Cardiorespiratory coupling in cetaceans; a physiological strategy to improve gas exchange?

Variation in heart rate (f_H) associated with breathing, often called respiratory sinus arrythmia (RSA), is seen in most vertebrates (Grossman and Taylor, 2007; Laske et al., 2010; Piccione et al., 2019; Zena et al., 2017). RSA results in the acceleration of f_H during inspiration, and deceleration during expiration (Mortola et al., 2015). The physiological significance of RSA is debated (Hayano et al., 1996; Yasuma and Hayano, 2004), but it has been suggested that RSA reduces cardiac work and improves gas exchange by enhancing ventilation–perfusion matching (Ben-Tal et al., 2012, 2014; Yasuma and Hayano, 2004). In terrestrial mammals that breathe almost continuously, RSA is minimal (Piccione et al., 2019), but has been shown to vary with both breathing frequency (f_R) and tidal volume (V_T) (de Burgh Daly, 1986; Guillén-Mandujano and Carrasco-Sosa, 2014; Hirsch and Bishop, 1981). This variation in f_H with f_R and V_T may not be critically important for comparative studies on terrestrial mammals where breathing is more or less continuous and its effect on cardiac function may not cause large variation in average f_H. However, in animals that breathe apneustically, like marine mammals, variation in f_R may cause large variation in f_H, thereby confounding temporal averages of f_H.

In marine mammals, and specifically cetaceans, where breaths are separated by long respiratory pauses (Fahlman et al., 2017), significant changes in f_H have been observed following a breath (Andrews et al., 1997; Bickett et al., 2019; Castellini et al., 1994a,b; Cauture et al., 2019; Elmegaard et al., 2016; Elsner et al., 1966; Fahlman et al., 2019b; Harms et al., 2013; Houser et al., 2010; Kaczmarek et al., 2018; McDonald et al., 2018; Miedler et al., 2015; Noren et al., 2004, 2012; Reed et al., 2000; Williams et al., 1993). Variation in f_R is expected to have a large effect on average f_H, but despite this, most studies on cardiac function in marine mammals do not report f_R or the magnitude of RSA. For this reason, it has been suggested that studies that investigate cardiovascular changes with diving should estimate and report the RSA-corrected f_H for an inactive animal (Cauture et al., 2019; Fahlman et al., 2019b). However, RSA-correcting data does not provide a true estimate of the average f_H, stroke volume (SV) or cardiac output (CO) as it omits the dynamic period from the end of the breath until the instantaneous f_H (if_H) and SV (iSV) have stabilized after about 10–12 s (Cauture et al., 2019; Fahlman et al., 2017). Still, the RSA-corrected f_H and SV provide a means to compare values between studies and species that are independent of f_R. Thus, a better understanding of the cardiorespiratory coupling between f_H and f_R is warranted and may provide improved understanding how marine mammals balance the trade off between gas exchange during apneustic periods with the periods of tachycardia while breathing.

Given the limited work that has been done on cardiorespiratory coupling, the confounding effect of f_R on f_H, the potential effect of RSA on gas exchange, and the possible common mechanism to the diving bradycardia, this study was aimed at providing comparative estimates of cardiorespiratory function in cetaceans. We measured semi-continuous f_H and SV, and calculated CO using transthoracic echocardiography immediately following inhalation in the bottlenose dolphin [Tursiops truncatus (Montagu 1821)]. We hypothesized that changes in CO during the respiratory cycle would mainly be governed by changes in f_H as has been reported in other marine mammals during short apneas (Blix et al., 1983; Blix et al., 1976; Elsner et al., 1964; Murdaugh et al., 1966). We also collected transthoracic echocardiography data from an adult false killer whale [Pseudorca crassidens (Owen 1846)] and a juvenile beluga whale [Delphinapterus leucas (Pallas 1776)] while breathing at the surface at rest. These data were used to test the hypothesis that cardiac parameters in cetaceans at rest are allometrically similar to those of terrestrial mammals. In addition, we also included previously published continuous if_H from the bottlenose dolphin, with new data collected from the false killer whale, beluga whale, killer whale [Orcinus orca (Linnaeus 1758)] and pilot whale (Globicephala macrorhynchus Gray 1846) to better define how f_H is modulated by breathing. These data were used to test the hypothesis that the relationship between f_R and f_H for inactive animals is similar in cetaceans as compared with terrestrial mammals.

