Kinematics and aerodynamics of the greater horseshoe bat, Rhinolophus ferrumequinum, in horizontal flight at various flight speeds

A gliding animal loses height because the wings generate a vertical force which supports the animal’s weight, but does not counteract drag - the sum of wing friction (profile drag), body friction (parasite drag) and the drag associated with wake vortex formation (induced drag). By flapping its wings the animal alters the direction of the force generated so that drag is counteracted and level flight sustained. As flight speed increases the relative magnitudes of the three drag components change, the relationship between flight speed and total drag taking the form of a U-shaped curve (Pennycuick, 1975). To sustain level flight at a range of speeds, the animal must alter its wingbeat kinematics so that thrust generated equals drag at all speeds.

In Columba livia the changes from one kinematic pattern to another appear to occur gradually in response to the gradual changes in the magnitudes of drag (Brown, 1948, 1953; Pennycuick, 1968). In flight immediately after take-off, the bird uses a ‘tip-reversal’ upstroke during which the primary feather region of the wing generates both weight support and thrust, only weight support being generated during the downstroke (Brown, 1948). When flying very slowly the secondary feather region is probably not moving fast enough to generate any useful aerodynamic forces (Boel, 1929; Brown, 1948). This type of upstroke has also been reported in bats, although it was thought that the upstroke force is essentially horizontal, taking the form of supplementary thrust (Eisentraut, 1936; Norberg, 1976; Altenbach, 1979; Aldridge, 1985a). At higher speeds C. livia uses a ‘feathered’ upstroke, where the wing is brought close to the body, with the primary feathers rotated so that they present a ‘Venetian blind’ appearance to the oncoming air. The wing is only active during the downstroke, the bird’s wake taking the form of a series of small-cored vortex rings (Kokshaysky, 1979; Spedding, Rayner & Pennycuick, 1984). This type of upstroke has not been described in bats and it has been suggested that this is because bats are unable to slot the handwing membrane (Aldridge, 1985b), although by maintaining incidence angles at or around zero during the upstroke the noctule Nyctalus noctula generates the same type of wake (Rayner, 1986). As the pigeon’s flight speed approaches normal commuting speeds the flight kinematics change again; the wing remains partially extended during the upstroke and force is generated throughout the wingstroke (Brown, 1948). The kestrel, Falco tinnunculus, and N. noctula both use this type of upstroke (the so-called ‘reduced span’) in cruising flight, force being generated as a result of the generation of a pair of distorted parallel vortices (Spedding, 1981).

The purpose of this paper is to address two questions concerning the flight of R.ferrumequinum. (i) What are the kinematic and aerodynamic changes accompanying increases in flight speed and (ii) are these changes gradual or do they occur abruptly?

Two male R. ferrumequinum were obtained and kept in captivity for 18 months. They were housed in a plywood box (0·6m × 0 6m × 0·6m), the walls of which were lined with non-toxic plastic netting. In the bottom of the box there was an aluminium tray lined with moist paper towelling which ensured that the relative humidity did not fall below 80%. The temperature in the animal room followed ambient unless it dropped below 10°C, when supplementary electric heating was automatically turned on. This ensured that the bats could enter torpor regularly but did not go into hibernation. The bats were fed on a diet of mealworm larvae (Tenebrio molitor), occasionally supplemented by insects caught in a mercury-vapour light trap. A vitamin supplement (Vionate) was added to the bran in which the mealworms were kept and sprinkled sparingly over the food before presentation. Water was always available, the bats quickly learning to drink from a normal small mammal water bottle.

