1.1. Small pelagic regimes
1.2. Climate variability
Many countries and regions, particularly some developing countries, rely heavily on fishery activity for both their economic development and the production of low cost, large quantities of animal protein for human consumption (Weber 1993). The collapse of fisheries often results in severe economic damage and in the loss of many jobs, as well as making infrastructure (boats, plants, harbors, etc.), that in most cases have no alternative use and were developed at a high cost, useless (Weber 1994, Lluch-Belda 1993). Therefore, the prevention of the collapse of fisheries and the optimal use of fishery resources have been both concerns and goals of a large number of management techniques. Approaches have ranged from the purely empirical, adaptive management to the application of population dynamics theory to the definition of maximum sustainable yield and the modeling and prediction of alternative management schemes. Between these two extremes, a number of intermediate levels combining the application of quantitative models with empirical knowledge and management can be found (Walters and Hilborn 1976).
Historically, the first steps towards the development of fishery science came from the experience with large stocks growing in temperate areas and harvested by European countries such as Great Britain. Faced with very complex physical-biological systems, and wishing to predict their variability with only limited information on the system conditions, fishery theory promptly relied on a series of basic assumptions as an attempt towards the consideration of only the major sources of variability but ignoring or assuming as constant some other effects, particularly the environmental (Sharp 1995). Therefore, the first proposed mechanism affecting the abundance of fisheries resources was overexploitation of the stocks, the so-called "fisheries problem". The basic idea was that heavy fishing exerts major predation pressure on the population; this being particular important if pre-spawning fish are affected.
While this simplification enabled the development of a solid conceptual framework and resulted in major advances of the fishery science, nowadays there is increasing evidence that a critical review of those basic assumptions, and particularly the recognition of environmental effects as major sources of variability, is required if fisheries are to be either managed or predicted (Mann and Lazier 1991, Mann 1992). For some fisheries, it has became very clear that the variation of the exploitation level is, by no means, the only significant cause for stock variability (Soutar and Issacs 1969, 1974; Lluch-Belda 1993). Moreover, some studies indicate that even relatively minor changes of climate are likely to result in major and drastic changes in the abundance of some marine species (Sharp and Csirke 1983, Kawasaki et al. 1991, Beukema et al. 1990).
Largely, this recognition has been derived from the experience with the small pelagic stocks. Because they are close to the base of the food web, these species are able to accumulate biomass at their trophic level and to grow into very large populations. Therefore, it is not surprising that their fisheries are by far the most important source of marine biomass for human use, accounting for an average of about a third of the world's total fisheries production. Despite these fisheries having developed in different countries and times, and therefore under very different management approaches, it has been generally observed that a rapid growing period of a fishery is followed by an unexpected collapse that leaves economic and social disasters (Everett et al. 1996). Moreover, paleoecological evidence has shown some small pelagic populations have undergone large changes in their abundance even in the absence of any human exploitation (Soutar and Issacs 1969, 1974; Baumgartner et al. 1992).
Sardine and anchovy fisheries have undergone wide changes in their production levels, with sustained periods of high and low catch occurring grossly during the same years in different systems (Kawasaki and Omori 1988, Kawasaki et al. 1991, Lluch-Belda et al. 1989, 1992). Intensive fisheries of sardine and anchovy are in five regions of the world: the western (Japan) and eastern (California) boundary areas of the North Pacific, the eastern (Humboldt) boundary of the South Pacific, and both the northern (Canary) and southern (Benguela) boundaries of the eastern Atlantic.
