Majka, D. 2005. Comparison of GIS-based modeling methods in predicting local avian distributions in the montane neotropics. In GIS-based modeling of avian distributions in a montane tropical forest. MS Thesis. Purdue Univerisity.

      In this chapter, Majka compared five different distribution modeling techniques: Generalized Linear Models (GLM), Generalized Additive Models (GAM), Artificial Neural Networks (ANN), Genetic Algorithm for Rule-set Production (GARP), and Ecological Niche Factor Analysis (ENFA), on 41 species distributed in the Tilaran Mountains, Costa Rica. To compare the model techniques, he used nine topographic predictor variables obtained from a NASA’s digital model Shuttler Radar Topography Mission (SRTM), and bird data collected in the field. In general, the results showed that simple techniques (GLM) performed the best to predict species occurrence. He also found that models using presence-absence data performed better than models using presence only data. Based on his results, Majka suggests using an Occam’s razor approach to select modeling techniques, especially if the distribution modeling is applied to small scale or along a gradient; thus simple techniques are the most favorable to model local and /or gradient distribution.  

More about Daniel Majka