TY - JOUR
AU - Garrick, D. J.
AB - Abstract The objective of this study was to quantify the role of maternal effects on docility in Limousin cattle. Docility scores were obtained at weaning while animals were restrained in a squeeze chute. Scores 1 through 6 represented a docile to aggressive temperament, respectively, and were provided by the North American Limousin Foundation. Observations with unknown age of dam, contemporary groups containing less than 10 observations, contemporary groups with no variation, and single-sire contemporary groups were removed, leaving 21,932 observations. A 2-generation pedigree file compiled from animals with observations contained 49,459 animals. Fixed effects were weaning contemporary group and age of dam (2, ≥3 yr). Six animal models encompassed combinations of random factors: direct genetic, maternal genetic, and maternal permanent environmental effects. The model D was the most basic, containing direct genetic and residual effects, and it resembled the method currently used by the North American Limousin Foundation for genetic evaluation of docility. Maternal genetic or permanent environmental effects were separately added to the model D, denoted as models DM and DC, respectively. Model DMC contained all random factors. Models DM-Zero and DMC-Zero were equivalent to models DM and DMC, respectively, but with zero direct-maternal genetic covariance. Direct heritability estimates were moderate for all models (0.29 ± 0.02 to 0.38 ± 0.03). Maternal heritability estimates were low, ranging from 0.01 ± 0.01 (DM-Zero) to 0.05 ± 0.02 (DM). Negative direct-maternal genetic correlations of −0.41 ± 0.09 and −0.55 ± 0.09 were estimated for models DM and DMC, respectively. The proportion of phenotypic variance accounted for by maternal permanent environmental effects was 0.03 ± 0.01, 0.04 ± 0.01, and 0.02 ± 0.01 for models DC, DMC, and DMC-Zero, respectively. Likelihood ratio tests indicated that model DMC best fit the data. Although maternal genetic and maternal permanent environmental effects were significant, they accounted for only 8% (model DMC) of the phenotypic variance, and a Spearman rank correlation of 0.99 between models D and DMC showed sires did not rank differently with or without inclusion of these effects. Given these results, inclusion of maternal effects to the genetic evaluation of docility in Limousin cattle does not seem warranted. INTRODUCTION Poor temperament in beef cattle has been associated with reduced performance, health, and carcass quality in beef cattle. Cattle with calm temperaments have been found to have greater ADG (Burrow, 1997; Voisinet et al., 1997b) and decreased incidence of dark cutting beef (Voisinet et al., 1997a; Scanga et al., 1998) compared with cattle with anxious temperaments. Feedlot cattle with excitable temperaments have been found to have lower immune function (Fell et al., 1999) and tougher meat (Voisinet et al., 1997a; King et al., 2006; Vann, 2006) than calm cattle. Busby et al. (2005) examined the effect of disposition on feedlot performance and carcass quality grade and reported that docile calves returned $62.19 per head more than aggressive calves. Additionally, aggressive cattle jeopardize stockperson safety and are more likely to become injured during handling (Grandin, 1989). The combination of these factors make docility an economically relevant trait (Golden et al., 2000) that should be strongly considered by beef producers when breeding or purchasing cattle. In 1998, the first national genetic evaluation of docility in beef cattle was published by the North American Limousin Foundation (NALF; NALF, 2004). Docility EPD reflects the probability that the offspring will inherit genes for acceptable behavior, with a greater docility EPD associated with progeny exhibiting calmer behavior. Traits such as birth weight and weaning weight are known to be influenced by maternal effects (Willham, 1972), as a portion of the environment experienced by a calf is provided by its dam. It was hypothesized that a similar dam-offspring relationship may exist in docility. Therefore, the objective of this study was to determine the role of maternal effects on calf docility. Additive maternal genetic effects, nongenetic maternal permanent environmental effects, and the correlation between direct and maternal genetic effects were examined. MATERIALS AND METHODS Animal Care Data for this study were obtained from an existing, historical database (NALF) and therefore were not subject to Animal Care and Use Committee approval. Current Methodology The genetic evaluation of docility in Limousin cattle currently utilizes a single trait, single component threshold model with random direct genetic and residual effects, along with fixed effects of weaning contemporary group and sex. Threshold analyses assume continuous distribution on an unobservable, underlying scale following the assumptions of a mixed linear model (Gianola and Foulley, 1983; Harville and Mee, 1984). A maximum a posteriori probit threshold model is used to generate genetic predictions of docility on the underlying scale, which are transformed to an observable scale, expressed as deviations from a 50% probability (Kuehn et al., 1998). In contrast, this study analyzed docility as a continuous trait and evaluated the possible influence of maternal effects on docility. Description of Data Docility scores, pedigrees, and other pertinent performance information were obtained from NALF. Guidelines for categorizing temperament (NALF, 2004) were assumed to have been used by producers to allocate docility scores at weaning. Basic assessment of the level of aggressiveness was determined while the cattle were restrained in the chute. Individuals with scores of 1 or 2 were considered docile or