Finding the perfect mismatch: Evaluating misspecification of population structure within spatially explicit integrated population models

Abstract

Spatially stratified integrated population models (IPMs) can account for fine-scale demographic processes and support spatial management for complex, heterogeneous populations. Although spatial IPMs may provide a more realistic representation of true population dynamics, few studies have evaluated the consequences associated with incorrect assumptions regarding population structure and connectivity. We utilized a simulation-estimation framework to explore how mismatches between the true population structure (i.e. uniform, single population with spatial heterogeneity or metapopulation) and various parametrizations of an IPM (i.e. panmictic, fleets-as-areas or a spatially explicit, tag-integrated model) impacted resultant fish population estimates. When population structure was incorrectly specified in the IPM, parameter estimates were generally unbiased at the system level, but were often biased for sub-areas. Correctly specifying population structure in spatial IPMs led to strong performance, whereas incorrectly specified spatial IPMs performed adequately (and better than spatially aggregated counterparts). Allowing for flexible parametrization of movement rates (e.g. estimating age-varying values) was more important than correctly identifying the population structure, and incorporation of tag-recapture data helped movement estimation. Our results elucidate how incorrect population structure assumptions can influence the estimation of key parameters of spatial IPMs, while indicating that, even if incorrectly specified, spatial IPMs can adequately support spatial management decisions.