We exploited the Exxon Valdez oil spill in Prince William Sound (PWS), Alaska, to evaluate effects of reduced sea otter densities on prey populations in sedimentary habitats. We considered the need for and characteristics of new models for trophic effects of sea otters on coastal marine benthic communities. We viewed evidence for nonlinear or uncertain patterns of prey response to varying sea otter density as particularly significant for new model structure.

We specifically examined responses of densities and size distributions of populations of mussels and clams (several taxonomic and habitat categories), all important sea otter prey in PWS, to reduction in sea otter density caused by the oil spill. We utilized two primary criteria for determining the consistency of prey demographic responses to reduced sea otter densities as predicted by null hypotheses consistent with existing published models. First, prey populations subject to reduced influence by sea otters should be denser and contain proportionately more large individuals than prey populations strongly influenced by sea otter predation. Second, response times of prey demography to reduced otter densities should be similar to response times of prey to increased otter densities, the latter as indicated in existing published models.  

Results were disparate with regard to expectation for the six categories of prey evaluated. With few exceptions, density data indicated nonconformance with demographic expectations. In contrast, size data for prey indicated conformance with expectation in about half the categories evaluated. We suggest that lingering effects of the oil spill, nonlinear relationships of sea otters and prey that involve thresholds in otter density, uncertainties in prey recruitment patterns, spatial differences in natural disturbance rate, and differences between areas in effects of competing predators are the main factors possibly accounting for patterns in our data. Recruitment and disturbance effects in particular may include significant stochastic components, especially in a temporal context. We suggest that recovered sea otter populations and their prey do not necessarily exist in long-term stable equilibria, and that development of new models incorporating both trophic thresholds and trophic stochasticity will be important in understanding community-level responses to variable sea otter numbers.