A general description of the catch trends in the main systems was provided by Lluch-Belda et al. (1989, 1992). Off Japan, sardine catch peaked around 193 5, when landings exceeded 1.5 million tons. From the 1940s, however, catches decreased rapidly and only 9,000 tons were caught in 1965. Catches increased again after 1975 and attained over five million tons during the 1980s. Anchovy catch reached maximum levels of more than 400,000 tons in the 1960s and early 1970s, decreasing thereafter during the same period that sardine were increasing. California sardine formed the bulk of the regional catch from the mid-1930s to the mid-1940s. Catches peaked at over 700,000 tons in 1936, but decreased rapidly between the mid-1940s and the early 1950s. They remained low until the mid-1970s when catch off Baja California began to increase. Anchovy catches peaked between the mid-1970s and the early 1980s, with over 340,000 tons taken in 1981. They then decreased to near zero by 1990. A low abundance of sardine off Peru and Chile occurred between 1963 and 1972, thereafter catches increased rapidly peaking near 6 million tons by the mid-1980s. The anchovy fishery started in the 1950s and grew rapidly to more than 13 million tons in 1970, but it then declined very quickly. Finally, off Namibia and South Africa, a sardine fishery started after World War II. Catches increased until 1968 and then collapsed to very low levels by the 1980s. Anchovy were not exploited before the mid-1960s; thereafter catch of anchovy increased concomitant with the decrease in sardine catch.
Lluch-Belda et al. (1989, 1992) also noted these changes are more or less synchronous between distant areas, and therefore suggested some global interdecadal phenomenon linking them rather than just the effect of independent fishing pressure. Therefore they proposed the term "regimes" (after the so-called climate regimes) for referring to this scale of small pelagic variability. Small pelagic regimes are thought to be determined by low frequency, long term changes of the climate of the earth. Globally sustained warm periods have had high sardine and low anchovy abundances, while globally sustained cold periods have had high anchovy high and low sardine abundances in the Japan, California, and Peru-Chile Systems (Kawasaki and Omori 1988). The Benguela System seems to be "out of phase" with the others, and shows the opposite pattern of sardine-anchovy abundances (Crawford et al. 1987). In any case, the shifts between consecutive abundance levels is often very sharp, with populations growing or collapsing quickly and unexpectedly (Lluch-Belda et al. 1989, 1992).
The specific mechanisms linking climate variability and small pelagic regimes are not fully understood even though several suggestions have been made. A general discussion can be found on Lluch-Belda et al. (1992a). A first approach towards the explanation of stock variability, as an alternative to the Fishery Problem, stemmed from the early work of Hjort (1914), and has been postulated by a number of authors. It is based on the assumption that the differential survival of the egg-to-larvae stage determines the amount of recruitment, and thus abundance in the following years. Most early papers dealing with this "Recruitment Problem" were reviewed and discussed by Lasker and MacCall (1983), who concluded that, despite the specific mechanism proposed by each author, it is the availability of food that determines larval survival.
Since then, a number of complementary hypotheses have been proposed on some specific mechanisms linking the environmental variability to the differential survival of eggs and larvae, particularly from the experience of the CalCOFI program with the California sardine. MacCall (1986) stated the sardine population growth and intraspecific competition are the main factors that force the expansion of the sardine population. Parrish et al. (1981) proposed that offshore advection during intensified upwelling may result in the loss of eggs and larvae to unsuitable areas for growth. Bakun (1990) showed that the intensification of long-term coastal upwelling is observed in several areas, and suggested some consequences for worldwide pelagic fish population (Bakun 1993). Ahlstrom (1965) noted that high productivity (resulting from intensified upwelling) results in poor sardine year-classes, and Lasker (1981) suggested that nonstratified ocean conditions may result in poor survival.
For the Japanese sardine, Kondo (1980) suggested the rapid increase in the stock after 1970 resulted from a very strong year-class, which was caused by a gradual expansion of the spawning area, greater egg abundance, and more favorable conditions for the postlarval stage as a result of a shift of the Kuroshio Current. This current changed from meandering to straight when it was affected by an anomalous southern intrusion of the cold Oyashio Current. This shift in the oceanic circulation created a broad area suitable for copepoda nauplii, thus allowing sardine postlarvae to survive the critical period after yolk absorption. Later, Kawasaki and Omori (1988) proposed that increased solar input may encourage the expansion of the Japanese sardine population by increasing phytoplankton production.
Lluch-Belda et al. (199 la, 1991b) proposed a mechanism based in the spawning temperatures and some upwelling-related factor, from their findings that the California sardine population expands its spawning habitat wherever proper sea surface temperature (SST) and moderate upwelling activity occur. Besides temperature, upwelling was found to be an important variable determining sardine spawning, with an optimum value and with sardine avoiding spawning at either too high or too low levels. In a later article, Lluch Belda et al. (1992b) suggested that the same mechanism can explain the variability of the Humboldt system stocks. This mechanism implies the major source of recruitment variability is to be found not in the eggs to larvae survival, but on the amount of spawning itself. Both approaches are not mutually exclusive, because climate variations may directly determine spawning, and also modify the egg-survival rate.
