We read ‘Whistling is metabolically cheap for communicating bottlenose dolphins (Tursiops truncatus)’ by Pedersen et al. (2020) and were concerned to see how our results (Noren et al., 2013; Holt et al., 2015) were presented. This was especially surprising, given our previous feedback to the authors (see acknowledgements, Pedersen et al., 2020). Although Pedersen et al. claim their work disproves our findings, their methods were not designed to accurately measure the low metabolic cost (MC) of whistle production. In fact, the differing conclusions are explained by differences in methods and interpretation of the findings.
Pedersen et al. (2020) ‘reject the hypothesis that whistling is costly for bottlenose dolphins.’ This ‘hypothesis’ is attributed to our publications. To be clear, we reported that the average metabolic rate (MR) for whistling dolphins is 1.2× resting metabolic rate (RMR) and concluded that ‘there is a measurable, though relatively small, metabolic cost to dolphins producing sounds, including whistles’ (Noren et al., 2013). Comparatively, MRs of whistling dolphins fall between those of sitting humans watching television (1.0× RMR) and typing (1.5× RMR, Ainsworth et al., 1993) and are comparable to MRs of stationary, sound producing birds (see Noren et al., 2013; Pedersen et al., 2020). We verified through video analysis that movement by whistling dolphins was minimal (Holt et al., 2015).
The experimental design in Pedersen et al. (2020) makes it challenging to detect the low MC of whistles. Noren et al. (2013) and Holt et al. (2015) measured dolphins whistling at the water surface via flow-through respirometry with a metabolic hood. Respiration rates (RRs) of whistling dolphins did not differ from RRs pre- and post- sound production (Noren et al., 2013). Importantly, we used statistical analyses to determine when MR returned to RMR during 10 min recovery periods (Noren et al., 2013). In contrast, Pedersen et al. used breath-by-breath respirometry with a pneumotachometer to measure dolphins following apneustic periods. The accuracy of breath-by-breath respirometry is impacted by dolphins' very high respiratory flows and short breath durations (Fahlman et al., 2015), which would make measuring the low MC of whistling difficult. Pedersen et al. did not evaluate changes in MR during recovery periods. This is critical, as dolphin MRs typically recover within <5 min following sound production (ranges, means±s.d.: 0–5.8 min, 3.2±1.9 min, unpublished data from Holt et al., 2015; 2.8–6.7 min, 4.9±1.2 min, Noren et al., 2013).
Although Pedersen et al. were unable to directly measure whistling MRs in submerged dolphins, they conducted a large number of trials, randomized over the experimental period. Using this experimental design, the most direct method to estimate the MC of whistling, while accounting for MR variability, is to compare post-apnea MC across trial types. Although Pedersen et al. reported no difference, the average MC of apnea+whistling was 1.15× the average MC of apnea during control trials and was nearly significantly different (P=0.06), indicating whistling MC similar to those reported in our studies.
Our studies compared daily matched RMR and sound production MR to account for daily MR fluctuations that could mask the MC of sound production (Noren et al., 2013; Holt et al., 2015). Pedersen et al. estimated whistling MR using assumptions that critically affected their results. First, oxygen consumption measured during 2 min pre-apnea is assumed to represent RMR. This is problematic because this short time period may not accurately estimate RMR (see Compher et al., 2006). Second, data from the entire 5 min recovery period are included. This is problematic because the small elevation in MR from whistling is present during a fraction of this period. Consequently, the average MR masks the MC of whistling. Thus, it is not surprising that the estimated whistling MR (1.04× RMR, Pedersen et al., 2020) is lower than our measured value (1.2× RMR, Noren et al., 2013; Holt et al., 2015). Interestingly, this 4% increase in MR is still greater than the maximum theoretical predicted increase (0.5–1%, Jensen et al., 2012). Consistent with our studies (Noren et al., 2013; Holt et al., 2015), Pedersen et al. showed highly variable metabolic responses in whistling dolphins. Some individuals had noticeably higher whistling MCs (Table 3, Fig. 4, Pedersen et al., 2020). Given these results, it would be informative to know how oxygen consumption changed over the recovery period and how these changes varied by individual, trial, and sound energy levels. This would be relatively easy to provide.
Comparisons between the metabolic cost of whistling and squawking, as presented in Pedersen et al., are erroneous. Squawking dolphins can have higher MRs (1.5× RMR, Holt et al., 2015) than whistling dolphins. This may be related to potentially differing metabolic demands associated with using different muscles to produce the distinct sounds (Ridgway et al., 1980). Pedersen et al. used data for squawking dolphins from our studies to estimate dolphin sound production efficiency (Fig. 6). Holt et al. (2015) clearly state that calculating sound production efficiency from our data is inappropriate, given a multitude of reasons, which were ignored. Consequently, Fig. 6 from Pedersen et al. is misleading and some readers may mistakenly assume that the value presented is for whistling dolphins.
Finally, three studies report that the MC of clicking, whistling, and squawking increases with sound duration and/or energy levels in dolphins (Noren et al., 2013, 2017; Holt et al., 2015). Yet, Pedersen et al. use results of their study, which aimed to measure sound production efficiency, rather than MCs of vocal modification, to conclude that there is no cost associated with the Lombard effect.
In summary, contrary to conclusions in Pedersen et al., their results align with ours. They were unable to detect the small MC of whistling in dolphins because of the use of less-sensitive respirometry methods and data modeling. Their conclusion that 1.2× RMR is a high MC is invalid, given that MRs of most sound producing dolphins fall within the large range of dolphin RMRs (Fahlman et al., 2015, 2019) and the greater MC of other activities in free-ranging cetaceans (Noren et al., 2013).