Ultrasound continuous flow Doppler ultrasound was used to measure f_H and SV at the level of the left ventricular outflow tract from one adult female false killer whale housed at Sea Life Park (Hawaii-USA, January 2017), and one juvenile male beluga whale calf at the Oceanogràfic (Valencia-Spain, March/April 2017) (Table 1). In addition, we included previously published cardiovascular and/or respiratory data from 10 adult male and 3 adult female Atlantic bottlenose dolphins (Fahlman et al., 2019a,b). The animal ID, sex, body mass (M_b), year of birth (known or estimated) and aortic valve orifice diameter are summarized in Table 1. For the false killer whale, the M_b was estimated from length and girth, while for all other animals, M_b was measured.

We used previously published electrocardiogram (ECG) data for up to 10 min in the bottlenose dolphin, and we added measured ECG from the juvenile beluga whale and the adult false killer whale. In addition, we collected ECG data from adult pilot whales (n=3) and killer whales (n=3), located at SeaWorld Orlando, while they were breathing at the surface (Table 1).

The study protocols were accepted at each facility, as well as by the Animal Care and Welfare Committee at the Oceanogràfic (OCE-17-16, amendments OCE-29-18 and OCE-3-19i) and the Bureau of Medicine (BUMED, NRD-1015).

All experiments were performed using operant conditioning as previously detailed (Fahlman et al., 2019a,b). Participation by each individual was voluntary, and the animals were not restrained and could refuse to participate or withdraw at any point during the experimental trial without any negative consequences. None of the animals participating in this study were fasted. Each experiment (trial) consisted of an animal floating stationary in the water in left lateral recumbency with the blow-hole out of the water, allowing ultrasound probe placement over the left thoracic wall to find the left ventricle. In a few dolphins (n=8), the false killer whale (n=1), the beluga whale (n=1), the pilot whales (n=3) and the killer whales (n=3), we also collected continuous ECG by placing three custom electrodes on the ventral surface (see below).

The ultrasound machine (Vivid-I or Vivid-IQ, General Electric), with a 1–3 MHz phased array probe, was used to obtain left ventricular blood flow and velocities in continuous movie recordings up to 16 s in duration (Fahlman et al., 2019b; Miedler et al., 2015). f_H was estimated from the flow traces, and SV was calculated from the velocity time integral of the systolic blood flow at the level of the left ventricular outflow tract multiplied by the cross-sectional area of the aortic valve orifice (the radius of the aortic valve orifice squared multiplied by π). The velocity time integral was evaluated using continuous flow Doppler in flow direction from the left ventricle into the aorta, and an angle correction was used to measure the flow parallel to its path. The aortic diameter at the level of valve insertion at the different cardiac phases was measured to confirm that it was constant for all flow rates. The CO was estimated as: CO=f_H×SV.

The ECG was recorded using three gold-plated electrodes mounted inside a silicone suction cup connected to a custom-built data recorder (UUB/1-ECGb, UFI, Morro Bay, CA, USA). The suctions cups were filled with conducting gel (Ten20 Conductive Neurodiagnostic Electrode paste, Waver and Company, Aurora, CO, USA) before being placed on the ventral side of the animal in a triangle or in a line down the sternum (Bickett et al., 2019; Cauture et al., 2019). The cetacean was instructed to turn ventral side up and the suction cups were placed on the skin; once a clear ECG signal was detected, the animal rolled over to the dorsal side up. Occasionally, the ECG signal was lost as the animal rolled and the procedure was repeated until a good signal was recorded when the cetacean was in a still position with the dorsal side up (Bickett et al., 2019; Cauture et al., 2019). The ECG was recorded at 400 Hz using a data acquisition system (Powerlab 8/35, ADInstruments, Colorado Springs, CO, USA), and displayed on a computer running LabChart (v.8.1.13, ADInstruments). Initially, the electrodes were adjusted to ensure a clear ECG trace. Next, the animal was allowed to breathe spontaneously for up to 10 min and the timing of each breath recorded.