mildly restless and were handled with little trouble. A score of 3, defined as the typical (average) temperament, was assigned if the animal was nervous, impatient, or exhibited a moderate amount of struggle. Animals who scored 4, 5, or 6 (flighty to very aggressive) were very nervous, difficult to handle, and aggressive, exhibiting attack behavior when handled individually. The 3 latter scores comprised animals deemed to possess unacceptable behavior. Initial data were filtered based on several criteria before being included in this study. Weaning contemporary groups with less than 10 observations, contemporary groups with no variation, as well as single-sire contemporary groups were removed from the analyses. Contemporary groups were formed by combining NALF weaning contemporary group, breeder management code, work group, sex, and creep designation. Preliminary analysis of age of dam effects showed a significant difference (P < 0.05) between docility scores of offspring reared by 2-yr-old females vs. offspring reared by 3-through 13-yr-old females (Table 1). Individuals were classified as having age of dam 2 or age of dam 3 and older, whereas animals with unknown age of dam observations were removed. The final data set contained a total of 21,932 animals, with docility observations in a total of 1,237 contemporary groups. Table 1. Estimates, SE, and t-values associated with the AOD1 fixed effect AOD  Estimates ± SE2  t-values2  2  −0.07 ± 0.02  3.423  3  0.01 ± 0.02  0.62  44  0.00  —4  5  0.01 ± 0.02  0.56  6  0.02 ± 0.02  1.2  7  0.03 ± 0.02  1.31  8  0.02 ± 0.02  0.83  9  0.01 ± 0.02  0.28  10  0.002 ± 0.03  0.08  11  −0.01 ± 0.03  0.5  12  −0.003 ± 0.03  0.09  13  0.06 ± 0.03  1.83  AOD  Estimates ± SE2  t-values2  2  −0.07 ± 0.02  3.423  3  0.01 ± 0.02  0.62  44  0.00  —4  5  0.01 ± 0.02  0.56  6  0.02 ± 0.02  1.2  7  0.03 ± 0.02  1.31  8  0.02 ± 0.02  0.83  9  0.01 ± 0.02  0.28  10  0.002 ± 0.03  0.08  11  −0.01 ± 0.03  0.5  12  −0.003 ± 0.03  0.09  13  0.06 ± 0.03  1.83  1 AOD = age of dam contemporary group. 2 Estimates ± SE and t-values from the initial analysis. 3 The t-value was statistically different from zero at P < 0.05. 4 Level of AOD constrained to remove the singularity in the mixed model equations. View Large Table 1. Estimates, SE, and t-values associated with the AOD1 fixed effect AOD  Estimates ± SE2  t-values2  2  −0.07 ± 0.02  3.423  3  0.01 ± 0.02  0.62  44  0.00  —4  5  0.01 ± 0.02  0.56  6  0.02 ± 0.02  1.2  7  0.03 ± 0.02  1.31  8  0.02 ± 0.02  0.83  9  0.01 ± 0.02  0.28  10  0.002 ± 0.03  0.08  11  −0.01 ± 0.03  0.5  12  −0.003 ± 0.03  0.09  13  0.06 ± 0.03  1.83  AOD  Estimates ± SE2  t-values2  2  −0.07 ± 0.02  3.423  3  0.01 ± 0.02  0.62  44  0.00  —4  5  0.01 ± 0.02  0.56  6  0.02 ± 0.02  1.2  7  0.03 ± 0.02  1.31  8  0.02 ± 0.02  0.83  9  0.01 ± 0.02  0.28  10  0.002 ± 0.03  0.08  11  −0.01 ± 0.03  0.5  12  −0.003 ± 0.03  0.09  13  0.06 ± 0.03  1.83  1 AOD = age of dam contemporary group. 2 Estimates ± SE and t-values from the initial analysis. 3 The t-value was statistically different from zero at P < 0.05. 4 Level of AOD constrained to remove the singularity in the mixed model equations. View Large Pedigree information from 1,805,286 individuals was provided by NALF. A 2-generation pedigree was compiled for animals with docility observations and contained 56,521 animals. Elimination of 5,946 dams and 1,116 sires with only 1 offspring in the data file (foundation animals) resulted in a final pedigree of 49,459 used in the analyses of docility. An average of 11.5 offspring per sire and 1.5 offspring per dam were represented. Seven percent (1,606) of the 21,932 individuals with docility scores were found to have contributed to the maternal component of the study, as they went on to become dams. Table 2 summarizes the number of docility observations in the final data file and the corresponding pedigree file. Table 2. Summary of docility observations, means,1 levels of fixed effects,2 and pedigree information Data file  Count/value  Observations  21,932  Mean docility score  1.93  SD  0.85  Sires  1,878  Dams  13,829  Contemporary groups3  1,237  Age of dam groups4  2  Pedigree file      Sires  4,724      Dams  23,074      Total  49,459  Data file  Count/value  Observations  21,932  Mean docility score  1.93  SD  0.85  Sires  1,878  Dams  13,829  Contemporary groups3  1,237  Age of dam groups4  2  Pedigree file      Sires  4,724      Dams  23,074      Total  49,459  1 Unadjusted means and SD. 2 Fixed effects included weaning contemporary group and age of dam. 3 Contemporary group = herd, year, sex, season of birth, weaning date, North American Limousin Foundation weaning contemporary group, and creep designation. 4 Age of dam = age of dam 2 and 3 through 13. View Large Table 2. Summary of docility observations, means,1 levels of fixed effects,2 and pedigree information Data file  Count/value  Observations  21,932  Mean docility score  1.93  SD  0.85  Sires  1,878  Dams  13,829  Contemporary groups3  1,237  Age of dam groups4  2  Pedigree file      Sires  4,724      Dams  23,074      Total  49,459  Data file  Count/value  Observations  21,932  Mean docility score  1.93  SD  0.85  Sires  1,878  Dams  13,829  Contemporary groups3  1,237  Age of dam groups4  2  Pedigree file      Sires  4,724      Dams  23,074      Total  49,459  1 Unadjusted means and SD. 2 Fixed effects included weaning contemporary group and age of dam. 3 Contemporary group = herd, year, sex, season of birth, weaning date, North American Limousin Foundation weaning contemporary group, and creep designation. 