Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Behavioral Ecology Vol. 12 No. 2: 150–156 Predator versus prey: on aerial hunting and escape strategies in birds Anders Hedenström and Mikael Rosén Department of Animal Ecology, Lund University, Ecology Building, SE-223 62 Lund, Sweden Predator and prey attack-escape performance is likely to be the outcome of an evolutionary arms race. Predatory birds are typically larger than their prey, suggesting different flight performances. We analyze three idealized attack-escape situations between predatory and prey birds: climbing flight escape, horizontal speeding, and turning and escape by diving. Generally a smaller bird will outclimb a larger predator and hence outclimbing should be a common escape strategy. However, some predators such as the Eleonora’s falcon (Falco eleonorae) has a very high rate of climb for its size. Prey species with an equal or higher capacity to climb fast, such as the swift Apus apus, usually adopt climbing escape when attacked by Eleonora’s falcons. To analyze the outcome of the turning gambit between predator and prey we use a Howland diagram, where the relative linear top speeds and minimum turning radii of prey and predator define the escape and danger zones. Applied to the Eleonora’s falcon and some potential prey species, this analysis indicates that the falcon usually wins against the example prey species; that is, the prey will be captured. Level maneuvering hunting is the most common strategy seen in Eleonora’s falcons. To avoid capture via use of this strategy by a predator, the prey should be able to initiate tight turns at high linear speed, which is facilitated by a low wing loading (weight per unit of wing area). High diving speed is favored by large size. If close enough to safe cover, a prey might still opt for a vertical dive to escape in spite of lower terminal diving speed than that of the predator. On the basis of aerodynamic considerations we discuss escape flight strategies in birds in relation to morphological adaptations. Key words: climbing flight, diving, Falco eleonorae, flight performance, Howland diagram, predation, turning gambit. [Behav Ecol 12:150–156 (2001)] D epending on size and morphology predatory birds use one or other preferred hunting technique, such as surprise attacks by the sparrowhawk (Accipiter nisus) or the legendary stoop by the perergine (Falco peregrinus) (Cresswell, 1996; Rudebeck, 1950–1951). Much recent work has focused on fat load management under perceived or experienced predation risk (Gosler et al., 1995; Lilliendahl, 1997; van der Veen, 1999). Another line of research has studied escape responses in relation to body mass of caged birds when attacked by a simulated predator (e. g., Kullberg et al., 1996; Lee et al., 1996; Lind et al., 1999; Veasey et al., 1998; Witter et al., 1994). However, it remains unclear if the responses observed in such experiments, usually a reduced take-off angle and speed with increasing body mass, represent the effect of a chosen strategy or a constraint on flight performance. This article is not, however, concerned with such take-off escape responses when a bird is attacked by a predator, but rather with the relative flight performance between a predator and its prey when both are already airborne. Lima (1993) reviewed the literature on escape flight strategies in North American birds when attacked by a predator. Size, morphology, and hunting strategy have probably coevolved among predatory species to maximize success in hunting their most common prey. Prey species, on the other hand, should evolve adaptations that maximize the chances of escaping a predator attack, leading to a co-evolutionary arms race between predator and prey (Dawkins, 1982). In birds, the predator is typically larger than the prey, and because size has profound effects on aerodynamic performance, we could expect that the size difference is exploited by the prey when selecting the best escape strategy. In this article we will con- Address correspondence to A. Hedenström. E-mail: anders. [email protected]. Received 15 March 2000; revised 14 June 2000; accepted 16 June 2000. 2001 International Society for Behavioral Ecology sider the interaction between avian predator and prey involved in attack-escape interactions in open air space. When attacked by a predator the prey bird only has one goal—to survive by reaching a safe site before being seized by the predator. In aerial combat the available strategy set for the prey consists mainly of three alternatives: (1) escape by outclimbing the predator (Cade, 1960), (2) escape by outmaneuvering the predator in a turning gambit (Howland, 1974), and (3) escape by diving away from the predator. From the prey’s point of view the choice of the optimal escape behavior will be context dependent. Factors likely to influence the choice of escape strategy are predator (species, sex, size), the relative position between prey and predator, speed vectors of prey and predator when the prey discovers the predator and the state of the prey (e.g., fuel load, muscle size, condition, etc.). A general model combining all these factors is unlikely to generate any clear insights. We will focus on the aerodynamic properties of birds during aerial attacks by considering the three main escape options by simple scaling analysis of flight performance. To improve realism and illustrate