A generalized discriminant for sexing fulmarine petrels from external measurements

Discriminant analysis can use morphometric differences between known male and female birds to predict the sex of unknown individuals in field studies. Geographic variation in size and shape often limits the predictive value of a discriminant function to the population from which it was derived. Specific discriminant functions for populations of five species of fulmarine petrels (Northern Fulmar, Fulmarus glacialis; Southern Fulmar, F. glacialoides; Antarctic Petrel, Thalassoica antarctica; Cape Petrel, Daption capense; and Snow Petrel, Pagodroma nivea) assigned 81 to 98% of birds in the samples to the correct sex, but the validity of each discriminant applied to alternative populations remained questionable. Our approach to overcome this limitation is to combine data from the different species into a single discriminant. Adequate performance of this generalized discriminant in samples of different species shows its validity for use in other populations of any of these species. The generalized function calculates the discriminant scores for individual fulmarine petrels as: Y = HL + 2.38BD + 0.41TL – 0.21CL, where HL is head length, BD is bill depth, TL is tarsus length and CL is bill length (measurements in millimeters). The cut point to split sexes is different in each sample and may be calculated directly from discriminant scores, without reference to sexed birds, by using a maximum-likelihood method. Depending on species, the generalized method results in 84 to 97% correct classifications and can be applied to other populations of fulmarine petrels without requiring samples of birds of known sex.