Abstract

Improvement in growth has been widely reported as the cause of increased incidence of leg problems in broiler chickens. We report herein the genetic relationship between growth and leg problems in a random mating broiler control population. The traits studied were valgus (VL), varus (VR), and tibial dyschondroplasia (TD), which were expressed on a binary scale of 0 (normal) and 1 (abnormal); growth rates from 0 to 4 wk (BWG 0-4) and from 0 to 6 wk of age (BWG 0-6); and residual feed intake from 5 to 6 wk of age (RFI 5-6). A threshold-linear mixed model was employed for the joint analysis of the categorical and linear traits. Incidences of VL, VR, and TD were 26, 4, and 2%, respectively. Heritability of leg problems ranged from 0.11 to 0.13. Phenotypic correlations alluded to an unfavorable relationship between growth and leg problems; however, the genetic relationship between growth and leg problems was extremely weak, ranging from 0.01 to 0.08. There is, therefore, a basis for genetic improvement in leg problems. However, improved management practices would also be important to reduce incidence of leg problems in broiler chickens.

INTRODUCTION

During the past 6 decades, poultry production has experienced an extraordinary improvement in performance. This advance has resulted from a better knowledge and understanding of nutrition, physiology, housing environment, and successful broiler breeding programs. Unfortunately, the improvement of some of the production traits has been accompanied by an incidence of metabolic disorders, including skeletal problems (EC, 2000; Kalmar et al., 2013; Rekaya et al., 2013).

Skeletal disorders reduce the general mobility of broilers and in extreme cases, limit their ability to walk to the feeder and drinker. Leg weakness has been described as the primary cause of mortality in late stages of the growing period (Kestin et al., 1999; Knowles et al., 2008). Chickens with these problems are at a disadvantage, and many are attacked by other birds and eventually die (Julian, 1998, 2005). Furthermore, animal health and welfare in animal production systems are of great importance to the poultry industry (Tuyttens et al., 2008).

Multiple reports highlight the skeletal problems in poultry (Sullivan, 1994; Julian, 1998, 2005; Cook, 2000). Tibial dyschondroplasia (TD), and varus-valgus deformations (VVD) are important examples of these issues. TD describes an avascular lesion characterized by abnormal cartilage formation resulting from an accumulation of immature chondrocytes with an extended life span in the proximal metaphyses of the tibiotarsus and tarsometatarsus (Leach and Nesheim, 1965; Leach and Monsonego-Ornan, 2007). On the other hand, VVD refers to a lateral or medial angulation of the shaft of the distal tibiotarsal bone resulting in deviation of the lower part of the leg (Randall and Mills, 1981; Julian, 1984). More specifically, valgus and varus deformations are outward and inward deviations of the tibiometatarsus, respectively (Hunter et al., 2008).

Although high growth rates have been associated with higher incidence of leg deformities (Sorensen, 1992; Julian, 2005; Shim et al., 2012), the direct genetic relationship between these two factors is not conclusive (Zhang et al., 1995; Kuhlers and McDaniel, 1996). Some results indicate that the relationship does not exist (Cook et al., 1984) or that the incidence of leg deformities is not a direct result of selection for broiler traits (Hocking et al., 2009). However, the definition of a breeding strategy for practical purposes requires adequate knowledge of the genetic relationship among traits of economic importance (Le Bihan-Duval et al., 1996).

Several studies have demonstrated that leg problems are heritable (Sheridan et al., 1978; Mercer and Hill, 1984; Le Bihan-Duval et al., 1997; Kapell et al., 2012; Rekaya et al., 2013). However, genetic correlations between leg problems and growth have been contradictory in the literature. In the case of TD, these values have been positive (Sheridan et al., 1978; Kapell et al., 2012), close to zero (Kuhlers and McDaniel, 1996; Rekaya et al., 2013), or negative (Burton et al., 1981; Zhang et al., 1995). Similar results were reported for VVD (Mercer and Hill, 1984; Le Bihan-Duval et al., 1997; Kapell et al., 2012; Rekaya et al., 2013).

Some of the inconsistencies in the relationship between growth parameters and leg problems as described by Rekaya et al. (2013) are due in part to the categorical nature of the data and the methodological and computational challenges associated with joint analyses of these traits. Also, the nature of the population used in the study could affect the relationships; e.g., the study of Rekaya et al. (2013) was conducted using a commercial line that is under selection. The objective of this study was to estimate genetic variance-covariance parameters for leg problems, growth, and feed efficiency in a random mating broiler control population.

