36 Productivity and Consumer Regulation-Concepts , Patterns, and Mechanisms Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power Introduction Productivity is a major factor affecting food web and ecosystem dynamics in natural sys- tems (Slobodkin, 1960, 1962; Odum, 1969; Fretwell, 1977, 1987; Oksanen et al., 1981, this volume). Productivity can influence as- pects of food webs like food chain length, stability, interaction strength, and species di- versity (Rosenzweig, 1971; Oksanen et al., 1981; DeAngelis et al., 1989a, 1989b; DeAngelis, 1992; Moore et al., 1993; and Abrams and Roth, 1994a, 1994b). Among these, food chain length has been the most discussed. Food chain length has been sug- gested to lengthen with productivity because trophic transfers from resources to consumers entail losses to heat and waste. Therefore more trophic levels (longer food chains) should be supported if the web receives more energy or limiting materials or if trophic transfers are more efficient. These classical trophic transfer arguments posed by Elton (1927) and developed by Hutchinson (1959) and Slobodkin (1960) are basic to productivi- ty-based food chain models (Oksanen et al., 1981; Fretwell, 1987). The actual support for the argument that energetic constraints limit food chain length in natural systems is, how- ever, open to debate (Pimm and Kitching, 1987; Oksanen, 1988; Lawton, 1989; Pimm, 1991; Persson et al., 1992; Hairston and Hair- ston, 1993; Wootton and Power, 1993). In view of the effort that has gone into the discussion of the effects of productivity on food web dynamics, we are struck by two circumstances. First, semantic and concep- tual confusion has impeded progress in the area. Many central concepts in studies of pro- ductivity and food webs have been used in- consistently, and some remain too vague to guide quantitative field measurements, or mathematical models. Second, the database from which conclusions about, for example, the effects of productivity on food chain length has been drawn is generally poor and limited to data on who eats whom (Briand and Cohen, 1987; Schoener, 1989; Pimm, 1991) rather than data on whether consumers regulate resource populations, as appropriate for tests of models of food chain dynamics. With respect to the first point, that seman- tic confusion prevails, a number of basic con- cepts like productivity, producer control, do- nor control, and trophic level are clearly in need of both clarification and operational mechanistic definitions. In this chapter, we begin by defining productivity, distinguish- ing extrinsic from intrinsic factors that con- trol the rates at which new tissue is elaborated by the web biota. Given the definition of productivity, we will ask how food chain length, biomass of different trophic levels, and food web stability are predicted to change when increasing extrinsic input in simple (mainly food chain) models. Responses of food webs to increased productivity will also depend on the nature of the consumer-re- source interactions, which leads to a consid- eration of donor control, an often misunder- stood concept, and its implications. Finally, the effects of productivity on food web dy- namics will depend on how food webs can be aggregated. This unavoidably leads to a 396 The Role of Indirect Effects in Food Webs I 395 Case, pp. 437-444. Harper and Row, New York. Wootton, J. T. 1992. Indirect effects, prey suscep- tibility, and habitat selection: Impacts of birds on limpets and algae. Ecology 73:981-991. Wootton, J. T. 1993. Indirect effects and habitat use in an intertidal community: Interaction chains and interaction modifications. American Naturalist 141 :7 1-89. Wootton, J. T. 1994. Predicting direct and indirect effects: An integrated approach using experi- ments and path analysis. Ecology 75151-165. Wootton, T. and M. E. Power. 1993. Productiv- ity, consumers, and the structure of a river food chain. Proceedings of the National Academy of Science, USA 90: 1384-1 387. Worthen, W. B. and J. T. Moore. 1991. Higher- order interactions and indirect effects: A resolu- tion using laboratory Drosophila communities. American Naturalist 138: 1092-1 104. Yodzis, P. 1988. The indeterminancy of ecologi- cal interactions. Ecology 69508-5 15. Yodzis, P. 1989. An introduction to theoretical ecology. Harper and Row, New York. Yodzis, P. and S. Innis. 1992. Body size and consumer-resource dynamics. American Natu- ralist 139:1151-1175. Zaret, T. M. 1980. Predation and Freshwater Communities. Yale University Press, New Ha- ven, CT. Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms I 397 discussion of the potentials and limitations of the trophic level concept. To address the second point, that surpris- ingly few data exist that are appropriate for examining links between productivity and community structure (including aspects such as interaction strength), we will also cover the empirical literature dealing with the impact of productivity on food web dynamics in ma- rine, stream, lake, and terrestrial systems. We will discuss aspects like effects of pro- ductivity on food chain length, community structure, interaction strength, and omnivory in light of the capacities of organisms at dif- ferent levels to track and exploit resources. Both our theoretical and conceptual treat- ments and the literature review of data sug- gest that productivity is likely to affect eco- system dynamics in a number of ways, but the effects cannot be predicted nor under- stood from consideration of productivity alone. Habitat heterogeneity (including ref- uges), disturbance and succession, size struc- ture and flexible, adaptive behavior (includ- ing defense) are other factors which interact with productivity to shape community and ecosystem patterns and dynamics. We sug- gest that synthetic models considering both energetic (organic matterhutrient) and dy- namic constraints will be more likely rather than one-dimensional models to enhance the development of new insights in this complex but crucial field. Production-Based Approaches Dejnition of Productivity Productivity is the rate by which new tissue (somatic tissue or offspring) is elaborated by organisms. Energy has often been viewed as the relevant variable limiting productivity, but other variables like precipitation, nutri- ents, or nutrient ratios may sometimes be more appropriate (Rosenzweig, 1968; Fret- well, 1977; DeAngelis, 1992; Sterner, this volume). In productivity-based food web models, the ultimate factor assumed to limit food chain length is the potential primary productivity (Oksanen et al., 1981). If nutri- ents for plants are not explicitly treated, the potential primary productivity is usually de- fined by the carrying capacity (in numbers or biomass) andlor the intrinsic rate of increase of the plant (Persson et al., 1992; Abrams and Roth, 1994a, 1994b). The potential primary productivity is different from the actual pri- mary productivity measured, because the lat- ter is affected by food web structure and dy- namics (Carpenter and Kitchell, 1988; Power, 1992~). Even with the simplistic as- sumption that the logistic growth equation describes the growth dynamics of the primary producer, at a given potential primary pro- ductivity both the per capita productivity and total primary productivity will depend on pri- mary producer biomass and turnover rate, which in turn are influenced by food web structure. To use the carrying capacity of primary producers is a phenomenological way to de- fine the potential primary productivity. In many situations a more mechanistic approach including nutrient dynamics is necessary be- cause of interactions between nutrient cycling and food web structure (DeAngelis, 1992; DeAngelis et al., 1989a, 1989b). For exam- ple, both timescale and spatial scale differ- ences between resource consumption and re- cycling of nutrients and other egested materials will in many situations require that these processes be treated explicitly. The im- portance of food web structure on nutrient recycling and measured primary productivity has been thoroughly demonstrated in lake ecosystems (Carpenter and Kitchell, 1984, 1987, 1988, 1993). There are also many ter- restrial examples of nutrient cycling affecting productivity including soil systems (Wedin and Tilman, 1990; Pastor and Naiman, 1992; Bengtsson et al., this volume). Nitrogen-fix- ation organisms will also affect the productiv- ity, this both in terrestrial and aquatic sys- tems. In streams, nutrient spiraling downstream (Newbold et al., 1981, 1982, 1983) adds an interesting twist to these inter- actions. In the spirit of making definitions in our field more explicit, we suggest that factors controlling productivity of a system can be separated into two components. One is en- ergy or materials coming from outside (allo- chtonous) the trophic system we have circum- scribed for study. When these enter the arena of the food web, they may influence its struc- ture or dynamics. We term this extrinsic con- trol of productivity. The other component is intrinsic control of productivity, which re- flects the rate and efficiency by which organ- 398 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power isms within the food web elaborate their own tissues from these resources and from each other. By definition, the renewal rate of the extrinsically controlled productivity does not depend on the interactions taking place at the scale of the circumscribed food web. Gradi- ents of extrinsic control of productivity are set up when different fluxes of phosphorus wash into lake food webs from their water- sheds (Persson et al., 1988, 1992); when dif- ferent photon densities reach benthic food webs in rivers (Wootton and Power, 1993); when different amounts of litter fall into phy- totelmata (Jenkins et al., 1992); when differ- ent annual amounts of precipitation fall on water-limited terrestrial communities (Ro- senzweig, 1968; Fretwell, 1977) and when different amounts of nitrogen are deposited from the atmosphere or used as fertilizer (Tamm, 1991). Our definition of extrinsic control of productivity includes the potential primary productivity as defined above (al- though in a more mechanistic way) but it also includes inputs to the system at higher levels of the food web. For example, zooplankton drifting into a stream from a lake is extrinsic input to the stream food web. Extrinsic input entering the system above primary producers may, in turn, affect the production of primary producers through nutrient excretion and egestion from higher trophic levels and through changes in food web structure (Polis and Hurd, this volume). The other component affecting the mea- sured productivity of the system is intrinsic (autochtonous) control of productivity. This component will, in contrast to extrinsic con- trol of productivity, depend on a number of abiotic and biotic processes taking place within the circumscribed system. The effi- ciency with which organisms transfer energy to other levels will also depend on the meta- bolic characteristic of the organisms (e.g., poikilotherms versus homotherms). In aquatic lake systems, nutrient recycling from the sediments to the open water depends on chemical processes, the impact of organisms (sediment-dwelling macroinvertebrates , rooted macrophytes, benthivorous fish) and their interaction (Anderson et al., 1988; Lodge et al., 1988). Additionally, recycling of nutrients depends on characteristics of or- ganisms (taxa, size, etc.). Herbivorous cope- pods produce fecal products in membrane- covered pellets, whereas herbivorous cladoc- erans egest fecal products continuously in relatively soluble form (Sterner, 1989) lead- ing to different responses of the phytoplank- ton community. In terrestrial ecosystems, plant species may affect intrinsic productivity by affecting litter quality (i.e., how easily litter is decomposed and nutrients are remin- eralized for plant uptake) (Wedin and Til- man, 1990; Berendse and Elberse, 1990; DeAngelis, 1992; Hobbie, 1992). The activi- ties of soil organisms clearly influence factors directly related to productivity, such as nitro- gen and phosphorus mineralization rates and soil structure (Bengtsson et al., this volume). In aboveground systems, plant community structure and rates of herbivory also seem to influence rates of nutrient cycling (McNaugh- ton et al., 1988; Wilson and Agnew, 1992; Hobbie, 1992). The above considerations lead to the con- clusion that in general productivity is not an independent factor determining food web structure (Hairston and Hairston (1993) offer more arguments supporting this view). The lack of theory relating extrinsic and intrinsic factors affecting productivity, however, has hampered the development of predictions that can be tested in the field (but see DeAngelis (1992)). This is an area where collaborations between population and ecosystem ecologists could lead to exciting new discoveries. Intrin- sic controls on productivity affect the rate at which nutrients and organic material leave the system. For example, nutrient spiral length in streams increases if local organisms are inefficient in absorbing and recycling nu- trients (see stream studies below). The rela- tive importance of extrinsic vs. intrinsic con- trol of productivity is also likely to depend on spatial and temporal context. Systems with a high edge-to-volume ratio such as seacoasts, streams, and small lakes may be more influ- enced by extrinsic control of productivity than systems with a low edge-to-volume ratio like pelagic oceans and large lakes. Temporal context (history) may matter in areas which have been subjected to glaciation where ex- trinsic control on productivity (by soil condi- tions, etc.) depend on time since the with- drawal of the glacial ice. Intrinsic control of productivity may also be more important than extrinsic control in late successional stages of communities. Finally, when inflow exceeds Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms I 399 outflow for a defined system, productivity of the system may increase, as in, for example, in the aging of lakes (Lindeman, 1942; Wet- zel, 1983). What is extrinsic or intrinsic will depend on the spatial scale that has been chosen, often somewhat arbitrarily, to delimit the in- vestigated system. In freshwater systems, for example, food web boundaries may be drawn so as to correspond to clear-cut physical boundaries, whereas in terrestrial (excluding more isolated islands) and marine systems food webs may be difficult to delineate. This is especially pronounced for measures such as food chain length, because, in general, organisms at higher trophic levels move over larger areas and hence consume productivity over a large area compared to lower levels (Cousins, this volume). Even when habitats appear physically distinct, their food webs can be linked in dynamically important ways. For example, zooplankton drifting from a lake into a stream can affect trophic control of zooplankton remaining in the lake if fish in the lake spawn in the stream and the re- cruitment of their progeny depends on re- source availability in the stream. Since most small-scale food webs are likely to be united at higher trophic positions by mobile interrne- diate and top predators (Moore and Hunt, 1988; Bengtsson et al., this volume; Polis and Hurd, this volume), compartmentalization at lower trophic positions can lead to some com- partments being affected by higher productiv- ity in other compartments via predators utiliz- ing both chains or subwebs (Holt, 1985, T. Oksanen, 1990; Polis and Hurd, this volume). Productivity and Food Web Dynamics- Theoretical Considerations Based on Some Simple Models Theories dealing with the response of trophic levels to enrichment have largely assumed that trophic levels are homogeneous (Hair- ston et al., 1960; Oksanen et al., 1981; see also Abrams (this volume)). The model of Oksanen et al. (1981) predicted that an in- crease in potential primary productivity at the bottom level should cause an increase in abundance of the top trophic level, until biomass at the top level can support a higher top trophic level. Increased primary produc- tion should also increase abundance of the top level and levels at alternate steps below it, but not at other levels. Fretwell (1977) and Oksanen et al. (1981) also proposed that an increase in potential primary productivity should cause an increase in the number of trophic levels. This model has been modified to account for evolutionary adaptation (plant defense) (Oksanen, 1990b), environmental heterogeneity (T. Oksanen, 1990) and sea- sonality (Oksanen, 1990a). In some of these modifications, adjacent trophic levels may both respond positively to enrichment. The above studies explored nonstable fixed-point dynamics in the form of popula- tion oscillations close to equilibrium (local stability). Abrams and Roth (1994a, 1994b) investigated how unstable food chains (Le., those far from equilibrium) responded to en- richment. They found that with a type 2 func- tional response, enrichment could (1) prevent the addition of trophic levels; (2) cause an increase in the abundance of both trophic levels in two-level systems; and (3) cause the extinction of the top level in three-level systems. In cycling populations, any re- sponse (-, 0, +) of the top predator was possible depending on the type of prey growth function, the form of the functional responses, and the specific growth parame- ters (r and/or K) of the basal level affected by enrichment. Abrams and Roth's analyses suggest that dynamic constraints not consid- ered in the energy constraint hypothesis (El- ton, 1927; Lindeman, 1942) may under some circumstances lead to a decrease in food chain length with increasing productivity. One of the many interesting results of this analysis is that it is possible that once a two-level system becomes unstable, the bottom level may start to increase with enrichment while the top level remains approximately constant (Abrarns and Roth, 1994b). This is the same pattern that Oksanen et al. (1981) predict should happen when a third trophic level is added to the system with enrichment. Unless the impact of the third trophic level can be evaluated (e.g., by experimental perturba- tions), observed patterns of trophic level abundances in relation to productivity cannot unequivocally test the importance of the high- est trophic level in controlling the system. Trophic levels are seldom if ever really homogeneous. Abrams ( 1993) analyzed how 400 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power predictions of enrichment models were af- fected by the existence of different types of species (heterogeneity) within trophic levels. He investigated both two- and three-level models. For two-level models, he found that, depending on configuration, enrichment could lead to (1) increases in both levels; (2) decreases in the top and increases in the bottom levels; and (3) decreases in the top level and no change in the bottom level. In three-level models, the third level most com- monly increased with productivity, although it could also decrease or not respond at all. To conclude, even the relatively simple models discussed above suggest that de- pending on the assumptions made, any re- sponse (-, 0, +) of the top level and the levels below to enrichment is possible. Major factors which will affect the predictions are (1) whether stable or unstable dynamics are assumed; (2) functional heterogeneity within trophic levels; and (3) spatial heterogeneity in the environment. In addition, the forms of the functional response of the consumers and of the resource growth function have major impact on the dynamics. The Nature of Relationships Between Resources and Consumers Bottom-Up versus Top-Down Regulation of Food Webs Observations that both consumer-prey inter- actions and resources can affect the dynamics of ecosystems have, especially in aquatic ecology, led to discussions concerning the conditions under which systems (or levels) are mainly controlled from below or from above. The resulting, so-called bottom-up/ top-down controversy (McQueen et al., 1986, 1989; Northcote, 1988) has grown, recently encompassing also terrestrial ecol- ogy (Hunter and Price, 1992; Strong, 1992). The dichotomization of interactions into bot- tom-up and top-down forces is unfortunate for several reasons. The strengths of bottom- up and top-down forces have been evaluated mainly from regressions of abundance pro- ductivity data (McQueen et al., 1986; Mc- Naughton et al., 1989; Moen and Oksanen, 