Anterior vena caval oxygen profiles in a deep-diving California sea lion: arteriovenous shunts, a central venous oxygen store and oxygenation during lung collapse

Blood oxygen profiles in deep-diving California sea lions [Zalophus californianus (Lesson 1828)] have revealed that arterial hemoglobin (Hb) saturation (S_O2) was maintained above 90% during the deepest portions of dives beyond 200 m in depth (McDonald and Ponganis, 2012). These high S_O2 values occurred despite (1) the presumed lack of gas exchange at the depths beyond when ‘lung collapse’ (compression of alveolar air into the tracheo-bronchial tree) occurs, and (2) low posterior vena caval S_O2 values that were sometimes near 0% at the bottom of these deeper dives (McDonald and Ponganis, 2012, 2013). The maintenance of high arterial S_O2 in combination with the large arteriovenous (a–v) oxygen content difference were remarkable because, in the absence of pulmonary gas exchange at depth, arterial S_O2 should decrease and reflect mixed venous (pulmonary artery) S_O2. These findings suggested that either (1) some gas exchange still occurred beyond the estimated depth of lung collapse, or (2) posterior vena caval S_O2 was not representative of the entire venous oxygen store, and a highly oxygenated central venous reservoir might contribute to maintenance of high arterial S_O2 values during lung collapse. We suspected the latter possibility was more likely for two reasons. First, pulmonary shunting was found to increase with depth in sea lions and complete lung collapse and lack of gas exchange has been predicted to occur by 160 m depth (Kooyman and Sinnett, 1982). Second, arterial partial pressure of oxygen (P_O2) profiles during deep dives were consistent with significantly decreased gas exchange at such depths (McDonald and Ponganis, 2012).

Previously, we had hypothesized that arterialization of venous blood (venous S_O2 approaching arterial S_O2) prior to deep dives and/or during early descent created a well-oxygenated central venous oxygen store (McDonald and Ponganis, 2013; Tift et al., 2017). Such arterialization of the venous blood may occur through a–v shunting, most likely through the fore- and hind-flippers as has been suggested for phocid seals (Blix et al., 2010). To avoid confusion about shunting, readers should remember that a–v shunts are distinct from pulmonary shunts. Arteriovenous shunts occur in the peripheral vasculature and, typically, increase venous oxygen content. Pulmonary shunts occur within the lungs and decrease arterial oxygen content.

The extreme bradycardia that occurs during the bottom phases of deep dives, when the lungs are collapsed, should slowly mete out this well-oxygenated blood. The low heart rate would also preserve oxygenation of the central venous blood owing to the slow return of desaturated venous blood from the periphery during the bradycardia (McDonald and Ponganis, 2014). Theoretically, arterialization of venous blood could be achieved through (1) a–v shunting prior to or early in dives, and/or (2) some splenic contraction and expulsion of oxygenated blood into the central vena cava due to further activation of the sympathetic nervous system to maintain blood pressure during the extreme bradycardia of the bottom phases of deep dives (McDonald and Ponganis, 2013; Ponganis, 2015; Ponganis et al., 1992). Lastly, arterialization of venous blood early or prior to the dive would require minimization of muscle blood flow and blood oxygen extraction by muscle; otherwise, venous saturation would decrease.

Our hypothesis of a well-oxygenated central venous oxygen store in deep-diving sea lions is analogous to the hepatic sinus oxygen reservoir proposed by Elsner et al. (1964) in northern elephant seals (Mirounga angustirostris). Otariids (sea lions and fur seals) do not have the capacious hepatic sinus/vena cava and elaborate posterior vena caval sphincter of phocid seals (Harrison and Tomlinson, 1956; King, 1977; King, 1983). However, the longest durations of lung collapse during the deep dives of sea lions are only 3 to 3.5 min (McDonald and Ponganis, 2012); therefore, even a relatively small oxygenated venous blood oxygen reservoir in the hepatic sinus may be sufficient to maintain high arterial S_O2 during the lung collapse phase of their dives. Furthermore, the rate of depletion of this proposed central venous oxygen store could be slowed by a high anterior vena caval S_O2 during lung collapse.

Ideally, to address whether an arterialized central venous oxygen store exists in sea lions and test our hypothesis that posterior vena caval oxygenation does not reflect the total venous oxygen store during a deep dive, we would have to measure pulmonary artery (mixed venous) P_O2 or S_O2 in diving sea lions. Although such catheterizations are feasible (Ponganis et al., 1991), the P_O2 electrodes we currently use are not long enough to reach the pulmonary artery, and other bio-logging devices and sensors are not available for such measurements in a free-diving sea lion. Therefore, as a surrogate measurement, we chose to investigate anterior vena caval P_O2 and S_O2 during dives with use of our available P_O2 electrodes/recorders.

