O₂ store management in diving emperor penguins

The concept of an aerobic dive limit (ADL) is central to most eco-physiological models of foraging behavior in marine mammals and diving birds. Although routinely calculated (but rarely measured), the ADL has remained a physiological black box. Conceptually, it should be dependent on the rate, pattern and magnitude of O₂ store depletion during dives. However, the management and utilization of O₂ stores during diving may vary in different species or situations due to (a) differences in the magnitude and distribution of O₂ stores and (b) differences in the intensity of physiological responses underlying the rate of O₂store depletion during a dive. The ADL has been defined as the dive duration associated with the onset of post-dive blood lactate accumulation, and it has been proposed that the lactate accumulation at the ADL is secondary to localized tissue O₂ depletion and increased glycolysis, most probably in active but ischemic locomotory muscle(Kooyman and Ponganis, 1998). On the other hand, less-than-expected decreases in muscle O₂saturation during dives of Weddell seals (Leptonychotes weddellii)have led to the suggestion that muscle blood flow persists during dives(Guyton et al., 1995), and to the hypothesis that maintenance of muscle blood flow during dives allows for the matched parallel depletion of blood and muscle O₂ stores(Davis and Kanatous, 1999). The primary difference between these two different ADL scenarios is dependent on the isolation of working muscle from the blood O₂ store during a dive.

Therefore, in order to increase our understanding of O₂ store management and the physiological basis of the ADL, we reviewed blood oxygen partial pressure (P_O2) profiles and analyses of blood samples from diving emperor penguins (Aptenodytes forsteri) for findings indicative of blood flow patterns and the utilization of O₂ stores during dives. Emperor penguins diving at an isolated dive hole are particularly appropriate for such investigations because the ADL has been determined by post-dive blood lactate measurements to be 5.6 min, and neither the respiratory nor the blood O₂ stores are depleted at the ADL (Ponganis et al., 1997; Ponganis et al., 2007; Stockard et al., 2005). For dive durations beyond the 5.6 min ADL, post-dive blood lactate concentrations are elevated.

We had two primary hypotheses. First, we hypothesized that the maintenance of gas exchange with the large respiratory O₂ store in emperor penguins contributes to significant O₂ transfer from the respiratory system to the blood during dives. Second, we hypothesized that muscle is isolated from the blood O₂ store during dives, i.e. muscle blood flow and blood-to-muscle O₂ transfer stop during dives. In particular, we reasoned that blood P_O2 profiles during dives should be a reflection of (a) pulmonary gas exchange, (b) a reduction in cardiac output secondary to the bradycardia of diving, and (c) changes in blood O₂extraction due to the reduction/redistribution of organ blood flow secondary to both the fall in cardiac output and peripheral vasoconstriction during dives.

In the austral springs of 2001, 2003–2005 and 2007, non-breeding emperor penguins (Aptenodytes forsteri Gray, 20–30 kg in body mass) were captured on the sea ice of McMurdo Sound or at Terra Nova Bay near Cape Washington (74°36′, 165°24′). They were maintained at the corralled isolated dive hole of the Penguin Ranch in McMurdo Sound(77°41′, 165°59′) for 6 weeks as in past studies(Kooyman et al., 1992; Ponganis et al., 2001; Ponganis et al., 2004; Ponganis et al., 2003), and then released at the McMurdo Sound ice edge. Blood-sampling catheters or P_O2 electrodes/thermistors (Licox C1.1 Revoxode; Integra LifeSciences, Plainsboro, NJ, USA; model 554, Yellow Springs Instruments, Yellow Springs, OH, USA) were inserted percutaneously into the aorta or vena cava of emperor penguins under general isoflurane anesthesia as described previously (Ponganis et al.,2007; Ponganis et al.,2001; Ponganis et al.,2004; Ponganis et al.,2003). In addition, in two birds in 2007, successful femoral artery catheterization was achieved with a 4.5 in, 20 g catheter (Arrow International, Reading, PA, USA); only a P_O2electrode was inserted through this small catheter. The birds were also equipped with a custom-made P_O2/temperature recorder (UFI, Morro Bay, CA, USA) and an Mk9 time depth recorder (TDR,Wildlife Computers, Redmond, WA, USA) as previously described(Stockard et al., 2005). After overnight recovery from anesthesia, birds were allowed to dive at the isolated dive hole. After 1–2 days of diving and data collection, catheters,probes and recorders were removed under general anesthesia. Data from the recorders were downloaded to a personal computer and analyzed with Excel(Microsoft, Redmond, WA, USA), Origin (Origin Lab., Northampton, MA, USA) and SPSS (SPSS Inc., Chicago, IL, USA) software. As previously reviewed(Ponganis et al., 2007), all P_O2 values were corrected to 38°C for construction of the P_O2 profiles. In the two birds without thermistors in 2007, blood temperature was assumed to be 38°C.

