Ventilation and gas exchange before and after voluntary static surface breath-holds in clinically healthy bottlenose dolphins, Tursiops truncatus

Marine mammals forage underwater but need to return to the surface to replenish O₂ stores and to remove the CO₂ produced during dives. While holding their breath underwater, they use the O₂ in their blood and tissues to fuel aerobic metabolism, which produces CO₂. Consequently, the O₂ levels decrease and the CO₂ levels increase during a dive (Scholander, 1940). Marine mammals have developed several strategies to increase time underwater and, therefore, foraging efficiency. For example, the O₂ stores are increased by elevated haematocrit, blood volume, muscle mass and muscle myoglobin levels (Ponganis, 2015). The lungs in most species are not comparatively larger than those in terrestrial species; instead, deep divers seem to have relatively smaller lungs to minimize gas exchange during diving and thereby decompression sickness (Piscitelli et al., 2010; Scholander, 1940).

Diving mammals also seem to increase foraging efficiency by minimizing time at the surface (Boutilier et al., 2001; Carbone and Houston, 1996; Fahlman et al., 2008; Halsey et al., 2003). The length of time required at the surface is constrained by the efficiency of recovery of blood and tissue gas levels. As the diver returns to the surface, cardiac output, which has been reduced during the dive, begins to increase, and the ascent tachycardia that increases perfusion is thought to help prepare the diver to more efficiently restore O₂ and remove CO₂ (Thompson and Fedak, 1993). It has also been proposed that this pre-surface tachycardia helps remove dissolved N₂, which helps to reduce the risk of gas emboli (Fahlman et al., 2007). Empirical data suggest that the animals incur an O₂ debt during a dive bout that is not repaid until the recovery period at the end of the dive bout. The dive behaviours observed in pinnipeds (Fahlman et al., 2008; Kooyman et al., 1973) are consistent with optimal foraging models, which indicate that the length of the surface interval is determined by the need to restore O₂ stores (Carbone and Houston, 1996; Halsey et al., 2003; Kramer, 1988). While insufficient uptake of O₂ would reduce the aerobic dive duration, insufficient removal of CO₂ could result in acid–base balance disturbances and lactate build-up. Thus, it has been proposed that that removal of CO₂ is the main driving force regulating surface interval duration (Boutilier et al., 2001; Fahlman et al., 2008; Reed et al., 1994, 2000). Recent reports that marine vertebrates may experience gas emboli and suffer decompression sickness (Fernández et al., 2005, 2017; García-Párraga et al., 2014) also suggest that at times accumulation of blood and tissue N₂ levels may limit diving and force animals to extend surface duration of the interval (Fahlman et al., 2014; Hooker et al., 2009). Thus, variation in the surface interval duration may have numerous causes, and changes in dive activity and behaviour may alter the duration at the surface. For example, if a species has to dive deeper to find food, this may result in accumulation of CO₂ and N₂, which may result in longer surface intervals to remove the additional gas taken up or produced. Consequently, understanding the variables that limit the surface interval and how gases are managed and recovered may help provide clues to improve our understanding of cardiorespiratory limitations.

The recovery time at the surface is reduced by increased cardiac output in bottlenose dolphins (Tursiops truncatus) (Miedler et al., 2015). Changes in ventilation following diving, in contrast, are not as obvious or not reported (Kooyman et al., 1971; Kooyman et al., 1973; Reed et al., 1994). Thus, the surface interval is a complex dynamic state that requires sufficient gas exchange to enable efficient foraging and the interaction between ventilation and perfusion will significantly alter gas exchange.

In the current study, we investigated the changes in respiratory effort [respiratory flow (V̇), breathing frequency (f_R) and tidal volume (V_T)] and gas exchange (expired O₂) in the bottlenose dolphin before and after periods of surface breath-hold ranging from 34 to 292 s. Our results indicate a complex relationship between respiratory effort and end-expired O₂ content that helps minimize recovery duration following breath-holds. Thus, bottlenose dolphins appear to vary V_T, f_R and breath duration depending on breath-hold duration to enhance gas exchange and minimize recovery duration. From our results, we estimate that a 196 kg bottlenose dolphin can restore its O₂ stores in approximately 1.2 min, following a surface breath-hold of 157 s.

