A state-space model to quantify density dependence, demographic heterogeneity, and spatial synchrony in Grande Ronde Basin Chinook salmon populations

Abstract

Pacific salmon face different mortality sources throughout life, requiring monitoring and modeling at various life stages to understand the relative influences of regulating processes. For example, density dependence may be important for freshwater juveniles, whereas ocean conditions may drive later-life outcomes; in addition, delayed effects further complicate these phenomena. State-space models offer a flexible and robust approach to analyze these complexities when faced with uncertain data. We constructed a state-space model for Grande Ronde Basin (NE Oregon, USA) spring Chinook salmon (Oncorhynchus tshawytscha) that tracks the abundance of $\sim$30 cohorts from 4 populations, modeling variability in freshwater juvenile growth and survival using density-dependent relationships and stochastic process noise that acknowledges synchronous dynamics. Model substructures include rearing origin setting (i.e., natural vs. hatchery), juvenile life history type, and adult age-of-return to account for heterogeneity at these scales. The model was fitted to empirical information collected by a variety of monitoring projects and included an index of freshwater habitat capacity to scale density-dependent processes. We found evidence of early-life density-dependent survival and growth, with subsequent over-wintering and out-migration survival mediated by early-life growth rates. Rearing capacity and growth rates showed positive, though uncertain, relationships with the habitat index. Life stage-specific covariances were overwhelmingly positive, indicating among-population synchronous dynamics throughout life. Post-hoc analyses showed juvenile life history diversity is important for increasing productivity and that increasing habitat availability would reduce density dependence. Model posteriors reflect current understanding of life cycle dynamics for these populations which can parameterize simulations of future population status.