The Effects of the Wingbeat Cycle on Respiration in Black-Billed Magpies (Pica Pica)

Marey (1890) suggested that forces generated by the flapping of wings during flight might have some effect on the thoraco-abdominal cavity of birds and the airsacs therein. More recently, in their cineradiographic studies of European starlings (Sturnus vulgaris) in flight, Jenkins et al. (1988) demonstrated that the sternum is moved in an expiratory direction during downstroke and in an inspiratory direction during upstroke. Jenkins et al. (1988) also documented that the furcula is bent laterally during downstroke and recoils with upstroke. These observations led them to suggest (see also Goslow et al. 1990) that skeletal movements during flight could alter the anterior and posterior airsac pressures in such as way as to affect the internal distribution of air within the respiratory system. Banzett et al. (1992) recently measured respiration in flying European starlings and suggested that the effect of wing movements on respiratory airflow (measured at the mouth with a mask-held pneumotach) was ‘insignificant’ (3–11 % of tidal volume as assessed by ensemble-averaging). The question remains, however, whether the kinematic events observed in starlings are present in larger birds. Furthermore, the issue of the potential effect of wing movements on anterior and posterior airsac pressures and internal airflows is still unresolved.

The purpose of our investigation, therefore, was to measure anterior and posterior airsac pressures and tracheal airflow in flying magpies while simultaneously visualizing skeletal movements cineradiographically. The magpie was chosen because of its more suitable size for instrumentation during flight and because we found that it has a variable ratio between wingbeat and respiratory cycles that includes the ratio of 3:1 most commonly observed among birds (Berger et al. 1970) as well as ratios of 5:2 and 2:1.

Some of the data presented here have been summarized in a review (Boggs, 1997).

Black-billed magpies (Pica pica L.) were trapped (under state and federal permits issued to K. P. Dial) in Western Montana and trained to fly in our windtunnel. The open windtunnel has a 76 cm×76 cm×91 cm acrylic flight chamber through which air is drawn by a Buffalo 36-b-vanaxial asymmetric fan driven by a 15 kW (20 horsepower) variable-speed d.c. motor. Upwind of the flight chamber, a 5 mm honeycomb baffling 10 cm thick straightens the airflow, which is laminar within the flight chamber farther than 2.5 cm from the walls (see Tobalske and Dial, 1994, for details of design and airflow regime). Four magpies were used in the cineradiographic studies, but respiratory measurements were made on eight birds (body mass 165±7.5 g, mean ± S.E.M.) and airsac pressure measurements on 10 birds. The birds were housed in flight cages or in an open room, so they were free to fly and perch. Birds were trained in the windtunnel for 30 min per day for 2–4 weeks prior to experimental measurements. During training and the experimental flights, they flew at their preferred speed (6–8 m s⁻¹) as assessed from field measurements (Olson, 1993).

Under anesthesia (25 mg kg⁻¹ ketamine and 2 mg kg⁻¹ xylazine, intramuscularly), cannulae made of PE 200 tubing with side-ports were inserted in the interclavicular and posterior thoracic airsacs. They were held in place with cyanoacrylate adhesive to the airsac wall and sutured to the furcula, in the case of the interclavicular sac cannula, or to the pubic bone, in the case of the posterior thoracic sac cannula. During the flights, each cannula was connected via silastic tubing (602-305, 2 mm i.d. × 3.2 mm o.d.) to the measurement side of a miniature piezoresistive pressure transducer (Endevco model 8507C-2) held on the bird’s back within a Velcro backpack (reference ports were within the still air of the Velcro pack). Each pressure transducer was calibrated against a water manometer before and after every experiment. The transducers have a sensitivity of 157 mV per 6.8 kPa, a non-linearity of only 0.82 % full-scale output and a rise time of 0.0179 ms (according to manufacturer’s specifications). In our own tests of the transducers, together with the length of tubing used in the experiments, we found the mean response time (i.e. the time between the application of a pressure and the 95 % recorded response) was 1.36±0.27 ms (mean ± S.E.M.). The transducers were found to be insensitive to acceleration.

To provide an indication of wingbeat cycle when the bird was not within the cineradiographic field, or in birds not filmed cineradiographically, we implanted bipolar silver wire electrodes (0.1 mm diameter, California Fine Wire Co.) into the pectoralis (as described by Dial, 1992). These wires were connected via ultra-miniature strip connectors (Microtech, model GM-2, GM-6, GF-2 or GF-6) within the backpack to a cable [Cooner Wire Co., model NMUF-2 (or 6)/30-40462J] that carried the electromyographic signal to a Grass P511 preamplifier.