the general principles we will calculate some relevant performance measures based on measured flight performance in the Eleonora’s falcon (Falco eleonorae) (Hedenström et al., 1999), which preys on migratory birds in the Mediterranean region during their autumn migration (Rosén et al., 1999; Walter, 1979). For comparison we selected some candidate prey species for which the relevant measures of flight performance were available (Hedenström and Alerstam, 1992, 1994). Even though our discussion will focus on the Eleonora’s falcon and its prey, our results are generally applicable to any aerial predator-prey interaction system where the relative locomotory performances of the predator and prey can be assessed using either biomechanical principles or experiments. METHODS Information on flight performance in the Eleonora’s falcon and seven potential prey species refers to published infor- Hedenström and Rosén • Predator versus prey 151 Table 1 Morphological data of Eleonora’s falcon and seven potential prey species used to derive aerodynamic properties Species m (kg) b (m) S (m2) Aspect ratio (b2/S) Wing loading (N/m2) Eleonora’s falcon Falco eleonorae Knot Calidris canutus Arctis tern Sterna paradiasaea Song thrush Turdus philomelos Dunlin Calidris alpina Swift Apus apus Chaffinch Fringilla coelebs Siskin Carduelis spinus 0.35 0.205 0.11 0.066 0.050 0.043 0.022 0.011 0.95 0.51 0.80 0.34 0.40 0.40 0.26 0.21 0.104 0.0286 0.0571 0.0193 0.0146 0.0171 0.0128 0.0075 8.7 9.1 11.2 6.0 11.0 9.4 5.3 5.9 33.0 70.3 18.9 33.5 33.6 24.7 16.9 14.4 Sources: Hedenström and Alerstam (1992, 1994), Hedenström et al. (1999); m, body mass; b, wing span; S, wing area. mation of sustained climbing flight (Hedenström and Alerstam, 1992, 1994; Hedenström et al., 1999). Climbing flight performance was measured by using tracking radar or an optical range finder. Track data were reduced in relation to wind measurements obtained by tracking ascending helium filled weather balloons. Sustained climbing flight was taken during at least 240 s to represent the climb rate that a bird can maintain by aerobic muscle work, which in principle is for as long as the fuel supply lasts. However, during very long climbs to high altitudes the climb rate will decline because the air density declines with increasing altitude (Pennycuick, 1978). We will assume that the birds exhibit their maximum power during these climbs or a constant proportion thereof (cf. Hedenström and Alerstam, 1992). Then we can estimate the maximum power (Pmax) as the sum of the aerodynamic power and the rate of work required to raise the body against gravity at speed Vz as: Pmax ⫽ P(V) ⫹ mgVz, (1) where m is body mass, g is acceleration of a body in free fall and P(V) is the mechanical power required to fly horizontally at airspeed V. Mechanical power was calculated according to the theory of Pennycuick (1989) with air density 1.23 kg/m3 representing sea level and standard atmospheric conditions, but differing by using a value of 0.1 for the body drag coefficient (Pennycuick et al., 1996). The estimated Pmax from the measured climbing flights was taken as the maximum available power from the flight muscles. Maximum horizontal airspeed (Vmax) was calculated by finding the speed where the power required for horizontal flight equals Pmax. This estimate of Vmax relies on current flight mechanical theory (Pennycuick, 1989), and possible future revision of this theory or parameters used may change the estimates. However, changes to the theory will affect the estimates approximately equally and presumably the relative difference between the species will remain largely unaffected. Hence, the general conclusions will not be subject to features of the aerodynamic theory used. In a steady horizontal turn a component of the lift force (L) has to be directed upwards to balance the weight, which determines the maximum bank angle by cos ⫽ mg/L. The remaining lift can be directed towards the center of rotation giving a minimum turning radius at a given angle of bank as: rmin ⫽ mV 2 , Lsin ⌽ sible bank angle and a maximum lift coefficient CL ⫽ 0.5 at the maximum speed. At low speeds birds can usually develop a maximum lift coefficient around 1.6, but at high speeds lift coefficients are in the range 0.3–0.5 (cf. Pennycuick, 1968). Morphological data (body mass, wing span and wing area) were taken from the original sources of flight performance (Table 1). Measurements refer to the standard methods used for aerodynamic calculations according to Pennycuick (1989), hence the wing area includes the area of the body between the wings. The prey species used as examples are not necessarily typical prey of the Eleonora’s falcon (cf. Walter, 1979), but they were chosen because measures of flight performance were available for these species. However, they do represent similar sizes of typical prey that should have similar performance measures, and they or close relatives have all been found among prey in breeding cliffs of Eleonora’s falcon (Spina et al., 1987; Walter, 1979). (2) where V is speed, L is lift and ⌽ is the bank angle (e.g., Thomas, 1996). For calculations we will assume the maximum pos- Escape by climbing flight If two birds, predator and prey, are on the same level when the prey discovers the predator, then a good strategy should be vertical escape upwards, that is, by climbing flight, if the prey possesses a capacity