MATERIALS AND METHODS

Experimental Population and Husbandry

In the present study, 2,301 pedigreed broiler chickens, produced in 8 consecutive hatches from mating 24 sires and 72 dams, were used. The Arkansas random mating population, which is an unselected broiler control population, was used (Aggrey et al., 2010). Once hatched, chicks were sexed and placed in pens (0.071 m2/bird) with litter. From hatching to 18 d, the chickens received a mash starter diet containing 225 g/kg protein, 52.8 g/kg fat, 25.3 g/kg fiber, 12.90 MJ ME/kg, 9.5 g/kg calcium (Ca), and 7.2 g/kg total phosphorus (P) (4.5 g/kg available P). Following this, chickens were fed a pelleted grower diet containing 205 g/kg protein, 57.6 g/kg fat, 25.0 g/kg fiber, 13.20 MJ ME/kg, 9.0 g/kg Ca, and 6.7 g/kg total P (4.1 g/kg available P). At 28 d of age, birds were transferred to individual metabolic cages (20.32 by 60.96 by 30.48 cm). They were allowed to acclimate for one week prior to measuring feed efficiency from 5 to 6 wk. Water and feed were provided ad libitum for the duration of the study. Birds were kept on a 20L:4D light regimen. Bird handling and experimental protocols were in line with the University of Georgia Animal Use and Care Guidelines.

Data

Weekly BW was recorded for each bird from hatch until 6 wk. Feed intake was measured from wk 5 to 6, and legs were scored VVD at wk 4 and TD at wk 6. Body weight gain (BWG 0-4) from wk 0 to 4 and (BWG 0-6) from wk 0 to 6 was calculated. Residual feed intake from 5 to 6 wk of age (RFI 5-6) was calculated according to Aggrey et al. (2010). Methods described by Leterrier and Nys (1992) were followed to score VVD in each leg at wk 4. Depending on the angle size of tibia-metatarsus, 4 categories and scores of VVD were defined: normal (score = 0), mild (10 to 25° angle; score = 1), intermediate (25 to 45° angle; score = 2), and severe (>45° angle; score = 3). Methods described by Edwards and Veltmann (1983) were followed to score TD in the right tibia. A longitudinal cut across the tibia was made, and white cartilage plug abnormality was observed. Depending on the severity of the cartilage abnormality, 4 categories and scores of TD were defined: normal (score = 0), mild (score = 1), intermediate (score = 2), and severe (score = 3).

Statistical Analysis

After data editing, there were 2,257 birds recorded for 8 traits: valgus in the right leg (VLR), valgus in the left leg (VLL), varus in the right leg (VRR), varus in the left leg (VRL), TD, BWG 0-4, BWG 0-6, and RFI 5-6. However, VLR and VLL were merged into valgus (VL), and VRR and VRL were merged into varus (VR) for the analysis. To ameliorate the grossly unequal representation of leg incidences in different hatch groups, and also for computational feasibility, the leg data were expressed on a binary scale of 0 (normal) and 1 (abnormal) for VL, VR, and TD so that birds having no leg problems were scored as 0 and those birds having leg problems in either one or both legs were scored as 1. Descriptive statistics were obtained using PROC UNIVARIATE procedures of SAS version 9.1.3 (SAS Institute, 2006). A threshold-linear mixed model similar to that used by Rekaya et al. (2013) was implemented for the joint analysis of the categorical traits (leg problems) and linear traits (growth and feed efficiency). The threshold-linear mixed model used was:  

(1)
\begin{equation} {\rm y}_{{\rm ijnk}} = {\rm H}_{\rm i} + {\rm S}_{\rm j} + {\rm u}_{\rm n} + {\rm e}_{{\rm ijnk}} \end{equation}
Where:

  • yijnk = Observed BWG 0-4, BWG 0-6, RFI 5-6, or the liabilities for the leg problems (VL, VR, TD) for bird n

  • Hi (i = 1-8) = Fixed effect of the hatch class i

  • Sj = Fixed effect of sex j (j = 1-2) of bird n

  • un = Random additive effect of bird n

  • eijnk = Random residual term

Assuming normality conditionally on the model parameters, the joint distribution of the liability for binary scores and continuous traits is expressed in matrix notation as:  

(2)
\begin{equation} {\bf y}|{\boldsymbol \beta} ,\; {\bf u,\;R}_0 \sim{\rm N}({\bf X}{\boldsymbol\beta } +{\bf Zu,\;R}_{\rm 0} \otimes{\bf I}) \end{equation}
Where:

  • y = (l1 ', l2 ', l3 ', y1 ', y2 ', y3 ')' = Vector of liabilities (l) and continuous responses (yi)

  • β and u = Vectors of systematic and random effects, respectively

  • R0 = An 6 × 6 residual (co)variance matrix with the first 3 diagonal elements, corresponding to the categorical traits, fixed to 1

  • X and Z = Known incidence matrices.

