Waddle and shuffle: gait alterations associated with domestication in turkeys

Artificial selection of domestic animals has led to alterations in morphology through the desire to enhance certain physical traits. The poultry industry has selected for traits that have decreased the time to market and increased meat production (Yost et al., 2002). The turkey, Meleagris gallopavo Linnaeus 1758, in particular has undergone large changes in muscle morphology and mass, with some broad-breasted white (domestic) strains reaching over 3 times the body mass of wild turkeys (Stover et al., 2018). Selection for breast meat has also resulted in pectoral hypertrophy in the domestic strains (Stover et al., 2018; Velleman et al., 2003), leading to an altered distribution of body mass (Abourachid, 1993; Stover et al., 2018). All of this extra body mass is supported by hindlimb bones that are only slightly longer than those of wild turkeys (Stover et al., 2018). Structurally, domestic turkeys are not scaled-up wild turkeys, with selection leading to altered skeletal dimensions and proportionally greater pectoral muscle mass.

Given the morphological differences between broad-breasted white and wild turkey strains, we expect that the locomotion of domestic turkeys should be affected by the structural changes resulting from artificial selection. Surprisingly, when Abourachid (1991) investigated the gaits of male traditional and domestic turkeys, no significant differences were detected in duty factory (stance phase percentage of cycle duration) or stride length. However, large lateral oscillations (side-to-side motions) of more than 15 deg in the broad-breasted strain were noted in the posterior view. Some chicken and duck varieties, which have undergone a similar selective regime, experience locomotor consequences of increased body mass including increased stride width, slow walking speeds, increased double support and large lateral motions accompanied by high medio-lateral ground reaction forces (GRFs) (Caplen et al., 2012; Corr et al., 2007; Duggan et al., 2016; Paxton et al., 2013). Selecting birds with a healthy gait is an integral part of the poultry breeding practice, and yet the kinetic characteristics of domestic turkey locomotion have yet to be described. Establishing a baseline gait description is important for being able to identify walking abnormalities associated with musculoskeletal disorders that continue to be a major animal welfare concern (Hocking, 2014; Julian, 1998, 2005).

Wild and domestic turkeys provide an opportunity to discern the relationships between evolved morphological alterations and adjustments in locomotor kinematics and mechanics. Besides domestication, there are other instances of body structure changes inducing locomotor variations within and among species. Obesity in humans is associated with altered features of gait dynamics, including increased double support phase, lateral swaying and higher GRFs (Browning and Kram, 2007). Carrying a heavy load can similarly elicit differences in gait, and the anatomical positioning of the load influences GRFs (Birrell et al., 2007). We can also gain insight by looking at gait differences across a wide size range of bipedal species. One comparative study on bipedal locomotor kinematics included three organisms with a large body mass disparity but similar limb lengths (emu, rhea and human) and found that they had remarkably similar stride frequencies (Gatesy and Biewener, 1991). Identifying gait similarities and variations can reveal links between body structure and motion.

The goal of this study was to identify whether wild and domestic turkeys display similar gait kinematics and kinetics, despite the significant morphological transformations of domestication. The increase in body mass, relatively short hindlimb bones and altered distribution of muscle mass in today's domestic turkeys may make maintaining equivalent gait dynamics difficult.

Eastern wild and broad-breasted white (Hybrid Converter 2013) turkey poults, Meleagris gallopavo, were obtained 3 days post-hatch from licensed breeders and housed in the Animal Care Facilities at Brown University. All turkeys were maintained on an ad libitum water and 28% protein commercial poultry diet for the first 8 weeks and then transitioned to regular poultry feed. The two strains were raised together in a common open pen environment where they could move freely. Domestic females were procured at 18 weeks of age from a local farm and had been raised on pasture until brought into the Animal Care Facility. A second group of laboratory-raised wild females were used for collecting multiple footfalls in the autumn of 2014. Wild female (n=5 and 4), wild male (n=5), domestic female (n=5) and domestic male turkeys (n=6) were used in this study. All animal use was approved by the Brown University Animal Care and Use Committee, IACUC no. 1602000189, and complied with state and federal legislation and regulation.