Eleven adult bottlenose dolphins, Tursiops truncatus (Montagu 1821), of varying age and size (Table 1) were used for all experiments. The dolphins were housed in a facility in either Hawaii (Dolphin Quest – Oahu) or Las Vegas (Siegfried and Roy's Secret Garden). The ID, body mass (M_b, measured within 2 weeks of each experiment), straight length (rostrum to midline fluke notch) and age of the animals are summarized in Table 1. Prior to initiating the study, animals were desensitized to the equipment and trained for novel research-associated behaviours. Therefore, all experimental trials were performed with the dolphins using operant conditioning and participation by the dolphins was voluntary (the animals were not restrained and could refuse to participate or withdraw at any point during the experimental trial). Each experiment (trial) began with one animal remaining stationary in the water, allowing placement of the appropriate equipment (see fig. 1 in Fahlman et al., 2015), breathing spontaneously while continuous measurements were made for between 210 and 400 s (including a 90 s period during which the dolphin was allowed to calm down, ‘pre-breath-hold’). This approach allowed collection of experimental data on lung function in dolphins that were in a relaxed, physiologically normal state.

Next, the dolphins were trained to turn upside-down (ventrum towards the water surface) and keep their blowhole submerged while holding their breath without moving (‘breath-hold’). Breath-holds of pre-determined duration were terminated by a signal from the trainer. Breath-hold durations were based on the history and comfort level of the individual dolphin. No animal was restrained, so they could terminate the breath-hold and/or withdraw from a trial at any time. At the end of the breath-hold, when the dolphin righted itself, the spirometry equipment was immediately placed over the blowhole, allowing measurement of V̇ and expired O₂ content (see below) during recovery (‘recovery’). For this study, the recovery period lasted 300 s from the first breath after the apnoeic period (please note that in four trials the recovery terminated early, with a minimum recovery period of 181 s, as the dolphin decided to end the session; see Fig. S1 for the layout of a trial).

The procedures and equipment were identical to those used in our previous studies (Fahlman et al., 2015, 2018a,c), and the procedure is briefly summarized here. V̇ was measured using a custom-made Fleisch-type pneumotachometer (Micah Brodsky, V.M.D. Consulting, Miami, FL, USA; Mellow Design, Valencia, Spain), which housed a low-resistance laminar flow matrix (item no. Z9A887-2, Merriam Process Technologies, Cleveland, OH, USA). A low-resistance diffuser was added to homogenize the flow (Fahlman et al., 2018c), which helped resolve the difference in calibration factors for inspired (V̇_insp) and expired (V̇_exp) flow (Fahlman et al., 2015). A differential pressure transducer (Spirometer Pod, ML 311, ADInstruments, Colorado Springs, CO, USA) was connected to the pneumotachometer with two firm-walled, flexible tubes (310 cm length×2 mm i.d.). The differential pressure transducer was connected to a data acquisition system (Powerlab 8/35, ADInstruments), and the data were captured at 400 Hz and displayed on a laptop computer running LabChart (v.8.1, ADInstruments). The differential pressure was used to determine V̇ and was calibrated using a 7.0 l calibration syringe (Series 4900, Hans-Rudolph Inc., Shawnee, KS, USA). The signal was integrated and the V̇ determined as detailed previously (Fahlman et al., 2015).

A normal breath was considered a respiration that began with an exhalation followed by an immediate inspiration. Some breaths were partial, i.e. only exhalation or inhalation, and these were excluded in the analysis specifically looking at the first breath following the breath-hold.

The concentration of expired O₂ was subsampled via a port in the pneumotachometer and passed through a firm-walled, flexible tube (310 cm length×2 mm i.d.) connected to 30 cm length of 1.5 mm i.d. Nafion tubing, fed into a fast-response gas analyser (ML206, Harvard Apparatus, Holliston, MA, USA) at a flow rate of 200 ml min⁻¹. The gas analyser was connected to the data acquisition system and sampled at 400 Hz. The gas analyser was calibrated before and after the experiment using a commercial mixture of 5% O₂, 5% CO₂ and 90% N₂ (product no. 17L-340, GASCO, Oldsmar, FL, USA). Ambient air was used to check the calibration before and after each experimental trial.

