Spatial modelling, phytogeography and conservation in the Eastern Arc Mountains of Tanzania and Kenya

Forests in the Eastern Arc Mountains are amongst the oldest and most biodiverse on Earth. They are a global priority for conservation and provide ecosystem services to millions of people. This thesis explores how spatial modelling can provide direction for conservation and botanical survey, and contribute to understanding of phytogeographical relationships. The ecoregion is rigorously defined by terrain complexity, vegetation distribution and established geoclimatic divisions, providing a coherent platform upon which to collate and monitor biological and socioeconomic information. Accordingly, 570 vascular plant taxa (species, subspecies and varieties) are found to be strictly endemic. The human population exceeds two million, with median density more than double the Tanzania average. Population pressure (accrued across the landscape) is shown to be greatest adjacent to the most floristically unique forests. Current knowledge on species distributions is subject to sampling bias, but could be systematically improved by iterative application of the bioclimatic models presented here, combined with targeted fieldwork. Tree data account for 80% of botanical records, but only 18% of endemic plant species; since conservation priorities differ by plant growth form, future fieldwork should aim to redress the balance. Concentrations of rare species correlate most strongly with moisture availability, whilst overall richness is better predicted by temperature gradients. Climate change impacts are projected to be highly variable, both across space and between species. Concordant with the theory that past climatic stability facilitated the accumulation of rare species, contemporary climates at sites of known endemic richness are least likely to be lost from dispersal-limiting mountain blocs during the 21st century. Faced with rapid population growth and the uncertainty of climate change, priorities for governance are to facilitate sustainable forest use and to maintain/restore habitat connectivity wherever possible. Overall, the thesis demonstrates that decision makers concerned with biodiversity conservation, particularly in mountain and coastal regions, should be wary of inferring local patterns of change from broad-scale models. The current study is a step toward spatially refined understanding of conservation priorities in the Eastern Arc Mountains, whilst novel methodologies have wider application in the fields of species distribution modelling and mountain analysis.