The robustness of Brownie tag return models to complex spatiotemporal dynamics evaluated through simulation analysis

Abstract

The development of a reliable tagging program requires simulation testing the experimental design. However, the potential for model misspecification, particularly in the underlying spatiotemporal dynamics, is often ignored. A continuous time, spatially explicit, age-structured, capture–recapture operating model was developed to better emulate real-world population dynamics typically overlooked in spatially aggregated or discrete time tagging models. Various spatiotemporal model parametrizations, including case studies with Atlantic bluefin tuna (Thunnus thynnus) and yellowfin tuna (Thunnus albacares), were explored to evaluate the bias associated with Brownie tag return estimation models. Simulations demonstrated that accounting for connectivity was essential for obtaining unbiased parameter estimates and that migration rates could be reliably estimated without the correlation associated with other parameters (e.g., between tag reporting and mortality). Mortality parameter estimates were particularly sensitive to the temporal dynamics of the tagging and fishing seasons, but accounting for the seasonality in tag releases and fishery recaptures allowed for relatively unbiased estimation. Our results indicate that parameter bias and uncertainty can be severely underestimated when discrete time or spatially aggregated operating models are used to determine optimal experimental design of tagging studies.