One bat was trained to fly through a flight tunnel on the execution of an auditory command (the clicking of my fingers). When the bats were first brought into captivity they had to be restrained when being fed or watered. After a few weeks they could be fed without restraint, but had to be captured (with a hand-net) and placed upon the feeding perch before they would take food. By this stage they had learnt to take water from a small mammal water bottle suspended in their box. Presentation of food was always accompanied by the auditory signal, and eventually both bats would respond by flying towards it. Only one, however, would land on the feeding perch, the other catching mealworms thrown into the air. The bat which landed on the feeding perch could be handled with ease, and it was this individual that was used for the experiments. The bat’s morphological parameters are presented in Table 1. Experiments were only carried out before the bat was fed, thereby ensuring that the bat was alert to the auditory signal. By clicking my fingers I could call the bat to the feeding perch and from there it could be carried to the flight tunnel where it was placed on a second perch and given a small mealworm fragment. After eating the fragment the bat would normally stay hanging from the perch grooming. By repeating the auditory signal the bat could be induced to fly through the tunnel to the feeding perch, where it was rewarded with another mealworm fragment; the procedure could then be repeated. The bat could be used in experiments for long periods without a noticeable decrease in performance or alertness, as long as only small pieces of mealworm were given as reward.

The photographic technique used by Rayner & Aldridge (1985) to analyse the turning flight of Plecotus auritus was also used in this study. The bat flew through the flight tunnel, triggering the two cameras (Nikon FM plus MD11 motordrive and Nikon FM2 plus MD12 motordrive, both with Nikkor 55 mm Micro lenses) by interrupting a light beam. The two cameras were arranged on mutually perpendicular axes (the FM suspended above the flight tunnel, the FM2 to one side) (Rayner & Aldridge, 1985). The camera shutters were set to open for up to Is while the bat flew through the tunnel, illuminated by a stroboscope (Dawe Instruments 1203c ‘Strobosun’) flashing at 100 or 200 Hz. A series of cotton threads was arranged within the flight tunnel so that they defined the corners of a 0·4 m cube (Rayner & Aldridge, 1985). The threads were permanently arranged, and therefore appeared in every pair of photographs. For each flight two multi-image photographs were produced, one showing the dorsal view, the other the lateral view of the bat’s flight path through the tunnel. The data were put into a Research Machines 380Z microcomputer through a Summagraphics Bit-Pad One digitizing tablet.

For each flight the positions, in three dimensions, of the body (represented by the eye in the photograph showing the lateral view and a point midway between the ears on the photograph showing the dorsal view), the wrist, the tips of the second, fourth and fifth digits and the wingtip (tip of the third digit) (Fig. 1) were obtained by resection for a sequential series of images (Rayner & Aldridge, 1985). From these data five kinematic parameters were calculated: flight velocity (V), wingbeat frequency (f), wingbeat amplitude (θ) and the strokeplane angles for the wrist (γ_w) and the wingtip (γ_wt). The velocities of the wrist, wingtip, second, fourth and fifth digits were calculated as described by Rayner & Aldridge (1985), total wingtip velocity (V_t) being defined as :

where v_x², v_y² and v_z² are wingtip velocities in the forward, lateral and vertical directions, respectively.

Wingbeat frequency was calculated by counting the number of images over a whole wingbeat, starting and finishing at the beginning of the downstroke. Wingbeat amplitude is here defined as the angle moved during the downstroke by the proximal portion of the wing (defined as a line drawn between the point representing the body and the wrist) (Pennycuick, 1968). The movement of the wing during the downstroke can be approximated to movement in a single plane, and is defined by a line drawn between wing position at the beginning and end of the downstroke (Pennycuick, 1968). The strokeplane angle is the angle between this plane and the horizontal. In this study two estimates of strokeplane angle were calculated, for the wrist and the wingtip (Scholey, 1982).

The accuracy of these measurements is determined by the accuracy of test cube construction, optical distortion, digitizing errors and difficulties in determining image locations on the photographs. Rayner & Aldridge (1985) have estimated the maximum possible error in point position estimation as being between 2 and 4 % of any linear distance, as long as the cameras are positioned not more than 10° away from the ideal camera axes (Rayner & Aldridge, 1985). In practice it was found that a test cube was relatively easy to construct accurately and by using a spirit level or a plumb line the cameras could be arranged with tilt angles of less than 5°.

