Altitude matters: differences in cardiovascular and respiratory responses to hypoxia in bar-headed geese reared at high and low altitudes

Birds exhibit enhancements at each step of their O₂ transport cascade that help to support the high flux of O₂ required for flight and endothermy (Scott, 2011). This cascade describes the flow of O₂ from the atmosphere to the mitochondria in vertebrates and the steps include ventilation, pulmonary O₂ diffusion, perfusion and tissue O₂ diffusion. Bar-headed geese (Anser indicus), which migrate biannually over the Himalayan mountain range (Bishop et al., 2015; Hawkes et al., 2011; Scott et al., 2015), exhibit further adaptations at each level of this transport cascade (Black and Tenney, 1980; Meir and Milsom, 2013; Scott and Milsom, 2006, 2007; Scott, 2011; Scott et al., 2015; Weber et al., 1993). While providing evidence of physiological differences between bar-headed geese and closely related low-altitude species, all studies to date have been conducted on groups of bar-headed geese born and raised for generations at sea level (Black and Tenney, 1980; Fedde et al., 1989; Hawkes et al., 2014; Scott and Milsom, 2007). Thus, relatively little is known about the influence of phenotypic plasticity (i.e. acclimatization) and developmental plasticity on the physiology of this species.

Hypoxic responses vary depending on when during an animal's development it is exposed to hypoxia and the duration of the exposure. In chickens, hypoxic exposure had no reported developmental effects when it occurred early in development or for an acute duration (Ferner and Mortola, 2009). When exposure to hypoxia occurred during the entire duration of incubation, or even during the final week, however, the hypoxic ventilatory response (HVR) of the chicks was blunted. This was a result of reduced ventilatory chemosensitivity (Ferner and Mortola, 2009; Mortola, 2011; Szdzuy and Mortola, 2007). Sustained hypoxic exposure in low-altitude birds also decreased whole animal oxygen consumption rate (V̇_O2) and growth rate (Mortola, 2011). However, embryos of some birds successfully hatch with normal growth rates and V̇_O2 at altitudes of 4000–6500 m (Carey et al., 1982; León-Velarde and Monge-C, 2004). For example, the high-altitude migrating bar-headed goose maintained V̇_O2 as an embryo when exposed acutely to extreme ambient hypoxia (11.7 kPa) (Snyder et al., 1982). Thus, although hypoxia exposure during development can alter physiological responses in birds, some high-altitude bird species have adapted to mitigate these effects.

In adult animals, many rapid physiological changes occur minutes to hours after acute hypoxic exposure and can be modified during chronic acclimatization (Ivy and Scott, 2015; Powell et al., 1998). Black and Tenney measured changes in V̇_O2, total ventilation (V̇_R) and cardiac output (Q̇) during progressive hypoxic exposure in bar-headed geese following short-term (4 week) acclimation to simulated altitude (5640 m) (Black and Tenney, 1980). The acclimated bar-headed geese did not become polycythemic, a trait characteristic of other species endemic to high-altitude regions. They also displayed higher resting V̇_O2, V̇_R and Q̇ under ambient conditions, as well as greater increases in V̇_R and Q̇ during exposure to progressive hypoxia (Black and Tenney, 1980). Thus, the HVR and hypoxic cardiovascular response of bar-headed geese can be affected by short-term high-altitude acclimation.

In the present study, we extended this work by examining the effects of high-altitude rearing and development on the physiological responses to hypoxia in bar-headed geese. This is significant, considering that the effects of pre- or postnatal hypoxic exposure can differ significantly, and persist throughout adult life (Bavis, 2005; Ivy and Scott, 2015). Our primary objective, therefore, was to compare the changes in the convective steps in the O₂ transport cascade, ventilation and circulation, of low-altitude-reared (LAR) bar-headed geese to a group of wild high-altitude-reared (HAR) bar-headed geese during short-term progressive hypoxic exposure. These responses were compared with those of LAR barnacle geese (Branta leucopsis), a member of a closely related genus that also migrates, but only at low altitude. We predicted that, because of lifelong exposure to high altitude, HAR bar-headed geese would show further enhancements in the overall magnitude of V̇_R and Q̇ compared with those previously reported during short-term high altitude acclimation of this species (Black and Tenney, 1980). In addition, because bar-headed geese are known to be capable of large increases in V̇_R during hypoxic exposure (Black and Tenney, 1980; Scott and Milsom, 2007), we predict that V̇_R would be greatest in the HAR bar-headed geese and lowest in the barnacle geese.