We used V̇ and expired O₂ to estimate the rate of O₂ consumption (V̇_O2) as previously detailed (Fahlman et al., 2015). To investigate the dynamics of gas exchange, we calculated the instantaneous V̇_O2 (iV̇_O2) from the volume of O₂ taken up since the last breath.

Mean (±s.d.) air temperature and humidity during trials were 28.2±1.9°C (n=80, range: 22.6–33.0°C) and 62.0±7.0% (44–76%) in Hawaii, and 35.6±5.9°C (n=33, range: 21.3–39.0°C) and 17.1±9.5% (10–45%) in Las Vegas. The average water temperature in the lagoon in Hawaii was 26.0±0.6°C (range: 25.6–26.7°C) and in Las Vegas it was 25.0±0.9°C (range: 22.7–26.0°C).

We removed the first 90 s of data at the beginning of the trial before the breath-hold from the time the pneumotachometer was placed over the blowhole to allow the animal to settle down. Metabolic data are reported as the mean V̇_O2 before a breath-hold. The relationship between a dependent variable and experimental covariates was analysed using linear-mixed effects models (lme, R v.3.1.0, 2014, http://www.R-project.org/). 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). Best models were chosen by the log-likelihood ratio test against nested models. In this study, P≤0.05 was considered as significant and P≤0.1 was considered a trend. Data are presented as the mean±s.d., unless stated otherwise.

As V_T, f_R, V̇, end-expired O₂ and iV̇_O2 decayed exponentially throughout the recovery period, we log₁₀-transformed these variables to linearize the relationship with the number of breaths following the breath-hold. This allowed us to determine how M_b and dive duration affected the recovery following a breath-hold.

The six dolphins at the Dolphin Quest site underwent thorough clinical and diagnostic assessment to establish general and pulmonary health within 2 weeks of the experiments. This included radiographic (x-ray) and/or ultrasonographic imaging of the thorax with evaluation by one of two board-certified veterinary radiologists with experience in marine mammal diagnostic imaging. Additional testing included complete blood count, serum chemistry, sedimentation rate and sputum cytology. The health of the five dolphins at Siegfried and Roy's Secret Garden – Las Vegas was evaluated from their medical records, which included routine radiographs every 6 months (except for animal M00001). These routine radiographs were evaluated by one of the veterinary radiologists.

Data from a total of 879 spontaneous breaths before and 1017 breaths following voluntary surface breath-holds were collected from 11 adult male bottlenose dolphins (Tables 1 and 2) housed at Dolphin Quest – Oahu in May 2016 or at Siegfried and Roy's Secret Garden – Las Vegas in May 2015. A total of 42 breath-holds were performed with a mean±s.d. breath-hold duration of 157±60 s (Table 2).

End-expired O₂ concentration (Table 1), expiratory and inspiratory duration (expiratory: 370±49 ms; inspiratory: 421±58 ms), V̇ (V̇_exp: 24.9±6.6 l s⁻¹; V̇_insp: 15.5±2.6 l s⁻¹) and V_T (expiratory: 4.9±1.5 l; inspiratory: 5.0±1.5 l) did not change with breath number (χ²<2.5, d.f.=1, P>0.1 for all). V̇_exp was significantly higher than V̇_insp (paired t-test, t-value=7.11, d.f.=10, P<0.01), but there was no difference in inspiratory (V_T,insp) or expiratory (V_T,exp) V_T (paired t-test, d.f.=10, t-value=1.20, P>0.1). The duration of the expiratory phase was significantly shorter than that of the inspiratory phase (paired t-test, d.f.=10, t-value=3.4, P<0.01). The average f_R and expired minute ventilation (V̇_E) before breath-hold were 5.3±1.5 breaths min⁻¹ (range: 2.3–6.8 breaths min⁻¹) and 23.0±8.8 l min⁻¹ (range: 4.1–38.7 l min⁻¹), respectively. All but two short breath-holds (<60 s) began after a normal respiratory cycle of exhalation followed by inspiration.

