Movement and growth in humpback chub: Using multistate models for inference from a mixed batch and individual tag dataset

Background/Question/Methods

The federally endangered humpback chub (Gila cypha) persists in six populations distributed throughout the Colorado River Basin and its tributaries. The largest population is dispersed downstream of Glen Canyon Dam throughout Grand Canyon National Park, where most individuals are found either in the Little Colorado River (LCR) or in reaches of the mainstem Colorado River  near the LCR confluence. While adults are known to migrate between spawning grounds in the LCR and return with remarkable site fidelity to locations within the mainstem, movements of juveniles have not been systematically studied. Understanding movement (adult and juvenile) is important both for designing and interpreting population surveys and for understanding how variation in the LCR and mainstem environments affects survival, growth and ultimately population size. More broadly, partial migratory systems suggest that there are life history tradeoffs that are dependent on fish size. Here, we use Bayesian multistate models to analyze three years of batch and individual mark-recapture data collected within the LCR and in a nearby reach of the mainstem to gain a better understanding of movement rates and their dependency on individual size. These models also allow us to compare apparent survival and growth rates between sampling locations.

Results/Conclusions

We describe and fit a multistate model where states are based on size class and location and where states are observed imperfectly. As predicted, movement rates vary systematically with size class. We also find evidence that apparent survival and growth rates vary between LCR and mainstem portions of the population. We interpret these results with respect to the ecology of G. cypha, in particular, and more generally in terms of other partial migratory systems. Future research will focus on incorporating ongoing sampling at the same locations as well as other sources of data, including otolith data, to gain a better understanding of temporal variation in state transitions.