Non-symmetric correspondence analysis: an alternative for species occurrences data

Abstract

Species occurrences gathered from the literature, from atlases or from field surveys are currently used to analyze multispecific patterns, such as species richness or species geographic ranges. Such occurrences result from the independent recognitions of specimens by several botanists in particular places and at particular occasions. Thereby, the analysis of the resulting occasional ‘releves’ involves the assignment of the species occurrences to spatial units such as a grid of quadrats. As a result, the distribution of occurrences among quadrats is controlled while their distribution among species is observed. In this paper we show how non-symmetric correspondence analysis (NSCA) enables the investigation of data structure by taking into account this fundamental asymmetry. We apply this new ordination technique to a list of endemic tree species occurrences in the Western Ghats (South India). We explore the interesting properties of NSCA as an ordination technique and demonstrate the usefulness of the method as a tool in biogeography. Regarding the Western Ghats, NSCA brings out the preponderance of deforestation over biogeographic history in explaining the observed multispecific patterns.