The ECG was analyzed using the heart rate detection software in LabChart, which automatically detected the R-peaks, using the following settings; typical QRS width 80 ms, R-waves 300 ms, pre-P baseline 120 ms, maximum PR 240 ms, maximum RT 400 ms. Following the automatic detection, the R-peaks were manually verified and the if_H determined from the time between R–R peaks. Noisy sections, or possible beats that did not contain a clear R-peak were removed.

We compared the data for inactive animals within and between individuals. The relationship between a dependent variable (f_H, SV and CO) and the experimental covariate, M_b, was analyzed using linear-mixed effects models (lme, R version 3.3.3, http://www.R-project.org/). We log₁₀-transformed the variables to generate linear functions that could be used with the lme function in R. The individual animal was treated as a random effect, which accounted for the correlation between repeated measurements on the same individual (Littell et al., 1998). Normality for all models was confirmed by the qqnorm plot. Best models of remaining variables were chosen by the log-likelihood (LL) ratio test. Acceptance of significance was set to the P<0.05 level, while 0.05<P<0.1 was considered a trend. Data are presented as the mean±s.d., unless otherwise stated.

We measured f_H and SV, and calculated CO in the bottlenose dolphin, false killer whale and juvenile beluga whale using transthoracic echocardiography. The changes in if_H and iSV associated with respiration are shown in Fig. 1 for the bottlenose dolphin. Similar dynamic changes in if_H and iSV were seen in both the false killer whale and juvenile beluga whale. A representative figure of continuous if_H shows that five species experienced a pronounced RSA, where the maximum if_H occurred approximately 3–5 s following the breath, and then slowly decayed towards a stable if_H (Fig. 2). This stable value we define as the RSA-corrected f_H for an inactive animal.

To better define the changes in cardiovascular function associated with breathing, we determined if_H and iSV in four bottlenose dolphins where it was possible to visualize the heart during all phases of the respiratory cycle. The measurements started at the end of the inhalation and continued for the next 16 s, the maximal duration possible to record continuous data using this ultrasound machine. The data are shown in Fig. 3, where both if_H (Fig. 3A) and iSV (Fig. 3B) increased, and then exponentially decayed towards the values for an inactive dolphin approximately 12–20 s following the breath (Fig. 2). The data were log₁₀-transformed, and the exponential decay function for if_H (beats min⁻¹) was: log₁₀(if_H)=1.912−0.172×log₁₀(time) (χ²=148, P<0.01; Fig. 3A), where time is in seconds. The same relationship for iSV (ml) was: log₁₀(iSV)=2.402−0.090×log₁₀(time) (χ²=131, P<0.01; Fig. 3B). There was a significant correlation within each individual between the if_H and iSV following the breath (iSV=112.6+1.719if_H, χ²=78, P<0.01; Fig. 3C).

For an analysis between animals and species, we used the RSA-corrected f_H, SV and CO for an inactive animal. This was done by extracting the average f_H, SV and CO at least 12–20 s following a spontaneous breath, and after the f_H and SV had been observed to stabilize for 5–7 s following the breath (Figs 1–3) (Fahlman et al., 2019b; Miedler et al., 2015). For each animal and measurement, a minimum of at least 3 repeated measurements of f_H and SV were made and averaged. Fig. 4 shows the average RSA-corrected f_H, SV and CO for the bottlenose dolphins, adult false killer whale and beluga whale calf. Also shown in this figure are the predicted f_H, SV and CO for a 170 kg and 520 kg terrestrial mammal (Bishop, 1997; Seymour and Blaylock, 2000; Stahl, 1967). For these data, the f_H was significantly higher, and SV and CO significantly lower in the dolphin as compared with both the similar-sized beluga whale and the heavier adult false killer whale. The average mass-specific f_H (sf_H), was significantly higher for the dolphins (one sample t-test, P<0.01, t=15.8, d.f.=10) as compared with the false killer whale (sf_H: dolphin 0.265±0.044 beats min⁻¹ kg⁻¹, false killer whale 0.056±0.010 beats min⁻¹ kg⁻¹, beluga whale 0.236±0.038 beats min⁻¹ kg⁻¹), while both mass-specific SV (sSV: dolphin 1.07±0.24 ml kg⁻¹, false killer whale 1.22±0.24 ml kg⁻¹, beluga whale 3.06±0.21 ml kg⁻¹) and CO (sCO: dolphin 47.1±10.5 ml min⁻¹ kg⁻¹, false killer whale 38.2±14.8 ml min⁻¹ kg⁻¹, beluga whale 116.6±17.3 ml min⁻¹ kg⁻¹) were significantly higher for the beluga whale calf as compared with the dolphins (sSV, P<0.01, t=27.5, d.f.=10; sCO, P<0.01, t=21.8, d.f.=10).