The mean V̇_O2 before breath-hold was 0.925±0.322 l O₂ min⁻¹, and the average mass-specific V̇_O2 was 4.9±2.3 ml O₂ min⁻¹ kg⁻¹.

Following a breath-hold, neither the inspiratory breath duration (χ²=1.9, d.f.=1, P>0.1, 524±400 ms, n=12) nor total breath duration (χ²=2.1, d.f.=1, P>0.1, 866±447 ms) changed with breath number (N_breaths) or time to recovery (P>0.1), but there was a slight decrease in expiratory duration with each breath (χ²=8.1, d.f.=1, P<0.01, expiratory duration=348−0.6×N_breaths).

The end-expired O₂ concentration (Fig. 1A), instantaneous V̇_O2 (Fig. 1B), V_T (Fig. 2A), f_R (Fig. 2B), V̇_E (Fig. 2C) and V̇ (Fig. 3) changed during the recovery period. Following a breath-hold, there was considerable variation in the end-expired O₂ concentration (Fig. 1A). The first post-breath-hold breath was successfully sampled from a total of 31 breath-holds, and the end-expired O₂ decreased significantly with the duration of the breath-hold (one-way ANOVA, F=4.8, P<0.05) from an average of 15.5±4.0% for a breath-hold less than 1 min to 10.4±3.1% for a breath-hold over 3 min (Fig. 1C). The best equation that explained the variation in the end-expired O₂ data included duration of the breath-hold, breath number following the breath-hold and increasing V̇_E (Table 3). iV̇_O2 decreased with breath number and the duration of the breath-hold, and increased with increasing V̇_E (Table 3).

There was large variation in both the expired and inspired V_T following breath-hold, but there were no differences in the mean±s.d. V_T,insp (5.0±1.5 l) or V_T,exp (4.9±1.5 l) (paired t-test, t-value: 1.3, d.f.=10, P>0.1). We therefore only analysed and report the V_T,exp. V_T,exp increased with the duration of the breath-hold and with M_b, but decreased with breath number following the breath-hold (Table 3). The V_T,exp following the breath-hold, expressed as a percentage of the estimated total lung capacity (TLC_est) from an equation based on excised lungs from a range of species (Fahlman et al., 2011; Kooyman, 1973) or from an estimate from a diving dolphin (Ridgway et al., 1969), was 30% (range: 4–106%) and 54% (range: 7–184%), respectively. f_R increased with the duration of the breath-hold and decreased with breath number (Table 3). The average inspiratory flow (V̇_insp: 15.1±2.5 l s⁻¹) was significantly lower than the average expiratory flow (V̇_exp: 27.7±4.8 l s⁻¹, paired t-test, t-value: 7.1, d.f.=10, P<0.01). The average V̇_exp (31.7±6.8 l s⁻¹, range 3.3–64.5 l s⁻¹, n=11, t-value: 2.61, P<0.05) and V̇_insp (20.2±2.8 l s⁻¹, range 0.8–46.6 l s⁻¹, t-value: 4.14, P<0.01) for the first breath following the breath-hold was significantly greater than the average V̇ before the breath-hold (Table 1). Both V̇_insp and V̇_exp decreased with breath number and scaled allometrically with M_b (Table 3). The inspiratory duration did not change with breath number following a breath-hold (χ²=2.02, P>0.1), but both expiratory and total breath duration increased with the duration of the breath-hold and decreased with breath number (Table 3).

Thoracic radiographs showed no evidence of clinically active pulmonary or other intrathoracic disease in any dolphin. In two cases (9FL3 and 99L7), a very mild interstitial pattern was identified on radiographs. Another animal (63H4) had two small pulmonary nodules, suggestive of incidental granulomas in this species. The findings were not considered to be clinically relevant and may indicate incidental age-related changes or very mild fibrosis from a previous pulmonary insult.

