Costs of sound production have been investigated only sparsely in cetaceans, despite recent efforts to understand how increasing anthropogenic noise affects these animals that rely extensively on sound for communication and foraging. Theoretical estimates suggest that metabolic costs of whistling for bottlenose dolphins should be <0.54% of resting metabolic rate (RMR) (Jensen et al., 2012), whereas empirical studies of a single whistling dolphin surprisingly claimed that sound production costs were around 20% of RMR (Holt et al., 2015; Noren et al., 2013). Addressing this discrepancy, we found that costs of whistling were significantly less than 20% RMR and not statistically different from theoretical estimates (Pedersen et al., 2020). In their correspondence, Noren et al., 2020 argue that they did not claim whistling was ‘costly’ and questioned aspects of our methods, and we address these points here.
The ‘costly sound production’ hypothesis
Sound production efficiency is the ratio between emitted acoustic energy and the metabolic energy consumed to generate sound. Mammals, frogs, and birds produce sound with an efficiency of ∼0.4–3.0% (Pedersen et al., 2020). Acoustic efficiency has not been measured for marine mammals; consequently, we assumed an efficiency of 1% to estimate a theoretical metabolic cost of 1.7 J per whistle (Jensen et al., 2012). Noren and colleagues (2013) stated that ‘a theoretical approach to determine the metabolic costs of sound production in dolphins may be inaccurate’ and that the discrepancy is due to ‘the incorporation of incorrect variables (e.g. efficiency factor)’. Holt et al., (2015) estimated that the trial with most acoustic energy contained ‘approximately 0.08 J of [acoustic] energy’, and incurred a metabolic cost of 82,067 J, suggesting an ‘extremely low calculated efficiency factor range (less than 0.1%)’. While the authors report an estimated efficiency resembling that of other species, their numbers yield an extremely low efficiency of 0.0001%. This discrepancy is independent of their acoustic measurements – even if trained animals had vocalized at peak output levels of wild animals (∼1 J emitted acoustic energy per trial), efficiency would have been no greater than ∼0.001%. Thus, while the authors deny labelling dolphin communication ‘costly’, their studies directly indicate that dolphins are 3–4 orders of magnitude less efficient at producing sound than any other species studied.
Was there in fact any evidence of costly sound production?
Noren et al. (2013) found a significant cost of whistling but a non-significant cost of producing a burst-pulse squawk and concluded that ‘there is a measurable, though relatively small, metabolic cost to dolphins producing sound’. Subsequently, Holt et al. (2015) found a significant correlation between metabolic costs and acoustic output for the squawking but not the whistling dolphin, concluding that ‘vocal performance affects metabolic rate’. To reach that conclusion, 9/29 trials (squawking) and 3/27 trials (whistling) were discarded because the metabolic rate after sound production was lower than RMR. Such data omissions and subjective statistical interpretations weakens support of their conclusions that theory is wrong by 3–4 orders of magnitude.
Improved methods for quantifying small changes in metabolism
Noren et al. (2020) argue that our experimental design was unsuitable for detecting the low costs of whistling. Pedersen et al. (2020) measured breath-by-breath respirometry before and after a breath-holding period to track the rate of oxygen consumption. In contrast, Noren and colleagues (2013) used a traditional flow-through respirometry system where each breath was diluted in a respirometry dome with a system lag time of 36.5 s. If this lag time is the system time constant, it would take ∼1.5 min to evaluate changes in metabolic rate (Fahlman et al., 2008). This could explain why their estimated metabolic rates took on average 3.2–4.9 min to return to baseline, compared with an average 1.2 min recovery after apnea when using a breath-by-breath system (Fahlman et al., 2019). Our finer temporal resolution facilitates shorter measurement periods, which in turn may improve results as it is often difficult to prevent animal movement during longer inactive periods. This might explain why vocal or post-vocal metabolic rates were higher than RMR >50% of the time for one animal (Noren et al., 2013) and in 11% and 31% of trials subsequently (Holt et al., 2015). The authors also claimed that the breath-by-breath system cannot detect the ‘small’ differences in metabolic cost because of the high respiratory flow and short breath durations of dolphins. As expiratory durations were on average 50% longer than required to accurately detect the O2 uptake, this is not an issue (Fahlman et al., 2015). In fact, this method accurately measures low metabolic costs (control measurements within 0.5% of RMR) and so is likely more suitable for quantifying small changes in metabolism than dome respirometry.
Attributing metabolic costs to sound production
Past studies attempting to assess the metabolic cost of sound production may have included confounding effects unrelated to sound production, such as changes in movement, posture, or even cognitive demands or stress associated with trained tasks. To account for such effects, it is necessary to include silent ‘control’ trials, which are identical but without sound production (Fig. 1; see also Oberweger and Goller, 2001). Noren et al. (2017) demonstrate the importance of this – by comparing metabolic costs during echolocation trials to metabolic costs during silent control trials, they found no measurable costs of echolocation; however, without a control trial, echolocation would have appeared costly. Such silent control trials are missing from the Noren et al. (2013) and Holt et al. (2015) studies, and thus, in repeating a problem known in the bird literature for >10 years, these studies make it impossible to attribute changes in energetic demands unequivocally to sound production.
In conclusion, Pedersen et al. (2020) provide results that agree with theory and show that dolphins, like any other animal relying on sound for communication and foraging, have evolved efficient ways of producing sound. Thus, metabolic costs of increasing vocal outputs in elevated anthropogenic noise, such as through the Lombard response (Kragh et al., 2019), are predicted to be low.