The diameter of the aortic valve orifice of the false killer whale (t=73.5, P<0.01) and that of the beluga whale calf (t=40.8, Table 1) were both significantly larger as compared with the average aortic diameter of the bottlenose dolphins (mean±s.d. 3.3±0.2 cm, P<0.01), but there was no difference in aortic valve orifice diameter with M_b within the dolphins (F_1,9=0.2, P>0.6; Table 1).

From Fig. 3, the average RSA-corrected f_H and SV (from around 12–16 s) of an inactive animal would be approximately 48 beats min⁻¹ and 192 ml, respectively, resulting in a CO of about 9.3 l min⁻¹. Using the regression equations for Fig. 3A,B (see ‘Cardiorespiratory dynamics’, above) for the uncorrected f_H and SV for an inactive dolphin, the average f_H and SV for an f_R of 2 breaths min⁻¹ would be approximately 54 beats min⁻¹ and 203 ml, respectively (this is the average f_H and SV from 0 to 30 s in Fig. 3). The same calculation for an f_R of 6 breaths min⁻¹ resulted in an average f_H and SV of 65 beats min⁻¹ and 223 ml, respectively, with a CO of 14.5 l min⁻¹. In comparison, the average f_H for an inactive animal over a 10 min period at the surface in the bottlenose dolphin, beluga whale and false killer whale was 65±9 beats min⁻¹ (dolphin f_R: 4.1±1.2 breaths min⁻¹, range: 2.8–6.0 breaths min⁻¹), 44 beats min⁻¹ (beluga whale f_R: 1.9±1.5 breaths min⁻¹) and 65 beats min⁻¹ (false killer whale f_R: 5.4±3.3 breaths min⁻¹), respectively.

To provide a comparison with an inactive terrestrial mammal with an M_b of 170 kg, the predicted f_R would be 14 breaths min⁻¹ (Stahl, 1967). Using this f_R and the regression results in Fig. 3, the average f_H and SV would be approximately 74 beats min⁻¹ and 240 ml, respectively.

The average f_R and f_H for the continuous f_H (ECG) from the inactive bottlenose dolphin, the juvenile beluga whale, the false killer whale, the pilot whale and the killer whale are plotted in Fig. 5. In the plot, we also included published data for the harbor porpoise (Phocoena phocoena), the adult beluga whale (Delphinapterus leucas) and gray whale (Eschrichtius robustus) (Bickett et al., 2019; Elmegaard et al., 2019; Ponganis and Kooyman, 1999; Wahrenbrock et al., 1974). The regression for the cetaceans was: f_H=24.8+11.9f_R (Fig. 5). As a comparison we plotted the regression for the allometric estimate from terrestrial mammals for f_R and f_H (Stahl, 1967). While the allometric equations developed by Stahl (1967) were from rat to human (200 g to about 80 kg), we added points for the donkey (Equus asinus, M_b 165–190 kg) (Yousef and Dill, 1969), the Thoroughbred horse (Equus ferus caballus, M_b 448 kg) (Kuwahara et al., 1996), the African elephant (Loxodonta africana), the Asian elephant (Elephas maximus) (M_b 1420–6000 kg) (Benedict and Lee, 1936; Honeyman et al., 1992; Olivares et al., 2019) and the hibernating American black bear (Ursus americanus) (Laske et al., 2010) (Fig. 5).

