Outside of relatively closed ecosystems such as lakes, the relevance and scale at which alternative stable states may emerge in demographically open systems such as coral reefs, grasslands, and kelp forests is unclear. In the presence of dispersal, feedback mechanisms typically described at local scales may not produce alternatively stable states as entire ecosystems transition into the most stable state through a domino effect. Other studies predict that depending on the intensity of dispersal, alternative stable states emerge either at local or system-wide scales. Therefore, the appropriate spatial scale (if any) of efforts to promote the magnitude of disturbances which desired ecosystem states can withstand (ecological resilience) is often unclear. In southern California kelp forests, large-scale (50-100km) consumer dispersal coincides with well-documented feedbacks within and abrupt shifts among forests and urchin barrens occurring at localized (1-10km) scales. Such transitions are often associated with intense fishing of urchin predators, and can be triggered by urchin recruitment pulses. To determine the spatial scale, if any, at which ecological resilience is relevant in this system, we develop a spatially explicit tritrophic model which incorporates high consumer dispersal and alternative stable states, as well as observed levels of local-scale temporal variability in urchin recruitment.
We find that despite intense dispersal of consumers during recruitment, modeled communities persist in and shift among distinct kelp forest and urchin barren states at localized (10-30km of coastline) spatial scales. This results from both a limited ability of recruit dispersal to homogenize community states across space and the spatial heterogeneity imposed by variability in urchin recruitment. At the system-wide scale (1200km of coastline), our model exhibits gradual and reversible transitions among forests and barrens in response to increasing fishing pressure on urchin predators. This contrasts intuition from conceptual models suggesting that intense dispersal in marine systems yields large-scale state shifts in response to press perturbations which are abrupt and difficult to reverse. However, at local scales undesired urchin barren states can persist for decades (10-30 years) even under moderate fishing levels, highlighting the relevance of ecological resilience at local rather than system-wide scales. In addition, the spatial scales of community shifts in our model also qualitatively match those reported in empirical studies of kelp forests. This suggests that the often-observed temporary and localized shifts among forested and barren states are consistent with the presence of alternative stable states in temporally variable environments.