A GIS-based model of Serengeti grassland bird species

Abstract

In this study we assess possible benefits of using satellite sensor data in large-scale landscape ecology. The study was conducted on the Serengeti Plains, Tanzania, combining (1) records from a bird survey, (2) local measurements of vegetation structure and precipitation, and (3) a habitat map derived from a Landsat satellite image classification. The question of whether ground-based or satellite data explained more of the species-environment relationships was explored by means of multivariate regression. On average across all 62 bird species recorded, the combination of satellite-based and groundbased data improved explained variance (R² = 0.26), as compared to satellite sensor data, or ground-based data alone (R² = 0.18 and 0.21, respectively). In spite of this low level of explained variance in the regressions, a classification of bird species according to utilised parameter space yielded reasonable results. Satellite image data seem to be suited to this kind of investigation.

Ostrich 2007, 78(2): 259–263