This is the first study to report comparative cardiac (f_H, SV and CO) measurements in inactive small and medium-sized cetaceans. We show that both f_H and SV vary during the respiratory cycle in the bottlenose dolphin, false killer whale and beluga whale, and that f_H varies in the killer whale and pilot whale. The data indicate that the RSA-corrected f_H for an inactive cetacean is lower in comparison to studies reporting a temporal average f_H, reflecting the importance of separating the effect of respiration on f_H for comparative purposes. Our data support that changes in CO are caused by dual changes in f_H and SV, but the former changes relatively more for a given change in CO. Of special interest is the higher mass-specific SV and CO in the beluga whale calf as compared with the similar-sized dolphins and the larger false killer whale, possibly reflecting increased metabolic demand due to active growth. A sensitivity analysis was done using the temporal responses in if_H and iSV to evaluate the variation in resting f_H with varying f_R. In addition, we show that in cetaceans the RSA appears to follow a similar time course and that f_H responds more strongly to variation in f_R in cetaceans as compared with terrestrial mammals. We propose that RSA may be physiologically important to enhance gas exchange matching during the surface interval to minimize recovery duration (Andrews et al., 2000; Fahlman et al., 2018; Fedak et al., 1988). The data presented in the current study provide valuable comparisons within and between species that are important to understand physiological limitations in cetaceans. Furthermore, these data highlight the significance of access to animals under managed care that provide physiological measurements under voluntary conditions.

In past studies, changes in CO during maximal exercise in land mammals appeared to be mainly regulated by an increase in f_H (Evans and Rose, 1988; Taylor et al., 1987). Similar results have been reported in pinnipeds during voluntary and forced breath-holds, but there have also been instances where SV changed substantially both during static breath-holds and active submerged swimming (Blix et al., 1983, 1976; Elsner et al., 1964; Murdaugh et al., 1966; Ponganis et al., 2006, 1991, 1990). Recently, considerable changes in SV were reported in voluntarily participating bottlenose dolphins following exercise, or during and following static breath-holds (Fahlman et al., 2019b; Miedler et al., 2015). Immediately following a brief bout of high-intensity exercise, the RSA-corrected if_H increased by 104% and iSV by 63% (Miedler et al., 2015). During a static breath-hold, the average RSA-corrected iSV decreased by between 1% and 21% while the average RSA-corrected if_H decreased by 18% (Fahlman et al., 2019b). During recovery, the RSA-corrected iSV and if_H increased by as much as 34% and 117%, respectively. Thus, similar to the studies in terrestrial and marine mammals, the changes in RSA-corrected CO in the bottlenose dolphin appear to be influenced more by changes in f_H, but there are also considerable changes in SV, during both static diving and exercise (Fahlman et al., 2019b). In the current study, the variation in if_H and iSV changed predictably following and between breaths, and a 56% decrease in if_H (from 90 beats min⁻¹ to 40 beats min⁻¹) caused a 32% decrease in iSV (Fig. 3C).

Studies have focused on assessing the magnitude of the dive response as the difference between the f_H during the dive and f_H for an inactive animal at the surface (Andrews et al., 1997; McDonald et al., 2018; Noren et al., 2004, 2012). In marine mammals, the aquatic breathing pattern and the cardiorespiratory coupling between f_H and f_R complicates the ability to estimate a ‘normal’ f_H for an inactive animal at the surface (Andrews et al., 1997; Cauture et al., 2019; Fahlman et al., 2019b; Fahlman et al., 2017). Because of this confounding effect of f_R, we propose that studies looking at the cardiovascular changes associated with diving have to account for this cardiorespiratory coupling. This is not a new suggestion, and previous studies have also expressed the importance of clearly defining a normal f_H as this would affect conclusions about whether animals experience a true bradycardia during a breath-hold (Belkin, 1964; Kooyman, 1985). In past studies, when using a more restrictive definition for normal f_H, it was shown that the freshwater river cooter did not experience a diving bradycardia, while the Weddell seal only experience a diving bradycardia during dives >8 min (Belkin, 1964; Kooyman, 1985). Thus, in species with substantial variation in f_H and SV with each breath, it is important to account for the confounding effect of RSA. Strictly defined or controlled conditions are common when measuring energy use, where, for example, the basal metabolic rate in homeotherms is measured under standard conditions (Kleiber, 1947; Pedersen et al., 2020; Rosen and Trites, 2013; Worthy et al., 2013). When an animal is not post-prandial, but the metabolic measurements are done with the animal inactive, the measurements are commonly referred to as resting metabolic rate. Comparative studies on cardiac function, e.g. f_H and SV, in contrast, generally lack strictly defined conditions that allow a direct comparison. For example, it is often not reported whether the study subjects were post-prandial (Fahlman et al., 2019b; Jones et al., 1973; Kaczmarek et al., 2018; Noren et al., 2004), even though this has been shown to alter metabolic rate and cardiac function (Young et al., 2011). Similarly, many studies investigating cardiac function either in marine or terrestrial mammals do not report f_R, despite its known effect on f_H (Cauture et al., 2019; de Burgh Daly, 1986; Guillén-Mandujano and Carrasco-Sosa, 2014). While this may not be critically important for studies on animals that breathe more or less continuously, even terrestrial mammals experience considerable variation in f_H when breathing intermittently (Laske et al., 2010). In marine mammals, and specifically cetaceans, where breaths are separated by long respiratory pauses and significant changes in both f_H and SV have been observed following a breath (Figs 1–3), how much variation in average f_H do changes in f_R cause?