Three animals had foci of surface lung pathology characterized ultrasonographically as alveolar-interstitial syndrome (63H4, 99L7 and 9FL3). This finding is abnormal but non-specific, and can be seen with lung fibrosis (i.e. from previous scarring), early pneumonia, atelectasis or pulmonary oedema (Lichtenstein et al., 1997). One other dolphin (01L5) had evidence of marginal lymph node enlargement, suspected to be due to a reactive process. The marginal lymph nodes are the primary lymphatic drainage site for the lung parenchyma in dolphins, and can readily be assessed with ultrasound (Cowan and Smith, 1999; Martony et al., 2017).

At the time of this study, there was no significant respiratory disease present that would require medical therapy for any of the dolphins in this study. This assessment statement was based on the judgement of the attending veterinarians at each facility, physical examination, cytological evaluation and thorough assessment of standard diagnostic laboratory parameters, and radiographic and ultrasonographic evaluation.

In the current study, we showed that bottlenose dolphins increase V̇, f_R and V_T following a static surface breath-hold. The V̇ during recovery increased more with increasing breath-hold duration, which helped to reduce the number of breaths before the end-expired O₂ had returned to levels equivalent to those before the breath-hold. Our results show that a 196 kg dolphin requires approximately 10 breaths to recover from a 159 s breath-hold, or approximately 1.2 min of recovery. Our results provide information about pulmonary function and gas exchange before and after a breath-hold in clinically healthy dolphins, which can be used as a comparison for wild populations.

One advantage of studying physiological function in animals under managed care is that the health of the study participants can be evaluated and monitored on an individual animal basis. In the current study, we confirmed that all animals were considered clinically healthy based on physical examination by experienced marine mammal veterinarians at each facility, and diagnostic imaging interpreted by one of the veterinary radiologists. In addition, studying trained animals under stimulus control that are participating voluntarily allows researchers to obtain baseline values with minimal stress. While some researchers rightly consider results obtained from animals under human care not to be perfectly representative of their free-ranging counterparts, certain information is virtually impossible to obtain from wild populations where controlled studies are extremely challenging to carry out. For that reason, we previously compared lung function values and metabolic rates in bottlenose dolphins under managed care, with both near-shore and pelagic populations, and have shown that results obtained with trained cetaceans do provide useful physiological information (Fahlman et al., 2015, 2018a,c). Thus, the data from the current study provide baseline lung function estimates from healthy dolphins under a situation with minimal stress.

One disadvantage of studying trained animals is that the training itself has the potential to cause a bias, which can be difficult to assess. For example, the large variation in V_T following a breath-hold may be the result of bias stemming from a dolphin anticipating additional requests for specific behaviours from the trainer following the breath-hold. At the beginning of the trial, the initial 90 s of the pre-breath-hold period were removed. Following this, there were no systematic changes in end-expired gas concentration or respiratory variables. If the dolphins had been stressed or uncomfortable with the procedure, we would have expected to see systematic changes with time. In addition, the f_R was within the range previously reported for bottlenose dolphins (Fahlman et al., 2015, 2018a,c), suggesting that the data reported prior to a breath-hold (Table 1) were normal.

During a surface breath-hold, gas is exchanged between the lungs and the blood; O₂ is taken up and CO₂ diffuses from the blood into the lungs. Thus, the pulmonary O₂ should be lower following the surface breath-hold and then increase as the blood and tissue O₂ levels are restored. Having the dolphins perform the breath-hold at the surface allowed us to determine how the pulmonary O₂ changes with minimal impact on ambient pressure and the shunt that develops during diving to depth (Kooyman and Sinnett, 1982; Ridgway et al., 1969). Thus, our experimental design allowed us to specifically look at changes in pulmonary gas content due to the breath-hold, and how the dolphins manage respiration during recovery.

