Proper statistical treatment of species-area data

The purpose of this report is to comment on the entire process of analyzing species-area data, particularly as performed by Rydin and Borgegaard (1988). They use three different models to test species-area relations for islands over a 100 year period. Several aspects of their analysis of species-area data could be improved, including their comparison of goodness-of-fit and testing of the expected value of z. The reason that these issues are important (their basic conclusions being correct) is that there is acrimonious debate over the best model to use for species-area curves and over whether the scope coefficient is constant or is an artifact, and because the species-area curve is being used for nature reserve design. The problems pointed out here are common to a large class of allometric-type analyses in ecology. The author attempts to show the potential pitfalls inherent in allometric analyses and demonstrate methods for avoiding these problems