A Range-wide Assessment of Migratory Connectivity for the Prothonotary Warbler Using Stable Hydrogen Isotopes

Populations of many species of migratory birds are declining, and an understanding of how populations are linked between the breeding and nonbreeding grounds is necessary in order to determine drivers of declines. While all current tracking technologies to study the movements of small songbirds are limited by their coarse resolution, tracking birds by measuring stable isotopes in keratin offers an advantage because it requires only a single capture and is cost-effective, which leads to robust sample sizes. While this tracking method is accurate, stable hydrogen isotope values measured in feathers (δ2Hf) are known to be variable within a site, and may be influenced by the hydrology of a site. In this study, we assessed sources of variation in δ2Hf values in a wetland-associated Neotropical migratory bird, the prothonotary warbler (Protonotaria citrea), by comparing δ2Hf values among ages, sexes, years and feather type. We found that age and year significantly influence δ2Hf values, and that differences between primaries and rectrices are statistically significant but generally small. We also tested the accuracy and precision of models to assign prothonotary warblers to their breeding origin using known-origin feathers. We assigned birds in a spatially-explicit manner using an interpolated surface of stable hydrogen values measured in precipitation (δ2Hp), and developed a species-specific calibration equation to account for the offset between δ2Hf and δ2Hp, which explained 51% of the variation in δ2Hf values. We incorporated breeding bird abundance as prior information in our assignment model, and compared two sources of abundance data: the Breeding Bird Survey and a spatio-temporal exploratory model developed with eBird data. The assignment model with no prior information was accurate (83% of birds correctly assigned to their true location of origin), but imprecise (50% of grid cells assigned as likely locations of origin). Incorporating abundance as prior information led to a decrease in accuracy (9-14% of birds correctly assigned) but higher precision (1% of grid cells assigned as likely). We also assigned prothonotary warblers to their breeding origin using feathers collected from across the nonbreeding range. We found that all nonbreeding sampling locations contained a mixture of birds whose origins spanned the majority of the breeding range, with no evidence of strong connectivity between the seasons. In the absence of strong connectivity, the influence of events occurring at one nonbreeding location will be widespread and diffuse across the breeding range. For prothonotary warblers, understanding connectivity can help conservation planners understand how nonbreeding season habitat destruction and other processes influence population dynamics. We recommend that future studies account for age and year variation in δ2Hf values when possible, and continue to examine the trade-off between precision gained and accuracy lost when using abundance as prior information