to climb at a faster rate than the predator. Pennycuick (1978) derived a formula for the climbing capacity in birds as: Vz ⫽ 2.16mm f 1.92m 2/3 ⫺ 1/2 3/2 , m b (3) where mm is the mass of the flight muscles, m is body mass, f is wingbeat frequency, is air density and b is wing span. We can use this formula to calculate how climb rate scales with body mass by substituting the body mass dependent parameters with their respective scaling relationship. Greenewalt (1962) found that flight muscle mass is relatively independent of body size at 17%, that is, mm ⫽ 0.17m. In isometrically scaled birds maximum wingbeat frequency scales as f⬁m–1/3 (Pennycuick, 1975) and wing span scales as b⬁m1/3. By substituting these relations into Equation 2 we get: Vz ⫽ a0m⫺1/3 ⫺ a1m1/6 (4) where a0 and a1 include physical constants independent of body mass. From this equation it is clear that the rate of climb should decline with increasing body mass in a series of isometrically scaled birds. If climbing flight is the selected escape strategy, then we might expect selection for smaller size in the prey species. From Equation 3 it is also clear that birds climbing performance is enhanced by increased flight muscle frac- Behavioral Ecology Vol. 12 No. 2 152 Table 2 Observed airspeed (Vh) and rate of climb (Vz) in the Eleonora’s falcon and seven potential prey species, and derived aerodynamic properties used for comparing relative flight performance between predator and prey Species N Vh (m/s ⫾ SD) Eleonora’s falcon Knot Arctic tern Song thrush Dunlin Swift Chaffinch Siskin 13 33 15 10 10 7 24 7 13.1 13.8 9.9 12.4 13.9 10.0 11.2 13.4 ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ 2.35 1.9 1.4 1.5 0.41 0.8 1.2 1.2 Vz (m/s ⫾ SD) 1.4 1.10 1.24 1.00 1.63 1.34 1.02 0.84 ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ 0.31 0.34 0.22 0.21 0.41 0.30 0.33 0.23 P(Vh) (W) mgVz (W) Ptot (W) Vmax (m/s) ⌽max (⬚) rmin (m) 2.45 2.48 0.46 0.70 0.43 0.27 0.18 0.11 4.81 2.21 1.34 0.65 0.80 0.57 0.22 0.091 7.26 4.69 1.80 1.35 1.23 0.84 0.40 0.20 28.4 25.8 23.6 22.1 24.6 22.1 19.4 18.3 82 70 84 77 80 81 82 82 11.0 24.8 6.3 11.4 11.3 8.3 5.6 4.8 Sources: Hedenström and Alerstam (1992, 1994), Hedenström et al. (1999). N, number of observations; P(Vh), mechanical power required for level flight; mgVz, the power expended to climb at speed Vz; Ptot, the total mechanical power (P(Vh)⫹mgVz), Vmax is the estimated maximum horizontal airspeed; ⌽max, the maximum angle of bank; and rmin, the calculated horizontal minimum turn radius. P(Vh) was calculated according to Pennycuick (1989) using a body drag coefficient CD,par ⫽ 0.1 and air density 1.23 kg/m3, and Vmax was calculated under the same assumptions; rmin was calculated for a lift coefficient CL ⫽ 0.5. tion, maximum wingbeat frequency and wing span. An increase in wingspan inevitably leads to a reduced maximum wingbeat frequency (Pennycuick, 1975, 1996), and so there is a trade-off between the two traits. However, wing span (b) is raised to 3/2 and wingbeat frequency is raised to one in Equation 3, and therefore we might still expect rather long and high aspect ratio wings in birds adapted for high climbing performance. Selection pressure on size and morphology should be in the same direction for a predator where climbing flight pursuit is an important hunting strategy. Real birds are not isometrically scaled, but wing span tends to increase faster with increasing body mass (Rayner, 1988), which partly compensates the adverse effects of size on climb rate. This is particularly obvious in the Eleonora’s falcon, which has an extreme climb rate for its size (Table 2; Hedenström et al., 1999). Comparing the prey species with Eleonora’s falcon, it is only the dunlin (Calidris alpina) that surpasses the predator, and allowing for the variation around the means, possibly also the arctic tern (Sterna paradisaea) and the swift (Apus apus) would escape successfully from the fal- con by climbing (Table 2). If the prey species has a lower climb rate than its predator, then escape climbing is a bad option, even if starting with an altitude advantage, because the predator will eventually close the gap in a sustained climb gambit. When attacked by Eleonora’s falcons small passerines such as chaffinch (Fringilla coelebs) and siskin (Carduelis spinus) should not try climbing flight escapes (Table 2). If a prey is attacked by a predator coming from above, that is, the prey has a lower initial position than the predator (Figure 1), it may still be advantageous to escape by climbing. If we assume that power available from the flight muscles is independent of forward speed (Pennycuick, 1968), then the maximum rate of climb will be associated with the minimum power speed (Vmp). If power available is speed dependent then maximum rate of climb will be at a forward speed greater than Vmp (Thomas and Hedenström, 1998). The escape should be directed away from the attacking predator to maximize the flight distance for the predator (Figure 1). If the prey reaches the same level as the predator before the predator has closed the distance, then climbing flight escape is a viable strategy when the horizontal distance to the predator (D) satisfies the inequality: D⬎ z Vz,prey (Vpred ⫺ Vmp,prey ), (5) where the variables are defined in Figure 1. This rather simplistic model assumes that the predator is already flying at its maximum speed when discovered