1991) although other techniques have also been used (Bartell et al., 1988). There are problems, however, in evaluating bottom-up versus top-down effects based on static pat- terns of abundance. Even in predator-prey models which assume strongly coupled inter- actions (no prey refuges, no time delays, etc.), very different patterns are predicted depending on assumptions (see above). Fur- ther, the failure of prey to respond to an increaseldecrease in the predator4 . e., an apparent weak top-down effect-may also be due to behavioral responses or physical refuges and not to productivity. In our view, the impact of these variables on producer- consumer interactions should be treated as mechanistically as possible. Producer and Donor Control There are strong connections between the bottom-upltop-down controversy and the concept of donor control (e.g., DeAngelis (1992) and Strong (1992)). These connec- tions are understandable but unfortunate. The term donor control has usually been con- nected to bottom-up and resource control of food webs and ecosystems (Vadas, 1989; DeAngelis, 1992; Strong, 1992; Polis, 1994). However, there are several different meanings of donor control in the literature, and they are not wholly compatible with each other (Table 36.1). First, the term has re- ferred to situations where resource renewal rates are independent of consumption, as with most allochthonous resources (Schoener, 1973; Pimm, 1982; Begon et al., 1986; Schmitz, 1993) (case I in Table 36.1). Sec- ond, donor control has been used as a descrip- tion of a particular type of trophic interaction, where the removal rate of the resource by a consumer depends on resource density but not on consumer density (DeAngelis et al., 1975; DeAngelis, 1992; Bengtsson et al., 1995) (cases 2-4 in Table 36.1). The mecha- nisms behind these two meanings of donor control and the dynamic consequences are very different (see below). Finally, the term has been rather loosely used to describe re- source or bottom-up control in food webs (Vadas, 1989; Strong, 1992; Polis, 1994). We propose that the connection between do- nor control and bottom-up regulation of food webs, and hence the third meaning of donor control, be abandoned for a more mechanistic Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms I 401 Table 36.1. Mechanisms suggested to give rise to donor-controlled interactions in food webs. Some of them are relevant under certain conditions only (see text). The list of references gives some examples only, and is not intended to be exhaustive. Observe that intermediates between the different cases are possible (e.g.. I and 2, I and4). Tentative examples Mechanism (references and synonyms) Remarks Allochthonous inputs of resources Consumers unable to directly influence in situ renewal rates of resources (because of interaction structure) Consumer interference 4. Prey refuges 5 Consumers only feed on individuals (tissues) that no longer contribute to population (organism) growth Photons, external nutrient inputs, deep-sea bottom-living organisms, hydrothermal vents, terrestrial insects, or zooplankton as drift in streams (De Angelis, 1992) (Subsidy, Polis and Hurd (this volume)). Detritus-detritivore interactions, some herbivore-plant interactions (DeAngelis, 1992; Schmitz, 1993; Bengtsson et al. 1995) (Noninteractive reactive herbivore- plant systems, Caughley and Lawton (1981)). Predatory arthropods, parasitoids (Hassell, 1981) (DeAngelis et al., 1975; Hassell, 1981) (Interactive inferential systems, Caughley and Lawton (1981)). Algae having refuges from grazing in crevices (Power (1990a). Littoral prey refuged from pelagic predators (Mittelbach, 1988). Parasitoid- insect interactions (Hawkins, 1993). See also e.g. Crawley (1992). (Ratio-dependent predation, Arditi and Ginzburg (1989)). Feeding on diseased or postreproductive prey individuals (Pimm, 1982; Fretwell, 1987). Resources enter from outside the system (extrinsic control of productivity). Consumers cannot have any indirect effects on resource renewal. Resources are mainly renewed within the system, but the interaction structure is such that consumers do not directly (i.e., through consumption) influence resource renewal rates. Consumers can have indirect effects on resource renewal. Intermediates between 1 and 2 are possible. consumer densities. Most likely to occur at higher Strength of donor control depends on size of refuge and prey density in relation to refuge size. Several different mechanisms are possible (e.g., Crawley (1992)). Depending on spatial separation between consumers and prey refuges, this type may grade into type I. unrealistic, as it assumes prey would die instantaneously in the absence of predation. This type may grade into type 2; e.g., scavengers on ungulates, detritus. Pimm's extreme version is view of both donor control and the terms bottom-up and top-down. The different types of donor control have different underlying mechanisms and differ- ent properties. Allochthonous inputs are in- puts of mass, numbers, or energy to the sys- tem (cf. extrinsic control of productivity), and this type of donor control applies regard- less of consumer and resource densities. Do- nor-controlled trophic interactions arise un- der certain assumptions about a predator's functional response (DeAngelis et al., 1975). These types of trophic interaction have some peculiar properties that usually are not ap- preciated. First, they assume that predator densities are close to their equilibria, since densities far from equilibrium lead to unreal- istic behavior (Schmitz, 1993; Bengtsson et al., 1995). With a more general formulation of trophic interactions (DeAngelis et al., 1975), it becomes clear that there is a contin- uum of trophic interactions or functional re- sponses, from Lotka-Volterra (the linear case of prey-dependent functional responses) to 402 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power donor control. The position along this contin- uum depends on densities and characteristics of predators and prey as well as environmen- tal conditions (DeAngelis et al., 1975). It is also worth noting that trophic interaction donor control is the linear limiting case for ratio-dependent functional responses (Arditi and Ginzburg, 1989). Hence there is a clear but often disregarded connection between do- nor control and ratio-dependence. Under allochthonous donor control, re- source renewal rates are independent of con- sumption. Steady-state densities of re- sources, however, may be affected by consumption in all cases of donor control. This is because consumer per capita resource acquisition rates may vary, and because con- sumption in itself depresses resource density. In general, prey death rates are due to both predation and other factors. Under such con- ditions, equilibrium prey (resource) densities may, depending on model formulation, be lower when the predator is present than when it is absent (cf. Schmitz (1993). This means that the proposition of Pimm (1982) (see also Begon et al. (1986)) that the removal of the consumer in a donor control interaction should not lead to a change in the steady- state density of the resource is erroneous in many cases. Table 36.1 summarizes the different mech- anisms that may give rise to donor-controlled dynamics. Only one of these, allochthonous inputs (case l), applies widely in nature. Ex- amples include the input of photons from the sun fueling plants, phosphorus entering lakes from their watersheds, and atmospheric nitro- gen deposition. These examples show that all food webs have a degree of donor control in them. Few if any food webs are devoid of this type of donor control. One important distinction between allo- chthonous inputs and the other mechanisms in Table 36.1 is that in the latter four cases, resources can be influenced by consumers via indirect pathways such as nutrient cycling (Menge, 1992, this volume; Bengtsson et al., this volume), which allochthonous inputs of resources by definition cannot be. Mecha- nisms 2-5 for donor control (Table 36.1) rep- resent situations which appear rather special and should have limited applicability to the field. Circumstances when donor control tro- phic interactions may occur are when preda- tors are so abundant as to make interference overwhelmingly important, or when only a small proportion of the prey is available, i.e., the prey has a refuge. However, since the impact of interference is likely to depend on consumer density and the impact of refuges on resource density, the conditions under which donor control trophic interactions oc- cur in natural systems are expected to be rare. The different cases of donor control also represent different degrees of spatial separa- tion between consumers and the renewal sites of their resources, e.g., from the remote ori- gins of allochthonous inputs (case l), to re- source refuges that may be adjacent to har- vestable resource components (case 4), and resources that are renewed within the habitat but only consumed after they die or senesce (e.g., the plant-detritus-detritivore system; case 2; see also Power et al. (this volume)). Because the degree of spatial separation may influence consumer-resource dynamics (Ok- sanen 1990a), it is obvious that spatial hetero- geneity in habitats may have effects on the dynamics of food webs, and the ways in which productivity affects food web structure. Both common sense and data suggest a continuum of possibilities from Lotka-Vol- terra to donor control trophic interactions in food webs (Lawton, 1989). Hence, we should only view these two as extreme points with most consumer-resource interactions ly- ing between them. For theoretical purposes it may be useful to explore the consequences of the extremes, i.e., Lotka-Volterra dynam- ics and trophic interaction donor control dy- namics. In order to preserve clarity when modeling complex food webs it may be nec- essary to use a few generic, phenomenologi- cal mathematical representations of the inter- actions, despite the fact that several mechanisms can produce them. In most cases, however, a more mechanistic form of trophic interactions in models of natural eco- systems is to be preferred. The Trophic Level Concept Interactive versus Descriptive Units Another central concept in ecology, and which, like donor control, has been used with Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms I 403 different meanings, is the trophic level con- cept. The idea that species in a community can be hierarchically sorted into levels repre- senting their distance from the source of en- ergy or food production is venerable in ecol- ogy (Elton, 1927; Lindeman, 1942). This view is central to the concept of trophic struc- ture in communities. It is also the basis of a major ecological controversy, for many (see Cousins (1987) and Polis (1991) for recent treatments) argue that the concept of a dis- crete trophic level is purely abstract and of limited or no use. For example, many organ- isms such as carnivorous plants, monkeys, corals, anemones, and filter-feeding inverte- brates may simultaneously occupy different trophic levels. Other animals, including am- phibians, fishes, and sea stars may occupy different levels at different stages of their life cycles. Various solutions to this problem have been proposed, including relaxing the strict hierarchical nature of trophic levels by portraying trophic structure as a food web, calculating average food chain length, or fractionating species into trophic levels in some way that reflects their intermediate po- sitions. Alternatively, while conceding that vagueness in trophic level is real and may sometimes be important, others argue that the trophic level concept is a powerful tool, particularly in understanding community dy- namics. For example, Hairston et al. (1960) argued that terrestrial communities comprise three trophic levels (predators, herbivores, and producers), and that community structure is regulated by direct and indirect conse- quences of interactions among these levels. Abundant, food-limited predators compete strongly, leading to intense predation on, and low abundance of herbivores. Herbivory is thereby weak, releasing plants to grow so abundant that they compete strongly for re- sources. It appears that part of the controversy sur- rounding the term trophic level may be based on confusion regarding its meaning. To some, trophic levels are descriptive, simply indicating how many energy transfers have occurred between the sun and a species or trophic group. As used in trophic dynamics models (Hairston et al. 1960; Hairston and Hairston, 1993; Oksanen et al., 1981; Fret- well, 1987), trophic levels are dynamic enti- ties including only trophically homogeneous groups which have an impact on, or are im- pacted by, comparable groups which can be arranged hierarchically. A key feature of this view is that the groups respond dynamically as a whole (i.e., abundances or biomasses of the level increase or decrease). This confusion seems similar to the confu- sion and controversy that has arisen from using food web to refer both to static descrip- tions of feeding links among species and to the strength of interaction links among spe- cies (Paine, 1980, 1988; Dayton, 1984). Consequently, the term interaction web (comparable to Paine's (1980) functional web) was introduced to refer to dynamic webs while food web (comparable to Paine's (1980) connectedness web) was reserved to indicate descriptive webs (Menge and Suther- land, 1987; Menge, 1995; Menge et al., this volume). Paine (1980) also distinguished an- other entity, the energy $ow web, which is more quantitative in that links reflect magni- tudes of energy transfer. Energy flow webs are intermediate between food and interaction webs, in that they represent the amount of energy flowing between trophic compart- ments (Lindeman , 1942). Nonetheless, en- ergy flow webs are still descriptive and pro- vide no indication of the factors regulating community structure. Interaction webs in- clude only those components of food webs which interact strongly, both trophically and nontrophically . Links in interaction webs are based on experimental evidence that abun- dance, distribution, or size of linked species are altered in an ecologically significant way by the interaction. To increase the precision of the terms used in these discussions, we propose the follow- ing definitions, which we will use in the fol- lowing: Trophic position refers to how many energy transfers that have occurred between the basal resource (photons or detritus) and the species or trophic group in question (cf. Moore and DeRuiter (199 1)). Trophic groups are defined as species or groups of species that have similar dynamics because they share the same resources and predators and have similar interactions with these. This definition shares some characteristics with the concept of trophic species (trophospecies; e.g., Yodzis (1993)), which are defined ac- cording to feeding interactions. However, 404 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power trophic groups are defined interactively and are linked by birth and death processes in that species or groups of species with life history omnivory may constitute one trophic group (Figure 36. Id, below). According to our definition, dynamically defined trophic levels are a special case of trophic groups (see also below). Finally, functional groups are groups of taxa that have similar functions with respect to some community attribute or ecosystem process (cf. Moore and deRuiter (1991), Vitousek and Hooper (1993), and Bengtsson et al. (this volume)). Trophic Levels and Vertical and Horizontal Distinctnesses The three-trophic-level Hairston et al. (1960) model has been criticized by many. Criti- cisms include (1) Several predator levels may occur, producing more than three trophic lev- els; (2) herbivores are sometimes food-lim- ited; and (3) omnivory is common (e.g., Ehr- lich and Birch ( 1967), and Murdoch ( 1966)). Rebuttals argued that in the Hairston et al. 1960 scheme, trophic levels are defined dy- namically, not by static features such as pres- ence, absence, abundance, or diversity (Slo- bodkin et al., 1967; Hairston, 1985; Hairston and Hairston, 1993). While trophic levels may indeed be blurred and omnivory may occur, it was argued that the operationally important feature of trophic levels is that the dynamics of each level are dominated by or- ganisms functioning in the predicted way (Hairston and Hairston, 1993). Terrestrial communities were therefore postulated to have three interactive trophic levels even when food chain lengths with more than three levels can be drawn, because predators de- signed as secondary or tertiary by descriptive criteria usually function at the primary preda- tor level, having their greatest effect on herbi- vores. While higher-level predators are thus omnivorous, this view argues that function- ally they are all primary carnivores, and have their greatest effect on herbivores. Hairston and Hairston (1993) noted that four-level community dynamics have been documented in aquatic environments (e.g., Carpenter et al. (1985), Persson et al. (1988), and Power (1990b)). Further, the postulates of Hairston et a]. (1960) are often met in aquatic communities (Estes et al., 1978; Van- Blaricom and Estes, 1988; Dayton, 1975; Strong, 1992; Wootton and Power, 1993). Hypotheses advanced to explain such varia- tion come from models predicting that pro- ductivity or environmental stress underlies variation in trophic level number (e.g., Fret- well (1987), Menge and Sutherland (1987), Menge and Olson (1990), Oksanen et al. (1981), Oksanen (1988), and Persson et al. (1988, 1992)). Alternatively, Hairston and Hairston (1993) suggested four-level dynam- ics result from habitat and trophic constraints, not variation in productivity or environmental stress. Specifically, four levels were held to occur uniquely in pelagic environments, where the lack of a substratum constrains plants to microscopic size, thereby limiting consumer sizes successively on up the food B) C) D) Life history omnivory Trophic levels A) Trophic level categorization Heterogenity within levels Omnivory Figure 36.1. Three principal examples (b-d) of interactions where the assumption of a trophic level categorization (a) are violated. In examples c and d, distinct trophic levels are absent due to the presence of omnivory. Each circle or oval represents a trophic group. In D3, L (large) and S (small) represent end points of a size distribution. See text for further explanations. Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms I 405 chain. Substratum-bound communities were still held to have three levels. While evidence has been marshalled in support of these vari- ous viewpoints (Hairston and Hairston, 1993), counterexamples are also available. For instance, four-level dynamics have been inferred or documented in substratum-bound habitats (Simenstad et al., 1978; Wootton and Power, 1993) and floating macrophytes may be important in many aquatic systems (Wet- zel, 1983). A central feature of the trophic dynamics hypothesis is alternating control of trophic levels by predation and competition (Fret- well, 1987). This feature, in turn, requires that ecologically significant omnivory be ab- sent or unimportant because omnivory in its various forms can compromise the dis- tinctness of trophic levels vertically. Alter- nating control cannot be possible if a con- sumer controls abundances of prey on two or more levels. In communities with strong omnivory , therefore, an alternating control model may no longer be relevant. Instead models predicting an increasing effect of con- sumers on community structure with increas- ing trophic complexity may be more relevant (e.g., Menge and Sutherland (1987)). Fur- thermore, even if alternate control is ob- served with omnivory present, the dynamic behavior of this system cannot be deduced from a categorization based on trophic levels (Matsuda et al., 1986; Polis et al., 1989; Polis and Holt, 1992). Similarly within levels (horizontally), eco- logically distinct units (i.e., trophic groups) A. 3 - Level (Hairston & Hairston 1993) Fish 1 Macrophytes Phytoplankton Detritus are common, and often exhibit different dy- namics (Hunter and Price, 1992). Different taxa within aquatic phytoplankton and herbi- vore levels, for example, can respond differ- ently to manipulations (e.g., of nutrients). Ecologically important within-level changes may be obscured by seeming nonresponses at the scale of the whole trophic level (Leibold, 1989; Leibold and Wilbur, 1992; Rosemond, this volume; see also above, Productivity- based models). For example, within a level, increases in one group may be canceled out by decreases in another when groups are lumped into levels. Omnivory and Trophic Groups The problem with a trophic level categoriza- tion can be illustrated by the dynamics of a tropical rocky intertidal community (Menge et al., 1986a, 1986b) (Figure 36.2a). If this community is represented in a three-level scheme, the predator level includes three tax- onomically, morphologically, and ecologi- cally distinct groups (fishes, crabs, and whelks). However, in addition to their strong effects on herbivores and plants, fishes have strong effects on both crabs and whelks, and crabs have strong effects on whelks. In this case, the distinctness of the trophic groups could as well justify the construction of a five-level, eight-trophic-group interaction web (Figure 36.2b). However, the five-level categorization illustrates another problem. Because fish and crabs have ecologically sig- nificant effects on several levels (fish, four B. 5 - Level 5 Fish Wh(f), Grazers Filter feeders 1 Macrophytes Phytoplankton , \ Detritus Figure 36.2. ing the problem with deriving an unambiguous trophic level categorization. Alternative schemes of a Panama interaction web (Menge et al., 1986a, 1986b) illustrat- 406 1 Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power levels, crabs, three levels), they are strongly interacting omnivores. As mentioned above, the dynamic impacts of omnivorous consum- ers on community dynamics cannot be de- duced from a system forced into a simple trophic level categorization. We suggest that the trophic group defini- tion should be used only when it allows a connection to birth, growth, and death pro- cesses in a food web context. Thus, members of a trophic group necessarily share the same resources and the same predators. In Figure 36.1 we illustrate examples where a trophic group categorization (Figure 36.lb-36. Id) differs in several from a trophic level catego- rization (Figure 36. la). In case b, more than one trophic group per level is justified when ecologically distinct groups within one level exhibit different dynamics (for example A1 and B1 use different resources). In case c with an omnivorous trophic group (B2), A2, B2, C2, and D2 are different trophic groups: either they eat different resources or are preyed upon by different predators. As ex- plained above, a strongly omnivorous trophic group violates the distinctness of a trophic level categorization (but see Power (1990b) and Power et al. (1992). Case d represents a situation with life history omnivory. Since life stages are coupled through reproduction and recruitment (here in D3), we suggest that they form a trophic group (see Mittelbach and Chesson (1987) for an example of coupling between life stages). The trophic level con- cept is also generally useless in this case be- cause just the presence of size structure in D3 may change the four-dimensional situation in a (four levels) to an infinite dimensional situation (i.e., with a continuous size distri- bution in D3). Still, a trophic level categori- zation may apply over short time intervals (i.e., if nonlife history omnivory is unimport- ant), but becomes problematic over longer, population-dynamic timescales, as birth, death, and growth alter size class abun- dances. Our definition of trophic group implies that a trophic level system (Figure 36.la) is a special case, resulting when a classification based on trophic groups can be collapsed into a classification based on trophic levels (no horizontal heterogeneity and no vertical inter- connections). How often trophic levels rather than trophic groups will be useful as abstrac- tions of nature depends, in large part, on how common and functionally important omni- vory is. Although arguments have been ad- vanced that omnivory is rare, rarely impor- tant, or limited only to certain habitats or community types (e.g., Pimm and Lawton (1978), Pimm (1982), and Hairston and Hair- ston (1993)), it is becoming increasingly evi- dent that this is not the case (e.g., Darnel1 (1961), Menge and Sutherland (1987), Wal- ter (1987), Sprules and Bowerman (1988), Vadas (1990), and Polis (1991)). Because more than 90% of all taxa are size structured, life history omnivory is common, suggesting that trophic groups may in most cases be more useful than trophic levels when long timescales are of interest. If, as we believe, omnivory occurs com- monly in natural communities, it is important to consider how its role varies with environ- mental conditions. Experimental results sug- gest that omnivory is sometimes important in determining community structure (Menge et al., 1986a; Diehl, 1992, 1993) and some- times not (Spiller and Schoener, 1990, this volume; Power, 1992a). The factors underly- ing differences in the importance of omnivory are not always clear, but one possibility is variation in the relative effectiveness of prey defenses. For example, in the rocky intertidal in Panama, predators were diverse in size, morphology, locomotion speed, sensory acu- ity, activity pattern, and diet breadth. Conse- quently, no prey seemed capable of escaping predation for longer than a few months (Menge and Lubchenco, 1981). In temperate rocky intertidal communities, in contrast, prey defenses are more commonly success- ful. Neither noncoexistence escapes (part of the population lives in habitats inaccessible to predators) or coexistence escapes (popula- tions coexist with predators by growing to invulnerable sizes, developing invulnerable morphological features, evolving toxicity, etc.) were available to prey, and their abun- dance was controlled by strong omnivory in predators. In contrast, in a California river, the dominant herbivore (midges) were invul- nerable to secondary carnivores (fishes) but vulnerable to primary carnivores (inverte- brate predators) (Power, 1990b; Power et al., 1992). Despite their omnivorous habits, in- cluding high consumption of another herbi- vore (mayflies), top-level fishes controlled Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms 1 407 only the third level (invertebrate predators), releasing the second level (midges) to control abundance of the first level (algae). While variation in the effectiveness of prey defenses could help explain variation in om- nivory, we are left with the question of what causes variation in prey (plant) defense effec- tiveness. Addressing this question in detail is beyond the scope of this chapter, but a relevant and important possibility is that ulti- mately, variation in the importance of omni- vory may at least partly depend on productiv- ity (Persson et al., 1988; Menge et al., this volume). Productivity affects densities of predators, predator size, the abundance of normally scarce predators, the rate of growth of young, trophically lower life stages to older, trophically higher life stages, resource levels, and resource variety. Predator diets tend to be broad, and many consumers can shift from being functionally carnivorous to omnivorous as resource availabilities decline (Polis, 1991). Elucidating responses and im- pacts of omnivores under varying productiv- ity regimes will be a fruitful challenging area for future investigation. Productivity and the Structure and Regulation of Communities In Different Systems With the above theoretical and conceptual background, we next discuss the evidence and counterevidence for the influence of pro- ductivity on the trophic dynamics and struc- ture of communities in different systems. In contrast to the food web analyses by Pimm (1982, 1991) and Briand and Cohen (1987; Cohen et al., 1990), we focus mainly on the functional aspects of the food web dynamics in different systems. Functionally important chains link predators or consumers to re- source populations which they potentially regulate if they themselves are not regulated by predation. Our treatment includes streams, lakes, ma- rine systems, and terrestrial systems (includ- ing soil systems). Each of us has experimen- tal and comparative experience in one or more of these habitats. Because the informa- tion on how productivity affects food web dynamics varies considerably among these systems, we must address somewhat different questions in each. Whether productivity has any effect on the dynamics at all, however, will be dealt with for all systems. We know more about how productivity affects ecosys- tem dynamics for lake and stream systems, for example, than for marine and terrestrial systems. For the former, we will discuss the impact of productivity (in combination with other factors such as succession, disturbance, and prey refuges) on food chain length and interaction strength. For lake systems, we will also consider the relationship between productivity and omnivory. In marine sys- tems, only recently has the question of whether productivity affects the growth rate of different levels and interaction strength been raised, mainly in rocky shore systems. One reason for the paucity of marine studies is the openness of these systems involving transport of nutrients and organic materials between habitats and systems (extrinsic con- trol). In terrestrial systems, the relation be- tween productivity and food web dynamics has to a large extent focused on whether or not productivity increases the length of the grazing food chain. Relatively few data are available in support of either view, however. It is also problematic that these studies have largely neglected the fact that most of the carbon fixed by plants enters the soil as detri- tus rather than first passing up through herbi- vore grazing chains (cf. Wiegert and Owen (1971)). Streams Whether or not productivity lengthens food chains in streams by allowing the introduc- tion, or the growth to functional importance, of a new top predator population capable of suppressing the current top trophic level de- pends on whether three conditions are met. First, potential new top predators must be biogeographically available. Second, the populations at lower trophic positions must be capable of tracking resource enhance- ments spatially and temporally, so that prey productivity increases. Third, these prey must not be capable of sequestering the in- creased resources without passing them up the food chain (Figure 36.3). These condi- tions all depend on natural history features of the organisms, including their defense against predation and their performance un- / 408 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power Figure 36.3. Conditions that must be met if increased extrinsic productivity is to lengthen food Elevation of flux of llrnltlng resource into food web 1 Is a new top predator biogeographically available? (by dispersal. or by increase of a local population to a level at which it can suppress its prey) No change in chain length -- no Yes Can lower trophic levels capture the resource to augment their own rate of i tissue elaboration? 4 Do any of the lower trophic levels sequester the productivity gained into / chains. Food chain lengthens der constraints imposed by the physical envi- ronment. To illustrate how the three conditions spec- ified above apply in natural systems, we first consider how organisms in nutrient-limited streams might respond to fertilization. Attached algae are the major primary produc- ers in sunlit streams. Nutrient availability for attached algae in streams is a complex func- tion of nutrient concentration and boundary layer dynamics over the algae and their sub- strate (Hart and Robinson, 1990). If algae are scoured by floods, or subject to severe grazing, only low-profile, adnate forms per- sist which can absorb only a small fraction of the nutrients dissolved in the water column. Most supplemented nutrients would simply wash over a thin algal skin, and be lost down- stream. As attached algae accrue (if faster growth rates are not offset by grazing or sloughing), they project higher into the boundary layer and increase roughness, hence turbulent mixing. These changes pro- mote nutrient uptake until periphyton stand- ing crops begin to trap stagnant pockets of water. At this point, algae may become self- limiting with respect to nutrients (and also light), so further nutrient supplementation will no longer stimulate primary productiv- ity. Responses to fertilization at basal trophic levels are in general hump-shaped, because of feedbacks imposed by the physical struc- ture of the accruing plant assemblage: Condi- tion 2 starts to fail at very low, or very high producer biomass. The idea that variation in stream biota de- termines the efficiency with which they can capture, retain, and reuse nutrients has been formulated as a theory of nutrient spiraling (Newboldet al., 1981, 1982). Nutrient mole- cules cycle as they are taken up and released from plants or other biota, but in streams, these cycles are stretched downstream by transport, turning nutrient cycles into nutrient spirals. Spirals lengthen if local biota are in- efficient at retaining and recycling nutrients, so that the downstream transport between cy- cles increases. A stream with short spiral lengths will generate more biological produc- tion from the same nutrient loading than a stream with long spirals. The condition and abundance of plant biomass, as described above, has a large effect on spiral length and the efficiency of use of nutrients for stream Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms I 409 food web production. Dead logs that impede downstream drift of materials (Sedell and Froggatt, 1984; Meyer et al., 1988) also shorten stream spiral lengths. In addition, bi- ota at higher trophic levels retain nutrients and can even transport them upstream (Fitt- kau, 1970; Hall, 1972). Will nutrients that enhance primary pro- ductivity stimulate grazer populations in streams? This depends on the nutritional quality and availability of the enhanced algal biomass. In Augusta Creek, Mich., two ses- sile grazing caddis flies responded positively to phosphorus enrichment over 105 days in in situ experimental stream channels with nat- ural beds (Hart and Robinson, 1990). Psy- chomyia increased both in densities and in individual mass in fertilized channels; Leuco- frichia (a fiercely territorial species (McAu- liffe, 1984; Hart 1985)) increased only in individual mass (Hart and Robinson, 1990). In the Kuparek River of Alaska, phosphorus has been dripped during the summer season since 1983. Over the first two years, algal biomass responded, but grazing insects did not show differences between the fertilized reach and the unfertilized reference reach up- stream. Over the third and fourth summers, however, total grazing insect production was increased by fertilization, and following this, algal biomass did not increase significantly downstream of the phosphorus input. Life cycles of Arctic insects are slow (one-year to three-year life cycles), which explains the two-year lag before fertilization stimulated grazer populations enough so that insects in turn could suppress algae (Peterson et al., 1993). Will increases in grazer biomass stimulate predator populations? Sessile grazers that live within retreats, like the caddis flies and midges of the Michigan and Alaska studies, are relatively invulnerable to visual predators like odonates and insectivorous fishes (Her- shey, 1987; Hart and Robinson, 1990; Power et a1 . , 1992). If defended grazers absorb most of the enhanced primary productivity, little may reach higher trophic positions (condition 3 for lengthening the food chain may fail). Increased productivity could conceivably re- duce energy flow to predators by two mecha- nisms. Tube-dwelling midges were found to abandon their cases when local grazeable al- gal growth was not fast enough to offset depletion (Wiley and Warren, 1992). In- creased productivity would decrease the need for grazers to abandon retreats or other ref- uges and forage more widely, making them less susceptible to predators. Alternatively, productivity might speed the replacement of early successional, vulnerable grazers by later successional armored or sessile grazers. In preliminary surveys of regulated and un- regulated rivers in northern California, mo- bile grazers were replaced by sessile, cased grazers about one month earlier in a sunny productive river than in its dark, nutrient- poor tributary (Power, 1992a). Further stud- ies are needed to examine the generality of this result, and its consequences for trophic structure and fish production (see Power et al. (this volume)). Productivity may not shorten food chain length, however, if invulnerable and vulnerable taxa do not compete. In situa- tions where both defended and undefended members of lower trophic levels can respond to enrichment, transfer of productivity to higher trophic levels, with the possible lengthening of the functional food chain, might occur. After three and four years of fertilization in the Kuparek, production of both adult and young-of-the-year grayling (the only fish in the system) was stimulated by increases in mayflies and caddisflies. Future results will show whether, with continued enrichment, fish production will increase to the point that fish suppress insect biomass. Enrichment with nutrients (Penin et al., 1987; Johnston et al., 1990) or carbon (War- ren et al., 1964) has been transmitted to up- stream food webs to enhance fish production in several other experimental studies. Primary productivity in the Eel River of northern California was manipulated in situ with flow-through stream channels with five levels of shading (Wootton and Power, 1993). Food chain length was constrained by 6-mm screens placed upstream and down- stream of each of the channels. In the Eel River, this mesh size admits small (third- level) predators (fish fry and carnivorous in- vertebrates) but excludes larger fish further up in the food chain. There was no evidence for changes in food chain length: the top pred- ators colonized even the darkest treatments. Biomasses at different levels responded to productivity as predicted by simple food chain models (Rosenzweig, 1973; Oksanen 410 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power et al., 1981) for response to enrichment by a community with three functionally signifi- cant trophic levels. Some of the underlying processes, however, differed from the popu- lation-dynamics mechanisms assumed by these models. While plant biomass and that of sessile and small grazers reflected totally or largely the balance of in situ growth and losses, densities of larger mobile grazers and predators were influenced by immigration and emigration. A model in Wootton and Power (1993) explores the implications of migration for dynamic responses of trophic groups to variation in environmental produc- tivity. Spatial scales of controls on primary pro- ductivity determine their impacts on organ- isms and food webs. If the spatial scale, for example in shading, is small relative to the foraging area of a consumer, consumers sup- ported in more productive habitats can oppor- tunistically spill over to forage in habitats that otherwise could not support viable popu- lations (Holt, 1985; T. Oksanen, 1990). For highly mobile predators like larger-stream fish, deep, dark habitats are highly desirable for cover, while sunnier habitats usually pro- duce more food, both autochthonous and al- lochthonous. In the Rio Frijoles of central Panama, primary productivity varied up to 28-fold with forest canopy cover over pools. Armored catfish tracked this variation so closely that algal standing crops and individ- ual somatic growth rates of immature catfish were indistinguishable in sunny, crowded pools, and in dark pools with sparser catfish densities. When productivity of pools was changed, by treefall for example, or when pools were created or destroyed during floods, armored catfish emigrated or immi- grated within months to adjust to altered food availability (Power, 1984b). Thus, in the Rio Frijoles, food chain length in habitat deeper than 20 cm remained constant at two trophic levels across a 28-fold gradient in primary productivity, over a spatial scale of 11 km, and over a 2.5-year period of observation. The grazing catfish could closely track varia- tion in their resource's productivity (condi- tion 2 was met), but could also avoid transfer- ring it to swimming predators (because of catfish morphological defenses), or to rapto- rial predators like fishing