The Bray-Curtis similarity index remains misidentified: Reply to Somerfield (2008)

Abstract

Somerfield (2008, Mar Ecol Prog Ser 372:303-306) provides numerous reasons for refuting my contention that the Bray-Curtis (BC) similarity index is misidentified by some software programs. He argues that I confuse the pretreatment of data with the calculation of similarity, and ends with the assertion that software programs are not the problem. I argue that the pretreatment of data is 'built into' the BC index, and that software packages are a major source for propagation of the misidentified BC index. Much more importantly, I think that Somerfield's arguments are peripheral to, and do not disprove, the key contention that the index in some software programs differs from the index described by Bray & Curtis (1957). Nevertheless, I agree with Somerfield's sentiment that the term Bray-Curtis should be maintained in community ecology. With respect to indices of percentage similarities I think that Bray and Curtis deserve greater recognition for developing a double standardization procedure that assigns equal weights to species and samples. © Inter-Research 2008.