The experiments on the LAR geese were performed at the University of British Columbia, where the geese were housed at the Centre for Comparative Medicine. Cardiovascular measurements were made on 6 cannulated bar-headed geese (Anser indicus Latham 1790) (2.5±0.2 kg) and respiratory measurements were made on 5 non-cannulated bar-headed geese (2.4±0.1 kg). All cardiorespiratory measurements were made on 7 cannulated barnacle geese (Branta leucopsis Bechstein 1803) (2.5±0.2 kg). Cardiorespiratory measurements were also obtained from 5 cannulated HAR bar-headed geese (2.1±0.1 kg) that were born in the wild at 3200 m at Lake Qinghai, China and reared in captivity for at least 1 year at the lake. All experimental animals were fed similar diets, housed in outdoor pens under natural conditions and experienced similar levels of (in)activity. The HAR bar-headed geese, however, were born and reared in hypobaric hypoxia. All experimental procedures were conducted according to guidelines approved by the Animal Care Committee at the University of British Columbia under the guidelines of the Canadian Council on Animal Care.

For cardiovascular measurements, surgery was conducted under general and local anesthesia 1 day before the hypoxic exposure. All geese were first weighed, gently restrained and induced with isoflurane (4%) supplemented with O₂ (100%) by facemask prior to intubation. General anesthesia was maintained with isoflurane and O₂. The right brachial artery and vein were accessed via a small incision and blunt dissection and cannulated with polyurethane cannulae (PU-90; 0.102 cm internal diameter×0.410 cm outer diameter) filled with 1000 IU ml⁻¹ heparinized saline. Geese were recovered for at least 24 h prior to experimentation.

Each goose was placed in a flexible cradle that permitted unrestricted breathing. Its head was placed in an opaque Plexiglass chamber large enough to accommodate free movement of the neck and head, sealed around the neck with a flexible latex collar and supported by the cradle. Geese in the experimental apparatus were allowed 60–90 min to adjust to their surroundings. Then, air with varying equivalent fractional compositions of inspired O₂ (Fi_O2) was delivered at a flow rate (V̇_C) through the box ranging between 5 and 10 l min⁻¹. Changes in Fi_O2 were produced by mixing nitrogen and air through a series of calibrated rotameters. Birds were exposed to 25 min step reductions in equivalent Fi_O2: ambient [0.21 at 0 m or 0.134 at 3200 m], 0.12, 0.09 and 0.07. For the respiratory trials, birds were exposed further to 0.05 Fi_O2. A 25 min recovery at ambient Fi_O2 followed the hypoxic exposures.

Whole animal V̇_O2 was calculated from V̇_C, Fi_O2 and the fractional O₂ composition of expired gas from the chamber (Fec_O2), which were directly measured by a gas analyzer (Sable Systems, Las Vegas, NV, USA). Water vapour was removed from the gas prior to analysis. Tidal volume and breathing frequency were measured from the head mask outflow using a pneumotachograph connected to a differential pressure transducer (Validyne, Northridge, CA, USA).

Mean arterial pressure and all respiratory variables were recorded to a computer using PowerLab data acquisition software (ADInstruments, Colorado Springs, CO, USA). Arterial blood pressure was continuously monitored throughout using a pressure transducer (Deltran, Utah Medical Products, Midvale, USA) connected to the brachial artery cannula. Strategic sampling of arterial and venous blood (0.4 ml per sample) occurred 15 min after exposure to each Fi_O2, as well as after 5 and 25 min in recovery. Any blood remaining after analysis was returned to the bird. Blood samples were immediately analyzed for partial pressures of O₂ and CO₂, O₂ content, Hb concentration ([Hb]), hematocrit (Hct), arterial pH (pH_a) and plasma ions including HCO₃⁻. Arterial O₂ content (Ca_O2; mmol l⁻¹) and venous O₂ content (Cv_O2; mmol l⁻¹) were determined at 41°C using the Tucker method (Tucker, 1967) with a FireSting O₂ probe (PyroScience, Aachen, Germany). The O₂ probe was calibrated with 0% O₂ (3 g l⁻¹ Na₂SO₃; Sigma-Aldrich) and water saturated with ambient air (21% at sea level and 12% at 3200 m) prior to each experiment. [Hb] (g dl⁻¹), Hct (%), arterial bicarbonate ion concentration ([HCO₃⁻]_a; mmol l⁻¹), Pa_O2 (kPa), arterial partial pressure of carbon dioxide (Pa_CO2; kPa), and arterial pH (pH_a) were analyzed from arterial blood at 41°C using CG8+ cartridges with the i-STAT VetScan Analyzer (Abaxis, Union City, CA, USA). All i-STAT values were corrected according to Harter et al. (2015). [HCO₃⁻]_a was calculated using the Henderson–Hasselbach equation, assuming a pK of 6.090 and a CO₂ solubility coefficient of 0.2117 mmol l⁻¹ kPa⁻¹ in plasma (Helbecka et al., 1964; Scott and Milsom, 2007).