Rekaya et al. (2013) stated that in the Bayesian implementation of a model like that observed in (1) via Markov chain Monte Carlo (MCMC) methods, the direct sampling of R0 is not feasible due to the fixation of some of its diagonal elements. To overcome this problem of sampling of the residual (co)variance matrix, the methods described by Rekaya et al. (2013) were used.

The multiplication of model (1) by a diagonal matrix D = D0I, where D0 is a 6 × 6 diagonal matrix and I is an identity matrix with appropriate dimensions, yields an equivalent model:  

(3)
\begin{equation} {\bf y^*} = {\bf X} {\boldsymbol \beta {\rm ^*}} + {\bf Zu{\rm ^*}} + {\bf e{\rm ^*}} \end{equation}
Where:  
\begin{equation*} {\boldsymbol \beta ^*} = {\bf ({\boldsymbol \beta _1^{ * '}} ,{\boldsymbol \beta _2^{ * '}} , {\bf \ldots ,\beta _{\boldsymbol t}^{ * '}} )',} \end{equation*}
 
\begin{equation*} {\bf u^*} ={\bf (u_1^{ * '} ,u_2^{ * '} , \ldots ,u_{\boldsymbol t}^{ * '} )',} \end{equation*}
with $${\boldsymbol \beta _i^*}$$ = βidii and $${\bf u_i^*}$$ = uidii, where βi and ui are vectors of fixed and random effects for the trait i in the identifiable model in (2), respectively, and dii is the diagonal element i of matrix D0. However, the model in (3) is not identifiable because D is not known. The residual (co)variance matrix of the nonrestricted model in (3) is given by:  
(4)
\begin{equation} {\bf {\rm var}(e{\rm *})} = {\bf DRD'} = \Sigma = \Sigma _0 \otimes {\bf I}, \end{equation}
This is a nonrestricted residual (co)variance matrix, where R is the original restricted residual (co)variance matrix in (2), and ∑0 is a 6 × 6 residual (co)variance matrix of the nonidentifiable model in (4). Thus, estimates of the restricted residual covariance matrix, R, could be easily obtained using model (4) and the nonrestricted matrix ∑0. The lack of restriction in ∑ facilitates enormously the Bayesian implementation via MCMC methods. However, to obtain the parameters of the identifiable model in (2) from the draws of the nonidentifiable parameters, D needs to be defined. The identifiable parameters, based on expressions in (3) and (4), can be retrieved as:  
(5)
\begin{equation} {\boldsymbol \beta} _i = \frac{1}{{d_{ii} }}{\boldsymbol \beta} _i^ * ; \end{equation}
 
(6)
\begin{eqnarray} {\bf u}_i = \frac{1}{{d_{ii} }}{\bf u}_i^ * ; \end{eqnarray}
 
(7)
\begin{eqnarray} {\bf R} = {\bf D}^{ - 1} \Sigma {\bf D'}^{ - 1} \;{\rm and}\;{\bf R}_0 = D_0^{ - 1} \Sigma _0 D_0^{ - 1} . \end{eqnarray}
Given that the diagonal elements of R0 corresponding to the binary responses are fixed to 1, the first 3 diagonal elements of the matrix D0 must be equal to the square root of their corresponding elements in ∑0, and the 3 diagonal elements of D0 corresponding to the continuous traits are set equal to one as indicated by Rekaya et al. (2013).

To complete the Bayesian formulation, the following priors were assumed to the unknowns in the model,  

\begin{equation*} {\rm p(}{\boldsymbol \beta ^{\rm *})}\sim{\rm U}[ - 10^6 ,10^6 ] \end{equation*}
 
\begin{equation*} {\rm p(}{\bf u}|{\bf A}{\rm ,}{\bf G}{\rm )}\sim{\rm N}(0,\;{\bf A} \otimes {\bf G}_0 ) \end{equation*}
and for the elements of matrix G0,  
\begin{equation*} {\rm p}(g_{ii} )\sim{\rm U}[0,10^5 ]\;{\rm for}\;i = 1,2, \ldots ,6, \end{equation*}
 
\begin{eqnarray*} &&\!\!\!\!{\rm and}\;{\rm p}(g_{ij} )\sim{\rm U}\left[ { - \sqrt {\sigma _{u1}^2 \sigma _{u2}^2 } ,\sqrt {\sigma _{u1}^2 \sigma _{u2}^2 } } \right]\;\\ &&\!\!\!\!{\rm for}\;{\rm i} \ne {\rm j} = 1,2, \ldots ,6, \end{eqnarray*}
where A is the additive genetic relationship between birds, G0 is the additive genetic (co)variance matrix, and gii is the genetic variance for the trait i. Similar priors are assumed for ∑0 but with 106 for the upper bound of the uniform distribution for the diagonal elements.