4 Age of dam = age of dam 2 and 3 through 13. View Large A skewed distribution of docility scores was observed due to a large percentage of animals with scores of 1 and 2 (Figure 1). Docility scores were transformed to expected normal scores. These standardized scores ensured the difference between docility observations 1 and 2 was consistent with the difference between observations 2 and 3, correcting for inadequacies due to the subjective scoring system. Dispersion and frequency of the 21,932 docility scores observed in the final data set, as well as subsequent standardized scores, are shown in Table 3. Figure 1. View largeDownload slide Docility score distribution on a percentage basis. Figure 1. View largeDownload slide Docility score distribution on a percentage basis. Table 3. Distribution of docility observations and standardized scores Docility score  Observations, No.  Frequency1  SS2  1  7,836  35.73  −1.05  2  8,620  39.30  0.14  3  4,805  21.91  1.13  4  567  2.59  2.14  5  68  0.31  2.74  6  36  0.16  3.23  Docility score  Observations, No.  Frequency1  SS2  1  7,836  35.73  −1.05  2  8,620  39.30  0.14  3  4,805  21.91  1.13  4  567  2.59  2.14  5  68  0.31  2.74  6  36  0.16  3.23  1 Frequency of docility scores expressed as a percentage of the number of observations. 2 SS = standardized score used in model analyses of docility in Limousin cattle. View Large Table 3. Distribution of docility observations and standardized scores Docility score  Observations, No.  Frequency1  SS2  1  7,836  35.73  −1.05  2  8,620  39.30  0.14  3  4,805  21.91  1.13  4  567  2.59  2.14  5  68  0.31  2.74  6  36  0.16  3.23  Docility score  Observations, No.  Frequency1  SS2  1  7,836  35.73  −1.05  2  8,620  39.30  0.14  3  4,805  21.91  1.13  4  567  2.59  2.14  5  68  0.31  2.74  6  36  0.16  3.23  1 Frequency of docility scores expressed as a percentage of the number of observations. 2 SS = standardized score used in model analyses of docility in Limousin cattle. View Large The Models Six models were produced by incorporating 3 factors: 1) additive direct genetic, 2) additive maternal genetic, and 3) maternal permanent environmental effects. The direct genetic component accounted for the effect an individual's genes have on its own performance (observed docility). Maternal genetic effects accounted for genes in the dam that influenced the phenotype of its offspring through the environment it provided for its calf. Maternal permanent environmental effects described the nongenetic environmental factors experienced by the dam that had a permanent effect on the environment it provided and were expressed in the phenotype of its offspring. The 6 models and their corresponding components are listed in Table 4. Table 4. Random components in models for analyses of docility in Limousin cattle   Components1  Models  σ2D  σ 2M  σDM  σ 2C  D  ✓  —  —  —  DM  ✓  ✓  ✓  —  DM-Zero  ✓  ✓  —  —  DC  ✓  —  —  ✓  DMC  ✓  ✓  ✓  ✓  DMC-Zero  ✓  ✓  —  ✓    Components1  Models  σ2D  σ 2M  σDM  σ 2C  D  ✓  —  —  —  DM  ✓  ✓  ✓  —  DM-Zero  ✓  ✓  —  —  DC  ✓  —  —  ✓  DMC  ✓  ✓  ✓  ✓  DMC-Zero  ✓  ✓  —  ✓  1 ✓= the effect was included in the model; σD2= random direct genetic; σ2M = maternal genetic; σDM = direct-maternal genetic covariance of phenotypic variance; and σ2C = maternal permanent environmental variance. View Large Table 4. Random components in models for analyses of docility in Limousin cattle   Components1  Models  σ2D  σ 2M  σDM  σ 2C  D  ✓  —  —  —  DM  ✓  ✓  ✓  —  DM-Zero  ✓  ✓  —  —  DC  ✓  —  —  ✓  DMC  ✓  ✓  ✓  ✓  DMC-Zero  ✓  ✓  —  ✓    Components1  Models  σ2D  σ 2M  σDM  σ 2C  D  ✓  —  —  —  DM  ✓  ✓  ✓  —  DM-Zero  ✓  ✓  —  —  DC  ✓  —  —  ✓  DMC  ✓  ✓  ✓  ✓  DMC-Zero  ✓  ✓  —  ✓  1 ✓= the effect was included in the model; σD2= random direct genetic; σ2M = maternal genetic; σDM = direct-maternal genetic covariance of phenotypic variance; and σ2C = maternal permanent environmental variance. View Large The basic, single component model equation  \[D:\ \mathbf{y}\ =\ \mathbf{Xb}\ +\ \mathbf{Z}_{D}\mathbf{u}_{D}\ +\ \mathbf{e}\] contained direct genetic and residual effects. Maternal genetic or maternal permanent environmental components were separately added to model D to form the model equations  \[DM:\ \mathbf{y}\ =\ \mathbf{Xb}\ +\ \mathbf{Z}_{D}\mathbf{u}_{D}\ +\ \mathbf{Z}_{M}\mathbf{u}_{M}\ +\ \mathbf{e}\] and  \[DC:\ \mathbf{y}\ =\ \mathbf{Xb}\ +\ \mathbf{Z}_{D}\mathbf{u}_{D}\ +\ \mathbf{Z}_{C}\mathbf{u}_{C}\ +\ \mathbf{e}.\] Direct genetic, maternal genetic, and maternal permanent environmental components were combined to form the model equation  \[DMC:\ \mathbf{y}\ =\ \mathbf{Xb}\ +\ \mathbf{Z}_{D}\mathbf{u}_{D}\ +\ \mathbf{Z}_{M}\mathbf{u}_{M}\ +\ \mathbf{Z}_{C}\mathbf{u}_{C}\ +\ \mathbf{e}.\] Two additional models, (DM-Zero) and (DMC-Zero), were formed from the DM and DMC model equations by assuming zero direct-maternal genetic covariance. Vector y contained transformed docility values; X was a known matrix relating the fixed effects in b (weaning contemporary group, age of dam) to the observations in y; ZD, ZM, and ZC were known incidence matrices relating the random animal effects uD, uM, and uC for direct, maternal, and maternal permanent environmental effects, respectively. Vector e represented random residual effects corresponding to docility values in y. The (co)variance structure of random effects in model (DMC) was written as:  \[V\left[\begin{array}{l}\mathit{\mathbf{u}_{D}}\\\mathit{\mathbf{u}_{M}}\\\mathit{\mathbf{u}_{C}}\\\mathbf{\mathit{e}}\end{array}\right]\ \mathbf{=}\ \left[\begin{array}{llll}\mathit{A}{\sigma}^{2}_{\mathit{D}}&\mathit{A}{\sigma}_{\mathit{D}M}&0&0\\\mathit{A}{\sigma}_{\mathit{D}M}&\mathit{A}{\sigma}^{2}_{\mathit{M}}&0&0\\0&0&\mathit{I_{M}}{\sigma}^{2}_{\mathit{C}}&0\\0&0&0&\mathit{I_{N}{\sigma}^{2}_{\mathit{e}}}\end{array}\right],\] with simpler models having structures represented by deleting relevant rows and columns. Wright's numerator relationship matrix was represented