With the exception of the blood variables, all data were acquired and analyzed using the PowerLab data acquisition and analysis software (ADInstruments, Colorado Springs, CO, USA) at a sampling frequency of 1000 Hz per channel. Mean values were derived for each variable for a 1–2 min period before each blood sample (e.g. after 12–15 min of each hypoxic Fi_O2 exposure, and after 3–5 min and 22–25 min of the normoxic recovery). Ca_O2 and Cv_O2 were acquired using software designed for the FireSting O₂ probes.

V̇_O2, air convection requirement and lung O₂ extraction were reported in terms of standard temperature and pressure and dry (STPD). Tidal volume was reported in terms of body temperature and pressure saturated (BTPS), assuming a constant body temperature of 41°C and taking into account changes in barometric pressure and air density at altitude (Dejours, 1975). V̇_R was reported in both BTPS and STPD for comparison.

We corrected the Hct and [Hb] data collected by the i-STAT VetScan Analyzer for all groups based on the calibrations derived for bar-headed geese in Harter et al. (2015).

Data are presented as means±s.e.m. unless stated otherwise. Within each species, all data were analyzed using one-way repeated measures analysis of variance (ANOVA) and Holm–Šidák post hoc tests. Comparisons between each species were made using two-way (species and Fi_O2) repeated measures ANOVA and Holm–Šidák post hoc tests within each Fi_O2. For statistical comparisons, P<0.05 was used to determine statistical significance. Variables analyzed with a one-way repeated measures ANOVA that did not meet assumptions for either normality or equal variance in barnacle geese were transformed with x′=ln(x) for Q̇, lung O₂ extraction, and blood convection requirement and with x′=x² for tidal volume. Similarly, variables were transformed when they did not meet assumptions for either normality or equal variance analyzed for a two-way repeated measures ANOVA (i.e. x′=ln(x) and x′=1/(1−x) for V̇_R, tidal volume and Pa_O2). Student's t-tests were used to compare Hct and [Hb] prior to and following the experiment to ensure that no blood dilution had been incurred throughout the experiment. Statistical analyses were carried out using SigmaStat (version 3.0; Systat Software).

Both bar-headed and barnacle geese maintained V̇_O2 during progressive hypoxia (Fig. 1A), with V̇_O2 increasing significantly during hypoxic exposure in LAR bar-headed geese (P<0.001), almost significantly in HAR bar-headed geese (P=0.051) and remaining unchanged in barnacle geese (P=0.72). The V̇_O2 of the HAR bar-headed geese was significantly lower than that of the LAR bar-headed geese at every level except 0.07 Fi_O2 (P=0.004).

