Higher heritabilities for gait components than for overall gait scores may improve mobility in ducks
© The Author(s) 2017
Received: 9 September 2016
Accepted: 19 April 2017
Published: 2 May 2017
Genetic progress in selection for greater body mass and meat yield in poultry has been associated with an increase in gait problems which are detrimental to productivity and welfare. The incidence of suboptimal gait in breeding flocks is controlled through the use of a visual gait score, which is a subjective assessment of walking ability of each bird. The subjective nature of the visual gait score has led to concerns over its effectiveness in reducing the incidence of suboptimal gait in poultry through breeding. The aims of this study were to assess the reliability of the current visual gait scoring system in ducks and to develop a more objective method to select for better gait.
Experienced gait scorers assessed short video clips of walking ducks to estimate the reliability of the current visual gait scoring system. Kendall’s coefficients of concordance between and within observers were estimated at 0.49 and 0.75, respectively. In order to develop a more objective scoring system, gait components were visually scored on more than 4000 pedigreed Pekin ducks and genetic parameters were estimated for these components. Gait components, which are a more objective measure, had heritabilities that were as good as, or better than, those of the overall visual gait score.
Measurement of gait components is simpler and therefore more objective than the standard visual gait score. The recording of gait components can potentially be automated, which may increase accuracy further and may improve heritability estimates. Genetic correlations were generally low, which suggests that it is possible to use gait components to select for an overall improvement in both economic traits and gait as part of a balanced breeding programme.
Increases in growth rate and breast muscle mass which have been achieved through selective breeding of poultry have been associated with welfare problems, notably an increased incidence of poor gait (which includes ‘leg weakness’) [1–5]. Birds with leg weakness may suffer pain and have difficulty reaching food and water [1, 6–8], which lead to economic losses for the producer and possible starvation for the animals. Gait problems were first reported in turkeys and broiler chickens [9, 10], although early studies focussed mainly on the emergence of skeletal leg defects rather than gait itself [10, 11]. Poor gait has since been observed in other heavy meat-producing birds [3, 12–15]. Although in Pekin ducks poor gait has not been reported as extensively, there is concern that gait problems may appear in the future if selection for production traits continues along its current trajectory, mirroring their emergence in other poultry species. It is important to consider that while gait problems may be associated with pain, sub-optimal gait may also be simply a functional consequence of an altered morphology in lines which have been heavily selected for increased muscle mass [16, 17].
Traditionally, in chickens and ducks, poor gait is assessed and selected against by using a visual gait score [9, 18], which is an ordinal score given to each bird based on a visual assessment of how that individual walks. Although efforts have been made to refine the visual gait score , it remains a subjective measure of walking ability and thus is prone to error. Previous studies have found relatively moderate kappa coefficients (a measure of agreement between observers) between 0.6 and 0.8 in ducks and chickens [14, 20]. This may suffice for flock-level welfare assessments but is below the accuracy required for selection. An EU report on the welfare of broiler chickens acknowledges the subjective nature of the gait scoring system and highlights the need to develop a more objective system of assessing gait .
Gait is a complex trait that requires the integration of sensory input, balance, conformation and fine motor control, and heritability estimates for poultry gait tend to be low [22–24]. Similarly low heritability estimates have been published for visual gait scores in other species . In addition, as the visual assessment of gait is a subjective measure , heritability estimates may be low, which limits potential genetic progress when selecting for such a trait. Attempts have been made to circumvent this problem of low heritability estimates by focusing selection on objectively measured traits such as tibial dyschondroplasia or bone deformity , although it remains unclear how these phenotypes affect the overall walking ability of birds. However, some gait components, such as step width, will certainly affect the overall walking ability of an animal and have not as yet been genetically evaluated.
