NEWS & COMMENT d Kobert Holt comment on Gary Polis, Michael Rose visiting colleagues - y killed in a boating What enables trophic cascades? Commentary on Polis et a/. olis and colleagues, in both a 1999 P paper` and in this issue of TREE @p. 473-476), have made the case for a needed distinction between community- level and species-level trophic cascades. Confusion over the level of aggregation in studies of predator impacts on prey and plants has muddied this issue for more than 30 years. Murdoch2, in his critique of Hairston et d3, pointed out the need to specify the unit of study: `It might be, for example, a single population or a group of populations of a recognizable type, etc.' He argued that if fossil fuels didn't accumulate, this needn't imply food limi- tation for all detritivores, because some of these populations could be limited by predators or other factors below the level that would have been set by their food, leaving other populations to `eat up the "left-over" food'. Slobodkin et al.4 rebutted that they weren't concerned in their 1960 paper with a majority vote (how many individual consumer or pro- ducer species were regulated by food, predators or some other factor), but rather with how their collective trophic level biomass responded. Ehrlich and Birch5 argued that, contrary to the generalization that herbivores were gen- erally not food limited, specialists such as the cabbage white butterfly might be. If the world were planted with more cabbages, its population would increase. Slobodkin et al.4 again replied by saying that if the world were planted with more cabbages, other plants, and their associ- ated herbivores, would decrease as a result of cabbage competition, causing collective trophic level abundance to respond as they predicted. Different questions require different levels of aggregation. The similar spec- tral irradiances of manzanita and mari- juana might not matter to scientists inter- ested in remote sensing of chaparral plant cover, but are inconvenient for drug enforcement officers. Community ecologists often seek some intermediate level of resolution (e.g. `edible' and `ined- ible' members of a trophic level6). Our j i conceptual pigeonholes accommodate natural continua and variation very imperfectly. Improving our approxi- mations requires specific research on which differences are trivial, and which are crucial, for a particular prediction. If ecologists accept the community- versus species-level cascade terminology pro- posed by Polis and colleagues, we could at least better communicate which phe- nomena we are attempting to predict. We still have a lot to learn about the distribution and abundance of both types of cascades, and the community charac- teristics that enable them. Schmitz et al.7 found trophic cascades frequent in pub- lished terrestrial studies (45/60 cases), with carnivore impacts on plants and her- bivores equal to or stronger than those documented in aquatic systems. Polis et a/, (in this issue) point out that the studies they reviewed documented species-level cascades. Documenting community-level cascades on land requires large spatial and temporal scales, in part because of the storage effects imposed by long-lived terrestrial producersg. By an ingenious synthesis of paleoecology and contempo- rary Siberian pony manipulations, Zimov and colleagues10 argued persuasively for topdown mediation of vegetation transi- tion from steppe graminoids to mossy tundra over Beringia, following the ex- termination by human hunters of Pleisto- cene grazing megafauna. Here, however, plant biomass did not increase, and a key mechanism (trampling) was nontrophic. Polis and other terrestrial ecologists here and elsewhere have argued that aquatic ecosystems that are homo- geneous, simple and closed are more likely to cascade. Aquatic ecologists, however, have been impressed with the subtle het- erogeneity that structures even pelagic communities, where thermal convection cells, gelantinous surfaces of salps, or algal aggregates provide key boundaries that delimit and intensify processes and interactionsll-13. The jury is still out on how diversity differs between aquatic and terrestrial systems, pending more thorough inventories, including of mi- crobial species identities, in both. It is common for a benthic or planktonic algal sample of a few cm3 to contain >30 species; however, a producer species density is difficult to match in a ter- restrial sample of a