and that the prey is flying at Vmp and can accelerate to its maximum rate of climb instantly. This is not true in the real world, but the model illustrates the point that climbing escape might be an option even with an initial height disadvantage to the prey and that the decision to climb or not should be distance dependent. Escape by horizontal turning gambit Figure 1 A simple model for a criterion of climbing flight escape with an initial height disadvantage. The horizontal distance to the predator at start of the climb is D. The predator has to cover the distance D⫹z/Vz,prey·Vmp, prey at horizontal speed Vpred, where z is the initial height difference between the predator and prey, Vz,prey is maximum climb rate of prey, and Vmp,prey its minimum power speed. A useful escape tactic to a prey is to initiate a turn before predator closure and rely on a tight turn radius for escape. The classical example is the cheetah and gazelle, where the cheetah has the highest top speed but the gazelle can execute tight turns that are often life-saving maneuvers. Howland (1974) analyzed the condition of a two dimensional turning gambit between predator and prey with respect to maximum linear speed and turn radii. The turning gambit starts with a linear escape away from the predator, but because the predator has the highest maximum speed it will close in on the Hedenström and Rosén • Predator versus prey 153 Figure 2 are calculated on the mean performances of the falcon and prey, but there is individual variation in these and so the border between escape and no escape should be considered as a band around the curve. Species or individuals falling close to the curve of Figure 2 can escape with some probability, depending on the relative individual flight performances of predator and prey. Three of the species (knot, dunlin and song thrush) even have greater turning radii than the predator (Table 2), and they should avoid getting involved in a turning gambit with Eleonora’s falcons altogether. Escape by diving Diving or stooping is a typical attack strategy by large falcons, famous by the performance of the peregrine (Falco peregrinus) (Alerstam, 1987; Peter and Kestenholz, 1998; Tucker, 1998). In a gliding dive inclined at an angle ␣ to the horizontal the bird must keep the wings partly open to provide the lift needed to maintain a constant glide angle. Using simple fixed wing theory of gliding flight the vertical speed (Vz) is: Vz ⫽ Figure 2 A Howland diagram showing the escape and no escape zones of a turning gambit between relative maximum linear speed (v) of prey and predator and minimum relative turning radius (r). The function describing the borderline between escape and no escape is v ⫽ 兹r (Howland, 1974). Also shown are the relative v and r for the species listed in Tables 1 and 2, using estimated maximum speeds and turning radii. The only species falling outside the no escape zone is the arctic tern. The three data points plotted on the right ordinate represent points of r ⬎ 1 and are actually further away from the escape demarcation than indicated in the diagram. The species are knot (Kn), arctic tern (At), song thrush (St), dunlin (Du), swift (Sw), chaffinch (Ch) and siskin (Si). prey. There is, however, a chance to escape if the prey has a smaller turning radius than the predator and there is an optimal moment (or distance) to execute the turn (Howland, 1974). The border between the danger and safe zones in a Howland diagram is defined by: v ⫽ 兹r , (6) where v is the ratio of maximum speeds of prey and predator (Vmax,prey/Vmax,pred) and r is the ratio between the minimum turning radii (rmin,prey/rmin,pred). Hence, knowing maximum speeds and turning radii it is possible to predict the outcome of a turning gambit. We used available data on flight performance in the Eleonora’s falcon and seven potential prey species representing different size and morphology to estimate maximum sustained horizontal flight speed and the minimum horizontal turning radius when initiated at the maximum speed (Table 2). During circular horizontal turns at the calculated minimum radii these birds would experience forces between 2.7g (knot) and 9g (arctic tern), which is of the same magnitude as the vertical force measured in quail (Coturnix coturnix) during take-off (Earls, 2000). The falcon has the highest maximum speed and hence would overtake all the prey species in a straight level escape flight. We have plotted relative speeds and turning radii for the species in Table 2 in a Howland diagram (Figure 2), which indicates that only one species, the arctic tern, would escape from a turning gambit with Eleonora’s falcon. The other species would be taken by the predator. However, the three smallest species (swift, chaffinch, and siskin; Table 2) fall near the border between the escape and no escape zones (Figure 2). The points in the (v,r)-plane of k 1 mg k SV 3 ⫹ 2 , 2 b V mg (7) where V is forward speed along the glide path, b is wing span, S is wing area, and k1 and k2 include physical constants. At the terminal speed the vertical speed is Vsin␣ and Equation 7 becomes: Vmg sin␣ ⫽ k 1 (mg)2 ⫹ k 1 SV 3 , b 2V (8) which after substitution of b and S and rearrangement gives the condition for terminal gliding speed as: V 4 ⫺ c1m1/3V ⫹ c2m2/3 ⫽ 0, (9) where c1 and c2 are new constants independent of body mass. Solving Equation 9 for terminal speed in a