Other hypotheses derived from the fact that sardine and anchovy populations, despite many similarities in their ecological and biological characteristics (Parrish et al. 1983, Crawford et al. 1987, Silvert and Crawford 1988), tend to alternate their abundance periods, with sardines showing high (low) abundance when anchovy population levels are low (high). Therefore, some degree of exclusive interspecific competition is suspected to occur (Nitba and Harding 1993), although some authors noted that this competition is probably not a major factor (Bakun 1996).
Paleoecological evidence suggests that sardines and anchovies do sometimes grow into large populations within the same decadal periods. Recent comparative studies (Arocena-Ponce-de León, 1996) indicate sardine and anchovy in the Gulf of California show some morphological differences regarding their buccal cavity and filtering structures. From this evidence, the authors proposed one species may feed more effectively than the other on particles of different size, and suggested long-term climate variability may favor one species over the other through this mechanism.
Up to now there has been no agreement among the different authors, although some factors such as food supply, ocean-circulation patterns, changing winds and resulting upwelling patterns, and larval food access as related to turbulence, temperature, and habitat selection have been considered important several times. The difficulty in discriminating among these factors is that they are closely related and mostly change simultaneously. Whatever the specific mechanism through which climatic regimes and abundance regimes may be related, no doubt should remain as to the existence of more than one mechanism as the cause of sardine population fluctuations. However, the relative importance of each of these may be quite different, and a great deal of research must be done to reach definitive conclusions.
Despite the many differing views, most authors agree that small pelagic abundance regimes imply major changes in many characteristics of the ocean physical environment, probably led by shifts of the ocean-atmosphere system (Lluch-Belda et al. 1989, 1992). Most authors also agree that the recognition of the environment as a major source of variability does not imply that human (i.e. fishery) impacts are to be denied, but rather that both the fishery and the environment act together in a complex and variable way (Everett et al. 1996). New and more integrated approaches are to be developed, thus demanding the collaborative effort of scientists with different orientations: fishery, oceanographers, meteorologists, ecologists, physiologists, etc. Such interdisciplinary efforts are currently being done by some international research programs such as SCOR Working Group 98 (SCOR-WG98 1996) and SPACC (Hunter and Alheit 1995).
Considering the sources of interannual climatic variability on the global scale, nowadays the best described is the ocean-atmosphere interaction known as El Niño-Southern Oscillation (ENSO). ENSO changes are known to have effects on many areas of the earth, and are major phenomena driving ocean climate variability. The following general description of ENSO nature and consequences was extracted from Kassoy (1996).
Major elements of ENSO involve a large-scale redistribution of mass, heat, and momentum within the ocean-atmosphere system, which results in large deviations from climatology of monthly, seasonal, and annual values of temperature and precipitation in many regions of the world. In general, the term ENSO refers to the general system that comprises both warm (the El Niño phase) and cold (so-called La Nina episodes) sea surface temperature extremes of Walker's Southern Oscillation (Diaz and Kiladis 1992). ENSO involves a disruption of the ocean-atmosphere system in the tropical Pacific with important consequences for weather around the globe. Among these consequences are increased rainfall across the southern tier of the USA and Peru, which has caused destructive flooding and drought in the west Pacific, sometimes associated with devastating brush fires in Australia.
In normal, non-ENSO conditions, the trade winds blow towards the west across the tropical Pacific. These winds pile up warm surface water in the west Pacific, so that the sea surface is about ½ meter higher at Indonesia than at Ecuador. The SST is about 8°C higher in the west, with cool temperatures off South America, caused by strong oceanic upwelling associated with trade winds along the equatorial regions and equatorward flow along the South American coast (Enfield, 1989). This cold water is nutrient-rich, and therefore supports high levels of primary productivity, diverse marine ecosystems, and major fisheries. Rainfall is found in rising air over the warmest water, whereas the eastern Pacific is relatively dry.
