Classifying boreal landbirds to facilitate multi-species approaches to national-level conservation and forest degradation analyses

Forest landbirds have high potential as indicators of changes in forest state including forest degradation. We used Canada-wide landbird survey data and remotely sensed land cover information to demonstrate a workflow for classifying 117 landbirds according to local (within 200 m) and regional (within 2 km) habitat preference across 16 ecological regions. We quantified preference as habitat use relative to availability along three major axes of forest variability: percent cover, proportion conifer, and forest age. Data files provided are: (1) "predictionlayer2025.tiff": the stacked, derived, 1-km rasters used for prediction; (2) "BCR boundaries": BCR/ecoregion shapefile of the boundaries used in our regional analyses (2) "TabulatedPredictedDensity.csv": the ecoregion (=BCR) density predictions (cumulative density =cdens) for all species tabulated across each value unit (=Val) of our covariate range (covariate ID = Var) for 32 bootstraps (=boot); (3) "GAMfit_allSppByBCRs.csv": the global HGAM fits (fit) and associated simultaneous 90% CI (=sim) and pointwise (=pw) 95%, 90% and 75% CIs of the natural log of the use:availability ratio of each value unit (=value) of our covariate range derived over all bootstraps for each species, ecoregion (=BCR), and covariate (=type) In both datafiles regional scale covariates are labelled as "Land" and local scale covariates are labelled as "Loc"