The elaborate plumage in peacocks is not such a drag

Sexual selection may favour the evolution of extravagant secondary sexual characteristics that are important in mate choice and confer reproductive benefits (Darwin, 1871). There are several theories about the origin and maintenance of such sexual traits, but central to ‘handicap’ and ‘good genes’ models of sexual selection is the assumption that such ornaments incur a cost (Zahavi, 1975; Andersson, 1994). It is hypothesised that exaggerated sexually selected traits negatively affect organismal performance (e.g. endurance or escape response) by directly (e.g. through increased predation risk) or indirectly (e.g. through increased metabolic energy expenditure) incurring costs (Kotiaho, 2001). However, while the sexual trait has a negative affect on organismal performance, it is an indicator of overall superior genetic quality and therefore has a net positive effect on fitness [e.g. fecundity or offspring success (Petrie, 1994; Rowe and Houle, 1996; Möller and Alatalo, 1999; Hale et al., 2009; Husak and Swallow, 2011)]. Examples of such traits are found throughout nature, and include the ornate plumage exhibited in the males of some bird species [e.g. long-tailed widowbirds, Euplectes sp. (Andersson, 1982); peafowl, Pavo sp. (Petrie, 1994); and paradise whydas, Vidua sp. (Alatalo et al., 1988)]. It has been suggested that these structures may impair flight performance (Balmford et al., 1993). However, empirical evidence quantifying the costs associated with sexually selected traits is generally lacking (Kotiaho, 2001). Demonstrating whether sexually selected traits do incur a cost is necessary to test current theories in evolutionary biology.

Arguably one of the most iconic examples of sexual selection is the extravagant plumage that forms the train in male peafowl (peacocks; Pavo cristatus Linnaeus 1758), which is thought to have evolved as a result of female preference to mate with males possessing more elaborate trains (Darwin, 1871; Petrie et al., 1991; Petrie and Halliday, 1994; Loyau et al., 2005). Possessing a more elaborate train reflects male health status (Loyau et al., 2005), and may confer reproductive benefits (Petrie, 1994). Take-off flight is an important means of avoiding predators and the ability to take-off at high speeds and at a steep angle of ascent demands a very high power output (Askew and Marsh, 2002) and may have a direct influence on an individual's survival (Cresswell, 1993; Witter et al., 1994). Peafowl spend most of their time on the ground or roosting in trees; in adult males, flight only accounts for 0.12% of their daily activity (Harikrishnan et al., 2010). While being capable of sustained flight (Hoyo et al., 1994), peafowl primarily use flight to access roost sites and to evade mammalian predators, such as leopards Panthera pardus fusca (Mondal et al., 2011). Possessing elaborate plumage with high aerodynamic drag and/or weight could negatively affect take-off flight performance and consequently could have a direct influence on an individual's survival (Witter et al., 1994).

The aim of this experiment was to determine whether possessing elaborate plumage during the breeding season adversely affects take-off flight performance in peacocks. High-speed videogrammetry was used to capture the wing and body kinematics of flights in birds possessing the train and following removal of the train. Take-off performance was quantified as the sum of the rates of change of the potential and kinetic energies of the body. In addition, wind tunnel tests were carried out to determine the aerodynamic forces on the train in order to estimate the drag it incurs during take-off flights. It was hypothesised that the train would increase the parasite drag of the bird's body during flights, and this, together with the additional weight, would impair take-off performance.

The morphological data for the birds are given in Table 1. Birds with trains performed 4.6±0.7 flights and birds without trains performed 4.8±0.9 flights.

The velocity of the bird at take-off and the flight velocity of the peafowl at the end of the second wingstroke were not significantly different between the two treatments (Table 2). The overall velocity decreased during the flight; the rate of loss of kinetic energy was similar across the two treatments (Table 2). The rate of change of potential energy was not significantly different between the two conditions (Table 2). Note that in all birds the rate of increase in potential energy far exceeded the rate of loss of kinetic energy; hence, net positive mechanical power is required from the flight muscles to move the centre of mass of the body (CoM) of the bird. The total mass-specific power of the centre of mass of the body (P_CoM; calculated relative to pectoralis muscle mass) was not significantly different between birds with and without trains (train P_CoM=222.6±43.3 W kg⁻¹; no train P_CoM=210.0±25.5 W kg⁻¹; t₄=0.41, ns; Fig. 1A,C, Table 2).

The drag on the train increased linearly with increasing air velocity for each angle and was highest at steeper angles (Fig. 2). At the mean flight velocity for birds (4 m s⁻¹) possessing a train, the drag on the train ranged from 0.12 N (level) to 0.15 N (−10 deg): during take-off, the train was held at an angle of between 0 and −10 deg relative to the angle of elevation of the flight trajectory of the bird. A small vertical lift force was produced when the train was held at angles below the horizontal (0.04 N at a train angle of −10 deg at 4 m s⁻¹). This was less than 0.08% of body weight and was ignored in the calculation of induced power.

