Design and Analysis of Surveys for Estimation of Temporal and Spatial Change in Animal Populations
Designing and analyzing large-scale animal surveys is an important focus of our research. Although we conduct research into analysis methods for many surveys, the primary focus of this project is to conduct analyses and develop web-based summaries of data from the North American Breeding Bird Survey (BBS).
The Challenge: Population status information is required for management of migratory bird populations, and structured decision making and adaptive management place additional emphasis on the need for rigorous survey designs and robust estimation methods. The North American Breeding Bird Survey (BBS) and Christmas Bird Count (CBC) provide continent-scale information on breeding and wintering populations of >450 species of North American birds, and for many species these two surveys are our only data source for population status and trend information. Appropriate analyses of these important surveys require sophisticated methods to accommodate variation in survey efficiency over the large areas covered by the surveys and to control for factors that influence detection of birds. Factors such as observer quality and effort, if not appropriately controlled for in the analysis, can lead to biased estimates of population change.
The Science: We are developing and implementing hierarchical log-linear models for analysis of BBS and CBC data. These models control for observer differences in counting along BBS routes by treating observers as random effects and including a start-up effect that models lower counts for the first year an observer surveys a route. Effort effects in CBC data are controlled for by including a 2-parameter model component, that allows the effort adjustment to vary by species or region. For both of these surveys, hierarchical components control for regional differences in survey quality, allowing for estimation of composite population indices at the continental scale.
The Future: Hierarchical models have been fit to BBS and CBC data, and comprehensive analyses of BBS data are updated yearly and provided to scientists and the public.. These analyses provide a reasonable means of summarizing these complicated data sets. Ongoing studies involve implementing analyses using alternative models that expand the scope of the survey, incorporation of spatial effects, covariates, integrating multiple surveys to extend scope of inference, and modeling complex dynamical features. Hierarchical model results for BBS and CBC are featured in State of the Birds reports (http://www.stateofthebirds.org).
Below are partners associated with this project.
Designing and analyzing large-scale animal surveys is an important focus of our research. Although we conduct research into analysis methods for many surveys, the primary focus of this project is to conduct analyses and develop web-based summaries of data from the North American Breeding Bird Survey (BBS).
The Challenge: Population status information is required for management of migratory bird populations, and structured decision making and adaptive management place additional emphasis on the need for rigorous survey designs and robust estimation methods. The North American Breeding Bird Survey (BBS) and Christmas Bird Count (CBC) provide continent-scale information on breeding and wintering populations of >450 species of North American birds, and for many species these two surveys are our only data source for population status and trend information. Appropriate analyses of these important surveys require sophisticated methods to accommodate variation in survey efficiency over the large areas covered by the surveys and to control for factors that influence detection of birds. Factors such as observer quality and effort, if not appropriately controlled for in the analysis, can lead to biased estimates of population change.
The Science: We are developing and implementing hierarchical log-linear models for analysis of BBS and CBC data. These models control for observer differences in counting along BBS routes by treating observers as random effects and including a start-up effect that models lower counts for the first year an observer surveys a route. Effort effects in CBC data are controlled for by including a 2-parameter model component, that allows the effort adjustment to vary by species or region. For both of these surveys, hierarchical components control for regional differences in survey quality, allowing for estimation of composite population indices at the continental scale.
The Future: Hierarchical models have been fit to BBS and CBC data, and comprehensive analyses of BBS data are updated yearly and provided to scientists and the public.. These analyses provide a reasonable means of summarizing these complicated data sets. Ongoing studies involve implementing analyses using alternative models that expand the scope of the survey, incorporation of spatial effects, covariates, integrating multiple surveys to extend scope of inference, and modeling complex dynamical features. Hierarchical model results for BBS and CBC are featured in State of the Birds reports (http://www.stateofthebirds.org).
Below are partners associated with this project.