As an example of the impact of f_R on f_H, the mean (±s.d.) f_H for the false killer whale was 48±7 beats min⁻¹ during a period when f_R was 1 breath min⁻¹, as compared with 70±9 beats min⁻¹ when f_R was 4 breaths min⁻¹. In the bottlenose dolphin, the if_H varies from around 80–100 beats min⁻¹ immediately following an inspiration, to between 35 and 50 beats min⁻¹ between breaths (Fig. 3 and see fig. 1 in Cauture et al., 2019). Consequently, the f_R significantly affects the f_H for an inactive animal when measuring the number of beats over a pre-determined duration, and is especially dependent on where in the breathing cycle f_H is measured. In previous studies in the bottlenose dolphin, where the effect of f_R was not evaluated or considered, the f_H for an inactive animal was between 78 and 105 beats min⁻¹ (Noren et al., 2004, 2012; Williams et al., 1993, 2015). From these comparisons it may be prudent to suggest that the measurement conditions, such as fasted or fed, and f_R should be reported. In addition, the conditions for f_H for an inactive animal need to be clearly defined to allow comparisons between studies and species (Belkin, 1964; Kooyman, 1985). This may be particularly important in studies that attempt to evaluate the evolutionary significance of the dive response, and those used to compare cardiac measurements across species, especially marine and terrestrial, as the results may be affected by f_R. By using RSA correction, comparisons within and between species and activity states (inactive, swimming, diving) could be made and be independent of f_R (Cauture et al., 2019; Fahlman et al., 2019b; Miedler et al., 2015).

From the data presented here and in previous studies (Andrews et al., 1997; Bickett et al., 2019; Cauture et al., 2019; Fahlman et al., 2019b; Harms et al., 2013; Kaczmarek et al., 2018; McDonald et al., 2018; Miedler et al., 2015), it appears that f_R can have a significant effect on measured cardiac function. While terrestrial mammals experience large changes in f_H when f_R is modified (Laske et al., 2010; Mortola et al., 2016), we are not aware of any study in terrestrial mammals that have attempted to remove the effect of f_R on f_H, making direct comparisons difficult. With increasing f_R, the RSA becomes less obvious as the average f_H during the interbreath interval remains elevated. The almost continuous breathing pattern in terrestrial mammals may be one reason why the effect of f_R on f_H has not been considered in comparative studies.

However, while RSA-correcting data may provide comparative cardiac indices, it does not provide a true estimate of the average f_H, SV or CO in an inactive cetacean as it omits the dynamic period from the end of the breath until the if_H and iSV have stabilized after about 12–20 s (Figs 1–3). Thus, the period of elevated if_H and iSV is not accounted for, and the estimated value would be lower than the value from continuous measurements. Using the temporal relationships between time, if_H and iSV (see Results, ‘Cardiorespiratory dynamics’) the average f_H and SV for different f_R were estimated. For an f_R of 2 breaths min⁻¹, the average f_H, SV and CO would be 11% (e.g. 54 beats min⁻¹ as compared with 48 beats min⁻¹; see Results, ‘Estimating average f_H, SV and CO for a range of f_R’), 6% and 16% higher than the RSA-corrected value. At an f_R of 6 breaths min⁻¹ the same values would be 25%, 14% and 35% higher than the RSA-corrected value, respectively. For an f_R of a terrestrial mammal of 170 kg, the average values would be similar to those predicted from a terrestrial mammal (Fig. 4). Thus, when accounting for the differences in f_R it appears that the average f_H and SV are similar between marine and terrestrial mammals of comparable body size. However, these calculations are done for the bottlenose dolphin, and clearly a better understanding of the cardiorespiratory coupling in mammals is important to correctly assess physiological differences without confounding effects between species.

