Gliding Flight of the Andean Condor in Nature

The Andean condor (Vultur gryphus) is an excellent subject for an analysis of flight. To fly, these big vultures depend on updraughts which in many areas are strongest near cliff faces and canyon walls. From cliffside observation points qualitative aspects of flight by this large bird can be studied in detail at close range while quantitative treatments of flight performance are possible at sites along foraging routes where air movement is steady and easily monitored. This paper presents a quantitative description of the condor’s gliding flight as recorded at one of these sites.

All data describing the aerodynamics of gliding flight were recorded at a Peruvian beach, Playa Chucho. This level south-facing beach is located about 30 km south of Pisco and 1 km east of the village Laguna Grande. The beach extends 620 m east west, separating a series of sandstone cliffs that ascend abruptly to 150 m on the west and slope gradually to a 70 m elevation on the east. Before rising in a series of small hills, 5–15 m high, the beach stretches south into the desert as a flat for 300 m on the west end and 120 m on the east. Condors flying along the coastline crossed the beach in straight smooth flight. At one end of the beach they left declivity currents produced by the cliff-deflected south wind to encounter them again at the other end. This predictable flight path in conjunction with the steady wind characteristic of the Peruvian coast provided an opportunity to measure parameters of their flight.

The observation segment of the beach was marked by two rock piles 60 cm in height and 250 m apart (Text-fig. 1). I stood midway between them and when a condor passed over each marker I recorded the lateral distance to the nearest metre separating the flight path projected on to the beach from the marker. My wife, posted 145 m north of and in line with the tower, recorded the angle of the bird’s position above each marker using a sextant mounted on a swivel-head tripod. The sextant, a vertical scale with 80 2·5 mm divisions mounted on a horizontal base 58·5 cm in length, permitted angular measurements used to compute values for altitude to the nearest metre. The tripod and sextant were adjusted so that for each sighting the base of the vertical scale was in line with the top of the marker. For each sighting I used the tangent of the angle recorded with the sextant and the distance measured from the sextant to the flight path projected on to the beach to calculate the altitude of the bird. Considering possible errors in measurement of both angle and distance I estimate each altitude value to be accurate within 3 %.

The duration of the flight interval from one marker to the next was measured to the nearest 0·1 sec with a stopwatch at the inland observation point. Simultaneously, changes in wind conditions were recorded on series of gauges cabled from a 5 m tower standing near the centre of the beach below the flight path. The tower supported a wind vane, a cup anemometer sensitive to horizontal currents and a propeller anemometer sensitive to vertical currents. With data on the velocity of air and bird relative to the ground, I used the law of cosines in a vector analysis to compute values for air speed of the bird. Measured values for the ground distance and time travelled are believed accurate within 2%, values for wind speed within 20%, and computed values for air speed of the condor within 4%.

where V is the air speed. The mean sinking speed and computed standard error for the gliding condors were 0·5 ± 0·08 m/sec. Differences in air speeds and sinking speeds according to sex were not significant. Similar data recorded on five turkey vultures (Cathartes aura) are also presented in Table 1.

where c is the average wing chord length and u is the viscosity of the air. Both air viscosity (u) and density (ρ) are temperature- and pressure-related ; in this study the ratio p\u used varied from 62500 to 65400.

The mean and standard error for lift coefficient calculated for the 15 gliding condors were 0·7 ± 0·04. Lift coefficients are dependent on wing shape and position and describe the effects produced by the angle of attack, the aspect ratio, the profile of the wings, and fluid viscosity of the air. In equilibrium gliding C_L is inversely related to air speed. Condors probably employed a narrow range of optimum speeds in crossing Playa Chucho ; I was unable to obtain estimates for minimum and maximum limits for C_L.

Data recorded on five turkey vultures presented in Tables 1 and 2 have been treated like the data for condors with one exception. I assumed that the general shapes of turkey vulture and black vulture were similar and estimated the wetted surface area of the turkey vulture by multiplying the value computed for the black vulture (Parrott, 1970) by the squared ratio of wing spans. Since values for the wetted area of the condor determined by this method and the method mentioned above differed only by 1 %, I expect the error for the turkey vulture estimate to be insignificant.