birds (because of their behavioral defense, of avoiding shallow water even when algae were abundant there, and catfish in deeper water were food-lim- ited) (Power, 1984a; T. Oksanen et a]., in press). This latter ability depended on the flexible and rather low metabolic rates of cat- fish, which could starve for prolonged peri- ods without dying (Power, 1984a, 1984b). Food chains along the shallow margins of the Rio Frijoles appeared to have three functional levels: bird predation significantly deterred catfish grazing, allowing bands of algae to accrue (Power, 1984a; Power et al., 1989). This particular increment in food chain length was related not to productivity, but to physi- cal properties of the habitat (water depth) which restricted the effectiveness of fishing birds, the potential top predators in this com- munity. In Oklahoma streams, the primary grazers were minnows (Campostoma anomalum). These thin, soft fish remain vulnerable to swimming predators (bass, Micropterus spp.) throughout their lives. In Brier Creek, some pools were filled with filamentous green algae, and others are barren. The barren pools contained grazing minnows (two-level chains); the green pools lack these minnows and contain bass (three-level chains). When bass or minnows were transferred among pools by experimentalists, or naturally, dur- ing floods, pools changed from green to bar- ren or vice versa within weeks (Power et al., 1985). Food chain lengths in pools of this Oklahoma stream depended on the presence or absence of bass, not on primary productiv- ity. In fact, it is likely that the pools with three-level chains were on average less pro- ductive than those with two-level chains, be- cause bass preferred deeper pools, whose beds were likely to be more light-limited than beds of shallower pools where Campostoma found refuge from bass. For bass, preference for depth may be related to cover from terres- trial predators. As with birds and catfish in the Rio Frijoles, responses by grazers and predators to physical features of the habitat outweighed energetic considerations in de- termining food chain length. Lakes Productivity and Food Chain Length The recognition of productivity as a factor affecting trophic dynamics and structure in Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms 1 41 I lakes has a long history. The nature of lakes as islands and the human-induced eutrophica- tion of many lakes have also resulted in a vast number of studies-both comparative and experimental-on the impact of enrich- ment on lake ecosystem dynamics. Nutrient levels have commonly been used to predict phytoplankton primary production and bio- mass (Dillon and Rigler, 1974; Schindler, 1978; Watson and Kalff, 1981). Phytoplank- ton biomass and production have, in turn, been used to predict herbivore (zooplankton) and fish biomass (McCauley and Kalff, 1981; Hanson and Leggett, 1982; Mills and Schia- vone, 1982; Hanson and Peters, 1984). Com- monly, phosphorus is the most frequently limiting factor for algae growth in freshwa- ters (Schindler, 1978), and phosphorus load- ing was used by Persson et al. (1992) as a measure of potential productivity in lake ecosystems. Transitions between phosphorus and nitrogen limitation often occur seasonally and in waters eutrophied by humans (Wetzel, 1983; Elser et al., 1988). In very productive lakes, factors other than nutrients such as light may limit production. It is now widely recognized that the pelagic system and the benthicAittora1 system of lakes are linked in many ways (Lodge et al., 1988). The biogeochemical interdependency between pelagic and benthic habitats involves both inputs and losses to the pelagic commu- nity (Wetzel, 1979). Nutrients may be pumped from sediment through littoral mac- rophytes to the water phase (Barko and Smart, 1980; Carpenter, 1980; Hansson et al., 1987). Fish may consume food in one habitat, and during seasonal and diel foraging migrations, excrete and egest nutrients into another (Brabrand et al., 1990; Schindler et al., this volume; Vanni, this volume). Through such transport, the littoral habitat of especially small lakes may have profound effects on the productivity of the pelagic habi- tat. The importance of the littoral zone as a nutrient source has been suggested to depend on the trophic status of the lake (Lodge et al., 1988) being highest in moderately productive systems where the relative importance of macrophyte primary production is at its maxi- mum (Wetzel, 1979; Sand-Jensen, 1979; Carpenter, 1981). Evidence for strong consumer regulation of trophic dynamics was demonstrated in the pioneer pond experiments by Hrbacek et al. (1961), which showed that the presence of planktivorous and omnivorous fish had sub- stantial effects on both lower trophic levels (zooplankton, phytoplankton) and water chemistry. Since then, a number of studies carried out at different temporal and spatial scales have provided strong evidence for con- sumer regulation (reviews in Persson et al. (1988), Carpenter (1988), Anderson et al. (1988), and Leibold (1989)). The recognition of consumer regulation in many aquatic sys- tems has also led to the question of whether potential primary productivity affects food chain length in lakes. Several authors have argued that this is not the case, arguing that lake food webs, if heavily aggregated into trophic levels, generally consist of four levels (the microbial loop excluded): phytoplank- ton, zooplankton, planktivorous fish, and pi- scivorous fish (Mittelbach et al., 1988; Hair- ston and Hairston, 1993). The only evidence that productivity affects food chain length comes from studies of unproductive Scandi- navian lake ecosystems which showed that pelagic food chain length may be correlated with productivity (Persson et al., 1991, 1992; Persson, 1994). Phosphorus loading of lakes with pelagic piscivores present was signifi- cantly higher than that of lakes lacking pe- lagic piscivores. Planktivores were positively related to phytoplankton biomass and produc- tion in systems lacking pelagic piscivores, whereas no relationship was present in lakes having pelagic piscivores. In addition, zoo- plankton biomass was positively related to the phytoplankton productionhiomass ratio in systems having pelagic piscivores but not in systems lacking pelagic piscivores. All these patterns are consistent with expecta- tions from the model by Oksanen et al. (1981). Persson et al. (1992) argued, how- ever, that habitat heterogeneity could be the causal factor affecting food chain length rather than productivity per se. Biomass in the benthic habitat also depended on the phos- phorus loading rate. Benthic piscivores, how- ever, were present in all lakes studied, so no relationship between food chain length and productivity was found for this habitat. As a potential result of the impact of pe- lagic piscivores, phytoplankton biomass was virtually constant over a gradient of phospho- rus loading varying tenfold (0.03-0.3 g/m2 412 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and May E. Power year) (Persson, 1994). However, above a P- loading of 0.3 g/m2 year, phytoplankton bio- mass increased sharply with P-loading con- comitant with a drastic decrease in biomass of piscivores (secondary carnivores). This decrease has been related to size-structured interactions where planktivorous/omnivo- rous species, by competing with piscivore juveniles, strongly limit the recruitment of the latter to larger piscivorous stages (Pers- son, 1988; Persson et al., 1991). The de- crease in piscivore biomass at high productiv- ities thus suggests that constraints set by life history phenomena may have major effects on how production interacts with trophic dy- namics. Interaction Strength, Inedible Algae, and Productivity Interaction strength between different lev- els in relation to lake productivity has been discussed in a number of papers (McQueen et al., 1986, 1989; Carney, 1990; Persson et al., 1992, 1993). Carney (1990) (see also Elser et al. (1990)) suggested that the grazing pressure and nutrient regeneration imposed by herbivores are strongest in moderately productive lakes. As an explanation for the decreased grazing pressure in highly produc- tive lakes, the appearance of inedible (large filamentous) bluegreen algae in highly pro- ductive lakes was advanced. Due to the pres- ence of bluegreen algae, McQueen et al. (1986, 1989) suggested that a decoupling at the level of zooplankton-phytoplankton may be present in highly productive lakes. In- creased herbivore grazing has also been sug- gested to foster a succession from small edi- ble to large inedible algae causing a positive, nonlinear relationship between zooplankton and phytoplankton biomass among lakes (McCauley and Kalff, 1981; Leibold, 1989; Gliwicz and Lampert, 1990). Although bluegreen algae dominate primary production during a substantial part of the season in highly productive lakes, the causes of this dominance are still far from understood. In a review of experimental studies camed out at different lake productivities, Sarnelle (1992) found no evidence for a decreased capacity of herbivores to control phytoplankton with increasing productivity. Furthermore, when comparing a grazer model including inedible algae as a component with a model not in- cluding inedible algae, he found no improve- ment in fit by using the more complex model with algae split into two components. In a subsequent study, Sarnelle (1993) also found that the appearance of bluegreen algae in a highly productive lake was difficult to explain based on a simple herbivore-phytoplankton interaction. When planktivorous fish biomass was high, filamentous algae appeared only after a period of intense Daphnia grazing. When planktivorous fish biomass was low due to a fish kill, filamentous algae did not appear at all. Under these circumstances, Daphnia may have prevented the develop- ment of bluegreen algae. In enclosure experi- ments Sarnelle found that Daphnia grazing was negatively related to both the absolute and relative (percentage filamentous algae of total algae) biovolume of filamentous algae. Sarnelle's (1992, 1993) results suggest that the appearance of bluegreen algae in highly productive lakes is not necessarily the cause to the observed positive and nonlinear rela- tionship between phytoplankton and zoo- plankton. McCauley et al. (1988) (see also Murdoch and McCauley (1985) and McCauley and Murdoch (1987)) discussed the increased proportion of bluegreen algae (negatively affecting the attack rate of Daph- nia) as a potential mechanism behind the ob- served positive and nonlinear relationship be- tween phytoplankton and zooplankton biomasses among lakes, but they also sug- gested that the observed pattern could be a result of an increase in the mortality rate of zooplankton (Daphnia) with increased pro- ductivity. The latter explanation is congruent with results of Persson et al. (1991, 1992) which suggested that planktivore predation pressure on herbivores should increase from moderately to highly productive systems. The two explanations (increased proportion of inedible algae and increased mortality rate of Daphnia) are not mutually exclusive. More studies are needed to analyze the com- plex interactions between nutrient levels, al- gae composition, grazing pressure, and planktivore predation pressure. Given an in- creased planktivore predation pressure in highly productive lakes, the impact of zoo- plankton grazing on phytoplankton is ex- pected to decrease (Carney, 1990; Persson et al., 1992). This circumstance does not in Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms I 413 itself lead to the conclusion that trophic cas- cades are less frequent in highly productive systems (Strong, 1992), but the decreased grazing impact of zooplankton may rather reflect a shift from piscivore- to planktivore- dominated systems with productivity. Omnivory and Productivity Omnivory is often important in aquatic systems. In pelagic habitats, omnivory is present throughout the food web. Certain flagellate species are autotrophic and hetero- trophic as well as phagotrophic (Porter et al., 1988). Among macrozooplankton, both her- bivores and carnivores are present in the sub- classes Cladocera and Copepoda (Sprules and Bowerman, 1988). Among vertebrate preda- tors, piscivorous species are zooplanktivor- ous as juveniles (Werner, 1986; Persson, 1988) and even strictly zooplanktivorous fish species feed on both herbivorous and carnivo- rous zooplankton as well as other invertebrate predators. It is also commonly the case that planktivorous/benthivorous fish feed on both animal and plantlalgae food items (Persson, 1983; DeVries and Stein, 1992). It has been suggested that invertebrate carnivores may play a minor role in the dynamics of pelagic community structure when vertebrate preda- tors are present, i.e., the link vertebrate carnivores-invertebrate carnivores-herbi- vores may be collapsed to a vertebrate carni- vore-herbivore link (Persson et al., 1992; Diehl, 1992, 1993). Diehl(l992, 1993) also argued that the dynamic role of omnivory should be greater in 1ittoraUbenthic habitats than in pelagic habitats, one reason being the presence of structurally complex refuges in benthic habitats for intermediate invertebrate consumers. Carney (1990) advanced the hy- pothesis that, because fish feed on detritus and algae more in highly productive systems, the importance of omnivory should increase with productivity (see also Menge et al. (this volume)). Although he did not support his argument with any empirical evidence, this hypothesis is in agreement with the observa- tion that omnivorous cyprinid species, which dominate total fish biomass in highly produc- tive European lakes, may extract as much as 65% of their energy from detritus and algae (Persson (1983), see also DeVries and Stein (1992) for a North American example). In contrast, the dominant planktivores in less productive Eurasian lakes like cisco (Corego- nus albula) and whitefish (Coregonus sp.) are strictly carnivorous (cf. Hamrin and Persson (1986). The presence of relatively nutritious bluegreen algae in highly productive lakes is actually a major factor behind the dominance of cyprinid species in highly productive Eur- asian lakes (Persson, 1983). Therefore, an omnivorous feeding link between primary producers and planktivores may increase the impact of planktivores on zooplankton. Marine Communities Because most marine systems are open and the spatial scale of variation in productivity is usually very large, little information is available regarding the influence of produc- tivity on marine communities. Although oceanographers have studied the relationship between nutrients and phytoplankton inten- sively, pelagic habitats are not amenable to ecological observation, let alone experimen- tation, so little is known of community dy- namics in these habitats (e.g., Pomeroy (1991)). More is known of the role of com- munity processes in coastal marine habitats, coral reefs, and in some respects, the deep sea. Little effort has been made to evaluate the importance of the cross-scale benthic-pe- lagic couplings between these benthic marine communities and the nutrientlproductivity conditions of the water bathing them, however. Rocky Shores Until recently, studies evaluating the effect of small-scale variation in nutrients or pro- ductivity on community structure demon- strated either that no relationship existed, or that the effect was relatively minor (e.g., Bosman and Hockey (1986), Bosman et al. (1986), and Wootton (1991)). Recent work along the Oregon coast suggests that be- tween-site differences in community struc- ture and dynamics on the scale of 70-80 km correlate to consistent differences in coastal oceanographic conditions (Menge, 1992; Menge et al., this volume). Specifically, at a site washed with high nutrients, chloro- phyll-a, and phytoplankton productivity, high abundances of filter-feeding inverte- 414 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power brates, herbivorous mollusks, carnivorous whelks, and sea stars and low abundances of seaweeds are observed. In contrast, at a site washed with lower concentrations of nutri- ents, chlorophyll-a, and productivity, lower abundances of sessile and mobile inverte- brates and higher standing crops of seaweeds are observed. Evidence suggesting a causal link between oceanographic and benthic pro- cesses came from field experiments indicat- ing producer-consumer forces were stronger at the more productive site. For example, predation grazing (consumer effects), prey recruitment, filter-feeder growth rates, and seaweed growth rates (presumed conse- quences of producer effects) were all greater at the more productive location. These pat- terns imply that increases in nearshore nutri- ents and productivity lead to greater second- ary production, including higher abundances of consumers. In turn, these higher consumer abundances have stronger effects on prey abundances. Specifically, seaweed abun- dances may be kept scarcer by herbivores despite higher seaweed growth rates at the more productive site. Furthermore, more in- tense predation by lowshore sea stars leads to higher rates of prey (e.g., mussel and bar- nacle) mortality and restricts mussels to higher levels on the shore. Thus, increased productivity may indirectly control the strength of trophic interactions and thereby determine patterns of community structure. While the generality of these apparent dif- ferences is undetermined, possible broader implications are intriguing. Variation in near- shore productivity may help explain intri- guing differences among shores (e.g., in abundances of filter feeders, seaweeds, and consumers (Dayton, 1971; Menge, 1976; Foster, 1990) and may even offer insight into one basis of large-scale variation (or similar- ity) in marine benthic community structure. Sandy Beaches While sandy beaches are far less amenable than rocky shores to experimental investiga- tion, McLachlan (1990) postulated a similar relationship between productivity and the Structure and dynamics of some sandy shore communities. Wide and flat dissipative (Short and Wright, 1983) beaches occurring along wave-beaten coastal Oregon were ob- served to have high abundances, biomass, and diversity of benthic invertebrates. Such beaches often have dense blooms of surf dia- toms (Lewin and Schaefer, 1983), which are partially controlled by wave energy (Le- gendre and Demers, 1984) presumably through constant resuspension of nutrients. The resultant high levels of primary produc- tion may support the higher animal abun- dances found on such shores. Whether such increases lead to increases in the intensity of trophic interactions is unknown, however. While this difference may depend primarily on local factors (wavelshore interactions), in- vertebrate abundances on other sandy beaches (e.g., west coast of South Africa) may depend on coastal phytoplankton blooms and kelp production in the Benguela current upwelling system (Branch and Griffiths, 1988). Subtidal Marine Benthic Habitats Pelagic-benthic coupling may underlie sig- nificant variation in nearshore benthic com- munity structure. In Alaska, growth of filter feeders (mussels and barnacles) was posi- tively correlated to local kelp abundance (Duggins et al., 1989). Kelp abundance de- pends on the activity of sea otters, which control sea urchins in a Hairston et al. (1960) type trophic cascade (Estes et al., 1978; Estes and Duggins, 1995). The basis of the higher growth of filter feeders, based on stable car- bon isotope analysis (Duggins et al., 1989), appears to be that high kelp abundance leads to high levels of kelp detritus. This work thus indicates that benthic intertidal community processes can be influenced by nearshore pe- lagic processes, but the key factor seems to be a difference in sea otter abundance, which may be independent of coastal production differences. In the Gulf of Maine, phytoplankton pulses forced downward by tidally influenced inter- nal waves evidently influenced community structure on subsurface topographic peaks (Witman et al., 1993). In this region, phyto- plankton abundance, as measured by chloro- phyll-a levels, was highest between 10 and 20 m in depth, about 10 m shallower than the peak of a subsurface seamount (Ammen Rock Pinnacle). The regular passage of inter- nal waves evidently pushes this chlorophyll- Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms I 415 maximum layer down to about 35 m, provid- ing filter-feeding invertebrates (benthic