posted online on 13 June 2016 as doi: /jeb

First posted online on 13 June 2016 as 10.1242/jeb.139709 J Exp Biol Advance Access the Online most recent Articles. version First at http://jeb.biologists.org/lookup/doi/10.1242/jeb.139709 posted online on 13 June 2016 as doi:10.1242/jeb.139709 Access the most recent version at http://jeb.biologists.org/lookup/doi/10.1242/jeb.139709 Differential sex-specific walking kinematics in leghorn chickens (Gallus gallus domesticus) selectively bred for different body size Kayleigh A. Rose, Jonathan R. Codd and Robert L. Nudds* Faculty of Life Sciences, University of Manchester, Manchester, M13 9PT, UK * Address for reprints and other correspondence: Dr. Robert Nudds Faculty of Life Sciences University of Manchester Manchester M13 9PT, UK E-mail: robert.nudds@manchester.ac.uk Tel: +44(0)161 275 5447 KEY WORDS: Froude number; locomotion; posture; sexual dimorphism; walking Summary statement: In two size classes of layer chicken, sexually dimorphic walking kinematics is linked to the differential muscle force, work and power demands of varied visceral and muscle proportions 2016. Published by The Company of Biologists Ltd.

ABSTRACT The differing limb dynamics and postures of small and large terrestrial animals may be mechanisms for minimising metabolic costs under scale-dependent muscle force, work and power demands; however, empirical evidence for this is lacking. Leghorn chickens (Gallus gallus domesticus) are highly dimorphic: males have greater body mass and relative muscle mass than females, which are permanently gravid and have greater relative intestinal mass. Furthermore, leghorns are selected for standard (large) and bantam (small) varieties and the former are sexually dimorphic in posture, with females having a more upright limb. Here, high-speed videography and morphological measurements were used to examine the walking gaits of leghorn chickens of the two varieties and sexes. Hind limb skeletal elements were geometrically similar among the bird groups, yet the bird groups did not move with dynamic similarity. In agreement with the interspecific scaling of relative duty factor (DF, proportion of a stride period that a foot has ground contact) with body mass, bantams walked with greater DF than standards and females with greater DF than males. Greater DF in females than in males was achieved via variety-specific kinematic mechanisms, associated with the presence/absence of postural dimorphism. Females may require greater DF in order to reduce peak muscle forces and minimize power demands associated with lower muscle to reproductive tissue mass ratios and smaller body size. Furthermore, a more upright posture observed in the standard, but not bantam, females, may relate to minimizing the work demands of being larger and having proportionally larger reproductive volume. Lower DF in males relative to females may also be a work-minimizing strategy and/or due to greater limb inertia (due to greater pelvic limb muscle mass) prolonging the swing phase.

INTRODUCTION The size of an animal influences its walking kinematics. When moving at the same speed (U, m s -1 ) larger animals generally take longer and fewer strides per unit time than smaller animals. Comparison of the walking kinematics of different sized animals can be conducted at speeds at which the ratios of inertial to gravitational forces acting upon the body centre of mass (CoM) are equal, using either the Froude number (Fr = U 2 /ghhip) or its square root, often termed relative speed: û =U / ghhip (1), where hhip is hip height (m) and g is gravitational acceleration (9.81 m s -2 ) (Alexander, 1976; Alexander and Jayes, 1983). Dynamic similarity of motion between different sized animals requires geometric similarity in body plan and equal values of dimensionless kinematic parameters (scaled appropriately to negate the effects of size) for a given relative speed (Alexander, 1976; Alexander and Jayes, 1983; Hof, 1996). Animals may move in such a way as to minimize metabolic cost. The metabolic cost of transport is the energy required to move a unit body weight over a unit distance ([power]/body weight x speed). Geometrically similar animals of different size moving in a dynamically similar fashion are expected to have equal metabolic costs of transport (Alexander and Jayes, 1983). The dynamic similarity hypothesis of Alexander and Jayes (1983) postulated that different quadrupedal mammals would locomote with dynamic similarity at equivalent relative speeds. Within non-cursorial (<1 kg) and cursorial (>10 kg) mammalian groups (Jenkins, 1971) the hypothesis was supported. Observed kinematic differences between the two groups, however, were not accounted for (Alexander and Jayes, 1983). Furthermore, between avian species of small and large body size, there is considerable

deviation from dynamic similarity of locomotion (Gatesy and Biewener, 1991; Abourachid and Renous, 2000; Abourachid, 2001). A general pattern, however, exists across these vertebrates, whereby smaller species move with greater relative duty factors (DF, proportion of a stride with ground contact for any given foot) and relative stride lengths. These deviations from dynamic similarity of locomotion have been attributed to differences in the relative lengths of the limb segments and limb posture (Alexander and Jayes, 1983; Gatesy and Biewener, 1991; Abourachid and Renous, 2000; Abourachid, 2001). Crouched and upright limb postures are generally adopted by small and large vertebrate species, respectively, which are clear departures from geometric similarity in body form (Biewener, 1989; Gatesy and Biewener, 1991). The differing gait kinematics and postures of small and large terrestrial animals may be mechanisms for minimising metabolic costs under scale-dependent muscle force, work and power demands; however, empirical evidence for this is lacking. Body weight ( l 3 v 1 ) increases at a faster rate with body size than the strength (i.e. ability to resist forces, crosssectional area l 2 v 2/3 ) of the biological materials, which must support it (Biewener, 1989). An erect limb aligns body weight with each limb bone s long axis reducing mechanical loading on the muscles associated with turning moments about the joints (Biewener, 1989). The cost of muscle force hypothesis for the scaling of limb posture and gait with body size states that the more upright limbs of larger species serve to reduce the large forces that would otherwise have to be exerted by the limb muscles (Biewener, 1989). An alternative to the cost of muscle force approach is that animals of differing size optimise active muscle volume under scale-dependent muscle work and power demands (Usherwood, 2013). A more erect limb requires shorter stance (push-off) periods, reducing fore-aft speed fluctuations and, consequently, muscle work (J kg -1 ) requirements (Usherwood, 2013). Although the same benefits of an upright limb (in terms of reducing muscle work) would apply to smaller