gliding dive yields: V ⬀ m1/6, (10) which shows that speed increases with increasing body size (cf. Andersson and Norberg, 1981). A similar analysis for terminal speed in a vertical dive with completely folded wings gives the same scaling relationship as Equation 10. Hence, this explains why large falcons may adopt this attack strategy since they have a speed advantage with respect to a usually smaller prey bird. However, depending on the starting positions a prey may escape, and reach a safe site before being seized, by diving even with a lower terminal speed than the attacking predator. In a very simple setting, the prey dives to cover with a vertical dive and the predator attack by an inclined gliding dive from a horizontal distance D to the prey. Starting at the same level, the prey escapes if the vertical distance to the ground (h) is: h⬍ DVz,prey 2 兹V 2 ⫺ V z,pred , (11) where V is the terminal speed of the predator and Vz is the birds’ respective vertical speeds. Again, this criterion assumes that both birds can accelerate instantaneously to their maximum speeds. Equation 11 shows that the prospects of escaping an attack by diving away from the predator increases with horizontal distance to prey and the terminal speed of the prey. This is perhaps not a very startling result, but the model could easily be modified to account for a more realistic situation, including time of accelerating to maximum speed and different relative altitudes when the hunt starts. Small passerines are often seen trying to escape by vertical dives at high altitudes over sea, even when Eleonora’s falcons are very close to them (Walter, 1979; Hedenström and Rosén, Behavioral Ecology Vol. 12 No. 2 154 unpublished observations). But often such escape dives are combined with a sharp pull-up, i.e. initiating of a loop. This is again an application of the turning gambit. Assuming that the prey bird reaches its terminal speed velocity in a dive with completely folded wings, then V⫽ 冢 冣 2mg Sb CD,par 1/2 , (12) where is air density, Sb is body frontal area and CD,par is the body drag coefficient. If the bird opens its wings they will create a lift force that can be used to create a centripetal acceleration in a turn away from the dive path. Initially all that lift is used to turn, but later along the loop path the same amount of lift also has to support the weight. The initial turning radius is: r⫽ 2m , SCL (13) where m is body mass, S is wing area and CL is the lift coefficient. The morphology and aerodynamic properties of the bird determine the relative speed and the turn radius to be plotted in a Howland diagram. For the species in Table 1 we need information about body mass, frontal area, wing area, body drag and lift coefficients to estimate the outcome of a vertical dive turning gambit. Such data are lacking for these species, but by assuming CD,par ⫽ 0.1, CL ⫽ 0.5 and otherwise the default values in Pennycuick (1989), we get a similar outcome as that of the horizontal flight turning gambit (cf. Figure 2). DISCUSSION Predator and prey interactions often have a great influence on the life of organisms, such as habitat selection, selection of feeding sites, sociality and group living and vigilance (Lima and Dill, 1990). They can also generate morphological adaptations and counter adaptations in the predator and prey species to enhance performance of capturing and escaping, respectively. We have focused our analysis on the flight performance in birds escaping by one of three main strategies: climbing away from the predator, outmaneuvering the predator in a horizontal turning gambit, or diving to safety. The general principles should however be applicable also to other situations and animals. In climbing flight escape, performance is enhanced by small size, long wings, large flight muscles and high wingbeat frequency. When encountering attacks from Eleonora’s falcons, a predator with an exceptional capacity to climb fast for its relatively large size (Hedenström et al., 1999), only the dunlin and possibly also the swift should escape by climbing (Table 2). In fact, this is the main escape tactic observed in swifts (Hedenström et al., 1999; Hedenström A, and Rosén M, unpublished observations). The dunlin is usually not a prey of Eleonora’s falcons, but when attacked by other predators climbing escapes have been observed (Lima, 1993). The wing morphology of both the swift and the dunlin is characterized by rather high aspect ratio wings, in line with the prediction for birds adopting climbing escapes (cf. Table 1). The skylark (Alauda arvensis) is another species escaping by climbing (song) flight on attacks by the merlin (Falco columbarius) (Cresswell, 1994). Skylarks are known to have capacity for fast climbs (Hedenström, 1995b). Morphologically the skylark has shorter wings than the swift and dunlin, but it has larger flight muscles than birds in general (about 25% of total body mass; Rayner, 1988). Lima (1993) also reports escape by climbing flight in the short-billed dowitcher (Limnodromus griseus), American crow (Corvus brachyrhynchos) and water pipit (An- thus rubescens), and Hedenström (1995a) reports an observation of a flock of white-winged black terns (Chlidonias leucopterus) escaping by climbing away from a pursuing peregrine. Some large species, such as the sage grouse (Centrocercus urophasianus), which is a typical quarry to large falcons, rely