The resulting full conditional distributions needed for the implementation of Gibbs sampling for the systematic and random effects, liabilities, and genetic and residual (co)variance matrices were in closed form being normal, truncated normal, and scaled inverted Wishart, respectively. A unique chain of 200,000 samples was implemented where the first 50,000 samples were discarded as burn-in period based on visual inspection of the behavior of the chain. In each iteration of the sampling process, the transformations indicated in Equations (5) to (7) were applied to draws of the nonidentified model to obtain samples corresponding to the identifiable parameters in model (2). Computer software developed by Rekaya et al. (2013) was used for analysis.

RESULTS

Incidence (%) of leg problems and phenotypic correlations of these traits with growth rate are shown in Table 1. Descriptive statistics of the traits studied are summarized in Table 2. Incidence of VL was greater than VR (P < 0.01). Phenotypic correlations of leg problems with growth rate were unfavorable and ranged from 0.02 to 0.10, but only those of VL and VR were significant (P < 0.01). Heritability of leg problems and growth and feed efficiency traits are presented in Table 3. While heritability values corresponding to leg problems ranged from 0.11 to 0.13, those of growth and feed efficiency traits ranged from 0.12 to 0.27. Genetic and residual correlations among the traits are presented in Table 4. Genetic correlation between VVD was low and negative (–0.02) but between these traits and TD was positive (0.01 to 0.05) but weak. Genetic correlations of VL and VR with growth rate were unfavorable although small and ranged from 0.02 to 0.08. In contrast, the genetic association of TD with growth rate tended to be still close to zero (–0.01). Genetic correlations of leg problems with RFI 5-6 ranged from 0.01 to 0.02.

Table 1.

Incidence (%) of leg problems1 and phenotypic correlations between these traits and growth rate1 in the Arkansas randombred chicken population

 VL VR TD 
Incidence 26.23 4.25 2.13 
BWG 0-4 0.10** 0.08** 0.02 
BWG 0-6 0.09** 0.03 0.05 
 VL VR TD 
Incidence 26.23 4.25 2.13 
BWG 0-4 0.10** 0.08** 0.02 
BWG 0-6 0.09** 0.03 0.05 

1VL = valgus (right and left legs combined); VR = varus (right and left legs combined); TD = tibial dyschondroplasia; BWG 0-4 = BW gain from 0 to 4 wk of age; BWG 0-6 = BW gain from 0 to 6 wk of age. **(P < 0.01).

Table 2.

Descriptive statistics of leg problems and growth and feed efficiency traits in the Arkansas randombred chicken population

Trait1 Mean SD Minimum Maximum 
VL 2,257 0.26 0.44 0.00 1.00 
VR 2,257 0.04 0.20 0.00 1.00 
TD 2,257 0.02 0.14 0.00 1.00 
BWG 0-4 (kg) 2,257 0.82 0.12 0.18 1.21 
BWG 0-6 (kg) 2,257 1.65 0.21 0.71 2.36 
RFI 5-6 (g) 2,257 0.00 0.11 −0.53 0.85 
Trait1 Mean SD Minimum Maximum 
VL 2,257 0.26 0.44 0.00 1.00 
VR 2,257 0.04 0.20 0.00 1.00 
TD 2,257 0.02 0.14 0.00 1.00 
BWG 0-4 (kg) 2,257 0.82 0.12 0.18 1.21 
BWG 0-6 (kg) 2,257 1.65 0.21 0.71 2.36 
RFI 5-6 (g) 2,257 0.00 0.11 −0.53 0.85 

1VL = valgus (right and left legs combined); VR = varus (right and left legs combined); TD = tibial dyschondroplasia; BWG 0-4 = BW gain from 0 to 4 wk of age; BWG 0-6 = BW gain from 0 to 6 wk of age; RFI 5-6 = residual feed intake from 5 to 6 wk of age.

Table 3.

Posterior means (posterior SD) of genetic, residual variances, and heritability of leg problems and growth and feed efficiency traits in the Arkansas randombred chicken population

Trait1 Genetic variance Residual variance Heritability 
VL 0.170 (0.141) 1.000 0.13 (0.10) 
VR 0.128 (0.112) 1.000 0.11 (0.08) 
TD 0.172 (0.142) 1.000 0.12 (0.11) 
BWG 0-4 0.003 (0.003) 0.0082 (0.0004) 0.27 (0.04) 
BWG 0-6 0.006 (0.006) 0.0216 (0.0009) 0.23 (0.04) 
RFI 5-6 0.002 (0.002) 0.0116 (0.0004) 0.12 (0.02) 
Trait1 Genetic variance Residual variance Heritability 
VL 0.170 (0.141) 1.000 0.13 (0.10) 
VR 0.128 (0.112) 1.000 0.11 (0.08) 
TD 0.172 (0.142) 1.000 0.12 (0.11) 
BWG 0-4 0.003 (0.003) 0.0082 (0.0004) 0.27 (0.04) 
BWG 0-6 0.006 (0.006) 0.0216 (0.0009) 0.23 (0.04) 
RFI 5-6 0.002 (0.002) 0.0116 (0.0004) 0.12 (0.02) 

1VL = valgus (right and left legs combined); VR = varus (right and left legs combined); TD = tibial dyschondroplasia; BWG 0-4 = BW gain from 0 to 4 wk of age; BWG 0-6 = BW gain from 0 to 6 wk of age; RFI 5-6 = residual feed intake from 5 to 6 wk of age.

