A comparison of two methods for estimating measurement repeatability in morphometric studies

1. Measurement repeatability is often reported in morphometric studies as an index of the contribution of measurement error to trait measurements. However, the common method of remeasuring a mounted specimen fails to capture some components of measurement error and could therefore yield inflated repeatability estimates. Remounting specimens between successive measurements is likely to provide more realistic estimates of repeatability, particularly for structures that are difficult to measure. 2. Using measurements of 22 somatic and genitalic traits of the neriid fly Telostylinus angusticollis, we compared repeatability estimates obtained via remeasurement of a specimen that is mounted once (single‐mounted method) versus remeasurement of a specimen that is remounted between measurements (remounted method). We also asked whether the difference in repeatability estimates obtained via the two methods depends on trait size, trait type (somatic vs. genitalic), sclerotization, or sex. 3. Repeatability estimates obtained via the remounted method were lower than estimates obtained via the single‐mounted method for each of the 22 traits, and the difference between estimates obtained via the two methods was generally greater for small structures (such as genitalic traits) than for large structures (such as legs and wings). However, the difference between estimates obtained via the two methods did not depend on trait type (genitalic or somatic), tissue type (soft or sclerotized) or sex. 4. Remounting specimens between successive measurements can provide more accurate estimates of measurement repeatability than remeasuring from a single mount, especially for small structures that are difficult to measure.