Every few years (El Niño years), this pattern is broken by periodic heavy rainfall lasting several months, associated with a dramatic increase of equatorial Pacific SST. The weakening of the South Pacific High is accompanied by a decrease in the strength of the trade winds, and the resulting weakening of the oceanic upwelling causes the SST to rise in the Eastern Tropical Pacific.
This warming of the ocean surface causes increased evaporation and heating of the surrounding troposphere and decreased atmospheric stability, creating conditions favorable for convection and rainfall. This process tends to be self-sustaining because the development of organized areas of convection near the equator tends to further weaken the trade winds within and to the west of the convection area, thereby causing SSTs in that area to remain anomalously warm. Forcing mechanisms of El Niño events are caused by a complex ocean-atmosphere instability that depends on the coupling of the two media (Díaz and Kiladis, 1992).
Another aspect of ENSO is the state opposite to El Niño, the so-called La Nina or cold event. During this phase, the climatological conditions of heavy rainfall over the monsoon regions and relatively cool SST over the Eastern Equatorial Pacific are amplified. The global-scale climatic anomalies during cold events are for the most part opposite in sign to those seen during warming events.
The direct impacts of an ENSO warm phase (El Niño) come in many forms, from storms and subsequent damage in the western United States as a result of changing weather patterns, to a disappearance of fish along the Peruvian coast. For example, the collapse of the Peruvian anchoveta has been blamed on the 1972 El Niño event. According to this idea, this significantly disturbed the cycle of upwelling along the coast, changing the ecological systems of the region. Instead of temperate water rich in nutrients, a wave of nutrient poor, warm water moved toward the coast, creating an environment unable to support the existing marine life (Cavides and Fik 1992).
Some authors have suggested that, just as the marine habitat was disrupted in South America as a result of the 1972 El Niño, the 1982-83 event altered the abundance of salmon in the US Northwest. The size and availability of coho and chinook salmon experienced reductions along the Washington, Oregon, and California, and the Fraser River, while sockeye migrated north into Canada. Extratropical effects of the very strong 1982-83 event contributed to the largest diversion in recent history of these sockeye into Canadian waters (Miller and Fluharty 1992).
However, ENSO changes are often regarded as high frequency, year-to-year variations, whereas the most significant changes of the pelagic stocks and their environment are believed to occur over the interdecadal - regime - time scale. The possible relationships between the ENSO cycle and the regime changes are not fully understood. Among the early suggestions, Bakun (pers. com.) proposed that regimes might be characterized as sustained periods of high-low ENSO activity, thus providing a link between them. Extending this idea, high(low) ENSO activity periods may be regarded either as periods with many (few) El Niño-La Niña events, periods of strong (weak) events, or as a combination of both. Similarly, regime shifts may somehow be triggered by very strong El Niño-La Niña events.
Climatic changes on the interdecadal scale have been the subject of several recent studies, particularly since it is necessary to understand the natural variability over this time scales before the proper assessment of the possible effects of the global warming can be achieved. Among the best documented cases, Beamish (1995) showed that a sudden, worldwide shift during 1976-1977 can be observed in many time series, including both atmospheric and oceanic variables (Trenberth 1990, Graham 1994, Meyers et al. 1996). At the biological level, this change of the system was experienced both by some exploited populations (Francis and Sibley 1991, Francis and Hare 1994, Drinkwater et al. 1996) and by others not under direct anthropogenic influence (Roemmich and McGowan 1995, Polovina et al. 1994, 1995). Though the origin of the 1976-77 shift is not clear, some authors believe this climatic event should not be considered an isolated phenomenon but rather an extreme phase of natural oscillations of the climate systems occurring in periods of 16 to 24 years (White and Cayan 1996).
Many workers have dealt with the possible causes of climate change over this time scale but it remains unclear whether the nature of the causes is atmospheric or oceanic, and both atmospheric forcing (Miller et al. 1994, Graham 1994) and changes in the depth of the ocean mixed layer (Polovina et al. 1995, Polovina 1996, Meyers et al. 1996) have been mentioned. Despite some major advances, no consensus has been reached yet. Some authors (Jacobs et al. 1994) have proposed that interdecadal variability is driven by the tropical system through ENSO-related mechanisms. Others believe this mechanism is likely to be found at the interaction between the tropics and the extratropics, and even others have proposed that interdecadal climate variations are basically extratropical phenomena.