Information recorded on condors and turkey vultures that flapped at some time while crossing the observation section of the beach also provided estimates for the forces and coefficients of parasite drag. I corrected sinking-speed values for these data, however, since flapping rate and sinking speed are related (McGahan, 1972). In an analysis of regression with flapping rate as the independent variable I computed mean sinking speeds of 0·5 m/sec for condors and 0·6 m/sec for turkey vultures for a projected flapping rate of zero.

The C_Dp values I obtained for gliding condors were implausibly low. With mean data on air speed and sinking speed I computed a C_Dp of 0·001 for the gliding female condor; for the male I obtained negative values for parasite drag forces, an impossible condition (Table 3). estimates from mean data on flapping birds were also low. I calculated 95 % confidence limits for the mean air speed and sinking speed observed and combined these upper and lower interval values to secure maximum and minimum estimates for the glide angle, and indirectly the coefficients of parasite drag. So, for example, I divided the lower confidence interval value of 13·6 m/sec for the mean air speed into the upper confidence interval of 0·68 m/sec for sinking speed corrected for zero flapping, to calculate a maximum estimate for the glide angle of flapping condors. For these maximum angles, C_Dp estimates were 0·003 for the flapping male condor, 0·005 for the female, and 0·020 for the flapping turkey vulture. All estimates for the condor are exceptionally low, either less than or within the range of optimum values defined for parallel airflow across a smooth flat plate (Table 3). Only the maximum estimate for the turkey vulture approximated to the values determined for the black vulture in Parrott’s wind-tunnel study. Apparently some source of error has not been taken into account.

I found that relatively small errors in the weight estimates for heavier birds produced substantial differences in the values. In computing total drag, weight values of the first power are used compared to second-power figures used in calculating induced drag. Thus, if the weight given for female condors is an overestimate of 1 kg or 12% then the C_Dp, value for flapping female condors would be twice that shown. An error of similar proportions in the weight of the smaller turkey vulture changed the by only 2%. This may help explain why data for the heavier condors, particularly the male, are more deviant.

Two other sources of error could account for the general trend toward low C_Dp values. (1) If a vertical wind component existed in the flight path area above the level of the anemometer tower then the values for glide angle θ would be unduly low and would provide underestimates for the total drag as calculated with equation (2). (2) If the assumption of equilibrium gliding were unfounded and condors were decelerating during the period of observation then they could glide at angles less than possible in equilibrium situations. A third possibility, where glide angle is reduced by an acceleration of wind acting in concert with inertial forces, is unlikely since the velocity of the even coast wind varied little if at all across the short span of the flight intervals. To obtain expected values for total drag force I calculated new values for parasite drag with equation (8) using a modified C_Dp value derived for the black vulture (Parrott, 1970). Although C_Dp values for the black vulture were secured at Re values lower than those recorded for the condor I adjusted the C_Dp value according to the factor K (Tucker & Parrott, 1970) for Re values that ranged from 3 · 3 × 10⁵ to 3 · 6 × 10⁵. The factor K is a ratio of C_Dp for the bird to that of a flat plate in parallel but turbulent airflow at a given Re value. I multiplied a K value of 2 · 2 derived for the black vulture (Tucker & Parrott, 1970; Parrott, 1970) by the C_Dp value for a flat plate at the Re computed for gliding and flapping condors (Table 3) to secure an adjusted of 0 · 012 for the condor. For the turkey vulture I used a C_Dp of 0 · 015, a value midway between the extremes obtained for the black vulture since the Re for the turkey vulture were within the range noted in Parrott’s study.

Three hypotheses were presented. (1) An undetected vertical component for wind above the tower should be stronger when the horizontal wind velocity increased. If this error contributed the discrepancies would be positively related to the horizontal wind speed. (2) Decelerating condors probably reduce their forward speeds to an optimum level for crossing the beach. Hence I would expect birds flying faster to decelerate more on the average, predicting a positive correlation between values for deceleration and air speed. (3) Condors could glide at steeper angles to attain above average speeds. If most of the condors started at comparable altitudes then those decelerating after gliding temporarily at a steeper angle would start across the beach at a lower initial altitude. A negative correlation between the discrepancy for drag forces and the initial altitude of flight would serve as evidence for the presence of deceleration forces. One of the 15 gliding condors actually gained altitude while crossing the beach ; this bird also started at the lowest altitude, 39 m. The condor may have dropped sharply just before entering the observation section and then, while decelerating, crossed it without descending-a typical pelican flight pattern seen when they skim along just above the water surface without flapping.

