Although multiple steady states may exist, many factors besides population size influence catch data; these include fishing-gear changes, the amount of fishing effort, and market-demand factors. Furthermore, most commercial species show long-term, secular trends of decreasing abundance (catch) that are not associated with climatic trends. There are strong indications that human harvesting of populations has introduced such a large added source of mortality at such an unusual point in the age structure that natural and climate-induced variability may be masked. Thus, the use of fisheries catch data (or other estimates of population change) to test competition and predation theory, or to assess the role of extrinsic, density-independent factors such as climate, is questionable. In this case anthropogenic and natural variability cannot be separated, and long-term studies of unharvested, co-occurring species of fish are clearly called for.
In an attempt to assess anthropogenic influence on fish populations, Soutar and Isaacs (1974) utilized a time-series of fish-scale counts from cores of the annually varved sediments from the anoxic Santa Barbara basin off Southern California. They looked chiefly at the Pacific sardine and the northern anchovy, two pelagic, schooling species with overlapping distributions in the ocean and almost identical food habits. Judging from 40 years of catch records (meager for the anchovy in the early days) and from egg and larval surveys, the abundances of these two fish had apparently oscillated out of phase. Soutar and Isaacs assumed that the fish-scale count from a core (7.6 cm in diameter) provided a reliable estimate of the abundance of fish in the overlying waters, and that the region of the Santa Barbara basin was representative of the much larger area occupied by these two fish. Their results, which covered at least 1800 years in 10-year increments, showed large oscillations in abundance of both fish as far back as A.D. 400; these variations had a red spectrum, and the two species may have varied in abundance with respect to one another before man began harvesting them. Subsequent studies (e.g., Baumgartner et al., 1992) from this core plus a second one, however, have shown that the two species' abundances are most often positively correlated with one another (Figure 7). Furthermore, attempts to calibrate the sedimentary record, by using 5-year blocks from much larger box cores of sediment from only the past 50 years or so that can be matched to actual in vivo population estimates, have not yielded very convincing results. Baumgartner et al. attempted to overcome some of the sampling problems by time-averaging the data into 10-year blocks and by using only four categories of abundance; they were thus able to resolve only variations of 50-year periodicity. Since both species have generation times of two to three years, these cases have not been very useful for tests of the importance of climate in theories of competition and predation. If those estimates of the lower-frequency variations can be validated, however, they should be examined with respect to the climatic-change indices that can also be found in the cores. That comparison would seem especially apt, since at low frequencies the two species appear to be positively correlated. Such a correlation would be expected if these two ecologically similar fish were sensitive to climatic changes.
Both Lange and Schimmelmann have examined high-frequency signals of several kinds found in Santa Barbara cores, with the goal of recovering climate-related information from the varved sediments in the Santa Barbara basin that have year-to-year resolution (Lange et al., 1990; Schimmelmann and Tegner, 1991; Schimmelmann et al., 1990; Schimmelmann et al., 1992). While in some cases there seems to be low-frequency (several decades) correspondence of climate indicators and biogenic flux, the existence of higher-frequency relationships is not clear.
There is a third body of biological research that has no very structured theory behind it, nor is it concerned with the details of the intermediary mechanisms responsible for population changes. It is empirical and observational. Its main aim, which does not depend on preconceived ideas of how systems function, is to describe complex ecosystems by asking what the temporal scales of variability of basic properties such as climate, hydrography, nutrients, and biological functional groups are. In other words, what are the frequency spectra? With such information, hypotheses may be tested. Are some frequencies more important—i.e., of greater amplitude—than others? Are there regular patterns of succession over time (Wiebe et al., 1987)? Are there connections (i.e., correlations) between climate change and ecosystem change?
The assumption behind the studies that constitute this third body of research is that various components of the physical-chemical-biological systems interact to influence each other's magnitude or concentration over time. If these interactions happen in a patterned or regular way, there should be detectable statistical relationships between them in spite of a large amount of noise. Most of the researchers in this field want to know whether there are cross-correlations and, if so, at which frequencies. Since both density-dependent and density-independent forcing are probable, the fact that this approach avoids the necessity of discriminating between them is in its favor. It does, however, presuppose a series of measurements long enough and frequent enough to sample adequately all of the potentially important frequencies of variation. Since at present there are not many cases in which we know what these are, we must begin with frequent sampling over long periods of time to capture