Cardiorespiratory adjustments of homing pigeons to steady wind tunnel flight

Flying vertebrates are of interest to cardiovascular physiologists because their aerobic metabolic capability is about twice that of running mammals of equal size (Thomas et al., 1987). Because homing pigeons Columba livia combine a relatively small body size (allometric variation) with superior athletic ability (adaptational variation) and tractable behavior,they are an excellent animal model for investigating the upper limits of vertebrate cardiorespiratory adaptation and performance during intense metabolic stress.

Over the years, much has been learned about the many aspects of the cardiorespiratory responses of birds to treadmill exercise (Bevan et al., 1994, 1995) and resting hypoxic stress (Maginniss et al.,1997; Novoa et al.,1991). Other studies have investigated various aspects of avian flight physiology that include metabolism and biochemistry(Christensen et al., 1994; George and John, 1993; Schwilch et al., 1996),thermoregulation and water balance (Adams et al., 1997; Carmi et al., 1993, 1994; Hissa et al., 1995; Giladi et al., 1997), and wing cycle and ventilation (Boggs et al., 1997a,b). However, only one in-depth cardiorespiratory study on an avian species during steady flight has been reported (Butler et al., 1977). Consequently, this central aspect of avian physiology remains poorly understood.

Two reasons for the scarcity of detailed avian flight cardiorespiratory data are the need for specialized equipment (e.g. a wind tunnel) and chronic vascular cannulation techniques, which provide access to the bird's circulatory system while allowing normal behavior of the bird. Accordingly,the two primary objectives of this study were: (1) to develop chronic cannulation techniques and blood sampling procedures for the pigeon in order to routinely obtain arterial and mixed venous blood samples from both resting and flying birds, and (2) to comprehensively characterize the basic cardiorespiratory parameters of our population of homing pigeons at calm rest and during steady wind tunnel flight under defined conditions.

Homing pigeons Columba livia L. to be investigated were of unknown sex and obtained from Dr Melvin Kreithen's outdoor loft at the University of Pittsburgh; mean body mass (± s.d.) was 340±45 g. The birds were housed in individual indoor cages at 21°C under a 12 h:12 h (light:dark) photoperiod with ample commercial pigeon food and water. All data reported in this study were collected between 09:00 h and 01:00 h.

The birds were first screened to determine whether they were behaviorally acceptable for use in the resting experiments. Screening involved placing the bird inside an open-circuit metabolic chamber (described below) for 1 h while its rate of carbon dioxide production was continuously recorded. Quiet,resting birds produced little variation in carbon dioxide production. These birds were chosen as potential candidates for resting studies. Birds passing this screening test were further familiarized to the chamber during three 1 h sessions on consecutive days before being scheduled for surgery.

After an initial screening for wind tunnel flight competence, birds showing promise were included in the daily training schedule. Each bird was trained to fly in a horizontal wind tunnel for 7 min at a constant air speed of 18.4 m s^–1 and an ambient air temperature of 10–12°C,while carrying the devices required for a specific experiment (e.g. metabolism mask and tube, blood pressure transducer, or pieces of heavy string simulating cannula extension lines). The trained bird was then scheduled for surgery.

Resting measurements were made from each bird as it sat quietly in a darkened, open-circuit Plexiglas^® metabolic chamber (40 cm×20 cm×20 cm L:W:H). This chamber was placed inside an environmental chamber set at 15°C to prevent thermal hyperventilation during the experiment. The environmental chamber kept the internal temperature of the metabolic chamber at 18.0±0.9°C, as monitored by a YSI Scanning tele-thermometer (Model 47, YSI, Yellow Springs, OH, USA) and thermistor probe (Model 427).

All flight experiments were performed in a 6.6 m long open-circuit horizontal wind tunnel (Aerolab, Laurel, MA, USA) with test section dimensions and air flow properties that have been described elsewhere(Thomas et al., 1984). All training and each wind tunnel experiment was carried out at an air speed of 18.4 m s^–1, and an ambient temperature of 10–12°C to keep the bird from overheating.

Surgery was performed on the day before an experiment using 2% Halothane anesthetic metered by a vaporizer (Flurothane, Model FR, Pittsburgh, PA, USA)and carried in a 2:1 ratio of oxygen and nitrous oxide. The total flow rate through the vaporizer was 1.5 l min^–1. Mixed venous blood samples were obtained from a silastic cannula (0.030 cm i.d.×0.064 cm o.d.; Dow Corning, Midland, MI, USA) placed in the bird's right atrium through the cervical cutaneous branch of the jugular vein. The position of the cannula's tip in the right atrium was verified by autopsy after completion of the experiments. Data from birds having the cannula tip located outside the right atrium were discarded. Arterial blood samples were obtained from a second silastic cannula (Dow Corning, 0.051 cm i.d.×0.094 cm o.d.)placed in the bird's right ischiadic artery. An 80% solution of polyvinylpyrrolidone (PVP-40T; Sigma, St Louis, MO, USA) in aqueous heparin[1000 U ml^–1; Elkins-Sinns, Inc., Cherry Hill, NJ, USA(hereafter called `PVP–heparin')] was used to maintain cannulae patency prior to an experiment.