The absolute accuracy with which the kinematic parameters can be determined is dependent upon the frequency of sampling. Ideally the interval between images should be as small as possible so that rapid changes of position or speed can be resolved. A limit was set on this interval by (1) the quantity of light delivered in each flash, which falls as frequency is increased, and (2) the need for each overlapping image to be distinguished. The maximum practical flash rate was found to be 100 Hz although in some cases 200 Hz could satisfactorily be used (the higher the flight speed, the higher the frequency had to be to produce enough images for analysis). Aldridge (1985a) recorded maximum wingtip speeds in R. ferrumequinum of approx. 13ms^-1, and at these high speeds the wingtip could move as much as 13 mm between consecutive flashes. Fortunately the mean wing speed is lower than this (approx. 5ms^-1) but at critical stages during the wingbeat, such as the beginning and end of the downstroke, the discrepancy between the estimated positions of the wingtip and wrist and their true positions may be as much as 4%. This source of error will be most noticeable in the estimates of wingbeat amplitude.

An indication of wing function can be obtained from estimates of the changes in incidence angles at regular stages during the wingbeat cycle. A wingchord is defined here as being a straight line between two points on the leading and trailing edges of the wing, respectively. For this investigation two wingchords were used, the wrist chord, a line drawn between the wrist and the tip of the fifth digit, and the wingtip chord drawn between the tips of the second and fourth digits (Fig. 1) (Norberg, 1976). The positions, in three dimensions, of the two wingchords were calculated at regular intervals, and the angles of incidence for both chords were calculated by determining the local velocity vectors of the two ends of the chords, and finding the angle between the chord and a line parallel to the mean of the two velocities by vector methods.

The relative airflow (V_ri) is the resultant of the bat’s forward speed (V), the wing’s flapping velocity (Vf) and the induced velocity (V_i) (Fig. 2). Induced velocity is the vector sum of bound and wake vorticities which, without techniques which either visualize or measure airflows around flying animals (e.g. Kokshaysky, 1979; Spedding et al. 1984), are difficult to estimate. Using the Rankine—Froude momentum theory of propellers it is possible to obtain an estimate of mean vertical induced velocity. V_i can be defined by the formula:

where W is the bat’s mass, S_d is wing disk area and ρ is air density (Pennycuick, 1975). Equation 2 assumes that V_i is constant across the wingspan, and throughout the wingbeat.

At high flight speeds the resultant airflow is dominated by V and the effect of V_i is probably negligible, and for this reason V_i was ignored in my estimations of incidence angles. As flight speed decreases, wake-induced velocity reaches a maximum and V_i becomes increasingly unreliable. In hovering animals, when wake-induced velocity is highest, the bound vorticity of one wing is likely to affect that of the other, thus altering the direction and magnitude of the instantaneous induced velocity and making the discrepancy between estimates of V_i and their true values very large (Ellington, 1984a). In this situation the estimates of incidence angles obtained are likely to be very inaccurate, and therefore result in serious error in the estimation of the direction and magnitude of the forces generated. Assuming that the ratio between. mean induced velocity and forward speed is likely to be highest in a slow-flying bat, I estimated V_i for R. ferrumequinum flying at 2·7ms’¹ and calculated incidence angles with and without this value, to see whether mean induced velocity had a significant effect on it.

Using equation 2, values of V_i for R. ferrumequinum flying at 2·7 ms^-1, 3·1 ms^-1 and 4·8 m s^-1 were calculated and the results are presented in Table 2. In Fig. 3 the angles of incidence for R. ferrumequinum flying at 2·7ms ¹ for V_i (calculated without an estimate of V_i) and V_ri (calculated with V_i = 0·39 m s^-1) are plotted.

