Résumé
Statistical methods for delimiting species with morphological data remain woefully understudied. The present study presents statistical approaches to study the intraspecific variations of continuous characters used in species diagnosis. For this, the data set comprised 14 species of the mite family Phytoseiidae (Acari: Mesostigmata) and 23 continuous characters (seta lengths). The statistical approaches aim to determine (i) how intraspecific variations differ depending on the characters and species considered, and (ii) how to choose the number of specimens to be considered for characterizing intraspecific boundaries. The means of the related confident limits at 95% (rCL95) of the 14 species and the 23 setae are around 20%, suggesting that 95% of the specimens of a same species are included in the interval mu +/- 0.2 mu. No relationship was found between seta variations and the number of specimens considered for each species, nor for seta variations and their position on the mite body. However, a correlation between seta lengths and rCL95 means was observed. An abacus was proposed to determine the coefficient that must be applied to mean seta length to calculate the lower and upper limits of the interval that will include 95% of the specimens of a same species. The higher the precision around the mean, the higher the number of specimens considered. Furthermore, this number varies considerably depending on the characters and species considered for high precisions. However, for a priori error around the mean of 15%, the number of specimens needed to characterize the intraspecific variation is quite homogeneous and ten specimens would be required. I discuss the utility of such results to resolve some persistent problems in the alpha-taxonomy of the family Phytoseiidae but also of other taxa. (c) The Willi Hennig Society 2012.