The hypoxic ventilatory responses of each study group and the differences present in the relative contributions of breathing frequency and tidal volume to V̇_R are depicted in a Hey plot (Fig. 1B), a graphical depiction of breathing patterns (tidal volume and breathing frequency) at different levels of V̇_R (Guz and Widdicombe, 1970). At a given level of V̇_R, bar-headed geese breathed at a slower rate with significantly larger tidal volumes (Fig. 1B, Fig. S1A,B) than the barnacle geese, and this difference in pattern was sustained in hypoxia. The increase in V̇_R was greatest in the HAR bar-headed geese and was lowest for the barnacle geese (Fig. 1C). V̇_R of HAR bar-headed geese was higher than that of barnacle geese during every exposure (P<0.001). LAR bar-headed geese trended towards having a larger V̇_R than barnacle geese at 0.05 Fi_O2, but this did not reach statistical significance (P=0.057). The air convection requirement – the ratio of V̇_R to  V̇_O2 – increased in hypoxia in all groups, significantly so in both LAR groups (Fig. 2A). Lung O₂ extraction, which is the percentage of the inspired O₂ extracted from inspired gas, increased initially in both LAR groups between 0.21 and 0.12 Fi_O2 (Fig. 2B), and then remained constant at 30–50% beyond 0.12 Fi_O2. All differences in both the resting levels of V̇_R between HAR and LAR bar-headed geese disappeared when our data were expressed as STPD rather than BTPS (Fig. 1D).

Hct and [Hb] were not significantly different among HAR bar-headed geese (Hct: 38.8±2.8%, [Hb]: 117.3±7.0 g l⁻¹), LAR bar-headed geese (Hct: 43.9±4.3%, [Hb]: 125.1±7.5 g l⁻¹), or LAR barnacle geese (Hct: 43.1±1.9%, [Hb]: 112.6±4.0 g l⁻¹). In all three groups of geese, Hct and [Hb] were unchanged during progressive hypobaric hypoxia.

Pa_O2 decreased with progressive decreases in Fi_O2, and was lower (P<0.001) in the bar-headed geese than in barnacle geese at or below an Fi_O2 of 0.07 (Fig. S2A). Ca_O2 was similar between the groups of geese and decreased with hypoxia (Fig. S2B). Plotting Ca_O2 as a function of Pa_O2 generated in vivo O₂ equilibrium curves (Fig. 3) that are representative of the arterial saturation given the prevailing acid-base conditions that accompanied hypobaric hypoxia (see Table 1). Differences in these O₂ equilibrium curves reflect the higher O₂ affinity of the bar-headed goose blood (Black and Tenney, 1980; Weber et al., 1993) compared with that of barnacle geese and differences in the pH_a at each Fi_O2 (Fig. 4).

The starting pH_a of the HAR bar-headed geese was higher than that of the LAR bar-headed geese (P<0.001) and was accompanied by higher starting [HCO₃⁻] (P=0.009) (Fig. 4A,B). Pa_CO2 decreased (P<0.001) and pH_a increased (P<0.001) in all three groups of geese during progressive hypoxia (Fig. 4, Table 1). Both bar-headed goose study groups experienced a respiratory alkalosis during hypoxic exposure down to 0.07 Fi_O2 (Fig. 4A,B). At that point, LAR bar-headed geese were recovered to normoxia (Fig. 4A), whereas HAR bar-headed geese were further exposed to 0.05 Fi_O2 (Fig. 4B). Between 0.07 and 0.05 Fi_O2 (Fig. 4B), the pH_a of HAR bar-headed geese remained unchanged, but [HCO₃⁻]_a decreased significantly (Fig. 4B), indicative of a metabolic acidosis. This also occurred in the barnacle geese, but the metabolic acidosis was triggered at a less extreme level of hypoxia (0.07 Fi_O2) (Fig. 4C). Intriguingly, only during ambient recovery did pH_a fall significantly in any group. LAR bar-headed geese recovered their pH_a within 5 min of normoxia after being exposed to 0.07 Fi_O2. The HAR bar-headed geese and barnacle geese that were exposed to 0.05 Fi_O2 both had a persistent acidosis and low [HCO₃⁻]_a even after 25 min of recovery (Fig. 4B,C).

All three groups of geese increased Q̇ by 2- to 3-fold, yielding similar maximum values during progressive hypoxia (Fig. 5A). The increase in Q̇ became significant at a different Fi_O2 between the groups: 0.09 in HAR bar-headed geese, 0.07 in LAR bar-headed geese and 0.05 in barnacle geese (Fig. 5A). The relative contributions of increases in heart rate and stroke volume to Q̇ are depicted in a cardiac equivalent of a Hey plot for individuals at all exposures (Fig. 5B). Increases in stroke volume accounted for most of the increase in Q̇ in barnacle geese and LAR bar-headed geese (Fig. 5B,C) - their heart rates increased only modestly (Fig. 5D) whereas stroke volume roughly doubled (Fig. 5C). This was also the case initially in the HAR bar-headed geese; however, at 0.07 Fi_O2 heart rate increased substantially, with an associated decrease in stroke volume. The net overall result, however, was a trend for Q̇ to increase earlier and more rapidly in bar-headed geese than barnacle geese, and more so in the HAR bar-headed geese than in the LAR bar-headed geese (Fig. 5A).