The aim of this study was to estimate the reliability, heritability and genetic parameters of the visual gait score which is currently used in Pekin ducks and to compare this to heritability estimates for particular components of gait. It was hypothesised that these components of gait may be more heritable than the overall gait score. This was previously found to be the case in dairy cattle . Components were chosen for ease of measurement as well as for their hypothesised influence on overall gait. This study focusses on two gait components: step width, which influences balance during the stride, and body roll, which is a proxy for medio-lateral centre of mass movement during walking. There may be other components of gait which are more central to the overall movement of the bird but we chose those due to their ease of measurement. The components were also chosen on the basis of our previous findings that poultry lines selected for breast muscle mass ambulate with a wider step width and at a slower velocity (which is likely to increase body roll for a given step width) . The purpose of this study was to ascertain the suitability of selecting for gait components, rather than to identify which components, in particular, should become the focus of future selection programmes.
Assessment of gait score
In order to assess the reliability of the standard visual gait score in ducks, seven-week-old Pekin ducks were scored for gait by four industry gait scorers. Scorers were shown three video sequences of 36 birds walking over a runway. The video camera (Microsoft LifeCam Studio, recording at 30 frames per second) was placed behind each bird at a height of 15 cm. The video sequence contained 144 walks–four walks (including one duplicate) for each bird. Each walk lasted approximately 3 s in order to replicate the high throughput of birds during assessments on breeding farms. Scorers were asked to rate each walk with a score of 1 (very poor gait) to 5 (perfect gait). None of the scorers were informed that the sequences contained duplicate recordings or multiple walks from the same birds. Agreement between and within scorers was assessed by using Kendall’s coefficient of concordance and the Minitab software (Minitab version 17, Minitab Inc.).
Measurement of gait components
Over the course of eight weeks, on one day per week, over 5000 Pekin ducks were visually scored for gait. On average, 650 birds were visually scored during each week. Two breeding lines (A and B) of Pekin duck were used, alternating each week. In total, data was collected from four hatches of each line (a different hatch was measured each week). These breeding lines are grandparent stock of the standard Cherry Valley commercial hybrid duck. Line A forms part of the maternal grandparent stock and Line B forms part of the paternal grandparent stock. All birds were hatched in the same hatchery and raised according to the Cherry Valley published guidelines. Water and feed (standard industry rations) were provided ad libitum. The photoperiod was 23 h light on day 1, reducing by 1 h per day until day 6 when the photoperiod was 18 h of light per day and this was maintained to the end of the trial. Phenotypic data collection for various traits was carried out on a breeding company farm by experienced members of staff. All phenotypic measurements took place at a single measurement station on the same breeding farm. After corralling birds at six weeks of age into a small area adjacent to the measurement station, each bird was weighed and its (ultrasonic) breast muscle depth was recorded. The birds were subsequently placed on a custom-built walkway (1.2-m wide and 4.8-m long) and allowed to walk away at their own pace, during which time each bird’s overall gait and gait components were scored (during normal selection procedures, birds are gait scored while walking over loose straw bedding). The walkway consisted of a wooden base (6-mm thick plywood) which was covered by a sheet of 7-mm green artificial turf in order to provide grip and to create a contrast so as to make the birds’ feet easier to see. Perspex sheeting (30-cm high) was fixed to the sides of the walkway to ensure that the birds walked straight to the end of the walkway. Gait was assessed using a visual gait score (which forms part of the company’s routine phenotypic measurement). Gait scores for Lines A and B were recorded by two different members of staff (each line was scored by only one individual), both of whom were experienced at scoring gait. The visual gait score used by the breeding company spans a 1 to 5 scale, with 1 representing a bird which is markedly lame and 5 representing perfect gait. The score for a bird was downgraded if, when walking, that individual displayed bowed or splayed legs, medially or laterally rotated feet, or if the angle of the back to the floor was outside the 35 to 65 degree range. Birds which were lame, immobile or walked on their hocks were given a score of 1. Most ducks were assigned scores between 2 and 4 (in this trial, 1% were given a gait score of 1; 29% a score of 2; 61% a score of 3 and 8% a score of 4).