All partial pressures are expressed, as measured, in mmHg (7.5 mmHg=1 kPa). Means are expressed ±s.d. Significance was assumed at P<0.05. All procedures were approved under a UCSD Animal Subjects Committee protocol and US Antarctic Treaty Permit.

The custom-designed blood sampler (1.25 kg, 24 cm×8.5 cm diameter)collected one sample per dive and consisted of an underwater housing, a programmable, microprocessor-based control board (UFI), a pressure transducer(Model 96, IC Sensors, Milpitas, CA, USA), a blood access port, and two solenoid valves (2W13W-1NR-A1C4, Snap Tite, Erie, PA, USA) which each controlled access to a syringe, one for waste (>3 times dead space volume)and one for the blood sample (Fig. 1). Collection of blood relied upon the pressure difference between the ambient pressure at depth and that inside the housing. The catheter port inside the housing was connected to the solenoid valves via high pressure Teflon tubing and swedgelock fittings. The sampler was programmed to obtain the sample after detection of a specified depth, or to obtain the sample when a specified time interval had elapsed while the bird was beneath a specified `start-of-dive' depth threshold. The sampler was also programmed to delay activation for a minimum of 1 h after the sampler was mounted on a bird in order to avoid any effects secondary to restraint. The unit was retrieved by recapture/restraint of the bird when blood was visible in the external transparent stopcock connection between the catheter and blood sampler port (no. 91041 stopcock, Mallinckrodt, Glen Falls, NY, USA; 3.5 in extension tubing no. 53035, Medex, Dublin, OH, USA). Catheters for blood sampling were in the femoral artery, femoral vein or axillary vein. Lyophilized heparin (210-6, Sigma Aldrich, St Louis, MO, USA) was placed in the sample syringe (Becton Dickinson, Franklin Lakes, NJ, USA) to prevent clotting. Because the volume of the blood sample syringe was diluted by the dead space flush volume of its solenoid valve, analyses for blood gases,O₂ content and lactate concentration were performed on samples collected anaerobically from the undiluted blood in the Teflon tubing in the sampler. As in past blood sampler studies(Falke et al., 1985), nitrogen content and, in some cases, lactate concentration determined from the sample syringe were corrected for dilution by comparison of hematocrit (determined by microcentrifugation) or hemoglobin (Hb) concentration in the sample syringe versus that in the tubing. Blood samples were analyzed within 120 min after collection of the sample due to continued diving activity and the time required to recapture the bird and dismantle the blood sampler.

In addition to collection of blood samples from birds at rest(Ponganis et al., 2007) and from diving birds, occasional blood samples were also manually collected from birds under anesthesia and from restrained birds during surface intervals. Blood gas (P_O2, P_CO2 and pH) and lactate concentration analyses on these samples were conducted with a Series 200 i-STAT Portable Clinical Analyzer (CG4+ cartridge, Abbott Point of Care Inc., East Windsor, NJ, USA) at 37°C(Stockard et al., 2007). O₂ content was determined with a Tucker chamber technique (Models SI 782 O₂ meter and 1302 O₂ electrode, Strathkelvin,Motherwell, UK) (Tucker,1967). Blood N₂ content was determined with the Van Slyke technique (Ponganis et al.,1999). The N₂ solubility coefficient (1.44 ml N₂ per 100 ml blood per atmosphere N₂) was determined in blood tonometered with ambient air or 100% N₂ at 38°C. Hb concentration was determined with a cyanmethemoglobin spectrophotometric technique (Stockard et al.,2007). Blood samples were analyzed within 10 min of collection. Blood gas, O₂ content, lactate concentration and P_N2 were stable for as long as 4 h at room temperature in the blood gas syringes (Model 4041, Sims Portex, Keene, NH,USA).