on acceleration and high maximum horizontal speed achieved by large flight muscles and small wings (Pennycuick et al., 1994). The maximum speed of the grouse is higher than that of the falcon, but the grouse would not be able to fly at this speed with aerobic muscle work, and it is questionable if it can fly at any speed without incurring an oxygen debt (Pennycuick et al., 1994). Hence, it needs to find cover soon after having out-speeded a pursuing falcon. In the turning gambit a prey will outmaneuver a predator by a combination of high relative linear top speed and a small turning radius (Figure 2; Equation 6). Of our example species only one, the arctic tern, would outperform the Eleonora’s falcon, while two (or possibly three) species were borderline cases (Figure 2). A small turning radius is achieved by a low wing loading (large relative wing area) and fast flight is facilitated by a streamlined body shape, low wing drag (i.e., small wings), and high power available from the flight muscles (large muscle fraction). Generally, wing loading increases with increasing body size and so a small turning radius is obtained by small birds with relatively large, but short, wings. These features represent a typical passerine, but apparently also terns are adapted for this sort of gambit by low wing loading and relatively high top speed (Tables 1 and 2). The Eleonora’s falcon mainly hunts by chasing prey by active maneuvering flight (Rosén et al., 1999), perhaps because their main prey (small passerines) most often try to escape by maneuvering flight. The Howland diagram (Figure 2) indicates that this predator should do well by this strategy when compared with some potential prey species. Field observations show however that success rate is quite low per attack (11%; Walter, 1979), but once a bird is attacked when passing a colony of Eleonora’s falcons, many falcons will attack in rapid succession resulting in rather low survival chances for the prey (Rosén et al., 1999; Walter, 1979). The survey by Lima (1993) shows that escape by a last moment dodge is a very common strategy. Hence, the Howland diagram is a useful tool for assessing the likely outcome of an attack-escape gambit between predator and prey in birds. The only condition is that the prey has the opportunity to first move away from the predator and the ability to execute turns with minimum radii (Howland, 1974). This generally requires open spaces, such as the aerial hunting of Eleonora’s falcons, which usually takes place at great altitudes where migrants cruise when passing the Mediterranean sea on autumn migration (Rosén et al., 1999). Howland diagrams are also applicable to other situations, such as birds hawking insects in the air (cf. Warrick, 1998), bats hunting moths (Roeder and Treat, 1961), fish hunting in open water (Arnott et al., 1999) and cheetah and other terrestrial carnivores hunting on open plains. Extra lift and still more reduced turning radius may be achieved by spreading the tail and hence augmenting the lifting surfaces when turning (Thomas, 1996). Also the prey can do this, thereby reducing its turning radius by the same amount as the falcon, provided the relative tail surfaces are the same, in which case the relative radii will remain unchanged with respect to those calculated in Table 2. Some species, however, such as dunlin and knot (Calidris canutus), have relatively small tails and would not be able to increase lift to the same degree as the falcon, which has a normal tail size, and consequently do even worse in a turning gambit than indicated in Figure 2. In escape by diving the predator has an advantage from its size and will reach higher maximum terminal speeds than the Hedenström and Rosén • Predator versus prey smaller prey. Birds will maximize the diving speed by having streamlined bodies and small wings. Many water birds escape by plunge diving (Lima, 1993), and especially ducks seem to fulfil the requirements for this. Alerstam (1987), using tracking radar, reports that his highest measured speed of birds refers to red-breasted mergansers (Mergus serrator) reaching 43 m/s in a shallow gliding dive. Passerines also dive when attacked by Eleonora’s falcons, in combination with pull-ups, hence executing a vertical turning gambit. Small birds should be able to execute relatively tighter turns when initiated at a very high speed, as during a vertical dive with completely folded wings, where the larger predator might not be able to achieve its theoretical minimum turning radius for structural safety reasons (cf. Howland, 1974). Swifts were never observed diving when escaping from Eleonora’s falcons, perhaps because their long wings generate too much drag even when folded. In this article we have focused on how aerodynamic theory may be used to analyze attack-escape performance in birds by indicating the direction of selection on morphology for improved flight performance. Our analyses show that depending on escape strategy, there may be partly diverging selection pressures on morphology. This, in turn, suggests that a species should be generally best adapted for escaping by a certain method, that is, ‘‘escape specialists’’ rather than ‘‘escape generalists.’’ The swift is an example of a species that invariably escape by climbing flight when attacked by Eleonora’s falcons. Lima’s (1993) survey indicated that many species seem to prefer one main strategy, but several species may select one of the alternative strategies. However, our analyses also show that the relative performances of the prey and predator should influence the escape strategy selected. Also, the relative positions between the predator and prey birds, such as vertical and horizontal distances, decide the escape strategy selected. By always maximizing the distance and trajectory to the predator, a prey will inflict the maximum energy cost to the predator (Weihs and Webb, 1984), which eventually can force the predator to abort the attack. In line with the life-dinner principle (Dawkins, 1982) the predator should consider energy costs associated with prey capture, while the prey should pay little attention to energy costs per se (Hedenström and Alerstam, 1995). Studying sparrowhawk attacks on birds, Cresswell (1995) found that the hawks preferred prey in the size range 101–150 g. This could be due to the relative maneuverability of this size class in relation to small prey (ⱕ50 g) that may escape more easily. It may also be that large prey was preferred because they are more profitable than small prey. In another study, Cresswell (1993) found that redshanks (Tringa totanus) responded differently on attacks by three different bird predators, supporting our flight mechanics results. Hence, when possible, the prey should obtain information such as species, distance, speed, and flight direction about the attacking predator for appropriate choice of escape response that maximizes the survival chances. When facing surprise attacks this might not be possible, and the prey bird might rather chose a standard escape response. This situation could be what cage escape flight experiments represent (e.g., Lind et al., 1999; Veasey et al., 1998; Witter et al., 1994). However, Kullberg et al. (1998) found that the escape trajectory of great tits (Parus major) depended on the attack angle of a model predator. It is of course naı̈ve to believe that selection for efficient escape performance is the only factor that matters to shape a bird’s morphology. Selection for efficient foraging, migration and display should also be important selective agents. Predation may, however, be such an important selective force that it results in significant features in the prey species’ size and morphology. Some typical prey species to the Eleonora’s fal- 155 con, for example swifts, swallows and flycatchers, are themselves aerial predators, which could involve selection on size and morphology in the same direction as selection for efficient escape from an aerial predator. But depending on the escape response of invertebrate prey, small birds may face conflicting selective demands for their own foraging efficiency and escape flight performance. Finally, even if the aerodynamic theory used is revised in the future, we think that the general conclusions will remain valid. Measuring the maximum flight performance of birds is a challenge (cf. Chai and Dudley, 1999; Marden, 1987), but such data can be used to assess the relative flight performance of predator and prey by using Howland diagrams. They can also help us understand why birds are designed as they are. It is extremely difficult to measure the flight trajectories of two birds during attack-escape maneuvers in the wild. However, such measurements and high-speed films of these interactions would be most welcome. We thank Fernanda Diana for inviting us to work with the Lega Italiana Protezione Uccelli (LIPU) project as a base and for her support during our stays. We are also grateful to Alberto Badami, Nicola Fara, Franco Fadda, Maurizio Medda, Fernando Spina, and Susanne Åkesson for assistance during fieldwork that inspired our thinking on aerial predation. Erik Svensson and S. Åkesson gave much appreciated comments on the manuscript. This research was supported by the Swedish Natural Science Research Council (given to A.H.). REFERENCES Alerstam T, 1987. Radar observations of the stoop of the peregrine falcon Falco peregrinus and the goshawk Accipiter gentilis. Ibis 129: 267–273. Andersson M, Norberg RÅ, 1981. Evolution of reversed sexual size dimorphism and role partitioning among predatory birds, with a size scaling of flight performance. Biol J Linn Soc 15:105–130. Arnott SA, Neil DM, Ansell AD, 1999. Escape trajectories of the brown shrimp Crangon crangon, and a theoretical consideration of initial escape angles from predators. J Exp Biol 202:193–209. Cade TJ, 1960. Ecology of the peregrine and gyrfalcon populations in Alaska. Berkeley, California: University of California Press. Chai P, Dudley R, 1999. Maximum flight performance of hummingbirds: capacities, constraints, and trade-offs. Am Nat 153:398–411. Cresswell W, 1993. Escape response by redshanks, Tringa totanus, on attack by avian predators. Anim Behav 46:609–611. Cresswell W, 1994. Song as a pursuit-deterrent signal, and its occurrence relative to other anti-predation behaviours of skylark (Alauda arvensis) on attack by merlins (Falco columbarius). Behav Ecol Sociobiol 34:217–223. Cresswell W, 1995. Selection of avian prey by wintering sparrowhawks Accipiter nisus in southern Scotland. Ardea 83:281–389. Cresswell W, 1996. Surprise as a winter hunting strategy in sparrowhawks Accipiter nisus, peregrines Falco peregrinus and merlins F. columbarius. Ibis 138:684–692. Dawkins R, 1982. The extended phenotype. Oxford: Oxford