by A; IM and IN were identity matrices accounting for the number of dams that have offspring with a docility observation and the total number of docility observations scored at weaning, respectively. The additive direct genetic variance, additive maternal genetic variance, and maternal permanent environmental variance were designated σ2D, σ 2M, and σ 2C, respectively. Direct-maternal additive genetic covariance was denoted σDM and was zero (σDM = 0) in models DM-Zero and DMC-Zero. The remaining environmental (residual) variance was signified by σ 2e Estimation of Variance Components Variance components were estimated with ASReml (Gilmour et al., 2002), which fits linear mixed models using REML. Convergence was presumed when the REML log likelihood changed by less than 0.002 in successive iterations, and individual variance parameter estimates changed by less than 1% (Gilmour et al., 2002). Convergence criterion was met at no more than 8 iterations for all models. Estimates were used to calculate heritabilities for direct and maternal genetic effects and the correlation between direct and maternal effects. A variance component ratio was used to calculate the amount of phenotypic variance explained by maternal permanent environmental effects (i.e., C2). Model Comparison Likelihood ratio tests (LRT) were used to determine whether a full model fit the data significantly better than a simple model. Use of LRT for model comparisons can only be applied when full models encompass the parameters of simpler models. Log-likelihood (logL) values from each model were compared using the test statistic  \[\mathit{D}\ =\ 2|log\mathit{L}_{\mathit{F}}\ {-}log\mathit{L}_{\mathit{S}}|,\] where D represented twice the absolute difference between the full model REML logL (logLF) and the simple model REML logL (logLS). The null hypotheses stated that the full models did not fit significantly better than the simple models. A χ2 distribution, with the difference between the number of parameters fit for the full and simple models as the degrees of freedom, was calculated to determine the associated significance value. Level of significance was set at P < 0.01. Comparison of EPD Docility EPD solutions for continuous models were obtained with ASReml (Gilmour et al., 2002). Spearman rank correlations (SAS Inst. Inc., Cary, NC) were performed to compare sire docility EPD estimates between the basic model (D) and the complete model (DMC). Nongenetic Dam-Offspring Covariance Model DMC was further analyzed to determine whether the direct-maternal genetic correlation was biased due to nongenetic dam-offspring covariances present within the data. A second data file was formed in which docility values of individuals occurring as both offspring and dam (7% of the records) were set to unknown. RESULTS Parameter Estimates Heritability estimates, direct-maternal genetic correlations, proportions of phenotypic variance accounted for by maternal permanent environmental effects, and log-likelihood estimates used for LRT are in Table 5. Table 5. Estimates of log likelihood, phenotypic variance, and model parameters for docility in Limousin cattle using the program ASReml       Parameters1  Models  logL2  σ2PHEN3  h2D  h2M  rDM  C2  D  −2,569  0.435 ± 0.01  0.34 ± 0.01  —  —  —  DM  −2,559  0.436 ± 0.01  0.37 ± 0.03  0.05 ± 0.02  −0.41 ± 0.09  —  DM-Zero  −2,564  0.433 ± 0.01  0.31 ± 0.02  0.02 ± 0.01  —  —  DC  −2,564  0.431 ± 0.01  0.31 ± 0.02  —  —  0.03 ± 0.01  DMC  −2,554  0.434 ± 0.01  0.38 ± 0.03  0.04 ± 0.03  −0.55 ± 0.09  0.04 ± 0.01  DMC-Zero  −2,562  0.431 ± 0.01  0.29 ± 0.02  0.01 ± 0.01  —  0.02 ± 0.01        Parameters1  Models  logL2  σ2PHEN3  h2D  h2M  rDM  C2  D  −2,569  0.435 ± 0.01  0.34 ± 0.01  —  —  —  DM  −2,559  0.436 ± 0.01  0.37 ± 0.03  0.05 ± 0.02  −0.41 ± 0.09  —  DM-Zero  −2,564  0.433 ± 0.01  0.31 ± 0.02  0.02 ± 0.01  —  —  DC  −2,564  0.431 ± 0.01  0.31 ± 0.02  —  —  0.03 ± 0.01  DMC  −2,554  0.434 ± 0.01  0.38 ± 0.03  0.04 ± 0.03  −0.55 ± 0.09  0.04 ± 0.01  DMC-Zero  −2,562  0.431 ± 0.01  0.29 ± 0.02  0.01 ± 0.01  —  0.02 ± 0.01  1 h2D = direct heritability; hM2 = maternal heritability; rDM = direct-maternal correlation; and C2 = maternal permanent environmental proportion of phenotypic variance. 2 logL = log-likelihood estimate. 3 σ2PHEN = phenotypic variance. View Large Table 5. Estimates of log likelihood, phenotypic variance, and model parameters for docility in Limousin cattle using the program ASReml       Parameters1  Models  logL2  σ2PHEN3  h2D  h2M  rDM  C2  D  −2,569  0.435 ± 0.01  0.34 ± 0.01  —  —  —  DM  −2,559  0.436 ± 0.01  0.37 ± 0.03  0.05 ± 0.02  −0.41 ± 0.09  —  DM-Zero  −2,564  0.433 ± 0.01  0.31 ± 0.02  0.02 ± 0.01  —  —  DC  −2,564  0.431 ± 0.01  0.31 ± 0.02  —  —  0.03 ± 0.01  DMC  −2,554  0.434 ± 0.01  0.38 ± 0.03  0.04 ± 0.03  −0.55 ± 0.09  0.04 ± 0.01  DMC-Zero  −2,562  0.431 ± 0.01  0.29 ± 0.02  0.01 ± 0.01  —  0.02 ± 0.01        Parameters1  Models  logL2  σ2PHEN3  h2D  h2M  rDM  C2  D  −2,569  0.435 ± 0.01  0.34 ± 0.01  —  —  —  DM  −2,559  0.436 ± 0.01  0.37 ± 0.03  0.05 ± 0.02  −0.41 ± 0.09  —  DM-Zero  −2,564  0.433 ± 0.01  0.31 ± 0.02  0.02 ± 0.01  —  —  DC  −2,564  0.431 ± 0.01  0.31 ± 0.02  —  —  0.03 ± 0.01  DMC  −2,554  0.434 ± 0.01  0.38 ± 0.03  0.04 ± 0.03  −0.55 ± 0.09  0.04 ± 0.01  DMC-Zero  −2,562  0.431 ± 0.01  0.29 ± 0.02  0.01 ± 0.01  —  0.02 ± 0.01  1 h2D = direct heritability; hM2 = maternal heritability; rDM = direct-maternal correlation; and C2 = maternal permanent environmental proportion of phenotypic variance. 2 logL = log-likelihood estimate. 