For birds equipped with the P_O2 recorder,most dive durations and maximum depths have previously been described, and were similar to those in past studies at the isolated dive hole(Ponganis et al., 2007; Stockard et al., 2005). Additional arterial P_O2 profiles obtained from two birds in 2007 increased the number to 57 dives in five birds; venous profiles were from 130 dives in nine birds. Mean dive duration of all 187 dives with P_O2 recorders was 5.68±3.99 min; mean dive depth was 34±22 m. The relationship of surface interval to dive duration (Fig. 2) was similar to that previously reported for emperor penguins at sea(Kooyman and Kooyman, 1995; Wienecke et al., 2007).

Dive durations and maximum depths of penguins equipped with the blood sampler ranged from 3 to 12.8 min and 28 to 55 m, respectively. These depths and durations were within the range of dive durations exhibited by birds without such a large recorder. However, as previously observed in emperor penguins equipped with and without a Crittercam camera(Ponganis et al., 2000), the increased drag of the larger unit probably resulted in dives of shorter duration. Nonetheless, dive durations as long as 12.8 min did occur in birds equipped with the sampler. Arterial blood samples were obtained 2.1, 2.8 and 5.2 min into dives of 4.4, 10.1 and 10.5 min duration in one bird. Venous samples were obtained 1.1, 1.7, 2.2, 2.3, 3.2 and 10.5 min into dives of 7.3,7.8, 4.1, 3.0, 6.6 and 12.8 min duration in four birds. In general, the number of samples obtained was limited by (1) the design of the sampler (one sample per dive and size of the unit), (2) the time required for filling the syringes at these shallow depths (1 min per syringe), (3) our inability to predict the duration of a dive when trying to obtain samples late in a dive, and (4) a frequent lack of interest in diving that occurred after a bird was restrained and equipped with the sampler but which resolved itself after removal of the sampler.

Arterial P_O2 profiles during dives (see Fig. 3) revealed that (1)pre-dive P_O2 was usually elevated above the previously reported level (68 mmHg) for emperor penguins at rest(Ponganis et al., 2007), (2) P_O2 consistently increased during the early part of the dive, usually peaking at the end of the initial descent of the dive, (3) at similar depths (Fig. 4) these peak arterial P_O2 values were distinctly less than the previously measured peak air-sac P_O2 values(Stockard et al., 2005), (4)arterial P_O2 declined as the dive progressed,resulting in final P_O2 values similar to those previously reported for air-sac and venous P_O2(Fig. 5), and (6) prior to the next dive, post-dive arterial P_O2 reached 68 mmHg within 1.92±1.89 min (N=53).

Venous P_O2 profiles during dives (Figs 6 and 7) were remarkable for their variable patterns. During the pre-dive period, venous P_O2 often fluctuated; it decreased prior to the start of the dive in 51% of dives (see Fig. 6 for an example). Pre-dive venous P_O2 ranged from 21 to 82 mmHg (mean 44±12 mmHg), and initial dive P_O2 ranged from 24 to 79 mmHg (mean 45±12 mmHg). In 61% of dives, pre-dive venous P_O2 was greater than that (41 mmHg) of birds at rest (Ponganis et al., 2007). Dive duration was very weakly, but significantly, related to each of these variables (Fig. 8; pre-dive P_O2: y=0.113x+1.18, R²=0.08, P<0.05; initial P_O2: y=0.099x+1.18, R²=0.06, P<0.05). In 90% of dives, the venous P_O2 increased transiently (within the first 3 min in 78% of dives) with a maximum P_O2 of 53±18 mmHg (range 28–129 mmHg) and a mean increase in P_O2 of 11±12 mmHg (range 2–76 mmHg). Small transient increases could occur later and more than once during a dive (see Fig. 7). Maximum venous P_O2 during a dive was >68 mmHg in 15%of dives, and >41 mmHg in 71% of dives; again, these levels correspond to the arterial and venous P_O2 values of emperor penguins at rest (Ponganis et al.,2007). Regression analysis revealed that dive duration was not significantly related to the maximum venous P_O2during a dive. During short dives, the increase in venous P_O2 could result in final venous P_O2 values that were greater than the start-of-dive venous P_O2 [see figure 2 in Ponganis et al.(Ponganis et al., 2007)].