Resting studies. Conventional open circuit methods and equation 2 from Tucker (1968) were used to determine the oxygen consumption rate(V̇_O2) of each bird. A mass flow controller (Model FM-4587, Linde-Union Carbide, Danbury, CT,USA) maintained a constant air flow rate of 0.80±0.01 l min^–1 (stpd) through the metabolic chamber. The composition of gas entering and leaving the metabolic chamber was continuously sampled by a pump (Model R-2, Applied Electrochemistry, Pittsburgh, PA, USA)and monitored by oxygen and carbon dioxide analyzers (Applied Electrochemistry, Models S-3A-N-22M and CD-38, respectively). Analyzers were calibrated with dry air and a carbon dioxide (5.00% CO₂ in nitrogen) Primary Standard grade gas mixture (Linde-Union Carbide, Jackson Welding, Pittsburgh, PA, USA). Individual oxygen consumption rates were calculated from the mean values of the oxygen and carbon dioxide gas analyzers while blood samples were being drawn.

Flight studies. The rate of oxygen consumption for five pigeons was determined using the open-circuit method and flight training procedures similar to those previously described by Thomas et al.(1984). Briefly, the birds were trained to fly in the wind tunnel for 7 min wearing a custom-fit celluloid mask and molded latex hood. A vacuum source was connected to the mask by a flexible tube that trailed below the flying bird and pulled room air through the mask at a rate of 13.1 l min^–1 (stpd). Composition of the expired gas was continuously measured as described above,and the oxygen consumption rate was calculated using mean data from the sixth minute of each flight.

To detect if leakage of expired gas occurred from the mask at the flow rate that was to be used for the flight oxygen consumption experiments, several lower flow rates were used during preliminary tests on each bird, and the rate of oxygen consumption(V̇_O2) calculated. These tests revealed that no unintended mask leakage occurred (i.e. there was no decrease in the bird's apparent V̇_O2) until the flow rate was reduced to 16% below the flow rate used in the actual V̇_O2 flight experiments.

The flight mass-specific oxygen consumption rate(V̇_O2kg^–1) of each homing pigeon used in the cardiovascular series of measurements was calculated from the mean mass-specific oxygen consumption rate measured from a different but equivalent group of homing pigeons(N=5). These were separate experiments carried out during the same times of the day (09:00 h–11:30 h), months of the year (summer) and wind tunnel conditions used for the cardiorespiratory measurements (i.e. horizontal flight at 18.4 m s^–1 and 10–12°C). This two-group approach was used to increase wind tunnel training success. Switching back and forth between normal flight and training a given pigeon to fly with the mask(oxygen consumption measurements) and the pressure transducer back pack (heart rate measurements) confused most birds, and resulted in substantially lower training success rates.

Resting and flight studies. Body temperature was measured to the nearest 0.1°C by inserting a thermistor probe (YSI, Model 427) 5 cm into the bird's cloaca. The probe was calibrated immediately before each experiment using a thermostatically controlled water bath (Isotemp Bath Model 8000,Fisher, Hanover Park, IL, USA) and a precision mercury thermometer (Fisher,15-000A).

Resting and flight studies. Arterial (a) and mixed venous(v̄) blood samples were analyzed for oxygen tension (Pa_O2, Pv̄_O2) carbon dioxide tension (Pa_CO2, Pv̄_CO2) and pH(pHa, pHv̄) using a blood gas analyzer(Instrumentation Laboratories Model 1306, Lexington, MA, USA). Total oxygen content (Ca_O2, Cv̄_O2) for the arterial and mixed venous blood samples was measured using a Model K analyzer(Lex-O₂-Con, Chestnut Hill, MA, USA). Oxygen and carbon dioxide tensions and pH values were corrected using the measured cloacal temperature and equations supplied in the blood gas analyzer manual. Blood samples were stored in glass syringes (Model 1750-LTN, Hamilton Gastight^®,Reno, NV, USA) and chilled in an ice–water slurry to quench the red cells' metabolism before analysis. Hematocrit (Hct) was measured for each blood sample to determine if contamination had occurred by saline admixing. Arterial and mixed venous blood gas tensions and pH analyses were completed within 5 min after the sample was collected. Total oxygen content analyses were completed within 7 min after the sample was collected.

Resting studies. Heart rate (fh) was recorded during the blood sampling period with silver cutaneous electrodes connected to a heart monitor equipped with a chart recorder (Model 1700, Mennen Medical,Trevose, PA, USA).