No correlations were significant for the data on the 15 gliding condors. For the flapping condors I used a regression coefficient for flapping rate and sinking speed (McGahan, 1972) to correct the vertical wind and deceleration units to levels of zero flapping. Then with these corrected values I computed correlation coefficients with the same variables used with the gliding condor data. First, I examined the association between the flapping rate and each of the test variables to check for any bias introduced by the correction factor for flapping. The correlations were not significant and in each case the bias acted to retain the null model. For units of deceleration satisfying the apparent drag discrepancy I obtained a negative correlation with the altitude of the flight path (r = – 0 · 13) and a positive correlation with flight speed that was significant (Text-fig. 2). The significance of the last coefficient provides some evidence that deceleration was responsible for the low values of C_Dp I obtained.

Although this analysis provided no evidence for the presence of an undetected vertical wind component, one field observation did. A turkey vulture turned around twice within the central third segment of the observation beach, glided above me three times without flapping during a 1 min period, and lost only 5 – 10 m of altitude. The horizontal anemometer recorded a wind speed of 8 m/sec but the vertical anemometer registered nothing. Deceleration in this case could not explain the unusually low sinking speed; a vertical wind component must have been present above the tower. An average difference of 8 ° in the angle of streamlines in the boundary layer at the level of the tower and the flight path could account for the theoretical discrepancy in the drag forces for the gliding condors. By dividing 5 m/sec, the mean horizontal wind speed for the observations (Table 1), into 0 · 7 m/sec, the mean vertical wind component expected (Table 4), I obtained an estimate of the tangent of the streamline angle necessary to explain the drag differences. A similar computation using 0 · 5 m/ sec, an expected value for the vertical wind in the absence of flapping, and 3 m/sec, the mean horizontal wind speed, for observations of flapping condors provided an estimate of 10 ° for the difference in streamline angles. From data on flapping turkey vultures I obtained a value of 9 °. These three sets of data recorded during periods with generally different wind conditions provided three similar estimates. This hypothetical flow pattern for different levels in the boundary layer of air striking a coastline would not be unusual.

Flex-gliding, a term from Hankin (1913), designates a gliding posture where the wings are partially flexed in the horizontal plane. In bending both wrist and elbow condors altered the wing configuration across a range from near full extension to some positions where the span was almost halved (Plate 1C). Moderate flexion characterized smooth straight gliding journeys over long cross-country distances. Wing area is reduced by overlapping the primaries in flexing the manus and by relaxing the patagium in bending the elbow (Text-fig. 3). Simultaneously the tail usually contracts to a more closed position. Circling condors generally initiated long cross-country flex-glides with a dipping motion by both manus ; then, the wings and tail, which were expanded during the circling ascent, moved to the flexed position as the bird began the straight descending glide.

Many of these straight flights, uninterrupted by bouts of flapping or circling, extended over periods of 5 min. Recorded durations of four flex-glides were particularly long: 7 min 40 sec, 9 min 42 sec, 12 min, and 14 min 40 sec. In the last observation the condor, before disappearing in the distance, had traversed about 13 km ground distance and had lost only one-fourth of the 1000 m altitude gained in a circling bout prior to the flex-glide. Hypothetically, if the bird continued to encounter the same air conditions he could have travelled 50 km in a period of about 35 min, a potential journey made possible by ascent in a circling bout only 8 min in duration. In this particular observation I was standing below the midpoint of the flight path described. Values for altitude were obtained by combining estimates of the angle of the bird’s position above certain mountains or ridges with data from maps on the altitude of the landmark and its distance from the observation point. I was often able to determine the bird’s position relative to the landmark by locating the bird’s shadow on the ground and then using the sun ‘s’line of sight’ for triangulation.

provided by Alexander (1968), where R is the aspect ratio and a parasite drag coefficient for wetted surface in terms of the projected wing area. I used an air density value ρ = 1 · 18 kg/m³, a parasite drag coefficient C_Dp = 0 · 012, and the mean dimension values for male and female condors given in Table 2 to compute estimates for the air speed of condors in the two flight postures shown in Fig. 3.