The anatomical arrangement of veins draining into the anterior vena cava (King, 1977) (Fig. 1), and the relative blood flows and S_O2 values in those veins, should determine the anterior vena caval S_O2 profile. The S_O2 in the anterior vena cava would be dependent on possible a–v shunting, the concomitant arterial S_O2, the Hb concentration of the individual animal and the influx of venous blood from peripheral tissues (dependent on the level of bradycardia and peripheral perfusion). Notably, unlike venous inflow from other tissues, venous drainage from the brain via the external jugular vein into the anterior vena cava should continue during the bradycardia of diving because cerebral blood flow is maintained during the dive response (Blix et al., 1983; Zapol et al., 1979).

We expected that anterior vena caval S_O2 profiles during the dive would differ from posterior vena caval profiles, supporting our hypothesis that the posterior vena caval S_O2 profile was not indicative of the entire venous oxygen store. We also hypothesized that the anterior vena caval S_O2 profiles would (1) provide evidence for arterialization of venous blood prior to deep dives, and (2) be high enough to allow maintenance of arterial S_O2 when it returned to the heart and mixed with well-oxygenated blood from the hepatic sinus.

This study was conducted on San Nicolas Island, California, in August 2013. A single adult female California sea lion was captured with a hoop net and anesthetized using a portable vaporizer-breathing circuit set up with an initial mixture of 5% isoflurane and 100% oxygen. Once anesthetized, a custom-built P_O2 datalogger (UUB-2PT; UFI, Morro Bay, CA, USA) with custom housing (Meer Instruments, Palomar Mountain, CA, USA), time-depth recorder with three-axis accelerometer (TDR10x; Wildlife Computers, Redmond, WA, USA) and VHF radio transmitter (mm160B; Advanced Telemetry Systems, Isanti, MN, USA) was affixed to the dorsal, midline pelage of the animal using fast-setting epoxy (Loctite, Henkel Corp., Westlake, OH, USA). Using an ultrasound machine (SonoSite Inc., Bothell, WA, USA), the external jugular vein was identified and percutaneously catheterized with a peel-away catheter (5 Fr, Cook Medical, Bloomington, IN, USA). A 30 cm P_O2 electrode (Licox C1.1, Integra Life Sciences, Plainsboro, NJ, USA) was then inserted into the peel-away catheter and threaded down into the anterior vena cava. The P_O2 electrode was connected to the P_O2 datalogger via a waterproof cable and connector (Impulse Enterprise, San Diego, CA, USA). Depth and P_O2 were recorded at 1 Hz, and three axes of acceleration were recorded at 16 Hz. Calibration and specifics on the P_O2 electrode have been explained in more detail in a previous publication (Stockard et al., 2005). In particular, it should be remembered that ambient pressure does not have a direct effect on P_O2 in the blood. Of course, ambient pressure does have an indirect effect on blood P_O2 through pulmonary gas exchange and the well-known effects of ambient pressure on the partial pressures of gases in the air of the lungs. Catheter placement has also been explained in more detail in previous publications (McDonald and Ponganis, 2012, 2013; Ponganis et al., 1997, 1991; Tift et al., 2017).

After instrumentation, the animal was placed in a large canine kennel to be weighed (±0.2 kg, MSI-7200 Dyna-link; Measurement Systems International, Seattle, WA, USA) and recover from anesthesia (∼60 min). Once fully alert, the animal was released back onto the same beach where she was originally captured. After 7 days and two trips to sea, the animal was recaptured with hoop nets, instruments were removed under manual restraint and then the animal was released (10 min procedure). All procedures were approved by the University of California, San Diego, Animal Subjects Committee (no. S11303) and National Marine Fisheries Services (no. 14676).

Data were analyzed as described in our previous publication (Tift et al., 2017). Briefly, dives were identified and summary data from each dive were produced using a custom-written MATLAB program (IKNOS; Y. Tremblay). This dive analysis program implements a zero-offset correction at the surface and identifies dives using a specified minimum depth (10 m) and duration (20 s). The following dive phases were identified in the data: descent (surface to 80% maximum depth), bottom (depths within 80% maximum depth) and ascent (80% maximum depth to surface). Most dives from California sea lions are shallower than 100 m (Feldkamp et al., 1989), yet many adult female California sea lions from San Nicolas Island are known to be deep divers (Kuhn and Costa, 2014; McHuron et al., 2016) and exhibit a bimodal pattern in their maximum depth of dives, with very few dives occurring between 100 and 150 m (Fig. S1). Therefore, to compare shallow and deep dives, dives to maximum depths >100 m were classified as deep, and dives to maximum depths ≤100 m were classified as shallow.