While the increase in Q̇ of HAR bar-headed geese was triggered at a higher Fi_O2, it is evident from Fig. 6A, which plots the changes in cardiac variables as a function of Pa_O2, that this response was associated with this group of geese having a lower Pa_O2 at any given Fi_O2 during hypoxia (Fig. S2A). Significant increases in Q̇ occurred at the same Pa_O2 in all three groups of geese (∼6 kPa). That this reflects the differences in the Hb–O₂ equilibrium curves is clear from the extent of the overlap when Q̇ is plotted as a function of Ca_O2 (Fig. 6B). Similarly, significant changes in the contributions of stroke volume and heart rate to Q̇ occurred at ≤6 kPa (Fig. 6C,D).

Neither blood convection requirement (the quotient of Q̇ and V̇_O2) nor tissue O₂ delivery (the product of Q̇ and Ca_O2) changed significantly in any of the three groups of geese during progressive hypoxia (Table 1). The percentage of the O₂ extracted from arterial blood fluctuated between 30 and 50% and also did not change significantly either during progressive hypoxia or between any of the three groups of geese (Table 1).

While Q̇ increased during hypoxic exposure, mean arterial pressure was generally maintained, decreasing minimally at 0.07 Fi_O2 in the two LAR groups of geese (Table 1).

In this study we comprehensively compared the metabolic, ventilatory and cardiovascular responses of HAR bar-headed geese with those of LAR bar-headed geese. In addition, we compared these responses to those of barnacle geese, a member of a closely related genus, to provide further insight into responses unique to bar-headed geese. We found that HAR bar-headed geese exhibited a reduced V̇_O2 compared with LAR bar-headed geese. When exposed to progressive hypoxia, HAR bar-headed geese exhibited a modestly increased HVR and initiated cardiac responses earlier than LAR bar-headed geese, supporting our initial hypothesis. However, the magnitude of these differences was not as large as those described for bar-headed geese during short-term acclimation to simulated high altitude (Black and Tenney, 1980). Explanations for these differences are discussed below.

While it cannot be determined with absolute certainty the extent to which the differences present between the HAR and LAR bar-headed geese can be attributed exclusively to differences in barometric pressure during rearing, many potentially confounding variables were controlled across study groups. All groups of birds were held in outdoor pens with access to indoor shelter during the winter. All groups were healthy, fed similar diets and had been housed for at least a year without flying. While body mass and body composition have been shown to vary seasonally in barnacle geese (Portugal et al., 2007), the relationship between heart rate and V̇_O2, when normalized for body mass, was unaffected in five of the six seasonal sampling periods and was also unaltered by molt (Portugal et al., 2009). Furthermore, many of the variables measured in this study have been previously measured on LAR bar-headed geese and barnacle geese, allowing us to compare results. Resting values for V̇_O2 and V̇_R in our LAR bar-headed geese in normoxia were comparable to those previously described in the literature (Black and Tenney, 1980; Fedde et al., 1989; Hawkes et al., 2014; Scott and Milsom, 2007). Our values for Ca_O2, Cv_O2, Hct and [Hb] also fell within the range of values previously reported in the literature (Black and Tenney, 1980; Fedde et al., 1989; Hawkes et al., 2014; Scott and Milsom, 2007). Literature values for Q̇, stroke volume and heart rate of LAR bar-headed geese in normoxia vary widely. Fedde et al. (1989) reported a high heart rate and low stroke volume, while Hawkes et al. (2014) reported a low heart rate and high stroke volume. Our values fall midway between the two.

V̇_O2 was significantly lower in the HAR bar-headed geese compared with LAR bar-headed geese, suggesting that rearing at altitude leads to a reduction in metabolism and in the demand for O₂. All groups maintained or increased V̇_O2 when exposed to hypoxia (Fig. 1A). The small increases seen in V̇_O2 in all groups may reflect an increased cost of ventilation and associated events. The net response suggests that the cardiorespiratory adjustments were sufficient to match O₂ supply to O₂ demand at all but the most severe levels of hypoxia.