Table 4.

Genetic (above diagonal) and residual2 (below diagonal) correlations of leg problems and growth and feed efficiency traits in the Arkansas randombred chicken population

Trait1 VL VR TD BWG 0-4 BWG 0-6 RFI 5-6 
VL  −0.02 (0.11) 0.01 (0.15) 0.08 (0.14) 0.08 (0.14) 0.02 (0.08) 
VR 0.00 (0.27)  0.05 (0.13) 0.02 (0.14) 0.02 (0.14) 0.02 (0.09) 
TD 0.02 (0.30) 0.02 (0.29)  –0.01 (0.16) –0.01 (0.15) 0.01 (0.10) 
BWG 0-4 0.00 (0.12) 0.00 (0.11) 0.00 (0.11)  0.84 (0.00) 0.23 (0.02) 
BWG 0-6 0.00 (0.14) 0.00 (0.13) 0.00 (0.13) 0.67 (0.02)  0.26 (0.02) 
RFI 5-6 0.00 (0.07) 0.00 (0.07) 0.00 (0.07) –0.01 (0.03) –0.05 (0.03)  
Trait1 VL VR TD BWG 0-4 BWG 0-6 RFI 5-6 
VL  −0.02 (0.11) 0.01 (0.15) 0.08 (0.14) 0.08 (0.14) 0.02 (0.08) 
VR 0.00 (0.27)  0.05 (0.13) 0.02 (0.14) 0.02 (0.14) 0.02 (0.09) 
TD 0.02 (0.30) 0.02 (0.29)  –0.01 (0.16) –0.01 (0.15) 0.01 (0.10) 
BWG 0-4 0.00 (0.12) 0.00 (0.11) 0.00 (0.11)  0.84 (0.00) 0.23 (0.02) 
BWG 0-6 0.00 (0.14) 0.00 (0.13) 0.00 (0.13) 0.67 (0.02)  0.26 (0.02) 
RFI 5-6 0.00 (0.07) 0.00 (0.07) 0.00 (0.07) –0.01 (0.03) –0.05 (0.03)  

1VL = valgus (right and left legs combined); VR = varus (right and left legs combined); TD = tibial dyschondroplasia; BWG 0-4 = BW gain from 0 to 4 wk of age; BWG 0-6 = BW gain from 0 to 6 wk of age; RFI 5-6 = residual feed intake from 5 to 6 wk of age. 2Most of the residual correlations appears as zero due to the rounding to 2 decimal places; however, they are all different from zero.

DISCUSSION

The observed differential incidence between VL and VR was in concordance with previous studies (Le Bihan-Duval et al., 1996; Cook, 2000). Phenotypic correlations suggested that the higher the growth rate during the first 4 wk of age, the greater the incidence of VL and VR; a similar pattern was observed between VL and BWG 0-6. This observation would coincide with the association between growth and leg problems that has been discussed by some authors (Sorensen, 1992; Julian, 2005). Heritability of TD was lower than values reported by Sheridan et al. (1978) and Burton et al. (1981) (0.22 to 0.36); however, it was within the range seen in more recent investigations (0.10 to 0.27) (Kapell et al., 2012; Rekaya et al., 2013). Heritability of VL and VR in previous reports ranged from 0.21 to 0.38 and from 0.22 to 0.29, respectively (Mercer and Hill, 1984; Le Bihan-Duval et al., 1996, 1997; Rekaya et al., 2013). These estimates are higher than the heritability values of VVD that were found in the present study. The low heritability estimates of leg problems imply that the expression of these traits is highly influenced by environmental causal effects (Visscher et al., 2008). Therefore, in addition to genetic improvement, management strategies are needed to reduce the incidence of leg problems (Rekaya et al., 2013). Heritability of growth rate was lower than values from different age periods (0.39 to 0.52) (Pym et al., 1991; Druyan et al., 2007). Likewise, heritability of RFI 5-6 was low compared to the value recently estimated in a commercial broiler population for the period 6 to 7 wk of age (0.26) (Rekaya et al., 2013). Taken together with other studies, it appears there is an extremely weak genetic association between growth and VVD. The weak negative genetic association estimated between VVD was similar to that found by Le Bihan-Duval et al. (1996), even though they used a generalized linear model and a multinomial logistic transformation model. Rekaya et al. (2013) reported a rather high negative genetic relationship (–0.70) between VR and VL; however, in the current study, an extremely weak association was observed. This might be due to the fact that the commercial population used by Rekaya et al. (2013) had undergone genetic improvement for growth, whereas the population used in this study was a control population.