Flight studies. Heart rate during flight was determined from the arterial pressure waveform measured with a small (3.0 cm×4.2 cm×1.3 cm W:L:H) 9.4 g pressure transducer (NAMIC, Glens Falls, NY, USA)attached to the bird's back with Velcro™ patches. The transducer's output was recorded by a MacLab analog-to-digital converter controlled by a Macintosh computer running MacLab Chart software.

On the day of the experiment, the venous cannula was cleared of the PVP–heparin solution, and the bird systemically heparinized with 0.3 ml of heparinized saline (50 U ml^–1) through the venous cannula. The venous cannula was then attached to a saline-filled extension line (0.051 cm i.d.×0.150 cm o.d.×90 cm L; S-54-HL, Tygon, Akron, OH, USA)connected to one port of a saline-filled three-way manifold. A Tuberculin syringe used to remove saline from the extension line before collecting the blood sample, and a Hamilton syringe used to collect the blood sample, were each attached to the remaining two manifold ports by 1.3 cm lengths of silastic tubing (Dow Corning, 0.051 cm i.d.×0.094 cm o.d.). The arterial cannula was prepared the same way, except that 0.3 ml of saline was used in place of the heparinized saline. After the electrocardiographic leads were attached to the bird it was placed inside the metabolic chamber that was placed inside the environmental chamber.

After a 45 min equilibration period, heart rate and carbon dioxide readouts were monitored for stability to determine an appropriate time for blood sample collection. Just before collecting the blood samples, the extension lines were simultaneously cleared of saline using the Tuberculin syringes. The silastic segments of the Tuberculin syringes were clamped and a 0.5 ml sample of arterial and mixed venous blood was collected simultaneously in each Hamilton syringe over 3 min while heart rate and metabolic rate information were simultaneously recorded. Immediately after collecting the blood samples, the extension lines were clamped near the manifolds, the silastic tubing segments on the Hamilton syringes were clamped and separated from the manifolds, then immersed in an ice–water slurry to quench the red cells' metabolism. The 3 min sampling period was chosen because preliminary tests revealed that faster blood sampling rates were detected by the birds as indicated by higher heart rates and increased oxygen consumption values. Total oxygen content analyses were completed within 7 min after sample collection, and arterial and mixed venous blood gas tensions and pH were completed within 5 min after sample collection. Deep body temperature was measured within 1 min after sample collection was completed. Cardiac output(Q̇) and stroke volume(Vs) were calculated for each experiment using the Fick equation.

Blood sampling and blood gas measurements. For these experiments,the arterial and venous cannulae were first prepared as described above. The three-way manifold, Tuberculin syringe, and the Hamilton glass sample syringe associated with each cannula's 90 cm long extension line were secured to a custom-built `dual-sampling device' that allowed all four syringes to be filled by using a single hand. By applying small clamps to the appropriate silastic tubing segments during the flight, a single researcher could first simultaneously clear saline from the lines into the Tuberculin syringes and then draw arterial and venous blood samples into the Hamilton glass syringes. During an experiment, the sampling device was held in a position 0.5 m downstream from the flying bird by a researcher lying on top of the wind tunnel with his arms protruding through snug-fitting arm holes in the ceiling of the wind tunnel's test section. Arterial and mixed venous blood (0.5 ml)was collected simultaneously in each Hamilton syringe during the sixth minute of flight. As soon as the bird landed on a perch lowered to signal the end of the flight, the extension lines were clamped near the manifolds, the silastic tubing segments on the Hamilton syringes were clamped and separated from the manifolds, and then submersed in an ice–water slurry to quench the red cells' metabolism prior to analysis. Blood gas tension and pH analyses were completed within 5 min after sample collection, and total oxygen content analyses were completed with in 7 min after sample collection. Deep body temperature was measured within 1 min after the bird landed by the previously described procedure. Cardiac output and stroke volume were calculated for each experiment using the Fick equation.

Heart rate measurements. Measurement of a bird's heart rate during flight was made on the day following the blood gas sampling experiment using the same wind tunnel conditions (horizontal flight at 18.4 m s^–1 and ambient temperature of 10–12°C). The PVP–heparin solution was removed from the arterial cannula and the bird systemically heparinized as previously described. The arterial cannula was then connected to the pressure transducer attached to the back of the bird with a Velcro^® square. The heart rate reported for each bird is the mean value calculated from data collected throughout the sixth minute of flight, to coincide with the same time in the flight when blood samples were collected.

After these measurements were completed, the bird was killed and the position of the tip of the venous cannula in the right atrium was checked. The heart was then excised, its chambers were open and blotted free of blood and the mass of the heart was (M_H) was determined to the nearest 0.01 g using a Mettler balance.

