Dynamic and context-dependent keystone species effects in kelp forests

Sea otters are an iconic keystone predator that can maintain kelp forests by preying on grazing invertebrates such as sea urchins. However, the effects of sea otters on kelp forests vary over their geographic range. Here, we analyze two 30-y datasets on kelp forest communities during the reintroduction of sea otters along the west coast of Vancouver Island, BC, Canada, and around San Nicolas Island, CA. We developed a community model to estimate species interactions as dynamic rates, varying with community state. We find evidence of a classic trophic cascade off Vancouver Island; the arrival of otters quickly led to depletion of urchins and recovery of kelp. However, this cascade was muted around San Nicolas Island, with otters, urchins, and kelp all coexisting at intermediate densities for multiple years. Our models show that this difference came from a pulse of strong otter impacts on urchins following recolonization off Vancouver Island, but not off San Nicolas Island. The mean effects of otters on urchins and urchins on kelp were not stronger in the north, indicating that interaction dynamics and not average interaction strength are key to explaining differences in community trajectories. We also find stronger multistep interaction chains in the south, arising from competitive interactions that indirectly buffered otter effects. These findings shed light on long-standing hypotheses about how interspecific interactions can alter the function of keystone species across community contexts. More broadly, we show how community change can be more accurately predicted by considering dynamic interaction strengths.