The total power requirement of take-off (P_aero) was not significantly different between the two conditions (train P_aero=414.1±31.2 W kg⁻¹; no train P_aero=393.3±27.1 W kg⁻¹; t₄=0.55, ns; Fig. 1B, Table 2). Parasite drag was significantly higher in birds with a train compared with birds following the train's removal (Table 2). Profile power and the power required to generate the induced velocity were not significantly different between the two conditions (Table 2).

The train of male peafowl is one of the most elaborate sexually selected structures found in nature. It is hypothesised that such structures incur a cost (Andersson, 1994), e.g. by reducing locomotor performance, and could have functional ecological consequences such as increased predation risk (Witter et al., 1994). However, contrary to this prediction, here it is shown that possessing a train does not significantly reduce flight take-off performance (Fig. 1A,C). P_CoM was not significantly different in birds possessing trains compared with those birds from which the train had been removed. There was no significant difference in P_aero between the two treatments, suggesting that birds with trains were not maintaining take-off performance (P_CoM) simply by increasing the mechanical power output of the flight muscles (P_aero).

The drag on the train depends on flight velocity and the angle at which it is held relative to the flight trajectory of the bird (Fig. 2) (see also Evans and Thomas, 1992). At the flight velocities and relative train angles used during take-off (4.1±0.2 m s⁻¹), the drag on the train was only 0.12–0.15 N. This results in a significant, twofold increase in parasite power, but as parasite power only represents 0.1% of the total aerodynamic power requirements, the effects of the drag of the train on flight performance are trivial. The train represents 6.9% of the bird's body mass. Increasing body mass is expected to increase induced power (through increases in thrust and induced velocity), and therefore to decrease flight performance via a reduction in P_CoM. However, this effect was not detected. Diurnal mass gains of 5.3% in great tits have also been shown to have no significant effect on escape flight performance (Macleod, 2006). Together, these findings suggest that although added mass must affect the induced power requirements, the magnitude of the effect has little meaningful functional relevance (given the high variability in individual take-off performance per se; Table 2, Fig. 1A) (Macleod, 2006). There are several other examples of sexually selected traits having rather subtle effects on bird locomotor performance. In Anna's hummingbirds (Calypte anna) and scarlet-tufted malachite sunbirds (Nectarinia johnstoni), artificial manipulations of the tail by the addition of feathers from a different bird species or the removal of the tail feathers demonstrate that elaborate tail ornaments can have relatively small effects on flight performance by reducing maximum speed (by 3%) and increasing metabolic cost (by 11%, but only at the highest speeds, which represent 1–7% of the birds' flight behaviour) (Clark and Dudley, 2009), and by reducing hawking efficiency (Evans and Thomas, 1992). The vertical display flights performed by collared doves (Streptopelia decaocto) have been estimated to incur a relatively low metabolic cost (~5% of basal metabolic rate) (Usherwood, 2008).

Not all flights performed were maximal, as indicated by the range of values measured for the total aerodynamic power requirements of take-off (P_aero; Fig. 1B). The flights were performed in a flight arena with which the birds were familiar, using captive-bred birds. These are conditions that could potentially compromise performance. If flights were submaximal performance, it would be feasible for a bird encumbered by some handicap to maintain take-off performance (P_CoM) by increasing the mechanical power output of the flight muscles (P_aero) and the conclusion about the absence of a significant effect on locomotor performance could be doubted. The maximal P_aero recorded across individuals was similar and averaged 403.7±19.8 W kg⁻¹. Whether these flights were maximal is not known; however, the total power is close to the maximal power reported in other species [395 W kg⁻¹ in magpie, 350 W kg⁻¹ in grey jay (Jackson and Dial, 2011); 390 W kg⁻¹ in blue breasted quail (Askew et al., 2001)]. The work output from the muscles (66 J kg⁻¹) is also close to the total mechanical work required from the muscles during take-off flights in common raven [48 J kg⁻¹ (Jackson and Dial, 2011)] and Harris hawks [56 J kg⁻¹ (Askew et al., 2001)], and close to the maximum work skeletal muscles are predicted to be capable of generating (Peplowski and Marsh, 1997). Together, the high power output and work generated by the flight muscles suggest that the flights obtained for each individual were likely very close to maximal, and performed with a similar level of effort.