Tracheal airflow was measured by inserting a dual thermistor bead anemometer into the bird’s trachea. The anemometer was made by Hector Engineering, model 2C, with Veco 0.25 mm diameter Ultrasmall thermistor beads mounted within a 5 mm section of stainless-steel tubing (3 mm o.d., 2.4 mm i.d.). Mean tracheal internal diameter at the point of insertion was 3.15 mm. The flow signal was linearized and integrated to provide tidal volume. Flows were calibrated using a Brooks Instrument Division 12 L volumeter, to deliver a flow through the anemometer, with pieces of silastic tubing on either end and with a cross-sectional area equivalent to a magpie trachea. Because the flow signal was affected by the slight difference in temperature between inspiratory and expiratory air, we calibrated the flow probe in the expiratory direction and integrated expiratory flow to obtain tidal volumes, since we could assume that the expiratory air was at body temperature but did not know what the slight difference from inspiratory air temperature was. The worst-case dynamic response of the flow meter is 90 % of full scale in 6 ms. In this device, the two thermistor beads are axially oriented in the flow stream, and the upstream thermistor is cooled faster than the downstream thermistor. With alternating flow directions, the thermistors alternate in terms of which one has the higher output in synchrony with the flow direction, independent of the magnitude of flow. The change in polarity of this signal indicates the flow direction, and this change in direction signal is used to trigger the integrator. The voltage output is a non-linear function of flow in this device. That signal is linearized and then integrated electronically over time to give tidal volume. Measurements of alternating flow direction are complicated by the complex momentary movement of air in directions other than axial at the instant of flow reversal. If bulk flow reversal occurs before this momentary non-axial translation subsides, then the flow signal will not come to zero when bulk flow reverses (as will be noted in some instances in Figs 2 or 4 where linearized flow is presented). The effect of this on the value of the integral of the flow curve is insignificant because, at the point of flow reversal, the slope of the flow curve is very steep and a very small percentage error is obtained by taking the integral as though the flow curve actually came to zero. The flow probe’s response characteristics were also assessed using the techniques described by Jackson and Vinegar (1979), and there was no aberrant response for magnitude or phase up to at least 30 Hz, which is well beyond the range of frequencies at which flying birds breathe or flap their wings. All respiratory and electromyographic records and synchronization signals from the cineradiographic camera were recorded on a Teac MR40 tape recorder.

We used the simultaneous records of airsac pressures and tracheal airflows to estimate airway resistance for five birds from the mean slope of the relationship between pressure and flow from four randomly selected breaths in each state, at rest and during flight. During flight, the effects of inertial impedance and the slower response time of the flow probe with respect to the pressure transducer create a phase shift between the pressure and flow signals that had to be accounted for, and the flow signal could be used to calculate resistance only during periods when flow was changing slowly.

The birds were flown 12–18 h after the surgery at their preferred speed (6–8 ms⁻¹) for a mean flight time of 20 s, with at least four repetitions. The short post-operative recovery period was sufficient for the relatively minor surgical procedures employed and was necessary to avoid contamination of the thermistor beads with tracheal mucus. Atropine was administered to some birds to try to avoid this contamination, but had no apparent effect. Many experiments were terminated because of mucus getting onto the thermistor beads. The thermistors are exquisitely sensitive to anything that alters heat conductance from their surface. It was apparent from the changing signals for both flow and flow direction when this occurred, and the measurements had to be terminated. Mucus accumulation in the tube containing the thermistor beads could also compromise the flow calibration as the radius of the lumen was reduced. In the cases where the experiments were ended without apparent problems with the flow probe, there was little mucus in the tube upon removal, but in those cases when experiments were terminated due to mucus on the beads, it is not possible to know for how long mucus may have reduced the lumen radius before it reached the beads. However, our impression is that once mucus got into the tube it adhered quickly to the beads and its effects were apparent in the records. The birds were flown as soon after implantation of the tracheal flow probes as they were willing because the longer the probe was in the trachea the more likely it was that mucus would accumulate. The birds will not fly if they do not feel well; hence, the time from instrumentation to flight measurement was dictated by the bird. The relatively brief flight durations are a product both of inherently short flight durations of magpies in the field and laboratory and of the influence of the large amount of apparatus attached to the animal.

