| Summary: | High-resolution sound and movement recording tags (e.g. Dtags, Acousonde tags, Atags) offer unprecedented views of the fine-scale foraging behaviour of cetaceans, especially those that use sound to forage, such as the sperm whale (Physeter macrocephalus). However, access to these tags is difficult and expensive, limiting studies of sperm whale foraging behaviour to small sample sizes and short time periods, preventing inferences at the population level. The development of accurate foraging indices from relatively inexpensive time-depth recorder (TDR) data would allow obtaining data from a larger number of individuals, and capitalizing on datasets already available, providing long-term analyses of foraging activity. In this study, data from high-resolution acoustic and movement recording tags from 8 sperm whales was used to build predictive models of the number of buzzes (i.e, indicative of prey capture attempts (PCA)) for dive segments of different lengths, using dive metrics calculated from timedepth data only. The number of buzzes per dive segments of 180s and 300s was best predicted by the average depth, depth variance, vertical velocity variance and number of wiggles. Model performance was best for 180s segments, accurately predicting the number of buzzes in 63% of the segments used to construct the model and in 58% of the segments for new individuals. Predictive accuracy reached 81%, when only presence or absence of buzzes in segments was assessed. These results demonstrate the feasibility of finding a reliable index of sperm whale foraging activity for time-depth data, when combining different dive metrics. This index estimates the number of buzzes over short dive segments (of 180s), enabling investigating and quantifying PCAs at very finescales. Finally, this work contributes to leverage the potential of time-depth data for studying the foraging ecology of sperm whales and the capacity of applying this approach to a wide range of cetacean species. O cachalote (Physeter macrocephalus) é um dos mais ...
|
|---|