3 σ2PHEN = phenotypic variance. View Large A moderate direct heritability estimate of 0.34 ± 0.01 observed in model D was comparable to an unweighted average (0.36) of various measurements of temperament reported by Burrow (1997), but proved to be greater than the estimate of 0.22 in Limousin cattle reported by Le Neindre et al. (1995). Direct heritability estimates ranged from 0.29 ± 0.02 (DMC-Zero) to 0.38 ± 0.03 (DMC). Maternal heritability estimates of 0.05 ± 0.02 and 0.04 ± 0.03 were observed for DM and DMC models, respectively, and were lower for models DM-Zero (0.02 ± 0.01) and DMC-Zero (0.01 ± 0.01). Negative direct-maternal genetic correlations of −0.41 ± 0.09 for model DM and −0.55 ± 0.09 for model DMC were estimated. The negative correlation implies that animals with superior genes for docility tend to have inferior genes for maternal docility. This suggests that a female scored as docile at weaning, who went on to become a dam, would provide a bad environment in terms of temperament for her offspring. Survival instincts might provide a possible explanation, because calves with docile mothers may develop aggressive behavior as a defense mechanism, compensating for a lack of protection provided by their dam. This interpretation, however, is still questionable. Burrow (2001) found a similar estimate of −0.59 between direct and maternal genetic effects when analyzing temperament in a composite breed of tropical cattle (25% Hereford, 25% Shorthorn and 50% Africander in herd AX, and 25% each of Africander, Brahman, Here-ford, and Shorthorn in herd AXBX) using a repeated animal model with maternal effects. It was deduced by Burrow (2001) that the significant negative direct-maternal correlation simply reflected the fact that maternal genetic effects were not important when evaluating temperament. However, results observed by Burrow (2001) may be attributable to the small data set of 1,871 animals (8,943 records) as well as the small 4,518-animal pedigree file compiled from animals with observations and their nonrecorded ancestors. Proportions of phenotypic variance accounted for by maternal permanent environmental effects in the current study were 0.03 ± 0.01, 0.04 ± 0.01, and 0.02 ± 0.01 for models DC, DMC, and DMC-Zero, respectively. Maternal permanent environmental effects are exemplified as the effect of chronic mastitis relative to the reduced level of milk production throughout the life of a dam (Beef Improvement Federation, 2002). Evidence would infer that environmental factors provided by the dam may play a minor role in the phenotypic expression of docility. Nongenetic Dam-Offspring Covariances A negative direct-maternal correlation of −0.50 ± 0.11 was estimated for model DMC using a second data file in which docility observations of females that went on to become dams were set to unknown. Similar estimates produced by the first −0.55 and second −0.50 data sets suggest the direct-maternal genetic correlation is not biased due to nongenetic dam-offspring covariances. Model Comparisons with LRT Model comparisons using LRT revealed the most complete model (DMC) provided the best fit of the 6 models. The simple model (D) did not fit as well as any other model. Models DM-Zero and DMC-Zero did not fit as well as models DM and DMC, which included a parameter to define the direct-maternal genetic correlation. Rank Correlations Spearman rank correlation coefficient between sire docility EPD estimated from model D and model DMC was 0.99 (P < 0.0001). Results indicate addition of maternal genetic and maternal permanent environmental effects does not cause sires to rank differently for direct genetic merit. DISCUSSION Measuring Temperament Temperament has been described as the reaction of beef cattle to handling by humans (Burrow, 1997). Various methods used to measure temperament in the literature include flight speed (FS) test (Burrow et al., 1988), docility test (LeNeindre et al., 1995; Grignard et al., 2001), and crush test (Tulloh, 1961; Grignard et al., 2001). Flight speed test objectively measures the time taken (in hundredths of a second) for an animal to pass through 2 light beams separated by a distance of 1.7 m after leaving a weight crush or chute (Burrow et al., 1988). Flight speed may be reflective of intrinsic fearfulness (Petherick et al., 2002) and has been used to measure temperament in cattle, in which faster FS reflects poorer temperaments and slower FS indicates calmer temperaments (Burrow, 1997). Direct heritability of 0.40 was estimated for FS in a tropical breed of beef cattle (Burrow 2001); however, both data and pedigree files were relatively small and comprised animals of 2 different, but very similar composite breeds. An association between FS and ADG in feedlot cattle has been shown in the literature (Burrow and Dillon, 1997; Petherick et al., 2002; Müller and von Keyserlingk, 2005). Burrow and Dillon (1997) found a negative correlation between FS and ADG in 1 group of Bos indicus cattle, but not in the second group. Petherick et al. (2002) reported significant but low correlations between FS and ADG in Bos indicus cross steers (−0.24 average from 7 correlations), and Müller and von Keyserlingk (2005) observed a quadratic rather than linear relationship between FS and ADG in Angus cross heifers (i.e., animals with the greatest FS had the lowest ADG, but many animals with low FS also had low ADG). Müller and von Keyserlingk (2005) also examined the correlation between ADG and personality traits measured in a social separation test (e.g., cattle were isolated from pen mates and video recorded to quantify locomotion, mobility, etc.) and concluded that fast animals were not the most fearful, as previously thought, but animals with lower ADG were the most fearful. Even though FS has proven to be an objective measure of flight speed, the relationship between FS and ADG is not clear (Müller and von Keyserlingk, 2005). Additionally, if temperament is reflective of an animals' reaction toward handling