animals, theoretically, their muscle power (J s -1 kg -1 ) requirements may be disproportionately high (Usherwood, 2013). Therefore, a more crouched limb, requiring a longer push-off period, may act to minimise power requirements in smaller animals (Usherwood, 2013). Indeed, for a given relative speed, human (Homo sapiens) toddlers were found to deviate more from work-minimising gaits than adults, via longer relative stance periods (Hubel and Usherwood, 2015). The understanding of the gaits and postures of different sized animals is compromised because the majority of comparisons are conducted between different species (Alexander and Jayes, 1983; Gatesy and Biewener, 1991; Abourachid and Renous, 2000; Abourachid, 2001). Intraspecifically, the sexes may differ not only in body size (Lislevand et al., 2009; Remes and Szekely, 2010), but also in morphological proportions, which are likely to influence muscle force, work and power demands. For example, in many vertebrate species, the relative proportions of total body mass (Mb) allocated to different somatic and reproductive components are usually biased towards males and females, respectively (Shine et al., 1998; Hammond et al., 2000; Lourdais et al., 2006). Furthermore, female reproductive specialisation may even require specific skeletal proportions (e.g. a wider pelvis (Baumel, 1953; Smith et al., 2002; Cho et al., 2004)), or posture, during pregnancy (Franklin and Conner-Kerr, 1998) or gravidity (Rose et al., 2015b). Most studies on gait kinematics, however, have been conducted using either individuals of only one sex (Reilly, 2000); without comparing sexes (Rubenson et al., 2004; Watson et al., 2011); or using individuals whose sexes were not reported (Gatesy and Biewener, 1991; Abourachid, 2000; Abourachid and Renous, 2000; Abourachid, 2001; Griffin et al., 2004; Nudds et al., 2010). Previous studies have identified sex differences in walking kinematics in humans (Bhambhani and Singh, 1985) and two species of bird (Lees et al., 2012; Rose et al., 2014), but whether size

variations alone, or both size, and additional unidentified sexual dimorphisms were behind the differences in kinematics was not determined. The leghorn chicken (Gallus gallus domesticus) is highly dimorphic, with males having greater body size and muscle mass than females (Mitchell et al., 1931; Rose et al., 2016a). Female leghorns have greater digestive organ masses than males and remain permanently gravid (Mitchell et al., 1931). Furthermore, leghorns are selectively bred for standard (large) and bantam (small) varieties, and only the standard variety is sexually dimorphic in limb posture, with females possessing a more upright limb than males at mid stance during a walking gait (Rose et al., 2015b). For a given sex, the two varieties are expected to be closer to geometric similarity in anatomical proportions (a prerequisite for dynamic similarity of motion). Whilst the males of the two varieties are geometrically similar in their axial and appendicular skeletons, the bantam males adopt a more upright posture at mid stance than the standards during a walking gait (Rose et al., 2015a). The morphological variations to have resulted from selective breeding in these leghorns provide a novel opportunity to investigate the effects of limb posture and differing relative locomotor muscle, digestive and reproductive tissue masses (i.e. varied muscle force, work and power demands) on walking dynamics. Here, high-speed videography and morphological measurements were used to test the hypothesis that male and female standard and bantam varieties of leghorn would show clear departures from dynamic similarity of motion associated with their morphological variations.

MATERIALS AND METHODS Animals Male and female bantam brown leghorns (B and B ) and standard breed white leghorns (L and L ) were obtained from local suppliers and housed in the University of Manchester s Animal Unit. All leghorns (> 16 weeks < 1 year) had reached sexual maturity and females were gravid. Sexes and varieties were housed separately with ad libitum access to food, water and nesting space. Birds were trained daily for a week to sustain locomotion for ~5 min within a Perspex chamber mounted upon a Tunturi T60 (Turku, Finland) treadmill. The kinematics of twenty-four of the twenty-eight leghorns used for the simultaneously collected metabolic measurements described in (Rose et al., 2015b) are presented here (B : N = 9; 1.39 ± 0.03 kg; B : N = 5; 1.04 ± 0.03 kg, L : N = 5; 1.92 ± 0.13 kg, L : N = 5; 1.43 ± 0.06 kg, mean ± s.e.m). All experiments were approved by the University of Manchester s ethics committee, carried out in accordance with the Animals (Scientific procedures) Act (1986) and performed under a UK Home Office Project Licence held by Dr Codd (40/3549). Kinematics The left greater trochanter of the hip of each bird was located by hand and any overlying feathers were removed and replaced with a reflective marker. Each leghorn was exercised at a minimum speed of 0.28 m s -1 and at increasing increments of 0.14 (in a randomised order), up to the maximum they could sustain without showing signs of fatigue. The birds were rested between speed trials. All trials were filmed from a lateral view (left of each bird) using a video camera (HDR-XR520VE, Sony, Japan, 100 frames s -1 ). All video recordings were analysed using Tracker software (Open Source Physics). Distance was calibrated for each video recording using a known distance from the front to the

back of the respirometry chamber. This allowed for the alignment of a calibration tool through the line of travel of each bird (always passing through digit 3), eliminating any error that could be incurred by a bird s displacement from it (i.e. the bird s position on the treadmill/distance from the camera did not affect our distance calibration). At each speed, the phasing of the sum of the vertical kinetic and gravitational potential energies with the horizontal kinetic energy of the body centre of mass (approximated by the trochanter marker) was determined using spatial and temporal data. Unlike the males, the female leghorns are either incapable or unwilling to use grounded running gait mechanics (Rose et al., 2015b). Hence, only data for speeds at which the birds used walking gait mechanics (out of phase fluctuation of gravitational potential and horizontal kinetic energy) were used in the analyses. The left foot of each bird was tracked across ~10 continuous strides (constant speed and position) to obtain the times of toe-on and toe off, which were used to calculate DF, stride frequency (fstride, Hz), stride length (lstride= U/fstride, m), swing duration (tswing, s) and stance duration (tstance, s). A single fixed measure of hip height (hhip) was used per individual (see the following morphological measurements section) as the characteristic length for normalising their speed and gait kinematic parameters. U was normalised for size differences as the square root of Froude, here termed relative speed ( û =U / gh hip ). Kinematic parameters were normalised based on the Hof (1996) record of non-dimensional forms of mechanical quantities as: relative stride length ( ˆL = lstride/hhip); relative stride frequency ( ˆF = f stride g hhip ); relative swing duration ( ˆTswing = t swing hhip g ); and relative stance duration ( ˆT stance = t stance ). hhip g

