Multi-state dead recovery mark-recovery model performance for estimating movement and mortality rates

Abstract Multi-state mark-recovery models are used to estimate movement and mortality rates of terrestrial and aquatic animals. These models have become especially popular in the last 20 years since technology and statistical techniques have improved to accommodate the extensive data requirements. However, the ability of multi-state mark-recovery models to estimate movement rates has received little evaluation, with few studies exploring the effects of alternative release and recovery designs on the bias and precision of estimates. Our objectives were to evaluate the effects of the spatio-temporal pattern of releases, pattern of recovery efforts, and number of releases on the performance of a multi-state mark-recovery model. We generated mark-recovery data from a spatial model and fitted them using a multi-state dead recovery model that included prior distributions on movement rates, natural mortality, and catchability. We generated data using a spatially variable schedule of releases and effort, and a release size of about 1 million individuals to mimic a mark-recovery study conducted for Atlantic Menhaden Brevoortia tyrannus in the late 1960s. We also ran alternative scenarios of sample size and spatially uniform releases and effort, either by themselves or in combination to determine their effects on the accuracy of the estimates. The model generally produced unbiased estimates of mortality rates with median error −1 for all scenarios, but some biases were present for the movement rates. Movement rates and catchability were more accurately and precisely estimated in scenarios that included spatially uniform fishing effort, while spatially uniform releases had little to no effect on bias or precision of estimated movement rates. Increased sample size improved accuracy of all parameter estimates except for the lowest movement rates. Future mark-recovery experiments that use a multi-state dead recovery model may benefit from distributing recovery effort uniformly over time and space.