A Siemens cineradiographic apparatus (grid controlled tube with a 0.6 mm focal spot and 27.9 cm Optilux 27D Triplex image intensifier) captured radiographic images with a Photosonics series 2000 16 mm cine camera run at 200 frames s⁻¹. To calibrate kinematic measurements, a 1 cm length of wire was affixed with cyanoacrylate adhesive to the skin in the midline of the posterior dorsal region.

Films were analyzed employing customized kinematic analysis software (S. M. Gatesy), a BITPAD PLUS digitizing tablet and a Quadra 950 computer. Skeletal landmarks that consistently appeared clearly in the X-ray images were used to measure excursions of the furcula, the sternum and the pelvis (Fig. 1). From dorsoventral projections, measurements were made of the distance between the dorsal ends of the furcula at the shoulder. Lateral views were used to measure the distance between the coraco-sternal junction and a 1 cm marker on the vertebral column or the acetabulum, the angle between the horizontal extension of the 1 cm vertebral column marker and the coraco-sternal junction, and the angle between the horizontal extension of the 1 cm marker and the acetabulum and posteroventral tip of the pelvis.

When differences between means were assessed, a two-tailed Student’s t-test was employed with significance taken at the 5 % level. Kinematic data were analyzed using time-series analysis (SPSS, 1993). As sequential data are frequently autocorrelated (that is, any one point in the series is a good predictor of the next), they violate the assumption of independence required for ordinary regression models and will thus overestimate the strength of the inferred causal relationship between the two variables being tested. Instead, a maximum-likelihood autoregressive model was used in which the autocorrelation effect, estimated by an autoregressive integrative moving average algorithm (ARIMA), essentially becomes a variable in a multiple regression. The resulting regression model thus accounts for the autocorrelation effect and reveals a better estimate of the relationships between the variables independent of any spurious relationship that might have been created by autocorrelation within each variable of the series (SPSS, 1993).

Kinematic analysis of X-ray film of flying magpies reveals that the skeletal movements observed in the European starling during flight also occur in this somewhat larger species of bird. The furcula is bent laterally on downstroke and recoils medially on upstroke (total excursion 2–2.5 mm). If the vertebral column is used as a reference, the sternum appears to move dorsally on downstroke and ventrally on upstroke (Fig. 1, see also Fig. 3) (excursions 2–9 mm). Time-series analysis of three birds for which we had the longest series of wingbeat cycles consistently showed a highly significant correlation (P<0.001) between humeral position and the angle between the coraco-sternal junction and the vertebral column marker. The increasing coracoid angle (Fig. 1) as the humerus descends indicates a dorsal movement of the coraco-sternal joint with downstroke. This was a consistent pattern in all three of the birds for which we had sufficiently long flight sequences within the 20 cm diameter cineradiographic camera field to allow for the continuous data needed for this analysis.

We also observed movements of the pelvis in two of the four birds with each wingbeat cycle. The posterior aspect of the pelvis moved ventrally on downstroke and dorsally during upstroke.

If the furcular bending on downstroke expanded the interclavicular airsac while the vertical force on the sternum with downstroke was compressing the posterior airsacs, then the changes in pressure in the interclavicular and posterior thoracic airsacs should be out of phase with one another and/or of differing magnitude (assuming that there is sufficient impedance between the airsacs to allow that to occur). Contrary to these predictions, we found that the anterior and posterior airsac pressure fluctuations occurred in phase with each other (Fig. 2). Also illustrated in Fig. 2 are the smoother airsac pressure profiles during a brief period without flapping (i.e. during gliding; Fig. 2). Peak pressures in the interclavicular and posterior thoracic airsacs ranged from 0.20 to 0.78 kPa, with the mean values presented in Table 1. There is a statistically significant (P<0.005) tendency for the anterior airsac peak pressures, whether below or above atmospheric pressure, to be lower than those of the posterior airsacs.

Since the magpie takes 2.5–3 wingbeats per breath, there are times when the flight-induced movements of the sternum are in the opposite direction from those driven by the respiratory muscles. Combining airsac pressure measurements with the simultaneous high-speed cineradiographic images of the sternal movements (in this case traced frame-by-frame onto acetate sheets to produce the composite image), we can discern a substantial effect of flight-induced sternal displacements upon airsac pressure both when downstroke occurs during an inspiration and when upstroke occurs during an expiration (Fig. 3). We can be sure that the pressure fluctuations we see are not artifacts because they correspond well to fluctuations in airflow (Figs 2 and 4).