Morphological measurements For each experimental bird, hhip was measured from a video recording (accuracy, ± 1 mm) of a slow walking speed (0.28 m s -1 ) for a minimum of 7 strides. hhip was taken (using the same method as Rose et al (2015a,b)) as the distance from the treadmill belt (where digit one meets the base of the tarsometatarsus) to the hip marker at 90 to the direction of travel during mid stance, when at its greatest. Back height (hback) was measured the same way as hhip. Digital vernier callipers (accuracy, ± 0.01 mm) were used to measure hind limb long bone (femur, tibiotarsus and tarsometatarsus) lengths (lfem, ltib, ltars) and widths (wfem, wtib, wtars). Total leg length (Σlsegs) was calculated as the sum of the three element lengths. Note that Σlsegs does not represent the true functional length of the hindlimb, because the measurements were taken from dried bones excluding the inter-joint soft tissue. The measurements were taken as the shortest distance between the most proximal and distal grooves of each element, which would further decrease the values of Σlsegs relative to the maximum potential length of the three leg segments if they were vertically aligned. Therefore, a posture index near the value of 1.00 at mid stance in the present study is not indicative of a completely upright limb. The width of the pelvis (wpelv = the distance between the left and right acetabula) was also measured. Soft tissue components from 5 experimental individuals of each bird group were dissected and weighed upon electronic scales (accuracy, ± 0.01 g). Thirteen major muscles of the right pelvic limb (m.iliotibialis cranialis, m. iliotibialis lateralis, m. iliofibularis, m. flexor cruris lateralis pars pelvica, m. flexor cruris medialis, m. iliotrochantericus caudalis, m. femerotibialis medialis, m. pubioischiofemoralis pars lateralis, m. pubioischiofemoralis pars medialis, m. gastrocnemius pars lateralis, m. gastrocnemius pars medialis, m. fibularis lateralis and m. tibialis cranialis) were dissected for a sister study on variety- and sexspecific muscle architectural properties (Rose et al., 2016a). The masses of these muscles

were summed to give a comparable total pelvic limb muscle mass between chicken groups. The right breast muscles (m. pectoralis and m. supracoracoideus) and the intestines (small and large combined) were also weighed. Reproductive mass (developing eggs, ovaries and oviduct) was measured from female birds only and was assumed negligible in males in terms of influencing locomotion dynamics. For the bantam females, 4 of the 5 reproductive masses measured were from individuals whose experimental data were not included in this present study. These individuals, however, were from the same cohort and underwent the same training and experimental process as the birds for which kinematic data are presented here. All anatomical components were compared between varieties and sexes as a percentage of dead bird body mass. Statistical analyses All statistical analyses were conducted in R (v. 3.0.2 GUY 1.2 Snow Leopard build 558) (Team, 2011). The Car package (Fox and Weisberg, 2011) was used for all analyses of variance (ANOVA) in which variety and sex were included as fixed factors. Shapiro-Wilk tests were performed on the standardised residuals generated by all statistical models to ensure the data conformed to a normal distribution. Where morphological data (Table 1) did not conform to a normal distribution even after log transformation, a Kruskal Wallis test was conducted to compare the means of the four groups: B, B, L, and L. Dunn post-hoc tests were used to indicate which groups differed. The relationships between absolute and relative kinematics variables with U and û were compared between the bird groups using linear models. U and û were included in the models as covariates and all potential interaction terms were considered before a stepwise backwards deletion of non-significant interaction terms was conducted to simplify the models. Outputs from the final models are reported. Best-fit

lines were obtained from the coefficients tables of the final statistical models and were back transformed where data had been log transformed. RESULTS Morphological measurements Body mass (Fig 1A) and total leg length (Σlsegs, Fig 1B) were greater in the standard than in the bantam variety and greater in males than females (Table 1). Hip height (hhip, Fig 1C), however, was greater in males than in females in the bantam variety, but, conversely, greater in females than in males in the standard variety (Table 1). Posture index (hhip:σlsegs, Fig 1D) did not differ between the sexes of the bantam variety. In contrast, the posture index of L was ~27 % greater than that of L, indicative of a more erect limb (Table 1). Each hind limb segment was a similar proportion of Σlsegs (Table 2) in all of the leghorn groups excluding B, which had a relatively longer lfem, and concomitantly shorter ltars, resulting in a small, but, nonetheless statistically significant difference (Table 2). The width of each limb segment (Table 2) was a similar proportion of its respective segment length in all groups (Table 2). The finding of a more erect posture in L when compared to the other three groups was further supported by indices incorporating the height of the back (hback). hhip:hback did not differ between the bird groups, and Σlsegs:hback was lower in L than in the other three leghorn groups (Table 2). wpelv, relative to Σlsegs (Table 2), did not differ between the sexes, but was ~1 % greater in the bantams than in the standards (Table 2). Total pelvic limb muscle mass (the sum of the masses of thirteen pelvic limb muscles) was a greater % of body mass (Fig 2) in males than in females in both varieties (Table 1). Total pelvic limb muscle mass was also greater in the bantam than in the standard variety (Fig 2, Table 1). The observed differences between varieties, however, were small in comparison to the sexual dimorphisms. The same statistical outcomes obtained for the total

pelvic limb muscle mass were mirrored by the breast muscles, m. pectoralis and m. supracoracoideus (Fig 2, Table 1). Intestine mass as a proportion of body mass (Fig 2) was greater in females than in males (Table 1). The two varieties, however, shared similar relative intestinal masses. The reproductive tissue mass as a percentage of body mass was greater in L (11.49 ± 0.56 %) than in the B (8.40 ± 0.08 %). Therefore, all four leghorn groups were similar in their hind limb skeletal bone geometry (a prerequisite of dynamic similarity of locomotion). The sexes, however, differed markedly in each of the measured anatomical masses when expressed as % body mass. In contrast, with the exceptions of limb posture and the relative mass of the female reproductive system, for a given sex, the two varieties of leghorn were more similar in their anatomical proportions. Walking kinematics and dynamics DF, tswing and tstance were negatively correlated, and lstride and fstride were positively correlated with U. The same correlations were also found for the relationship between size-normalised kinematics parameters and û. For all four groups of leghorns, the exponents describing the relationships between absolute or size-normalised parameters and U or û were similar, unless otherwise stated below. Across all speeds, DF (Fig 3A) was greater in females than in males (~ 2 %) and greater in the bantam than in the standard variety by ~ 2 % (Table 3). At comparable û, DF was, similarly, greater in females than in males in both varieties and this sex difference was greater in the bantams. In addition, DF was greater in the bantam than in the standard variety (Fig 3B; Table 3). fstride (Fig 3C) was greater in females compared to males at any given U (0.11 Hz) (Table 3). Absolute fstride and the rate of increase in fstride with U were significantly greater in