There was evidence of a metabolic acidosis, indicative of recruitment of anaerobic metabolism, in barnacle geese at 0.07 Fi_O2 and in HAR bar-headed geese at 0.05 Fi_O2. On return to control conditions, the HAR bar-headed geese recovered to control acid-base status faster (within 25 min) than the barnacle geese. Both groups exposed to 0.05 Fi_O2 experienced a significant decrease in pH_a upon recovery to normoxia (Fig. 4B,C). This may reflect the sequestering of lactate and H⁺ during hypoxia that was rapidly released into the blood upon return to resting conditions. Because LAR bar-headed geese were recovered to ambient conditions after breathing 0.07 Fi_O2 rather than 0.05 Fi_O2, we cannot ascertain whether the ability of bar-headed geese to avoid metabolic acidosis until more severe levels of hypobaric hypoxia is an adaptation or a consequence of high-altitude rearing. Nevertheless, the ability to recover quickly from severe hypoxia would be an asset during the high-altitude migration of bar-headed geese.

The magnitude by which V̇_R increased during progressive hypoxia in the LAR bar-headed geese and barnacle geese of our study was within the range reported in previous studies (Black and Tenney, 1980; Fedde et al., 1989; Hawkes et al., 2014; Scott and Milsom, 2007). In addition, as previously reported, bar-headed geese exhibited a higher overall tidal volume and lower breathing frequency at any given V̇_R than barnacle geese – a pattern hypothesized to be a more effective breathing pattern that reduces effective dead space ventilation (Scott and Milsom, 2007). The demonstration that this breathing pattern was common to LAR and HAR bar-headed geese supports the suggestion that this is an adaptation specific to bar-headed geese. Furthermore, we found that the increase in V̇_R in hypoxia was greatest in magnitude in the HAR bar-headed geese and lowest in the barnacle geese (P<0.001; Fig. 1C). These findings support our hypothesis and are consistent with previous findings of a greater increase in V̇_R in bar-headed geese following short-term acclimation to simulated altitude (Black and Tenney, 1980).

Differences in both the resting levels of ventilation and the HVR disappeared when our data were expressed as STPD rather than BTPS (Fig. 1D, Fig. S3B). Expressing volume as a function of STPD reveals the molar amount of air (and thus O₂) moved. In this instance, STPD values are not significantly different for either LAR or HAR bar-headed geese, indicating that the differences in ventilation reported in BTPS were due to the thinner air.

Expressing volumes as a function of BTPS, however, is standard for showing how much gas an animal ventilates. Black and Tenney found that the resting ventilation in bar-headed geese acclimated under hypobaric conditions (equivalent to 5640 m) for 4 weeks was approximately double that of sea level-acclimated birds when measured at similar levels of Pa_O2 under normobaric conditions. In the present study, while the level of V̇_R at an inspired partial pressure of O₂ of 12 kPa (Fi_O2=0.21 at 3200 m; sea level equivalent Fi_O2=0.12) was roughly 30% higher in the HAR bar-headed geese relative to the LAR bar-headed geese, this difference was not significant. This suggests that despite apparent similarities, the changes seen following short-term acclimation (Black and Tenney, 1980) are more akin to ventilatory acclimatization to hypoxia, while those seen in the HAR bar-headed geese appear to reflect hypoxic desensitization (Powell et al., 1998).

Ventilatory acclimatization to hypoxia is defined as the further increase in ventilation, compared with the rapid initial response, which occurs over hours to days of acclimatization (Powell et al., 1998). This secondary increase has been ascribed to plasticity in O₂ sensing by the carotid body chemoreceptors and in central integration of chemoreceptor input (Powell, 2007). Over many months at high altitude, however, this hypoxic ventilatory response can be gradually attenuated by hypoxic desensitization (Brutsaert, 2007; Powell et al., 1998). While the increases in breathing during ventilatory acclimatization to hypoxia improve O₂ uptake, hypoxic desensitization could be representative of longer-term high-altitude exposure and the ability to effectively transport O₂ without magnified convective transport. This would help reduce respiratory water loss, and reduce the metabolic cost of breathing (Powell, 2007; Storz et al., 2010). Despite the apparent hypoxic desensitization, V̇_R remained elevated relative to V̇_O2 in the HAR bar-headed geese, indicative of a reduction in lung O₂ extraction (Fig. 2).

