Sexing shags phalacrocorax aristotelis from external measurements using discriminant analysis

Abstract

Discriminant Function Analysis was used to sex dead Shags collected from the Shetland Islands in January 1993. An equation based on bill depth and wing length measurements correctly assigned 94.8% of individuals and performed similarly on a second sample of known sex individuals, correctly sexing 92.9%. Consideration of the effect of feather abrasion on wing length indicated that the function would have considerably reduced discriminatory power on birds caught at different stages in the moult cycle. A second function based solely on bill depth, (which correctly assigned 92.8% of cases in the original sample and 90.3% of the second independent sample), would be more appropriate for studies involving birds with either very fresh, or very worn plumage. The preparation of specimens as museum skins resulted in a significant, but small, shrinkage in wing length. However, the shrinkage in wing length did not reduce the accuracy of classification of those specimens prepared as skins. The limitation of such discriminant functions, in relation to geographical range, season and age composition of the sample is discussed. The level of repeatability of measurements by a single observer was high, although the use of multiple observers to record biometrics is likely to have increased the sample variances greatly. The ability of a discriminant function, derived from measurements taken by many observers, to perform accurately on a second data set, itself obtained by many observers, indicates that the function is highly robust, and may be confidently applied by other workers, within the limitations discussed. © 1997 British Trust for Ornithology.