A state–space model for estimating fishing and natural mortality using live mark–recapture and dead mark–recovery data

Objective The objectives of this study were to develop joint tag-recovery models that make use of both live- and dead-tag-recovery data to estimate fishing and natural mortality and allow for an evaluation of hypotheses at the individual fish scale. Such models are useful for fisheries that have relatively high rates of catch-and-release fishing in addition to harvest. Methods State–space models were developed that make use of tag-return data from fish that were harvested and/or released in a joint-likelihood model using Bayesian methods. The accuracy of the models was assessed using simulation techniques. Additionally, the models were fit to data from four fisheries in Idaho, including Smallmouth Bass Micropterus dolomieu in Dworshak Reservoir and Lake Pend Oreille, Yellow Perch Perca flavescens in Cascade Reservoir, and Lake Trout Salvelinus namaycush in Priest Lake. In addition to the null model, which contained time- and individual-invariant parameters, a time-varying model and a model that allowed fishing and natural mortality to vary based on length at tagging were also fit. Results The simulation results suggested that the state–space models were relatively unbiased and the models that used both live- and dead-tag recoveries were more accurate than those that used only harvest data. Estimated natural mortality was greater than fishing mortality at all four fisheries in the study. The direction and magnitude of the effect of length on fishing and natural mortality varied among the fisheries. The model that included length was the top model for Smallmouth Bass in Dworshak Reservoir, Yellow Perch in Cascade Reservoir, and Lake Trout in Priest Lake, whereas the time model was the top model for Smallmouth Bass in Lake Pend Oreille. Conclusions The state–space models described in this study provide a framework for estimating the demographic parameters that is more flexible than those of the previously published instantaneous rate models. The models are easily extensible to evaluate numerous hypotheses or accommodate additional data that can greatly benefit management or conservation of fisheries with concurrent harvest and catch-and-release components.