Bird-habitat associations: Accounting for stop-level location uncertainty in the Breeding Bird Survey dataset
Many ecological inferences are based on the locations at which species are detected and on the characteristics (e.g., habitat, climate) of those locations. Yet many species records, particularly in historical datasets, lack precise location information. An example is the North American Breeding Bird Survey (BBS), a 55-year record of bird surveys that has revealed large-scale trends in bird populations but lacks precise survey-specific location coordinates. In this project, we develop and test a Bayesian modeling method that incorporates location uncertainty into analyses to understand local-scale habitat associations of forest and grassland birds in the central United States.
This study aims to: (1) quantify location uncertainty for BBS stops (i.e., bird survey events) and use this information to then quantify the resulting habitat uncertainty for each stop, (2) incorporate this information into Bayesian multi-species models using ‘informative priors’ (i.e., background knowledge*), and (3) use these models to quantify the responses of forest and grassland bird species to habitat at different spatial scales. Researchers will use land cover information, including the USGS National Land Cover Database, to classify and quantify land cover and habitat uncertainty around BBS bird survey routes and National Parks for use in these models. One application of this information will be in predicting the effects of forest and grassland fragmentation around National Park Units on bird communities in these areas. Breeding birds in North America have declined in abundance over the past 50 years, and grassland- and forest-dependent species are among those most affected. Grassland birds have declined by more than 50%, with nearly three of four species in decline. Forest birds have fared only slightly better, with overall rates of decline of 18%-30% and more than half of all species declining. Habitat fragmentation and loss are responsible for many of these changes. We will use models based on BBS data and that incorporate habitat uncertainty in order to predict bird prevalence in National Parks in the central United States, then test these predictions using bird count data collected by the National Parks Service Inventory and Monitoring Program.
This project is ongoing and collaborates with the USGS Breeding Bird Survey office, U.S. National Park Service, U.S. Fish and Wildlife Service, and the Wisconsin Department of Natural Resources. Funding is provided in part by the USGS National Resource Preservation Program (NRPP).
*informative priors are a way to incorporate background knowledge into a model
Many ecological inferences are based on the locations at which species are detected and on the characteristics (e.g., habitat, climate) of those locations. Yet many species records, particularly in historical datasets, lack precise location information. An example is the North American Breeding Bird Survey (BBS), a 55-year record of bird surveys that has revealed large-scale trends in bird populations but lacks precise survey-specific location coordinates. In this project, we develop and test a Bayesian modeling method that incorporates location uncertainty into analyses to understand local-scale habitat associations of forest and grassland birds in the central United States.
This study aims to: (1) quantify location uncertainty for BBS stops (i.e., bird survey events) and use this information to then quantify the resulting habitat uncertainty for each stop, (2) incorporate this information into Bayesian multi-species models using ‘informative priors’ (i.e., background knowledge*), and (3) use these models to quantify the responses of forest and grassland bird species to habitat at different spatial scales. Researchers will use land cover information, including the USGS National Land Cover Database, to classify and quantify land cover and habitat uncertainty around BBS bird survey routes and National Parks for use in these models. One application of this information will be in predicting the effects of forest and grassland fragmentation around National Park Units on bird communities in these areas. Breeding birds in North America have declined in abundance over the past 50 years, and grassland- and forest-dependent species are among those most affected. Grassland birds have declined by more than 50%, with nearly three of four species in decline. Forest birds have fared only slightly better, with overall rates of decline of 18%-30% and more than half of all species declining. Habitat fragmentation and loss are responsible for many of these changes. We will use models based on BBS data and that incorporate habitat uncertainty in order to predict bird prevalence in National Parks in the central United States, then test these predictions using bird count data collected by the National Parks Service Inventory and Monitoring Program.
This project is ongoing and collaborates with the USGS Breeding Bird Survey office, U.S. National Park Service, U.S. Fish and Wildlife Service, and the Wisconsin Department of Natural Resources. Funding is provided in part by the USGS National Resource Preservation Program (NRPP).
*informative priors are a way to incorporate background knowledge into a model