These tests were carried out for two reasons: (1) to learn the best method of preserving the blood gas tensions of whole blood samples during collection and temporary storage, and (2) to determine whether whole blood metabolism and/or tubing permeability were factors to consider in reporting final blood gas tensions and pH values.

Mixed venous pigeon blood (0.5 ml) was collected in a Hamilton syringe at room temperature (22°C) and a sample immediately injected into the blood gas analyzer for analysis. The Hamilton syringe was then attached to an extension line of equal length and composition used in the resting experiments. The extension line was filled with blood and its free end was clamped to prevent contaminating the blood with room air. Samples of blood from the free tip of the extension line were then injected into the analyzer every 5 min until the syringe was empty. This procedure was repeated at 12°C (i.e. the ambient temperature of the environmental chamber and wind tunnel). Oxygen tension data were plotted against time and the slope was calculated for both temperatures.

A 0.5 ml sample of mixed venous pigeon blood collected in a Hamilton syringe was immediately injected into the blood gas analyzer for assay. After sealing off the syringe's needle with a short segment of silastic tubing, the syringe was submersed in an ice–water slurry. Injections were repeated every 5 min until the syringe was empty. This procedure was carried out at the temperatures of 22 and 12°C, and with the Hamilton syringe submersed in the ice-water slurry (1°C) during injections. Oxygen tension results were plotted against time and the slope was calculated for each temperature.

Results from these preliminary tests are summarized in Table 1. As expected, lowering the temperature of the blood sample from 22°C to 1°C substantially decreased the metabolic rate of the whole blood samples as reflected in the slopes (Table 1). The similarity between the slopes of the Hamilton syringe and the extension line data at 22°C show that the effect of the extension line permeability on measured oxygen tension values was minimal during sample collection. The question then turned to influence of the whole blood's metabolic rate on oxygen tension during the collection and storage time before analysis.

During both resting and flight experiments, preservation of the oxygen tension of blood samples during collection is seen by the notable reduction of the ΔP_O2 rate in going from 22°C to 12°C (Table 1). This reduction in blood temperature takes place as the blood passes through the extension line, which acts as a heat exchanger and brings the blood into thermal equilibrium with the air of environmental chamber or wind tunnel,preserving the oxygen tension of each arterial and mixed venous blood sample used in our results.

Using an ice–water slurry for storage of blood samples prior to analysis proved to be the most effective method of preserving the oxygen tension of blood samples (Table 1).

Although not measured, we would expect similar drifts in the carbon dioxide tension and pH values of whole blood samples using the methods of collection and storage described here.

In summary, these preliminary tests indicated that under the sampling and temporary storage conditions used in this study, the influences of cannula extension line permeability and whole blood metabolism prior to analysis on the blood gas tensions and pH of analyzed samples was negligible. Accordingly,no attempt was made to correct the final measured values for these influences.

Our pigeons satisfied their 17.4-fold increase in V̇_O2 with a 7.4-fold increase in cardiac output (Q̇)and a 2.4-fold increase in blood oxygen extraction (E_B). Cardiac output was increased primarily by an increase in the heart rate(fh) (sixfold), and stroke volume (Vs)increased only modestly (Table 2). The similarity of the resting and flight Pa_O2 and Ca_O2values indicates a close matching of respiratory adjustments and metabolic requirements (Table 3). The fall in Pa_CO2 and the associated respiratory alkalosis that accompanied flight, however, indicate hyperventilation during flight (Table 3). These data suggest some exchange limitations for oxygen during flight that required the higher relative ventilation to maintain Pa_O2.

As indicated previously, only one in depth cardiorespiratory study has heretofore been reported for an avian species undertaking steady flight(Butler et al., 1977). Whereas Butler's group also investigated the pigeon, both the resting conditions and the flight speed they utilized differed substantially from those of the present study (see below). An in depth understanding of avian cardiorespiratory physiology is clearly central to advancing our knowledge of avian flight physiology and adaptation. Accordingly, we will begin our discussion by examining how closely the resting and flight data obtained from these two studies agree. In so doing, particular attention will be given to evaluating which data set for this `bird model' most closely approximates the cardiorespiratory adjustments of a pigeon flying in nature.

Aside from these two studies, all other cardiorespiratory studies on flying birds have been of limited scope in the sense that they did not directly measure all of the Fick equation parameters needed to fully assess a given animal's cardiorespiratory performance(Bishop, 1997; Bishop and Butler, 1995). To circumvent this problem, various researchers have combined data from incomplete studies on different avian species flying at disparate flight conditions in order to derive allometric equations for estimating the various Fick equation parameters for a generalized bird(Bishop, 1997; Bishop and Butler, 1995). In the second part of our discussion, we will evaluate how accurately Bishop's allometric equations (Bishop,1997) can estimate the cardiorespiratory parameters that we measured from our flying pigeons.

