Application of multi-scale spatial and spectral analysis for predicting primate occurrence and habitat associations in Kibale National Park, Uganda

Remote sensing has made important contributions to biodiversity conservation planning through the measurement of deforestation rates, habitat fragmentation, habitat degradation, and isolation of protected areas in human altered landscapes. Progress in using remote sensing together with field data for the development of robust habitat suitability models has been more modest. In this paper, we present a habitat suitability model developed by relating field data on habitat characteristics and habitat selection of a medium-sized primate (redtail monkey, Cercopithecus ascanius) with a small home range (in contrast to many studies that focus on large-bodied, wide-ranging animals) to satellite-derived data, for predicting animal occurrence. We examined the relative performance of two high-resolution satellite sensors (2.5 m resolution Quickbird and 30 m resolution Landsat ETM+) in capturing patterns of habitat selection by redtail monkeys in Kibale National Park. Overall, our analyses suggest that a model combining data from both the Quickbird and ETM+ sensors predicts monkey presence–absence best, and that, individually, ETM+ data are better predictors than Quickbird data. Moreover, model fit was best at larger scales (ETM+ 120 m, 150 m and 240 m pixels). These results have important implications for future habitat modeling, biodiversity analyses and conservation studies given Landsatʹs temporal and spatial resolutions, long history of use, relatively well-developed methods for processing and analysis, and relatively low cost in comparison with many higher resolution sensors.