ascid- ians, bryozoans) higher levels of food than occurs deeper (50 m), thus supporting higher growth rates. Again, the influence of such differences on processes higher in the food chain is unknown. At much larger spatial scales, evidence from El Niiio Southern Oscillations suggests a strong link between nutrients and phyto- plankton productivity, and abundance of con- sumers (Glynn, 1988). Populations of sea- birds, marine mammals, marine iguanas, fishes, and even coastal invertebrates in the Peru current upwelling system collapsed dur- ing the 1982-1983 El Niiio, presumably due to the lack of nutrient input into both pelagic and coastal food webs (Glynn, 1988). The community effects of these declines in con- sumers were not investigated, however. Birkeland (1987, 1988) suggested that variation in nutrients/productivity underlies geographic scale differences in coral reef community structure. Reefs in oligotrophic areas (oceanic islands) are dominated by hard corals and other sessile invertebrates, most of which recycle nutrients via animavplant symbioses. With increased nutrients, benthic and planktonic algae become more abundant, both supporting higher-level consumers (thereby leading to more intense grazing and predation) and competing with corals. Birke- land (1987, 1988) postulates that tropical benthic community structure may ultimately depend on productivity-related differences in biotic processes such as recruitment, compe- tition, grazing, and predation. Deep Sea Except for hydrothermal vents, most or- ganic input to the deep sea is external. The rate of organic input to the ocean bottom generally decreases with depth, such that the deep sea is regarded as severely energy lim- ited (Rex, 1973, 1976). With increasing depth, abundance of benthic invertebrates de- clines while diversity first increases then de- creases. Rex (1983) proposed that these pat- terns reflected changes in the importance of competition (decreases with depth) and pre- dation (increases then decreases with depth). At the ocean's floor, in turn, he postulated that productivity was so low that higher tro- phic levels could not be supported and abun- dances were extremely low. The discovery of hydrothermal vent com- munities (Ballard, 1977; Corliss et al., 1979), where chemosynthetic primary production rates are comparable to those in upwelling regions (Karl et al., 1980), provides tantaliz- ing hints that trophic structure varies with productivity. Communities around hydro- thermal vents (Corliss et al., 1979; Grassle, 1985) include mussels, limpets, crabs, tube- worms, fish, shrimp, and octopus, and appear comparable in structure to shallow benthic communities. Food webs have at their base chemosynthetic bacteria (plant equivalents), and include bacterivores (herbivore equiva- lents), carnivores, and invertebrates depen- dent on animalhacteria symbioses (Arp and Childress, 1983). These communities are rel- atively short-lived (around 10 years), sug- gesting a highly dynamic relation between productivity and the development (or disap- pearance) of dense assemblages of animals with trophic complexity comparable to shal- low hard-bottom communities (Hessler et al., 1988). As in most marine examples, how- ever, we do not yet know if the influence of productivity affects predation effects. Productivity and Food Web Structure in Terrestrial Ecosystems In terrestrial ecosystems, the major part of the carbon fixed by plants does not enter the herbivore food chains, but instead reaches the soil as detritus (Wiegert and Owen, 1971; Hairston and Hairston, 1993), forming the basis for a diverse and complex food web (e.g., Bengtsson et al., this volume; Moore and De Ruiter, this volume). There is a major difference between these pathways. Herbi- vores, both above- and belowground, are able to directly influence plant abundance and hence the renewal of the basal resource through grazing. In the detrital food web, in contrast, resources are mainly in situ donor- controlled, and consumers can only influence resource renewal through a number of indi- rect effects (cf. Bengtsson et al., this vol- ume). The detrital food web, through its ef- fect on nutrient cycling contributes, along with plant litter quality, to variation in exter- nally determined productivity. Nonetheless, the plant- and detritus-based pathways are not 416 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power independent, and are often joined at higher trophic positions. For example, the detrital food web may be used as an alternative en- ergy channel for predators of herbivores, re- inforcing the regulatory potential of predators (Polis and Hurd, this volume). Belowground herbivores may be partially dependent on de- tritus (Strong, personal communication) and hence joined with the detrital food web. Be- lowground predators can join the root- and microbial-based energy channels in the soil (Moore and Hunt, 1988). Productivity-Related Changes in Aboveground Food Webs Two major theories or hypotheses have explicitly incorporated productivity to ex- plain patterns of food web structure in terres- trial ecosystems: A purely productivity-based hypothesis formally developed by Oksanen et al. (1981; and see Oksanen (1990a) for an extension to patchily distributed prey) and Schoener's (1989) productive space hypothe- sis. Both of them have mainly addressed aboveground food webs. The productive space hypothesis was for- mulated to explain why small islands often lack the top predators occurring on larger ones. Its basis is the observation that it may not be productivity (mass x unit area-' x time-') but total productivity (area X pro- ductivity) over the area occupied by a food web that may determine the number of spe- cies (Wright, 1983) and the number of trophic levels (Schoener, 1989). This may not only lead to shorter food chains on small islands, but also variation in the processes structuring the food webs (cf. Olpanen et al. (1981), Schoener (1989), and As et al. (1992)). This hypothesis emphasizes the importance of the population processes determining the persis- tence of the top predators. It can also contain many of the other hypotheses explaining food chain length within it. However, it shares with them the same problems of scale depen- dency in defining the food web. The same large predator populations can in many cases use several smaller patches, confounding the relationships between area, productivity, and food chain length. Furthermore, by substitut- ing productivity per unit area for total produc- tivity, the contributions of area (defining the scale of population dynamics of larger spe- cies) and productivity are still not distin- guished. In contrast to most of the other hypotheses for productivity influencing food chain length, there is some unequivocal support for the productive space hypothesis in terrestrial environments. On the Bahama islands (re- view in Schoener (1989)), larger ones have lizards and other larger carnivores (four tro- phic positions), smaller islands lack these but have high densities of predatory spiders (three positions), and even smaller ones may lack spiders (two or even one trophic posi- tion). Although other factors such as environ- mental variability and hurricane damage may differ between these islands, the role of total productivity is unambiguous. Larger preda- tors are lacking on many other islands, for example in the Baltic (Angerbjom, 1985) and on land bridge islands in southeast Asia (Heaney, 1984), leading to shortened food chains or altered food web structure. For ex- ample, the lack of larger mammal predators could lead to other taxa with lower energy requirements dpminating the third trophic level (cf. e.g., As et al. (1992)). The produc- tive space hypothesis seems most likely to apply where the same patch boundaries are experienced by organisms at all trophic posi- tions. The other major theory of productivity and food web structure in terrestrial environments is that of Oksanen and coworkers (Oksanen et al., 1981, this volume; Oksanen, 1988; Moen and Oksanen, 199 1 ; see above Produc- tivity and food web dynamics). A main pre- diction of this theory is that in two-level eco- systems, herbivore biomass should increase with productivity, but as predators start to regulate herbivores, increased productivity will no longer lead to increased herbivore biomass, but rather be shunted into predators. Hence a test of it would be to relate herbivore biomass to productivity along a wide range of ecosystem productivities and examine if herbivore biomass levels off at higher pro- ductivities. McNaughton et al. (1989) did exactly this, and found a steady linear increase in herbi- vore biomass along a wide productivity gradi- ent, without any sign of leveling off. This result is consistent with, among others, theo- ries of producer control of the terrestrial her- bivore food chain. However, Moen and Ok- Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms I 41 7 1000 b 100 - A v) 2 .e 0 10 - 2 > 1- z 0.1 - e 0 + 1 1 g 0.01 sanen ( 199 1) reexamined these and other data, and suggested that herbivore biomass increased at a significantly slower rate at pro- ductivities above a suggested threshold where predators can regulate herbivores (Figure 36.4). Although a better set of data on pro- ductivity and biomasses at different trophic levels could shed some light on the issue, the types of regression analyses used are prob- lematic. Net aboveground productivity (the independent variable) may not be indepen- dent of food web structure and regulation (see above Definition of productivity). The scatter around the regression line is fairly large (Fig- ure 36.4), suggesting that a large proportion of the variation in herbivore biomass may be attributed to factors other than productivity. The decelerating slope in the relationship be- tween herbivore biomass and productivity is also predicted by other models (see above Productivity and food web dynamics; Osen- berg and Mittelbach, this volume). Although the theory of Oksanen et al. is conceptually important, the empirical support for (or against) it is usually ambiguous and open to other interpretations. For example, factors such as habitat heterogeneity may be more directly related to food chain length and the degree of predator regulation than productiv- ity per se. Nonetheless, Oksanen et al. (1983, Desert o Tundra O Temp. grassland 0 Old-field A Trop.grassland A Temp. forest 4k Trop. forest this volume) provide support for their views from tundra ecosystems. Productivity-Related Changes in Soil Food Webs Average primary productivity varies more than 30-fold between major terrestrial bi- omes, the range of productivity more than 100-fold. The main part of this production usually enters the detrital-based food webs. Community composition and soil food web structure also differ substantially between the major ecosystem types (Swift et al., 1979). Differences in productivity affect the soil community, but the problem has received lit- tle attention. In low-productive environments such as deserts, soil food webs seem to be both complex and have several trophic posi- tions (Polis, 1991). Predators also appear to regulate microbivores in some desert soils (Santos and Whitford, 1981). An exception may be extremely unproductive soil environ- ments, such as caves, where the number of trophic positions may increase with produc- tivity (Moore, personal communication). It is therefore difficult to find evidence for the views of Oksanen et al. (1981) that productiv- ity affects the number of trophic levels, even if we accept the argument that a trophic level 418 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power doesn't count if it doesn't potentially (when not suppressed by the level above) regulate the next lower level. Primary productivity affects both the amount of detritus entering the soil system as plant litter, and the quality of the litter (i.e., how easily the litter is attacked and decomposed by microorganisms). Litter quality can be measured as, for example, nitrogen concentration, C/N, or lignin/N-ra- tio (Hobbie, 1992). Usually litter quality in- creases with productivity. Substantial parts of the impacts of productivity on the soil community can probably be ascribed to re- sponses to varying plant litter quality (cf. Heal and Dighton (1985)), but few studies where these factors have been distinguished and manipulated can be found. Some experiments manipulating plant lit- ter inputs to the soil and monitoring the re- sponse of the soil community have been per- formed. The results have been mixed. For example, David et al. (1991) added or ex- cluded litter from an oak forest soil for five years. Litter exclusion decreased the abun- dance of several soil fauna groups, but more than doubling the amount of litter on one plot did not change the soil fauna community appreciably. Others have found that soil or- ganism abundances such as earthworms can be increased by litter addition (Nielsen and Hole, 1964; see David et a]. , 199 1). In a Swed- ish pine forest, experimenters manipulated both quantity and quality of detritus input (pro- ductivity) by adding or removing harvest resi- dues. Animal groups higher in the food web, e.g., fungivorous springtails and predatory ar- thropods, tended to be less abundant 15 years after whole-tree harvesting compared to where harvest residues were left (Bengtsson et al., unpublished manuscript). On the other hand, the saprovorous enchytraeids (Lund- kvist, 1983), and nematodes (Sohlenius, per- sonal communication), although both clearly favored by harvesting residues a few years after clear-cutting, did not show any treatment differences after 15 years. Similar decreases in soil arthropod abundances after whole-tree harvesting were observed two years after clear-cutting in amixed conifer-hardwood for- est by Bird and Chatarpaul(l986). Although all major groups (and species within groups) still were present, the lower litter input at for- est harvesting seemed to have effects on the soil community and the structure of the food web. In Swedish agricultural land, increased carbon (and nitrogen) inputs to the soil led to increased animal biomasses in the soil food web (Sohlenius, 1990). This increased graz- ing pressure on microbes, which may have ac- counted for the small response of microbial biomass to increased carbon inputs. This find- ing suggests that grazers (and predators) can regulate microbial dynamics. Before any gen- eral conclusions about the effects of productiv- ity on soil food web structure can be drawn, however, more explicit studies of these issues are needed. Also, the slow turnover of organic matter in temperate and boreal soils necessi- tates long-term studies. Food Chain Length and Dynamic Constraints Factors Affecting Dynamic Constraints Although the information, especially for ma- rine and terrestrial systems, is very incom- plete, the documentation presented in the pre- vious section suggests that productivity may affect the dynamics of natural systems in a number of ways. Support for the existence of a simple positive correlation between pro- ductivity and food chain length as assumed by the energy constraint hypothesis is, however, less obvious. Overall, the data from the dif- ferent systems suggests that besides produc- tivity, five major factors profoundly affect the dynamics of natural systems: presence/ absence of functionally important top preda- tors, habitat heterogeneity (including ref- uges), disturbance and succession, flexible and adaptive behavior (including inedibility) and size(stage) structure in populations. In the following, we briefly discuss the four last factors and propose that all of them will affect the dynamic constraints of ecological sys- tems. In dynamic constraints we include fac- tors which will affect the stability of the sys- tem (e.g., flexible behavior), but also factors which, by preventing the system to reach equilibrium (e.g., disturbance), will affect the probability that species constituting food chains will become extinct. Our definition of dynamic constraints has similarities to Pimm's (1982) use of the term dynamic sta- bilitylinstability but, as is apparent, our defi- Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms 1 41 9 nition of dynamic constraints has a broader meaning. We recognize that phenomena re- lated to timescales (including history) and spatial scales other than those covered under heterogeneity and disturbance may have ma- jor impacts on food web dynamics (see Strong et al. (this volume) for an illustrative example and Oksanen (1990b) for effects of seasonality). However, as this topic is treated in another chapter (Polis et al., this volume), we will here restrict ourselves to the last four factors. Because these factors affect how en- ergy is transferred through the food web, they also set energetic constraints in different parts of the food web. We conclude that future research should profit from simultaneously considering the energetic (organic matter, nu- trient) and dynamic constraints imposed on the system as a function of productivity, het- erogeneity, disturbance, flexible/adaptive be- havior and sizehtage-structured dynamics. Habitat Heterogeneity ana` Food Chain Length Evidence that habitat heterogeneity affects food chain length is available from several systems. Warren (1989) found that structur- ally complex habitats in freshwater ponds were associated with longer food chains. In streams, Power and coworkers (Power, 1992a; Power et al., 1985) demonstrated that heterogeneity profoundly affects the func- tional significance of fish. Food chain length could be negatively related to productivity, as deeper pools which had a lower productiv- ity had predators (bass) present, while shal- lower pools (with higher productivity) lacked predators (see Stream section, above). Pers- son et al. (1992) suggested that increasing heterogeneity of lakes (increasing with lake area) increases food chain length in lakes. In a terrestrial desert system, Abramsky (1988) has demonstrated a positive relationship be- tween species diversity, potentially related to food chain length, and habitat heterogeneity. This has also been observed on many islands, where the habitat diversity hypothesis has been suggested to be one of the explanations for theo species-area relation (Williamson, 1981; As et al., 1992). Habitat structure can affect population and trophic interactions in a number of ways. Heterogeneity in the environment can gener- ate relationships between predator and prey biomasses not predicted by models assuming a homogeneous environment (T. Oksanen, 1990). Heterogeneity may also mediate coex- istence between competitors and between predators and prey (Stenseth, 1980; Abrams, 1988). The mechanisms behind the stabiliz- ing effects of habitat heterogeneity often in- volve the presence of structurally complex refuges which prevent overexploitation by the predator (Hixon and Menge, 1991). Ref- uges in the form of availability of submerged vegetation has been shown to affect competi- tive and predator-prey interactions between key interactors in lake fish communities (Mit- telbach, 1988; Persson, 1993; Persson and Eklov, 1995). In marine benthic systems, the impact of epibenthic predators has been dem- onstrated to be stronger in rocky intertidal and unvegetated soft-sediment habitats compared to more structurally complex sea grass beds (Peterson, 1979; Wilson, 1991). Another ex- ample of the influence of structural complex- ity comes from a pond experiment (Diehl, 1992) where fish decreased macroinverte- brate densities more in unvegetated than in vegetated treatments. In a mesocosm experiment, Leibold and Wilbur (1992) documented strong interac- tions between a benthic food chain consisting of periphytic algae and amphibians and an open water food chain consisting of phyto- plankton and zooplankton. The biomasses of different levels also differed from those ex- pected assuming homogeneous trophic lev- els. This result corresponds to Abrams' (1993; see above Productivity-based models) theoretical analyses which showed that the introduction of heterogeneity within trophic levels may alter predictions of productivity- based models. Environmental Stress and Disturbance Gradients of environmental stress are com- mon in nature. Gradients in physical stress (force) are imposed by moving water (from streams to rivers, from coastal headlands to coves, from shallows to depths in lakes and oceans) or ice (from rivers to ocean in winter, from oceanic coves to rocky