According to these computations an adult male with wings fully extended would glide with minimum sinking speed when his forward air speed was 13 · 9 m/sec. Upon flexing them to the degree shown in Text-fig. 3 (a span reduction of 20%) this optimum forward air speed would increase to 15 · 2 m/sec. Corresponding values for the adult female are 12· 6 and 13 · 8 m/sec, respectively. These figures can serve only as rough estimates, however, since C_Dp variation at different speeds is neglected as well as airfoil effects of the tail.

An ability to change forward air speed provides the condor with a flexibility for crossing areas that differ in the types of air movement, food availability and potential danger. Efficient increases in air speed are necessary for flight against a headwind; the condor must sacrifice the shallow glide angle of a moderate air speed for a speed providing some forward progress relative to the ground. As air speed increases the condor can minimize increased sinking speed by reducing airfoil area. Once I saw an adult female circling in a 40 km/h wind flex her wings and tail more while travelling upwind than downwind, apparently accelerating upwind flight to reduce ground speed downwind. Five condors flushed from a cliffside roost circled up about 500 m above the sea and then, in postures flexed to the degree shown in Text-fig. 3 (dotted line), glided in steep descent south-east along the coast heading against an evening cross wind of 30 km/h. Ground speeds were noticeably reduced compared to those observed when wind speeds were less. Every 20 – 30 sec each bird flapped in a bout of three wing-beats. Presumably these bouts served to reduce sinking speed (McGahan, 1972) and thus extended the potential endpoint of the descending flight path farther up the coast.

Frequently, condors flying in intense declivity winds near cliffs glided with strongly flexed wings and lowered feet (Plates 1 A, C, 2). This posture occurred in flight patterns preceding landing and often during periods when the birds flew near my observation point and inspected me carefully. Fluttering wing covert feathers (Plate 2) and raised alulas (Plate 1 A) indicate that the angle of attack with the steep vertical wind is large enough to detach part of the boundary layer of air moving over the wings. These stalling effects coincide with an increase in sinking speed and a reduction in forward speed. Lowered feet can function as air brakes to retard forward air speed. In an air mass that ascends rapidly the bird is permitted greater sinking speeds without losing altitude; then, the reduction in forward speed can augment conditions for examining objects on the ground, enhance the precision of landing manoeuvres, or provide prolonged access to local declivity currents of limited size. I watched an adult male advance with a fairly constant ground speed of only 0 · 5 m/sec for several minutes while flex-gliding near the edge of a cliff in a 35 km/h headwind. Another adult male in a 30 km/h headwind both advanced and ascended at the rate of 0 · 3 m/sec. One condor, flex-gliding in a deflexion current, maintained a high pitch angle of 35 ° and climbed in steep almost vertical ascent without stalling. Sometimes in strong deflexion currents condors with fully extended wings made little forward progress relative to the ground for extended periods; in one instance a condor, apparently gliding in the updrafts of a beach-deflected wind, did not move, except for a gradual descent during a 40 sec period. Then she glided downwind, ascended near the face of a bluff, and returned again to assume this stationary glide above a carcass on the beach. Advancing slightly, another condor descended for 1 min at a mean rate of 2 · 4 m/sec in a straight path that formed a 70 ° angle with the ground. The wings were completely extended and pitch was parallel to the horizon.

Increasing the weight loading ratio of the airfoil by flexion may have functioned at times to provide more stability for condors flying in turbulent updrafts. In pursuits and in flight patterns in restricted areas temporary flexed-wing postures provided manoeuvrability. Acceleration forward and down succeeded wing flexion, and deceleration succeeded extension. Sometimes circling condors alternated between extended and flexed wing postures; in alternating between altitude gains and losses they tended to remain in the same general area for extended periods. Eight times an immature female alternately ascended and descended in a coastal declivity current. Flexing her wings to half span and simultaneously lowering her feet she began a descent losing about 10 m altitude; then, extending her wings fully and lifting her feet back to the position against the body she initiated the ascent that carried her back to the original level. This series of manoeuvres was conducted while circling in the same general area.

I am indebted to my wife, Libby, for her many hours of help in the field and on the manuscript. Karen Craighead assisted in the field and David Thompson, Dr John Emlen, Dr John Neess, Dr John Magnuson and Daniel Smith provided editorial assistance. For their help I am most grateful. Figure drawings are by John Dallman and Cheryle Hughes. This study was supported by a grant (GB-19449) from the national Science Foundation and a fellowship from the Danforth Foundation.