Re-evaluating the significance of the dive response during voluntary surface apneas in the bottlenose dolphin, Tursiops truncatus

The dive response is well documented for marine mammals, and includes a significant reduction in heart rate (fH) during submersion as compared while breathing at the surface. In the current study we assessed the influence of the Respiratory Sinus Arrhythmia (RSA) while estimating the resting fH while breathing. Using transthoracic echocardiography we measured fH, and stroke volume (SV) during voluntary surface apneas at rest up to 255 s, and during recovery from apnea in 11 adult bottlenose dolphins (Tursiops truncatus, 9 males and 2 females, body mass range: 140–235 kg). The dolphins exhibited a significant post-respiratory tachycardia and increased SV. Therefore, only data after this RSA had stabilized were used for analysis and comparison. The average (±s.d.) fH, SV, and cardiac output (CO) after spontaneous breaths while resting at the surface were 44 ± 6 beats min−1, 179 ± 31 ml, and 7909 ± 1814 l min−1, respectively. During the apnea the fH, SV, and CO decreased proportionally with the breath-hold duration, and after 255 s they, respectively, had decreased by an average of 18%, 1–21%, and 12–37%. During recovery, the fH, SV, and CO rapidly increased by as much as 117%, 34%, and 190%, respectively. Next, fH, SV and CO rapidly decreased to resting values between 90–110 s following the surface apnea. These data highlight the necessity to define how the resting fH is estimated at the surface, and separating it from the RSA associated with each breath to evaluate the significance of cardiorespiratory matching during diving.

Heart rate, stroke volume and cardiac output during surface apnea. Of the 13 dolphins that participated in the study, the diameter of the aortic valve orifice could be measured in 11 animals (Table 1). These 11 dolphins participated in a total of 173 voluntary surface apneas at rest, ranging in duration from 20 s to 255 s ( Table 2). For if H , the best model to describe the effect of surface apnea included the maximal duration of the dive, and the time during the dive (Fig. 1A, Table 3). The best models to describe the effect of apnea on iSV included M b and apnea duration (Fig. 1B, Table 3). For iCO the best model included M b , time during the dive, and the maximal duration of each breath-hold (Fig. 1C, Table 3).
Heart rate, stroke volume and cardiac output during recovery. The average (n = 10) if H , iSV, and iCO after the 10 first seconds following the first breath from a voluntary surface apnea at rest was 78 ± 8 beats · min −1 (t-test compared to pre-dive rest, t-value = 11.1, df = 21 P < 0.01), 225 ± 41 ml (t-test compared to pre-dive rest, t-value = 2.9, df = 19 P < 0.01), and 17612 ± 3047 ml · min −1 (t-test compared to pre-dive rest, t-value = 9.0, df = 19 P < 0.01), respectively. For if H ( Fig. 2A), iSV (Fig. 2B), and iCO (Fig. 2C, Table 3) the best models to describe the recovery from surface apnea included recovery time. Neither the duration of the surface apnea, nor M b warranted inclusion in the model (P > 0.3 for all).

Discussion
Given the importance to understand the effects of O 2 delivery during exercise and submersion, determining the cardiovascular responses are important to understand physiological function in cetaceans. The objective with the current study was to evaluate if H , iSV and iCO before, during and following voluntary surface apneas at rest to help improve our understanding of cardiorespiratory responses associated with diving. Similar to previous studies, we report a significant RSA 10,16,17,36,37 , which alters how the resting f H is evaluated and thereby the magnitude of the apparent dive response. As RSA is known to be affected by both tidal volume (V T ) and f R in humans and dolphins 37,41 , we propose that studies that are aimed at evaluating the cardiovascular responses associated with diving and exercise need to evaluate the effect of RSA when estimating the surface f H that is made as a comparison. For example, if if H increases to approximately 120 beats · min −1 immediately following a dive, what www.nature.com/scientificreports www.nature.com/scientificreports/ proportion of this increase is merely due to elevated f R ? To answer this, the cardiorespiratory coupling should be determined for each species to accurately determine the effect of RSA (see Fig. 1 in 37 , and Supplemental Material).