The favorable genetic association between VVD and TD was also observed by Kapell et al. (2012) in 2 of 3 broiler lines studied (0.10). Overall, the genetic correlations of leg problems and growth rates were similar to earlier reports in the literature. While genetic correlations of VVD with growth rate were unfavorable, those of TD with growth seemed slightly favorable. This latter result would tend to agree with the additive genetic association of TD with BW at 7 wk (–0.12 to –0.46) reported by Burton et al. (1981) and Zhang et al. (1995). However, current results are different from reports of Sheridan et al. (1978) and Kapell et al. (2012), who found unfavorable genetic correlations of TD with BW (0.16 to 0.38). On the other hand, the genetic association between VVD and growth found in the current study differs from other reports. For example, Rekaya et al. (2013) reported negative genetic correlations of BW at 6 wk with valgus (–0.20) and varus (–0.06). Likewise, Le Bihan-Duval et al. (1997) provided estimates of favorable genetic correlations of VVD with BW at 3 (BW3) and 6 wk (BW6), in 2 broiler lines. Genetic correlation of valgus and varus with BW3 ranged between –0.02 and –0.03 and between 0 and –0.10, respectively, in one line. The other line showed a favorable genetic correlation of varus with BW6 (–0.06). In contrast, the current weak unfavorable genetic correlation of VVD with growth rate is similar to what was reported by Mercer and Hill (1984) from commercial broiler strains with genetic correlation of valgus and varus with BW6 ranging from 0.02 to 0.25 and between 0.28 and 0.34, respectively, when estimated from half-sib analysis. Furthermore, using a full-sib analysis, genetic correlation of valgus and varus with BW6 ranged between 0.07 and 0.11 and between 0.17 and 0.25, respectively. Other unfavorable genetic correlations between VVD and BW6 were reported by Le Bihan-Duval et al. (1997). Genetic correlation of valgus with BW6 ranged between 0.01 and 0.09; and in one of the broiler lines that they studied, the genetic correlation of varus with BW6 was 0.08. Recently, Kapell et al. (2012) also reported unfavorable genetic associations of VVD with BW at 6 wk, which ranged from 0.09 to 0.25 in the 3 broiler lines that they studied. Genetic correlations of VL and TD with RFI 5-6 were similar to the values reported (0.04 and 0.02, respectively) by Rekaya et al. (2013); however, the genetic correlation between VR and RFI was contradictory. Residual correlations were low, except those between growth traits. In fact, the posterior means of residual correlations between leg soundness traits were practically equal to zero. However, their associated posterior SD were large. A further inspection of these results showed that although the Markov chain Monte Carlo has good mixed and converged to the target distribution (Figure 1), the marginal distributions of these correlations (Figure 2) are somewhat flat and covered almost the whole parametric space of a correlation coefficient (–1, 1). This is very likely due to the small size of the data and the low incidence of these traits. Thus, point estimates such as the posterior mean, although unbiased, should be interpreted with caution.

Figure 1.

Trace plot of the sampling process of the residual correlation between valgus (VL) and varus (VR)

Figure 1.

Trace plot of the sampling process of the residual correlation between valgus (VL) and varus (VR)

Figure 2.

Marginal posterior distribution of the residual correlation between valgus (VL) and varus (VR)

Figure 2.

Marginal posterior distribution of the residual correlation between valgus (VL) and varus (VR)

From all the aforementioned results, the genetic correlations between growth and leg problems is very low, varying from slightly negative to slightly positive. Considering the SD associated with such estimates, it becomes practically impossible to determine the true genetic relationship between growth and leg problems with certainty. Despite this uncertainty, Kapell et al. (2012) outlined a strategy for the simultaneous improvement of these traits that considers the discarding of breeder candidates with clinical symptoms and the identification, through breeding values, of breeder candidates that have not developed clinical symptoms but have increased propensity to develop the problems. Therefore, to achieve a simultaneous improvement, it has been suggested to include the fitness traits in the general breeding goal along with a sufficient weight for each trait (Wall et al., 2007; Berglund, 2008). The management strategies under which these birds are raised would contribute more significantly to reducing the incidence of leg problems, even though Kapell et al. (2012) clearly demonstrated that genetic selection for reduced leg problems can also achieve favorable results.

CONCLUSION

From the current study and studies already reported in the literature, it is amply clear that leg problems have a genetic component; however, the additive genetic variation is small. Despite the favorable response of leg problems to genetic improvement, optimal management strategies could go a long way toward reducing the incidence of leg problems in broilers. The current study showed that there is genetic variability underlying varus, valgus, and TD, but their association with increased growth is extremely weak at best.