While the RSA-corrected f_H and SV do not account for the values during the dynamic period, they do provide values that are not confounded by breathing and can therefore be used for comparative purposes between species. The data in Fig. 4 show that the RSA-corrected f_H was lower than that predicted from terrestrial mammals by between 34% and 51% for all species (Fig. 4). The RSA-corrected SV for an inactive animal was 94% greater in the juvenile beluga whale as compared with the predicted value from a 170 kg terrestrial mammal (Fig. 4; Bishop, 1997; Seymour and Blaylock, 2000; Stahl, 1967). In the dolphin, the RSA-corrected SV was 9% lower compared with that of a similar-sized terrestrial mammal (Fig. 4), while for the false killer whale, the RSA-corrected SV was almost the same as that of a similar-sized terrestrial mammal (Fig. 4) (Seymour and Blaylock, 2000). The RSA-corrected CO in the dolphin and false killer whale was 34% and 33% lower, respectively, as compared with that of a similar-sized terrestrial mammal (Bishop, 1997; Stahl, 1967). In the beluga whale, in contrast, the RSA-corrected CO was 19% higher as compared with that of a terrestrial mammal of similar size, mainly due to the high SV. Interestingly, both the RSA-corrected resting SV and CO were considerably higher in the beluga whale calf as compared with the bottlenose dolphins of similar size (Fig. 4). As heart size correlates directly with M_b during ontogeny in smaller cetaceans (Mallette et al., 2016; McLellan et al., 2002), it is unlikely that these differences are caused by variation in heart size and body mass growth. However, the aortic valve orifice was greater in both the larger adult false killer whale and juvenile beluga whale as compared with dolphins. Thus, aortic valve orifice diameter may be an important anatomical bottleneck in cardiovascular performance as its size appears to limit SV, and thereby CO. Anatomical data on aortic valve orifice diameter from different species, age classes and sexes would be interesting to investigate how anatomy may limit cardiovascular performance.

To compare the cardiorespiratory coupling between species, we plotted f_R and f_H from bottlenose dolphin, false killer whale, juvenile beluga whale, killer whale and pilot whale. Reported data for the harbor porpoise, adult beluga whale and gray whale that were available from the literature were also included (Bickett et al., 2019; Elmegaard et al., 2019; Ponganis and Kooyman, 1999; Wahrenbrock et al., 1974). This regression provides an estimate how f_H varies with changes in f_R between cetaceans (Fig. 5). As a comparison we plotted f_R and f_H for terrestrial mammals using allometric equations (Stahl, 1967). We also added data points for the donkey, Thoroughbred horse and elephant as the allometric equations developed by Stahl (1967) did not account for animals larger than humans. The resulting regression showed that the slope of the regression of f_R and f_H is much greater in the cetacean than in terrestrial mammals extrapolated from the allometric equations by Stahl (1967). This is a conservative estimate, as the addition of data points for the elephant, horse and donkey in the M_b range from 165 to 6000 kg suggests that the regression for terrestrial mammals is even less steep. In other words, for each breath in a cetacean there are almost 12 heart beats, while for a terrestrial mammal there are 3 (Fig. 5). This difference is due to a large change in f_H for each breath in the cetacean, causing a greater RSA as compared with a terrestrial mammal. In humans, a breath causes a change in f_H of between 10 and 20 beats, while in the bottlenose dolphin this variation is between 40 and 60 beats (Cauture et al., 2019; Mortola et al., 2016). The results here further support this and show that f_H changes associated with f_R are greater in cetaceans (Fig. 5), and the dynamic changes between breaths are consistent between cetacean species (Fig. 3) and affect both f_H and SV (Fig. 2). Past work has shown that the f_H f_R⁻¹ is significantly greater not only in cetaceans but also in a larger range of marine mammals (Mortola, 2015). It was suggested that this is a non-respiratory adaptation to provide additional buoyancy. While this is an interesting suggestion, we propose that this greater cardiopulmonary coupling may be a physiological trait that help improve gas exchange. Interestingly, when overlaying the data for the hibernating bear, although lower, it is closer to the line for the cetaceans as compared with the terrestrial mammals (Fig. 5). It was suggested that this extensive RSA in the hibernating bear would help conserve cardiac work while respiratory coupled tachycardia would help to maintain gas exchange and tissue perfusion (Laske et al., 2010). In the elephant seal, it was suggested that the differences in cardiorespiratory responses at sea versus on land could be a way to ensure rapid recovery rather than adequate gas exchange (Andrews et al., 2000). Thus, the steeper slope for cetaceans may indicate a greater cardiovascular response for each breath, which may help maximize gas exchange during short surface intervals.