Fore-flipper stroke rate was calculated using a custom-written algorithm in MATLAB. The low-frequency static acceleration was filtered out using a 0.2 Hz high-pass Butterworth filter. The resulting dynamic acceleration was analyzed using power spectral density analysis to identify the dominant frequency of a flipper stroke for each axis (∼0.8–1.2 strokes s⁻¹). Flipper strokes were identified using a MATLAB peak detection algorithm similar to those in other studies (Jeanniard-du-Dot et al., 2016; Sato et al., 2011). A single flipper stroke was identified when there was a prominent acceleration peak (≥0.45 m s⁻²) in either the forward surge direction or the heave surge direction. Stroke rate was calculated both as an average stroke rate along a moving window of 10 s throughout the diving record, and as an instantaneous stroke rate (strokes s⁻¹) based on the time between two individual flipper strokes.

The P_O2 electrode was calibrated in the laboratory before deployment (McDonald and Ponganis, 2012, 2013). The percent hemoglobin saturation was calculated using an equation from a previously measured hemoglobin-dissociation curve in this species (McDonald and Ponganis, 2013). The minimum calculated aerobic dive limit for this species is 3.4 min (Weise and Costa, 2007), and increased P_CO2 accumulated during a breath hold is known to contribute to a decrease in pH as dive duration increases (Kooyman et al., 1980; Qvist et al., 1986). Therefore, P_O2 values collected prior to 3 min into the dive were converted to S_O2 using the equation for a pH of 7.4 {log[S_O2/(100– S_O2)]=2.473×log(P_O2)–3.632} (r²=0.97); all P_O2 values collected beyond 3 min into dives were converted to S_O2 using the equation for a pH of 7.3 {log[S_O2/(100–S_O2)]=2.363×log(P_O2)–3.576} (r²=0.96). The core body temperature of California sea lions is thought to remain relatively constant at 37°C while diving (McDonald and Ponganis, 2013). Therefore, the P_O2 electrode was only calibrated at this temperature, and effects of temperature variation on electrode output and S_O2 values were not considered. The time on the TDR/accelerometer instrument and P_O2 logger were synchronized to the same internet-synced computer clock. Mean change in S_O2 (ΔS_O2) was calculated for the same 10 s moving window as flipper stroke rates to facilitate comparison between the two parameters. ΔS_O2 (initial S_O2–final S_O2) was also calculated for the duration of the three dive phases (descent, bottom and ascent) and for the total dive duration of every dive.

Depth of lung collapse was estimated for all dives >200 m using the linear regression provided by McDonald and Ponganis (2012): depth of lung collapse (m)=0.62×maximum depth (m)+35.66. Values and patterns of anterior vena caval S_O2 were compared with the estimated depths of lung collapse by extracting S_O2 values during depths beyond the estimated depth of lung collapse. Mean values are reported ±s.e.m., unless otherwise noted.

In this equation, Q̇_AVC–Q̇_jugular equals Q̇_{a–v shunt}, i.e. a–v shunt flow. In this model, it was assumed that other venous inflow into the anterior vena cava was minimal.

The same type of formula was also used for calculation of the flows needed in the anterior vena cava and posterior vena cava (from the hepatic sinus) during lung collapse in order to achieve the observed maintenance of high arterial saturations. Calculations were based on the known cardiac output, the observed saturation in the anterior vena cava and the assumed saturation (95%) of blood coming from the hepatic sinus into the proximal posterior vena cava (and then into the right atrium to mix with the blood from the anterior vena cava).

A total of 198 dives with simultaneous and continuous venous P_O2, depth and acceleration data were collected from one adult, female, 77.8 kg California sea lion. The mean depth of dives was 149±116 m, with a maximum depth of 399 m (Fig. S1). The mean dive duration was 194±124 s, with a maximum dive duration of 443 s. During the 7 days that the deployment lasted, she performed two trips to sea. The first trip was approximately 5 h, and the second trip was just over 24 h of continuous diving.

