Analysis of vegetation fragmentation and impacts using remote sensing techniques in the Eastern Arc Mountains of Tanzania.
The Eastern Arc Mountains of Tanzania forms part of the Eastern Afromontane Biodiversity Hotspots, listed among the global World Wide Fund for Nature's (WWF) priority ecoregions. However, the region is threatened by fragmentation and habitat modification resulting from competing forms of land uses, which is in turn threatening biodiversity conservation, planning and management efforts. To determine vulnerability that can inform long-term conservation and management of the biodiversity hotspots, an in-depth understanding of the qualitative and quantitative nature of ecosystems is a pre-requisite. The overall goal of this study was to quantify fragmentation, investigate its impacts on tree species diversity, abundance and biomass and to identify management interventions in the Eastern Arc Mountains of Tanzania. Using ecological field based measurements and a series of LANDSAT and RapidEye satellite imagery, fragstats metrics showed dynamic fragmentation patterns at both spatial and temporal scales. Furthermore, species diversity was predicted better with customized environmental variables using the Generic Algorithm for Rule-Set Prediction (GARP) model, which recorded an Area under Curve (AUC) of 0.89. In addition, Poisson regression results showed different responses by individual tree species to patch area dynamics, habitat status and soil nitrogen. Partial Least Squares and Random Forest models were used to determine above ground biomass prediction based on a combination of edaphic variables and vegetation indices. Total biomass estimations decreased from 1162 ton ha-1 in 1980 to 285.38 ton ha-1 in 2012. As a reference point in formulation of policy insights based on strong scientific and empirical knowledge, socio-economic factors influencing vulnerability of ecosystems and management interventions were examined using remotely sensed and empirical data from 335 households. The multiple logistic regression model indicated habitat fragmentation and forest burning as key conservation threats while low income level (54.62%) and limited knowledge on environmental conservation (18.51%) were identified as major catalysts to ecosystem vulnerability. The study identified livelihood diversification, effective institutional frameworks and afforestation programmes as major intervention measures. The overall study shows the effectiveness of remote sensing techniques in ecological studies and how results can be used to inform decisions for addressing complex ecological challenges in the tropics.