After 84 dives, post-dive venous P_O2 reached 41 mmHg prior to the next dive; this required 2.23±2.64 min. After a 23.1 min dive, venous P_O2 reached this value in less than 2 min. Regression analysis revealed that the relationship of prior dive duration to the time for venous P_O2 to recover to 41 mmHg was not significant. The relationship of surface interval to the time to reach a venous P_O2 of 41 mmHg was also not significant.

In the post-dive period, venous P_O2initially continued to decline below the final dive value after 48% of dives. The time for post-dive venous P_O2 to return to 41 mmHg was not significantly longer than that required for arterial P_O2 to return to 68 mmHg, but both intravascular return times were significantly longer than the 0.92±0.44 min [N=73, data from Stockard et al.(Stockard et al., 2005)]required for air-sac P_O2 to return to the level of birds at rest (ANOVA, F=9.308, P<0.05; Tukey HSD post-hoc analysis, P<0.05).

Fig. 9 represents a comparison of typical air-sac, arterial and venous P_O2 profiles and a heart rate profile of different birds for shallow (<50 m) dives of approximately 8 min duration. The data are from this and prior studies(Meir et al., 2008; Stockard et al., 2005).

Arterial and venous N₂ partial pressures(P_N2) in emperor penguins at rest were 0.86±0.04 atmospheres absolute (ATA; N=3) and 0.85±0.06 ATA (N=9), respectively. Blood gas(P_O2, P_CO2, pH),O₂ content and lactate concentrations of these birds at rest have previously been reported (Ponganis et al.,2007).

Venous P_O2 and P_CO2 from the brachial vein (wing) of a restrained penguin during a post-dive surface interval were 81 and 38 mmHg, respectively. During anesthesia of one bird on 100% O₂, near-simultaneous sampling of the brachial artery and brachial vein yielded P_O2 values of 247 and 260 mmHg, respectively. Hemoglobin concentrations of three penguins during anesthesia and at rest were 16.5±0.6 and 16.4±0.8 g dl^–1, respectively.

Blood gas (P_O2, P_CO2,pH) and O₂ content were measured in two arterial and three venous samples obtained during dives (Table 1). It is notable that arterial and venous P_O2 values at between 1 and 3 min into the dives were above or near the respective values of penguins at rest. Blood lactate concentration was less than 2.0 mmol l^–1 as far as 10.5 min into a dive (Fig. 10). Mean arterial P_N2 at depths of 25–32 m at 2.1–5.2 min into a dive was elevated at 2.1±0.7 ATA (N=3). Venous P_N2 at 20–37 m depth at 1.7–3.2 min into a dive was 2.1±0.7 ATA (N=4). These arterial and venous P_N2 values were not significantly different (Student's two sample t-test).

Hb concentration during dives of the same three penguins sampled for Hb under anesthesia and at rest was 16.3±0.7 g dl^–1. The Hb concentrations during anesthesia, at rest and during dives were not significantly different (ANOVA, F=5.14, P=0.94).

The initial transient increases in arterial P_O2 profiles during dives reflected the compression hyperoxia previously observed in the air sacs of diving emperor penguins and confirmed the maintenance of pulmonary gas exchange and transfer of O₂ from the respiratory store into the blood O₂ store during the dive. As a result, during some dives, final arterial P_O2 values were greater than the mean arterial P_O2 of 68 mmHg of birds at rest(Fig. 5) and, in other dives,equivalent to or even greater than start-of-dive P_O2 values(Ponganis et al., 2007). Similarly, final venous P_O2 values of some dives, especially short-duration dives, were not only greater than the mean venous value of penguins at rest but also greater than start-of-dive P_O2(Ponganis et al., 2007).

Evidence for the maintenance of gas exchange during dives was also provided by the elevations of arterial/venous P_O2 and P_N2 recorded in the blood sampler data. In Table 1, arterial and venous P_O2 values at 1–3 min into dives were greater than or near the corresponding values of penguins at rest(Ponganis et al., 2007). In addition, during dives, mean arterial P_N2 rose to 2.1 ATA at depths of 25–32 m; this is again indicative of continued gas exchange during these shallow dives. During simulated dives, similar increases have been recorded in arterial P_N2 in Adelie penguins (Pygoscelis adeliae) and in venous P_N2 in king penguins (Aptenodytes patagonicus) (Kooyman et al.,1973; Ponganis et al.,1999).