Differences between species in blood-O₂ carrying capacity were driven primarily by differences in intrinsic O₂ affinity (i.e. P₅₀ of Hb) and in vivo blood O₂ affinity (i.e. blood O₂ affinity subject to in vivo changes in temperature, pH and allosteric modulators). There were no inter- or intraspecies differences in [Hb] or Hct either in normoxia or with progressive hypoxia. The greater O₂ affinity of the HbA isoform of bar-headed geese is well documented (Weber et al., 1993), although the properties of the HbD isoform have yet to be studied. Neither Hct nor [Hb] in bar-headed geese were affected by rearing altitude, a finding also reported after short-term (4 weeks) acclimation of bar-headed geese to simulated altitude (5640 m) (Black and Tenney, 1980). This is also similar to patterns described in high-altitude-acclimatized Tibetan humans (Simonson et al., 2015) and deer mice (Lui et al., 2015). As a result of the differences in intrinsic and in vivo blood O₂ affinity, however, at any given Pa_O2 bar-headed goose blood will be more saturated than that of barnacle geese (Fig. 3). Furthermore, the data suggest that at any given Pa_O2, the blood of the HAR bar-headed geese would be slightly more saturated than that of the LAR bar-headed geese (Fig. 3). These small differences are most likely explained by the higher levels of [HCO₃⁻]_a (P=0.009) at 0.12 Fi_O2 and pH_a (P<0.001) at all levels of hypoxia in the high-altitude study group. Such an alkalosis would left-shift the O₂ equilibrium curve and enhance O₂ loading at the lung, which are possibly another features of high-altitude rearing. A respiratory alkalosis occurred in all groups during progressive hypoxic exposure due to heavy ventilation, further enhancing blood-O₂ carrying capacity (Fig. 4).

Based on the in vivo O₂ equilibrium curves, the blood P₅₀ of the two groups of bar-headed geese are unlikely to be appreciably different. While birds have organic phosphates (inositol pentophosphate) for altering the O₂ affinity of Hb, there is little evidence of an IPP-induced change in P₅₀ with high-altitude exposure (Weber, 2007). This suggests that isoHb switching did not occur in response to environmental hypoxia. Although large reversible changes in blood P₅₀ could be achieved by altering the expression levels of HbA and HbD, our data suggest that bar-headed geese do not do this. Similar results have been reported for high- versus low-altitude hummingbirds (Projecto-Garcia et al., 2013), sparrows (Cheviron et al., 2014), house wrens (Galen et al., 2015) and waterfowl (Natarajan et al., 2015).

All groups in the present study increased Q̇ 2.0- to 2.5-fold during severe hypoxia. Previous reports of the magnitude and direction of change in Q̇ during severe hypoxia in bar-headed geese vary widely. Hawkes et al. (2014) reported that, in bar-headed geese breathing 0.07 Fi_O2, Q̇ decreased by ∼20%, whereas Fedde et al. (1989) and Black and Tenney (1980) reported no change in Q̇ at this level of hypoxia. However, when Black and Tenney exposed their birds to a further reduction in O₂ to ∼0.05 Fi_O2, corresponding to a Pa_O2 of ∼3.5 kPa, Q̇ increased by a remarkable 7-fold (Black and Tenney, 1980). These differences most likely reflect the steepness of the exponential cardiovascular response curve beyond the inflection point and small differences in Pa_O2. Analysis of the data based on Fi_O2 suggests that increases in Q̇ in the present study were initiated first in the HAR bar-headed geese (0.09 Fi_O2), next in the LAR bar-headed geese (0.07 Fi_O2) and last in the barnacle geese (0.05 Fi_O2) (Fig. 5A). When expressed as a function of Pa_O2 (or Ca_O2), however, all groups produced significant increases in Q̇ at a similar Pa_O2 of ∼6 kPa (Fig. 6A, Fig. S3C), indicating that the differences in which Q̇ increases are initiated when plotted as a function of Fi_O2 reflect differences in the blood O₂ affinity. Black and Tenney made a similar observation. They too noted that the differences they saw in the changes in Q̇ during progressive hypoxia following short-term acclimation in bar-headed geese could be accounted for by differences in Ca_O2 (Black and Tenney, 1980). Unlike Black and Tenney, however, we did not see an increase in the overall magnitude of Q̇ at a given Pa_O2 with lifelong exposure to high altitude. Under resting conditions, Q̇ does not differ in high- versus low-altitude native or domestic mammals either. Total blood flow has been found to be unaltered or slightly reduced in humans, alpaca, llama, rats and wild mice living at altitude (Banchero et al., 1971; Klausen, 1966; Monge et al., 1955; Sillau et al., 1976; Turek et al., 1973). Thus, the differences present between the study of short-term acclimatization by Black and Tenney (1980) and of high-altitude rearing in our study not only suggest that both acclimatization to hypoxia and hypoxic desensitization occur with ventilatory responses, but also that similar acclimatization and desensitization to hypoxia occur with regard to blood flow in bar-headed geese, and may act to reduce the costs of convective transport of blood as they do for respiratory gases.