The end-expired O₂ of the first breath following a breath-hold decreased with dive duration from an average of 9.4% for a breath-hold of up to 70 s to 6.4% when the breath-hold exceeded 180 s (Fig. 1C). The lowest end-expired O₂ was 4.2% following a 212 s breath-hold. Following the first breath, the end-expired O₂ increased throughout the recovery period, and increased with increasing V̇_E (Fig. 1A, Table 3). There was a significant increase in end-expired O₂ and a decrease in iV̇_O2, V_T, f_R and V̇_E with increasing recovery time after the breath-hold (Figs 1–3). The end-expired O₂ increased with the number of breaths following the breath-hold and increased with increasing V̇_E (Table 3, Figs 1 and 3). Thus, the recovery from a breath-hold was enhanced by increased respiratory effort. The V_T and end-expired O₂ following a dive were similar to those reported in the bottlenose dolphin [mass-specific V_T (sV_T)=36–43 ml kg⁻¹, end-expired O₂=5.4–13.2%; Ridgway et al., 1969], the harbour porpoise (Phocoena phocoena, sV_T=35–41 ml kg⁻¹, end-expired O₂≈9.5–13.5%; Reed et al., 2000) and the grey seal (Halichoerus grypus, sV_T=30–39 ml kg⁻¹, end-expired O₂≈8.6–10.1%; Reed et al., 1994). As dive duration increased, more O₂ was consumed, which depleted the O₂ stores. Following the breath-hold, the dolphins increased f_R and V_T to help replenish the O₂ stores.

There was considerable variation in the end-expired O₂ and V̇_E following a breath-hold (Figs 1 and 2), and when all post-breath-hold data were analysed, the end-expired O₂ increased with dive duration. While this may seem counterintuitive, the longer breath-holds also caused greater V_T, f_R and V̇_E, which may shorten the recovery and counteract the greater decrease in pulmonary O₂ during longer dives (Fig. 1C, Table 3). Thus, the animals compensated for the reduced O₂ levels by increasing respiratory effort to reduce recovery time. The end-expired O₂ returned to normal values within approximately 10 breaths (Fig. 1A), which is similar to the data reported by Ridgway et al. (1969). Similar responses were observed in the freely diving Weddell seal (Leptonychotes weddellii) where both iV̇_O2 (Kooyman et al., 1973) and V̇_E (Kooyman et al., 1971) changed throughout the recovery period. In the grey seal, in contrast, changes in V_T were not observed following breath-holds (Reed et al., 1994). It is possible that the changes in end-expired O₂ with breath number are affected by the changes in V_T.

During a breath with a V_T that is greater than the dead space volume (V_D), the measured O₂ content initially decreases to a plateau. This plateau reflects alveolar gas content, which, in a healthy lung, indicates arterial gas tension (West, 2012). If V_T<V_D, the measured O₂ content reflects a mixture of gas in the dead space. Consequently, as V_T changed with the number of breaths during recovery, it is possible that the temporal changes seen in end-expired O₂ are an artefact of these changes. To evaluate this, we estimated the V_D for each dolphin by assuming it was similar to the average minimum air volume (MAV=7% of TLC_est) (Fahlman et al., 2011; Kooyman and Sinnett, 1979) measured from water displacement of excised lungs from a range of marine mammals, including dolphins. Using these results, we calculated V_A=V_T−V_D, where V_A is the volume of V_T from the alveolar space. Between 0.5% and 2% (depending on which method was used to obtain TLC_est) of all breaths during recovery from a breath-hold were negative, representing breaths of lower volume than the V_D. However, while V_T changed throughout the recovery period, the proportion of breaths below the V_D remained constant with either method. Consequently, the changes in end-expired O₂ were not dependent on an increasing proportion of breaths not representing alveolar gas content, and we suggest that the end-expired O₂ can be used as a reflection of changes in arterial O₂ content. Looking at Fig. 1A, we can estimate that a 196 kg dolphin has recovered its end-expired O₂ content after 10 breaths following a 157 s breath-hold, i.e. when the red trend line crosses the average end-expired O₂ content before the breath-hold. The duration of this recovery will depend on the f_R. From Fig. 2B, we can see that f_R is about 8 breaths min⁻¹ during the first 10 breaths, so this dolphin would have recovered its arterial O₂ content in about 1.2 min.

