Introduction

Since its first appearance (Hairston, Smith & Slobodkin, 1960), the hypothesis that predation can structure communities has courted controversy (Shapiro, Lamarra & Lynch, 1975; Strong, 1992; McCann, Hastings & Strong, 1998). Nearly 50 years later there is still ongoing debate over the importance of predation relative to other factors limiting the growth of populations (Pace et al., 1999; Holt, 2000; Polis et al., 2000; Power, 2000), and the conditions that cause the effect of predation to cascade through the community (Polis & Strong, 1996; Schmitz, Krivan & Ovadia, 2004; Borer et al., 2005; Vander Zanden, Essington & Vadeboncoeur, 2005). With the discovery of predator impacts on the structure and dynamics of a diversity of real communities (Paine, 1980; Power, Matthews & Stewart, 1985; Carpenter & Kitchell, 1993), it became apparent that higher trophic levels could affect the biomass and dynamics of not only their prey, but of their prey's prey and, hence, the whole community. Earlier it was assumed that communities were typically pyramidal in structure, with declining biomass in each successive trophic level, and the dynamics of each trophic level dependent upon those of their prey and ultimately the primary producers/basal resources (Whittaker, 1961). It is now clear from habitats as diverse as Californian islands (Roemer, Donlan & Courchamp, 2002), the forests of Yellowstone Park (Ripple & Beschta, 2004) and the cod banks of the North Atlantic (Worm & Myers, 2003; Frank et al. 2005) that this assumption is not correct, such that nowadays the predictions of the trophic cascade influence how we manage our natural environment (Scheffer, 1998).