Modeling spatial patterns of limits to production of deposit-feeders and ectothermic predators in the northern Bering Sea

Network models can help generate testable predictions and more accurate projections of food web responses to environmental change. Such models depend on predator–prey interactions throughout the network. When a predator currently consumes all of its prey's production, the prey's biomass may change substantially with loss of the predator or invasion by others. Conversely, if production of deposit-feeding prey is limited by organic matter inputs, system response may be predictable from models of primary production. For sea floor communities of shallow Arctic seas, increased temperature could lead to invasion or loss of predators, while reduced sea ice or change in wind-driven currents could alter organic matter inputs. Based on field data and models for three different sectors of the northern Bering Sea, we found a number of cases where all of a prey's production was consumed but the taxa involved varied among sectors. These differences appeared not to result from numerical responses of predators to abundance of preferred prey. Rather, they appeared driven by stochastic variations in relative biomass among taxa, due largely to abiotic conditions that affect colonization and early post-larval survival. Oscillatory tendencies of top-down versus bottom-up interactions may augment these variations. Required inputs of settling microalgae exceeded existing estimates of annual primary production by 50%; thus, assessing limits to bottom-up control depends on better corrections of satellite estimates to account for production throughout the water column. Our results suggest that in this Arctic system, stochastic abiotic conditions outweigh deterministic species interactions in food web responses to a varying environment.