At the end of shallow dives, the mean P_O2 and S_O2 were typically higher than values at the start of the dive (Table 1). This pattern was reflected by positive ΔS_O2 values for the entire dive (i.e. overall increase in blood oxygen throughout the dive, see Figs 2, 3 and 4B). Most of the increase in S_O2 occurred early in the dive, during the descent phase of shallow dives (Fig. 5; Fig. S2). Despite these consistent patterns, the overall range in anterior vena caval S_O2 was highly variable among shallow dives (Figs 3 and 4B). There was no clear relationship between flipper stroke rate and ΔS_O2 across dive phases, or for the entire dive (Figs 4B, 5 and 6). There were often brief periods of gliding during the last portion of the ascent, when the animal was approaching the surface (Fig. S2). Rest periods at the surface and shallow swim periods revealed that surface resting S_O2 was near 75% (Fig. S3A).

At the start of deep dives, anterior vena caval P_O2 and S_O2 were typically much higher than those at the start of shallow dives; similarly, anterior vena caval P_O2 and S_O2 at the end of deep dives were lower than those at the end of shallow dives (Table 1, Fig. 4A). On deeper dives, S_O2 did not decline initially; instead, high S_O2 was maintained until the animal reached the bottom portion of the dive, when values began to decrease and then continued to drop until the animal surfaced (Fig. 4A; Figs S2 and S4). Hence, the largest reductions in S_O2 of the anterior vena cava occurred during ascent. There were consistent negative relationships between ΔS_O2 for the entire dive duration with both maximum dive depth and duration (i.e. longer and deeper dives had more oxygen extracted from blood) (Fig. 3).

In 74 deep dives, ranging from 205 to 399 m maximum depth, mean anterior vena caval S_O2 at the estimated depth of lung collapse was 82.9±3.0% (Fig. S4). At the estimated depth of lung re-expansion and resumption of gas exchange, mean anterior vena caval S_O2 was 63.8±8.4%. The maximum and minimum S_O2 at the start of the lung collapse period were 88.4 and 76.4%; the corresponding values at the end of the lung collapse period were 78.9 and 40.1%, respectively.

Unlike shallow dives, the rate of decline in blood oxygen during deep dives (>100 m) mirrored flipper stroke rate throughout deep dives (Fig. 6). This relationship was due primarily to the rapid drop in S_O2 that occurred during the bottom and ascent phases of dives, when animals were frequently stroking (Figs 4A, 5 and 6; Fig. S2).

The diving behavior of the sea lion in this study, including dive depths and flipper stroke rate profiles (Fig. 4; Fig. S2), was similar to that seen previously in other free-diving adult female California sea lions (McDonald and Ponganis, 2012, 2013, 2014; McHuron et al., 2016; Tift et al., 2017). For example, the animal from this study consistently exhibited prolonged glides during the descent of deep dives (>100 m maximum depth), as well as short duration glides during the last 25–30 m of ascent to the surface (Fig. 4A, Fig. S2). Therefore, despite the limitation in sample size, we believe the anterior vena caval blood oxygen profiles from this study are representative of the range of dives performed by adult female California sea lions during foraging trips to sea.

Several logistical and technical factors contributed to the limited sample size in this study: (1) frequent migration of the P_O2 electrode out of the blood vessel during dive activity, (2) internal damage to the wires and solder joints within the underwater connector cable secondary to presumed torque on the cable, and (3) failure of either the P_O2 logger or TDR/accelerometer while the animal was out at sea. Out of a total of 12 deployments of anterior vena caval P_O2 electrodes from 2013 to 2016, the animal in this study was the only animal where all instruments and the probe functioned properly. We think that the technical problems of P_O2 electrode migration and cable dysfunction were due to the high degree of flexibility and maneuverability of the sea lion neck and head while swimming (Fish et al., 2003).

Similar to prior findings in the posterior vena cava of sea lions (McDonald and Ponganis, 2013; Tift et al., 2017), anterior vena caval P_O2 and S_O2 profiles were highly variable during short duration, shallow dives (<100 m; Fig. 1). Start-of-dive S_O2 values during shallow dives ranged from 40 to 80% and indicated that the animal did not always fully replete the venous oxygen store prior to the next shallow dive, whereas start-of-dive S_O2 values on deep dives were consistently higher (Table 1, Fig. 2). Such incomplete restoration of the oxygen store prior to shallow dives has been suggested based on metabolic rate studies of Steller sea lions (Eumetopias jubatus) (Fahlman et al., 2008). Notably, however, anterior vena caval P_O2 and S_O2 frequently increased during shallow dives (Figs 2, 3 and 4B, Table 1). This increase in venous oxygenation during shallow dives has also been observed in the posterior vena cava, although less frequently (Tift et al., 2017). A net increase in anterior vena caval S_O2 suggests minimal muscle blood flow and possible a–v shunting during many of these shallow dives. In other words, the blood oxygen content contributed by the a–v shunt must be greater than the tissue oxygen extraction from blood in other vessels returning into the anterior vena cava. The large range in anterior vena caval S_O2 during shallow dives (Fig. 4B) and even during surface intervals suggests that blood oxygen delivery to muscle and potential use of a–v shunts for thermoregulation are highly variable.

