Flapping Flight of the Andean Condor in Nature

Air movement around the flapping wing of a bird is strikingly complex; at every stage of the wingbeat cycle the angle of attack and the strength of the relative wind is in flux and uniquely so for each spanwise section of the wing. The motor pattern of a wingbeat has been described in some detail (Storer, 1948) and mathematical models have been proposed (Cone, 1968) yet few measurements have been recorded on the flight of flapping birds.

This paper presents a quantitative analysis of flapping flight by the Andean condor (Vultur gryphus) in nature. Although condors flap infrequently when flying the pattern does occur predictably in certain circumstances. On a Peruvian beach below one of their foraging routes I was able to measure the effects of flapping on sinking and forward air speed. By combining these data with those taken from cine film of the wingbeat pattern I was able to analyse in some detail the aerodynamics of flapping flight by this big vulture.

The procedure in measuring forward and sinking air speeds of condors crossing Playa Chucho, a Peruvian beach, is described in the preceding paper (McGahan, 1972). When condors flapped during the crossing I recorded the number of wingbeats that occurred within the observation segment of the beach.

I filmed condors in flapping bouts from different positions with an 18–86 mm Switar lens on a Bolex 16 mm camera running at 64 frames/sec. The duration of some flapping bouts was timed to the nearest 0·1 sec with a stopwatch.

Condors usually flap when they take off and land. In level flight they often flap when air movement has no upward component. Long flapping bouts are typical of chases among condors. Gliding condors may flap when they are suddenly alarmed. On the ground they flap in apparent attempts to maintain balance while fighting, mating, or climbing up and down steep slopes. Adults flap their wings during courtship displays, during and after bathing and sometimes after sunning. Perched nestlings flap in begging displays and in long bouts that simulate flapping in flight.

The pattern usually occurs in short bouts. For level flight I observed 175 bouts ranging from 2 to 6 wingbeats in extent with a mean and standard deviation of 3·5 ± 0·8. No instance of a lone wingbeat was recorded for level flight suggesting that efficiency relative to effort is reduced. In take off, wingbeat frequency is more variable and depends largely on the type of take-off site. For 11 instances of take off from cliffside perches, bouts varied in extent from 0 to 8 wingbeats with a median of 2. Bouts for 21 instances of take off, all made downhill on a 30° slope at a bait area, varied in extent for 3 to 23 wingbeats with a median of 8. Bouts were significantly longer (P < 0·01, median test) for take off from the more level site; apparently the increase is necessary to attain a minimum air speed for flight when a descent is not immediately available.

I recorded average durations of the wingbeat cycle in different activities (Table 1). A frequency of 1·8 cyc/sec, representative of normal level flight, for an adult female is seen to be less than the frequencies recorded for (1) three take-off bouts that, in each case, began on the ground and ended after the bird was airborne, (2) an extended bout by a condor in escape flight, and (3) a bout by a black vulture in level flight. Frequency differences in the first comparison are probably due to a reduction of cycle amplitude early in the take-off bout, which is necessary to prevent the wingtips striking the ground. Flapping sequences observed in courtship and copulation patterns on the ground were less rapid than those recorded for flight.

I recorded information describing wing movement in a frame-by-frame inspection of cine film taken of flying condors. When a condor in gliding flight begins to flap the first movement carries the wings downward from the gliding position. At the onset of downward movement the angle of incidence of the midwing area is reduced slightly either by active rotation of the wing or by passive deflexion of a less rigid trailing edge to the new angle of the resultant wind force. Near the end of the downstroke the wing flexes slightly at the elbow and the wrist. The midwing section begins to rise before the primaries have completed the downstroke (Figs. 1a-c; 2). Once the primary tips reach the end of the downstroke, a nearly vertical position, the forearm and wrist section of the wing have already passed from their nadir below the body to a position level with it. The wrist and elbow remain partially flexed until the wing begins the downstroke (Fig. 1 d-f). The downward movement of the wing may act passively to extend the wing fully. The dihedral angle of the wings is about 45 ° at the beginning of the downstroke (Fig. 1e). When the flapping bout ceases the gliding position is resumed with the upstroke terminating at body level. The half stroke that begins and ends a series of wingbeats seems not to differ from half the phase of a full stroke.