The cardiovascular changes associated with diving and exercise while submerged have long fascinated eco-physiologists trying to understand how marine vertebrates manage gases during diving. Most initial studies were performed in a laboratory setting during forced submersions, but the development of bio -logging tools, or use of medical technology have opened up new avenues for research and understanding the cardiorespiratory changes while submerged under natural conditions, or in studies on animals under voluntary control 7,10,17,21,35,37,39,42 . The changes in f H associated with diving have been studied intensively in numerous species 26 . Only a few studies have measured CO in live marine mammals, and most of these studies have been made in pinnipeds 5,9,15,22,[27][28][29][30] . Of these, only three have been performed in unrestrained and/or free swimming animals 9,15,22 . In the sea lion, SV and f H were not reported 15 , but in the seal, CO and f H increased while SV decreased as the animal was swimming at the surface 9 . A similar increase in CO was observed during exercise in the submerged harbor seal, while SV also increased slightly 9 . We are only aware of one study that has measured SV in cetaceans either at rest or following high intensity exercise 36 . In a study looking at the development of the dive response in bottlenose dolphins, it was reported that there were no differences with age in resting f H at the surface, but the f H during submersions was reduced to lower values in older animals 7 . In the harbor porpoise, the f H before and during diving was altered depending on the anticipated dive duration, suggestive of cognitive control of f H 21 . In exercising dolphins there was a linear increase in f H with increasing metabolic rate 10 . A similar complex relationship was observed in freely swimming and exercising harbor porpoises where the degree of diving bradycardia  www.nature.com/scientificreports www.nature.com/scientificreports/ was altered by the duration and activity level of the submersion 35 . An important detail in the current study is whether the post-breath period that cause variation in f H is included in calculating the resting f H as this response is pronounced in this taxonomic group and depends on the f R 37 .
In our past work 36,37 , and based on the result in the current study, we propose that for better interpretation of the magnitude of the cardiac adjustment during a breath-hold, the comparison should be made with surface values that separate the confounding influence of the RSA. There could be different ways to account for this, and one would be to allow for a long enough resting period that assured that the animals f R was normal, or at least report the f R so that comparisons could be made between studies. However, as that would still involve confounding the autonomic cardiac response with the voluntary process of breathing, an alternative would be to measure the resting heart rate with minimal influence of the RSA. Thus, we propose that it can be done by measuring the if H as we have done in the current study, or continuous ECG measurements that allows the changes in if H to be modeled and the stabilized f H predicted 37 . We propose to call this value the RSA-corrected resting f H , to differentiate this with the resting f H reported in most studies.
The average RSA-corrected resting f H in the current study was not different from those reported in past studies using the same individual dolphins and the same methodology, or continuously measuring the ECG (f H = 41 ± 9 beats min −1 n = 13, 2-tailed t-test, P = 0.33, t-value = 1.0, df = 24) 36,37 , but was considerably lower and less variable as compared with resting f H 's reported in previous studies in the bottlenose dolphin (ranging from 60-105 beats min −1 ) 7,17,39 . The RSA-corrected resting f H , SV and CO, where estimated to exclude the changes in f H , SV and CO associated with respiration 36,37 . As both f H and SV increase following a breath, our average readings were lower as compared with an average obtained if the f H was estimated over the whole inter-breath cycle, which included the RSA (see Fig. 1 in 37 , or Supplemental Material). Conversely, estimating resting f H over a determined period of time includes the variation associated with the RSA, and the average will therefore depend on the duration of the measurement and the f R within this period.