A 6.25 m track-way covered with treadmill tread for traction was constructed with a space for a force platform at the halfway point. Three different force plate arrangements were used. A single force plate (Kistler 9281B, Kistler USA, Amherst, NY, USA) was masked off to a contact region of 30 cm, about the step length of a wild turkey, to capture single footfalls. The same force plate with no mask or two force plates (AMTI MC3A-100 and MC3A-6-250, Advanced Mechanical Technology, Inc., Watertown, MA, USA) in sequence were used to acquire consecutive footfalls and periods of double support. The width of the track-way was restricted to 0.61 m to keep the turkeys moving along a straight path. Unrestrained turkeys were encouraged to walk across the force platform and GRFs were recorded in the vertical, fore–aft and medio-lateral directions. Single and multiple steps were collected from each turkey moving at different speeds, starting with slow walking and progressing up to aerial phase running if the bird was capable and willing. Data were A/D converted at 4 kHz (USB-6259 DAQ, National Instruments, Austin, TX, USA) and recorded onto a PC running Igor Pro 6.0 (Wavemetrics, Lake Oswego, OR, USA).

Movements in the lateral and anterior views were captured using two high-speed digital video cameras (Photron Fastcam 1024 PCI or Photron Fastcam SA4, San Diego, CA, USA). Video was collected at 250 Hz and analyzed using DLTdv5 (Hedrick, 2008) in Matlab (MathWorks, Inc., Natick, MA, USA). Velocity was obtained by tracking a marker placed 10 cm above the greater trochanter of each turkey.

Forces were analyzed in Igor Pro 6.0. The minimum and maximum forces in each direction were recorded from each trial and the difference between them was used as the force range produced by an individual turkey. Forces in the vertical, fore–aft and medio-lateral directions were divided by the body mass of each turkey to normalize the data.

Three female turkeys from each strain were used to determine the position of the center of mass (CoM). Biplanar fluoroscopy was used to capture the position of the body while the turkeys stood on a force plate (Kistler 9281B, Kistler Instruments AG, Winterthur, Switzerland). Force plate signals were A/D converted at 4 kHz (NI-6259 DAQ) and recorded with a PC running Igor Pro 6.0 (Wavemetrics). The location of the center of pressure in force plate coordinate space was calculated using the relative force components and moments acting on the force plate.

The origin of the coordinate system of the force plate was located in x-ray coordinate space by hanging a plumb line studded with metal beads directly over the force plate origin and recording x-ray images of the plumb line. The center of pressure on the force plate was used to determine the location of the CoM in the standing turkey in relation to the force plate origin. Each turkey was CT scanned to create 3D bone models of the pelvis and femora. The videos were undistorted and calibrated, and the markers on the plumb line were digitized using XMALab (Knörlein et al., 2016). Then, Scientific Rotoscoping (Gatesy et al., 2010) in Autodesk Maya (2013, Autodesk Inc., San Rafael, CA, USA) was used to determine the position and orientation of the pelvis in the standing birds. Center of pressure coordinates from the force plate were input to pinpoint the CoM position in 3D coordinate space. The distance from the CoM to the center point between the acetabula on the pelvis was measured in the cranial/caudal plane. This distance was normalized by pelvis length.

The SMATR package in R was used to perform standard major axis (SMA) regressions for the GRFs normalized to body mass and stride frequency versus velocity at which the turkey was moving (http://www.bio.mq.edu.au/ecology/SMATR; Warton et al., 2006). Turkeys were separated by strain and sex for this analysis. SMATR uses a likelihood ratio test comparing it with a chi-squared distribution to test for common slopes and shifts in elevation using the Wald statistic. If no common slope is found between the groups, then a post hoc pairwise comparison was performed. A nested ANOVA was used to compare the energy magnitudes, energy phase shift, duty factor, stride width and CoM positions by strain, with individual as a nested effect within strain, using JMP Pro 12.0 (64 bit, SAS Institute, Cary, NC, USA).

The wild female turkeys locomoted over the widest range of velocities, followed by domestic females, wild males and finally domestic males (Fig. 1, Table 1). Only the females of each strain reached a Froude number over 0.5; however, the wild males did achieve a maximum relative velocity over 0.6, indicating they may be within the gait transition zone from walking to running (Alexander, 1977). Stance duration, or foot contact time, decreased across velocity very quickly at first and then much more slowly (Fig. 1). All groups of turkeys fell on the same general trend line for stance duration. Domestic turkeys tended to have a lower stride period for any given speed and a higher stride frequency (F=101.391, P=0.001, n=11 domestic turkeys, 109 strides, and n=12 wild, 89 strides). Duty factor at a given speed was higher in domestic turkeys overall (F=6.1449, P=0.018), but did not decrease as expected with speed in either strain, indicating that both wild and domestic turkeys may transition from walking to grounded running. Domestic turkeys had a significantly greater stride width (41±3% of hip height) during walking than the wild turkeys (17±2%) (F=44.1165, P<0.001).