Relative durations of the upstroke and the downstroke vary along the length of the wing (Table 2). For an adult male in level flight the duration of the downstroke in the midwing area was approximately the same as that of the upstroke, while at the wingtip the downstroke occupied 61 % of the cycle period and the upstroke only 39%. For a wounded adult male flying only 1-2 m above the sea the upstroke duration at the wingtip was even more abbreviated, accounting for only 28% of the cycle period. The amplitude of the wingbeats seemed no less than that of a flap in normal flight; however, the wings did appear to flex more at the wrist and elbow on the upstroke. An immature male in take off also flapped with reduced upstroke durations at the wingtip. I could not clearly see midwing movement in the film segment on the wounded bird but for the condor in take off I found that the upstroke duration for that part of the wing was somewhat less than the downstroke duration.

The wingbeat movement in take off appears similar to that in level flight but the pattern in landing differs considerably. Flexion of the wrist and elbow is greater in the upstroke. The long axis of the wingtip path is orientated more in an anteroposterior than in a dorsoventral direction and most of this path is restricted to a zone above the horizontal body plane (Fig. 3a-g) Rotation of the path axis apparently serves to direct the wingbeat forces posteriorly so as to counteract the inertia of forward motion. Lift forces produced above the body support weight with a lower centre of gravity and provide more stability than would occur if the forces were below or level with the bird. Also, in this higher position the wings do not interfere with perch contact.

As the wings sweep forward in the downstroke the pitch of the body increases, perhaps in response to the downstroke thrust (Fig. 3 c). During the succeeding upstroke the pitch angle of the body decreases again (Fig. 3 e). As in level flight the elbow and wrist flex to begin midwing recovery before the primaries have completed the downstroke (Fig. 3,b, c). The angle of incidence for the proximal half of the wing decreases rapidly, approaching zero as the primaries begin the upstroke (Fig. 3d); near the end of the upstroke the angle becomes negative (Fig. 3e). The manus swings back and is extended while the leading edge is rotated downwards until the wing has about the same negative angle of incidence along the full length (Fig. 3e). Then the downstroke begins and the wings move forward and down, rotating at the same time to a positive angle of incidence. The alula rises in the downstroke (Fig. 3 a, b) and probably helps to retard boundary-layer detachment by deflecting air-flow downward over the midwing area. Coverts fluttering on the back of the proximal wing area and vibrating secondaries (Fig. 3 c) indicate, however, that stalling effects are present near the end of the downstroke.

To provide lift while landing the wings in the downstroke compensate for diminishing forward speed of the bird by increasing their angle of attack up to the stalling angle and by accelerating air speed across them through the flapping movement. Maximum lift forces are produced just before the boundary layer detaches in the stall. Apparently lift potential is approached and then exceeded for different spanwise sections of the wing at different stages of the downstroke ; the tip moves faster than the inner wing so the effect moves distally. The condor can control decelerations in either forward or downward directions by changing the angle of the downstroke. Air speed of the bird is reduced so the resultant relative wind striking the bird’s wing, particularly the tip, is largely dependent on the direction of the wingbeat.

Flapping can provide lift. An adult female condor with a full crop and an adult male with an empty one flew over Playa Chucho together ; they both maintained the same speed and altitude but the female flapped and the male did not. In another instance four adult males crossed the beach together maintaining the same altitude and air speed. One bird had a full crop and flapped and the others, all with empty crops, glided. A full crop increases the weight-loading ratio for the wings by a factor of nearly 10% and should act to accelerate forward and sinking air speeds. Flapping apparently offset these increases. On three different occasions I observed condors flap while circling, apparently in an attempt to reduce sinking speed below the rate of ascent of the ambient air mass. Increased forward air speed would require the bird to exert a greater centripetal force to maintain a circular path with the same radius and this, subsequently, would increase the sinking speed. Flapping, then, seems important to a circling bird by its capacity to reduce sinking speed without increasing forward air speed.

Relations between forward air speed and flapping rate for 42 condors and 5 turkey vultures (Cathartes aura) crossing Playa Chucho were not statistically significant (r = +0·14 and 0·03, respectively). Other instances where flight speed seemed to be independent of wingbeat frequency were reported by Greenewalt (1960) for a hummingbird and Tucker (1966) for a budgerigar. I did find, however, a significant correlation between the flapping rate of the condors and turkey vultures and their sinking speeds (Figs. 4, 5).

For the upstroke calculations I adjusted the value for the diameter from 2·77 m to 2·33 m to secure an estimate of the arc length traversed by the tip of a wing flexed to the degree shown in Fig. 1 (d). Taking the upstroke duration to be 39% of the wingbeat period I was able to compute an estimate of 8 m/sec for average wingtip speed in the upstroke relative to the body. Air-speed values relative to the body of 4 m/sec for both downstroke and upstroke were calculated for the midwing area with estimates of wing movement across an arc of 75° for a circle with a radius of half the wing-span. I estimated the duration of the upstroke at midwing as 43 % of the wingbeat period and the downstroke as 46% (Table 2).