Fernando González-Cerón was supported by CONACYT (National Council for Science and Technology, Mexico). We thank Arthur Karnuah and Mi Yeon Shim, Department of Poultry Science, University of Georgia for assisting in the collection of the data. We also thank Chris McKenzie, Poultry Research Center, University of Georgia for technical support.

REFERENCES

Aggrey
S. E.
Karnuah
A. B.
Sebastian
B.
Anthony
N. B.
Genetic properties of feed efficiency parameters in meat-type chickens
Genet. Sel. Evol.
 , 
2010
, vol. 
42
 pg. 
25
 
Berglund
B.
Genetic improvement of dairy cow reproductive performance
Reprod. Dom. Anim.
 , 
2008
, vol. 
43
 
Suppl. 2
(pg. 
89
-
95
)
Burton
R. W.
Sheridan
A. K.
Howlett
C. R.
The incidence and importance of tibial dyschondroplasia to the commercial broiler industry in Australia
Br. Poult. Sci.
 , 
1981
, vol. 
22
 (pg. 
153
-
160
)
Cook
M. E.
Patterson
P. H.
Sunde
M. L.
Leg deformities: Inability to increase severity by increasing body weight of chicks and poults
Poult. Sci.
 , 
1984
, vol. 
63
 (pg. 
620
-
627
)
Cook
M.E.
Skeletal deformities and their causes: Introduction
Poult. Sci.
 , 
2000
, vol. 
79
 (pg. 
982
-
984
)
Druyan
S.
Shlosberg
A.
Cahaner
A.
Evaluation of growth rate, body weight, heart rate, and blood parameters as potential indicators for selection against susceptibility to the ascites syndrome in young broilers
Poult. Sci.
 , 
2007
, vol. 
86
 (pg. 
621
-
629
)
Edwards
H. M.
Jr
Veltmann
J. R.
Jr
The role of calcium and phosphorus in the etiology of tibial dyschondroplasia in young chicks
J. Nutr.
 , 
1983
, vol. 
113
 (pg. 
1568
-
1575
)
European Commission (EC)
The welfare of chickens kept for meat production (broilers). Report of the Scientific Committee on Animal Health and Animal Welfare
European Commission, Health & Consumer Protection Directorate-General (adopted March 21, 2000)
 , 
2000
 
Hocking
P. M.
Sandercock
D. A.
Wilson
S.
Fleming
R. H.
Quantifying genetic (co)variation and effects of genetic selection on tibial bone morphology and quality in 37 lines of broiler, layer and traditional chickens
Br. Poult. Sci.
 , 
2009
, vol. 
50
 (pg. 
443
-
450
)
Hunter
B.
Whiteman
A.
Sanei
B.
Dam
A.
Valgus/varus leg deformities in poultry
Keeping Your Birds Healthy
 , 
2008
 
Julian
R. J.
Valgus-varus deformity of the intertarsal joint in broiler chickens
Can. Vet. J.
 , 
1984
, vol. 
25
 (pg. 
254
-
258
)
Julian
R. J.
Rapid growth problems: Ascites and skeletal deformities in broilers
Poult. Sci.
 , 
1998
, vol. 
77
 (pg. 
1773
-
1780
)
Julian
R. J.
Production and growth related disorders and other metabolic diseases of poultry: A review
Vet. J.
 , 
2005
, vol. 
169
 (pg. 
350
-
369
)
Kalmar
I. D.
Vanrompay
D.
Janssens
G. P. J.
Broiler ascites syndrome: Collateral damage from efficient feed to meat conversion
Vet. J.
 , 
2013
, vol. 
197
 (pg. 
169
-
174
)
Kapell
D. N. R. G.
Hill
W. G.
Neeteson
A. M.
McAdam
J.
Koerhuis
A. N. M.
Avendaño
S.
Twenty-five years of selection for improved leg health in purebred broiler lines and underlying genetic parameters
Poult. Sci.
 , 
2012
, vol. 
91
 (pg. 
3032
-
3043
)
Kestin
S. C.
Su
G.
Sorensen
P.
Different commercial broiler crosses have different susceptibilities to leg weakness
Poult. Sci.
 , 
1999
, vol. 
78
 (pg. 
1085
-
1090
)
Knowles
T. G.
Kestin
S. C.
Haslam
S. M.
Brown
S. N.
Green
L. E.
Butterworth
A.
Pope
S. J.
Pfeiffer
D.
Nicol
C. J.
Leg disorders in broiler chickens: Prevalence, risk factors and prevention
PLoS One
 , 
2008
, vol. 
3
 
