Forecasting capelin Mallotus villosus biomass on the Newfoundland shelf

Forage fish play a central role in marine ecosystems, acting as a trophic link between plankton and larger marine species. They also contribute to global economies by directly and indirectly supporting commercial and recreational fisheries. Managing forage fish is problematic due to their high spatial-temporal variation in biomass and strong responses to environmental variability. A variety of mechanisms have been proposed to explain variation in capelin Mallotus villosus biomass, a keystone forage fish in the Northwest Atlantic, including factors influencing cohort strength such as larval abundance and larval food (i.e. Pseudocalanus spp.) availability. Alternately, pre-spawning mortality of capelin may be regulated by the timing of the retreat of sea ice (t(i)(ce)) via its effects on the spring bloom and the availability of food (e.g. Calanus spp.) for adult capelin, or by the condition of adult capelin the previous fall. Here, we used a Bayesian approach in a multimodel inference framework to assess support for a series of hypotheses explaining the population dynamics of capelin. The most parsimonious model included larval abundance, t(i)(ce), and capelin condition in the fall, and explained 68% of the variance in capelin biomass. These results are consistent with much of the literature that suggests capelin in the Northwest Atlantic are driven by bottom-up forces. This model allows us to produce forecasts of capelin biomass 1-2 management cycles in advance, which will be a valuable contribution to the management of this forage fish species.