To obtain estimates for the resultant wind velocity striking the wing during different phases of the cycle I first simplified the diagram of wingtip movement by reducing it to a trapezoidal configuration (Fig. 6). Then I computed the resultant wind vector for each of the four sides or phases of the stroke. The sides of the trapezoid approximate to the direction of movement, and the values for air speed produced by the wingbeat are only averages; thus the computations are, at best, rough and average estimates describing complex and rapidly changing conditions. They are valuable, however, for comparing large-scale differences and for illustrating how a wingbeat may provide lift.

The lift generated by an airfoil is directly related to the square of the air velocity passing over it. Thus, in the downstroke the change in air speed of 15 m/sec to 17 m/ sec at the tip, an increase in air speed of less than 25%, would act to increase the lift force by almost 50% assuming other conditions remained constant. It is unlikely that this substantial increase is offset by a reduction in lift during the upstroke. At the tip, for example, an air speed of 20 m/sec at the top of the upstroke compensates for the reduction to 10 m/sec at the beginning so that, in general, gliding air speed seems to be approximated by the average resultant air speed during the upstroke while it is exceeded during the downstroke. The duration of the downstroke is longer than that of the upstroke hence an increase in air speed across the tip predominates during the wingbeat period.

As the wing moves, conditions other than air speed change to affect the lift force provided; the angle of attack of the airfoil is a critical variable. To provide optimum lift forces the flapping wing must rotate or twist to accommodate the changing angle of the resultant relative wind especially at the tip where these changes are greatest. Vector differences in Fig. 6 suggest that the wingtip angle of incidence should tend toward the negative during the downstroke when the relative wind is directed upward more steeply. In the upstroke when the relative wind moves downward the angle of incidence providing the greatest lift must be positive. I examined the film and found that the wing posture conforms to this model (Fig. 1 a-g). In the downstroke (Fig. 1,a) most of the wing has a negative angle of incidence and the angle increases toward the tip. For the upstroke (Fig. 1 d) the angle is positive and, here as well, the effect is more pronounced distally.

At the tip each primary may act as a separate airfoil that effects and is affected by highly complex airflow conditions characteristic of wings with slotted tips. To secure more information on the relation of angle changes for wind and wing I used estimates for the midwing area where the pattern of airflow is simpler even though wingbeat effects are reduced. Vector quantities describing resultant wind for the midwing are shown in the inset of Fig. 6. I obtained an estimate for the angle of incidence during an upstroke phase by measuring certain features of the image projected from the frame of cine film used to trace Fig. 1 (d). Using a ratio of image wingspan to an estimate of the actual span I secured a scale for computing an estimate of 13 cm for the width of the projected midwing area. Into this value I divided 35 cm, an estimate of the length of the wing chord, to obtain an approximation for the sine of the angle necessary to project the surface area observed. Since the long axis of the bird and the flight path were directed straight away from the camera the full 22° angle computed should represent the angle of incidence. If the relative wind strikes the bird from an angle of 13° above the horizon as computed and illustrated in Fig. 6 then the angle of attack would be the difference; that is, 9°. If the secondaries bent upward on the trailing edge in passive response to the pressure forces then the angle would approach zero and lift forces would diminish. Instead the wing apparently remains rigid and in doing so maintains an angle of attack that should produce a substantial lift force. For an airfoil with an aspect ratio and profile like that of the condor wing an angle of 9° is within the range of angles of attack most effective, i.e. those associated with maximum glide ratios for a wide range of air speeds.

The wingtip path described for a condor wingbeat resembles that noted for other birds (Storer, 1948; Cone, 1968); the forward orientation of the downstroke is not peculiar to condors. If this axis were directed posteriorly 15° from the vertical instead of anteriorly as shown in Fig. 6 then the high air speeds attained in the downstroke would occur in the upstroke instead and relatively reduced air speeds would characterize the downstroke. Mechanical efficiency for the lever system of the downstroke is much greater than that of the upstroke because in the bird the muscles are ventral to the wings. Thus, the downstroke is the optimum stage of the wingbeat cycle for producing and withstanding maximum lift forces-the forces incurred with increased air speed over the surface area of fully extended wings.

