Two floodplain-associated wood-warblers segregate along a hydrologic gradient, in large river floodplains of the southeastern USA.
Prothonotary and Swainson's Warblers are neotropical migrants that breed in floodplains of large rivers in the southeast Piedmont and Coastal Plain. Although reported habitat requirements for these two species are quite different, habitat selection and breeding success for both are affected by the local flood regime. To understand the relationship of the two species' distributions to local hydrology and to one another, we combined observations made during extensive surveys of birds and plants in large river floodplains in Georgia, USA with a flood model built from direct observation of inundation by satellite-borne Synthetic Aperture Radar (SAR). We derived metrics describing components of local flood regime for 205 survey sites, and used Bayesian variable selection within logistic regression models to choose the most influential metrics for each species. Then we used model selection to find the best among a set of a priori models describing various mechanisms by which hydrology might influence Prothonotary and Swainson's Warbler occurrence, including indirect effects through vegetation or geomorphology. Finally, we built a multinomial logistic model to represent both species' occurrence patterns simultaneously.
Both species' distributions were best predicted by a model containing direct measures of local flooding, together with an estimate of conspecific density along a 10-rivermile stretch of river surrounding each site. For the Prothonotary Warbler, the annual average number of days flooded during April-May was the most important flood metric. Swainson's Warbler occurrence was negatively related to both the annual return interval of April-May flooding, and the average duration of flood events. Our two-species model shows that, given Swainson's Warblers are present in the 10-rivermile neighborhood, they occupy areas that see brief but consistently recurring flooding, whereas Prothonotary Warbler occurrence increases with the number of days of flooding in the breeding season. Where floods in April-May occur almost every year, and Swainson's Warblers are present in the neighborhood, the probability of encountering Swainson's Warbler exceeds that for Prothonotary Warbler below a flood duration of ~18 days (credible interval: [16,20]). Co-occurrence of the two species is predicted to be unlikely. Our models imply that considerable portions of hydrologically suitable habitat remain unoccupied, especially for Swainson's Warbler, indicating an opportunity for floodplain habitat restoration. Changing hydrologic conditions in the southeast will have disparate effects on these species.