OOS 12-2 - Herbivore behavior regulates large-scale patterns of community states

Background/Question/Methods
Distinct community regimes characterize many systems and arise from environmental heterogeneity or biotic interactions. Recent work shows that interaction strength can depend on density-dependent changes in animal behavior. This raises the question of whether behavior can affect large-scale community patterns. On temperate rocky reefs, kelp forests and urchin barrens can span large ( >500m) patches in California but co-occur across adjacent depth zones along the coast of New Zealand. These spatial patterns might arise from spatial heterogeneity in wave stress and urchin density or because urchins switch from passive to active grazing at low kelp densities, dramatically increasing total grazing rate. We quantify the extent to which spatial heterogeneity versus scale-specific feedbacks in urchin grazing behavior pattern communities by comparing the fit of dynamical models incorporating each feature to field data spanning 201 reefs.

Results/Conclusions
We find that feedbacks in urchin behavior best explain observed community patterning on temperate rocky reefs. Large kelp forest and barren patches in California are most consistent with large-scale grazing feedbacks, where insufficient kelp frond subsidies to the bottom induce active urchin grazing across entire reefs. Best-fitting models in New Zealand include depth-dependent wave stress on kelp and local (1-5m) feedbacks where urchins avoid dense kelp stands that increase abrasion and predation risk. This combination of stress gradients and behavior feedbacks drives a sharp transition from shallower urchin-dominated zones to deeper kelp-dominated zones. Given these regional differences in behavior, our best-fitting models predict alternatively stable kelp- and urchin-dominated states that form (i) local mosaics at intermediate 3-8m depths in New Zealand but (ii) span entire reefs in California (32% of reefs). Thus, behavior can explain community patterning and why only some systems exhibit community-wide alternative stable states. As our modeled feedbacks arise from a widespread Type IV functional grazing response, our results also suggest that behavior can underlie community patterns in many ecosystems.