the bantam than in the standard variety. At comparable û, ˆF, was greater in females compared to males in the standard variety but, contrastingly, greater in males compared to females in the bantam variety (Fig 3D; Table 3). At any given U, lstride was greater in males than in females by a greater amount in the standard variety (70 mm) than in the bantam variety (20 mm) (Fig 1C; Table 2) and was associated with a greater difference between the males of the two varieties than between the females of the two varieties. At any given û, ˆL (Fig 1D), was greater in females than in males in the bantam variety, but, contrastingly, was greater in males than in females in the standard variety (Table 3). At each U, tswing (Fig 1G) was greater in males than in females and this difference was lower in the bantam (0.02 s) than in the standard (0.04 s) variety (Table 2). In the bantams, Tˆ swing (Fig 1H), was similar in the two sexes, whilst in the standard variety, it was significantly greater in males compared to females at a given û. tstance (Fig 1I) was similar in B and B, but was greater in L compared to L by 0.03s at all U (Table 2). Across all û, T ˆ stance (Fig 1J; Table 2) was greater in B than in B, yet lower in L than in L. Therefore, none of the sex or variety differences in gait kinematics were accounted for correcting for body size differences. The two varieties differed in the mechanisms by which females had elevated DF relative to males. In the bantams, females had relatively longer stance durations than males (Fig 1J) and the sexes shared similar swing dynamics (Fig 1H). In the standard variety, females had relatively shorter swing and stance durations than males, but the sex difference in Tˆ swing was much greater than that in T ˆ stance (Fig 1H and J).

DISCUSSION This study represents the first detailed comparison of the gait of the sexes in any species. Leghorn chickens, although all similar in their hind limb segment geometry, show considerable variation in limb posture and the relative contributions of anatomical components (skeletal muscle, digestive organs and reproductive tissues) to total body mass. In association with these morphological differences, and in agreement with our hypothesis, none of the leghorn groups walked with dynamic similarity. Incremental responses of absolute kinematics parameters to increasing U are generally greater in smaller species (Gatesy and Biewener, 1991). With the exception of fstride, which increased at a faster rate in bantams than in the standard variety, all birds in the present study, however, showed similar incremental kinematic responses to U, despite the size differences associated with sex and variety. Most of the sex differences in absolute kinematic parameters paralleled inter-species differences, associated with body size (Gatesy and Biewener, 1991). In females of the two varieties, fstride was greater, and lstride, tswing and tstance, smaller, at any given U compared to that in their conspecific males, which had greater body size (except for tstance in the bantams, which was similar between the sexes). Similarly, fstride was greater, and lstride, tswing and tstance shorter at any given U in the bantams compared to the standards. The only parameter that was not comparable to inter-species patterns associated with body size for a given absolute speed was DF. Interspecific scaling patterns, would predict the heavier and longer-legged, animal to have a greater DF than the lighter, shorterlegged one, at the same U (Gatesy and Biewener, 1991). In contrast, here, females walked with greater DF than males, and bantams walked with greater DF than standards. In agreement with body size dependent inter-species differences in DF measured at equivalent relative speeds (Alexander and Jayes, 1983; Gatesy and Biewener, 1991), at any

given relative speed, DF here, was still higher in the smaller bantam relative to the standard variety, and in females relative to males. Deviations from dynamic similarity of motion with increasing body mass are usually associated with increasing limb erectness, i.e., an increasing hhip to Σlseg ratio (Biewener, 1987, 1989; Gatesy and Biewener, 1991). Smaller, more crouched, species can achieve greater lstride relative to their hhip because they can extend the crouched limb, which in turn, allows a greater DF (Gatesy and Biewener, 1991). In contrast, a more erect limb is constrained in terms of the range of lstride and DF it can achieve, relative to a given hhip (Gatesy and Biewener, 1991). The similar pelvic limb skeletal geometry of the birds in the present study provides a control for the potential effects of limb segment proportions on walking dynamics and allows investigation into the potential influences of additional morphological characteristics including limb posture. Despite L having the most upright limbs, and the lowest relative stride lengths (Fig 1 E), however, they still produced a greater DF relative to hhip than did the L, whose limbs were more crouched. Furthermore, since sexual dimorphism in limb posture was exclusive to the standards, limb posture cannot explain the similar sex difference in DF in the two varieties. The sexual dimorphism in posture in the standard variety only was reflected in the opposite sex-specific dynamics of ˆL, ˆF, T ˆ swing and T ˆ stance at any given û between the two varieties (i.e the sex differences in gait were variety-specific, yet ultimately led to greater DF in females than males). The lower ˆL in L than in B and higher ˆF in L than in B are consistent with the general consensus that a more upright limb limits the length of a step relative to hhip (Gatesy and Biewener, 1991). By adopting a more upright limb posture, larger animals reduce the forces that the muscles must exert and that the bones must resist to counteract joint moments, which would otherwise scale geometrically ( Mb 1/3 ) (Biewener, 1989). Until recently, this has been considered the principal reason for the scaling of limb posture and gaits in vertebrates (Biewener, 1989). Why smaller animals do not also have upright limbs so that they could