Reed et al. (1994) showed end-expired O₂ and CO₂ recovery curves that changed with time in an opposite pattern in seals. For O₂, the levels were initially high, then decreased and finally increased again at the end. It was suggested that this could indicate reduced gas exchange at the lung level and/or reduced tissue perfusion (Reed et al., 1994). Either situation would reduce pulmonary gas uptake, and the partial pressures in the lung would remain high for O₂ and low for CO₂. As blood flow and gas exchange resume during the post-breath-hold surface interval, the venous oxygen (P_O2) and CO₂ tension (P_CO2) would be lower and higher, respectively, causing a drop in pulmonary O₂ and an increase in CO₂. As O₂ is restored and CO₂ is removed, the partial pressures again approach normal values. Interestingly, this U-shaped pattern was not reported in the harbour porpoise (Reed et al., 2000), in the bottlenose dolphins in the current study or in the work by Ridgway et al. (1969). A new hypothesis suggesting how cetaceans control pulmonary gas exchange through management of ventilation–perfusion matching was recently published (García-Párraga et al., 2018). Ventilation–perfusion matching would enable active pulmonary shunting, which would allow cetaceans to selectively access pulmonary gases, facilitating exchange of O₂ and CO₂ with minimal exchange of N₂. Thus, the variation in results may reflect differences in the mechanisms by which marine mammals manage gases during diving.

Past measurements have shown that cetaceans are capable of exchanging as much as 90% of their TLC_est (Fahlman et al., 2015, 2017a, 2019; Kooyman and Cornell, 1981; Olsen et al., 1969; Ridgway et al., 1969). From these estimates, numerous studies have assumed that the V_T is close to TLC_est, and during exercise or during recovery from a dive it may be logical to assume that V_T is close to TLC_est (Dolphin, 1987a; Fahlman et al., 2016; Folkow and Blix, 1992). However, our results suggest that the average V_T during recovery from a breath-hold was only between 30% and 54% of TLC_est, depending on which value is used to estimate TLC_est. The latter value was based on measured V_T (5–6 l) following diving in a 138 kg bottlenose dolphin. By assuming that the measured V_T represented a diving lung volume of 80% of total lung capacity, TLC_est was 6.9 l (Ridgway et al., 1969). Using this TLC_est, the V_T during recovery ranged from 7% to 184%. The other equation for TLC_est was based on work on excised lungs from a range of marine mammal species (TLC_est=0.135M_b^0.92) (Fahlman et al., 2011; Kooyman, 1973), resulting in a range of V_T from 4% to 106% of TLC_est. With either method, it appears that during recovery from a surface breath-hold, V_T is mostly considerably lower than TLC_est. These results are similar to those of our previous study measuring V_T during recovery from a bout of exercise in bottlenose dolphins (Fahlman et al., 2016). Our results indicate that the V_T,exp during recovery increased with the length of the surface breath-hold (Table 3, Fig. 2A), but most breaths were still considerably lower than TLC_est. Based on these results, it appears that bottlenose dolphins have considerable scope to further increase respiratory effort following a breath-hold. While it is difficult to extrapolate this to freely swimming and diving animals, bottlenose dolphins may be capable of much longer (Fahlman et al., 2018b) or more active breath-holds with similar recovery rates. However, this warrants further study if respiratory variables are to be used to assess field metabolic rate.