by humans, allocating temperament (good vs. poor) based solely on FS may not be appropriate (Müller and von Keyserlingk, 2005). The docility test, similar to the social separation test (Müller and von Keyserlingk, 2005) and the handling test (Boivin et al., 1992), measures total time in locomotion and changes in mobility in an animal, along with aggressiveness toward humans. Using the handling test, Boivin et al. (1994) found that significant sire effects influenced aggressiveness toward humans in Limousin heifers (P < 0.05). Le Neindre et al. (1995) estimated the heritability of docility in Limousin cattle to be 0.22 using procedures similar to Boivin et al. (1992). The crush test (Tulloh, 1961) allows for assessment of animals confined in a chute. Following Tulloh (1961), Hearnshaw and Morris (1984) reported heritability estimates for temperament of 0.03 ± 0.28 for Bos taurus-sired calves and 0.46 ± 0.37 for Bos indicus-sired calves. Grignard et al. (2001) evaluated temperament in Limousin heifers using both the docility test (similar to Le Neindre et al., 1995) and the crush test and found sire effect was significant for both tests (P < 0.05), and heifer responses to the docility test were significantly correlated with their responses to the crush test (P < 0.001). Overall, results indicated a general reaction of beef cattle to handling by humans, which was influenced by sire (Grignard et al., 2001). Although subjective, NALF guidelines (NALF, 2004) used to allocate docility score encompass many aspects illustrated in other tests. Several of these factors include general behavior when restrained in a chute (i.e., crush test, Tulloh, 1961), rate at which a calf exits the chute (e.g., slow vs. fast), vocalization (Watts et al., 2001), and aggressiveness toward humans (i.e., docility test; social separation test, Müller and von Keyserlingk, 2005; handling test, Boivin et al., 1992). Results of direct heritability estimates from the current study correspond to the average, unweighted estimate of 0.36 obtained from various measures of temperament reviewed by Burrow (1997). Additionally, docility scores were standardized to account for subjectivity (i.e., the difference between docility observations 1 and 2 consisted with the difference between observations 2 and 3, etc.). Efficient scoring systems must reflect typical, everyday handling practices used by cattle producers, while also being simple and inexpensive to implement. Considering labor and equipment necessary for other methods discussed, the system used in this study would likely be more effective for improvement of temperament. Direct-Maternal Genetic Correlations Negative direct-maternal genetic correlations of −0.41 (DM) and −0.55 (DMC) for docility are similar to estimates of genetic correlations between direct and maternal effects for weaning weight in beef cattle, which have typically been large and negative (Baker, 1980; Robinson, 1996; Meyer, 1997). Biologically, the correlation for weaning weight has been described as the result of a dam with superior maternal ability overfeeding its daughter, which inhibits proper development of its daughter's udder (i.e., fatty udder syndrome), resulting in below-average maternal ability of the daughter (Baker, 1980). These negative estimates have been attributed to negative dam-offspring covariances, but have also been shown to be due to fixed effects fitted (Bijma, 2006). Maternal models including a linear regression on maternal phenotype (i.e., regression of observed phenotype of an individual on observed phenotype of that individual's dam; Falconer, 1965; Koch, 1972; Robinson, 1996; Koerhuis and Thompson, 1997; Meyer, 1997; Dodenhoff et al., 1999) have been implemented for a better understanding of the negative direct-maternal genetic relationship, whereas others have incorporated parameters accounting for grandmaternal effects (Dodenhoff et al., 1998, 1999). Similar to Koch (1972), who reported estimates of −0.1 to −0.2 for gain of beef calves from birth to weaning using regression on dam phenotype, Meyer (1997) obtained estimates of up to −0.2 for weaning weight in Australian and New Zealand beef breeds. However, Meyer (1997) and Dodenhoff (1999) found that regression models produced similar log likelihoods and breeding values, respectively, to models not including the regression term. Also, Robinson (1996) and Koerhuis and Thompson (1997) found regression models yielded biased estimates when using simulated data. The regression model used in these studies introduces an additional genetic component that is confounded with other genetic and environmental components (Bijma, 2006), resulting in biased estimates. Another approach considered the correlation between maternal permanent environmental effects of 2 adjacent generations (Quintanilla et al., 1999; Iwaisaki et al., 2005) and reported estimates to be approximately −0.2. However, rank correlations of >0.99 between models with and without the correlation suggest the parameter is not relevant or necessary for the genetic evaluation of weaning weight in Limousin or Gelbvieh beef cattle (Iwaisaki et al., 2005). Recently, Bijma (2006) combined the model of Willham (1963) with standard quantitative genetic theory described by Falconer and Mackay (1996; e.g., phenotypic covariances between traits results from summing additive genetic covariances and environmental covariances) to estimate the residual correlation between an individual's record and that of its dam. Bijma (2006) explained that excluding the residual covariance term assumes that either maternal effects are fully heritable or that direct and maternal effects are genetically but not environmentally correlated. It was also suggested that environmental covariances between dam and offspring are