The North American Breeding Bird Survey Results and Analysis 1966-2009


Technical Notes:


(1) This is our first implementation of the hierarchical model analysis. This analysis approach has been described in the peer-reviewed literature (Link and Sauer 2002, Sauer and Link 2010), and a comprehensive comparison of the hierarchical model results with results from the route regression method used in earlier summaries is available at this link . These results have been reviewed, but we greatly value your opinions about the new analysis and results. Remember the disclaimer!, and please let us know about errors or other issues that limit the use and value of this site.
(2) As in earlier summaries, the primary quantities we present are trends (% change/yr) and annual indices (yearly composite population indexes). However, the hierarchical model approaches to estimating these quantities are very different from route regression approaches, and the estimates, although generally consistent with earlier analyses, are displayed somewhat differently than in the past. Hierarchical model analysis results are shown in terms of estimates and credible intervals (Bayesian Confidence Intervals). Precision of estimates are summarized as credible intervals rather than p values. Sample sizes are now listed as the number of routes on which the species was ever observed; in earlier analyses we presented the number of routes on which trends could be estimated. Finally, relative abundance (i.e., the level of the annual indexes) is now estimated in the model rather than taken as a simple average of counts as was done in earlier analysis.
(3) The hierarchical models provide credible intervals for annual indices, and hence the graphics that display annual indices also present credible intervals.Earlier summaries of annual indices did not provide precision estimates for annual indices. Unfortunately, the additional information in the graphics sometimes leads to "visual clutter," as the graphics must be scaled to show both the annual index and the credible intervals. Please note that annual indices are often imprecise, and graphic displays often omit portions of the upper credible interval.
(4) Maps (e.g., distribution maps) are not presented in this release of the website. We have been integrating relative abundance and trend maps into species accounts, which will be submitted to an online Journal and ultimately linked to this website. However, older versions are still available on the archived websites from earlier analyses.
(5) Some species are grouped for analysis. These are indicated on the selection lists.
(6) Annual index data are presented as graphical images, and are accessed via links from the species trend results. Actual data for these indices can be calculated in the regional analysis website.
(7) Credibility codes for trend estimates have been modified to accommodate the structure of the hierarchical model analysis. See the help file associated with the credibility code to see these modifications.
(8) Results are provided for states, Provinces, Bird Conservation Regions, BBS regions, Canada, United States (lower 48), and Survey-wide (excluding Alaska and NF). Due to analytical limitations, this initial summary using hierarchical models does not provide results for some regions analyzed in earlier summaries. In particular, analysis results are no longer presented for the Breeding Bird Survey Physiographic Regions.