Abstract: Male humpback whales (Megaptera novaeangliae) produce long songs that contain predictably repeated sound patterns. Other animals (including humans) identify patterns in acoustic sequences based on regularities in transitions between sounds. The present study examined transitional probabilities within humpback whale songs to determine whether relative acoustic changes from unit to unit are sufficient for identifying repeating patterns within humpback whale songs. To identify such patterns, four humpback whale songs were analyzed by first classifying song units using a self-organizing map, and then calculating transitional probabilities based on these classifications. Two separate analyses of transitional probabilities were conducted: one involved units classified based on their absolute acoustic features (e.g., duration, peak frequency, and amplitude) as well as changes in these features relative to adjacent units, and the other used units classified based on the relative changes alone. Both analyses revealed repeated sequences of units within humpback whale songs, but the analysis based on relative changes alone yielded a larger number of predictable transitions. This finding suggests that relative acoustic changes within humpback whale songs may provide robust indicators of repeating patterns.


Document Type: Research article

DOI: 10.1578/AM.33.2.2007.202

Page Numbers: 202 - 213

$12.00 each Vol. 33, Iss. 2, Green et al