When flying at 2·7ms^-1 the downstroke is initiated with the wing fully extended, and held at an angle of 54° above the horizontal (Fig. 4A). The wing is accelerated forwards and downwards reaching a maximum wingtip downstroke speed of 6·9ms^-1 midway through the downstroke (Figs 6, 9). When the proximal portion of the wing reaches 36° below the horizontal, it begins to rise, the wingtip continuing to move downwards and forwards for a further 0·01 s. Flexion of the wrist at this point accelerates the wing inwards; overall wingtip speed is, however, falling (Fig. 9). Midway through the downstroke incidence angles become positive and high (Fig. 12). After a further 0·02s the wing begins to decelerate (Fig. 9), and the incidence angles begin to fall (Fig. 12). The upstroke starts with the elbow and wrist fully flexed and the digits fully extended (Fig. 4 f)- As the upstroke proceeds the elbow and wrist extend, and as a result the handwing is extended perpendicular to the long axis of the body (Fig. 4g-h). This movement is accomplished rapidly as the wing is accelerated upwards and backwards (Fig. 9), the angles of incidence becoming momentarily negative and large (Fig. 12).

At 3·1 ms^-1 the bat moves its wings in essentially the same manner. There are, however, a number of significant differences. The upward and backward movement of the wing during the upstroke is reduced (Fig. 7) and there is no dramatic upstroke acceleration of the wingtip (Fig. 10). The angles of incidence are lower during the downstroke, and do not become large during the upstroke (Fig. 12).

The changes observed at 3·1 m s^-1 are more noticeable at 4·8 m s^-1. The backward movement of the wing during the upstroke is now totally lost, and the strokeplane becomes vertical (Fig. 8). The angles of incidence during the upstroke are large and positive (Fig. 12). There is no marked upstroke acceleration, wingtip speed remaining almost constant throughout the wingbeat (Fig. 11). At the beginning of the downstroke the wing is not fully extended, extending during the first half-stroke (Fig. 5a-c). The wing remains partially extended during the upstroke and, as seen from above, the wingchords appear almost parallel to the long axis of the body (Fig. 5e-g).

As the bat’s flight speed increases there is a significant, but small, fall in its wingbeat frequency (b = –0·71, r = –0·39, P<0·01) (Fig. 13A), and a significant increase in wing strokeplane angles (b = 3·5, r = 0·33,.P<0·01 for the wrist and b = 5·71, r=0·33, P<0·01 for the wingtip) (Fig. 14). There is no significant difference between the slopes of the two regression lines (F = 2·2, P> 0·05). There is no significant correlation between wingbeat amplitude and flight speed (r = –0·05, P>0·05) (Fig. 13B), although there is a tendency for amplitude to be lower at higher speeds (Table 2). As discussed above, of all the kinematic parameters, amplitude is probably the most sensitive to errors in pinpointing the exact beginning and end of the downstroke, and this may be the reason for there being no significant trend.

A flying animal flaps its wings primarily to generate thrust; if circulation is kept constant throughout the wingbeat no net thrust will be generated, the forces generated during the up- and downstrokes having equal magnitude but opposite sign. To generate net thrust it is essential that there is asymmetry between the up-and downstrokes. By changing either incidence angles or wing planform geometry, the animal can reduce the magnitude of upstroke thrust and therefore achieve wingbeat asymmetry. At low and high speeds [speeds below minimum power speed (V_mp) and above maximum range speed (V_mr)], when drag is high, upstroke wing angles of incidence should be low and no force should be generated (Rayner, 1986). At intermediate speeds, when total drag is low, the upstroke could be used to generate weight support, net thrust ensured by the reduction of wing planform geometry. The fact that the changes in drag are gradual suggests that there should also be a gradual change in upstroke function as flight speed increases.

When flying at 2·7ms^-1R. ferrumequinum uses a ‘tip-reversal’ upstroke; the proximal region of the wing is highly flexed with wristchord angles of incidence remaining low, suggesting that no aerodynamic work is being done; the digits are extended into the airstream and wingtip chord angles of incidence become momentarily large and negative, suggesting that a large propulsive force is being generated. Theoretically at low flight speeds the upstroke should be inactive, because induced drag is high (Rayner, 1986). However, Brown (1948) points out that because the wing is moving in the opposite direction to the body, drag will tend to propel the animal forward. Also the bat may have to generate a propulsive force during the upstroke because the force generated during the downstroke is not large enough to counteract drag, only the wingtip region of the wing generating any lift (Boel, 1929).