The large domestic turkeys had relatively high vertical GRFs because of their large body mass. For comparison, all forces were normalized to body mass (Fig. 2A,B). The relationships between normalized GRFs and velocity were compared among the four turkey groups. The large male turkeys maintained very low normalized peak vertical forces across their low range of absolute velocity, keeping the vertical forces very close to 1.0 body mass (M_b) (Fig. 3A). The resultant vector magnitudes showed the same pattern as the normalized vertical force but were shifted to marginally greater magnitudes; for example, the wild females' resultant vector magnitudes were on average 0.041±0.045 M_b greater than the normalized vertical forces. It should be noted that for the few trials where the normalized vertical force was slightly less than 1.0 M_b, the resultant vector magnitudes confirmed that the turkeys were supporting their body mass during the peak of the GRF. The low vertical forces at slower speeds are associated with a long double support period in both strains of turkeys, as seen by the high duty factors (Fig. 1D). The slopes of the normalized peak vertical force versus absolute speed among the four groups were not significantly different from each other; however, there were shifts in elevation and shifts along common slopes (i.e. same slopes over a different range in velocities) (Table 2).

Fore–aft GRFs increased significantly with absolute speed (Fig. 3B, Table 2). The slopes for the male and female domestic turkeys were significantly higher than those for both the male and female wild turkeys. The wild female turkeys had the lowest slope but as they covered the largest range of speeds, they still reached some of the highest fore–aft forces.

Medio-lateral GRFs also increased with absolute speed for all turkey groups (Fig. 3C). The domestic male turkey slope was the highest and was significantly different from that of all the other groups (Table 2). The difference in large male domestic turkey medio-lateral force magnitude was easily distinguishable from that of the small female wild turkeys at any given speed (Fig. 2B). This also translated into large visible lateral oscillations of the body during walking in the male domestic turkey, versus no distinguishable medio-lateral movement in the female wild turkeys (Fig. 2C,D). The turkey groups with intermediate body masses, female domestic and male wild turkeys, had intermediate medio-lateral force slopes and the female wild turkeys had the lowest slope. Overall, the medio-lateral forces increased at a higher rate with absolute velocity as the turkeys increased in body mass.

Gravitational PE of the CoM and kinetic energy in the fore–aft (KE_FA) and medio-lateral (KE_ML) direction were calculated across two or three steps on the force plate and normalized to M­_b (Fig. 4). The KE fluctuation magnitudes were significantly different between strains in both the fore–aft and medio-lateral direction (Fig. 5A). The domestic turkeys had greater fluctuations in KE_ML, 3.5 times that of the wild birds, easily discernible in the raw data (Fig. 4), while the wild turkeys had KE_FA magnitudes 1.7 times that of the domestic turkeys. The KE_ML of the domestic turkeys remained high compared with that of the wild turkeys across all velocities, even at speeds approaching potential gait transitions. PE fluctuations were not significantly different between the two strains. In the domestic turkey, fluctuations in KE_ML, KE_FA and PE were similar in magnitude (Fig. 5A).

Exchange between KE and PE requires that they fluctuate out of phase. Both KE_ML and KE_FA phase shifts were not significantly different between strains (Table 3), with averages falling midway between in phase and out of phase for both strains (Fig. 5B).

The CoM in the domestic turkeys was positioned 26% farther anterior from the acetabulum than in the wild turkeys, when expressed as a fraction of pelvis length (Fig. 6). The absolute distance of the CoM from the acetabulum was 10.9±1.6 cm in the domestic turkeys and 7.0±1.0 cm in the wild turkeys.

The domestic turkeys in this study move differently from the wild turkeys in a few immediately observable ways. First, they locomote across a lower range of speeds, keeping their vertical GRFs relatively low (Figs 1A and 3A). Second, male and female domestic turkeys have large lateral oscillations during walking, evident in their large fluctuations in medio-lateral GRF and energies (Figs 2, 3C and 5). Finally, domestic birds also walk with a higher stride frequency, meaning that they are taking shorter, faster steps to maintain the same speed as a wild turkey, while also taking wider steps. Hence, domestic turkey locomotion has its own set of kinematic and kinetic gait characteristics and is not simply a slowed-down version of wild turkey locomotion.