similar volume. Size disparity between consumers and re- sources, and relatively fast prey dy- namics, can sometimes enable cascades (Refs 1,14 and on pp. 473-476). Fast resource dynamics allow aquatic sys- tems to respond to consumer or nutri- ent manipulations over short timescales convenient for experimentalists, but could work against top-down control by permitting demographic escape. The small size of freshwater algae might make them more uniformly ingestible but, with high resource loading, algae can accrue enough biomass to outcom- Pete animals for oxygen, putting an abrupt end to top-down control. Similar density related escapes were reported for whelks, which when abundant turn the tables on spiny lobsters, their former predators, and collectively rasp them to deathl5. This `run away production' is the converse of Strong's'6 `run away con- sumption', and is yet another manifes- tation of strong nonlinearity in ecological relationships. Polis and colleagues (and many oth- ers) have pointed out that trophic cas- cades might be more likely in a hom- ogeneous, closed ecosystem. These reasonable arguments are not supported by a model of Carpenter et al.17, in which they summarize conditions influencing whether eutrophication of lakes could be reversed (e.g. by biomanipulations of higher trophic levels). Eutrophication becomes irreversible when nutrient fluxes from the watershed, or from inter- nal recycling, overwhelm nutrient sinks (higher trophic levels, sedimentation out of the euphotic zone or flushing through outflowing rivers). In their model, hetero- geneity (having a sediment compartment in which nutrients become unavailable to algae) or an open system (river washout of excess nutrients) enable rather than preclude trophic cascades that allow higher trophic levels to affect plant bio- mass. As always, we need specific under- standing of causal processes to predict trophic cascades, or any other commu- nity or species-level phenomena. NEWS & COMMENT What we are really struggling to understand in our investigations of tro- phic cascades is how and why the strength of ecological interactions varies over space and time and across taxa. Our understanding clearly is contingent on the scales (scope and resolution) of our investigation, both taxonomic (lumping versus splitting and how much of the web to include) and spatio-temporal (how to delimit systems in space and time and how intensively to sample them). Gary Polis stimulated the quest for clarity and larger understanding of these issues tremendously, and led by example to provoke community ecologists to `stop looking at our feet', and ecosystem ecol- gists to deepen their consideration of natural history. As we continue Paine'sls `profitably frustrating' quest for predic- tive understanding of trophic dynamics, Gary Polis' insights and impetus will con- tinue to energize and illuminate food web research. Mary E. Power Depf of Integrative Biology, University of California, Berkeley, CA 94720, USA (mepower@garnet. berkele y.edu) References 1 Polis, G.A. et a/. (1999) Why are parts of the world green? Multiple factors control productivity and the distribution of biomass. Oikos 86,2-15 2 Murdoch, W.W. (1966) Community structure, population control, and competition - a critique. Am. Nat. 100,219-226 structure, population control, and competition. Am. Nat. 94,421-425 4 Slobodkin, L.C. (1967) Regulation in terrestrial ecosystems and the implied balance of nature. Am. Nat. 101,109-124 5 Ehrlich, P.R. and Birch, L.C. (1967) The balance of nature and population control. Am. Nat. 6 Marks, J.C. et a/. (2000) Flood disturbance, algal productivity, and interannual variation in food chain length. Oikos 90, 20-27 7 Schmitz, O.J. et a/. (2000) Trophic cascades in terrestrial systems. A review of the effects of carnivore removals on plants. Am. Nat. 8 Warner, R.R. and Chesson, P.L. (1985) Coexistence mediated by recruitment fluctuations: a field guide to the storage effect. Am. Nat. 125,769-787 9 Terborgh, J. et a/. (1997) Transitory states in relaxing ecosystems of landbridge islands. In Tropical Forest Fragments (Laurance, W.F. and Bierregaard, R.O., eds), pp. 256-274, University of Chicago Press 3 Hairston, N.G. et:/. (1960) Community 101,97-I07 155,141-153 Trophic cascades in terrestrial ecosystems. Reflections on Polis et a/. n its simplest form, the concept of a I `trophic cascade' is an ecological vari- ant of a basic truism, `the enemy of my enemy