Talk Whale to Me: Classifying the Low Frequency Social Calls of Feeding Humpbacks
Faculty Mentor Name
Stacie Hooper
Research or Creativity Area
Natural Sciences
Abstract
Humpback whales (Megaptera novaeangliae) are migratory, traveling between feeding areas in the summer and breeding grounds in the winter (Cerchio & Dahlheim, 2001). On the feeding ground in the Gulf of Alaska, some use a cooperative feeding strategy called bubble-net feeding to capture fish. To perform this behavior a coordinated groups of whales produce a series of short, repetitive “food calls” while simultaneously creating a net of bubbles. Food calls cause the fish to move into tighter shoals, and are also likely used to organize the group’s behavior during feeding events (Sharpe, 2001). In addition to food calls, humpback whales produce a large variety of social sounds while performing this behavior. The complexity of the vocal repertoire of humpback whales is similar to that of human language. In order to understand the role that these social sounds play in coordinating and performing bubble-net feeding, we first need to be able to categorize them into distinct, consistent types. Prior studies have attempted to organize humpback whale social calls into broad categories using both qualitative auditory classification and quantitative measures of acoustic variables like mean frequency and duration (Dunlop et al., 2007). We were able to identify some similar call types to those reported in Dunlop et al. (2007) in recordings of humpback whales performing bubble-net feeding in the Gulf of Alaska. The objective of this study was to utilize auditory classification methods to build a set of standardized descriptions for five low frequency social call types: burps, growls, thwups, roars, and creaks. There was considerable human-generated noise during our sample recordings, which made quantitative methods difficult to perform, especially for low frequency calls. Using Audacity analysis software, we isolated good quality exemplars of each call type (determined by lab group consensus), and asked trained individuals to classify unknown calls to a particular type based on the acoustic features of each call. Performing call categorization is a challenge as considerable inter-individual variation exists in social call production, and social calls are less stereotyped than food calls. Our classification aims to improve consistency in call identification and will contribute to a more objective understanding of humpback whale vocalizations.
Talk Whale to Me: Classifying the Low Frequency Social Calls of Feeding Humpbacks
Humpback whales (Megaptera novaeangliae) are migratory, traveling between feeding areas in the summer and breeding grounds in the winter (Cerchio & Dahlheim, 2001). On the feeding ground in the Gulf of Alaska, some use a cooperative feeding strategy called bubble-net feeding to capture fish. To perform this behavior a coordinated groups of whales produce a series of short, repetitive “food calls” while simultaneously creating a net of bubbles. Food calls cause the fish to move into tighter shoals, and are also likely used to organize the group’s behavior during feeding events (Sharpe, 2001). In addition to food calls, humpback whales produce a large variety of social sounds while performing this behavior. The complexity of the vocal repertoire of humpback whales is similar to that of human language. In order to understand the role that these social sounds play in coordinating and performing bubble-net feeding, we first need to be able to categorize them into distinct, consistent types. Prior studies have attempted to organize humpback whale social calls into broad categories using both qualitative auditory classification and quantitative measures of acoustic variables like mean frequency and duration (Dunlop et al., 2007). We were able to identify some similar call types to those reported in Dunlop et al. (2007) in recordings of humpback whales performing bubble-net feeding in the Gulf of Alaska. The objective of this study was to utilize auditory classification methods to build a set of standardized descriptions for five low frequency social call types: burps, growls, thwups, roars, and creaks. There was considerable human-generated noise during our sample recordings, which made quantitative methods difficult to perform, especially for low frequency calls. Using Audacity analysis software, we isolated good quality exemplars of each call type (determined by lab group consensus), and asked trained individuals to classify unknown calls to a particular type based on the acoustic features of each call. Performing call categorization is a challenge as considerable inter-individual variation exists in social call production, and social calls are less stereotyped than food calls. Our classification aims to improve consistency in call identification and will contribute to a more objective understanding of humpback whale vocalizations.