2
pg. 
e1545
 
Kuhlers
D. L.
McDaniel
G. R.
Estimates of heritabilities and genetic correlations between tibial dyschondroplasia expression and body weight at two ages in broilers
Poult. Sci.
 , 
1996
, vol. 
75
 (pg. 
959
-
961
)
Le Bihan-Duval
E.
Beaumont
C.
Colleau
J. J.
Genetic parameters of the twisted legs syndrome in broiler chickens
Genet. Sel. Evol.
 , 
1996
, vol. 
28
 (pg. 
177
-
195
)
Le Bihan-Duval
E.
Beaumont
C.
Colleau
J. J.
Estimation of the genetic correlations between twisted legs and growth or conformation traits in broiler chickens
J. Anim. Breed. Genet.
 , 
1997
, vol. 
114
 (pg. 
239
-
259
)
Leach
R. M.
Jr.
Monsonego-Ornan
E.
Tibial dyschondroplasia 40 years later
Poult. Sci.
 , 
2007
, vol. 
86
 (pg. 
2053
-
2058
)
Leach
R. M.
Jr.
Nesheim
M. C.
Nutritional, genetic and morphological studies of an abnormal cartilage formation in young chicks
J. Nutr.
 , 
1965
, vol. 
86
 (pg. 
236
-
244
)
Leterrier
C.
Nys
Y.
Clinical and anatomical differences in varus and valgus deformities of chick limbs suggest different aetio-pathogenesis
Avian Pathol.
 , 
1992
, vol. 
21
 (pg. 
429
-
442
)
Mercer
J. T.
Hill
W. G.
Estimation of genetic parameters for skeletal defects in broiler chickens
Heredity
 , 
1984
, vol. 
53
 (pg. 
193
-
203
)
Pym
R. A.
Johnson
R. J.
Etse
D. B.
Eason
P.
Inheritance of plasma insulin-like growth factor-I and growth rate, food intake, food efficiency and abdominal fatness in chickens
Br. Poult. Sci.
 , 
1991
, vol. 
32
 (pg. 
285
-
293
)
Randall
C. J.
Mills
C. P. J.
Observations on leg deformity in broilers with particular reference to the intertarsal joint
Avian Pathol.
 , 
1981
, vol. 
10
 (pg. 
407
-
431
)
Rekaya
R.
Sapp
R. L.
Wing
T.
Aggrey
S. E.
Genetic evaluation for growth, body composition, feed efficiency, and leg soundness
Poult. Sci.
 , 
2013
, vol. 
92
 (pg. 
923
-
929
)
SAS Institute
SAS User's Guide: Statistics. Version 9.1.3 ed
 , 
2006
Cary, NC
SAS Inst. Inc.
Sheridan
A. K.
Howlett
C. R.
Burton
R. W.
The inheritance of tibial dyschondroplasia in broilers
Br. Poult. Sci.
 , 
1978
, vol. 
19
 (pg. 
491
-
499
)
Shim
M. Y.
Karnuah
A. B.
Anthony
N. B.
Pesti
G. M.
Aggrey
S. E.
The effects of broiler chicken growth rate on valgus, varus, and tibial dyschondroplasia
Poult. Sci.
 , 
2012
, vol. 
91
 (pg. 
62
-
65
)
Sorensen
P.
Whitehead
C. C.
The genetics of leg disorders
Bone Biology and Skeletal Disorders in Poultry. Poultry Science Symposium 23
 , 
1992
Abingdon, UK
Carfax Publishing Co.
(pg. 
213
-
229
)
Sullivan
T. W.
Skeletal problems in poultry: Estimated annual cost and descriptions
Poult. Sci.
 , 
1994
, vol. 
73
 (pg. 
879
-
882
)
Tuyttens
F.
Heyndrickx
M.
De Boeck
M.
Moreels
A.
Van Nuffel
A.
Van Poucke
E.
Van Coillie
E.
Van Dongen
S.
Lens
L.
Broiler chicken health, welfare and fluctuating asymmetry in organic versus conventional production systems
Livest. Sci.
 , 
2008
, vol. 
113
 (pg. 
123
-
132
)
Visscher
P. M.
Hill
W. G.
Wray
N. R.
Heritability in the genomics era: Concepts and misconceptions
Nature Rev. Genet.
 , 
2008
, vol. 
9
 (pg. 
255
-
266
)
Wall
E.
Coffey
M. P.
Brotherstone
S.
The relationship between body energy traits and production and fitness traits in first-lactation dairy cows
J. Dairy Sci.
 , 
2007
, vol. 
90
 (pg. 
1527
-
1537
)
Zhang
X.
McDaniel
G. R.
Yalcin
Z. S.
Kuhlers
D. L.
Genetic correlations of tibial dyschondroplasia incidence with carcass traits in broilers
Poult. Sci.
 , 
1995
, vol. 
74
 (pg. 
910
-
915
)