Sexing of Phylloscopus based on multivariate probability of morphological traits

ABSTRACT Sexes cannot be distinguished with certainty by human observers in many avian species. However, some apparently monomorphic species have small but measurable sexual dimorphisms in biometry, which can be used to determine sex. Here, we develop a method based on multivariate probabilities to improve sexing in three Phylloscopus species. We captured 16 124 Common Chiffchaffs (P. collybita), Iberian Chiffchaffs (P. ibericus) and Willow Warblers (P. trochilus) during postnuptial migration in the Doñana National Park (SW Spain). We estimated the probability of them being male or female based on a combination of the density distributions of wing and tarsus lengths. Density distributions were derived as two normal components of the mixture distribution in both traits. Discriminatory ability was compared under different assumptions. 95% of Common Chiffchaffs, 92% of Iberian Chiffchaffs and 87% of Willow Warblers were sexed with 95% confidence. These percentages are greater than those reported for these species using sexing methods based only on wing length. Our method was not affected by changes between years in the degree of dimorphism or mixture of populations from different geographical origins. In the Iberian Chiffchaff and the Willow Warbler, sexing was improved when our method was applied to immatures and adults separately. Sex ratio was estimated to be two females per male in the Common Chiffchaff and the Willow Warbler. In the latter species, females migrated one week later than males. Sex ratio in the Iberian Chiffchaff was 1:1. Our sexing method could be used for other species with known and apparent dimorphism or to any data set of birds with biometric measures.