headlands in severe winters in temperate regions). Gradi- ents in physiological stress imposed by ex- treme temperatures or low moisture (up 420 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power mountainsides, from forests to deserts, away from rivers in arid habitats). Stress gradients may be complex. For example, low physical stress (flow) in streams may lead to high physiological stress, because turbulence is insufficient to remove metabolic wastes or deliver nutrients. On rocky shores, biota in habitats of low physical stress (small waves) may experience high physiological stress from high temperature or desiccation. Effects on organisms can be direct (dislodgement by waves) or indirect (waves can suppress con- sumers, thus indirectly sparing prey, or weaken prey, thus making them more suscep- tible to consumers) (Menge and Olson, 1990). Effects can be lethal (a rock smashing a benthic invertebrate) or sublethal (high tem- peratures can slow prey evasion behavior). Lethal effects have been termed physical or physiological disturbance (Menge and Suth- erland, 1987), corresponding to the ecologi- cal definition of disturbance as a discrete event that removes organisms, empties habi- tat, and frees resources (Sousa, 1985). The potential ecological significance of en- vironmental stress and its interaction with biotic forces have been incorporated into models predicting variation in ecological pro- cesses (disturbance, competition, predation, grazing) along environmental gradients, and the consequent responses of community structure (Connell, 1975; Menge and Suther- land, 1976, 1987; Menge and Olson, 1990). Evidence consistent with predictions of these models (e.g., that stress can shorten food chains, that high stress can inhibit competi- tion and predation, or lower diversity) has accumulated (Menge and Farrell, 1989; Menge and Olson, 1990; Power, 1990b; Tonn, 1990; Dunson and Travis, 1991; McClanahan, 1992; Arnott and Vanni, 1993; Locke, 1992). Both sublethal and lethal stress (distur- bance) can harm individuals, but moderate levels tend to maintain species diversity in communities (Connell, 1978; Sousa, 1984, 1985). Power et al. (this volume; see also Power (1992b)) suggest that food chains may be longest in moderately disturbed systems. Furthermore, they suggest that disturbance and productivity may interact to influence trophic structure. In terrestrial systems, envi- ronmental stress and primary production are both influenced by moisture levels (Rosen- zweig, 1968; Louda and Collinge, 1992). Hence, distinguishing their respective effects on variation in terrestrial community struc- ture will be difficult. For instance, some arc- tic islands have very low primary productiv- ity and weak herbivory (e.g., Oksanen (1988)), but it is not clear whether herbivory is inefficient because of low productivity, high environmental stress, both, or neither (a historical explanation, failure of herbivorous mammals to colonize, is also a possibility). To some degree the effects of stress and productivity may be separated if they operate on different timescales and/or spatial scales. In aquatic environments, these two factors may be of different significance in different habitats, at different times in the same habi- tat, or both (Menge and Olson, 1990). For instance, shallow benthic areas of lakes and oceans may experience strong effects of me- chanical stress from wave action or ice scour while offshore pelagic communities may be influenced more by nutrientlproduction-re- lated processes. Streams may alternate be- tween periods of strong physical stress influ- ences (during periods of high, scouring flow (Grimm and Fisher, 1989; Peterson and Ste- venson, 1992; Power and Stewart, 1987) and periods of strong influences of factors lim- iting productivity (during periods of low flow (Grimm and Fisher, 1986; Power, 1990b; Wootton and Power, 1993). SizelStage-Structured Interactions In most plant and animal taxa, conspecific individuals differ substantially in size (Nor- berg, 1988; Persson, 1988; Werner, 1988; Ebenman and Persson, 1988). The presence of size structure in the majority of taxa com- plicates trophic interactions because an or- ganism may feed at several trophic levels over its ontogeny (ontogenetic omnivory) (Polis, 1991), different trophic levels will become dynamically connected over the life cycle, and ontogenetic niche shifts involving habitat shifts will link trophic dynamics in different habitats. Study of the dynamics of sizelstage-struc- tured populations are complex and analyses of these dynamics are just beginning to ap- pear (Metz and Diekmann, 1986; Ebenman and Persson, 1988; DeAngelis and Gross, 1992). Embedding sizelstage-structured pop- Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms I 421 ulation dynamics into the context of commu- nity and ecosystem dynamics will be even more complex. Still, several lines of evidence suggest that sizektage-structured interactions have substantial effects on trophic dynamics. Mittelbach and Chesson (1987) showed in an age-based model with two stages (juveniles, adults) that an increase in resource productiv- ity for one stage produced an increase in the numbers of the other stage despite the fact that the resource productivity for the latter stage had not increased (Figure 36.5). With density dependence in both stages (juvenile survival and adult fecundity), the stage having its resource productivity increased did not re- spond as fully as expected from Lotka- Volterra dynamics due to density-dependent interactions in the other stage. As a result, the model predicted a positive correlation be- tween resource biomass and the biomass of the stage whose resource productivity increased and a negative relationship between resource biomass and the biomass of the stage whose resource did not increase. In lakes where zoo- plankton (the main diet of adult bluegill, Lep- omis macrochirus) levels varied, adult blue- gill growth was positively related to their own biomass, whereas the growth of juvenile blue- gill was negatively related to their own bio- mass, as predicted by the model (Mittelbach and Osenberg, 1993). These stage-structured interactions also had a stabilizing effect on the system (Mittelbach et al., 1988). The above example suggests that life cycle omnivory can prevent a stage from respond- ing fully to an increase in the productivity of its resource, producing patterns not predicted by nonstructured models. Neill and Peacock (1980; Neill, 1988) showed in experiments with the phantom midge (Chaoborus) that juvenile bottlenecks could prevent an adult predatory stage from responding to an in- crease in the biomass of its resource. In pisc- ivore-planktivore interactions, size-struc- tured processes may even result in negative relationships between a predatory stage (pi- scivorous size classes) and their prey (plankti- vores) due to competition between plankti- vores and juvenile piscivores (Persson, 1988; Persson et al., 1991). More complex size-structured interactions are often modeled using physiologically based models (Metz et al., 1988; DeRoos et al., 1992; DeAngelis and Gross, 1992). Analyses of one-consumer, one-resource models suggest that the introduction of size structure in these cases will have a destabiliz- ing effect on consumer-resource interactions (DeRoos et al., 1992; Gyllenberg et al., un- published manuscript). Flexible and Adaptive Behavior and Food Web Dynamics With differences in productivity, organ- isms also experience differences in the avail- abilities of particular types of food resources, and often, differences in habitat structure (if biomass of plants, corals, or other basal spe- cies are important components of this struc- ture). Few if any consumers maintain con- stant behavior as resources or habitat structure change. Flexible prey behavior has been shown to affect the impacts of plant biomass on structure and dynamics of higher trophic groups in a number of ways. For ex- ample, choice of floating algal patches which serve as food and resting sites by native frog tadpoles in the Eel River appears to involve adaptive trade-offs (Kupferberg , unpublished data). In the absence of predatory garter snakes, tadpoles choose Cludophoru mats Bluegill sunfish 7 + + adult juvenile I I $1 $1 +I I - +I I - and, hence, increase juvenile density Adult resource littoral macroinvertebrate (zooplankton) is unaffected (Mittelbach and Chesson, 1987). Figure 36.5. Stage-based interac- tions in the bluegill sunfish. An in- crease in resource production of zoo- plankton will increase adult fecundity 422 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power which are superior as food (because of diatom epiphytes). In the presence of snakes, tadpoles leave these dense mats, where garter snakes forage, and enter looser, gelatinous clouds formed by Zygnematales (Mougeotia, Spiro- gyra). The latter are avoided by snakes, but are poorer food for tadpoles, as Zygnematales are not epiphytized (Kupferberg et al., 1994). In a pond experiment, Turner and Mittelbach ( 1990) demonstrated that piscivore-induced habitat shifts in bluegill sunfish had substantial indirect effect on both zooplankton and phyto- plankton. At the scale of whole lakes, Carpen- ter et al. (1987) found strong effects on phyto- plankton and zooplankton of piscivore- induced habitat shifts in planktivorous fish. Complementing these experimental stud- ies, a number of theoretical studies show that adaptive, flexible behavior in predatodprey will have considerable effects on food web dynamics. Abrams (this volume) thoroughly reviews of the impact of adaptive foragers on the dynamics and interactions in food webs. In general, the introduction of flexible/ adaptive behavior will stabilize interactions. Gleeson and Wilson (1986) showed that an optimally foraging predator may promote co- existence between competing prey. Ives and Dobson (1987) showed that flexible habitat use in prey will enhance the stability of preda- tor-prey interactions by reducing predation rates at high predator densities, damping os- cillatory tendencies. Behavioral refuges have also been shown to stabilize host-parasite population dynamics in the same way as spa- tial refuges do (Mangel and Roitberg, 1992). Other adaptive defense responses are also likely to enhance stability of predator-prey interactions, especially if they are predator- induced rather than constitutive (Karban and Myers, 1989). Among such responses are morphological changes (e.g., in size or spine lengths), crypticity , chemical defenses, and changes in other characters aimed at reducing predation risk (Stenberger and Gilbert, 1987; Scrimshaw and Kerfoot, 1987; Karban and Myers, 1989). The Interactions between Energetic and Dynamic Constraints The amount of energy (or organic matter or nutrients) entering a system has, no doubt, major impacts on community and ecosystem processes. However, the effects of this extrin- sic control of productivity will be modulated by other factors which interact with produc- tivity to shape the ecological system. Above, we have advanced four factors which we sug- gest profoundly affect the dynamics of food webs. We suggest that food chain length and other basic food web properties are generally outcomes of the interaction between ener- getic and dynamic constraints (Figure 36.6). Spatial heterogeneity and flexible behavior (including defense) will generally have a sta- bilizing effect on the system. Size-structured interactions will affect stability, but how is yet incompletely understood. Heterogeneity and flexible behavior will often decrease en- ergy transfer up the food web, allowing diver- sion of more energy to detritus, for example, if refuges decrease predation. With more ref- uges for prey, the probability of predator ex- tinction will also increase, decreasing food chain length. Disturbance will negatively af- fect local stability but intermediate distur- bance levels may sometimes increase food chain length resulting in a unimodal relation- ship between food chain length and distur- bance (Power et al., this volume). Because heterogeneity and flexible behavior may both increase stability and decrease energy trans- fer, we hypothesize that food chain length, at a given productivity level, may also show a unimodal relationship with the degree of spatial heterogeneity and flexible behavior. Depending on how productivity interacts with these factors, it may also have hump- shaped relation with food chain length (cf. Persson et al. (1988, 1992)). Testing the gen- erality of hump-shaped relationships between food chain length and productivity, spatial heterogeneity, degree of flexible behavior and disturbance is an interesting topic for future research. Our hypothesis of unimodal relationships between food chain length and factors affect- ing dynamic constraints of food webs is re- lated to the commonly observed relationship between species diversity and productivity (Rosenzweig and Abramsky, 1993). Many studies on the relationship between produc- tivity and diversity are restricted to one func- tional group (grasses, rodents, etc.) (cf. Ro- senzweig and Abramsky (1993)) and do not address food chain length. If overall species diversity is positively related to food chain Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms 1 423 I Dynamics I # . .. Habitat Structure Population Disturbance Size Behavior, Fluctuations Structure Defense n Energy Flow Figure 36.6. Habitat heterogeneity (including refuges), disturbance and fluctuations, size structure and behavior, and defense are all features of organisms or habitats and their interaction that modulate the interaction of energy flow and trophic dynamics. length, however, we have provided several mechanisms (habitat heterogeneity, distur- bance, flexible behavior, size-structured in- teractions) which may produce hump-shaped relationships of food chain properties to pro- ductivity. Our review of different systems also suggests that the general pattern docu- mented by Rosenzweig and Abramsky (1993) for different systems (desert, marine, tropi- cal, and other systems) may arise from differ- ent mechanisms in different systems. When defining productivity, we argued that intrinsically controlled productivity was not independent of food web structure. Simi- larly, habitat heterogeneity, stresddistur- bance, flexible behavior, and sizekitage- structured dynamics are affected by food web dynamics. For example, primary producers form both food resources for herbivores and physical structure for many organisms lead- ing to a potential for consumer regulation of habitat heterogeneity (Power, 1990b, 1992b; Diehl, 1993; Persson, 1994). In river sys- tems, for example, if long turfs of algae de- velop, they detach to form extensive floating mats. These provide sun-warmed, food-rich incubators that are also partial refuges for residents from fish. Extensive floating mats, which do not develop unless algal growth outstrips grazing during the early growing season, probably greatly enhance secondary production of aquatic insects and tadpoles, as well as the fraction of that production that is exported from the river to the watershed (Power, 1990b). In lakes, fish may affect interactions between phytoplankton and mac- rophytes and hence the structural complexity of the system (Blindow et al., 1993; Diehl, 424 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power 1993; Persson, 1994). For example, plankti- vorous fish, by depressing zooplankton, may release phytoplankton from grazing leading to enhanced shading and die-offs of sub- merged vegetation. In terrestrial environ- ments, heavy grazing by voles may eliminate the shelter from predators provided by grass. Grazers which strip fouling epiphytes from macroalgal fronds reduce the probability that entire fronds will be tom off rocky shores by storm waves (D'Antonio, 1985). Large primary producers can also by virtue of their structure, modify disturbance regimes, as when trees encroach downstream of diver- sions on rivers, and immobilize the river bed (Carter, 1994). Although productivity, habitat heterogene- ity, disturbance, flexible behavior (including inedibility), and sizektage-structured dynam- ics are all likely to affect community and ecosystem dynamics to some extent in any system, it is likely that their relative impor- tance will vary between systems. Productiv- ity and heterogeneity may be the main factors to consider in lake and nontemporary pond systems. In contrast, disturbance should be important in streams, temporary ponds, and intertidal systems. Size-structured interac- tions are likely to matter in all aquatic sys- tems, but may be less critical in terrestrial systems dominated by mammal and bird spe- cies in which adults provide their young with food. Some variables like stress and primary production may often also be inversely corre- lated, and to a certain extent, interchange- able. For example, if energy that could be used for growth or progeny in a benign cli- mate must be diverted to maintenance (e.g., of body temperature) under stressful condi- tions, productivity and stress effects are linked by allocations by organisms of a com- mon energy currency. In the same vein, or- ganisms may be able to withstand more stress or longer periods of stress in productive habi- tats where they are not food-limited. Clarify- ing the relationships of such factors may help to reduce the number that must be understood to predict the dynamics of a specific system. We see such studies in diverse ecosystems as an essential area for future research. Acknowledgments Valuable comments on a previous draft of this chapter were given by Steve Carpenter, Dan Schindler, and Don Strong. The authors' research has been sponsored by the National Science Foundation (Bruce Menge and Mary Power), the Swedish Council for Forestry and Agricultural Research (Jan Bengtsson and Lennart Persson), NUTEK (Jan Bengtsson) and the Swedish Natural Research Council (Jan Bengtsson and Lennart Persson). The Swedish Natural Research Council also pro- vided traveling grants to Jan Bengtsson and Lennart Persson to attend the food web meet- ing in Pingree Park. Jan Bengtsson wrote his parts of the chapter while a visiting scientist at the Centre for Population Biology at Silwood Park, UK, and thanks to Jim Grover and John Lawton for comments and discussions. References As, S., J. Bengtsson, and T. Ebenhard. 1992. Archipelagoes and theories of insularity. In The Ecological Principles of Nature Conservation, ed. L. Hansson, Chapman & Hall, London. Abrams, P. A. 1988. Resource productivity-con- sumer species diversity: Simple models of com- petition in spatially heterogeneous environ- ments. Ecology 69:1418-1433. Abrams, P. A. 1993. Effects of increased produc- tivity on the abundandances of trophic levels. American Naturalist 141:351-371. Abrams, P. A. and J. Roth. 1994a. The effects of enrichment on three-species food chains with nonlinear functional responses. Ecology 75: 1 11 8-1 130. Abrams, P. A. and J. Roth. 1994b. The responses of unstable food chains to enrichment. Evolu- tionary Ecology 8:150-171. Abramsky, Z. 1988. The role of habitat and pro- ductivity in structuring desert rodent communi- ties. Oikos 52:107-114. Andersson, G., W. Graneli, and J. Stenson. 1988. The influence of animals on phosphorous cy- cling in lake ecosystems. Hydrobiologia 170:267-284. Angerbjorn, A. 1985. The evolution of body size in mammals on islands. Some comments. Amer- ican Naturalist 125:304-309. Arditi, R. and L. R. Ginzburg. 1989. Coupling in predator-prey dynamics: Ratio-dependence. Journal of Theoretical Biology 139:311-326. Amott, S. E. andM. J. Vanni. 1993. Zooplankton assemblages in fishless bog lakes: Influence of biotic and abiotic factors. Ecology 74:2361- 2380. Arp, A. J. and J. J. Childress. 1983. Sulfide bind- Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms I 425 ing by the blood of the hydrothermal vent tube work Riftia pachyptila. Science 2 19:295-297. Ballad, R. D. 1977. Notes on a major oceano- graphic find. Oceanus 20:35-44. Barko, J. W. andR. M. Smart. 1980. Mobilization of sediment phosphorous by submersed fresh- water macrophytes. Freshwater Biology Bartell, S. M., A. L. Brenkert, R. V. 0. O'Neill, and R. H. Gardner. 1988. Temporal variation in regulation of production in a pelagic food web model. In Complex Interactions in Lake Communities, eds. S. R. Carpenter, pp. 101- 118. Springer-Verlag, New York. Begon, M., J. L. Harper, and C. R. Townsend. 