Community Ecology

One of the fundamental challenges of ecological science is to blend population and community theory, to examine the relationships among phenomena occurring on different scales and the dynamic processes underlying the emergence of pattern. It is a challenge incompletely met; yet community ecology, in its search for integration, is leagues ahead of ecosystems ecology. There, the need and desire for synthesis are at least as great, but the gap separating the subject from population biology remains virgin territory. In each of these quests, reductionistic and holistic approaches must be wedded; in each, the goals are to understand system structure and function in relation to the dynamics at lower levels of organization, and to understand how changes at higher levels may filter down to influence lower levels.

It has long been known that marine planktonic organisms are patchily distributed at a variety of spatial scales (Steele, 1976, 1978; Haury et al, 1978). Until recently, however, most studies of plankton patchiness have focused on spatial scales larger than the order of 10 cm (vertical) and of 10 m (horizontal). Recently advances in instrumentation have now made it possible to examine patches smaller than 1 m in horizontal direction (Denman and Mackas, 1978; Platt, 1978). Evidence indicates that microscale patches as small as or smaller than i cm in diameter are common and persistent in the turbulent sea (Mitchell, 1988). These patches may have a significant role in ecosystem function. Most importantly, although organisms and nutrients are on average sparsely distributed in oceanic waters, their coassociation in micropatches could be key to energy and nutrient flows (McCarthy and Goldman, 1979).

The theoretical ecologist faces a dilemma when deciding whether to include size and age structure in an attempt to understand species dynamics. There is a large cost in terms of difficulty, so the benefit in terms of increased understanding must be substantial. This question of whether to include age structure (or size structure) in models is really a question of scale — organizational scale. At the organizational scale of species interactions, when can one ignore information at a lower level, namely at the level of differences among individuals?

Ecological theory is in an obvious state of flux. Earlier attempts to model communities using equilibrium analyses of Lotka-Volterra equations have fallen into disfavor (see the volumes edited by Price et al. 1984, and by Diamond and Case 1986). Indeed, it seems that many experimental ecologists have given up on theory altogether — rather than testing models, field ecologists are now mostly concerned with proper experimental design and the detection of significant effects attributable to one manipulation or another. Theoreticians, of course, continue to develop models; but even among theoreticians there is no consensus about the sorts of models that most deserve our attention. While everyone may agree that we need to move beyond simple Lotka-Volterra equations, few can agree upon the ways in which models might be made more pertinent to the natural world. For example, in this volume alone, arguments are made for extending theory to include: age structure (Hastings), stochastic environments (Chesson), the effects of spatial scale (Levin), fluid dynamics (Okubo), and food web architecture (Cohen, Pimm, and Yodzis). Yet clearly no model can handle all of these complications at once. This leaves us with the problem of determining which theoretical elaborations are called for by different real-world systems, and how these theories might be applied to experimental studies.

One of the major theoretical and empirical challenges in ecology today is elucidating the role of various kinds of heterogeneity, such as environmental fluctuations, in the dynamics of populations and the organization of communities. There is substantial evidence that stochastic environmental fluctuations have a strong role in population and community processes (Andrewartha and Birch 1954, 1984, Hutchinson 1961, Sale 1977, Connell and Sousa 1983, Sale and Douglas 1984, Grubb 1977, Wiens 1977, 1986, Murdoch 1979, Underwood and Denley 1984, Simberloff 1984, Murdoch et al 1985, Strong 1986, Victor 1986), and as a consequence, a variety of verbal theories of community structure in a stochastic environment have been developed (Hutchinson 1951, 1961, Paine and Vadas 1969, Sale 1977, Grubb 1977, Wiens 1977, Connell 1978). However, the mathematical theories of stochastic environments have not provided an adequate alternative to the classical theory based on deterministic models. There seem to be two reasons for this. First, the existing stochastic models have generally not provided the sort of simple quantitative results that often follow from deterministic models. Second, there is a widespread perception that models incorporating temporal variability may give unfathomable or inconsistent results (Hastings and Caswell 1979, Levin et al 1984). Indeed, an examination of models of communities of competitors reveals that a stochastic environment can have essentially any effect depending on the specific model involved and the assumptions that are made about it. Some models say that environmental variability promotes coexistence (Chesson and Warner 1981, Abrams 1984, Ellner 1984, Shmida and Ellner 1985), others say that environmental variability has little effect on coexistence (Turelli 1981, Chesson and Warner 1981), while others say that environmental variability promotes competitive exclusion (Chesson and Warner 1981).

This paper is an expository and nontechnical review of some recent discoveries about food webs. The discoveries are those I have been privileged to make jointly with two splendid collaborators: Frédéric Briand, formerly at the University of Ottawa and now at the International Union for the Conservation of Nature, Gland, Switzerland; and Charles M. Newman, at the University of Arizona, Tucson. These discoveries depend on data collected by scores of field ecologists, so the circle of contributors is much wider.

Like others in this volume, this chapter has a lot to do with scale. But it is organizational scale rather than spatial or temporal scale that I wish to consider. Much of what is commonly called community ecology deals with a very small number of species — often two. Yet communities are large sets of species — all the species in an area or at least those in a particular taxonomic or trophic group. Clearly, we need to ask how species interactions are built into large sets. Such is the aim of the various studies of food webs: food webs are the diagrams depicting which species in a community interact trophically. Among the achievements of recent food web studies is a long catalogue of food web patterns reviewed recently by Lawton (1989). These patterns can be grouped into six broad categories: the general patterns of connectance and trophic grouping across trophic levels (e.g. Briand and Cohen, 1984, Cohen, this volume, Cohen and Briand, 1984, Yodzis, 1981), the number of trophic levels (e.g. Cohen, this volume, Pimm and Lawton, 1977, Pimm and Kitching, 1987), the patterns of species which feed across trophic levels (e.g. Pimm and Lawton, 1978), the degree to which interactions are grouped into compartments (e.g. Pimm and Lawton, 1980, Yodzis, 1982), and such ratios as that between the number of species at one trophic level and the number of species at the trophic level on which these predators feed (e.g. Cohen, 1977).