For deep, long-duration dives, initial P_O2 and S_O2 increased with dive duration while end-of-dive values consistently decreased (Figs 2 and 3). Surface interval P_O2 and S_O2 values between deep dives were typically greater than those for shallow dives, and for extended rest periods at the surface that occasionally occurred (see Fig. S3B). Based on the higher surface interval heart rates before deeper dives (McDonald and Ponganis, 2014), we think that elevated cardiac outputs and a–v shunting prior to the dive contribute to these elevations in anterior vena caval S_O2. Presumably, myoglobin is already re-saturated, and muscle blood flow/blood oxygen extraction by muscle are not exceptionally high before the dive. These patterns suggest an element of planning prior to deep dives in this species.

The S_O2 profiles in the anterior vena cava during deep dives differed markedly from those previously recorded in the posterior vena cava (McDonald and Ponganis, 2013; Tift et al., 2017); S_O2 remained elevated in the anterior vena cava during most of descent (Figs 4 and 5; Fig. S2) whereas it typically declined dramatically in the posterior vena cava. During the latter portions of deep dives, anterior vena caval S_O2 always decreased in contrast to those in the posterior vena cava, which often increased during ascent. These findings support our hypothesis that posterior vena caval oxygen profiles are not representative of the entire venous system.

Anterior vena caval S_O2 may be affected by flipper stroke effort because blood flow from the primary locomotory muscles of sea lions returns to the heart via the anterior vena cava (Fig. 7). However, the relationship appears to be weak between anterior vena caval blood oxygen profiles and flipper stroke rate patterns during shallow dives (Figs 4B and 5). The net increases in S_O2 during many shallow dives suggest that there was (1) minimal muscle blood flow (insignificant blood oxygen extraction by muscle), (2) a–v shunting or (3) a high ratio of a–v shunting to locomotory muscle blood flow. Such variability in peripheral blood flow patterns has also been suggested in diving emperor penguins (Aptenodytes forsteri), where increases in venous P_O2 and S_O2 were also often seen in their short duration dives (Ponganis et al., 2009; Williams et al., 2011).

During the initial active flipper stroking of early descent in deep dives, anterior vena caval S_O2 remained high (Figs 4–6; Figs S2 and S4). Maintenance of these high venous S_O2 values suggests that muscle blood flow and blood oxygen extraction by active muscle are minimal during the initial phase of the dive. In contrast, when the prolonged glide to the bottom of the dive ended and foreflipper stroking began, anterior vena caval S_O2 began to decline (Figs 4–6; Figs S2 and S4). These declines in anterior vena caval S_O2 were consistent with increased inflow of more desaturated blood from peripheral tissues (including muscle) that may be associated with the increased heart rate observed during the latter phases of the dive, especially during ascent (McDonald and Ponganis, 2014). As discussed above, such variations in peripheral blood flow patterns have also been suggested for emperor penguins. An increase in muscle blood flow and blood oxygen extraction by muscle during the latter phases of deep dives would support previous suggestions that changes in heart rate may be coupled to exercise in diving mammals (Davis and Williams, 2012; Hindle et al., 2010; Signore and Jones, 1996).

As discussed in a prior paper (McDonald and Ponganis, 2013), a normal a–v oxygen content difference of 5 ml O₂ dl⁻¹ in a sea lion with a Hb content of 18 g dl⁻¹, and an arterial S_O2 of 95 to 100% should result in a venous S_O2 of 74 to 79%. This coincides with the S_O2 range observed in the anterior vena cava of the sea lion in this study during rest activity at the surface (Fig. S3). However, just as was found in the posterior vena cava (McDonald and Ponganis, 2013), S_O2 in the anterior vena cava was usually elevated above this level, sometimes to >90%, prior to deep dives (>100 m maximum depth) (Fig. 4; Fig. S3B). Exceptionally high P_O2 and S_O2 values were also frequently observed during shallow dives and during the early segments of deep dives (Figs 2 and 4; Figs S3B and S5). These high venous S_O2 values suggest that sea lions utilize some form of a–v shunting prior to and even during dives, most likely through the fore- and hind-flippers as has been suggested for phocid seals (Blix et al., 2010), and as has been reviewed in prior papers (McDonald and Ponganis, 2013; Tift et al., 2017). These high S_O2 values in both the anterior and posterior vena cava prior to deep dives support our hypothesis of a highly oxygenated central venous oxygen store in the sea lion.