Thus, the higher f H reported in previous studies may be confounded through inclusion of RSA, and without reporting the f R a comparison is difficult. Consequently, estimating resting f H without accounting for the RSA result in a higher resting value, and if used to assess the dive response it will result in higher values than when using the RSA-corrected resting f H . For example, in one study in the bottlenose dolphin, the estimated resting f H while at the surface was 105 ± 8 beats min −1 , and decreased to 40 ± 6 beats min −1 , while submerged at 15 m for an average dive duration of 85 ± 51 s (range 14-160 s) 39 . The RSA-corrected resting f H during a voluntary surface apnea up to 160 s in the current study was similar to the resting diving f H reported in the past study 39 , and similar to the resting f H for spontaneous breathing at the surface after accounting for the RSA (27 to 67 beats min −1 , see refs 36,37 ). Thus, our results suggest that the changes in f H associated with diving, after accounting for RSA, are less pronounced than previously reported, and only become significantly lower than resting surface values during apneas exceeding 2.5-3 min (Fig. 1A). Thus, the results presented here provide an interesting perspective that when accounting for the RSA dolphins do not exhibit the pronounced diving bradycardia associated with the stress during forced dives or those that include the RSA when estimating the resting f H , except during extended apneas. Whether this is a universal trait in other marine mammals, or how exercise or depth influence these changes remains to be determined once past and future studies evaluate the RSA-corrected resting f H in voluntary diving animals. Interestingly, similar results were found in elephant seals where the RSA minimum and f H during apnea appeared similar 43 . In addition, the RSA developed with age, and its magnitude correlated with dive duration 43 . Thus, the RSA may be an important physiological index that is more correlated with diving ability in addition to cognitive ability to alter f H and SV to accurately match ventilation and perfusion to improve gas exchange 24 .
Both resting sSV, and sCO were significantly higher as compared with a previous study assessing cardiovascular changes following a bout of exercise (sSV = 0.79 ± 0.14 ml · kg −1 , 2-tailed t-test, P < 0.01, t-value = 3.4, df = 22; sCO = 32 ± 9 ml · min −1 · kg −1 , 2-tailed t-test, P < 0.01, t-value = 3.6, df = 22) 36 . However, both sSV and sCO were considerably lower as compared with those measured at rest in the harbor seal (Phoca vitulina, sSv = 1.8-3.1 mL · kg −1 , sCO = 102-394 mL · min −1 · kg −1 ) 9 , and California sea lion (Zalophus californianus, sSv = 2.0 mL · kg −1 , sCO = 150-180 mL · min −1 · kg −1 ) 15 using thermodilution techniques. We believe that these differences between species could in part be explained by the lower f R , breath duration, and higher flow-rates, and V T 's in the cetaceans as compared with the pinnipeds 38,44 . Thus, if respiration significantly alters cardiac function (f H and SV), it is vital that these variables are compared during periods between breaths to minimize the impact of respiration on cardiac function. By standardizing these measurements to prevent the confounding effect of RSA, it would allow inter-and intra-species comparisons by reducing the variability caused by the RSA.
It has been suggested that the extreme changes in f H associated with forced submergence, when an animal does not know or has no control over the duration of the apnea, reflect an animal that prepares for a maximal asphyxic challenge 20 . For this reason, Blix 20 proposed that studies that assess the dive response should clearly distinguish between forced and voluntary diving. An additional issue that may significantly alter variation in f H , especially in  . Shown are the parameters (±s.e.) for the best model, the χ 2 , and P-value against nested models and the r 2 m /r 2 c are the marginal and conditional r 2 , respectively. SV is in ml, CO in ml · min −1 , durations in s, and M b in kg.
www.nature.com/scientificreports www.nature.com/scientificreports/ studies on trained animals, is the suggestion that marine mammals have cognitive control of f H 20-23 . As repeated measurements were performed for each trial, the time during the dive allowed us to investigate temporal changes during a breath-hold. The maximal dive duration was the total duration of the apnea, which allowed us to assess whether the initial cardiovascular changes were different for a long or short breath-hold, possibly indicating a learning effect or cognitive ability to alter f H and SV. Our results showed a greater initial drop in f H during longer dives (Fig. 1A), which provide additional support of voluntary control of the diving related changes in f H 20-23 . Previously, it has been proposed that by fine tuning the alveolar ventilation (  V A ) and lung perfusion, marine vertebrates are able to selectively exchange O 2 and CO 2 during diving, while minimizing N 2 exchange 24 . This mechanism relies in part on voluntary control of pulmonary and systemic perfusion, and the ability to selectively perfuse collapsed regions of the lung 24 . Disruption of such refined control of pulmonary blood flow could have severe consequences in gas management, and may explain how species that are normally able to avoid diving related problems experience gas bubble disease when exposed to stressful situations while diving 32,[45][46][47][48][49][50] . While our study does not provide direct evidence of voluntary control of perfusion, we show indirect evidence that the www.nature.com/scientificreports www.nature.com/scientificreports/ dolphins alter f H and SV depending on the length of the dive, similar to the results in the harbor porpoise 21 . As proposed by Mottishaw et al. 19 , the cardiovascular changes during diving may be a complex physiological response altered by a number of factors such as voluntary/anticipatory adjustment, submersion, exercise, stress, and fear. Thus, the cardiovascular responses during diving may be a much more plastic physiological trait rather than a purely autonomic response, or potentially merely an extension of the RSA.
