Male corncrakes Crex crex produce repetitive calls the elements of which are composed of individual pulses. The inter-pulse intervals were measured for elements of calls from ten birds recorded on one to three nights. The abilities of discriminant analysis and two artificial neural nets to correctly attribute calls to birds across nights, using sets of vectors of inter-pulse intervals, were compared. Discriminant analysis correctly attributed all calls to the correct individual without error. One of the two neural nets incorrectly attributed less than 1% of calls. Using coded data the second neural net incorrectly attributed less than 6% of calls overall but the individual error rate varied between zero and more than 50%.
corncrake, multivariate, discriminant analysis, neural network, classifier, threshold