The mean changes in airsac pressure caused by either downstroke occurring during inspiration or upstroke occurring during expiration are summarized in Table 2 for 10 birds. These changes were not measured by ensemble-averaging. They were measured in terms of the peak difference in pressure or flow with respect to a line joining the preceding and succeeding maxima and are expressed in Table 2 either as that pressure or as the fraction representing that pressure or flow divided by the pressure or flow represented by the line joining the preceding and succeeding maxima (as indicated in the diagram in Table 3). The downstroke-induced thoraco-abdominal compression can have a substantial effect on the subatmospheric airsac pressure during inspiration, changing it by as much as 115 % and by an average of 92 % or 0.35 kPa. The upstroke-induced thoraco-abdominal expansion can change the positive expiratory pressures by an average of 63 % or 0.23 kPa. The number of events from which the mean for each bird was calculated differed (over a range from 10 to 86), and hence the more appropriate weighted means are also presented in Table 2. These changes in pressure are mirrored by changes in tracheal airflow; changes in airsac pressure that exceed 100 % can cause airflow reversals in the trachea (Fig. 4). A more complex and unusual expiration pressure profile seen in Fig. 4 illustrates that, during an expiration, both an upstroke and a downstroke may be associated with a reduction in pressure, suggesting that the effect of downstroke on sternal position may be position-sensitive or that something else (as yet unidentified, but such as leg movements) may be causing some of the pressure fluctuations. Heart rate in the flying magpie is approximately 1.4–1.7 times wingbeat frequency (i.e. of the order of 703 beats min⁻¹). Although heart beat was apparent in a few pressure recordings as a very small pressure fluctuation, it was not usually apparent and seems an unlikely candidate for these occasional fluctuations in pressure during flight due to something ‘other than’ wingbeat. If heart beat were inducing that much variation in pressure, its effect would also have been apparent during gliding and resting periods.

The durations of compressions or expansions of the thoraco-abdominal cavity induced by the wingbeat cycle are quite short (3–8 ms). Data on changes in airflow in birds from which we obtained both airsac pressure and tracheal airflow (N=5) simultaneously (Table 3) show that the measured flow depression is often not as great as the airsac pressure deviation. The pressure values for some individuals differ in Table 3 from those in Table 2 because only the pressure values from those experiments when both flows and pressures were measured simultaneously are presented in Table 3. By integration over the time of the diminution of flow caused by the pressure disturbance of the downstroke during inspiration, or upstroke during expiration, we can calculate the estimated volume effect, expressed as a percentage of the actual tidal volume of that breath (column 3, Table 3). The mean pressure change caused by a downstroke occurring during inspiration (94 %) is greater (P<0.001) than the mean pressure change caused by an upstroke occurring during expiration (51 %). The corresponding changes in flow and volume are likewise greater when downstroke occurs during inspiration (75 % and 23 %, respectively) than when upstroke occurs during expiration (45 % and 11 %, respectively).

Mean values for a number of respiratory parameters in magpies at rest and during flight (Table 4) show that minute volume is increased sixfold during short flights and is achieved by a tripling of respiratory frequency and a doubling of tidal volume. Airway resistance tends to be reduced during flight compared with airway resistance in the resting bird.

Cineradiographic analysis of magpies in flight reveals a pattern of movement of the furcula and sternum similar to that described for starlings (Jenkins et al. 1988), which suggests that these may be common kinematic events in most flying birds. Airsac pressure and tracheal airflow measurements indicate a substantial effect of the wingbeat cycle on respiratory system function in flying magpies.

The observation that the anterior and posterior airsac pressures change in phase with one another (Figs 2, 4) suggests that the lateral expansion of the furcula on downstroke and the medial recoil on upstroke probably have little effect on the interclavicular airsac, which resides largely posterior to this structure. The somewhat smaller pressure excursions in the anterior airsac compared with the posterior airsac must be due to the smaller volume changes occurring anteriorly than posteriorly. The same phenomenon is observed in resting birds (although the anterior–posterior pressure difference is smaller) and hence is probably not entirely attributable to flight-induced furcular movements.