Past studies have attempted to estimate field metabolic rate from f_R in a number of species by assuming static respiratory values (Armstrong and Siegfried, 1991; Blix and Folkow, 1995; Christiansen et al., 2014; Dolphin, 1987a; Dolphin, 1987b; Folkow and Blix, 1992; Williams and Noren, 2009), or by measuring the variation in V_T and O₂ exchange fraction (ΔO₂=inspired O₂−expired O₂) following exercise (Fahlman et al., 2016). While this method is attractive in its simplicity, considerable variation in V_T and ΔO₂ yields extensive error (Fahlman et al., 2016, 2017b; Folkow and Blix, 2017). The results from the current study show a large variation in V_T, f_R and end-expired gas content following breath-holds ranging from 34 to 292 s. Maximal and average dive durations in shallow-diving near-shore dolphins are within this range (Fahlman et al., 2018b; Wells et al., 2013), and this ecotype may therefore show similar variation following diving. In addition, the change in respiratory effort following longer breath-holds to reduce the duration of recovery resulted in a complicated relationship between V_T, end-expired gas content and recovery duration. Our results are from static breath-holds and it is possible that the variability in f_R and V_T in free-ranging dolphins diving to depth and breathing while swimming is reduced. New methods are being developed that may help estimation of f_R and V_T in free-ranging marine mammals (Cauture et al., 2019; Sumich, 2001; van der Hoop et al., 2014). These may contribute to our understanding of respiratory effort and expired gas content, allowing us to continue to develop more accurate and useful estimates of field metabolic rate from methods based solely on f_R.

Both V̇_insp and V̇_exp increased immediately following a breath-hold, and decreased throughout the recovery period, but were not affected by the duration of the breath-hold (Fig. 3, Table 3). V̇ of the first breath following the breath-hold was significantly higher than the average V̇ before the breath-hold (Fig. 3). The expiratory breath duration increased positively with breath-hold duration and decreased during recovery. Thus, the increased V_T following a surface breath-hold was achieved by increased breath duration and V̇_exp, but only by increasing V̇_insp. During voluntary breaths, exhalation is passive and mainly driven by the elastic recoil of the chest, whereas the inspiratory flow is active for both voluntary and forced breaths (see fig. 5A in Fahlman et al., 2017a). As the respiratory effort increases, a greater proportion of the expiratory phase becomes active where the respiratory muscles assist in generating high V̇ (see fig. 5B,C in Fahlman et al., 2017a). The relative importance of the diaphragm, cranio-cervical, thoracic and lumbo-pelvic muscles to drive inspiration during rest in dolphins is not known (Cotten et al., 2008; Dearolf, 2003). Understanding the mechanism of respiration both at rest and while actively swimming could provide important information for conservation efforts, and this could be an interesting topic for future studies.

There are conflicting reports in the literature about whether marine mammals dive on inhalation or exhalation, but some studies clearly show that cetaceans dive on full or partial inhalation (Ridgway, 1986; Ridgway and Howard, 1979; Scholander, 1940). Our results concur with this, and most breath-holds began following a normal breath, which in many cases had a higher than normal V_T,insp. Following a breath-hold, the initial V_T was highly variable and occasionally the dolphins began with a small exhalation and inhalation. In these cases, a second breath of larger volume followed shortly after, but these were exceptions to the pattern. During the study trials, when a trainer signalled the beginning of a breath-hold, most dolphins took a deep breath before initiating the breath-hold. While the deep breath before the breath-hold may be a learned behaviour to extend the breath-hold, this implies a certain level of behavioural control and it is likely that free-ranging animals perform a similar inspiration before the breath-hold. Assessment of respiratory effort in free-ranging animals could be made possible by phonospirometry (Sumich, 2001), or by assessing respiratory sinus arrhythmia and relating these changes to V_T,insp (Cauture et al., 2019; Fahlman et al., 2017c).

In the current study, we measured respiratory flow and end-expired O₂ content in bottlenose dolphins before and after surface breath-holds. Our data show a complex relationship between respiratory effort (V_T, f_R, V̇_E) and end-expired gas content, suggesting that the dolphins maximized effort following long dives in order to minimize recovery duration. While V_T increased following a breath-hold, most breaths were considerably lower than TLC_est. Thus, dolphins may have considerable scope to increase respiratory effort following longer or more strenuous breath-holds to rapidly replenish O₂ stores.

A special thanks to all the Marine Mammal Specialists and staff at Dolphin Quest – Oahu, and Siegfried and Roy's Secret Garden – Las Vegas who made this study possible. Thanks also to Angela Casas Quirante for help with the data analysis. We thank the anonymous referees, who provided constructive criticism that helped improve the manuscript. We thank Dolphin Quest and Mirage for providing support for animals and crew, and for access to resources.