have relatively more gracile and lighter bones, however, is not accounted for by this explanation. Explanations proposed to account for a more crouched limb include that it may improve manoeuvrability (Biewener, 1989) stability (Gatesy and Biewener, 1991) or minimize the cost of work associated with bouncing viscera (Daley and Usherwood, 2010). Another potential explanation, however, is that animals optimise muscle mechanical work and power demands during push-off (which are scale-dependent) in order to minimise the volume of active muscle for a given size (Usherwood, 2013). In this case, a crouched limb at mid stance (allowing longer DF) for small animals may serve to reduce power demands (which are high because at any given U shorter legs require quicker stances), whilst a more upright limb suits the work demands of being large (which are high because a disproportionate amount of body weight must be supported). The females in the present study may, therefore, benefit from greater DF, which would decrease the elevated power demands associated with having small limbs, yet greater body weight to support per unit of muscle mass because of gravidity. The L in the present study may have adopted kinematic and postural mechanisms for reducing both the elevated work demands due to gravity (via upright limbs) and the power demands of being small (via longer DF, achieved without a crouched posture). It is possible that L may require a more upright limb than B because of their greater relative reproductive tissue mass. In B minimising power via a greater relative DF (exceeding that in L ) appears to be more important. In guinea fowl, Numida meleagris, adding trunk loads equivalent to 23% of body mass did not affect tswing but led to a 4% increase in tstance (Marsh et al., 2006). In several additional avian species, however, no changes in gait kinematics were associated with the application of trunk loads (McGowan et al., 2006; Tickle et al., 2010; Tickle et al., 2013). These experiments, however, involve unnatural loads applied in backpacks and may not represent a true gravid loading condition. The hypothesis that the kinematics of leghorn hens

are influenced by muscle mechanical demands associated with gravidity is further supported by the finding than DF increases with the onset of egg laying during sexual maturation in leghorn hens (Rose et al., 2016b). Equally, males may benefit from lower DF via the minimization of work demands associated with changes in fore-aft acceleration and deceleration, because of being larger. The hypothesis of Usherwood (2013), that animals adopt kinematic and postural mechanisms to minimise active muscle volume (according to work and power demands) in order to minimize metabolic costs is supported by the present findings together with previously published data on locomotor energy metabolism collected simultaneously from the same birds (Rose et al., 2015a,b). The metabolic cost of transport in gravid female leghorns is in fact lower than allometric predictions based on body mass (Vankampen, 1976; Rose et al., 2015b) and also lower than that of male leghorns (Rose et al., 2015b), which can be linked to their comparatively greater DF. A lower metabolic cost of transport in L than in B (Rose et al., 2015b) can also be linked to more upright limbs. Additionally, greater relative DF and more upright limb posture may contribute to a lower than expected metabolic cost of transport in B for their body mass and the lack of scaling in the metabolic cost of transport between males of the two varieties (Rose et al., 2015a). Alternatively/additionally, the greater DF in females relative to males may be a mechanism for reducing peak muscle forces in supporting body weight, which may again be particularly important when carrying proportionally more weight (due to greater digestive/reproductive tissue volume) with proportionally less muscle volume. Furthermore, chickens artificially selected for egg production are well known to suffer from osteoporosis associated with the utilisation of calcium from limb bone medullary in order to form eggshells (Dacke et al., 1993; Whitehead, 2004). A greater proportion of ground contact throughout the stride to reduce peak forces exerted on the bones may reduce the risk of bone

fracture. This argument may be particularly pertinent to L due to the fact that the pelvic limb skeletal element diameters of the two varieties conformed to geometric, and not elastic (positive allometry), scaling as is found between species (Doube et al., 2012). The bones of L are, therefore, not expected to be any more robust than those of B to assist with supporting proportionally more weight of the reproductive system. The same reasoning may also explain the upright limbs of L. It is also possible that additional sexual dimorphisms, perhaps in muscle physiology or morphology, are linked to the sex differences in dynamics. For example, simply the distribution of mass across the limb segments may be responsible. In a recent study in which the swing phase kinematics of different charadriiform birds were compared, Northern lapwings (Vanellus vanellus) and Eurasian oystercatchers (Haematopus), which share similar hind limb long bone proportions, shared similar tswing at any given speed despite oystercatchers having longer and heavier limbs overall (Kilbourne et al., 2016). In comparison to these two species, pied avocets (Recurvirostra avosetta) moved with longer swing durations at higher speeds linked to a more distal concentration of hind limb mass (Kilbourne et al., 2016). The greater relative pelvic limb muscle mass in males, relative to females (Mitchell et al., 1931; Rose et al., 2016a), may similarly increase limb moment of inertia and prolong the swing phase of the limb, increasing its contribution to the stride period. In summary, this study represents the first detailed comparison of male and female gait in a bird. Clear departures from dynamic similarity of motion were evident between the sexes in standard and bantam varieties of leghorn chicken. Females walked with greater DF than males at any given relative speed, but this sex difference was achieved through alternative kinematic mechanisms in each variety and linked to variety differences in sexspecific posture. L carry a greater relative reproductive mass than B and potentially

represent the first documented example of an animal adopting mechanisms for minimising the demands of both work (via an upright limb, relative to B ) and power (via a longer DF than their heavier, more crouched male conspecifics). List of abbreviations B B DF fstride Fr ˆF lfem lstride hhip hhip: Σlsegs L L ˆL ltib ltars tstance Tˆ stance tswing Tˆ U swing female bantams male bantams duty factor stride frequency Froude number normalised stride frequency femur length stride length hip height posture index female standards male standards normalised stride length tibiotarsus length tarsometatarsus length stance duration normalised stance duration swing duration normalised swing duration speed

û wfem wtib wtars Σlsegs relative speed femur width tibiotarsus width tarsometatarsus width sum of the hind limb long bone lengths

Acknowledegments We thank two anonymous reviewers for their helpful comments on an earlier version of this manuscript. Competing interests The authors have no competing interests Author s contributions R.L.N, J.R.C and K.A.R designed the study and contributed to the manuscript. K.A.R conducted the experiments and statistical analyses with advice from R.L.N and J.R.C. Funding This research was funded by the BBSRC (G01138/1 and 10021116/1 to J.R.C). K.A.R was supported by a NERC DTA stipend and CASE partnership with the Manchester Museum.

