Individual Distinctiveness and Possible Function of Song Parts of Short-toed Treecreepers (Certhia brachydactyla). Evidence from Multivariate Song Analysis

Abstract

Sonagram analyses of short-toed treecreeper (= STT) song by use of discriminant analysis revealed consistent individuality in the songs predominantly sung by each male (P-songs). The P-songs had been digitized and over 30 song variables were used in the analysis to separate different individuals. In the end, 7 of these variables were needed for a correct classification of 100% of the 203 P-songs recorded of 39 individuals in 1986. Most of these variables were derived from the first part of the song. A similar analysis with data from 1987 showed that by use of all variables with unskewed distribution complete discrimination of all 271 P-songs of 47 individuals ensued. The reduction of variables to the 10 most important led to the correct classifications of 99.5%. Using the 7 best discriminators of the 1986 data set 94.2% of the 1987 data were correctly classified, thus confirming the 1986 results. Wrong assignments were mainly due to P-songs from second or third recording dates which in some cases could not be separated from other P-songs of the same males by the analysis. Further analysis of the data showed that although the most important discriminators were found in the first part of the song, the combination of all characters of the end part would also allow for a virtually complete discrimination of individuals. The song features of STT P-song are so distinctive that territory holders should have no difficulties in separating individuals (e.g. neighbours from strangers). Corresponding playback experiments have yet to be carried out on STT.