The arterial P_O2 profiles and P_N2 values also provide insight into possible mechanisms affecting gas transfer in the lungs. Although the mean P_N2 of 2.1 ATA at 25–32 m depth is approximately twice the arterial value of penguins at rest, it is notable that the calculated air-sac P_N2 at these depths is 2.8–3.4 ATA, assuming an air-sac fraction of 0.8. The actual air-sac P_N2 is probably even higher since the O₂ fraction will decrease secondary to O₂ consumption(Kooyman et al., 1973). Thus,there appears to be an air sac-to-arterial difference not just for P_O2 (Fig. 4) but also for P_N2. These differences may be due to ventilation–perfusion mismatch in the lung during the dive and/or due to the thickened parabronchial capillary blood-to-air barrier (Powell,2000; Welsch and Aschauer,1986). It is also notable that a large air sac-to-arterial difference in P_O2 has been observed in emperor penguins at rest (Ponganis et al.,2007). The typical increases in arterial P_O2 prior to the start of diving activity (see Fig. 3) suggest that the tachycardia and hyperventilation of the pre-dive period(Kooyman et al., 1971; Kooyman et al., 1992; Meir et al., 2008) improve ventilation–perfusion matching and account for these changes in P_O2 during the pre-dive period.

Despite air sac-to-arterial differences in P_O2 and P_N2, the arterial P_O2 profiles and blood gas analyses obtained in this study all support the concept that pulmonary gas exchange is maintained during these shallow dives of emperor penguins with net transfer of O₂ from the respiratory O₂ store to the blood(Stockard et al., 2005).

Venous P_O2 profiles and intradive blood lactate concentration were first examined for evidence regarding muscle blood flow patterns during diving. We had hypothesized that a lack of muscle perfusion during dives would prevent muscle O₂ extraction from blood and also prevent the washout of lactate from muscle during dives longer than the ADL. In particular, despite the fact that buoyancy, stroke frequency and, therefore, muscle workload were highest during the initial descent period(Sato et al., 2002; van Dam et al., 2002), we expected the decline in venous P_O2 during the initial descent to be slow because of the hypothesized lack of muscle perfusion.

Although venous P_O2 profiles were variable in pattern and in overall rate of decline of P_O2, 71% of dives actually had an increase in venous P_O2 during the initial descent period(Figs 6 and 7). In 15% of dives, venous P_O2 increased to values that were greater than the mean arterial P_O2 of birds at rest. The P_O2 of venous blood from which muscle has extracted O₂ should decline, not rise. Such increases in venous P_O2 during the initial descent period are thus not consistent with perfusion of active locomotory muscle. This hypothesis is further supported by dives in which final venous P_O2 values were greater than start-of-dive venous P_O2 values (see range of final P_O2 values for short-duration dives in Fig. 5) (see also Ponganis et al., 2007). In addition, blood lactate concentrations remained less than 2 mmol l^–1 even as far as 10.5 min into a dive, well beyond the ADL(Fig. 10). Together, all these findings support a classical model of bradycardia and peripheral vasoconstriction in diving emperor penguins, in which muscle is isolated from the circulation during the dive, and lactate washout occurs during the eupneic period when muscle perfusion is re-established(Scholander, 1940; Scholander et al., 1942). This postulated lack of muscle perfusion during dives is also consistent with the lack of a relationship between stroke frequency and heart rate during dives of emperor penguins (Meir et al.,2008) and with increases in pectoral muscle temperature during dives of both emperor penguins and king penguins(Ponganis et al., 2003;Schmidt et al., 2007).

Increased muscle perfusion and blood O₂ extraction during both the pre-dive period and the post-dive period were suggested by decreases in venous P_O2 during these times (see Fig. 6 as an example). Although venous P_O2 may decrease due to variety of factors, including O₂ consumption by tissues other than muscle, we suggest that these declines in venous P_O2 and previously documented pre-dive and post-dive tachycardias(Froget et al., 2004; Kooyman et al., 1992; Meir et al., 2008) support the concept of increased muscle perfusion and muscle O₂ extraction in both the pre-dive and post-dive period. Such pre-dive muscle hyperemia has also been postulated in king penguins because of decreases in muscle temperature prior to dives (Schmidt et al., 2006). If this hypothesis is correct, the declines in pre-dive venous P_O2 imply that muscle with high myoglobin content is not always fully saturated with O₂ during some rest periods and interdive intervals. This last suggestion is supported by the wide range of pre-submersion muscle O₂ contents in the early research of Scholander and colleagues(Scholander, 1940; Scholander et al., 1942). During the post-dive period, muscle reperfusion is, of course, expected. Venous P_O2 may not always decline during the post-dive period, however, because, even after muscle O₂ extraction from re-oxygenated arterial blood, the resulting venous P_O2 may still be greater than the end-of-dive venous P_O2.

