Ecology, 91(3), 2010, pp. 637–643 Ó 2010 by the Ecological Society of America Habitat isolation moderates the strength of top-down control in experimental pond food webs JONATHAN M. CHASE, 1 AMBER A. BURGETT, AND ELIZABETH G. BIRO Department of Biology and Tyson Research Center, Washington University in Saint Louis, Saint Louis, Missouri 63130 USA Abstract. Habitat isolation is well known to alter patterns of species’ abundance, richness, and the ratios of predators : prey. Less clear, however, is how isolation alters interactions within food webs. Here, we present the results from an experiment performed in artificial ponds (mesocosms) manipulating habitat isolation crossed with a predator reduction treatment to disentangle how isolation mediates the top-down effect of predators. The strength of the trophic cascade, from predators, through herbivores, to producers, was considerably stronger in connected than in isolated habitats. We further found that the overall richness of both predator and herbivore species declined strongly with isolation. Experimental predator reductions suggest that the mechanism underlying the herbivore response was likely mediated by a keystone predator effect; when predators were reduced, herbivore richness was lower, and there was no discernable effect of isolation on herbivore richness. Finally, we found that the composition of predators in more isolated habitats consisted of species that were smaller and likely less effective predators than species that persisted in less isolated habitats. In all, our experiment showed that habitat isolation can alter the structure of communities by a combination of direct effects of the species in question, as well as effects mediated through their interactions in the food web. Key words: dispersal; experimental ponds; food webs; habitat isolation; metacommunity; trophic cascade. INTRODUCTION Most natural habitats are patchy (Levin 1992). As a result, distances among patches (isolation) can influence the rate by which individuals disperse among patches, and can influence many aspects of metapopulation and metacommunity dynamics (Holyoak et al. 2005). Variation in dispersal among patches can influence population persistence (e.g., Hanski 1999), the strength of interspecific interactions (e.g., Amarasekare 2003, Ryall and Fahrig 2006), and the diversity of species in communities (e.g., Cadotte 2006). However, habitat isolation does not influence all members of a community equally. For example, within a trophic level, species often have different dispersal strategies and thus are differentially influenced by habitat isolation (Damschen et al. 2008). Further, among trophic levels, predators tend to exhibit disproportionate responses to isolation relative to their prey (Holt and Hoopes 2005). This is not only due to differences in dispersal, but also extinction rates because predators typically have larger body sizes, higher metabolic needs, and smaller popu- lation sizes than prey (e.g., Schoener 1989). While there are theoretical and empirical reasons to suspect that isolation can alter the role that predators play in food webs (e.g., Holt and Hoopes 2005), it remains unclear as to exactly how these differential effects manifest in food webs. First, if isolation influences predators as a group more than prey, we might expect to see that changes in the structure of the food web (e.g., more prey relative to predators) can emerge simply due to differential dispersal experienced by members of the different trophic levels. Second, isolation can sometimes disproportionately affect the number of predator species that can exist relative to prey species (Gilbert et al. 1998, Shulman and Chase 2007). Finally, species that are favored and that can persist in isolated habitats are often smaller and less effective predators than those that are favored in more connected habitats (reviewed in Holt and Hoopes 2005). In addition to simply changing food web structure through these potential differential effects on predators and prey, isolation’s effects on predator abundance, species richness, and/or composition can mediate their top- down control on the rest of the food web (Gripenberg and Roslin 2007). To date, studies on the effects of habitat isolation on food web interactions have explored dynamics in simplified theoretical (e.g., Shurin and Allen 2001) or microcosm (Holyoak 2000) systems. However, no study has yet attempted to examine how habitat isolation influences the structure of food webs and the relative strengths of top-down control in complex natural communities. Here, we experimentally examined the Manuscript received 14 February 2009; revised 28 September 2009; accepted 13 October 2009. Corresponding Editor: D. A. Spiller. 1 E-mail: jchase@wustl.edu 637 R eports