The contribution of an a–v shunt to the elevation in anterior vena caval S_O2 is based on venous anatomy (Figs 1 and 7A). The anterior vena cava (the site where venous P_O2 was measured) is formed by the union of the two brachiocephalic veins, which each receive primary input from the external jugular (jugular), omocervical and axillary veins (King, 1977). Cerebral venous drainage into the jugular vein should result in an a–v oxygen content difference of 5–6 ml O₂ dl⁻¹ of blood (Kerem and Elsner, 1973; Macmillan and Andrews, 2000; Zauner and Muizelaar, 1997), corresponding to a jugular S_O2 of 74-79% in the sea lion (McDonald and Ponganis, 2013). Given that the sea lion brain is approximately 300 g, and cerebral blood flow is approximately 55 ml 100 g⁻¹ brain tissue min⁻¹, jugular venous flow from the brain would be approximately 165 ml min⁻¹ (Montie et al., 2009, 2010; Zauner and Muizelaar, 1997). With this jugular inflow into the anterior vena cava, the a–v shunt flow required to raise anterior vena caval S_O2 to a given level can be calculated with Eqn 3 (West, 1972) (see Materials and methods), where fractional S_O2AVC is 0.92, corresponding to the observed 92% S_O2 in the anterior vena cava; Q̇_jugular is the estimated 165 ml min⁻¹ inflow of jugular blood, and fractional S_O2jugular is 0.74, corresponding to 74% S_O2 estimated in the jugular vein; Q̇_AVC–Q̇_jugular is a–v shunt flow, and fractional S_{O2a–v shunt} is 0.95, reflecting the observed arterial S_O2 value of 95%. Use of these values in the equation results in a Q̇_AVC of 1155 ml min⁻¹. The calculated Q̇_AVC–Q̇_jugular is 990 ml min⁻¹ under these conditions.

The tachycardia of 120 beats min⁻¹ prior to deep dives results in a cardiac output of 14.4 l min⁻¹ based on a minimum stroke volume of 1.5 ml kg⁻¹ in an 80 kg sea lion (Ponganis et al., 1991). Thus, the 990 ml min⁻¹ a–v shunt flow would be only 6.9% of cardiac output. A slightly larger a–v shunt flow and percentage of cardiac output could compensate for additional desaturated blood from any venous tributaries, such as the omocervical vein. Even during late descent, with a heart rate of 15 beats min⁻¹, cardiac output would be 1.6 l min⁻¹, which should be more than sufficient to account for brain, a–v shunt and coronary blood flow. Furthermore, the bradycardia and peripheral vasoconstriction during this phase of the dive would also minimize inflow of any desaturated blood from the omocervical vein and other vessels into the brachiocephalic veins.

The magnitude of the hypothesized a–v shunt may decrease during late descent because elevated carbon dioxide increases cerebral blood flow in sea lions (Dormer et al., 1977) and decreases the percentage of cardiac output available for an a–v shunt. However, in regard to S_O2, the hypercarbia-induced increase in cerebral blood flow will result in more oxygen being delivered to the brain and an increased S_O2 in the jugular vein, because brain metabolism will remain constant (Macmillan and Andrews, 2000; Schell and Cole, 2000). This would mitigate the decrease in a–v shunting and again contribute to maintenance of elevated anterior vena cava oxygen content. In addition, hypercarbia can also depress brain metabolic rate (Xu et al., 2011), which should also elevate or at least conserve jugular S_O2. These processes could help account for the maintenance of relatively high anterior vena caval S_O2 during this phase of the dive.

Comparison of S_O2, heart rate and stroke rate profiles collected during deep dives of different animals provide a general physiological overview of deep dives of sea lions (Figs 7B, 8 and 9). The near equivalence of aortic S_O2 with both anterior and posterior vena caval S_O2 values early in deep dives (Fig. 8) supports the hypothesis of arterialization of anterior and posterior vena caval blood via a–v shunts prior to a deep dive. During the first minute of deep dives (Fig. 8; Figs S5 and S6), maintenance of high venous S_O2 values and low a–v O₂ content differences (<2 ml O₂ dl⁻¹) suggest continued a–v shunting and insignificant muscle blood flow at a time when stroke rate is highest during the dive. Otherwise, blood oxygen extraction by muscle should decrease venous S_O2, especially in the anterior vena cava (due to flow of desaturated blood from fore-flipper locomotory muscles into the axillary vein and anterior vena cava).