The return time of post-dive venous P_O2 (2.3 min) to the mean value of penguins at rest was slightly longer than the 1.9 min value for arterial P_O2, but these times did not differ significantly. This was contrary to our hypothesis that O₂ uptake/consumption by muscle and other organs during the surface period would slow the replenishment of the venous blood O₂ store relative to the arterial rate. As expected, the return time (0.9 min) for air-sac P_O2 to reach the level at rest was significantly less than either of the blood values (ANOVA). Overall, the post-dive increases in air-sac, arterial and venous P_O2 values during the surface interval were rapid and consistent with prior reports of (1) a decrease in the post-dive tachycardia and resumption of a respiratory sinus arrhythmia within 2 min(Meir et al., 2008) and (2) a rapid decline in post-dive respiratory rate to levels of birds at rest in 3 min (Kooyman et al., 1971). These observations all suggest that the blood O₂ store is replenished and most CO₂ is exhaled within 3 min after a dive. The relationship of surface interval to the time for post-dive venous P_O2 to reach 41 mmHg was not significant. A surface interval could be as much as 30 times longer than the time for venous P_O2 to return to a resting level. Although the time course of muscle re-oxygenation during the surface interval remains to be investigated, other factors such as sated appetites, food digestion, metabolic processing, social interactions and visits to the dive hole by Weddell seals probably also contribute to the duration of the surface interval of birds diving at the isolated dive hole.

The existence and function of arterio-venous shunts are often documented by elevations in venous P_O2 or O₂content above control values or average values of animals at rest. Pre-dive venous P_O2 was greater than that of penguins at rest prior to 61% of dives. Arterio-venous anastamoses, which have been described anatomically in the extremities of birds(Arad et al., 1989; Thomas and Fordyce, 2007), may contribute to these elevations during surface periods. In addition, high blood flow through the vasculature of an extremity such as the wing may act as an effective shunt given the relatively low O₂ requirements of the bones, tendons and ligaments of the wing. Such flow through the wings and feet during the surface tachycardia is consistent with the re-warming previously recorded in the extremities during the surface period(Ponganis et al., 2003). In addition, this suggestion is supported by a brachial vein (in the wing) blood sample obtained during a surface interval. The P_O2 (81 mmHg) and P_CO2 (38 mmHg) not only were characteristic of arterial blood but, as would be expected during hyperventilation, also were above (P_O2)and below (P_CO2) the respective arterial values of penguins at rest. The similarity of the brachial vein P_O2 of 260 mmHg to that in the brachial artery during anesthesia on 100% O₂ was also consistent with arterio-venous shunting through the wing. Indeed, this mechanism probably contributes to the high venous P_O2 values reported during avian anesthesia (Jaensch et al., 2002). Therefore, we propose that arterio-venous shunting through the extremities during the post-dive period contributes to elevated pre-dive venous P_O2 values as well as to the rapid recovery of post-dive venous P_O2.

Another site of arterio-venous shunting may be the brood patch vasculature. It has already been suggested that increased perfusion accounts for rewarming of the brood patch during surface periods of king penguins at sea(Schmidt et al., 2006). Since the brood patch contains arterio-venous anastamoses(Midtgard et al., 1985) and is not a site of high metabolic activity, increased flow in brood patch vessels could also contribute to elevations in venous P_O2.

Although the relationship of dive duration to pre-dive or initial venous P_O2 was weak (R²<0.1, Fig. 8), the potential significance of arterio-venous shunting during the surface period is exemplified by the pre-dive venous P_O2 of 63 mmHg prior to a 23.1 min dive, the longest dive ever reported for an emperor penguin. The Hb saturation, blood oxygen content and total blood O₂store at that P_O2 should be considerably greater than the respective values at the mean venous P_O2 of 41 mmHg of birds at rest. Examination of the relationship of venous Hb saturation to dive duration awaits determination of the emperor penguin O₂–Hb dissociation curve.

