Experimental biology: Sometimes Bayesian statistics are better

A worrying trend over the past decade has been the taxonomic splitting of mammal species, mostly by raising subspecies to species. Because of its potential bearing on conservation, we advise caution in this practice, which we maintain should be based solely on peer-reviewed evidence of biological validity. This trend is mainly the result of a shift from the biological to the phylogenetic species concept. The biological species concept holds that species are groups of (actually or potentially) interbreeding populations. The phylogenetic species concept and its variants, by contrast, define species either as the smallest cluster sharing genetically transmitted characters, such that all individuals are unequivocally diagnosable on the basis of those characters, or as monophyletic assemblages. In these, all individuals sharing a common ancestor belong to one species, with common ancestry inferred on the basis of shared derived characters (see, for example, As well as confusing the functional meaning of a species, taxon splitting could be detrimental to conservation. If threatened species are incorrectly split into several units and managed as such, for example in captive breeding or meta-population management, there could be unnecessary loss of genetic variation and an increased risk of extinction. Such newly designated species call into question the suitability of Red List assessments and the legality of species identified under national laws and international agreements. It is vital to identify true species as conservation units, based on adequate sample sizes and on information pertaining to genetics, morphology and behaviour. *On behalf of 6 cosignatories (see go.nature.com/4urduh for full list). David Vaux argues that experimental biologists should be better versed in classical statistics (Nature 492, 180–181; 2012). We suggest that they might also join the shift to Bayesian statistics that is already under way in many other areas of science. He defines the 95% confidence interval (CI) as " with 95% confidence, the population mean will lie in this interval " , adding that it is commonly used " to infer where the population mean lies, and to compare two populations ". However, a 95% CI merely tells us that if we were to sample from the population many times and calculate a 95% CI for each sample, 95% of the calculated CIs would, on average, contain the true population mean. Because classical statistics concern conditional probabilities of data based on assumed true parameter values (namely, the plausibility of the observed or more extreme data, given our …