As a deep dive progresses into the bottom phase of the dive, S_O2 profiles diverge with maintenance of S_O2 in the aorta and anterior vena cava and a progressive decline in S_O2 in the posterior vena cava (Fig. 8). By 3 min into a deep dive, the posterior vena caval a–v O₂ content difference is up to 20 ml O₂ dl⁻¹, while the anterior vena caval O₂ content difference is still only 2 ml O₂ dl⁻¹ (Fig. S5). As already suggested (McDonald and Ponganis, 2013; Tift et al., 2017), we suspect that the severe bradycardia during the bottom phase of the dive results in extremely low influx of highly desaturated blood from the distal posterior vena cava into the central blood oxygen store. These low S_O2 values in the posterior vena cava are probably due to prolonged tissue transit time and almost complete blood oxygen extraction from the small volume of blood perfusing tissues in the posterior regions of the body. Such intense vasoconstriction in the posterior body during the severe bradycardia is, in part, secondary to the dense sympathetic nerve fiber innervation of the proximal segments of arteries supplying abdominal organs and the posterior body in pinnipeds (White et al., 1973). In contrast, in the anterior regions of the body, such innervation is sparse, the brain is still well perfused (Zapol et al., 1979) and some element of an a–v shunt may still exist. As discussed in previous paragraphs, these blood flow patterns could account for the maintenance of higher S_O2 values in the anterior vena cava at that time.

When heart rate increases as the dive progresses into the ascent phase, we suspect that the observed decreases in anterior vena caval S_O2 are due to influx of relatively desaturated blood from other tissues in the anterior region of the body (potentially even including muscle tissue). At the same time, increases in peripheral perfusion of the posterior regions of the body would actually increase S_O2 in the posterior vena cava because of the increase in peripheral perfusion and decrease in tissue transit time of more oxygenated blood from the anterior region of the animal that has mixed in with blood from the posterior portion of the body. This description provides a general overview and we suspect that there will be considerable variability in saturation profiles depending on circumstances and physiological responses of individual dives.

Although adult sea lions have been reported to have large hepatic sinuses and even a modified posterior vena caval sling, these features are not as well developed as in phocid seals (King, 1977). We hypothesized that a–v shunting prior to a dive oxygenates venous blood, and that the hepatic sinus could serve as an important venous oxygen store in sea lions (Fig. 9). It was not likely to be as large as in phocids (Elsner et al., 1964), but it might be sufficient for the shorter dive durations and periods of lung collapse in sea lions. The volume of the hepatic sinus of the adult female California sea lion is typically 1 liter, maximally 2 liters (J. St. Leger, personal communication). Given that blood volume in an average 80 kg adult female California sea lion is 120 ml kg⁻¹ (Weise and Costa, 2007), a 1–2 liter hepatic sinus would only store approximately 10 to 21% of total blood volume. Therefore, we set out to determine, with the use of shunt calculations, whether the reported size of the hepatic sinus and proposed central venous blood oxygen store was large enough to account for maintenance of arterial S_O2 during the period of lung collapse (see Materials and methods for review of shunt calculations).

Despite the hypothesized central venous oxygen store in sea lions, it is still difficult to account for maintenance of arterial S_O2 values above 90% during the period of estimated lung collapse when gas exchange is presumed to not occur (Fig. 7B; Fig. S6). First, although we only have a limited amount of arterial and anterior vena caval data, the observed S_O2 values in the anterior and posterior venae cavae at the start of the dive never reach levels high enough to create a central oxygen reservoir with greater than 95% saturation (Figs 7B and 8). Conceivably, splenic contraction during deep dives could elevate venous S_O2 and Hb content, but sea lion spleens are not especially large (Ponganis et al., 1992). Second, during periods of lung collapse in deep dives, the anterior vena caval S_O2 values do not remain high the entire time (Fig. S4). The S_O2 values can range from as high as 88% at the start of complete lung collapse to as low as 40% at the start of lung re-expansion.

results in an anterior vena caval flow of 208 ml min⁻¹, and hepatic sinus (posterior vena caval) flow of 632 ml min⁻¹ into the heart (note: fractional S_O2a is 0.92, corresponding to an aortic S_O2 of 92%, Q̇_AVC is anterior vena caval flow into the right atrium, fractional S_O2AVC is 0.83, corresponding to 83% S_O2 in the anterior vena cava; cardiac output–Q̇_AVC is the posterior vena caval flow from the hepatic sinus into the right atrium, and fractional S_{O2hepaticsinus} is 95%, the assumed S_O2 in the hepatic sinus).