REFERENCES Abourachid, A. (2000). Bipedal locomotion in birds: the importance of functional parameters in terrestrial adaptation in Anatidae. Can J Zool 78, 1994-1998. Abourachid, A. (2001). Kinematic parameters of terrestrial locomotion in cursorial (ratites), swimming (ducks), and striding birds (quail and guinea fowl). Comparative Biochemistry and Physiology a-molecular and Integrative Physiology 131, 113-119. Abourachid, A. and Renous, S. (2000). Bipedal locomotion in ratites (Paleognatiform): examples of cursorial birds. Ibis 142, 538-549. Alexander, R. M. (1976). Estimates of Speeds of Dinosaurs. Nature 261, 129-130. Alexander, R. M. and Jayes, A. S. (1983). A dynamic similarity hypothesis for the gaits of quadrupedal mammals. J Zool 201, 135-152. Baumel, J. J. (1953). Individual variation in the white-necked raven. Condor 55, 26-32. Bhambhani, Y. and Singh, M. (1985). Metabolic and Cinematographic Analysis of Walking and Running in Men and Women. Med Sci Sport Exer 17, 131-137. Biewener, A. A. (1987). Scaling body support in sammals - limb posture and the relative mechanical advantage of limb muscles. Am Zool 27, A25-A25. Biewener, A. A. (1989). Scaling body support in mammals - limb posture and muscle mechanics. Science 245, 45-48. Cho, S. H., Park, J. M. and Kwon, O. Y. (2004). Gender differences in three dimensional gait analysis data from 98 healthy Korean adults. Clin Biomech 19, 145-152. Dacke, C. G., Arkle, S., Cook, D. J., Wormstone, I. M., Jones, S., Zaidi, M. and Bascal, Z. A. (1993). Medullary Bone and Avian Calcium Regulation. J Exp Biol 184, 63-88.

Daley, M. A. and Usherwood, J. R. (2010). Two explanations for the compliant running paradox: reduced work of bouncing viscera and increased stability in uneven terrain. Biol Lett 6, 418-421. Doube, M., Yen, S. C. W., Klosowski, M. M., Farke, A. A., Hutchinson, J. R. and Shefelbine, S. J. (2012). Whole-bone scaling of the avian pelvic limb. J Anat 221, 21-29. Fox, J. and Weisberg, S. (2011). An {R} companion to applied regression, second edition. Thousand Oaks CA: sage. Franklin, M. E. and Conner-Kerr, T. (1998). An analysis of posture and back pain in the first and third trimesters of pregnancy. J. Orthop. Sports Phys. Ther. 28, 133-138. Gatesy, S. M. and Biewener, A. A. (1991). Bipedal locomotion - effects of speed, size and limb posture in birds and humans. J Zool 224, 127-147. Griffin, T. M., Kram, R., Wickler, S. J. and Hoyt, D. F. (2004). Biomechanical and energetic determinants of the walk-trot transition in horses. J Exp Biol 207, 4215-4223. Hammond, K. A., Chappell, M. A., Cardullo, R. A., Lin, R. S. and Johnsen, T. S. (2000). The mechanistic basis of aerobic performance variation in red junglefowl. J Exp Biol 203, 2053-2064. Hof, A. L. (1996). Scaling gait data to body size. Gait Posture 4, 222-223. Hubel, T. Y. and Usherwood, J. R. (2015). Children and adults minimise activated muscle volume by selecting gait parameters that balance gross mechanical power and work demands. J Exp Biol 218, 2830-2839. Jenkins, F. A. (1971). Limb Posture and Locomotion in Virginia Opossum (Didelphis- Marsupialis) and in Other Non-Cursorial Mammals. J Zool 165, 303-&. Kilbourne, B. M., Andrada, E., Fischer, M. S. and Nyakatura, J. A. (2016). Morphology and motion: hindlimb proportions and swing phase kinematics in terrestrially locomoting charadriiform birds. The Journal of experimental biology 219, 1405-1416.

Lees, J. J., Nudds, R. L., Folkow, L. P., Stokkan, K. A. and Codd, J. R. (2012). Understanding sex differences in the cost of terrestrial locomotion. Proc R Soc B 279, 826-832. Lislevand, T., Figuerola, J. and Szekely, T. (2009). Evolution of sexual size dimorphism in grouse and allies (Aves: Phasianidae) in relation to mating competition, fecundity demands and resource division. J Evolution Biol 22, 1895-1905. Lourdais, O., Shine, R., Bonnet, X. and Brischoux, F. (2006). Sex differences in body composition, performance and behaviour in the Colombian rainbow boa (Epicrates cenchria maurus, Boidae). J Zool 269, 175-182. Marsh, R. L., Ellerby, D. J., Henry, H. T. and Rubenson, J. (2006). The energetic costs of trunk and distal-limb loading during walking and running in guinea fowl Numida meleagris: I. Organismal metabolism and biomechanics. The Journal of experimental biology 209, 2050-2063. McGowan, C. P., Duarte, H. A., Main, J. B. and Biewener, A. A. (2006). Effects of load carrying on metabolic cost and hindlimb muscle dynamics in guinea fowl (Numida meleagris). J Appl Physiol 101, 1060-1069. Mitchell, H. H., Card, L. E. and Hamilton, T. S. (1931). A technical study of the growth of White Leghorn chickens. Illinois agricultural experiment station bulletin 367, 81-139. Nudds, R. L., Gardiner, J. D., Tickle, P. G. and Codd, J. R. (2010). Energetics and kinematics of walking in the barnacle goose (Branta leucopsis). Comp Biochem Physiol A Mol Integr Physiol 156, 318-324. Reilly, S. M. (2000). Locomotion in the quail (Coturnix japonica): The kinematics of walking and increasing speed. J Morphol 243, 173-185. Remes, V. and Szekely, T. (2010). Domestic chickens defy Rensch's rule: sexual size dimorphism in chicken breeds. J Evolution Biol 23, 2754-2759.

