A generalized modeling framework for spatially extensive species abundance prediction and population estimation
BAM's landbird models bridge the gap between local studies and large-scale management needs by compiling and harmonizing data from many sources to predict avian abundance at a fine resolution and broad extent.
Version 4 of BAM's landbird models (Stralberg et al. In Review) can be reproduced using the data object stored on Zenodo and the generalized model script available in this repository. Please see Version 5 below for additional steps in the modelling and interpretation workflow.
Version 5 of the modelling workflow is currently under active development and is available in this repository.
New developments to BAM's Landbird Models for Version 5 include:
- Expanded study area to include hemi-boreal regions of the continental United States using natural biographic boundaries rather than political boundaries to delineate regions
- Expanded list of environmental covariates including time-matched variables for vegetation biomass, human disturbance, and annual climate
- Acquisition and inclusion of new datsets for data-sparse regions, including appropriate eBird data
- Predictions for five year intervals from 1985 to 2020
- More rigorous model tuning and covariate thinning
Stralberg, D., P. Solymos, T.D.S Docherty, A.D. Crosby, S.L. Van Wilgenburg, S. Hache, E.C. Knight, L. Leston, J.D. Toms, J. Ball, S. Song, F. Schmiegelow, S.G. Cumming, and E.M. Bayne. In Review. A generalized modeling framework for spatially extensive species abundance prediction and population estimation: an example for landbirds in subarctic Canada. Submitted to Ecosphere.