Because of their different evolutionary histories, birds and mammals use different cardiorespiratory designs and resources to satisfy their metabolic requirements. Nevertheless, the maximal aerobic capabilities(V̇_O2max) of flying birds and bats are essentially the same, but are about twice those of running mammals of similar size (Thomas,1975). This is because flying vertebrates combine the V̇_O2max-enhancing influences of small body size (allometric variation) and an athletic lifestyle(adaptational variation), whereas similar-sized non-flying mammals (e.g. mice and rats) have consistently adapted `non-athletic' lifestyles. Only after body size exceeds about 2 kg do examples of `highly athletic' running mammals begin to appear. Finally in our discussion, we consider what Fick parameter adaptations and adjustments distinguish our exercising pigeons from those of running mammals of various sizes. Particular attention will be given to how the heart-mass-specific stroke volumes, cardiac outputs and cardiac work capabilities of our flying pigeons compare with those of exercising mammals of different body masses and aerobic capabilities.

Resting data from the present study (`PS') are compared in Table 4 with corresponding pigeon data reported by Butler et al.(1977) (`B'). As is readily apparent from the PS/B ratios, the resting V̇_O2/M_b, Q̇/M_b, fh and Vs/M_bvalues of our pigeons are substantially lower than those reported by Butler. One important reason for this discrepancy is the difference in `resting'conditions that were used. While Butler's resting data were obtained from masked pigeons perched inside a wind tunnel, our data were collected from birds quietly resting in a darkened chamber to which they were previously familiarized. Consistent with this interpretation are the V̇_O2/M_band Q̇/M_b values reported by Grubb (1982) for pigeons resting at conditions similar to those of the present study. Grubb's values agree to within 3% and 12%, respectively with our values. A second probable reason for discrepancies in these PS/B ratios is that the V̇_O2/M_bvalue used by Butler in his 1977 study to calculate resting Q̇/M_b and Vs/M_b came from his earlier study(Butler, 1970), which used resting conditions different from those in his later study. Finally, the different hematocrit values measured in these two studies are an important factor accounting for the observed Ca_O2 and Cv̄_O2discrepancies, as indicated by the similarity of the PS/B ratios for Hct, Ca_O2, Cv̄_O2 and Ca_O2–Cv̄_O2(Table 4). The hematocrit value reported by Butler's group is substantially lower than the weighted mean value of 55.2% that we calculated from data reported for a total of 293 pigeons by five different groups of investigators(Gayathri and Hegde, 1994;Kalomenopolou and Koliakos, 1989; Viscor et al., 1985; Bond and Gilbert, 1958; Kaplan, 1954). In summary,differences in resting conditions used, methodologies concerning the source of the oxygen consumption value used in the Fick equation, and very different hematocrit values, appear to be primary factors contributing to the observed PS/B ratio discrepancies observed in these two studies (Tables 4 and 5).

The flight V̇_O2/M_bof our pigeons was about 1.6 times that of Butler's pigeons(Table 4). A reason for this difference could be the substantially different wind tunnel air speeds used in the current study (18.4 m s^–1) and Butler's study (10 m s^–1). Most birds, however, show little change in oxygen consumption over the range of air speeds at which they fly in a wind tunnel(Ellington, 1991). Grippler and Grivuni strains of pigeons show only a 10% change in V̇_O2/M_bover a range of wind tunnel flight speeds from 8 to 14 m s^–1(Rothe et al., 1987). These investigators reported a V̇_O2/M_bvalue of 310 ml (kg min)^–1 for their pigeons flying at 10 m s^–1, a value 1.6 times greater than Butler's value despite the same air speed, but essentially the same as that measured from our pigeons flying at 18 m s^–1.

Concerning flight speed, pigeons have been reported flying in nature at speeds of 16.1 m s^–1(LeFebvre, 1964), 18.3 m s^–1 (Skutch,1991), 19.1 m s^–1(Polus, 1985), and 19.7 m s^–1 (Levy,1986). Record-holding homing pigeons have achieved average flight speeds of 22 m s^–1 during long distance races in calm air(Levy, 1986). Thus, the air speed used in the current study (18.4 m s^–1) is within the range of air speeds that pigeons normally fly at in nature. The highest speeds that unencumbered Grippler pigeons will fly at in a wind tunnel is 18 m s^–1 (Rothe and Nachtigall, 1987). Our birds, however, were equipped with a mask and associated gas sampling tubes (metabolic determinations), or pressure transducers (heart rate determinations), or cannula extension lines (blood sampling), which contributed extra drag during these measurements. Based on determinations from other species of flying vertebrates(Thomas, 1975), we assume that the metabolic rates of our pigeons may have been about 10% to 15% higher than those expected during unencumbered flight at the same air speed, but may approximate the metabolic requirements of unencumbered flight at the higher speeds that these bird sometimes achieve in nature(Levy, 1986).