1986. Ecology-Individuals, Populations and Communities. Blackwell, Oxford, U.K. Bengtsson, J., D. W. Zheng, G. I. Agren, and T. Persson. 1995. Food webs in soil: An inter- face between population and ecosystem ecol- ogy. In Linking Species and Ecosystems. eds. C. G. Jones and J. H. Lawton, pp. 159-165. Chapman & Hall, New York. Berendse, F. and W. T. Elberse. 1990. Competi- tion and nutrient availability in heartland and grassland ecosystems. In Perspectives on Plant Competition, eds. J. B. Grace and D. Tilman, pp. 93-1 16. Academic Press, New York. Bird, G. A. and L. Chatarpaul. 1986. Effect of whole-tree and conventional harvest on soil mi- croarthropods. Canadian Journal of Zoology 64: 19861993. Birkeland, C. 1987. Nutrient availability as a ma- jor determinant of differences among coastal hard-substratum communities in different re- gions of the tropics. In Comparisons between Atlantic and PaciJic tropical coastal marine ecosystems: community structure, ecological processes, and productivity, Vol. 46, ed. C. Birkeland, pp. 45-97. UNESCO Reports in Marine Science, Rome. Birkeland, C. E. 1988. Geographic comparisons of coral-reef community processes. In Proceed- ings of the 6th International Coral Reef Sympo- sium, Vol. 1, pp. 211-220. Australia. Blindow, I., G. Andersson, A. Hargeby, and S. Johansson. 1993. Long-term pattern of alter- native stable states in two shallow eutrophic lakes. Freshwater Biology 30:159-167. Bosman, A. L., J. T. DuToit, P. A. R. Hockey, and G. M. Branch. 1986. A field experiment demonstrating the influence of seabird guano on intertidal primary production. Estuarine, Coastal and Shelf Science 23:283-294. Bosman, A. L. andP. A. R. Hockey. 1986. Sea- bird guano as a determinant of rocky intertidal 10~229-238. community structure. Marine Ecology Progress Series 32:247-257. Brabrand, A., B. A. Faafeng, and J. P. M. Nils- sen. 1990. Relative importance of phosphorus supply to phytoplankton production: Excretion versus external loading. Canadian Journal of Fisheries and Aquatic Sciences 47:36-372. Branch, B. M. and C. L. Griffiths. 1988. The Benguela ecosystem Part V. The coastal zone. Oceanographic and Marine Biology Annual Re- views 26:395-486. Briand, F. and J. E. Cohen. 1987. Environmental correlates of food chain length. Science 238:956-960. Carney, H. J. 1990. A general hypothesis for the strength of food web interactions in relation to trophic state. Verhandlungen der Internatio- nalen Vereinigung fur Theroretische und An- gewandte Limnologie 24:487-492. Carpenter, S. R. 1980. Enrichment of lake Win- gra, Wisconsin, by submersed macrophyte de- cay. Ecology 61:1145-1155. Carpenter, S. R. 1981. Submersed vegetation: An internal factor in lake ecosystem succession. American Naturalist 118:372-383. Carpenter, S. R., ed. 1988. Complex Interactions in Lake Communities, Springer-Verlag, New York. Carpenter, S. R. and J. F. Kitchell. 1984. Plankton community structure and limnetic primary pro- duction. American Naturalist 124: 159-172. Carpenter, S. R. and J. F. Kitchell. 1987. The temporal scale of variance of limnetic primary production. American Naturalist 129:417-433. Carpenter, S. R. and J. F. Kitchell. 1988. Strong manipulations and complex interactions: Con- sumer control of lake productivity. Bioscience 38:764-769. Carpenter, S. R. and J. F. Kitchell, eds. 1993. The Trophic Cascade in Lakes. Cambridge Uni- versity Press, Cambridge, U.K. Carpenter, S. C., J. F. Kitchell, and J. R. Hodg- son. 1985. Cascading trophic interactions and lake productivity. Bioscience 35:634639. Carpenter, S. R., J. F. Kitchell, J. R. Hodgson, P. A. Cochran, J. J. Elser, M. M. Elser, D. M. Lodge, D. Kretchmer, X. He, andC. N. Ende. 1987. Regulation of lake primary productivity by food web structure. Ecology 68:1863-1876. Carter, W. C. 1994. Woodland expansion in the Platte River, Nebraska: Patterns and causes. Ecological Monographs 64:45-84. Caughley, G. and J. H. Lawton. 1981. Plant- herbivore systems. In Theoretical Ecology, 2nd Ed., ed. R. M. May, pp. 132-166. Blackwell, Oxford, U.K. 426 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power Cohen, J. E., F. Briand, and C. M. Newman. 1990. Community Food Webs: Data and The- ory. Springer-Verlag, New York. Connell, J. H. 1975. Some mechanisms producing structure in natural communities: A model and evidence from field experiments. In Ecology and Evolution of Communities, eds. M. L. Cody and J. M. Diamond, pp. 460-490. Belknap Press, Cambridge, MA. Connell, J. H. 1978. Diversity in tropical rain forests and coral reefs. Science 199: 1302-1310. Corliss, J. B., J. Dymond, L. I. Gordon, J. M. Edmond, R. P. von Herzen, R. D. Ballard, K. Green, D. Williams, A. Bainbridge, K. Crane, and T. H. van Andel. 1979. Subma- rine thermal springs on the Galapagos rift. Sci- ence 203: 1073-1082. Cousins, S. 1987. The decline of the trophic level concept. Trends in Ecology and Evolution Crawley, M. J. 1992. Population dynamics of natural enemies and their prey. In Natural Ene- mies, ed. M. J. Crawley, pp. 40-89. Blackwell, Oxford, U.K. D'Antonio, C. 1985. Epiphytes on the rocky inter- tidal red alga Rhodomela lark (Turner) C. Agardh: negative effects on the host and food for herbivores? Journal of Experimental Marine Biology and Ecology 86: 197-2 18. Darnell, R. M. 1961. Trophic spectrum of an estuarine community based upon studies of Lake Ponchartrain, Louisiana. Ecology 42553- 568. David, J.-F., J.-F. Ponge, P. Arpin, and G. Van- nier. 1991. Reactions of the macrofauna of a forest mull to experimental perturbations of lit- ter supply. Oikos 61:316-326. Dayton, P. K. 1971. Competition, disturbance, and community organization: The provision and subsequent utilization of space in a rocky inter- tidal community. Ecological Monographs Dayton, P. K. 1975. Experimental evaluation of ecological dominance in a rocky intertidal algal community. Ecological Monographs 45: 137- 159. Dayton, P. K. 1984. Processes structuring some marine communities: Are they general? In Eco- logical Communities: Conceptuallssues and the Evidence, eds. D. R. Strong, Jr., D. Simberloff, L. G. Abele, and A. B. Thistle, pp. 181-197. Princeton University Press, Princeton, NJ. DeAngelis, D. A. 1992. Dynamics of Nutrient Cycling and Food Webs. Chapman & Hall, London. 2: 3 12-3 16. 41:351-389. DeAngelis, D. L., S. M. Bartell, and A. L. Bren- kert. 1989a. Effects of nutrient recycling and food chain length on resilience. American Natu- ralist 134:778-805. DeAngelis, D. L., R. A. Goldstein, and R. V. O'Neill. 1975. A model for trophic interaction. Ecology 56:881-892. DeAngelis, D. L. and L. J. Gross, eds. 1992. Individual-based Models and Approaches in Ecology-Populations, Communities and Eco- systems. Chapman & Hall, New York. DeAngelis, D. L., P. J. Mulholland, A. V. Pa- lumbo, A. D. Steinman, M. A. Huston, and J. Elwood. 1989b. Nutrient dynamics and food- web stability. Annual Review of Ecology and Systematics 20:7 1-95. DeRoos, A. M., J. A. J. Metz, and0. Diekmann. 1992. Studying the dynamics of structured pop- ulation models: A versatile technique and its application to Daphnia. American Naturalist DeVries, D. R. and R. A. Stein. 1992. Complex interactions between fish and zooplankton: Quantifying the role of an open-water plankti- vore. Canadian Journal of Fisheries and Aquatic Sciences 49: 12 16- 1227. Diehl, S. 1992. Fish predation and benthic com- munity structure: The role of omnivory and hab- itat complexity. Ecology 73: 1646-1661. Diehl, S. 1993. Relative consumer sizes and the strengths of direct and indirect interactions in omnivorous feeding relationships. Oikos 68: Dillon, P. J. and F. H. Rigler. 1974. The phospho- rous-chlorophyll relationships in lakes. Limnol- ogy and Oceanography 19:767-773. Duggins, D. O., C. A. Simenstad, and J. A. Estes. 1989. Magnification of secondary production by kelp detritus in coastal marine ecosystems. Science 245:170-173. Dunson, W. A. and J Travis. 1991. The role of abiotic factors in community organization. American Naturalist 138:1067-1091. Ebenman, B. and L. Persson, eds. 1988. Size- Structured Populations-Ecology and Evolu- tion. Springer-Verlag, Berlin. Ehrlich, P. R. and L. C. Birch. 1967. The "balance of nature" and "population control ." American Naturalist 101:97-107. Elser, J. J., H. E. Carney, and C. R. Goldman. 1990. The zooplankton-phytoplankton interface in lakes of contrasting trophic status: an experi- mental comparison. Hydrobiologia 2001 Elser, J. J., M. M. Elser, N. MacKay, and S. R. 139: 123- 147. 151-157. 201 169-82. Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms I 427 Carpenter. 1988. Zooplankton-mediated transi- tions between N- and P-limited algal growth. Limnology and Oceanography 33: 1-14. Elton, C. 1927. Animal Ecology. Sidgwick and Jackson, London. Estes, J. A. and D. 0. Duggins. 1995. Sea otters and kelp forests in Alaska: Generality and varia- tion in a community ecological paradigm. Eco- logical Monographs 65:75-100. Estes, J. A., N. S. Smith, and J. F. Palmisano. 1978. Sea otter predation and community orga- nization in the western Aleutian Islands, Alaska. Ecology 59:822-833. Fittkau, E. J. 1970. Role of caimans in the nutrient regime of mouth-lakes of Amazon affluents: A hypothesis. Biotropica 2: 138-142. Foster, M. S. 1990. Organization of macroalgal assemblages in the Northeast Pacific: The as- sumption of homogeneity and the illusion of generality. Hydrobiologia 192:21-33. Fretwell, S. D. 1977. The regulation of plant com- munities by food chain exploring them. Per- spectives in Biology and Medicine 20: 169-1 85. Fretwell, S. D. 1987. Food chain dynamics: The central theory of ecology? Oikos 50:291-301. Gleeson, S. K. and D. S. Wilson. 1986. Equilib- rium diet: Optimal foraging and prey coexis- tence. Oikos 46:139-144. Gliwicz, Z. M. and W. Lampert. 1990. Food thresholds in Daphnia species in the absence and presence of blue-green filaments. Ecology Glynn, P. W. 1988. El Niiio-Southern oscillation 1982-83: Nearshore population, community and ecosystem responses. Annual Review of Ecology and Systematics 19:309-345. Grassle, J. F. 1985. Hydrothermal vent animals: Distribution and biology. Science 229:7 13- 725. Grimm, N. B. and S. G. Fisher. 1986. Nitrogen limitation in a sonoran desert stream. Journal of the North American Benthological Society Grimm, N. B. and S. G. Fisher. 1989. Stability of periphyton and macroinvertebrates to distur- bance by flash floods in a desert stream. Journal of the North American Benthological Society 8:293-307. Hairston, N. G. Sr. 1985. The interpretation of experiments in interspecific competition. Amer- ican Naturalist 125:321-325. Hairston, N. G. Jr. and N. G. Hairston Sr. 1993. Cause-effect relationships in energy flow, tro- phic structure, and interspecific interactions. American Naturalist 142:379-411. 7 1 :69 1-702. 5:2-15. Hairston, N. G., F. E. Smith, and L. B. Slobod- kin. 1960. Community structure, population control, and competition. American Naturalist 94:421-425. Hall, C. A. S. 1972. Migration and metabolism in a temperate stream ecosystem. Ecology 53585- 604. Hamrin, S. F. and L. Persson. 1986. Asymmetri- cal competition between age classes as a factor causing population oscillations in an obligate planktivorous fish species. Oikos 47:223-232. Hanson, J. M. and W. C. Leggett. 1982. Empiri- cal prediction of fish biomass and yield. Cana- dian Journal of Fisheries and Aquatic Sciences Hanson, J. M. and R. H. Peters. 1984. Empirical prediction of crustacean zooplankton biomass and profunda1 macrobenthos biomass in lakes. Canadian Journal of Fisheries and Aquatic Sci- ences 41:439-445. Hansson, L.-A,, L. Johansson, and L. Persson. 1987. Effects of fish grazing on nutrient release and succession of primary producers. Limnol- ogy and Oceanography 32:723-729. Hart, D. D. 1985. Causes and consequences of territoriality in a grazing stream insects. Ecol- ogy 66:404414. Hart, D. D. and C. T. Robinson. 1990. Resource limitation in a stream community: Effects of phosphorus enrichment on periphyton and graz- ers. Ecology 71:14941502. Hassell, M. P. 1981. Arthropod predator prey cycles. In Theoretical Ecology, 2nd Ed., ed. R. M. May, pp. 105-131. Blackwell, Oxford, U.K. Hawkins, B. A. 1993. Parasitoid-host food webs and donor-control. Oikos 65:159-162. Heal, 0. W. and J. Dighton. 1985. Resource qual- ity and trophic structure in the soil system. In Ecological Interactions in Soil, eds. A. H. Fit- ter, D. Atkinson, D. J. Read, and M. B. Usher, pp. 339-354. Blackwell, Oxford, U.K. Heaney, L. R. 1984. Mammalian species richness on islands on the Sunda shelf, Southeast Asia. Oecologia 61:ll-17. Hershey, A. E. 1987. Tubes andforagingbehavior in larval Chironomidae: Implications for preda- tor avoidance. Oecologia 73:236-241. Hessler, R. R., W. M. Smithey, M. A. Boudrias, C. H. Keller, R. A. Lutz, and J. J. Childress. 1988. Temporal changes in the megafauna at the Rose Garden hydrothermal vent (Galapagos rift: eastern tropical Pacific). Deep-sea Re- search 35: 168 1-1709. Hixon, M. A. and B. A. Menge. 1991. Species 39:257-263. 428 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power diversity: Prey refuges modify the interactive effects of predation and competition, Theoreti- cal Population Biology 39: 178-200. Hobbie, S. E. 1992. Effects of plant species on nutrient cycling. Trends in Ecology and Evolu- tion 7:336-339. Holt, R. D. 1985. Population dynamics of two- patch environments: some anomalous conse- quences of an optimal habitat distribution. The- oretical Population. Biology 28: 18 1-208. Hrbicek, J., M. Dvorakova, V. Korinek, and L. Prochazkova. 1961. Demonstration of the effects of fish stock on the species composition and the intensity of the metabolism of the whole plankton association. Verhandlungen Interna- tionale Verhandlungen der Internationalen Ver- einigung fur Theroretische und Angewandte Limnologie 14:192-195. Hunter, M. D. and P. W. Price. 1992. Playing chutes and ladders. Bottom-up and top-down forces in natural communities. Ecology 73:724- 732. Hutchinson, G. E. 1959. Homage to Sant Rosalia or why are there so many kinds of animals. American Naturalist 93: 145-159. Ives, A. R. and A. P. Dobson. 1987. Antipredator behaviour and the population dynamics of sim- ple predator-prey systems. American Naturalist l30:43 1-447. Canadian Journal of Fisheries and Aquatic Sci- ences 41:2-19. Leibold, M. A. 1989. Resource edibility and the effects of predators and productivity on the out- come of trophic interactions. American Natural- ist 134:922-949. Leibold, M. A. and H. M. Wilbur. 1992. Interac- tions between food-web structure and nutrients on pond organisms. Nature 360:341-343. Lewin, J. and C. T. Schaefer. 1983. The role of phytoplankton in surf ecosystems. In Sandy Beaches as Ecosystems, eds. A. McLachlan and T. Erasmus, pp. 133-144. JunkPublishers, The Hague, The Netherlands. Lindeman, R. L. 1942. The trophic-dynamic as- pect of ecology. Ecology 23:399-418. Locke, A. 1992. Factors influencing community structure along stress gradients: Zooplankton responses to acidification. Ecology 73:903- 909. Lodge, D. M., J. W. Barko, D. Strayer, J. M. Melack, G. G. Mittelbach, R. W. Howarth, B. Menge, and J. E. Titus. 1988. Spatial heter- ogeneity and habitat interactions in lake com- munities. In Complex Interactions in Lake Com- munities, ed. S. R. Carpenter, pp. 183-208. Springer-Verlag, New York. Louda, S. M. and S. K. Collinge. 1992. Plant resistance to insect herbivores: A field test of the environmental stress hypothesis. Ecology 73:153-169. Lundkvist, H. 1983. Effects of clear-cutting on the enchytraeids in a Scots pine forest soil in central Sweden. Journal of Applied Ecology 20:873-885. Mangel, M. andB. D. Roitberg. 1992. Behavioral stabilization of host-parasite population dynam- ics. Theoretical Population Biology 42:308- 320. Jenkins, B., R. L. Kitching, and S. L. Pimm. 1992. Productivity, disturbance and food web stmcture at a local spatial scale in experimental container habitats. Oikos 65249-255. Johnston, N. T., C. J. Penin, P. A. Slaney, and B. R. Ward. 1990. Increased juvenile salmonid growth by whole-river fertilization. Canadian Journal of Fisheries and Aquatic Sciences 47 :862-872. Karban, R. and J. H. Myers. 1989. Induced plant responses to herbivory. Annual Review of Ecol- ogy and Systematics 20:33 1-348. Karl, D. M., C. 0. Wirsen, andH. W. Jannasch. 1980. Deep-sea primary production at the Gala- pagos hydrothermal vents. Science 207: 1345- 1347. Kupferberg, S. J., J. C. Marks, and M. E. Power. 1994. Effects of variation in natural algal and detrital diets on larval anuran (Hyla regilla) life history traits. Copeia 2:446457. Lawton, J. H. 1989. Food webs. In Ecological Concepts, ed. J. M. Cherrett, pp. 43-78. Black- well, Oxford, U.K. Legendre, L. and S. Demers. 1984. Towards dy- namic biological oceanography and limnology. Matsuda, H., K. Kawasaki, N. Shigesada, E. Ter- amoto, and L. M. Ricciardi. 1986. Switching effect on the stability of the prey-predator sys- tem with three trophic levels. Journal of Theo- retical Biology 122:251-262. McAuliffe, J. R. 1984. Competition for space, disturbance, and the structure of a benthic stream community. Ecology 65:894-908. McCauley, E. and J. Kalff. 1981. Empirical rela- tionships between phytoplankton and zooplank- ton biomass in lakes. Canadian Journal of Fish- eries and Aquatic Sciences 38:458-463. McCauley, E. and W. W. Murdoch. 1987. Cyclic and stable populations: Plankton as a paradigm. American Naturalist 129:97- 12 1. McCauley, E., W. W. Murdoch, and S. Watson. Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms I 429 1988. Simple models and variation in plankton densities among lakes. American Naturalist 132:383403. McClanahan, T. R. 1992. Epibenthic gastropods of the Middle Florida Keys: The role of habitat and environmental stress on assemblage compo- sition. Journal of Experimental Marine Biology and Ecology 160:169-190. McLachlan, A. 1990. Dissipative beaches and macrofauna communities on exposed intertidal sands. Journal of Coastal Research 657-71. McNaughton, S. J., M. Oesterheld, D. A. Frank, and K. J. Williams. 1989. Ecosystem-level pat- terns of primary productivity and herbivory in terrestrial habitats. Nature 341:142-144. McNaughton, S. J., R. W. Ruess, and S. W. Seagle. 1988. Large mammals and process dy- namics in African ecosystems. Bioscience 38:794-800. McQueen, D. J., M. R. S. Johannes, and J. R. Post. 1989. Bottom-up and top-down impacts on freshwater pelagic community structure. Ecological Monographs 59:289-309. McQueen, D. J., J. R. Post, and E. L. Mills. 1986. Trophic relationships in freshwater pe- lagic ecosystems. Canadian Journal of Fisher- ies and Aquatic Sciences 43:1571-1581. Menge, B. A. 1976. Organization of the New England rocky intertidal community: Role of predation, competition, and environmental het- erogeneity. Ecological Monographs 46:355- 393. Menge, B. A. 1992. Community regulation: Un- der what conditions are bottom-up forces impor- tant on rocky shores? Ecology 73:755-765. Menge, B. A. 1995. Indirect effects in marine rocky intertidal interaction webs: Patterns and importance. Ecological Monographs 65:21-74. Menge, B. A. andT. M. Farrell. 1989. Commu- nity structure and interaction webs in shallow marine hard-bottom communities: Test of an environmental stress model. Advances in Eco- logical Research 19: 189-262. Menge, B. A. and J. Lubchenco. 1981. Commu- nity organization in temperate and tropical rocky intertidal habitats: Prey refuges in relation to consumer pressure gradients. Ecological Monographs 5 1 :429-450. Menge, B. A,, J. Lubchenco, L. R. Ashkenas, and F. Ramsey. 1986a. Experimental separation of effects of consumers on sessile prey in the low zone of a rocky shore in the Bay of Panama: Direct and indirect consequences of food web complexity. Journal of Experimental Marine Biology and Ecology 100:225-269. Menge, B. A., J. Lubchenco, S. D. Gaines, and L. R. Ashkenas. 1986b. A test of the Menge- Sutherland mode of community organization in a tropical rocky intertidal food web. Oecologia Menge, B. A. and A. M. Olson. 1990. Role of scale and environmental factors in regulation of community structure. Trends in Ecology and Evolution 552-57. Menge, B. A. and J. P. Sutherland. 1976. Species diversity gradients: Synthesis of the roles of predation, competition, and temporal heteroge- neity. American Naturalist 110:35 1-369. Menge, B. A. and J. P. Sutherland. 1987. Com- munity regulation: Variation in disturbance, competition, and predation in relation to envi- ronmental stress and recruitment. American Naturalist 130:730-757. Metz, J. A. J. and 0. Diekmann, eds. 1986. The Dynamics of Physiologically Structured Popu- lations. Springer-Verlag , Berlin. Metz, J. A. J., A. M. DeRoos, and F. van den Bosch. 