Repo documenting the entire workflow used to generate the Canadian Wildlife Service's analysis of population status and trends from the North American Breeding Bird Survey, 1970-2022. Using the 2023 release version of the North American Breeding Bird Survey, and the spatially explicit GAMYE model defined in this paper.
The R-scripts are numbered to indicate the workflow. In general, each script requires that the previous scripts have been run (e.g., 1_bbs_script_doparallel.R depends on output created in 0_species_list_prep.R).
Note: these analyses require significant computational resources. This full set of scripts were conducted on a computer with 64 cores and 128GB of RAM. Fitting the Bayesian models using Stan is done in script 1_bbs_script_parallel.R, which requires the greatest time and resources (minimum 4-cores for a single species). There are numerous uses of the packages foreach and doParallel to analyse and/or summarise across many species. These foreach loops can be replaced with a for loop if analysing or summarising output for a single species.
The analyses use the R-package bbsBayes2.
The summaries and functions in all scripts after 1_bbs_script_parallel.R (e.g., all summaries of annual indices, trends, maps, etc.) can be run using saved model output for a given species. If you would like access to that saved model output for this analysis, please get in touch. In the near future, we are hoping to have a simple way to archive this saved model output in an openly accessible format, so check back here for updates.
This Google Drive contains all estimates of trends and indices of annual abundance (population trajectories) in a downloadable format. It also includes population trajectory graphs, maps of estimated trends for long-term, short-term, and three-generation periods, plus other saved output. There are readme files in both English and French in the folder that explain the contents in more detail.