The most remarkable and unexpected increases in venous P_O2 occurred in 78% of dives during early descent. In 15% of all dives, the peak venous P_O2 values were greater than the arterial value of birds at rest. Values could be as high as 90 mmHg(Fig. 6). Such high P_O2 values, which reflect arterial blood values and fully saturated Hb, suggest the existence of a significant arterio-venous shunt in addition to muscle ischemia during the dive. The elevated venous P_N2 values collected during the early parts of dives were also not significantly different from arterial values in the same depth range. This suggests minimal tissue uptake of N₂ and is again consistent with the existence of an arterio-venous shunt during the dive.

We propose that the blood flow through an arterio-venous shunt could be supported by the transient increase in heart rate that occurs during the early descent period of dives by penguins (Froget et al., 2004; Green et al.,2003; Meir et al.,2008). Therefore, as illustrated in Fig. 9, we hypothesize that a primary function of the transient increase in heart rate early in the dive is to enhance the transport of O₂ from the air sac and lungs into the arterial system and then, through a still-open arterio-venous shunt, into the venous system. This essentially represents a mechanism by which the size of the venous O₂ reservoir can be increased during the dive by transport of O₂ from the large respiratory O₂ store of the penguin into the blood. As such large increases in venous P_O2 were not always seen, the magnitude of such a shunt may vary in different dives. Indeed, the variability in venous P_O2 profile patterns during dives suggests that the vascular response may be quite plastic and adaptive to different conditions.

However, the anatomical basis of such an intradive arterio-venous shunt remains to be determined. Based on observations that bleeding from a nicked wing continued during the first 2.5 min of a dive of an emperor penguin(Kooyman et al., 1971), it is possible that blood flow through the wings contributes to these elevations in venous P_O2 during dives. If so, the countercurrent heat exchangers of the extremities(Arad et al., 1989; Thomas and Fordyce, 2007) are extremely efficient, given the preservation of core temperature and the significant cooling of both the wings and feet during dives(Ponganis et al., 2001; Ponganis et al., 2004; Ponganis et al., 2003). Conceivably, shunt flow may also occur in arterio-venous anastamoses described in the proximal regions of the extremities of birds(Arad et al., 1989). In addition, blood flow through the brood patch region could contribute to an arterio-venous shunt during a dive. It has already been suggested that such flow contributes to paradoxical increases in brood patch temperature during dives of king penguins (Schmidt et al.,2006). Another mechanism of arterio-venous shunting may involve the muscle vascular bed because avian muscle has been considered to have`luxuriant', non-nutrient blood flow that contributes to higher than expected P_O2 in the venous effluent of muscle(Folkow et al., 1966; Grubb, 1981). In summary, there may be several anatomical sites for the hypothesized arterio-venous shunt during dives.

It should also be noted that it is unlikely that contraction of a large spleen and expulsion of arterialized blood into the venous system contribute to such elevations in venous P_O2 during dives of emperor penguins. First, we have never observed a large spleen in any of our many dissections of emperor penguin carcasses. Second, Hb concentration in three birds did not decrease during anesthesia or increase during diving in this study. Such changes would be expected if a large spleen dilated during anesthesia or contracted during diving activity as in seals(Hurford et al., 1996; Ponganis et al., 1993; Ponganis et al., 1992; Qvist et al., 1986). Third,blood introduced from the spleen should have a P_N2 near that at the surface and should delay the rise in venous P_N2 at depth, thus increasing, not decreasing, the arterial-to-venous difference in P_N2.

Blood P_O2 profiles and analyses of blood samples during dives of emperor penguins are consistent with (a) maintenance of gas exchange between the air sacs, lungs and blood during dives, (b) muscle ischemia during dives, (c) muscle hyperemia both before and after dives, and(d) utilization of arterio-venous shunts during the surface period and during the dive. These physiological processes allow (1) utilization of the large respiratory O₂ store of penguins, (2) regulated depletion of the blood O₂ store, (3) isolation of the respiratory and blood O₂ stores from working muscle, and (4) optimized loading of O₂ stores both before and after dives. The proposed muscle ischemia and the incomplete depletion of the respiratory and blood O₂ stores at the ADL suggest that muscle O₂ depletion and subsequent lactate accumulation represent the primary source of the post-dive increase in lactate at the ADL. The potential existence of an arterio-venous shunt during a dive was unexpected. Such increased flow through an arterio-venous shunt could be supported by the transient increases in heart rate previously reported during the early descent periods of dives.