Fluctuating asymmetry and sexual performance in the drumming wolf spider Hygrolycosa rubrofasciata

Abstract

Recently, there has been much interest in estimating fluctuating asymmetry (FA) of morphological traits as a short-cut measure of individual quality. FA deals with small differences around the symmetry value of zero. Thus, measurement error is often relatively large. However, repeated measurements. and large sample sizes allow reliable estimates of FA that can be corrected for errors. The purpose of this study was two-fold: at the biological level, we examined with a large sample size (N = 804) whether pedipalp FA could be used as a short-cut measure, of individual quality in the wolf spider Hygrolycosa rubrofasciata. This was done by estimating how strongly FA correlates with male sexual performance, i.e. drumming rate and mobility. At the statistical level, we examined how the deflating effect of measurement error could be eliminated to get unbiased correlations between FA and any repeatable trait. We also examined which is the more economical way to get accurate population-level estimates of FA-trait associations, to increase sample size or the number of within-subject repeats. Our results show that there was a very weak, but significant negative relationship between pedipalp FA and mobility (Kendall's partial r(k) = -0.086). However, this value inevitably underestimates the true relationship, given the large measurement error. It is possible to estimate the unbiased relationship by correcting the above correlation coefficient with effective reliability estimates of both FA and sexual performance. After the correction, the unbiased relationship between FA and mobility was r = -0.183. Our results indicate that FA is weakly related to male sexual performance in H. rubrofasciata. Our results also show that leptokurtosis, which is characteristic of signed FA distributions, was entirely caused by 17 outlier individuals. This indicates that large sample sizes are important to illustrate the true between-individual heterogeneity in FA. Furthermore, our power analysis indicates that it pays to measure more individuals than to increase the number-of within-subject repeats to obtain accurate population-level estimates of FA-trait associations.