In addition to the overall visual gait score, two gait components (step width and body roll) were recorded simultaneously by one of the authors (BMD). The same author scored components of gait for both lines. Step width was scored visually as the estimated distance (perpendicular to the direction of travel) between the most posterior parts of the feet on a 1 to 3 scale, a score of 1 denoting the feet as being very close together (or overlapping) and a score of 3 denoting that the feet were widely spaced during walking. Body roll (also on a 1 to 3 scale) was recorded as the degree of rolling of the shoulders during walking. This was considered an approximation of medio-lateral centre of mass movement since the position of centre of mass was impossible to ascertain visually. This trait was deemed important because the degree of medio-lateral movement of the centre of mass can affect the birds’ balance. Birds which display greater variation in centre of mass position during walking may be at greater risk of stumbles or falls. A score of 1 represented very little rolling of the shoulders, whereas a score of 3 was given to birds which rolled their shoulders to a large degree while walking. The repeatability for both gait component scores was assessed in a small trial at the beginning of this study and deemed to be satisfactory, although no larger scale repeatability test using video was carried out as was the case for the standard gait score. Birds which moved too quickly or too slowly to reliably score gait components were treated as missing values. Hence, gait components were recorded on 4252 of the 5251 birds that were phenotypically measured (Table 2). Among the birds measured, 5% were given a step width score of 1, 79% were given a score of 2 and 16% were given a score of 3. For body roll, 9% were scored 1, 74% were given a score of 2 and 17% were given a score of 3. In addition to standard phenotypic measures of breast depth and body mass, feed conversion ratios (FCR) for each bird were calculated by automated measurement of each bird’s individual feed intake and body mass. Data collected at the phenotypic measurement station was collated with information of the FCR of each bird. The pedigree of all birds was known, stretching back 15 generations.
This study was approved by the Veterinary Ethical Review Committee at the University of Edinburgh.
Pedigree structure for Lines A and B
Individuals in pedigree
Generations in pedigree
Sires of sires
Dams of sires
Sires of dams
Dams of sires
Means (and standard deviations) for all traits measured in Lines A and B
Number of phenotyped males
Number of phenotyped females
Finish weight (g)
Breast depth (mm)
2.80 (0.66) 
2.13 (0.46) 
2.08 (0.53) 
3760 (290) 
152 (15.3) 
1.90 (0.17) 
2.70 (0.56) 
2.10 (0.43) 
2.06 (0.47) 
3362 (297) 
146 (16.7) 
2.02 (0.23) 
Kendall’s coefficient of concordance, calculated between four experienced observers who scored gait in short video clips, was equal to 0.49 (df = 132, p < 0.001). The Kendall’s coefficient of concordance within observers (scoring duplicate videos) was equal to 0.75 (df = 135, p < 0.001). No clear observer drift effect was detected, i.e. scorers deviated to a similar degree when scoring the first 60 walks compared to the last 60.
Heritability estimates and correlations for gait and other major economic traits in Line A
Heritability estimates and correlations for gait and other major economic traits in Line B
Phenotypic correlations between traits varied between lines. Generally, phenotypic correlations between gait traits and economic traits were very low and correlations between economic traits were also low, with the exception of finish weight and breast depth (Tables 3 and 4). Since in this study, relatively small sample sizes were used, estimates of genetic correlations between traits were associated with relatively high standard errors. Most genetic correlations between gait traits and economic traits were not significant (p > 0.05), with the exception of Line B, where significant genetic correlations were observed between step width and breast depth (t = 2.16, p < 0.05), and between gait score and FCR (t = 3.26, p < 0.01). The standard gait score had moderate to good genetic correlations with body roll (−0.51 to −0.69). Genetic correlations between gait score and step width were not significant. The significant genetic correlations between economic traits were moderate (0.23 to 0.61).
Gait problems are a major animal welfare issue facing modern poultry in intensive production systems. Our results suggest that a more targeted approach to assessing gait by focussing on gait components has the potential to improve progress in selecting for better gait in breeding birds.
The pilot study that involved a limited amount of data suggests that the current visual gait scoring system, while showing some level of agreement between scorers, may not be optimal for long-term use in breeding programmes, but can be improved. The Kendall’s coefficient of concordance suggests that low concordance exists between scorers. Indeed, when scoring video clips of the same walks (using the standard visual gait score described above), all four scorers agreed 28% of the time and three of the four scorers agreed 74% of the time. Individual scorers failed to allocate the same score to two duplicate walks 26% of the time. Some of these inconsistencies may be due to the short duration of each video recording. Short recordings were chosen in order to replicate conditions during assessments on farm; however for certain birds on farm, the scorer will observe a walk for longer than 3 s before allocating a score for that bird. The viewing angle of the camera, which was chosen to give a clearer view of the birds’ gait, is also different from the viewpoint used when scoring during selection on farm, which is from a standing position.