Brown (1948) predicted and confirmed (Brown, 1953) that as the pigeon’s flight speed increased two processes should occur; (1) the backward ‘flick’ of the wingtip should be reduced and (2) there should be a transfer of lift generation to the downstroke. This sequence was also observed in R. ferrumequinum. When flying at 3·1 ms^-1R. ferrumequinum uses a ‘feathered’ upstroke, during which no force is generated. Presumably the increase in the wings’ speed during the downstroke increases the force generated, and as a result reduces the need for upstroke thrust generation. At this speed (above V_mp) induced drag is still high and therefore no advantage would be gained by upstroke lift generation.

The minimum power speed (V_mp) for the individual used in this study is 4·84 ms^-1. When flying at this speed the total drag on the bat will be minimal. The upstoke is therefore used to generate a backward- and upward-directed force, net thrust being ensured by upstroke wingspan reduction. The advantage of this kinematic pattern is that by generating lift over the whole wingbeat cycle the bat can reduce downstroke wing angular velocity and therefore flight muscle power output.

As the bat’s flight speed increases there is a significant decrease in wingbeat frequency and an indication of a similar decrease in amplitude, which will result in a decrease in flight muscle power output. Presumably the increase in flight velocity and the decrease in drag enable the bat to reduce flight costs while ensuring that weight is supported and drag counteracted. The fact that wingbeat frequency changes with speed suggests that the flight muscles of R. ferrumequinum contain mixed fibre populations, similar to those found in Hipposideros speoris (George & Naik, 1957), because muscles consisting of a single fibre type should contract over a narrow frequency range to maximize efficiency (Goldspink, 1977).

The range of incidence angles at which a wing generates steady-state lift is determined by the wing’s zero-lift angles and the angle at which steady-state stall occurs. As V increases the angle between V_ri and the horizontal decreases, assuming that V_f remains constant and unsteady phenomena are unimportant. The bat alters wing strokeplane angles to ensure that wingchord incidence angles remain between the zero-lift angle and the steady-state stall angle as the direction of V_ri changes. Fig. 12 shows that as the bat’s flight speed increases there is a slight reduction in wingchord incidence angles, which may be a consequence of the lower thrust required as V approaches V_mp.

The advance ratios for R. ferrumequinum at 2·7ms^-1, 34 ms^-1 and 4·8ms^-1 are 0·38, 0·54 and 0·93, respectively. Although Norberg (1976) found that the flight of P. auritus flying at 2·35 ms^-1 (J = 0·5) was explicable in steady-state terms, advance ratios below 0·5 are normally taken to indicate that unsteady effects are important. My interpretation of the directions of the forces generated by the wings during flight in R. ferrumequinum are likely only to be accurate at high speeds, when it is probably safe to assume that the wings are more or less functioning in accordance with the assumptions of steady-state aerodynamic theory, even though it has recently been shown that unsteady effects are probably also very important in fast forward flight (Clopeau, Devilliers & Devezeaux, 1979). At low speeds, when the advance ratio is large, the assumptions of the steady-state aerodynamic theory do not hold and therefore any estimates of incidence angles assuming steady-state conditions are likely to be very inaccurate because of unsteady airflow. For this reason it is not possible to say with confidence that the function of the upstroke in slow forward flight is the generation of thrust, because small inaccuracies in the estimates of the direction of the relative airflow during the upstroke might have implied negative angles of incidence, when in reality the angles are either near zero or even positive. Without some form of flow visualization or measurement it will not be possible to resolve this problem.

In conclusion, the results presented here indicate a gradual change in kinematics as the bat’s flight speed increases up to V_mp. The changes observed are consistent with the hypothesis that upstroke function changes as drag decreases. Above V_mp drag increases again and further experiments are required to see whether, as predicted, the upstroke becomes inactive again.

I am grateful to the following people who assisted me during the course of this work; Dr J. M. V. Rayner, Dr K. D. Scholey and Dr R. E. Stebbings. I am particularly grateful to Professors C. J. Pennycuick and M. B. Fenton and two anonymous referees, who read and commented on an earlier draft of this paper. Finally I am grateful to the Natural Environment Research Council for financial assistance.