Estimating breeding bird survey trends and annual indices for Canada: how do the new hierarchical Bayesian estimates differ from previous estimates?

Adam C. Smith, Marie-Anne R. Hudson, Constance Downes, Charles M. Francis


Canadian data from the North American Breeding Bird Survey (BBS) provide information on the population status and trends for over 300 species that regularly breed in Canada. Since the first assessments were made in the mid-1970s, both the dataset and the suite of statistical tools and techniques available to researchers have grown. As a result, Canadian BBS trend estimates have been derived from numerous statistical models. Because the BBS data are relatively complex, different statistical models can generate different trend and status estimates from the same data. In 2013, Environment Canada’s Canadian Wildlife Service began producing BBS status and trend estimates using a hierarchical Bayesian model. To give users of BBS trends and annual indices of abundance a better understanding of these estimates, we demonstrate and explain some of the similarities and differences between the new hierarchical Bayesian estimates and those from the previous model; discuss the philosophical and inferential consequences of estimating trends with the new model; and describe how the hierarchical Bayesian model differs from the model currently used in the United States. Overall, trends and annual indices from the new model are generally similar to estimates from the previous model; however, they are more precise, less variable among years, better represent the spatial variation across Canada in population status, and allow for more intuitive and useful assessments of uncertainty.


breeding bird survey; hierarchical Bayesian model; maximum likelihood model; population trend; population monitoring


Volumes that are more than six years old are freely available courtesy of the Biodiversity Heritage Library.


Questions or problems with the website? Contact William Halliday (info -at- canadianfieldnaturalist -dot- ca).