There are reasons to believe that the flight V̇_O2 value reported in Table 2 for our pigeons is near the maximal value that these birds are capable of achieving in nature. The flight V̇_O2 of our birds is only 10% less than the highest specific V̇_O2 that Rothe et al. (1987) were able to measure from their somewhat smaller sized Grippler and Grivuni strains of pigeons during flight at their maximum encumbered speed (14 m s^–1). Also, most of our pigeons would only fly steady for a maximum period of about 10 min while carrying the various types of measuring and blood sampling equipment. Accordingly, we will hereafter assume that our homing pigeons were operating either very close to, or at their V̇_O2maxcapability, and the flight cardiorespiratory data we obtained represent the maximal, or very near the maximal, values that these birds are capable of.

Despite the substantially higher flight V̇_O2/M_bof our pigeons, the flight fh, Vs/M_b and thus, the specific Q̇ of our birds, are generally similar to those of Butler's birds (Table 4). This relationship appears to be primarily attributable to the higher hematocrits (Hct) of our birds, as previously discussed, that allow them to achieve a much higher Ca_O2(Table 4). This, along with their somewhat lower Cv̄_O2, allowed our pigeons to deliver 1.6 times more oxygen to their tissues per unit blood flow compared to Butler's pigeons (Table 4). Thus, our pigeons' higher Ca_O2–Cv̄_O2value allowed them to satisfy their substantially higher flight V̇_O2/M_bwith approximately the same Q̇/M_b as Butler's pigeons.

How do the factors by which our pigeons increase their resting cardiorespiratory parameters to accommodate the increased metabolic requirement of flight compare with those of Butler's pigeons? Flight/Rest(F/R) ratios for these two studies are summarized in Table 5. As expected from the foregoing relationships, each Fick parameter F/R ratio shown in Table 5 is higher for our birds than for those in Butler's study (Butler et al., 1977).

Allometric equations are a powerful tool for understanding fundamental relationships between the physiological capabilities and the morphological dispositions and design constraints of different animals(Taylor and Weibel, 1991). Bishop (1997) has recently published a very comprehensive allometric evaluation of the maximum cardiorespiratory capabilities of exercising birds and mammals. He concluded that avian heart muscle has the same biomechanical performance as that of a terrestrial animal. Accordingly, by making some simplifying assumptions about the magnitude of the Ca_O2–Cv̄_O2term in the Fick equation, Bishop(1997) showed that heart mass(M_H) and V̇_O2max data from mammals could be used to calculate reasonable estimates of a flying bird's V̇_O2max.

Bishop (1997) presented two different methods for predicting a flying bird's V̇_O2max. Method 1 takes a hierarchical approach, and has the reader use his allometric equations to estimate fh from M_b (equation 3)and Vs from M_H (equation 6), if they are not known, and to estimate Ca_O2–Cv̄_O2from blood hemoglobin concentration (Hgb). Bishop's Method 2 simply assumes that Q̇_max is a function of M_H and that there is no difference between birds and mammals during maximal cardiac performance. Thus, for both birds and mammals,Bishop's equation 10 (Bishop,1997) is used to calculate Q̇_max from the animal's M_H, and Ca_O2–Cv̄_O2is estimated from Hgb as in Method 1. How precisely do Bishop's Method 1 and Method 2 allometric equations (Bishop,1997) estimate the cardiorespiratory parameters that we directly measured from our flying pigeons?

Most of Bishop's allometric equations(Bishop, 1997) relate the cardiorespiratory parameter to heart mass, rather than to body mass, so M_H is an important attribute to consider(Table 6). Bishop's equation 11, in turn, lets one calculate a mean population M_H of a bird from its M_b. The calculated M_Hfor our pigeons is more than 20% less than the measured value(Table 6, Row 2). Concerning our M_H value Magnan(1922), in his comprehensive study of birds, reported a mean pigeon M_H of 1.294% of M_b, which corresponds to a mean M_H of 4.57 g for the 0.353 kg flying pigeons we studied. This differs by only 2%from our measured M_H value(Table 6).

Bishop's equation 6 (Bishop,1997) predicts a flying bird's maximal Vs from M_H. Substituting our pigeon's mean measured M_H into this equation yields a predicted Vs that is almost 20% higher than our measured value(Table 6, Row 3). What are the reasons for this discrepancy? Bishop's equation 6(Bishop, 1997) also comes directly from an earlier study (Bishop and Butler, 1995) that used Vs data for pigeons flying at minimum power speed (Bishop et al., 1977) to estimate the coefficient of this equation. As was discussed previously, even though the flight V̇_O2 of Butler's pigeons was only about two-thirds that of our birds, their Fick-calculated Vs value was about 6% higher than our value because of their low Ca_O2 value, which in turn, resulted from the abnormally low Hct values in their birds(Table 4). Accordingly, only part of the observed 20% discrepancy is explained by the Vs value that Bishop and Butler used to determine their equation's coefficient.