The fundamental mechanisms for the apparent thoraco-abdominal compression with downstroke and expansion with upstroke remain under investigation. If one stimulates the pectoralis in a non-flying bird or in a bird suspended by its wings, as we have done in preliminary studies with pigeons, there is insufficient movement of the sternum caused by the pectoralis contraction to induce a change in airsac pressure, even if the airsacs are held in an inflated state. However, if the bird is accelerated upwards, the inertia of its body does compress the airsacs and elevate the pressure within them. Thus, it seems likely that, when the downstroke is generating lift (which is greatest in mid-stroke), the vertical force transmitted through the coracoid ‘struts’ to the sternum accelerates the pectoral girdle upwards against the weight of the bird’s dorsal viscera and vertebral column, creating a compressive force on the airsacs. Essentially the opposite occurs during the upstroke. This kind of inertial mechanism creating internal pressure fluctuations that can contribute to a coordination between respiratory and locomotor cycles has been described in hopping and running mammals such as the wallaby (Baudinette et al. 1987; Alexander, 1989; Baudinette, 1991; Bramble, 1989) and the dog (Bramble and Jenkins, 1993).

In some magpies, we observed ventral pelvic movements on the downstroke that would contribute to abdominal compression and dorsal movements on the upstroke that could contribute to abdominal expansion. These pelvic movements are not observed in all birds, however, and appear to be passive effects of the lift and recovery phases of the wingbeat cycle. They could be associated with the active use of the tail as a lift-generating surface (Gatesy and Dial, 1993) or they may represent active recruitment of the pelvic muscles for breathing. Baumel et al. (1990) have described active pelvic movements with respiration in pigeons resting on their sternae while nesting. The depression of the pelvis with expiration results from contraction of what they refer to as the suprapubic muscles, the M. caudofemoralis, M. pubocaudalis internus and M. pubocaudalis externus. The elevation of the pelvis during inspiration is associated with activity in the M. longissimus, with the pelvis moving at the notarial–synsacral joint. The activity of these muscles needs to be recorded during flight to determine whether they are recruited for breathing during flight, as has been reported for pigeons resting on their sternae (Baumel et al. 1990).

The magnitudes of the absolute airsac pressure fluctuations with each breath observed in flying magpies in this study are similar to those observed by Brackenbury (1986) in chickens running at 5 km h⁻¹ or exposed to 10 % O₂, 3 % CO₂ while running at 3.2 km h⁻¹. The existence of slight pressure differences between the posterior and anterior airsacs can affect the patterns of airflow through the parabronchi both at rest and during flight, as demonstrated in models of the fluid dynamics of the avian respiratory system presented by both Kuethe (1988) and Butler et al. (1988). The magnitude of the pressure difference between the anterior and posterior sacs is also similar to that observed in chickens (Brackenbury, 1971) and ducks (Kuethe, 1986; Boggs et al. 1996) and is considerably greater than the 0.001 cm H₂O (0.1 Pa) elastic pressure differences between airsacs that Butler et al. (1988) suggest would be capable of influencing the patterns of gas flow and would ‘warrant more careful investigation’.

The differences in peak pressure observed between the anterior and posterior sacs in flying birds (Table 1) could theoretically be an artifact of common-mode rejection and/or of fluid in one of the tubes. However, the transducers were tested by exposing both to a common pressure at ‘in flight’ frequencies (i.e. 8 Hz). The signals were in phase with one another, and the amplitudes (i.e. pressures) were within 0.8 % with identical tubing configurations (i.e. a 0.02 cm H₂O or 0.00196 kPa difference with an applied pressure of 2.41 cm H₂O or 0.236 kPa). This tiny difference is probably attributable to our ability to calibrate and read these small pressures with consistent accuracy more than to real differences in response characteristics of the transducers. With the tubing configuration used in the bird (the PE 200 tubing emerging from the posterior sac had a bend in it and was slightly longer than that emerging from the anterior sac), there could be as much as a 2 % difference in the pressures (e.g. posterior 0.05 cm H₂O or 0.0049 kPa greater than anterior with an applied pressure of 2.2 cm H₂O or 0.216 kPa). The transducers and tubing were, therefore, reasonably well matched, with minimal common-mode rejection problems. However, if one tube had an accumulation of fluid and the other did not, this could create apparent pressure differences, because the pressure signals from the one containing fluid would be damped. For example, 5 mm of human saliva in one tube can cause the difference between the two pressure signals to rise to 25 %, i.e. the order of magnitude of the difference observed among the birds. We were aware of this potential problem and ‘blew out’ the tubes prior to the measurements to avoid any damping effects of fluid in them. Furthermore, any such problem would have occurred randomly in either tube and hence could not have contributed to a consistently greater pressure change in the posterior than in the anterior airsacs.

