An integrated analysis for estimation of survival, growth, and movement of unmarked juvenile anadromous fish

Managers invest substantial resources to promote recovery of declining anadromous fish stocks. Recovery strategies are manifold and often include management actions intended to stimulate somatic growth, increase in-river survival, and motivate juvenile outmigration during favorable environmental conditions. Evaluating the efficacy of these management actions is difficult, however, because monitoring data that explicitly track individuals from egg deposition to juvenile outmigration are typically lacking. We developed an integrated population model that links two different and often collected types of anadromous fish monitoring data: spawning ground surveys and rotary screw trap juvenile catch data. The integrated model accounts for incomplete detection and uses the two sources of data to estimate juvenile demographic parameters in a multistate framework. We evaluated the model's performance using simulated data under a range of conditions typically encountered in similar surveys. Simulation results indicated that the model estimated juvenile survival, growth, and movement with no-to-minimal bias (i.e., ≥ 50 % of simulations ± 0–0.05). As an example case study, we fit the model to empirical fall-run Chinook Salmon (Oncorhynchus tshawytscha) monitoring data collected in California's Central Valley, U.S.A. In doing so, we evaluated the influence of environmental conditions (e.g., discharge, water temperature) and habitat availability on juvenile demographic rates. We demonstrated that through our integrated approach we could estimate state transition probabilities that are typically inestimable for naturally produced, unmarked juvenile fish when using traditional statistical approaches to analyze these types of monitoring data. Furthermore, the structure of our model can serve as a useful foundation for decision-support models within adaptive management programs by directly linking management actions, decision-support-model predictions, and monitoring.