Context-dependent effects of fishing: variation in trophic cascades across environmental gradients.

Marine reserves provide a large-scale experimental framework to investigate the effects of fishing on food web dynamics and how they vary with environmental context. Because marine reserves promote the recovery of previously fished predators, spatial comparisons between reserve and fished sites are often made to infer such effects; however, alternative explanations for differences between reserve and fished sites are seldom tested (e.g., environmental variation among sites). We investigated the context dependency of the predator-urchin-kelp trophic cascade reported in northeastern New Zealand by comparing the abundance of herbivorous sea urchins (Evechinus chloroticus), the extent of urchin barrens habitat, and macroalgal biomass between reserve and fished sites within six locations that span an environmental gradient in wave exposure, sedimentation, and water clarity. At depths where differences in urchin abundance or macroalgal biomass were found between reserve and fished sites we used a model selection approach to identify which variables (fishing or environmental factors) best explained the variation among sites. Differences between reserve and fished sites were not ubiquitous across the locations examined and were highly depth specific. At sheltered locations, urchins were rare and barrens absent at both reserve and fished sites. At moderately exposed coastal locations, actively grazing urchins were most abundant at 4-6 m depth, and significant differences in macroalgal biomass between reserve and fished sites were observed. In contrast, at offshore island locations, urchins extended into deeper water, and differences between reserve and fished sites were found at 4-9 m depth. These differences could only be attributed to trophic cascades associated with protection from fishing in two of the six locations examined. In other cases, variation between reserve and fished sites was equally well explained by differences in sediment or wave exposure among sites. These results suggest that trophic cascades are not ubiquitous to northeastern New Zealand's subtidal reefs and the importance of sea urchins, and indirectly predators, in controlling macroalgal biomass will vary at local and regional scales in relation to abiotic factors. A better mechanistic understanding of how environmental variation affects the strength of species interactions across multiple spatial scales is needed to predict the ecosystem-level effects of fishing.