Wanted dead or alive: a state-space mark–recapture–recovery model incorporating multiple recovery types and state uncertainty
We developed a state-space mark–recapture–recovery model that incorporates multiple recovery types and state uncertainty to estimate survival of an anadromous fish species. We apply the model to a dataset of outmigrating juvenile steelhead trout (Oncorhynchus mykiss (Walbaum, 1792)) tagged with passive integrated transponders, recaptured during outmigration, and recovered on bird colonies in the Columbia River basin (2008–2014). Recoveries on bird colonies are often ignored in survival studies because the river reach of mortality is often unknown, which we model as a form of state uncertainty. Median outmigration survival from release to the lower river (river kilometre 729 to 75) ranged from 0.27 to 0.35, depending on year. Recovery probabilities were frequently ≥0.20 in the first river reach following tagging, indicating that one out of five fish that died in that reach was recovered on a bird colony. Integrating dead recovery data provided increased parameter precision, estimation of where birds consumed fish, and survival estimates across larger spatial scales. More generally, these modeling approaches provide a flexible framework to integrate multiple sources of tag recovery data into mark–recapture studies.