Walking with large lateral oscillations is often identified in the literature as waddling, a behavior that has been noted in penguins, geese and ducks (Abourachid, 2001; Kurz et al., 2008; Pinshow et al., 1977). The larger lateral oscillations we detected in the domestic turkey (Fig. 2C) confirm previous results (Abourachid, 1991), and are consistent with a waddling gait. Our GRF measurements indicate that this motion corresponds to higher medio-lateral forces in the domestic turkey as well (Fig. 3C).

These higher medio-lateral forces are consistent with results from other species with increased body mass, such as broiler chickens and obese humans, who also employ lateral oscillations and can exhibit high medio-lateral GRFs (Browning and Kram, 2007; Caplen et al., 2012; Paxton et al., 2013). Increased body mass is just one of the morphological changes that correlate with waddling; others include wide stance width and altered CoM position. Domestic turkeys have increased body mass, only slightly longer limbs, wider stride width during walking (current study) and muscle mass distribution differences compared with wild turkeys (Stover et al., 2018), as well as a more anterior CoM position (Fig. 6). Any or all of these structural modifications could contribute to the waddle.

It remains to be determined how waddling affects the metabolic and mechanical cost of locomotion in domestic turkeys. Waddling was once thought to be energetically wasteful (Pinshow et al., 1977), but there is evidence that waddling can save mechanical energy during walking, as is the case in penguins (Griffin and Kram, 2000). In penguins, the lateral KE from waddling allows them to recover more mechanical energy from each stride by increasing the total KE that can be converted to PE. In addition, penguins' lateral movements also make the total KE more out of phase with PE. Likewise, the domestic turkey has much higher medio-lateral energy fluctuations than the wild turkey, somewhat out of phase with PE, making mechanical energy savings possible for them as well.

Waddling is liable to have health repercussions for domestic turkeys. Current gait scoring methods to evaluate poultry locomotor problems use the descriptor ‘wobble’ to depict medio-lateral movement, deeming it an abnormality (Garner et al., 2002). Our study demonstrates that increased medio-lateral forces are associated with the wobble, which may contribute to loads that the joints experience during walking (Figs 2B and 3C). In humans, there are many musculoskeletal disorders associated with obesity due to increased external knee adduction moments and increased joint loading (Wearing et al., 2006). It is possible that increased medio-lateral GRFs and lateral oscillations could put more stress on the joints of the domestic turkey, possibly contributing to some of the knee and hip issues often associated with domestication, such as angular bone deformity, straddle legs and rotated tibia, as well as exacerbating tibial dyschondroplasia and osteochondrosis (Julian, 1998; Riddell, 1980; Siller, 1970). Gait scoring is a subjective visual scoring method, which, while fairly repeatable between observers (Garner et al., 2002; Kestin et al., 1992), as a selection method in turkeys (Neeteson et al., 2016) has not eliminated the waddling movement.

Many of the kinematic and kinetic differences between wild and domestic birds are characteristic of the ‘shuffling gait’ of some human populations. Short quick steps, higher step frequency, shuffling feet and a stooped stance without arm swing illustrate a shuffling, or festinant, gait (Knutsson, 1972; Murray, 1967). The most common human example of a shuffling gait is found in Parkinson's disease patients. During walking, the trunk is flexed forward, stooping, and the movement of the feet is described as shuffling, with long stance durations and decreased walking speed (Morris et al., 2001; Murray et al., 1978). This is similar to the slow-walking domestic turkeys with relatively long stance durations and high stride frequency (Fig. 1). Additionally, Smith–Magenis syndrome cases have been reported to be associated with a festinant gait (Elsea and Girirajan, 2008). Turkeys also share certain gait features with other human disorders. Prader–Willi syndrome is a chromosomal disorder with clinical features including muscular hypotonia and severe obesity. Prader–Willi gait is characterized by slow short stride lengths in the anterior direction that limit the velocity of progression, long stance phases seemingly to avoid overloading a single limb, as well as medio-lateral hip movements (Cimolin et al., 2010; Vismara et al., 2007). Medio-lateral movements are also seen in patients with ataxia, obese juveniles and those displaying a Trendelenberg gait due to muscle weakness (McGraw et al., 2000; Mitoma et al., 2000; Trendelenberg, 1998). The gaits characteristics of the human disorders described above have largely been attributed to an effort to maintain stability and balance.