Following a voluntary surface apnea up to 255 s, f H , SV, and CO increased by 78% (maximal value: 164%), 26% (maximal value: 73%), and 123% (maximal value: 229%), respectively (Figs 1 vs. 2). Similar changes in f H at maximal exercise are difficult to assess as they generally do not account for the RSA 10,51 . However, maximal f H , SV, and CO 10 s following a single bout of high intensity exercise increased by as much as 307%, 294%, and 727%, respectively 36 . Consequently, the cardiovascular changes during recovery from a voluntary surface apnea appear less extreme than measured following a high intensity bout of exercise. There are several possible reasons for the more moderate cardiovascular changes during recovery in the current study. First, the dolphins did not perform high intensity exercise and therefore did not have elevated metabolic rate. The elevated f H , SV and CO following the breath-hold in the current study was only used to replenish the O 2 stores and remove any CO 2 produced caused by the surface apnea. During exercise, the dolphins incur an O 2 debt that may be exacerbated by the elevated rate of O 2 consumption. This may require greater cardiovascular recruitment to rapidly repay the greater O 2 debt following exercise and may extend recovery as compared with voluntary apnea at rest. Increased aerobic work and greater O 2 debt may explain why if H , iSV and iCO took at least 4 min to recover following a high intensity bout of exercise 36 , while in the current study these variables had returned to baseline after approximately 90-110 s ( Fig. 2A-C). This agrees with a previous study showing that the O 2 stores had recovered after approximately 1.2 min following a surface apnea of up to 5 min 52 . Consequently, for an actively diving dolphin we would expect the recovery f H , SV and CO to reach higher values and/or increase for a longer duration as higher activity would increase the O 2 debt. In grey seals, the surface interval increased with dive duration for short dives (<7 min) but not for long dives 53 . In extreme divers, such as Weddell and elephant seals there does not appear to be a clear relationship between dive duration, even those exceeding the aerobic dive limit, and surface interval duration 54,55 . One reason could be that seals partially recover while diving 55 . However, the surface f H increased with dive duration in the elephant seal 54 , but there was no clear relationship in the grey seal 53 . Similarly, the recovery f H during single dives correlated with dive duration and activity in the Steller sea lion 14 . Thus, these data provide evidence for excellent control of cardiovascular function to rapidly and efficiently manage metabolic gases.
In summary, this study provides the first non-invasive measurements of f H and SV before, during, and after voluntary surface apneas at rest in cetaceans. To account for the RSA, we measured the RSA-corrected resting if H and iSV only during inter-breath periods when f H and SV had stabilized following the breath. We therefore report lower average f H while resting at the surface, and propose that these results have lower variability and provide improved comparable values across individuals or species with varying respiratory effort. Further work should be done to define the significance of the RSA, its role in managing gas exchange, and the temporal changes to allow the RSA-corrected resting f H and SV to be defined within and between species. The if H , iSV and iCO decreased slightly with breath-hold duration as compared with the RSA-corrected resting if H . However, compared to past studies, the level of the diving bradycardia was considerably reduced. Following the apnea, the if H , iSV and iCO immediately increased as much as 164%, 73%, and 229%, respectively even during the stabilization period, but rapidly returned to resting levels within 50-60 s after several respirations. Based on our results, we propose that future studies that assess the diving related changes in cardiac function have to do similar correction to avoid the confounding effect of RSA, and past studies may have to be re-evaluated to tease apart the contradictory effects of submergence and exercise. Our data show that the initial diving if H is lower during longer breath-hold, which provides additional support that cetaceans have partial cognitive control over their diving bradycardia. This, provides additional support for the hypothesis that cetaceans can voluntarily alter gas exchange during diving.

Material and Methods
Animals. The study protocols were approved by the Animal Care and Welfare Committee of the Oceanogràfic Foundation (OCE-17-16 and amendment OCE- , and all experiments were performed in accordance with relevant guidelines and regulations. Ultrasound continuous flow Doppler was used to measure f H and SV at the level of the left ventricular outflow tract (at the level of the aortic valve orifice), during and following a surface apnea (Table 1) Table 1).
