Input data, analytical methods and biogeography of Elegia (Restionaceae)

Aim The aim of this paper is to determine the optimal methods for delimiting areas of endemism for Elegia L. (Restionaceae), an endemic genus of the Cape Floristic Region. We assess two methods of scoring the data (presence–absence in regular grids, or in irregular eco‐geographical regions) and three methods for locating biogeographical centres or areas of endemism, and evaluate one method for locating biotic elements. Location The Cape Floristic Region (CFR), South Africa. Methods The distribution of all 48 species of Elegia was mapped as presence–absence data on a quarter‐degree grid and on broad habitat units (eco‐geographical areas). Three methods to delimit areas of endemism were applied: parsimony analysis of endemism (PAE), phenetic cluster analysis, and NDM (‘end em ism’). In addition, we used presence–absence clustering (‘Prabclus’) to delimit biotic elements. The performances of these methods in elucidating the geographical patterns in Elegia were compared, for both types of input data, by evaluating their efficacy in maximizing the proportion of endemics and the number of areas of endemism. Results Eco‐geographical areas perform better than quarter‐degree grids. The eco‐geographical areas are potentially more likely to track the distribution of species. The phenetic approach performed best in terms of its ability to delimit areas of endemism in the study area. The species richness and the richness of range‐restricted species are each highest in the south‐western part of the CFR, decreasing to the north and east. The phytogeographical centres identified in the present study are the northern mountains, the southern mountains (inclusive of the Riviersonderend Mountains and the Cape Peninsula), the Langeberg range, the south coast, the Cape flats, and the west coast. Main conclusions This study demonstrates that (1) eco‐geographical areas should be preferred over a grid overlay in the study of biogeographical patterns, (2) phenetic clustering is the most suitable analytical method for finding areas of endemism, and (3) delimiting biotic elements does not contribute to an understanding of the biogeographical pattern in Elegia. The areas of endemism in Elegia are largely similar to those described in other studies, but there are many detailed differences.