In the absence of actual flight Ca_O2–Cv̄_O2measurements (remembering that only pigeon values are currently available),Bishop used blood hemoglobin concentration data ([Hgb]; g ml^–1) to estimate blood oxygen carrying capacity([Hgb]×1.36 ml O₂ ml^–1 blood). When [Hgb]data were unavailable, Bishop assumed the mean value of 0.1513 g ml^–1 reported by Prinzinger and Miscovic(1994) from a comprehensive survey of birds. Bishop then estimated Ca_O2 by assuming that the arterial blood of a bird flying at V̇_O2max conditions was 94% saturated with oxygen, which is the mean value reported for seven species of mammals (mostly of large body size) running at V̇_O2max conditions(Bishop, 1997). Bishop estimated the flying bird's Cv̄_O2 by assuming a value of 0.038 ml O₂ ml^–1, which again is the mean value for seven species of mammals running at V̇_O2maxconditions. These assumptions are embedded in the two methods Bishop used for estimating the V̇_O2max of a flying bird. How closely do these largely mammalian-based assumptions and the resulting Ca_O2–Cv̄_O2estimates correspond to those measured from our homing pigeons flying at what we assume are V̇_O2maxconditions?

While we did not measure [Hgb] for our birds, McGrath(1971) reported a mean blood hemoglobin of 15.6 g% for his pigeons whose mean Hct was similar to our value(i.e. 52.6% vs 51% in our study). Proportional scaling and units conversion indicates a [Hgb] of 0.151 g ml^–1 for our birds,which is in excellent agreement with the default avian [Hgb] that Bishop assumes in his equations (see above). According to Bishop's assumptions, this[Hgb] would translate to an estimated flight Ca_O2 of 0.193 ml O₂ml^–1 for our pigeons, which is only slightly higher than our measured value of 0.188 ml O₂ ml^–1(Table 3). Finally, our finding that the Ca_O2 values of our resting and flying pigeons are almost the same (Table 3) indicates that arterial blood saturation does not fall as these birds go from quiet rest to heavy exercise.

Based on Bishop's assumptions, our pigeons should be capable of an estimated flight Cv̄_O2 of 0.038 ml O₂ ml^–1; a value about 20% lower than the measured value (Table 3). Again, it is not known whether this difference reflects the possibility that our pigeons were not flying at V̇_O2max conditions(as we have assumed) or the possibility that flying birds, for thermoregulatory or other reasons, may be incapable of achieving Cv̄_O2 values as low as highly athletic exercising larger-size mammals. In summary, the predicted Ca_O2–Cv̄_O2measured from our pigeons flying at V̇_O2max is about 10% lower than the calculated value (Table 6, Row 4).

Method 1: Our measured Q̇ is 18% lower than the value predicted by multiplying our measured fh by the Vs value estimated from Bishop's equation 6 (Bishop,1997; i.e. Method 1, see Table 6, Row 5). Surprisingly, our measured Q̇ is more accurately predicted by multiplying the fh value estimated from Bishop's equation 3 by the Vs value estimated from his equation 6. In the latter case, our measured Q̇ is only about 8% lower than the predicted value. These overestimates of Q̇ are primarily attributable to the previously discussed fact that Bishop's equation 6(Bishop, 1997) overestimates the Vs of our pigeons(Table 6, Row 3).

Method 2: Bishop's Method 2 (equation 10; Bishop, 1997), however, very precisely predicts the flight Q̇ of our pigeons (Table 6, Row 5). Accordingly, our Q̇ data provide strong support for Bishop's assumption that Q̇_max is simply a function of M_H (Bishop,1997).

How accurately do Bishop's two methods(Bishop, 1997) estimate the mean V̇_O2max value we determined for our flying pigeons? Our measured V̇_O2max value is about 18% less than the value predicted by Bishop's Method 1 when our M_H value was used to estimate Vs, and all other variables were estimated by Method 1 procedures. Our measured V̇_O2max value is only about 11% less than that predicted by Bishop's Method 2, however(Table 6, Row 6). This level of agreement is quite good considering all the variables that can influence V̇_O2max. This discrepancy might reflect an inaccuracy in our assumption that our pigeons were flying at V̇_O2maxconditions, or inaccuracies in one or more of the previously discussed assumptions or relationships that Bishop used to formulate his equation 10. These relationships support Bishop's(Bishop, 1997) suggestion that his Method 2 is the preferred way to estimate a flying bird's V̇_O2max.

How do the cardiorespiratory parameters of our flying pigeons compare with those predicted by Bishop's allometric equations for a mammal of this size running at V̇_O2maxconditions (Bishop, 1997)?These relationships are summarized in Table 7 and the findings are reported below.