The compressive effect of downstroke and expansive effect of upstroke on the thoraco-abdominal cavity are reflected in substantial changes in airsac pressures in flying magpies (Table 2; Figs 2–4). These flight-induced changes in pressure also have a substantial effect upon airflow, which may be retarded, stopped or reversed in the middle of an inspiration or expiration (Figs 2, 4; Table 3). These effects on airflow could induce mixing in the small dead space of the avian respiratory system, although this effect may be minimal at high convective transport rates, as described by Banzett and Lehr (1982), or they could enhance evaporative heat loss from the airways and affect the distribution of air through the lungs and primary bronchus. The oscillations in flow superimposed on the inspiratory and expiratory flow profiles are somewhat reminiscent of the thermoregulatory compound ventilation pattern of panting described in pigeons by Ramirez and Bernstein (1976). It has been estimated that starlings and white-necked ravens lose 19 % and 20 %, respectively, of their metabolic heat production during flight via respiratory evaporation (Torre-Bueno, 1978; Hudson and Bernstein, 1983). Oscillations in tracheal airflow induced by the wingbeat cycle could contribute to the dissipation of the potentially high heat loads incurred during flight without influencing lung ventilation.

The aerodynamic valving mechanisms (both inspiratory and expiratory) responsible for the unidirectional flow of air through the avian lung are highly dependent on convective inertial forces and, therefore, on flow velocity profiles (Butler et al. 1988; Wang et al. 1988; Brown et al. 1995) and may also be influenced, in the case of expiratory valving, by dynamic compression of the intrapulmonary bronchus (Brown et al. 1995). Expiratory valve failure is unlikely at the high flow velocities represented by the peak expiratory flows of flying birds, although it has been observed at the low flow velocities of resting breathing in ducks (Powell et al. 1981) and chickens (Piiper et al. 1970; Bouverot and Dejours, 1971). Whether the brief depressions in flow induced by the wingbeat cycle could result in some air bypassing the parabronchi, and passing instead through the primary bronchus, can only be determined in future studies using larger birds that allow for flow probes to be implanted in the primary bronchus (or mesobronchus) rather than the trachea.

The range of volume changes induced by the wingbeat cycle measured in magpies (11–23 %) (Table 3) exceeds that recorded in starlings by Banzett et al. (1992) (3–11 %). These differing assessments of the magnitude of the impact of flight on airflows and volumes may be due to the use of different species or to the different techniques employed or both. Banzett et al. (1992) recorded flow at the mouth with a mask-held pneumotachograph, which could represent an impedance that would dampen the recorded flow fluctuations. Also, the ensemble-averaging analysis would probably give smaller changes than our method of assessing the magnitude of these changes. Our approach may overestimate, while the ensemble-averaging method probably underestimates, the wingbeat effects.

It must be emphasized that the effects of the wingbeat cycle upon pressures (94 % change) and flows (75 % change) we have observed correspond to the difference between those points in time when the respiratory muscles and lift forces are opposing one another in their effects upon thoraco-abdominal volume and those points in time when they are acting in concert; these changes are with respect to those points in time when they are working together (i.e. when downstroke occurs with expiration and upstroke with inspiration). These out-of-phase ‘interference’ effects may therefore overestimate the ‘assisting’ contribution of the wingstroke when it is in phase with respiration, but it is not possible to know by how much. Preliminary data on respiratory muscle activity patterns in magpies (Boggs, 1997) indicate that the inspiratory muscles are not continuously active during inspiration as they are during non-flapping breaths, but rather are far less active during early inspiration, corresponding to peak subatmospheric pressure and upstroke, than later in response to the interfering effects of a downstroke during mid-inspiration. Therefore, the relative contributions of respiratory muscles and wingstroke dynamics to airsac pressures cannot be measured quantitatively. In a bird such as the magpie, with a 3:1 ratio of wingbeat to breath cycles, there would be two periods of ‘assistance’ and one period of ‘interference’ during each phase of the breath cycle; that is, the periods of assistance would not only cancel out the periods of interference but could contribute substantially to the total flow and volume. As described by Berger (1980), increasing the rate of flow reversals can contribute to achieving a greater tidal volume. Pigeons with a 1:1 ratio of wingbeats to breaths coordinate the wingstrokes to contribute to the flow reversals, i.e. the transition from inspiration to expiration occurs with downstroke and the transition from expiration to inspiration occurs with upstroke (as described by Butler et al. 1977; and as we observe as well). Likewise, the magpie, when using wingbeat to breath cycle ratios of 2.5–3:1, ensures that upstroke usually occurs with the transition into or with early inspiration and that downstroke occurs with early expiration or with the transition into expiration (see Figs 3–5 in Boggs et al. 1997). Hence, the phasic coordination of respiratory to locomotory cycles may not only enhance the tidal volumes achieved by accelerating airflow reversals but may also constitute an energy-conserving mechanism by minimizing the times when respiratory muscles must work against the kinematic effects of flight and by maximizing the times when the respiratory muscles and lift and inertia work together to expand and compress the thoraco-abdominal cavity to achieve respiratory airflow.