likely to be a general phenomenon and should not be restricted to special cases, such as the fatty udder syndrome in beef cattle (Bijma, 2006). Simulations fitting the residual covariance yielded unbiased estimates of genetic (co)variances, and interpretation of results was not challenging (Bijma, 2006). Estimates of direct-maternal genetic correlations may be negative due to sources other than genetic antagonisms. Biased correlation estimates have been observed when sire × year or sire × herd interactions were not incorporated into the model (Robinson, 1996; Lee and Pollak, 1997; Dodenhoff et al., 1999). Using simulated data, Robinson (1996) found that 6% of the phenotypic variation in weaning weight was the result of sire or sire × year variation, and when not accounted for, negative estimates of approximately −0.5 for direct-maternal genetic correlations resulted. When analyzing weaning weight in Simmental cattle, Lee and Pollak (1997) found sire × year interaction represented only 3% of phenotypic variance but explained 62% of the covariance between direct and maternal genetic effects. Dodenhoff et al. (1999) concluded that likelihood values showed sire × herd interaction effects were more important than regression or grandmaternal effects when evaluating weaning weight in Angus cattle. The current study demonstrated that environmental correlation between dam and offspring, if it exists, is not a source of bias in the estimation of the direct-maternal genetic correlation. However, neither sire × year nor sire × herd interactions were considered. Further research should evaluate alternative models (Robinson, 1996; Bijma, 2006), incorporating such components to assess their possible influence on the direct-maternal genetic correlation of docility in Limousin cattle. IMPLICATIONS Docile beef cattle gain more efficiently, harvest more desirable carcasses, and are easier to handle than aggressive cattle. Direct heritability estimates suggest that selection for cattle with more favorable docility scores would be effective in producing cattle with more acceptable temperaments. Although maternal genetic and maternal permanent environmental effects were significant, they explained only small proportions (approximately 8%) of the phenotypic variation, and sires did not rank differently with or without inclusion of these components. Considering these results, addition of maternal effects to the genetic evaluation of docility in Limousin cattle does not seem warranted. Further research should be conducted to assess the strong negative relationship between direct and maternal genetic effects observed in docility. LITERATURE CITED Baker, R. L. 1980. The role of maternal effects on the efficiency of selection in beef cattle: A review. Proc. N. Z. Soc. Anim. Prod.  40: 285– 303. Beef Improvement Federation 2002. Guidelines for Uniform Beef Improvement Programs . BIF, Athens, GA. Bijma, P. 2006. Estimating maternal genetic effects in livestock. J. Anim. Sci.  84: 800– 806. Google Scholar CrossRef Search ADS PubMed  Boivin, X., P. Le Neindre, J. M. Chupin, J. P. Carel, and G. Trillat 1992. Influence of breed and early management on ease of handling and open-field behaviour of cattle. Appl. Anim. Behav. Sci.  20: 259– 273. Boivin, X. P., P. Le Neindre, J. P. Garel, and J. M. Chupin 1994. Influence of breed and rearing management on cattle reactions during human handling. Appl. Anim. Behav. Sci.  39: 115– 122. Google Scholar CrossRef Search ADS   Burrow, H. M. 1997. Measurements of temperament and their relationships with performance traits of beef cattle. Anim. Breed. Abstr.  65: 477– 495. Burrow, H. M. 2001. Variances and covariances between productive and adaptive traits and temperament in a composite breed of tropical beef cattle. Livest. Prod. Sci.  70: 213– 233. Google Scholar CrossRef Search ADS   Burrow, H. M., and R. D. Dillon 1997. Relationships between temperament and growth in a feedlot and commercial carcass traits of Bos indicus crossbreds. Aust. J. Exp. Agric.  37: 407– 411. Google Scholar CrossRef Search ADS   Burrow, H. M., G. W. Seifert, and N. J. Corbet 1988. A new technique for measuring temperament in cattle. Proc. Aust. Soc. Anim. Prod.  17: 154– 157. Busby, W. D., P. Beedle, D. Strohbehn, L. R. Corah, and J. F. Stika 2005. Effects of disposition on feedlot gain and quality grade. J. Anim. Sci.  83( Suppl. 2): 63. (Abstr.) Dodenhoff, J., L. D. Van Vleck, S. D. Kachman, and R. M. Koch 1998. Parameter estimates for direct, maternal, and grandmaternal genetic effects for birth weight and weaning weight in Hereford cattle. J. Anim. Sci.  76: 2521– 2527. Google Scholar CrossRef Search ADS PubMed  Dodenhoff, J., L. D. Van Vleck, and D. E. Wilson 1999. Comparison of models to estimate genetic effects for weaning weight of Angus cattle. J. Anim. Sci.  77: 3176– 3184. Google Scholar CrossRef Search ADS PubMed  Falconer, D. S. 1965. Maternal effects and selection response. Pages 763– 774 in Genetics Today: Proc. XIth Int. Congr. Genet. Vol. 3 . Pergamon Press, Oxford, UK. Falconer, D. S., and T. F. C. Mackay 1996. Introduction to Quantitative Genetics . 4th ed. Addison-Wesley Longman Ltd., Essex, UK. Fell, L. R., I. G. Colditz, K. H. Walker, and D. L. Watson 1999. Associations between temperament, performance and immune function in cattle entering a commercial feedlot. Aust. J. Exp. Agric.  