Abstract

Machine learning (ML) models are a powerful tool for land use and land cover (LULC) mapping. In the African tropics, and particularly in the Congo Basin, there is a need to better assess the performance and reliability of ML-based LULC classification using coarse-resolution satellite images. In the context of ongoing climate change and socioeconomically-driven forest disturbances, it is important to understand and quantify the extent of forest cover loss in the Congo Basin, as well as the impact of this loss on suitable habitat for key wildlife species. In this dissertation, I address these key issues in three manuscript-based chapters. In Chapter 2, I compared the classification performance of four ML algorithms (k-nearest neighbor (kNN), support vector machines (SVM), artificial neural networks (ANN), and random forests (RF)) for LULC mapping within a tropical region in Central Africa (the Mayo Rey department of northern Cameroon). All four classification algorithms produced high accuracy (overall classification accuracy > 80%), with the RF model (> 90% classification accuracy) outperforming the other algorithms. In Chapter 3, I used the RF model, together with the Idrissi TerrSet land change modeler, to map and project LULCC for the Congo Basin under historical and future scenarios of socioeconomic impacts and climate change. I found that over 352642 km2 of dense forests have been lost in this region between 1990 and 2020, with projected continued loss of about 174860 - 204161 km2 by the year 2050. In Chapter 4, I produced spatially explicit species distribution models to map habitat suitability for great apes (chimpanzees and gorillas) and elephants within the Dzanga Sangha Protected Areas (DSPA) of the Congo Basin. I found that priority habitat areas for the three mammal species mostly occurred and overlapped spatially within the DSPA national parks. However, priority habitat areas for the three species declined by 4, 4.5 and 9.8 percentage points respectively between 2015 and 2020, mostly due to increased human pressures. This research provides a new understanding of the extend and implications of forest cover loss in the Congo Basin, highlighting the critical conservation challenges that remain in this region.