low yields, thus corroborating the decadal changes. Decadal-scale changes in distribution and concentration, influenced by environmental variations, are not only associated with seawa-ter temperature, but also with other factors such as dissolved oxygen levels in the water column, food availability, and prey size (^{Bertrand et al. 2011}).

Variations in fish catches can be seen as indicators of resource abundance and availability (^{Melo-Barrera et al. 2010}). In Peruvian waters, the high fishing pressure exerted on S. sagax by the industrial purse-seine fleet (translated into number of vessels and HC) strongly affected the abundance of this species that began declining in the mid-1990s. Sardine landings decreased from an average of 2.5 million tons during the period 1987-1992 to 304 t between 2002 and 2005, clear evidence of the collapse of the fishery, which was also affected by the extraordinary 1997-1998 El Niño event.