We hypothesize that domestic turkeys, like humans, may assume a shuffling gait in order to increase their stability during locomotion. The shorter, faster steps necessary for the domestic turkeys to move at any given velocity may help explain the speed limit we observed (Fig. 1). Biomechanical and/or neurological limits could be responsible for the required increase in stability during walking. We have shown that the massive increase in pectoralis muscle mass has shifted the CoM position more anterior in the domestic turkey relative to that of the wild turkey (Fig. 6). This change in CoM position is reminiscent of the stooped stance associated with Parkinson's disease, which requires bringing the CoM closer to the stance foot to increase balance and dynamic stability. Domestic turkeys with increased body mass are also similar to humans carrying a load, approximating a wild turkey with a heavy backpack, although front-pack may be a more appropriate descriptor of the heavy pectoralis muscle. Significant increases in medio-lateral forces have been reported in humans carrying loads, likely caused by a continual shift of the CoM further from the neutral position, contributing to decreased stability (Birrell et al., 2007). Ducks also shift their CoM via trunk translations during terrestrial locomotion, ostensibly to increase stability (Provini et al., 2012). Finally, the extra girth present in the domestic turkey due to increased body mass and pectoralis mass could hinder the path of the turkey's legs during walking.

Alternatively, changes in gait may reflect challenges to stability that are cognitive or neurological in origin. Domestication has decreased animals' brain size (Clutton-Brock, 1999), thought to result from pedomorphosis in domestic animals, with certain sensory centers undergoing significant reductions (Ebinger, 1995). In addition, domestic turkeys continue to increase in both body mass and girth, even after bone growth slows, which could affect their sensory perception for resolving their own position and distances, therefore affecting motor control for balance (Carrier, 1996). It is also possible that the selection for increased body mass has resulted in decreased sensory or motor resolution; for example, the number of axons innervating a given amount of muscle tissue may be decreased, again compromising control (More et al., 2010). Any or all of these neural deficits could contribute to the domestic turkeys' issues with stability and balance.

Domestic turkeys may move slower than wild turkeys to maintain stability, but there are other potential explanations for the observed slow speeds and altered kinematics. The hindlimb muscles likely do not produce as much force per kilogram body mass, as their cross-sectional areas would only be expected to scale with body mass^0.67, not keeping up with the extra body mass support demands. In addition, domestic turkeys have been selected for desirable traits such as increased muscle mass and tenderness (Fletcher, 2002), probably influencing the muscle composition, which in turn could affect muscle function. Cardiovascular and respiratory changes associated with domestication may also impose limits on locomotor function. The heart and lung mass of chickens scale with negative allometry across ontogeny (Tickle et al., 2014), which continues to be a major animal welfare concern for poultry as it is presumed that they outgrow their cardiovascular capacity (Julian, 1993; Wideman, 2007). Finally, motivation should not be overlooked when considering the top speed of the turkeys. It is difficult to know that an animal is reaching a true performance limit; however, we feel that the combination of treats (dried meal worms and white bread) and exuberant chasing from behind gave the domestic turkeys incentive to move at close to their maximum speed.

The domestic turkey is a prime example of how an organism's morphology is closely linked to its locomotor mechanics in a delicate balance to maintain stability. Domestic turkeys are limited to a lower speed range than wild turkeys; however, when comparing their movements at the same speeds, domestic turkeys' gait is distinctive. Artificial selection for increased body mass in domestic turkeys has given rise to locomotor changes that are apparent in the GRF signature and fundamental mechanics, such as the motion of the center of mass. Domestic turkeys seem to have adopted a shuffling gait in order to maintain stability, despite or perhaps due to their suite of morphological alterations.

Selective practices have attempted to breed turkeys that minimally waddle, but this movement may be necessary to sustain a stable gait. Others have suggested improvements to gait scoring by focusing on characteristics of a balanced gait and even applying camera monitoring with image analysis of lateral oscillations, which we also support to improve gait heritability (Aydin, 2017; Duggan et al., 2016, 2017). The domestic turkeys displayed a consistent pattern of kinematic and kinetic gait parameter relationships with speed. Deviations from these established relationships in turkeys may be a more repeatable assessment of potential gait abnormalities than gait score, as originally suggested in ducks (Duggan et al., 2017). Future studies should address the skeletal stresses experienced during walking across ontogeny to determine the effect that a shuffling gait has on the domestic turkeys' joints and development.

The authors would like to thank Erika Tavares, Allison Rubenstein, Elizabeth McGinn, Jackie Alois and Elizabeth Rao, as well as Drs Chris Arellano, Ariel Camp, Terry Dial and Nicolai Konow for their assistance during turkey walking data collection and/or video analysis.