Graham (1994) suggested that climate oscillations over the interdecadal time scale may be caused in the Tropical Pacific by particularly strong El Niño events that have extended and prolonged extratropical effects. These prolonged effects are believed to arrive via reflected Rossby waves generated when the ENSO-related tropical Kelvin waves reach the eastern boundary of the Pacific basin. Modeling and sea level variations suggest that after the reflected Rossby waves reach the western boundary, they promote a northward shift in the direction of the Kuroshio Current extension; these in turn force higher than normal sea surface temperatures over the northwest Pacific for periods longer than a decade (Jacobs et al. 1994).
Similar results were obtained by Latif and Barnett (1994, 1996) from their study of the effects of the seasonally varying insolation integrated over 70 years, by means of a coupled model of global ocean-atmosphere general circulation (ECHO). Their results show a 20 year cyclic signal on the heat content of the surface layer over the Northwest Pacific, the signal being stronger over the Kuroshio Current extension. Simulated sea surface temperature in this region showed an irregular oscillatory behavior on the decadal time scale. The model also showed that positive and negative anomalies change their relative position in a pattern that reminds one of the North Pacific Gyre, completing a cycle after 20 years. The authors claimed this variable pattern of the position of the anomalies may in turn result in changes of the subtropical gyre strength, therefore providing a positive feedback and accounting for the observed cyclic behavior.
However, a major difference between this mechanism and that proposed by Graham (1994) is embodied in the fact that Latif and Barnet (1994,1996) obtained similar results even when the information on the tropical Pacific variability was excluded from the model, thus suggesting the actual mechanism is not ENSO related. Similarly, White and Cayan (1996) suggested the 20-year oscillation of the North Pacific sea surface temperature is to be explained by the poleward migration of the Aleutian Low and the resulting intensification of the extratropical wind fields. The 20 year signal is detectable not only at the surface but through the upper 100 to 200 m of the ocean as previously suggested by Polovina et al. (1995) from the identification of this signal in the variations of the mixed layer depth and also in some indices of marine biological production.
The debate is still ongoing, and has been enriched by other authors who searched for tropical-extratropical interactions as the origin of interdecade variability. Ghil and Vautard (1991) used time-series analysis of global and regional air temperature records, and found that significant cyclic components occur both over interannual, ENSO-like periods (5 to 6 years) and interdecadal periods (16 to 21 years). They suggested a tropical nature for the former and a extratropical origin for the latter. More recently, Zhang et al. (1996) worked on an ENSO-like EOF (Empirical Orthogonal Function) mode in the SST global field and were able to separate two components, one identified with the ENSO variability, but the other being linearly-independent residuals comprising all the decade-to-century variability and exhibiting a larger amplitude over the extratropical North Pacific.
From the above, it seems clear this line of research is just starting and a lot of future work will be required before the causes and mechanisms of the interdecadal variability can be agreed upon, particularly regarding the linkages between the physical and the biological systems that are in the very nature of the small pelagic regimes. Therefore, the purpose of this contribution is only to provide additional evidence on the relationship between interdecadal regimes, both climatic and biological, and the ENSO as the best-known source of the variability of interannual global climate. Despite the approach being largely empirical, we hope that some of the results will be useful towards the long-term goal of explaining the causes of small pelagic world-wide fluctuations.
The work has been divided in two main parts. In the first, we deal with those results concerning the analysis of the global climate regimes as identified within the GSAT series. Global climate regimes are then compared to large-coverage air temperature series, to define wherever identified regimes should be considered truly global phenomena. Then, we present some evidence regarding possible linkages between global climate regimes and ENSO activity. The second part deals with the small pelagic regimes, as defined from the catch records of the main sardine and anchovy fisheries. Small pelagic regimes and global climate regimes are then compared. Evidence is presented concerning the possible relationships between small pelagic regimes and interdecadal tropical and extratropical variability. Some connection to decade-scale solar activity variations is also suggested. Finally, suggestions are made regarding possible connections between global small pelagic regimes and regional climate variability in those areas where large sardine and anchovy stocks occur.