The V̇_O2max of our pigeons was about 1.5× that predicted for an athletic running mammal of the same size (Table 7, Row 6). Of course, very few (if any) running mammals of this size are athletic, so this V̇_O2maxdifferential is usually significantly greater when a typical small running mammal is compared to our pigeons. For example, the mass-specific V̇_O2max of our pigeons is more than 3× greater than that of a trained running rat(Gleeson et al., 1983).

The higher (1.5-fold) V̇_O2max of our pigeons is associated with a disproportionately larger (1.8-fold) heart mass,and Q (1.7-fold) compared to values predicted for an athletic running mammal of this size (Table 7,Row 5). The higher Q̇ of our pigeons results from its ability to achieve fh and Vs values that are both about 20% higher than those expected for a running mammal of the same size(Table 7, Rows 1 and 3). The fact that the ratio of `present study/predicted' for M_Hand Q̇ in Table 7 are generally similar supports Bishop's assumption that M_H is a valid indicator of an animal's maximum Q̇ capability(Bishop, 1997).

In summary, the superior blood oxygen convection capabilities of our pigeons over running mammals of equivalent size are primarily attributable to their ability to achieve higher heart rates and greater stroke volumes simultaneously from their larger size hearts.

As stated previously, a primary underlying assumption of Bishop's comprehensive analysis was that avian heart muscle has the same maximal physiological performance capability as mammalian heart muscle(Bishop, 1997). In this final part of our discussion, we will examine how closely the heart mass-specific cardiac performances of our flying pigeons correspond to those reported for certain non-athletic and highly athletic mammals exercising at or near V̇_O2maxconditions.

It has been known for some time that high M_b-specific heart mass (M_H/M_b) distinguishes highly athletic animals from their less athletic counterparts. The 4.6-fold M_H/M_b difference between the pigeon and the exercise-trained rat is particularly striking(Table 8), and reflects the fact that the body mass-specific V̇_O2max of our pigeons is more than three times greater than that of a rat(Gleeson et al., 1983). Although M_H/M_b is generally similar in highly athletic mammals of very different body sizes, our pigeons' value is the highest of any of the animals shown in Table 8.

The relatively high flight fh of our pigeons compared to a flying bird or a running mammal of the same size has been discussed previously (Tables 6 and 7). The latter relationship is further supported by the fact that our pigeon's flight fhexceeds the running rat's maximum value despite the latter's substantially smaller body mass (Table 8).

Data summarized in Table 8show no consistent relationship between athletic ability or body size and M_H-specific Vs(Vs/M_H). Since available data indicate that running mammals and flying birds have similar mean systemic arterial blood pressures (Bishop, 1997),the data in Table 8 suggest that the maximum M_H-specific stroke work capability of our pigeons is somewhat less than that of a running rat. Perhaps this is a trade-off that pigeons have had to accept in gaining the advantages of having a relatively large size heart that can achieve a relatively high fh.

Due in part to the inverse relationship between maximum fh and body mass, M_H-specific Q̇(Q̇/M_H) decreases with increasing body mass (Table 8). Accordingly, the most valid Q̇/M_H comparisons are between animals of generally similar body mass. There is reasonably close correspondence between the pigeon's Q̇/M_H value and that reported for the running rat or estimated for two species of flying bats(Table 8). Since flying birds and running mammals have similar mean systemic arterial blood pressures(Bishop, 1997), the M_H-specific cardiac work rate capabilities(W_H/M_H) of birds and smaller size mammals also appear to be generally similar. In summary, Q̇/M_H data from our pigeons provide additional direct support for one of Bishop's central assumptions, that avian heart muscle has the same maximal physiological and biomechanical performance as that from terrestrial mammals. Particularly impressive to us is the ability of 1 g of pigeon myocardium to achieve a Q̇/M_H value that approaches that of a running rat, even though this bird is operating at a Q̇/M_b level during flight that is about four times greater than that of the running rat. The physiological mechanisms that enable homing pigeons to achieve such a high level of cardiovascular performance are still not completely understood. Some important contributing factors, however, may be the robust plasma catecholamine response that accompanies pigeon flight(Liang and Thomas, 1994), and the possibility that flying pigeons maintain advantageous phase relationships between their cardiac and pectoral muscle (wing-beat) cycles(Thomas et al., 1996).

This paper is dedicated to Dr Melvin Kreithen. His abrupt death leaves a void in friendship, interesting conversation and unique insights into the area of avian physiology. He was a professor who was sincere when teaching students, both undergraduate and graduate. He will be deeply missed. Special thanks to all the undergraduate students that participated in completing this project. Without their participation and enthusiasm, the project would have not come to completion. A number of surgical techniques and analytical procedures used in this study were perfected by Grant William Peters while carrying out research for his MSc thesis.