Rose, K. A., Nudds, R. L. and Codd, J. R. (2015a). Intraspecific scaling of the minimum metabolic cost of transport in leghorn chickens (Gallus gallus domesticus): links with limb kinematics, morphometrics and posture. The Journal of experimental biology 218, 1028-1034. Rose, K. A., Nudds, R. L. and Codd, J. R. (2016a). Variety, sex and ontogenetic differences in the pelvic limb muscle architectural properties of leghorn chickens (Gallus gallus domesticus) and their links with locomotor performance. J Anat 228, 952-964. Rose, K. A., Nudds, R. L., Butler, P. J. and Codd, J. R. (2015b). Sex differences in gait utilization and energy metabolism during terrestrial locomotion in two varieties of chicken (Gallus gallus domesticus) selected for different body size. Biol Open 4, 1306-1315. Rose, K. A., Bates, K. T., Nudds, R. L. and Codd, J. R. (2016b). Ontogeny of sex differences in the energetics and kinematics of terrestrial locomotion in leghorn chickens (Gallus gallus domesticus). Sci Rep-Uk 6, 24292. Rose, K. A., Tickle, P. G., Lees, J. J., Stokkan, K. A. and Codd, J. R. (2014). Neither season nor sex affects the cost of terrestrial locomotion in a circumpolar diving duck: the common eider (Somateria mollissima). Polar Biol 37, 879-889. Rubenson, J., Heliams, D. B., Lloyd, D. G. and Fournier, P. A. (2004). Gait selection in the ostrich: mechanical and metabolic characteristics of walking and running with and without an aerial phase. Proc R Soc B 271, 1091-1099. Shine, R., Keogh, S., Doughty, P. and Giragossyan, H. (1998). Costs of reproduction and the evolution of sexual dimorphism in a 'flying lizard' Draco melanopogon (Agamidae). J Zool 246, 203-213.

Smith, L. K., Lelas, J. L. and Kerrigan, D. C. (2002). Gender differences in pelvic motions and center of mass displacement during walking: Stereotypes quantified. J Women Health Gen-B 11, 453-458. Team, R. D. C. (2011). R: A language and environment for statistical computing: Vienna, Austria: R Foundation for Statistical Computing. Tickle, P. G., Richardson, M. F. and Codd, J. R. (2010). Load carrying during locomotion in the barnacle goose (Branta leucopsis): The effect of load placement and size. Comp. Biochem. Physiol. Part A Mol. Integr. Physiol. 156, 309-317. Tickle, P. G., Lean, S. C., Rose, K. A., Wadugodapitiya, A. P. and Codd, J. R. (2013). The influence of load carrying on the energetics and kinematics of terrestrial locomotion in a diving bird. Biol Open 2, 1239-1244. Usherwood, J. R. (2013). Constraints on muscle performance provide a novel explanation for the scaling of posture in terrestrial animals. Biol Lett 9. Vankampen, M. (1976). Activity and Energy-Expenditure in Laying Hens.2. Energy- Cost of Exercise. J Agr Sci 87, 81-84. Watson, R. R., Rubenson, J., Coder, L., Hoyt, D. F., Propert, M. W. and Marsh, R. L. (2011). Gait-specific energetics contributes to economical walking and running in emus and ostriches. Proc R Soc B 278, 2040-2046. Whitehead, C. C. (2004). Overview of bone biology in the egg-laying hen. Poultry Sci 83, 193-199.

Figures

Figure 1. Morphological measurements for the variety and sex combinations of leghorn chicken. (a) Body mass. (b) Total leg length (sum of the three skeletal element lengths). (c) Hip height. (d) Posture index (hip height: total leg length). Bars represent means ± s.e.m for standard males (grey), standard females (purple), bantam males (blue) and bantam females (red). Significant variety, sex and variety x sex interaction terms are denoted by V, S and I, respectively. Significance levels of 0.05, 0.01 and 0.001 are denoted by *, ** and ***, respectively. Results of the two way ANOVAs conducted to test for variety and sex differences are in Table 1. These morphological differences have been reported previously in (Rose et al., 2015b) for a different sample size.

Figure 2. Anatomical components as a % of body mass for the variety and sex combinations of leghorn chicken. Bars represent means ± s.e.m for standard males (grey), standard females (purple), bantam males (blue) and bantam females (red). Limb muscle mass was calculated as the sum of thirteen major pelvic limb muscles on the right limb. M. pectoralis and M. supracoracoideus expressed as a % of body mass are for the right side of the body only. Reproductive mass was assumed negligible in the males in terms of influencing gait. Significant variety and sex effects are denoted by V and S, respectively. Significance levels of 0.05, 0.01 and 0.001 are denoted by *, ** and ***, respectively. Results of the one- and two-way ANOVAs conducted to test for variety and sex differences are in Table 1.

Figure 3. Absolute gait kinematics parameters versus speed (left column) and relative gait kinematics parameters versus relative speed (right column). Duty factor (a-b), stride length (c-d), stride frequency (e-f), swing duration (g-h) and stance duration (i-j). Each data point (standard males = grey circles, standard females = purple crosses, bantam males = blue circles and bantam females = red crosses) represents a single trial for an individual bird. Bestfit line equations and the results of the linear models conducted to test for variety and sex differences are in Table 3.

Table 1. Results of the one- and two-way ANOVAs that tested for variety and sex differences in morphological measurements/indices. Measurement/index Body mass (kg) Leg length (mm) Hip height (mm) Posture index 13 pelvic limb muscles (%body mas) M. pectoralis (% body mass) M. supracoracoideus (%body mass) Intestines a (%body mass) Reproductive (%body mass) b ANOVA (final model) variety: F1,20= 47.34, P<0.001, sex: F1,20=33.96, P<0.001 R 2 = 0.76 variety: F1,20=125.26, P<0.001, sex: F1,20=84.44, P<0.001 R 2 = 0.89 variety: F1,19=95.44 P<0.001, sex: F1,19=0.29, P=0.594 variety x sex: F1,19=29.55, P<0.001, R 2 = 0.85 variety: F1,19=14.42, P=0.001, sex: F1,19=19.44, P<0.001 variety x sex: F1,19=42.30, P<0.001, R 2 =0.75 variety: F1,17=9.10, P=0.008, sex: F1,17=161.53, P<0.001, R 2 =0.90 variety: F1,17=18.50, P<0.001 sex: F1,17=31.00, P<0.001, R 2 =0.71 variety: F1,17=8.35, P=0.010 sex: F1,17=29.01, P<0.001, R 2 =0.65 variety: F1,16=0.71, P=0.411, sex: F1,16=48.09, P<0.001, R 2 =0.73 variety: F1,8= 8.74, P=0.018, R 2 = 0.46 a N=4 for standard males b Females only as reproductive mass was assumed negligible in males Body mass, leg length, hip height and posture index were measured from the full sample of experimental birds Soft tissue masses (% body mass) were calculated for 5 individuals of each bird group Breast muscle %body masses are for the right side of the body only The adjusted R 2 values of the final statistical models are reported