In each of four different aerial chases when a pursued condor flapped more than the pursuer the distance between them increased in the horizontal but not in the vertical plane. The wingbeat pattern may have been different from that discussed previously or perhaps the bird flew faster with wings somewhat flexed while compensating for increased sinking speed by flapping. In another situation an alarmed female took off, flapping 53 times, to travel an air distance of 227 m horizontally and 14 m vertically (Table 4). In take off the relative wind velocity is less than in gliding flight and average resultant lift forces act in different directions. The wing encounters the wind at such steep angles of attack that the L/D ratios must be much less than those of level flight; even so, the resultant force has a positive thrust component directed toward the flight path during both upstroke and downstroke (Fig. 7). In the diagrams of Fig. 7 it can be seen that forces of an upstroke in level flapping flight produce a small component opposing forward motion. Although the L/D ratios are stylised in this diagram the effect would also be present with actual ratios that are larger; this explains why condors flapping across Playa Chucho maintained significantly lower air speeds than those that glided (P < 0·01, Mann-Whitney U test).

Although diagrams in Figs. 6 and 7 may describe general conditions for the major wing area the complex airflow pattern around the primaries probably produces wingtip forces with quite different vectors. Dissimilarities in individual wingbeat action also were not considered in this analysis. Substantial differences noted for descent distance related to flapping (Table 3) are probably due to variations in wingbeat movement, which in turn, are related to unique flight conditions.

In addition to these estimates of power required for flight I was able to secure one measure of the power generated by recording certain parameters of an escape flight by a startled condor (Table 4). This condor, a female, fed on a beach below a long ridge. Simultaneously, two assistants and I flushed her from different points so that her only escape route was over the ridge. She flew toward the ridge flapping steadily in the longest uninterrupted series of flaps that I ever recorded for a condor. Apparently unable to fly over the ridge she alighted near the top and ran to the crest. The maximum power output possible was probably approached in this exceptional effort. Considering both the power necessary for level flight and that expended in the ascent, the wing muscles of this female condor produced a minimum of about 54 W/kg muscle. The output estimate for this brief and exceptional effort is about half that recorded for a pigeon in similar circumstances (Pennycuick & Parker, 1966).

To evaluate the relationship between estimates of the power output generated by the exceptional effort of this female and the output necessary for sustained flight I examined similar data on human muscle. In a brief and maximum effort, human muscle can produce a power output of about 40 W/kg muscle (Dickinson, 1928; Parry, 1949) while the maximum sustained output has been calculated at about 17 W/kg muscle (Henderson & Haggard, 1925; Parry, 1949). If this output ratio for exceptional and sustained efforts (40:17) is similar for condor muscle then about 23 W/kg muscle are available for sustained flight, assuming that the mechanical efficiency of the flight systems for exceptional and sustained efforts are similar. Apparently then, less than the minimum requirement of 33 W/kg muscle necessary for level flight are present; oxygen would not be supplied fast enough to maintain the necessary power output. Behavioural data support the hypothesis that the condor is incapable of sustained level flight in still air; in situations where the air seemed still I never saw a condor fly for an extended period without losing altitude rapidly.

I obtained no comparable data for maximum outputs in exceptional efforts by turkey vultures so I could not obtain similar estimates for this smaller bird. Three other comparisons, however, indicate that the turkey vulture probably can maintain level flight in still air for longer periods than can the condor. (1) The estimates for the minimum power output/kg wing muscle necessary to sustain level flight are greater for the condor than for the turkey vulture largely because a greater proportion of the turkey vulture mass is composed of wing muscle. (2) The average mass of turkey vulture per unit of projected wing area is about half that of the condor (McGahan, 1972). This twofold difference in weight loading is significant in considering the effect of flapping on sinking speed; assuming similar conditions for both flapping systems the condor must flap at a greater rate than the turkey vulture to obtain the same effect. (3) Some evidence for this difference in flapping efficiency is presented in the regression analyses of sinking speed and flapping rate. Changes in the flapping rate appeared to affect sinking speed of the turkey vulture more than that of the condor (Figs. 4, 5). By dividing the regression coefficient for flapping rate and sinking speed into estimates of the average sinking speed I obtained values of 160 wingbeats/min for the condor and 75/min for the turkey vulture as rough estimates of the rates necessary to maintain level flight (Table 5). The data in this comparison are in good agreement with expectations projected from the weight loading comparison but they can only serve as possible indications since the difference between the two regression coefficients is not statistically reliable.

In sum, on the basis of flight performance and physical dimensions, the turkey vulture appears capable of more extended flights in still air than does the condor. Whether or not the smaller vulture can maintain level flight for extended periods in still air depends, of course, on the maximum level of sustained power output by the muscles and the efficiency of the flapping system.

In addition to those acknowledged in the preceding paper I am grateful to Betsy Brauer for her help in the analysis of muscle power output.