Body size and fitness in Odonata, stabilising selection and a meta-analysis too far?

Sokolovska et al. (2002) used 57 weighted correlation coefficients derived from 33 published studies of 20 species to produce the first meta-analysis of body size and fitness in odonates. They concluded that 'there is a general fitness benefit to large size of odonates'. Given the inherent biases of their analyses, such a conclusion is hardly surprising. In this paper, their study was used to illustrate major problems that arise from (1) ignoring the possibility of stabilising selection on body size, (2) using statistics from the literature indiscriminately, when the original studies are inappropriate for comparative analysis, and (3) relying heavily on transformed statistics derived from small samples. Suggestions are made as to how data can be presented and analysed in ways conducive to future comparative analyses on size relations to fitness in odonates and other insects. [ABSTRACT FROM AUTHOR]