The mean value for the resting tidal volume (2.95 ml BTPS) reported here for magpies is close to the value that would be predicted for a bird of this size by the allometry of Bech et al. (1979), but the frequency is higher (at 52 breaths min⁻¹) than predicted (30 breaths min⁻¹) by the allometric equation of Lasiewski and Calder (1971). Several other studies (reviewed in Bernstein, 1987) have found higher frequencies than that equation would predict, however, which may emphasize the ‘preliminary’ nature of that study and the need to re-examine that allometric relationship using the additional data that are now available. These birds were probably not entirely ‘at rest’ since measurements were taken before flights while they sat on a perch in the windtunnel. The tidal volume of 6 ml BTPS during flight is also close to the prediction from the equation derived by Bernstein (1987) from what little data exist on tidal volumes in flying birds. More information on the respiratory mechanics of flying birds is needed to help us understand the breathing patterns selected (see Butler, 1991, for a review): some birds increase both tidal volume and frequency and do not achieve 1:1 synchrony, while others, such as the pigeon, increase only frequency while maintaining the same tidal volume as at rest and do synchronize respiratory and locomotory cycles.

Our estimate of airway resistance in the resting bird (0.22 cm H₂O ml⁻¹ s⁻¹ or 21.6 Pa ml⁻¹ s⁻¹) is lower than would be predicted for a mammal of the same mass (0.34 cm H₂O ml⁻¹ s⁻¹ or 33.4 Pa ml⁻¹ s⁻¹; Bennett and Tenney, 1982). A lower airway resistance in birds than in mammals has been observed in other studies (e.g. Kampe and Crawford, 1973) and is consistent with the lower work of breathing in birds compared with mammals (Perry and Duncker, 1980). A reduced airway resistance during the increased ventilatory drive of exercise is also consistent with the dilation of the segmentum accelerans, the region of the primary bronchus cranial to the junctions of the ventrobronchi, observed by Wang et al. (1992) in response to the increased ventilatory drive of elevated CO₂ levels.

Variations in airsac pressures induced by a pump ventilator can entrain the respiratory frequency (as indicated by respiratory motor neuron output) to the ventilatory frequency (Ballam et al. 1985). The variations in airsac pressure caused by the wingbeat cycle reported here probably create such a coordinating signal to the respiratory control centers from airsac stretch receptors. Funk et al. (1992) have also demonstrated a coordination of respiratory to locomotor cycles with passive wing flapping in Canada geese, and suggest a role for afferent signals from both the chest wall and the wings in creating that coordination. Hence, afferent feedback to respiratory control centers in birds that communicates the mechanical impacts of the wingbeat cycle upon the respiratory pump, as described in this study, are likely to be important in achieving phasically coordinated patterns between respiratory and wingbeat cycles, as described by Boggs et al. (1997) for magpies and by Berger et al. (1970), Butler and Woakes (1980) and Funk et al. (1993) for other species.

We greatly appreciate the contributions of several students (Jerred Seveyka, Keith Bradley, Bret Tobalske, Doug Warrick, Andi Rogers and Randy Trenary) in collecting and training birds and/or analyzing data. Dwight Hector provided valuable assistance in the proper use of his anemometry system and in preparing replacement probes rapidly. The manuscript was greatly improved by the helpful comments of the anonymous reviewers whose efforts we very much appreciate. This work was supported by NSF grant IBN-9206673 and was performed under an IACUC-approved protocol (BSDB-91-B).