Mémoire numérisé par la Direction des bibliothèques de l'Université de Montréal.

In this study, geographic information systems (GIS) and remote sensing (RS) tools were integrated and used to investigate the plant species diversity of the Nallihan forest ecosystem. Two distinct indices, Shannon Wiener and Simpson, were employed in order to express species diversity. The relationships between the indices and pertinent independent variables (topography, geology, soil, climate, supervised classes, and Normalized Difference Vegetation Index (NDVI) classes) were investigated to develop two distinct models for each index. After detecting important components with factor analysis, two models were developed by using multiple regression statistics. Running the models, two plant species diversity maps in grid format were produced. The validity of the models were tested by (1) mapping residuals to predict the locations where the models work perfectly, and (2) logical interpretations in ecological point of view. Elevation and climatic factors formed the most important component that are effective on plant species diversity. Geological formations, soil, land cover and land-use characteristics were also found influential for both models. Considering the disturbance and potential evapotranspiration (PET), the model developed for Shannon Wiener index was found out more suitable comparing the model for Simpson index.