The suboptimal reliability of the visual gait score that was recorded by using these video clips suggests that an alternate and more rigorous method of gait assessment is required to make progress on selection for optimal gait as weight increases. Previous work on gait in cows suggested that assessing components of gait may yield higher heritability estimates . Certain gait components such as step width and the ratio of double to single support time are known to have changed to a similar extent in both ducks and chickens which have undergone selection for increased body weight and meat yield , and selection decisions based on these components may yield greater progress than the current subjective gait scoring system.
This study estimated genetic parameters for components of gait and compared these to those of the overall visual gait score. The heritability of step width was higher than that of the original gait score in both lines and standard errors were approximately the same for both estimates. This is to be expected since the gait score is a subjective measurement based on a visual assessment of overall body movements, without any tangible reference points, whereas step width is a simpler score based on only one aspect of foot placement and therefore one would expect this score to be more objective. In addition, the recorder that measures step width can make use of reference points on the ground to compare successive birds. The heritability of body roll was similar to that of the gait score, probably because unlike step width, the assessment of body roll is a more subjective assessment. Heritability estimates for other economic traits (finish weight, breast depth and FCR) were in the range expected, with some differences observed between lines. For example, the mean estimate for the heritability of body weight in this study (0.34) is in a similar range than the heritabilities of 0.28 to 0.45 which were estimated in recent poultry studies [24, 28–30]. Heritability estimates presented in this study were calculated from only one phenotyped generation; thus, it is expected that these heritabilities would be estimated with more accuracy if more generations had been phenotyped, as is the case within commercial breeding programmes.
Phenotypic correlations between the gait score and production traits were generally low, which suggests that the gait score is indeed a measure of gait, rather than a proxy measure of body mass or breast depth. Phenotypic correlations between the gait score and components of gait were low to moderate, whereas those between each component of gait were generally moderate. Due to the relatively small sample size, genetic correlations were generally associated with relatively large standard errors that were of a similar magnitude to the genetic correlation estimates. The notable exceptions were the genetic correlations between gait score and body roll and between step width and body roll in both Lines A and B. In Line B, breast depth (considered a proxy for pectoral muscle mass) was negatively correlated with step width; continued selection for greater breast depth may result in a narrower step width. The effect of this narrower step width on balance will depend on the degree to which the body’s centre of mass moves laterally during gait. Genetic correlations between gait components and production traits were generally low and the data suggest that selection for improved gait will not be compromised by negative responses in economic traits.
These data demonstrate that the visual assessment of gait components during selection is both feasible and yields promising heritability estimates. While some caution must be exercised when interpreting these results (given the presence of categorical traits in the model), the use of gait components holds promise for future progress in selection for improved gait in ducks; since they are simpler traits, the assessment of gait components can be automated, for example by using pressure sensing technology as in Duggan et al. . Automation of measurement has the potential to bring about greater objectivity and to increase breeding success. However, it is important to note that although the gait components that are the focus of this paper can be measured satisfactorily and have reasonable heritability estimates, it is not yet known which components should be selected to improve gait. For example, it could be argued that a wide step width would be beneficial to a bird with large lateral displacement of the centre of mass, whereas a narrow step width would be beneficial to a bird with little lateral centre of mass movement. However, it is also difficult to differentiate cause and effect associations between step width and lateral body movement. A more thorough understanding of how gait components are integrated to perform overall locomotion is therefore necessary before recommendations can be made on which particular gait components should be used in breeding programmes. It is likely that most of the improvement will be achieved using a selection index which combines weighted measurement of various gait components. Indeed, current overall gait scoring methods use a combination of components, which are subconsciously weighted in different ways depending on the observers’ opinions of what optimal gait entails. By focussing only on the measurement of gait components, this differential weighting among observers can be avoided.
