Bias in survival estimates from tag-recovery models where catch-and-release is common, with an example from Atlantic striped bass (Morone saxatilis)

Survival rate is underestimated when tag-recovery models include tags recov ered from harvested and caught-and-released fish. The magnitude of the bias depends on tag-recovery rate, proportion of catch released alive, and repo rting rate; changes in these factors over time confound temporal changes in survival. The bias is of potential concern for any tagging study where cat ch-and-release is mandatory or practiced voluntarily. The bias is of concer n particularly for the Atlantic striped bass (Morone saxatilis) tagging stu dy where catch-and-release is common and anglers commonly remove the tag up on capture regardless of fish disposition. Biased estimates of striped bass survival did not change with changes in harvest regulation during the mid- 1990s. However, bias-adjusted estimates of survival showed a decrease, whic h corresponds to the regulatory change made in 1995. Year-specific reportin g rate is critical to bias adjustment, underscoring the need for reward tag s in fish tagging studies. Tag-recovery modeling allows for a diverse set o f models, each of which can produce widely different estimates with far-rea ching consequences for management. We applied model averaging to base infer ence on a weighted average of parameter estimates and to account for model selection uncertainty.