University Press. Earls KD, 2000. Kinematics and mechanics of ground take-off in the starling Sturnus vulgaris and the quail Coturnix coturnix. J Exp Biol 203:725–739. Gosler AG, Greenwood JJD, Perrins C, 1995. Predation risk and the cost of being fat. Nature 377:621–623. Greenewalt CH, 1962. Dimensional relationships for flying animals. Smithsonian Misc Coll 144(2):1–46. Hedenström A, 1995a. Ecology of avian flight (PhD dissertation). Lund: Lund University. Hedenström A, 1995b. Song flight performance in the skylark Alauda arvensis. J Avian Biol 26:337–342. Hedenström A, Alerstam T, 1992. Climbing performance of migrating birds as a basis for estimating limits for fuel-carrying capacity and muscle work. J Exp Biol 164:19–38. Hedenström A, Alerstam T, 1994. Optimal climbing flight in migrating birds: predictions and observations of knots and turnstones. Anim Behav 48:47–54. 156 Hedenström A, Alerstam T, 1995. Optimal flight speed of birds. Phil Trans R Soc Lond B 348:471–487. Hedenström A, Rosén M, Åkesson S, Spina F, 1999. Flight performance during hunting excursions in Eleonora’s falcon Falco eleonorae. J Exp Biol 202:2029–2039. Howland HC, 1974. Optimal strategies for predator avoidance: the relative importance of speed and manoeuvrability. J Theor Biol 47: 333–350. Kullberg C, Fransson T, Jakobsson S, 1996. Impaired predator evasion in fat blackcaps. Proc R Soc Lond B 263:1671–1675. Kullberg C, Jakobsson S, Fransson T, 1998. Predator induced take-off strategy in great tits (Parus major). Proc R Soc Lond B 265:1659– 1664. Lee SJ, Witter MS, Cuthill IC, Goldsmith AR, 1996. Reduction in escape performance as a cost of reproduction in gravid starlings Sturnus vulgaris. Proc R Soc Lond B 263:619–624. Lilliendahl K, 1997. The effect of predator presence on body mass in captive greenfinches. Anim Behav 53:75–81. Lima SL, 1993. Ecological and evolutionary perspectives on escape from predatory attacks: a survey of North American birds. Wilson Bull 105:1–47. Lima SL, Dill LM, 1990. Behavioral decisions made under the risk of predation: a review and prospectus. Can J Zool 68:619–640. Lind J, Fransson T, Jakobsson S, Kullberg C, 1999. Reduced take-off ability in robins (Erithacus rubecula) due to migratory fuel load. Behav Ecol Sociobiol 46:65–70. Marden JH, 1987. Maximum lift production during takeoff in flying animals. J Exp Biol 130:235–258. Pennycuick CJ, 1968. Power requirements for horizontal flight in the pigeon Columba livia. J Exp Biol 49:527–555. Pennycuick CJ, 1975. Mechanics of flight. In: Avian biology (Farner DS, King JR, eds). New York: Academic Press; 1–75. Pennycuick CJ, 1978. Fifteen testable predictions about bird flight. Oikos 30:165–176. Pennycuick CJ, 1989. Bird flight performance: a practical calculation manual. Oxford: Oxford University Press. Pennycuick CJ, 1996. Wingbeat frequency of birds in steady cruising flight: new data and improved predictions. J Exp Biol 199:1613– 1618. Pennycuick CJ, Fuller MR, Oar JJ, Kirkpatrick SJ, 1994. Falcon versus prey: flight adaptations of a predator and its prey. J Avian Biol 25: 39–49. Behavioral Ecology Vol. 12 No. 2 Pennycuick CJ, Klaassen M, Kvist A, Lindström Å, 1996. Wingbeat frequency and the body drag anomaly: wind-tunnel observations on a thrush nightingale (Luscinia luscinia) and a teal (Anas crecca). J Exp Biol 199:2757–2765. Peter D, Kestenholz M, 1998. Sturzflüge von Wanderfalke Falco peregrinus und Wüstenfalke F. pelegrinoides. Orn Beob 95:107–112. Rayner JMV, 1988. Form and function in avian flight. Curr Ornithol 5:1–66. Roeder KD, Treat AE, 1961. The detection and evasion of bats by moths. Amer Sci 49:135–148. Rosén M, Hedenström A, Badami A, Spina F, Åkesson S, 1999. Hunting flight behaviour of the Eleonora’s falcon Falco eleonorae. J Avian Biol 30:342–350. Rudebeck G, 1950–1951. The choice of prey and modes of hunting of predatory birds with special reference to their selective effect. Oikos 2:65–88, 3:200–231. Spina F, Scappi A, Berthemy B, Pinna G, 1987. The diet of Eleonora’s falcon Falco eleonorae in a colony of the western coast of Sardinia with some remarks on the migration of small passerines through the Mediterranean. Suppl Ric Biol Selvaggina 12:235–254. Thomas ALR, 1996. The flight of birds that have wings and tail: variable geometry expands the envelope of flight performance. J Theor Biol 183:237–245. Thomas ALR, Hedenström A, 1998. The optimum flight speeds of flying animals. J Avian Biol 29:469–477. Tucker VA, 1998. Gliding flight: speed and acceleration of ideal falcons during diving and pull out. J Exp Biol 201:403–414. van der Veen IT, 1999. Effects of predation risk on diurnal mass dynamics and foraging routines of yellowhammers (Emberiza citrinella). Behav Ecol 10:545–551. Veasey JS, Metcalfe N, Houston DC, 1998. A reassessment of the effect of body mass upon flight speed and predation risk in birds. Anim Behav 56:883–889. Walter H, 1979. Adaptations to prey and habitat in a social raptor. Chicago: University of Chicago Press. Warrick DR, 1998. The turning- and linear-maneuvering performance of birds: the cost of efficiency for coursing insectivores. Can J Zool 76:1063–1079. Weihs D, Webb PW, 1984. Optimal avoidance tactics in predator prey interactions. J Theor Biol 106:189–206. Witter MS, Cuthill IC, Bonser RHC, 1994. Experimental investigations of mass-dependent predation risk in the European starling, Sturnus vulgaris. Anim Behav 48:201–222.