Scoring overall gait visually is a subjective measure which generally generates low (but useable) heritabilities. We demonstrate that focussing on gait components, rather than overall gait, may result in heritability estimates that are equal to or higher than those of the conventional visual gait score in ducks. The benefit of using components of gait is that their measurement can be automated to generate greater accuracy and easily combined to create an index score of overall gait. Genetic correlations, while difficult to ascertain, are generally low; therefore it is possible to use gait components to select for an overall improvement in both economic traits and gait as part of a balanced breeding programme.
BMD and AMR collected the data. PMH obtained funding for the project. BMD, AMR, DNC and PMH contributed to the design of the study, analysis of data and drafting of the manuscript. All authors read and approved the final manuscript.
We would like to acknowledge Cherry Valley Farms for provision of birds and financial support. Thanks to Ozzie Matika and Tanya Englishby for very useful and helpful comments and feedback. Many thanks to the farm staff at Cherry Valley for assistance with data collection. This work was funded through a Biotechnology and Biological Sciences Research Council Industrial CASE studentship (Grant BB/K501621/1) in collaboration with Cherry Valley Farms. The Roslin Institute is supported by an Institute Core Strategic Grant from the BBSRC.
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Bradshaw RH, Kirkden RD, Broom DM. A review of the aetiology and pathology of leg weakness in broilers in relation to welfare. Avian Poult Biol Rev. 2002;13:45–103.View ArticleGoogle Scholar
- Knowles TG, Kestin SC, Haslam SM, Brown SN, Green LE, Butterworth A, et al. Leg disorders in broiler chickens: prevalence, risk factors and prevention. PLoS One. 2008;3:e1545.View ArticlePubMedPubMed CentralGoogle Scholar
- Jones TA, Dawkins MS. Environment and management factors affecting Pekin duck production and welfare on commercial farms in the UK. Br Poult Sci. 2010;51:12–21.View ArticlePubMedGoogle Scholar
- Paxton H, Daley MA, Corr SA, Hutchinson JR. The gait dynamics of the modern broiler chicken: a cautionary tale of selective breeding. J Exp Biol. 2013;216:3237–48.View ArticlePubMedGoogle Scholar
- Farm Animal Welfare Council. Report on the welfare of broiler chickens. 1992.
- McGeown D, Danbury TC, Waterman-Pearson AE, Kestin SC. Effect of carprofen on lameness in broiler chickens. Vet Rec. 1999;144:668–71.View ArticlePubMedGoogle Scholar
- Caplen G, Colborne GR, Hothersall B, Nicol CJ, Waterman-Pearson AE, Weeks CA, et al. Lame broiler chickens respond to non-steroidal anti-inflammatory drugs with objective changes in gait function: a controlled clinical trial. Vet J. 2013;196:477–82.View ArticlePubMedGoogle Scholar
- Danbury TC, Weeks CA, Chambers JP, Waterman-Pearson AE, Kestin SC. Self-selection of the analgesic drug carprofen by lame broiler chickens. Vet Rec. 2000;146:307–11.View ArticlePubMedGoogle Scholar
- Kestin S, Knowles TG, Tinch AE, Gregory NG. Prevalence of leg weakness in broiler chickens and its relationship with genotype. Vet Rec. 1992;131:190–4.View ArticlePubMedGoogle Scholar
- Nestor KE. Genetics of growth and reproduction in the turkey. 9. Long-term selection for increased 16-week body weight. Poult Sci. 1984;63:2114–22.View ArticlePubMedGoogle Scholar
- Mercer JT, Hill WG. Estimation of genetic parameters for skeletal defects in broiler chickens. Heredity (Edinb). 1984;53:193–203.View ArticleGoogle Scholar
- Martrenchar A. Animal welfare and intensive production of turkey broilers. Worlds Poult Sci J. 1999;55:143–52.View ArticleGoogle Scholar