39: 795– 802. Google Scholar CrossRef Search ADS   Gianola, D., and J. L. Foulley 1983. Sire evaluation for ordered categorical data with a threshold model. Genet. Sel. Evol.  15: 201– 224. Google Scholar CrossRef Search ADS PubMed  Gilmour, A. R., B. J. Gogel, B. R. Cullis, S. J. Welham, and R. Thompson 2002. ASReml User Guide Release 1.0 . VSN Int. Ltd, Hemel Hempstead, UK. Golden, B. L., D. J. Garrick, S. Newman, and R. M. Enns 2000. Economically relevant traits a framework for the next generation of EPDs . Pages 2– 13 in Proc. Beef Improv. Fed., Wichita, KS. Grandin, T. 1989. Behavioral principles of livestock handling. Prof. Anim. Sci.  5: 1– 11. Grignard, L., X. Boivin, A. Boissy, and P. Le Neindre 2001. Do beef cattle react consistently to different handling situations? Appl. Anim. Behav. Sci.  71: 263– 276. Google Scholar CrossRef Search ADS   Harville, D. A., and R. W. Mee 1984. A mixed-model procedure for analyzing ordered categorical data. Biometrics  40: 393– 408. Google Scholar CrossRef Search ADS   Hearnshaw, H., and C. A. Morris 1984. Genetic and environmental effects on a temperament score in beef cattle. Aust. J. Agric. Res.  35: 723– 733. Google Scholar CrossRef Search ADS   Iwaisaki, H., S. Tsuruta, I. Misztal, and J. K. Bertrand 2005. Estimation of correlation between maternal permanent environmental effects of related dams in beef cattle. J. Anim. Sci.  83: 537– 542. Google Scholar CrossRef Search ADS PubMed  King, D. A., C. E. Schuehle Pfieffer, R. D. Randel, T. H. Welsh Jr., R. A. Oliphint, B. E. Baird, K. O. Curley Jr., R. C. Vann, D. S. Hale, and J. W. Savell 2006. Influence of animal temperament and stress responsiveness on the carcass quality and beef tenderness of feedlot cattle. Meat Sci.  74: 546– 556. Google Scholar CrossRef Search ADS PubMed  Koch, R. M. 1972. The role of maternal effects in animal breeding. VI. Maternal effects in beef cattle. J. Anim. Sci.  35: 1316– 1323. Google Scholar CrossRef Search ADS PubMed  Koerhuis, A. N. M., and R. Thompson 1997. Models to estimate maternal effects for juvenile body weight in broiler chickens. Genet. Sel. Evol.  29: 225– 249. Google Scholar CrossRef Search ADS   Kuehn, L. A., B. L. Golden, C. R. Comstock, and K. J. Andersen 1998. Docility EPD for Limousin cattle. J. Anim. Sci.  76( Suppl. 1)/ J. Dairy Sci. 81(Suppl.1): 334. (Abstr.) Lee, C., and E. J. Pollak 1997. Relationship between sire × year interactions and direct-maternal genetic correlation for weaning weight of Simmental cattle. J. Anim. Sci.  75: 68– 75. Google Scholar CrossRef Search ADS PubMed  Le Neindre, P., G. Trillat, J. Sapa, F. Ménissier, J. N. Bonnet, and J. M. Chupin 1995. Individual differences in docility in Limousin cattle. J. Anim. Sci.  73: 2249– 2253. Google Scholar CrossRef Search ADS PubMed  Meyer, K. 1997. Estimates of genetic parameters for weaning weight of beef cattle accounting for direct-maternal environmental covariances. Livest. Prod. Sci.  52: 187– 199. Google Scholar CrossRef Search ADS   Müller, R., M. A. G. von Keyserlingk 2005. Consistency of flight speed and its correlation to productivity and to personality in Bos taurus beef cattle. Appl. Anim. Behav. Sci.  99: 193– 204. Google Scholar CrossRef Search ADS   North American Limousin Foundation 2004. Limousin breeders improve temperament. Limousin Breeder's Edge . Vol. 2. NALF, Englewood, CO. Petherick, J. C., R. G. Holroyd, V. J. Doogan, and B. K. Venus 2002. Productivity, carcass and meat quality of lot-fed Bos indicus cross steers grouped according to temperament. Aust. J. Exp. Agric.  42: 389– 398. Google Scholar CrossRef Search ADS   Quintanilla, R., L. Varona, M. R. Pujol, and J. Piedrafita 1999. Maternal animal model with correlation between maternal environmental effects of related dams. J. Anim. Sci.  77: 2904– 2917. Google Scholar CrossRef Search ADS PubMed  Robinson, D. L. 1996. Models which might explain negative correlations between direct and maternal genetic effects. Livest. Prod. Sci.  45: 111– 122. Google Scholar CrossRef Search ADS   Scanga, J. A., K. E. Belk, J. D. Tatum, T. Grandin, and G. C. Smith 1998. Factors contributing to the incidence of dark cutting beef. J. Anim. Sci.  76: 2040– 2047. Google Scholar CrossRef Search ADS PubMed  Tulloh, N. M. 1961. Behaviour of cattle in yards. II. A study of temperament. Anim. Behav.  9: 25– 30. Google Scholar CrossRef Search ADS   Vann, R. C. 2006. Relationships between carcass quality and temperament in beef cattle. Pages 69– 72 in Proc. Beef Improv. Fed., Chocktaw, MS. Available: http://www.beefimprovement.org/pro-ceedings.html. Accessed Sept. 2006. Voisinet, B. D., T. Grandin, S. F. O'Connor, J. D. Tatum, and M. J. Deesing 1997a. Bos indicus cross feedlot cattle with excitable temperaments have tougher meat and a greater incidence of borderline dark cutters. Meat Sci.  46: 367– 377. Google Scholar CrossRef Search ADS   Voisinet, B. D., T. Grandin, J. D. Tatum, S. F. O'Connor, and J. J. Struthers 1997b. Feedlot cattle with calm temperaments have greater average daily gains than cattle with excitable temperaments. J. Anim. Sci.  75: 892– 896. Google Scholar CrossRef Search ADS   Watts, J. M., J. M. Stookey, S. M. Schmutz, and C. S. Waltz 2001. Variability in vocal and behavioural responses to visual isolation between full-sibling families of beef calves. Appl. Anim. Behav. Sci.  70: 255– 273. Google Scholar CrossRef Search ADS PubMed  Willham, R. L. 1963. The covariance between relatives for characters composed of components contributed by related individuals. Biometrics  19: 18– 27. Google Scholar CrossRef Search ADS   Willham, R. L. 1972. The role of maternal effects in animal breeding. III. Biometrical aspects of maternal effects in animals. J. Anim. Sci.  35: 1288– 1293. Google Scholar CrossRef Search ADS PubMed  Footnotes 1 We thank the North American Limousin Foundation for providing the data used in this study. Copyright 2007 Journal of Animal Science
TI - Maternal effects on docility in Limousin cattle
JF - Journal of Animal Science
DO - 10.2527/jas.2006-450
DA - 2007-03-01
UR - https://www.deepdyve.com/lp/oxford-university-press/maternal-effects-on-docility-in-limousin-cattle-oRRrlDSh3L
SP - 650
EP - 657
VL - 85
IS - 3
DP - DeepDyve
ER -