Biodiversity patterns in Indian Ocean corals, and effects of taxonomic error in data

Abstract

Corals from 26 sites in the Indian Ocean, from numerous taxonomic sources, are analysed for distribution patterns after applying a consistent synonymy. The data set contains double the amount of distribution information used previously. Cluster analysis shows several contiguous regional groupings within this ocean, and no geographically dispersed groupings exist. Between-site species similarity correlates negatively with distance between sites. Coral species diversity and genus diversity plotted with latitude and longitude confirm that a band of high diversity stretches across the entire Indian Ocean, but that in this Ocean, unlike the Pacific and greater Caribbean area, there is no clear gradient with latitude at least up to the latitudinal limits of coral growth. About half the species are widespread throughout the Indian Ocean. Species frequency/distribution curves and cumulative frequency curves show that 150 species occur at only 2–4 sites. Multi-dimensional scaling (MDS) analysis using only these species confirms that they are important in forming sub-regional groupings which are superimposed on a general Indian Ocean homogeneity. The perennial problem of error in taxonomic data sets is examined. Increasing taxonomic error is introduced into the data using random methods. It is found that about 25% more error can be added to this data set before the relationship between similarity coefficient and geographical distance between pairs of sites is lost. Measures extracted from the clustering procedure using the original data and the data sets with added taxonomic error, show a sharp loss in cluster formation after addition of about 10 or 20% more error.