Selective reactions to different killer whale call categories in two delphinid species

Abstract

The risk of predation is often invoked as an important factor influencing the evolution of social organization in cetaceans, but little direct information is available about how these aquatic mammals respond to predators or other perceived threats. We used controlled playback experiments to examine the behavioral responses of short-finned pilot whales (Globicephala macrorhynchus) off Cape Hatteras, NC, USA, and Risso's dolphins (Grampus griseus) off the coast of Southern California, USA, to the calls of a potential predator, mammal-eating killer whales. We transmitted calls of mammal-eating killer whales, conspecifics and baleen whales to 10 pilot whales and four Risso's dolphins equipped with multi-sensor archival acoustic recording tags (DTAGs). Only playbacks of killer whale calls resulted in significant changes in tagged animal heading. The strong responses observed in both species occurred only following exposure to a subset of killer whale calls, all of which contained multiple non-linear properties. This finding suggests that these structural features of killer whale calls convey information about predatory risk to pilot whales and Risso's dolphins. The observed responses differed between the two species; pilot whales approached the sound source while Risso's dolphins fled following playbacks. These divergent responses likely reflect differences in anti-predator response mediated by the social structure of the two species.

Keywords: Acoustic discrimination; Antipredator behavior; Non-linear acoustics; Pilot whales; Risso's dolphins.

Fig. 1.

Playback analysis. Examples of non-response (left, Gm12_162a) and response (right, Gm12_125a) of short-finned pilot whales to playbacks of mammal-eating killer whale calls grouped into sound types with responses (right) and those without (left). The reactions included increased flow noise, overall dynamic body acceleration (ODBA) and heading variance. Each panel displays a 30 s period before, during and after presentation of a killer whale call and provides values for: (A) spectrogram of the acoustic record – the first received signal from the playback is highlighted with a black box; (B) flow noise, which is a proxy for the speed of the tagged animal; (C) depth profile of the tagged animal; (D) ODBA, which is a proxy for movement of the tagged animal; and (E) circular plots of heading of the tagged animal.

Fig. 2.

Spectrograms of calls from mammal-eating killer whales used in playback experiments with short-finned pilot whales and Risso's dolphins. Playback exemplars are identified using the two-letter system denoting species (Oo, Orcinus orca) as in Table 1. *A strong response was observed during playback experiments.

Fig. 3.

Circular plots of heading relative to the sound source for 30 s before (top) and after (bottom) playbacks of killer whale exemplar Oo-10 to a pilot whale (left) and a Risso's dolphin (right). Bearing of the sound source (green) to the focal animal was determined using known location of the sound source and the estimated location of the focal animal as reconstructed from the focal follow points. Heading of the animal was obtained from the DTAG.

Fig. 4.

Parameter estimates of Poisson general estimating equation (GEE) of pilot whale (top) and Risso's dolphin (bottom) call counts before and after playbacks of each stimulus type. Horizontal lines represent parameter estimates and vertical lines represent the 95% confidence interval (CI) derived from parametric bootstrap of the fit parameters of the GEE. Call counts following playback of the group of mammal-eating killer whale calls that elicited a response (highlighted in red for pilot whales) were significantly higher than baseline or before levels. There were no significant changes in call counts following playback of any of the other stimulus types.

Fig. 5.

Parameter estimates and 95% CI of Gaussian GEE for ODBA before, during and after playbacks of various exemplars to short-finned pilot whales (top) and Risso's dolphins (bottom). Horizontal lines represent parameter estimates and vertical lines represent the 95% CI derived from parametric bootstrap of the fit parameters of the GEE. OBDA levels during playback of the identified group of mammal-eating killer whale calls that elicited a response (highlighted in red) were significantly higher than baseline levels. No other period (before, during or after) for any other stimulus was found to be significantly different from baseline levels.