is my friend'. Recent definitions of the term `trophic cascade' include: `recip rocal predator-prey effects that alter the abundance, biomass, or productivity of a population, community, or trophic level across more than one link in a food web" and the `propagation of indirect mutualisms between nonadjacent levels in a food chain'2. In principle, these defini- tions apply throughout a food web, but, in practice, there has been a focus on indirect carnivore impacts on plants via shifts in herbivore abundance and activ- ity's. This emphasis reflects a fundamen- tal ecological question: to understand the forces that govern plant community composition and dynamics, must one pay attention to the food webs supported by those plant communities? If trophic cascades are ubiquitous and large in magnitude, the answer is `yes'. Gary Polis4 and his colleagues (see pp. 473-476, this issue) sensibly observe that it is important to distinguish between `species-level' and `community-level' cas- cades. In species-level cascades, altering predator numbers indirectly influences just one or a few plant species; whereas, in community-level cascades, there is a substantial impact on plant biomass dis- tribution for entire communities. They urge ecologists to agree on objective measures of strengths of cascades. In addition to these useful methodological and terminological suggestions, Polis et al. suggest that `community cas- cades.. . [ are] apparently absent or rare in terrestrial habitats' as compared with aquatic habitats and that `support for even species-level cascades is limited in terrestrial systems's. They argue that this putative difference between biomes reflects the great complexity of terres- trial ecosystems and the reticulate pat- terning of food webs. I would like to 10 Zimov, S.A. eta/. (1995) Steppe-tundra transition: a herbivore-driven biome shift at the end of the Pleistocene. Am. Nat 146,765-794 I1 Silver, M.W. eta/. (1995) Marine snow, what it is and how it affects ecosystem functioning. In Linking Species and Ecosystems (Jones, C.G. and Lawton, J.H., eds), pp. 45-51, Chapman &Hall 12 Alldredge, A.L. and Cohen, Y. (1987) Can microscale chemical patches persist in the sea? Microelectrode study of marine snow, fecal pellets. Science 235,689-691 13 Powell, T.M. (1989) Physical and biological scales of variability in lakes, estuaries, and the coastal ocean. In Perspectives in Ecological Theory(Roughgarden, J. et a/., eds), pp. 157-176, Princeton University Press 14 Power, M.E. eta/. (1996) Disturbance and food chain length in rivers. In Food Webs, Integration of Pafferns and Dynamics (Polis, G.A. and Winemiller, K.O., eds), pp. 286-297, Chapman &Hall 15 Barkai, A. and McQuaid, C. (1988) Predator-prey role reversal in a marine benthic ecosystem. Science 242,62-64 16 Strong, D.R. (1992) Are trophic cascades all wet? Differentiation and donor-control in speciose ecosystems. Ecology 73,747-754 17 Carpenter, S.R. et a/. (1999) Management of eutrophication for lakes subject to potentially irreversible change. Ecol. Appl. 9,751-771 18 Paine, R.T. (1980) Food webs, linkage, interaction strength and community infrastructure. J. Anim. fcol. 49, 667-685 respectfully suggest that the jury is still out on these substantive claims. Unfortunately, the proper timescale for assessing trophic cascades at the plant community level extends well be- yond that of typical field studies. In the recent review of terrestrial studies by Schmitz et al.3, 81% of the studies in- volved measurements within a single annual growing season of the focal plants, many of which were long-lived shrubs or trees. A fair test of trophic cascades would have to extend over multiple plant generations. Extending the timescale could either enhance or weaken cas- cades. A small increase in herbivory that seems `insignificant' in any given year could be greatly magnified in its ultimate impact (for instance, if it lowers the com- petitive ability of a plant species). Or, a large impact within a single growing season might induce compensatory or defensive mechanisms, or act at life stages unimportant in determining den- sity, and thus become weakened over longer timescales. The timescale issue in terrestrial stud- ies is a tough nut to crack. For instance, consider the ambitious, large-scale study reported recently by Sinclair et a1.6 in Canadian boreal forest. Mammalian and avian predators were excluded to exam- ine indirect impacts upon vegetation.