Effects of the analysed variable set composition on the results of distance-based morphometric surveys

Abstract

Abstract Distance-based morphometry is still widely used in ichthyology. Beside others, this methodology is often used to characterise species or to compare intraspecific group (i.e. population level) differences. However, scarce information is available about: (a) which are the most widely used variables for these purposes, (b) how certain variables are selected for the morphometric analyses, and (c) how variable set compositions and the number of variables affect the results. To answer these questions, a literature review was compiled and our own three datasets were analysed. The results showed that although a high number of variables can be used, previous authors have measured “common” ones most often, regardless of the taxonomic position of the studied group and the goal of the survey. Additionally, our review showed that authors rarely made a variable selection and often did not standardise their datasets; these are methodical problems that make the accuracy and usability of the results questionable. Analyses of our own three datasets showed that the number of variables and the variable set compositions in most cases strongly influenced stock subdivision and the percentage of correctly classified individuals. It was also shown that the most useable variable sets for morphometric purposes can differ considerably depending on the taxon and goal of the survey.