- Abourachid A. Comparative gait analysis in two strains of turkey, Meleagris gallopavo. Br Poult Sci. 1991;32:271–7.View ArticlePubMedGoogle Scholar
- Makagon MM, Woolley R, Karcher DM. Assessing the waddle: an evaluation of a 3-point gait score system for ducks. Poult Sci. 2015;94:1729–34.View ArticlePubMedGoogle Scholar
- Da Costa MJ, Grimes JL, Oviedo-Rondón EO, Barasch I, Evans C, Dalmagro M, et al. Footpad dermatitis severity on turkey flocks and correlations with locomotion, litter conditions, and body weight at market age. J Appl Poult Res. 2014;23:268–79.View ArticleGoogle Scholar
- Duggan BM, Hocking PM, Schwarz T, Clements DN. Differences in hindlimb morphology of ducks and chickens: effects of domestication and selection. Genet Sel Evol. 2015;47:88.View ArticlePubMedPubMed CentralGoogle Scholar
- Corr SA, Gentle MJ, McCorquodale CC, Bennet D. The effect of morphology on walking ability in the modern broiler: a gait analysis study. Anim Welf. 2003;12:159–71.Google Scholar
- Kestin SC, Gordon S, Su G, Sorensen P. Relationships in broiler chickens between lameness, liveweight, growth rate and age. Vet Rec. 2001;148:195–7.View ArticlePubMedGoogle Scholar
- Garner JP, Falcone C, Wakenell P, Martin M, Mench JA. Reliability and validity of a modified gait scoring system and its use in assessing tibial dyschondroplasia in broilers. Br Poult Sci. 2002;43:355–63.View ArticlePubMedGoogle Scholar
- Webster AB, Fairchild BD, Cummings TS, Stayer PA. Validation of a three-point gait-scoring system for field assessment of walking ability of commercial broilers. J Appl Poult Res. 2008;17:529–39.View ArticleGoogle Scholar
- Berg C, Bessei W, Faure JM, Jensen P, Porin F, San Gabriel Closas A, et al. The welfare of chickens kept for meat production (broilers). In Report of the scientific committee in Animal Health and Animal Welfare. Brussels European Commission; 2000.
- EFSA Panel on Animal Health and Welfare. Scientific opinion on the influence of genetic parameters on the welfare and the resistance to stress of commercial broilers. EFSA J. 2010;8:1666.View ArticleGoogle Scholar
- Whitehead CC, Fleming RH, Julian RJ, Sørensen P. Skeletal problems associated with selection for increased production. In: Muir WM, Aggrey SE, editors. Poultry genetics, breeding and biotechnology. Wallingford: CABI Publishing; 2003. p. 29–52.View ArticleGoogle Scholar
- Kapell DNRG, Hocking PM, Glover PK, Kremer VD, Avendaño S. Genetic basis of leg health and its relationship with body weight in purebred turkey lines. Poult Sci. 2017;. doi:10.3382/ps/pew479.PubMedGoogle Scholar
- Chapinal N, Sewalem A, Miglior F. Short communication: estimation of genetic parameters for gait in Canadian Holstein cows. J Dairy Sci. 2012;95:7372–6.View ArticlePubMedGoogle Scholar
- Kapell DNRG, Hill WG, Neeteson AM, McAdam J, Koerhuis AN, Avendaño S. Twenty-five years of selection for improved leg health in purebred broiler lines and underlying genetic parameters. Poult Sci. 2012;91:3032–43.View ArticlePubMedGoogle Scholar
- Duggan BM, Hocking PM, Clements DN. Gait in ducks (Anas platyrhynchos) and chickens (Gallus gallus): similarities in adaptation to high growth rate. Biol Open. 2016;5:1077–85.View ArticlePubMedPubMed CentralGoogle Scholar
- Rekaya R, Sapp RL, Wing T, Aggrey SE. Genetic evaluation for growth, body composition, feed efficiency, and leg soundness. Poult Sci. 2013;92:923–9.View ArticlePubMedGoogle Scholar
- Bailey RA, Watson KA, Bilgili SF, Avendano S. The genetic basis of pectoralis major myopathies in modern broiler chicken lines. Poult Sci. 2015;94:2870–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Mignon-Grasteau S, Beaumont C, Poivey JP, de Rochambeau H. Estimation of the genetic parameters of sexual dimorphism of body weight in ‘label’ chickens and Muscovy ducks. Genet Sel Evol. 1998;30:481–91.View ArticlePubMed CentralGoogle Scholar