Table 2. Mean (± s.e.m) morphometric indices and results of the statistical tests conducted to investigate whether the indices differed between varieties and sexes. Index B B L L Statistical results lfem:σlsegs 0.28 0.29 0.28 0.28 Kruskal Wallis: X 2 = 19.6, df=3, P=<0.001 Dunn test: B v B : Z=3.71, P<0.001, B v L : Z=-0.61, P=0.272, B v L : Z=0.00, P=0.500, B v L : Z=-3.89, P<0.001, B v L : Z=-3.35, P<0.001, L v L : Z=0.55, P=0.292 ltib:σlsegs 0.42 0.42 0.42 0.42 Kruskal Wallis: X 2 = 0.13, df=3, P=0.988 ltars:σlsegs 0.30 0.29 0.30 0.30 Kruskal Wallis: X 2 = 12.47, P=0.006 Dunn test B v B : Z=-2.48, P=0.007, B v L : Z=0.90, P=0.18 B v L : Z=0.90, P=0.18, B v L : Z=3.05, P=0.001, B v L : Z=3.05, P=0.001, L v L : Z=0.00, P=0.500 wfem:lfem 0.11 0.11 0.11 0.11 Kruskal Wallis: X 2 = 0.46, df=3, P=0.929 wtib:ltib 0.07 0.06 0.07 0.07 Kruskal Wallis: X 2 = 7.20, df=3, P=0.066 wtars:ltars 0.10 0.10 0.10 0.10 Kruskal Wallis: X 2 = 0.36, df=3, P=0.948 hhip:hback 0.78 ± 0.01 0.77 ± 0.02 0.76 ± 0.00 0.77 ± 0.02 variety: F1,20=0.45, P=0.512, sex: F1,20=0.00, P=0.948, R 2 =0.00 Σlsegs:hback 1.01 ± 1.02 ± 1.03 ± 0.76 ± variety: F1,19=20.58, P<0.001, sex: F1,19=22.16, P<0.001, 0.03 wpelv:σlsegs 0.17 (N=6) 0.02 0.01 0.02 variety x sex: F1,19=37.91, P<0.001, R 2 =0.78 0.18 0.16 0.17 variety: F1,18=11.368, P=0.003, sex: F1,18=0.347, P=0.079, R 2 =0.39 The adjusted R 2 values of the final models are reported. Only the results of the final two-way ANOVAs are reported Standard errors are not presented where they were 0.00 to 2 decimal places The sample size is indicated in brackets if lower than the total sample size of the leghorn group

Table 3. Results of the linear models that tested for sex differences in absolute/normalised kinematics with speed. Parameter Final model Lines of best fit DF U (F 1,111=315.80, P<0.001), variety (F 1,111=7.38, P=0.008), sex (F 1,111=27.79, P<0.001), R 2 = 0.79 B : = -0.16U+0.79 B : = -0.16U+0.81 L : = -0.16U + 0.77 L : = -0.16U+0.79 DF û (F 1,110=276.78, P<0.001), B : = -0.23 û + 0.79 variety (F 1,110=30.71, P<0.001), B : = -0.23 û + 0.82 sex (F 1,110=25.30, P<0.001), L : = -0.23 + 0.77 variety x sex (F 1,110=5.58, û P=0.020), L : = -0.23 + 0.78 f stride R 2 = 0.77 U (F 1,110=706.18, P<0.001), variety (F 1,110=204.13, P<0.001), sex (F 1,110=24.64, P<0.001), U x variety (F 1,110=16.74, P=<0.001), R 2 = 0.87 (F 1,110=615.06, P<0.001), variety (F 1,110=1.32, P=0.253), sex (F 1,110=26.84, P<0.001), variety x sex (F 1,110=80.66, P<0.001), R 2 = 0.87 log l stride logu (F 1,110=662.71, P<0.001), variety (F 1,110=222.07, P<0.001), sex (F 1,110=40.77, P<0.001), variety x sex (F 1,110=10.93, P=0.001), R 2 = 0.92 log log (F 1,110=663.00, P<0.001), variety (F 1,110=0.03, P=0.862), sex (F 1,110=52.48, P<0.001), variety x sex (F 1,110=151.57 P<0.001), R 2 = 0.88 log t swing logu (F 1,110=94.88, P<0.001), variety (F 1,110=225.41, P<0.001), sex (F 1,110=71.69, P<0.001), variety x sex (F 1,110=4.51, P=0.036), R 2 = 0.73 log log (F 1,110=97.77, P<0.001), variety (F 1,110=40.02, P<0.001), sex (F 1,110=77.88, P<0.001), variety x sex (F 1,110=57.83, P<0.001), R 2 = 0.69 log t stance logu (F 1,110=1431.42, P<0.001), variety (F 1,110=214.12, P=0.079), sex (F 1,110=22.35, P<0.001), variety x sex (F 1,110=13.76, P<0.001), R 2 = 0.93 log log (F 1,110=1057.41, P<0.001), variety (F 1,110=1.85, P=0.177), sex (F 1,110=17.74, P<0.001), variety x sex (F 1,110=126.77, P<0.001), R 2 = 0.92 The adjusted R 2 values of the final statistical models are reporte û B : = 1.76U+ 0.69 B : = 1.76U+ 0.80 L : = 1.24U+ 0.61 L : = 1.24U+ 0.72 ˆF û û B : = 0.30 + 0.12 B : = 0.30 û + 0.10 L : = 0.30 û + 0.08 L : = 0.30 + 0.13 û B : = 0.43U 0.46 B : = 0.41U 0.46 L : = 0.56U 0.46 L : = 0.49U 0.46 ˆL û û 0.23 B : = 2.50 B : 0.23 = 2.78 û L : 0.23 = 2.95 û L : 0.23 = 2.30 û B : = 0.16U -0.22 B : = 0.14U -0.22 L : = 0.22U -0.22 L : = 0.18U -0.22 ˆT swing û û -0.11 B : = 1.04 B : -0.11 = 1.04 û L : -0.11 = 1.31 û L : -0.11 = 1.00 û B := 0.28U -0.64 B : = 0.28U -0.64 L : = 0.36U -0.64 L : = 0.33U -0.64 ˆT stance û û -0.32 B : = 1.60 B : -0.32 = 1.85 û L : -0.32 = 1.85 û L : -0.32 = 0.47 û