The animal ID, body mass, and age (known or estimated) at the time of the study are summarized in Table 1.

Experimental trials.
All experiments were performed using operant conditioning. Participation by the dolphins was voluntary, and the animals were not restrained and could refuse to participate or withdraw at any point during the experimental trial. Each experiment (trial) consisted of an animal staying stationary in the water in left lateral recumbence with the blow-hole out of the water, allowing ultrasound probe placement to find the left ventricle. To evaluate the effect of a surface apnea, the animals were conditioned to hold their breaths voluntarily as much as possible up to a maximum of 255 s ( Table 2). The dolphins were asked by the trainer to turn on their side with the blowhole submerged until they decided to end the breath-hold. The ultrasound examination continued throughout the apnea and recovery period to assess the cardiovascular changes during a breath-hold and recovery. For measurements of f H , and SV during apnea and recovery, the breath-hold and recovery durations were recorded. Out of the 13 animals, we were only able to measure the aortic diameter in 11 and SV and CO were therefore only estimated in 11 dolphins (Table 1). For that reason, only 11 animals participated in the voluntary breath-hold (Table 2).
www.nature.com/scientificreports www.nature.com/scientificreports/ Ultrasound data acquisition. The ultrasound machine (Vivid-I, General Electric) with a 1-3 MHz phased array probe was used to obtain left ventricular if H and iSV, as previously detailed 36 . CO was estimated as: iCO = if H × iSV. The if H was estimated from the flow traces, and the iSV calculated from the surface area of the aortic valve orifice multiplied by the velocity time integral of the systolic blood flow along the left ventricular outflow tract. The cross-sectional area of the aortic valve was calculated from the aortic valve orifice diameter as: surface area = radius 2 × π. In order to obtain left ventricular systolic blood flow, the "apical" long axis view of the left ventricular outflow tract and the aortic root was used to place the continuous flow Doppler parallel to the blood flow in the left ventricle and through the middle of the aortic valve orifice. We measured the aortic valve orifice diameter at the level of valve insertion at the different cardiac phases to confirm that its diameter was circular and constant for all flow-rates.
Our pas studies and initial assessments indicated significant changes in both if H and iSV following respiration 36,37 . Because of this significant respiratory sinus arrhythmia (RSA), at least 3 repeated measurements of f H and ventricular flow were made for each measurement and the average used as an estimate of the f H following a breath. To obtain resting data that were not affected by the RSA, we only analysed f H at least 12-20 s following a spontaneous breath and after the if H had stabilized for 5-7 s following the breath (see Supplementary  Material) 36,37 . At this point, the if H and iSV were averaged for at least 3 R-R intervals for each measurement. These were then averaged for each individual. The ultrasound data were saved either as 8-16 s movies or 2-dimensional images and later analysed using manufacturer specific software (See S1 in Supplementary Material).
The duration to locate the aortic valve orifice differed between animals based on the size of the cardiac acoustic window, animal movement, and whether the animal was performing the surface apnea or breathing spontaneously. We standardized the recovery phase to begin immediately following the first breath, and cardiovascular measurements began as early as 10 s following this breath when the if H again stabilized following the RSA.
Data assessment and statistical analysis. We separated the analysis into one of three conditions: pre-apnea, apnea, post-apnea. The relationship between a dependent variable (CO, SV and f H ) and experimental covariates; body mass (M b ), time during dive, the total dive duration for a given dive (max duration), or time since end of apnea (recovery time), was analyzed using linear-mixed effects models (lme, R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, version 3.3.3, 2016). We log 10 -transformed f H , SV, CO, time during dive, max dive duration, recovery time and M b 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 56 . Initially, a univariate analysis on each independent variable was performed; only those variables with a P-value < 0.10 (Wald's tests) were considered in a multivariate analysis. Best models of remaining variables were chosen by the log-likelihood (LL) ratio test, and the Akaike information criterion (AIC) and significant parameters assessed by the t-value between the estimate and its standard error. 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 ± standard deviation (s.d.), unless otherwise stated.

Data Availability
The data used in this study are freely available at the following link: osf.io/wdfmz.