Conservation of Avian Resources Through Species Distribution Models and Strategic Habitat Conservation

This dataset was created in 2013 using Breeding Bird Survey Data (www.pwrc.usgs.gov/bbs) from 2005 – 2011. Stop locations within BBS routes were validated by the Rainwater Basin Joint Venture and species count data at each stop were used in the statistical analysis. We used Poisson or binomial (depending on species) generalized linear mixed models in [R] statistical software (R Development Core Team. 2013. A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.)) package ‘lme4’ (http://cran.r-project.org/web/packages/lme4/vignettes/Theory.pdf) to model species abundance in relation to land cover, climatic, and elevation spatial data. An information theoretic approach (AICc) was used to test a model set containing a priorimodel hypotheses. The parameter estimates from the top ranked model were entered in to ESRI’s ArcGIS raster calculator to produce a spatially-explicit species distribution model for each species.

In order to help understand the predictive power of this dataset, we used a 10-fold cross validation method to refit our top ranked model with 90 percent of the BBS data and validated the re-parameterized model with the remaining 10 percent of the data. This process was repeated 10 times (10-fold) to calculate a Root Mean Square Error (RMSE), the Mean Absolute Error (MAE), and the Pearson’s Correlation Coefficient Squared (Cor^2).