1988. Population models incorporating physiological structure: A quick survey of the basic concepts and an application to size-struc- tured population dynamics in waterlleas. In Size-Structured Populations-Ecology and Evolution, eds. B. Ebenman and L. Person, pp. 106-126. Springer-Verlag, Berlin. Meyer, J. L., W. H. McDowell, T. L. Bott, J. W. Elwood, C. Ishizaki, J. M. Melack, B. L. Peck- arsky, B. J. Peterson, and P. A. Rublee. 1988. Elemental dynamics in streams. Journal of the North American Benthological Society 7:410- 432. Mills, E. L. and A. Schiavone. 1982. Evaluation of fish communities through assessment of zoo- plankton populations and measures of lake pro- ductivities. North American Journal of Fisher- ies Management 2:14-27. Mittelbach, G. G. 1988. Competition among ref- uging sunfishes, and effects of fish density on littoral zone invertebrates. Ecology 69:6l4- 623. Mittelbach, G. G. and P. L. Chesson. 1987. Pre- dation risk: Indirect effects on fish populations. In Predation: Direct and Indirect Impacts on Aquatic Communities, eds. W. C. Kerfoot and A. Sih, pp. 315-332. University Press of New England, Hanover. Mittelbach, G. G. and C. W. Osenberg. 1993. Stage-structured interactions in bluegill: Conse- quences of adult resource variation. Ecology Mittelbach, G. G., C. W. Osenberg, and M. A. Leibold. 1988. Trophic relations and ontoge- 71 175-89. 74~238 1-2394. 430 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power netic niche shifts in aquatic ecosystems. In Size- Structured Populations-Ecology and Evolu- tion, eds. B. Ebenman andL. Persson, pp. 315- 332. Springer-Verlag, Berlin. Moen, J. and L. Oksanen. 1991. Ecosystem trends. Nature 353:5 IO. Moore, J. C. and P. C. deRuiter, 1991. Temporal and spatial heterogeneity of trophic interactions within below-ground food webs. Agriculture, Ecosystems and Environment 34:371-397. Moore, J. C., P. C. de Ruiter, and H. W. Hunt. 1993. Influence of productivity on the stability of real and model ecosystems. Science 26 1 :9O6-909. Moore, J. C. and H. W. Hunt. 1988. Resource compartmentation and the stability of real eco- systems. Nature 314:261-263. Murdoch, W. W. 1966. "Community structure, population control, and competition"-A cri- tique. American Naturalist 100:219-226. Murdoch, W. W. and E. McCauley. 1985. Three distinct types of dynamic behaviour shown by a single planktonic system. Nature 326:628- 630. Neill, W. E. 1988. Complex interactions in oligo- trophic lake food webs: Responses to nutrient enrichment. In Complex Interactions in Lake Communities, ed. S. R. Carpenter, pp. 31-65. Springer-Verlag, New York. Neil], W. E. and A. Peacock. 1980. Breaking the bottleneck: Interactions between nutrients and invertebrate predators in oligotrophic lakes. In Evolution and Ecology of Freshwater Zooplank- ton Communities, ed. W. C. Kerfoot, pp. 175- 724. University Press of New England, Hanover. Newbold, J. D., J. W. Elwood, R. V. O'Neill, and A. L. Sheldon. 1983. Phosphorus dynamics in a woodland stream ecosystem: A study of nutrient spiralling. Ecology 64:1249-1265. Newbold, J. D., J. W. Elwood, R. V. O'Neill, and W. van Winkle. 1981. Measuring nutrient spiralling in streams. Canadian Journal of Fish- eries and Aquatic Sciences 38:86&863. Newbold, J. D., R. V. O'Neill, J. W. Elwood, and W. van Winkle. 1982. Nutrient spiralling in streams: Implications for nutrient limitation and invertebrate activity. American Naturalist 120:628-652. Nielsen, G. A. andF. D. Hole. 1964. Earthworms and the development of coprogenous A1 hori- zons in forest soils of Wisconsin. Soil Science Society of America Proceedings 28:426-430. Norberg, R. A. 1988. Self-thinning of plant popu- lations dictated by packing density and individ- ual growth geometry and relationships between animal population density and body mass gov- erned by metabolic rate. In Size-Structured Populations-Ecology and Evolution, eds. B. Ebenman and L. Persson, pp. 259-279. Springer-Verlag, Berlin. Northcote, T. G. 1988. Fish in the structure and function of freshwater ecosystems: A "top- down" view. Canadian Journal of Fisheries and Aquatic Sciences 45:361-379. Odum, E. P. 1969. The strategy of ecosystem development. Science 164:262-2790. Oksanen, L. 1983. Trophic exploitation and arctic phytobiomass patterns. American Naturalist Oksanen, L. 1988. Ecosystem organization: Mu- tualism and cybernetics or plain Darwinian struggle for existence? American Naturalist 13 1:424-444. Oksanen, L. 1990a. Exploitation ecosysetms in seasonal environments. Oikos 57: 1624. Oksanen, L. 1990b. Predation, herbivory and plant strategies along gradients of primary pro- ductivity. In Perspectives on Plant Competi- tion, eds. J. B. Grace and D. Tilman, pp. 445- 474. Academic Press, San Diego, CA. Oksanen, L., S. D. Fretwell, J. Anuda, and P. Niemela. 1981. Exploitation ecosystems in gradients of primary productivity. American Naturalist 11 8:240-261. Oksanen, T. 1990. Exploitation ecosystems in het- erogeneous habitat complexes. Evolutionary Ecology 4:220-234. Oksanen, T., M. E. Power, and L. Oksanen. in press. Habitat selection and predator resource dynamics. American Naturalist (in press). Paine, R. T. 1980. Food webs: Linkage, interac- tion strength and community infrastructure. Journal of Animal Ecology 49:667-685. Paine, R. T. 1988. Food webs: Road maps of interactions or grist for theoretical develop- ment? Ecology 69: 1648-1654. Pastor, J. andNaiman, R. J. 1992. Selective forag- ing and ecosystem processes in boreal forests. American Naturalist 1391690-705. Perrin, C. J., J. L. Bothwell, and P. A. Slaney. 1987. Experimental enrichment of a coastal stream in British Columbia: Effects of organic and inorganic additions on autotrophic periphy- ton production. Canadian Journal of Fisheries and Aquatic Sciences 44: 1247-1256. Persson, L. 1983. Food consumption and the sig- nificance of detritus and algae to intraspecific competition in roach Rutilus rutilus in a shallow eutrophic lake. Oikos 4 1 : 1 18- 125. 122:42-52. Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms I 431 Persson, L. 1988. Asymmetries in competitive and predatory interactions in fish populations. In Size-Structured Populations-Ecology and Evolution, eds. B. Ebenman and L. Persson, pp. 203-218. Springer-Verlag, Berlin. Persson, L. 1993. Predator-mediated competition in prey refuges: The importance of habitat de- pendent prey resources. Oikos 68: 12-22. Persson, L. 1994. Natural patterns of shifts in fish communities-mechanisms and constraints on perturbation sustenance. In Rehabilitation of FreshwaterFisheries, ed. I. G. Cowx, pp. 421- 434. Fishing News Books, Blackwell, Oxford, U.K. Persson, L., G. Andersson, S. F. Hamrin, and L. Johansson. 1988. Predator regulation and primary production along the productivity gra- dient of temperate lake ecosystems. In Complex Interactions in Lake Communities, ed. S. R. Carpenter, pp. 45-68. Springer-Verlag, New York. Persson, L., S. Diehl, L. Johansson, G. Anders- son, and S. F. Hamrin. 1991. Shifts in fish communities along the productivity gradient of temperate lakes-patterns and the importance of size-structured interactions. Journal of Fish Biology 38:28 1-293. Persson, L., S. Diehl, L. Johansson, G. Anders- son, and S. F. Hamrin. 1992. Trophic interac- tions in temperate lake ecosystems: A test of food chain theory. American Naturalist 14059- 84. Persson, L. and P. EMOv. 1995. Prey refuges affecting interactions between piscivorous perch (Perca Juviatilis) and juvenile perch and roach (Rutilus rutilus). Ecology 76:70-81. Persson, L., L. Johansson, G. Andersson, S. Diehl, and S. F. Hamrin. 1993. Density depen- dent interactions in lake ecosystems-Whole lake perturbation experiments. Oikos 66: 193- 208. Peterson, B. J., L. Deegan, J. Helfrich, J. E. Hobbie, M. Hullar, B. Moller, T. Ford, A. Hershey, A. Hiltner, G. Kipphut, M. A. Lock, D. M. Fiebig, V. McKinley, M. C. Miller, J. R. Vestal, R. Ventullo, andG. Volk. 1993. Biological responses of a tundra river to fertilization. Ecology 74:653-672. Peterson, C. G. and R. J. Stevenson. 1992. Resis- tance and resilience of lotic algal communities: Importance of disturbance timing and current. Ecology 73:1445-1461. Peterson, C. H. 1979. Predation, competitive ex- clusion, and diversity in the soft-sediment ben- thic communities of estuaries and lagoons. In Ecological Processes in Coastal and Marine Systems, ed. R. J. Livingston, pp. 233-264. Plenum, New York. Pimm, S. L. 1982. Food Webs. Chapman &Hall, London. Pimm, S. L. 1991. The Balance of Nature? The University of Chicago Press, Chicago. Pimm, S. L. and R. L. Kitching. 1987. The deter- minants of food chain length. Oikos 50:302- 307. Pimm, S. L. and J. H. Lawton. 1978. On feeding on more than one trophic level. Nature Polis, G. A. 1991. Complex trophic interactions in deserts: An empirical critique of food-web theory. American Naturalist 138: 123-155. Polis, G. A. 1994. Food webs, trophic cascades and community structure. Australian Journal of Polis, G. A. and R. D. Holt. 1992. Intraguild predation: The dynamics of complex trophic interactions. Trends in Ecology and Evolution 7: 151-154. Polis, G. A., C. A. Myers, and R. D. Holt. 1989. The ecology and evolution of intraguild preda- tion: Potential competitors that eat each other. Annual Review of Ecology and Systematics Pomeroy, L. R. 1991. Relationships of primary and secondary production in lakes and marine ecosystems. In Comparative Analyses of Eco- systems: Patterns, Mechanisms, and Theories, eds. J. Cole, G. Lovett, and S. Findlay, pp. 97-1 19. Springer-Verlag, New York. Porter, K. G., H. Paerl, R. Hodson, M. Pace, J. Priscu, B. Rieman, D. Scavia, and J. Stockner. 1988. Microbial interactions in lake communities. In Complex Interactions in Lake Communities, ed. S. R. Carpenter, pp. 209-227. Springer-Verlag, New York. Power, M. E. 1984a. Depth distributions of ar- mored catfish: Predator-induced resource avoid- ance? Ecology 65:523-528. Power, M. E. 1984b. Habitat quality and the distri- bution of algae-grazing catfish in a Panamanian stream. Journal of Animal Ecology 53:357- 374. Power, M. E. 1990a. Benthic turfs versus floating mats of algae in river food webs. Oikos 58:67- 79. Power, M. E. 1990b. Effects of fish in river food webs. Science 250:811-814. Power, M. E. 1992a. Habitat heterogeneity and the functional significance of fish in river food webs. Ecology 73:1675-1688. 2751542-544. Ecology 19: 12 1- 136. 20~297-330. 432 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power Power, M. E. 1992b. Hydrologic and trophic con- trols of seasonal algal blooms in northern Cali- fornia rivers. Archive fur Hydrobiologie 125:385410. Power, M. E. 1992c. Top-down or bottom-up forces in food webs: Do plants have primacy? Power, M. E., T. L. Dudley, and S. D. Cooper. 1989. Grazing catfish, fishing birds, and attached algae in a Panamanian stream. Envi- ronmental Biology of Fishes 26:285-294. Power, M. E., J. C. Marks, and M. S. Parker. 1992. Variation in the vulnerability of prey to different predators: Community-level conse- quences. Ecology 73:2218-2223. Power, M. E., W. J. Matthews, and A. J. Stewart. 1985. Grazing minnows, piscivorous bass and stream algae: Dynamics of a strong interaction. Power, M. E. and A. J. Stewart. 1987. Distur- bance and recovery on an algal assemblage fol- lowing flooding in an Oklahoma stream. Ameri- can Midland Naturalist 117:333-345. Rex, M. A. 1973. Deep-sea species diversity: De- creased gastropod diversity at abyssal depths. Science 18 1 : 105 1-1053. Rex, M. A. 1976. Biological accommodation in the deep-sea benthos: Comparative evidence on the importance of predation and productivity. Deep-sea Research 23:975-987. Rex, M. A. 1983. Geographic patterns of species diversity in the deep-sea benthos. In Deep-sea Biology, ed. G. T. Rowe, pp. 453-472. John Wiley and Sons, Inc., New York. Rosenzweig, M. L. 1968. Net primary productiv- ity of terrestrial communities: Prediction from climatological data. American Naturalist Rosenzweig, M. L. 1971. Paradox of enrichment: Destabilization of exploitation ecosystems in ecological time. Science 171:385-387. Rosenzweig, M. L. 1973. Exploitation in three trophic levels. American Naturalist 107:275- 294. Rosenzweig, M. L. and Z. Abramsky. 1993. How are diversity and productivity related? In Spe- cies Divers@ in Ecological Communities, eds. R. R. Ricklefs and D. Schluter, pp. 52-65. The University of Chicago Press, Chicago. Sand-Jensen, K. 1979. Balancen mellan autotrofe komponenter i tempererade s$er med forskellig naeringsbelastning. Vatten 2:104-115 (in Dan- ish with English summary). Santos, P. F. and W. G. Whitford. 1981. The effects of microarthropods on litter decomposi- Ecology 73 ~733-746. Ecology 66: 1448-1456. 102~67-74. tion in a Chihuahuan desert ecosystem. Ecology 62:654-663. Samelle, 0. 1992. Nutrient enrichment and grazer effects on phytoplankton in lakes. Ecology Samelle, 0. 1993. Herbivore effects on phyto- plankton succession in a eutrophic lake. Ecolog- ical Monographs 63:129-149. Schindler, D. W. 1978. Factors regulating phyto- plankton production and standing crop in the world's freshwaters. Limnology and Oceanog- raphy 23:478436. Schmitz, 0. J. 1993. Trophic exploitation in grass- land food chains: Simple models and a field experiment. Oecologia 93:327-335. Schoener, T. W. 1973. Population growth regu- lated by intraspecific competition for energy or time: Some simple representations. Theoretical Population Biology 456-84. Schoener, T. W. 1989. Food webs from the small to the large. Ecology 70:1559-1589. Scrimshaw, S. and W. C. Kerfoot. 1987. Chemi- cal defenses of freshwater organisms: Beetles and buds. In Predation: Direct and Indirect Impacts on Aquatic Communities, eds. W. C. Kerfoot and A. Sih, pp. 240-262. Univeresity Press of New England, Hanover, N.H. Sedell, J. R. and J. L. Froggatt. 1984. Importance of streamside forests to large rivers: The isola- tion of the Willamette River, Oregon USA from its floodplain by snagging and streamside forest removal. Verhandlungen der Internationalen Vereinigung fur Theroretische und Angewandte Limnologie 22: 1828- 1834. Short, A. D. and L. D. Wright. 1983. Physical variability of sandy beaches. In Sandy Beaches as Ecosystems, eds. A. McLachlan and T. Eras- mus, pp. 381-389. Junk Publishers, The Hague, The Netherlands. Simenstad, C. A., J. A. Estes, andK. W. Kenyon. 1978. Aleuts, sea otters, and alternate stable- state communities. Science 200:403-411. Slobodkin, L. B. 1960. Ecological energy rela- tionships at the population level. American Nat- uralist 94:213-215. Slobodkin, L. B. 1962. Energy in animal ecology. In Advances in Ecological Research, Vol. 1. ed. J. B. Cragg, pp. 69-101. Academic Press, London. Slobodkin, L. B., F. E. Smith, and N. G. Hair- ston. 1967. Regulation in terrestrial ecosys- tems, and the implied balance of nature. Ameri- can Naturalist 101 : 109-124. Sohlenius, B. 1990. Influence of cropping system and nitrogen input on soil fauna and microor- 73:551-560. Productivity and Consumer Regulation-Concepts, Patterns, and Mechanisms 1 433 ganisms in a Swedish arable soil. Biology and Fertility of Soils 9: 168- 173. Sousa, W. P. 1984. The role of disturbance in natural communities. Annual Review of Ecology and Systematics 15:353-391. Sousa, W. P. 1985. Disturbance and patch dynam- ics on rocky intertidal shores. In The Ecology of Natural Disturbance and Patch Dynamics, eds. S. T. A. Pickett and P. S. White, pp. 101- 124. Academic Press, Orlando, FL. Spiller, D. A. andT. W. Schoener. 1990. A terres- trial field experiment showing the impact of eliminating top predators on foliage damage. Nature 347:469-472. Sprules, W. G. and J. E. Bowennan. 1988. Omni- vary and food chain length in zooplankton food webs. Ecology 69:418-426. Stenberger, R. S. and J. J. Gilbert. 1987. Defenses of planktonic rotifers against predators. In Pre- dation: Direct and Indirect Impacts on Aquatic Communities, eds. W. C. Kerfoot and A. Sih, pp. 227-239. University Press of New England, Hanover. Stenseth, N. C. 1980. Spatial heterogeneity and population stability: Some evolutionary conse- quences. Oikos 35:165-184. Sterner, R. W. 1989. The role of grazers in phyto- plankton succession. In Plankton Ecology- Succession in Plankton Communities, ed. U. Sommer, pp. 107-170. Springer-Verlag, Berlin. Strong, D. R. 1992. Are trophic cascades all wet? Differentiation and donor-control in speciose ecosystems. Ecology 73:747-754. Swift, M. J., 0. W. Heal, and J. M. Anderson. 1979. Decomposition in Terrestrial Ecosys- tems. Blackwell, Oxford, U.K. Tamm, C.-0. 1991. Nitrogen in Terrestrial Eco- systems. Springer-Verlag, Berlin. Tonn, W. M. 1990. Climate change and fish com- munities: A conceptual framework. Transac- tions of the American Fisheries Society Turner, A. M. andG. G. Mittelbach. 1990. Preda- tor-avoidance and community structure: Inter- actions among piscivores, planktivores, and plankton. Ecology 71:2241-2254. Vadas, R. L. Jr. 1989. Food web patterns in eco- systems: A reply to Fretwell and Oksanen. Oikos 56:339-343. Vadas, R. L. Jr. 1990. The importance of omni- vory and predator regulation of prey in freshwa- ter fish assemblages of North America. Environ- mental Biology of Fishes 27:285-302. 119~337-352. VanBlaricom, G. R. and J. A. Estes. 1988. The Community Ecology of Sea Otters. Springer- Verlag, New York. Vitousek, P. M. andD. U. Hooper. 1993. Biologi- cal diversity and terrestrial ecosystem bio- geochemistry. In Biodiversity and Ecosystem Function, eds. E. D. Schulze and H. A. Moo- ney, pp. 3-14. Springer-Verlag, New York. Walter, D. E. 1987. Trophic behavior of "myco- phagous" microarthropods. Ecology 68:22f% 229. Warren, C. F., J. H. Wales, G. E. Davis, and P. Douderoff. 1964. Trout production in an ex- perimental stream enriched with sucrose. Jour- nal of Wildlife Management 28:617-660. Warren, P. H. 1989. Spatial and temporal varia- tion in the structure of a freshwater food web. Oikos 55:299-3 1 1. Watson, S. and J. Kalff. 1981. Relationships be- tween nannoplankton and lake trophic status. Canadian Journal of Fisheries and Aquatic Sci- ences 38:960-967. Wedin, D. A. and D. Tilman. 1990. Species ef- fects on nutrient cycling: A test with perennial grasses. Oecologia 84:433-441. Werner, E. E. 1986. Species interactions in fresh- water fish communities. In Community Ecol- ogy, eds. J. Diamond and T. Case, pp. 344- 358. Harper and Row, New York. Werner, E. E. 1988. Size, scaling and the evolu- tion of life cycles. In Size-Structured Popula- tions-Ecology and Evolution, eds. B. Eben- man and L. Person, pp. 60-81. Springer- Verlag, Berlin. Wetzel, R. G. 1979. The role of the littoral zone and detritus in lake metabolism. Archive fur Hydrobiologie und Beihefte Ergebnisse der Limnologie 13:145-161. Wetzel, R. G. 1983. Limnology, 2nd ed. Saunders College Publishing, Philadelphia. Wiegert, R. G. and D. F. Owen. 1971. Trophic structure, available resources and population density in terrestrial vs. aquatic ecosystems. Journal of Theoretical Biology 30:69-81. Wiley, M. J. and G. L. Warren. 1992. Territory abandonment, theft, and recycling by a lotic grazer: A foraging strategy for hard times. Oikos Williamson, M. H. 1981. IslandPopulations. Ox- ford University Press, Oxford, U.K. Wilson, J. B. and A. D. Q. Agnew. 1992. Posi- tive-feedback switches in plant communities. Advances in Ecological Research 23:263-336. Wilson, W. H. 1991. Competition and predation 631495-505. 434 I Lennart Persson, Jan Bengtsson, Bruce A. Menge, and Mary E. Power in marine soft-sediment communities. Annual Review of Ecology and Systematics 21:221- 241. Witman, J., J. J. Leicter, S. J. Genovese, and D. A. Brooks. 1993. Pulsedphytoplankton sup- ply to the rocky subtidal zone: Influence of inter- nal waves. Proceedings of the National Acad- emy of Science, USA 90:168&1690. Wootton, J. T. 1991. Direct and indirect effect of nutrients on intertidal community structure: Variable consequences of seabird guano. Jour- nal of Experimental Marine Biology and Ecol- ogy 15 1 : 139- 153. Wootton, J. T. andM. E. Power. 1993. Productiv-' ity, consumers, and the structure of a river food chain. Proceedings of the National Academy of Science, USA 90:1384-1387. Wright, D. H. 1983. Species-energy theory: An extension of species-area theory. Oikos 41:49& 506. Yodzis, P. 1993. Environment and trophodiver- sity . In Species Diversity in Ecological Commu- nities, eds. R. R. Ricklefs and D. Schluter, pp. 26-38. The University of Chicago Press, Chicago.