The absence of a detectable effect of the male peafowl train on flight take-off performance does not necessarily mean that there is no cost to possessing a train: there are other potential costs. It could affect the bird's moment of inertia, which could affect flight control and stability. The train could impair running performance and increase the metabolic cost of carrying the additional mass during terrestrial locomotion (Marsh et al., 2006). There is also the cost of feather production during the seasonal growth of the train. The cost of feather production, C_f, is estimated to be 54.9 kJ g⁻¹ dry feathers [following Lindstrom et al. (Lindstrom et al., 1993) and using an estimate of basal metabolic rate from Lasiewski and Dawson (Lasiewski and Dawson, 1967)]. The mass of the train in this study was 320 g, giving an estimated cost of producing the train of 17.5 MJ. The train is produced over ~6 months (M. Caunce, personal communication), such that the train production costs are estimated to be 96.3 kJ day⁻¹, representing ~10% of basal metabolic rate [947 kJ day⁻¹ (Lasiewski and Dawson, 1967)] and 3% of field metabolic rate [2805 kJ day⁻¹, estimated from a scaling equation for Galliformes (Nagy et al., 1999)]. None of these potential costs are mutually exclusive and all could individually be relatively subtle and difficult to detect: the ultimate fitness costs could be the result of the cumulative effect of each of these factors on locomotor performance.

The peacock's train is often cited as a classic example of a ‘costly’ product of sexual selection. However, here it is demonstrated that possessing a train does not detrimentally affect take-off flight performance. This is partly due to the low drag of the train and its consequent trivial effects on take-off power. These results do not necessarily mean there are no costs associated with possessing an ornate train; rather, any such costs are small with limited meaningful functional significance to take-off performance.

Five mature, adult male Indian peafowl (P. cristatus) were obtained from a UK population of captive-bred birds.

A flight arena (1.7×4.3×4.6 m, width×length×height) was constructed from galvanised steel scaffolding covered with galvanised steel chicken wire netting (50 mm hexagonal sections). Two wooden perches were placed within the flight chamber (separated by a total distance of 3.5 m and by a vertical distance of 2.7 m). The peafowl were flown within the flight arena and recorded using two synchronised digital high-speed video cameras (Troubleshooter, Fastec Imaging, San Diego, CA, USA) operating at 125 frames s⁻¹ and shuttered at 0.8 ms and mounted on tripods. The flight arena volume was calibrated using the direct linear transformation method (Hedrick, 2008). Peafowl were stimulated to fly from the lower to the higher perch (usually by clapping or rattling a stick on the perch) in an effort to elicit a maximal response. The flight was recorded using a centre trigger mode that recorded a sequence of images both before and after the trigger, thereby capturing the flight in its entirety. There were two experimental conditions: flights were recorded in birds possessing the train and following removal of the train. The train was removed by cutting through the calamus (shaft) of the feathers close to the body using secateurs, either 2 or 5 days after the recording of the flights in which the birds had trains. The flights in birds from which the train had been removed were recorded on the same day that the train was removed. However, birds were given time to become accustomed to the missing train, prior to recording.

To allow the total effort of take-off to be assessed, it was necessary to estimate the total power requirements of take-off (P_aero). An aerodynamic model was used to calculate the induced (P_ind), profile (P_pro) and parasite power (P_par) components of flight, the sum of which equals P_aero (following Askew et al., 2001; Morris and Askew, 2010; Wakeling and Ellington, 1997).

For each individual, the flight with the highest P_CoM for the two treatments was selected for statistical comparisons, as any potential differences in performance are most likely to be apparent during maximal effort flights.

Following the flights, animals were killed using an overdose of sodium pentobarbital injected into the brachial vein. The outstretched wings were photographed in order to determine wing area and wing length. The pectoralis and supracoracoideus muscles were dissected and their masses determined. The mass of the train was also determined. All power components were expressed relative to the total pectoralis muscle mass, determined post-mortem.

The train was secured inside a hollow cylindrical tube (170 mm long×68 mm outer diameter) with a hemispherical nose profile, which held the train in a manner that reproduced the in-flight train morphology. The cylinder was mounted on a six-component under-floor balance (Aerotech ATE Ltd, UK) via a 20 mm diameter cylindrical shaft that passed through a clearance hole in the floor of the wind tunnel (see Johl et al., 2004). Tests were performed with the train mounted at angles of 0, −10 and −20 deg at free-stream velocities (v) of 3.6–8 m s⁻¹, encompassing the range of speeds and angles (relative to the path of motion of the bird) observed during take-off. Equivalent force measurements were also made on the cylinder and mount, and these values were subtracted from the force measurements made with the train in place. All force measurements were sampled at 300 Hz and averaged over 20 s to obtain a mean.

Reported values are means ± s.e.m. for the five individuals. Flight performance variables and aerodynamic power components were analysed using paired t-tests with significance levels adjusted using the Bonferroni–Holm correction for multiple comparisons.

I thank Chris Wright for assistance in carrying out the experiments and his care of the birds. I am grateful to Dr Martin Passmore, Dr Simon Tuplin and Rob Hunter at Loughborough University for their assistance with the wind tunnel testing and to Russ Jeanes for technical assistance in the video analysis.