January 23rd, 2018 9:00 (PST): FSH 203
Development of a size-structured spatiotemporal population model for invertebrates: individual growth, size-transitions, and natural and fishing mortality
Population dynamics and stock assessment models have historically modeled populations by tracking total abundance (often structured by age, size, and/or sex) across the entire stock. This practice implicitly assumes that individuals are equally mixed within each stratum, and overlooks the fact that marine populations are spatially patchy and locally structured. In this study, we apply spatiotemporal approach to population models by combining theory and methods from population dynamics and geostatistics. Specifically, spatiotemporal population models define population variables (density) as varying continuously across space, while estimating spatial variation as a random effect. Size-structured population models are convenient for populations where ageing information is unavailable or inaccurate, and such models are commonly used for assessment of crustacean species worldwide, e.g. crab, shrimp, and lobster. This proposed size-structured spatiotemporal model allows external specification of survival, growth, and size transition rates, and assumes that recruitment can be approximated as varying around a constant mean that varies spatially. Spatially-explicit catch data from fisheries are used for estimating annual fishing mortality across space. We use simulation approach to demonstrate that the spatiotemporal population model provides (1) accurate and precise estimates of spatial variation in population density and fishing mortality of each size class over years, (2) unbiased estimates of total abundance and model parameters, such as fishery selectivity. We apply this model to eastern Bering Sea snow crab, where spatially-explicit fishery-dependent and fishery-independent data are directly used.