Previous studies on geese reported changes in heart rate as the primary contributor to changes in Q̇, with stroke volume remaining largely unchanged (Faraci, 1986; Fedde et al., 1989; Smith et al., 2000). In the present study, this was true of the low-altitude groups down only to an Fi_O2 of 0.12. Below this, all groups increased Q̇ during progressive hypoxia down to a Pa_O2 ∼6 kPa (Fig. S3C) more by increasing stroke volume than heart rate (Fig. 5C and Fig. 6C). HAR bar-headed geese were the only group in which Pa_O2 fell <6 kPa, at which point heart rate increased substantially (Fig. 6D) associated with a decrease in stroke volume (Fig. 6C). The large increases in stroke volume seen in the present study could have been mediated either by extrinsic factors (circulating hormones or neurotransmitters) or intrinsic factors [cardiac muscle fiber contractile properties associated with the Frank–Starling response – the relationship between cardiac contractility and venous return (Smith et al., 2000)]. As outlined in Shiels and White (2008), limited information exists on the Frank–Starling response in avian cardiomyocytes, although it is known to facilitate large increases in stroke volume during hypoxia in fish (Farrell, 1991; Shiels and White, 2008). Further studies are required to determine the underlying mechanisms of this response.

The primary differences present between the HAR and LAR bar-headed geese were ventilatory and metabolic in nature. But, at this point, we cannot discern the differential effects of phenotypic plasticity (i.e. acclimatization) from developmental plasticity on the physiology of this species. The reduction in resting V̇_O2 was one of the most significant differences observed in the HAR bar-headed geese. We also observed an increase in resting V̇_R and in the HVR that could be explained by the differences in barometric pressure at which the measurements were made. Even taking this into account, however, HAR bar-headed geese still exhibited a large air convection requirement (ratio of V̇_R to V̇_O2), compensating for a reduction in lung O₂ extraction (Fig. S3A). This may help to maintain blood acid–base balance at the expense of O₂ uptake.

All geese increased Q̇ by ∼2-fold to a similar overall magnitude, but Q̇ increased earlier and more rapidly in bar-headed geese than barnacle geese as environmental O₂ fell, and more so in the HAR bar-headed geese than in LAR bar-headed geese. However, this could be explained by the differences in in vivo blood O₂ affinity. All groups increased perfusion at a similar Pa_O2 during hypoxic exposure. An unexpected finding was the prominent role of increases in stroke volume in increasing Q̇ in all groups, including the barnacle geese. Further studies are required to determine the underlying mechanisms of the differences reported here between HAR bar-headed geese and LAR bar-headed geese, and the extent to which these differences may also facilitate high-altitude flight.

We thank J. Meir, M. Pamenter, M. Qu, E. J. Ross, J. York, and J. You for their assistance in performing these experiments. We also thank Dr G. R. Scott for his insights concerning these data. We express our gratitude to Dr Yang Zhong of Fudan University for assisting us in logistical arrangements in the field, as well as to the Lake Qinghai Wildlife Conservation Office for their field and lab assistance. We thank the staff at the Centre for Comparative Medicine at the University of British Columbia for their care of the low-altitude experimental animals, as well as B. Gillespie and V. Grant at the University of British Columbia for building the experimental apparatus. Finally we would like to thank the anonymous reviewers whose comments greatly helped shape this manuscript.