PATTERNS OF SPECIES RICHNESS AND DISTRIBUTION OF PTERIDOPHYTES IN RWANDA (CENTRAL-AFRICA) - A NUMERICAL APPROACH

Rwanda is one of the smallest African states, but its environmental co nditions are highly differentiated. Its pteridophyte flora is fairly r ich (173 taxa). The distribution of pteridophyte taxa in a grid of 146 squares, 7'30''x7'30'' in size, were analyzed numerically. The analys is of the DCA results permitted the distinction of seven local distrib ution types, which were then applied in the classification of grid squ ares by the constrained block clustering method. The results obtained permitted the delimitation of five floristic areas. Geographical regio ns were delimited by the block clustering method using twenty-two envi ronmental variables (different ranges of altitude, annual rainfall, de nsity of water-courses, and types of geological substratum and soils). Highly significant associations were found between the independently delimited floristic areas and the geographical regions. Stepwise regre ssion analysis was used-with nine independent variables taken into acc ount-to examine the influence of environmental conditions upon the spe cies richness of pteridophytes. It was found that the number of taxa i n the squares depended chiefly on the level of rainfall, whereas the n umber of vegetation types was of minor importance. The results of the analyses show that the humidity gradient is the main factor influencin g the differentiation of the flora and species richness of pteridophyt es in Rwanda. The block clustering method seems to be suitable for fin ding geographically and floristically homogeneous areas. With the spec ies groups fixed, the constrained version of block clustering appeared to be a very good method for classifying the squares.