Under these conditions, for maintenance of aortic S_O2 at 92% during 3 min of lack of gas exchange (complete lung collapse), the hepatic sinus would have to be at least 1.9 l in volume. Given that these conditions were optimal for maintenance of aortic S_O2 (low heart rate and high anterior vena caval S_O2), it is unlikely that the hepatic sinus is large enough to contribute a sufficient volume of highly saturated blood to venous return in order to maintain high aortic S_O2 during 3 min of complete absence of gas exchange. Furthermore, any venous inflow from the coronary sinus, thebesian veins and bronchial veins into the heart would also decrease aortic S_O2.

The decrease in anterior vena caval S_O2 and the increase in heart rate during the lung collapse period (Fig. 8) also reinforce the argument that the hepatic sinus is not large enough to maintain aortic S_O2. At the start of lung re-expansion, mean anterior vena caval S_O2 was 64% and heart rate was near 33 beats min⁻¹ (cardiac output=3.9 l min⁻¹). In order to maintain aortic S_O2 at 92% during those conditions, the above shunt calculation resulted in an anterior vena caval flow of 383 ml min⁻¹ and hepatic sinus flow of 3577 ml min⁻¹. At this flow rate, the assumed 95% saturated blood in a 1-liter hepatic sinus would be emptied in approximately 15 s and be replaced with desaturated blood returning from the posterior vena cava. There is also no mechanism to account for an almost 10-fold greater return of blood to the heart from the posterior vena cava than from the anterior vena cava. In addition, even at the lower heart rates earlier in the lung collapse period, there would be some desaturated blood returning to the hepatic sinus and decreasing the assumed 95% value in the equation.

These calculations indicate that although a–v shunts may arterialize the venous blood and increase the magnitude of the total blood oxygen store, the size of the hepatic sinus in the sea lion is inadequate to maintain aortic saturation during a period of complete lung collapse and lack of gas exchange. How then does aortic S_O2 remain elevated above 90% during the presumed period of lung collapse? We suggest that some gas exchange may still occur even at or below depths where lung collapse has been predicted to occur. These predictions of lung collapse have been based on the depths at which there were characteristic changes in the arterial P_O2 profile of deep-diving sea lions (McDonald and Ponganis, 2012). As originally pointed out in that study, although the changes in arterial P_O2 profile were consistent with a significant decrease in gas exchange, continuation of some gas exchange at great depths could not be ruled out. Resolution of this question requires further anatomical and physiological investigations.

Blood oxygen profiles in the anterior vena cava demonstrated that previously documented oxygen profiles in the posterior vena cava were not representative of the entire venous oxygen store. Similar to findings in the posterior vena cava, there was marked variation in anterior vena caval S_O2 profiles during shallow dives (<100 m in maximum depth). Notably, these shallow and short duration dives (1) often began before full re-saturation of the venous oxygen store, and (2) often had a net increase in venous S_O2 during the dive. During deep dives, although S_O2 values in both the anterior and posterior venae cavae were often >90% at the start of dives, the S_O2 profiles markedly differed between the two blood vessels as the dives progressed. Anterior vena caval S_O2 remained high until the bottom phase of the dive and then decreased during ascent, while posterior vena caval S_O2 declined markedly during descent and then often increased during ascent. These differences likely result from variations in peripheral blood flow patterns and tissue transit times of blood entering the respective venae cavae. We conclude that regulation of peripheral blood flow, tissue oxygen extraction and a–v shunting is complex and variable during dives of sea lions, and that mixed venous S_O2 measurements are needed to completely assess venous blood oxygen profiles and net oxygen depletion during dives.

Although our findings support the hypothesis of a highly oxygenated central venous oxygen reservoir at the start of deep dives, declines in anterior vena caval S_O2 and increases in heart rate during periods of lung collapse indicate that the hepatic sinus of the sea lion is not large enough to account for the maintenance of arterial S_O2 values greater than 90% during the period of lung collapse. Although arterial P_O2 profiles are indicative of periods of significantly decreased gas exchange during deep dives of sea lions (McDonald and Ponganis, 2012), our findings suggest that some gas exchange may still persist. Further investigations are needed to address this question.

The logistics and field work for this field work were made possible by the intense efforts of many volunteers and we are very thankful for their help. Specifically, we thank J. Ugoretz, G. Smith, P. Thorson, R. Walsh, C. Stehman, N. Hauff, B. Ruddick, A. Livingstone, M. Jeffko, A. Currylow and C. Williams.