In the northern elephant seal (Mirounga angustirostris), development of RSA appears to correlate positively with the duration of an apneic event, and negatively with both apneic and eupneic f_H (Castellini et al., 1994b). It was also reported that the changes in f_H during diving at sea, during a sleep apnea on land, or with RSA were similar and it was suggested that these may be regulated by a similar mechanism (Castellini et al., 1994a). In another study, the average f_H and f_R while at the surface at sea were considerably higher (f_H: 107±3 beats min⁻¹; f_R: 22.0±2.0 breaths min⁻¹) as compared with those on land (f_H: 65±8 beats min⁻¹; f_R: 9.2±1.3 breaths min⁻¹), but the mean apnea f_H was not different at sea or on land (Andrews et al., 2000). Thus, the higher f_H at the sea surface may, at least in part, be confounded by a higher f_R as the animal returns to the surface to breath. In a study on harbor seals, it was noted that at times the seals performed what was termed ‘top-up’ dives, where the seal spent a brief moment at the surface, only long enough to take a single breath, and then submerged again (Fedak et al., 1988). It was suggested that these top-up dives could be important circulatory events that helped improve gas exchange and reduce recovery time (Andrews et al., 1997; Fedak et al., 1988). A more recent theoretical study showed that the cardiac responses during the surface interval following shallow and deep diving in bottlenose dolphins is to maximize gas exchange, which helps to replenish more rapidly the O₂ stores and remove the CO₂ and thereby minimize the surface duration (Fahlman et al., 2018). Thus, the greater cardiovascular response to breathing in marine mammals may be a physiological trait that ensures rapid gas exchange and helps reduce the surface interval during a foraging bout.

This is the first study to report comparative semi-continuous measurements of f_H, SV and CO, and continuous ECG in a range of small and medium-sized odontocetes. We have shown that both if_H and iSV initially increase and then decrease exponentially following a breath, which agrees with data from static breath-holds and following high-intensity exercise in the bottlenose dolphin (Fahlman et al., 2019b; Miedler et al., 2015). Of special interest is the higher mass-specific SV and CO in the beluga whale calf as compared with that in similar-sized dolphins and the larger false killer whale, possibly reflecting increased metabolic demand due to active growth. In addition, we show that in cetaceans, f_H responds more strongly to variation in f_R as compared with that in terrestrial mammals, resulting in a greater RSA in the former. We propose that RSA may be important to maximize gas exchange at the surface to minimize the time at the surface between dives. The data presented in the current study provide valuable comparisons within and between species that are important to understand physiological limitations in cetaceans.

A special thanks to all the trainers and staff at Dolphin Quest-Oahu, the Oceanogràfic and Sea Life Park-Oahu, and to Laura Surovik and the animal training staff at SeaWorld Orlando who made this study possible through their dedication and motivation. Dolphin Quest, the Oceanogràfic, SeaWorld Orlando and Sea Life Park kindly supported us by providing animals, crew and access to resources. This is a SeaWorld Parks and Entertainment Corporation technical contribution number TC-2020-01. A special thanks to Ann Pabst for the discussion about anatomical ontogeny in dolphins, and for sending the data. We would like to thank Peter Madsen for lending us the mold for the ECG suction cups. A special thank you to Jacopo Mortola for stimulating discussions about cardiorespiratory function in marine mammals. We are grateful for the excellent comments made by the referees on earlier versions of the manuscript, which helped change the scope and direction of the objectives, and we would like to thank them for their helpful criticism.