Ecology and Development Series No. 10, 2003 - ZEF

Conservation of the wild Coffea arabica populations in situcan be used to predict distribution patterns of species and diversity (Reyers et al. 2002). Anumber of techniques are used for prediction and mapping of plant species distributionwithin a landscape. Three most often used methods: (1) mosaic or block diagrams, (2)regression models, and (3) geostatistical techniques like kriging and co-kiriging.The digital elevation model (DEM) provides topographic information such aselevation, slope, landform, aspect, etc. Such terrain-derived data can be used to map actual orpotential vegetation across the landscape. Bolstad et al. (1998) used a terrain-based linearregression to predict and map overstorey tree species in a mountainous region. Several otherresearchers also similarly used terrain based data for vegetating mapping (e.g., Meentemeyeret al. 2001; Guisan et al. 1999; Ostendorf and Reynolds 1998). This study aims to find outthe spatial distribution patterns of the abundance of coffee and diversity index, and to classifythe forest into different reserve zones based suitability of the forest areas for conservation ofthe wild coffee population and plant species of the forest.6.2 Materials and methods6.2.1 Data sourcesForest vegetation data and terrain data were used in this analysis. For prediction of thedistribution patterns of the abundances of coffee and species diversity index, vegetation datafrom undisturbed forest were used. The forest vegetation data were those described inChapter 4.2.1 of this thesis and have been collected from 58 sample plots. However, the dataof one sample plot (Plot68), which was an outlier in the cluster analysis described in Chapter4.2.2, was excluded. Hence, only the vegetation survey data of 57 sample plots were used.The types of vegetation data used in the analysis were species abundances, species diversityindex and the site scores of the Principal Component analysis (PCA) axes I and II.Abundance refers to the number of stems per sample plot, while the species diversity indexused in this analysis is the Shannon diversity index, since it crystallizes both species richnessand evenness into one figure (Magurran 1988). The site scores of axes I and II of the PCAwere obtained from the results of the analysis presented in Chapter 4.3.3 of this thesis. Theterrain data were derived from a digital elevation model (DEM), which in turn was developedfrom topographic maps. The area of topographic map used in DEM development includes theundisturbed forest and the surrounding managed forests and farm lands (793870-830490 m E101