Weaning weight revisited : estimates of genetic parameters fitting a regression on maternal phenotype.

Abstract

Proc. Assoc. Ahmt. Anim. Breed Genet. Voll2 WEANING WEIGHT REVISITED : ESTIMATES OF GENETIC PARAMETERS REGRESSION ON MATERNAL PHENOTYPE K. Meyer Animal Genetics and Breeding Unit, University of New England, Armidale, NSW 2351 PITTING A SUMMARY Restricted maximum likelihood estimates of genetic parameters for weaning weight were obtained fitting a regression on maternal phenotype to account for direct-maternal environmental covariances. For Herefords there was a substantial negative regression on dam's phenotype (up to -0.2), accompanied by small, negative estimates of the direct-maternal genetic covariance. For Angus and Limousin, the direct-maternal genetic covariance was clearly more important than its environmental counterpart, i.e. an estimate of the direct-maternal genetic correlation of about -0.5 could not be attributed to a negative environmental relationship not taken into account. Fitting a sire x herd-year interaction as an additional random effect reduced estimates of the direct-maternal genetic covariance for these breeds, resulting in corresponding correlation estimates of -0.3 to -0.2. Keywords: Beef Cattle, weaning weight, maternal effects, genetic parameters INTRODUCTION Numerous studies of preweaning growth of beef cattle and also sheep, have found an antagonistic correlation between direct and maternal genetic effects, often in the range of -0.5 to -0.7. While a slightly adverse genetic relationship between direct and maternal effects has been considered plausible, such estimates have been met with justified scepticism. Repeatedly this has been attributed to a negative direct-maternal environmental covariance which is expected to bias the estimate of the direct-maternal genetic covariance (CAM) and correlation (e.g. Koch, 1972). Falconer (1965) modeled maternal effects on litter size in mice by fitting a regression on maternal phenotype. This accounted for direct-maternal environmental covariances. This paper reports analyses of weaning weight records for beef cattle, fitting a regression on maternal phenotype in addition to genetic and permanent environmental maternal effects, i.e. an `integrated Falconer-Willham' model (Koerhuis and Thompson, 1996). MATERIAL AND METHODS Data. Data consisted of records for Australian Polled Herefords, Herefords and Limousin, respectively, extracted from the National Beef Recording data base. Basic edits included consistency and range checks for birth and weighing dates and weights. Weights within the range of 80 to 300 days allowed by (GROUP)BREEDPLAN closest to to target weaning age of 200 days were selected. For Herefords and Polled Herefords only calves born from 1980 onwards were considered, and only herds with 200 or more calves with maternal phenotype known were included. In addition, subsets of data for Australian (AUS) and New Zealand (NZ) Angus, analysed previously assuming GA&,= 0 (Meyer, 1995) were obtained, extracting records for herds contributing at least 200 (AUS) and 100 (NZ) weights with dams' records available. Characteristics of the data structure are summarised in Table 1. Analyses Analyses were carried out by Restricted Maximum Likelihood (RBML) using a derivativefree algorithm. For Limousins all pedigree information available was used while only one (Herefords) 479 Proc. Assoc. Advmt. Anim. Breed. Genet. Voll2 Table 1. Characteristics of the data structure for weaning weight Polled H. No. of . . . with No. of No. of No. of No. of No. of Weight records dam's rec. animalsa siresb damsb contemp. groups SxHYC (kg) xl sde Age (days) x sd Dam age (years) X sd 61,787 23,234 84,520 2089 23,967 6224 7337 221.9 50.4 218.2 42.0 5.33 2.30 Hereford 79,434 33,426 107,154 2689 30,866 6551 7837 219.0 55.7 214.1 39.1 4.91 2.38 Angus AUS 87,389 46,274 109,841 2589 31,272 6903 7719 232.9 47.3 214.6 32.6 4.88 2.29 Angus NZ 57,375 29,319 69,817 1819 19,456 2553 3776 216.6 50.5 201.8 39.8 5.41 2.57 Limousin 16,635 3926 28,236 1086 8502 4304 3291 232.3 49.4 215.4 40.1 'in the analysis, including parents without records hwith progeny in the data csire x herd-year interaction effects dmean <standard deviation or two (Angus) passes through the pedigree were performed to locate parental identities data sets to restrict the total number of animals in the analysis. for the other Random efsects. An animal model including maternal genetic as well as permanent environmental effects in addition to animals' direct additive genetic effects was fitted throughout. Ignoring maternal phenotype, analyses were carried out under Models 5 (assuming c&t,$f 0; M5) and 6 (allowing o,`tAM = # 0; M6) of previous analyses (e.g. Meyer, 1993). Furthermore, Model 7 (M7) was like M6 but fitted a sire x herd-year interaction with assumed covariance matrix $1 as group-sex subclasses, with an 'age calves born not more than 45 days in age at weighing were taken into was fitted as a linear and quadratic vs. cows : older herd-year-management slicing' of 45 days, i.e. subclasses were divided further, so that only apart were directly compared with each other. Additional differences account by fitting a linear regression on age within sex. Age of dam covariable. Age status of dam (heifers : 28 month or less at calving RESULTS AND DISCUSSION Estimates of genetic parameters and corresponding error and phenotypic variances are given in Table 480 Proc. Assoc. Advmt. Anim. Breed. Genet. Vol12 2. The most parameterised model (M7p) fitted best in all cases, but there were distinct differences the relative importance of genetic and environmental correlations between dams and offspring. in Herefords. Results for Polled Herefords and Herefords followed the same pattern. Estimates of /!l were negative, -0.16 to -0.20. Koch (1972) speculated that its magnitude was -0.1 to -0.2 for preweaning growth of beef calves. Ignoring p but allowing for a non-zero GA,+,(M6) resulted in substantial, negative estimates for DAM. Likelihoods for M5p were markedly higher than for M6, indicating that the negative dam-offspring covariance was environmental rather than genetic. In line with this, estimates of p under M6p were only slightly reduced (in absolute value) compared to those under M5P, while estimates of HAM though negative, amounted to only 5 to 8% of the phenotypic variance. Allowing for a direct-maternal covariance (either kind), increased estimates of both the direct (h2) and maternal (m*) heritability (M5p, M6 and M6p vs. M5), while fitting a regression on maternal phenotype tended to reduce the estimate of the maternal environmental variance. Including a sire x herd-year `interaction' (M7 and M7p) gave estimates of cri amounting to about 4% of c$ and substantially increased likelihoods, while reducing estimates of h2, m2 and GA,+,(absolute value) compared to M6 and M6fl. In contrast, estimates of the permanent environmental maternal effects (c2) and p were virtually unchanged. Angus and Limousin. Maternal phenotypes proved to be considerably less important for these breeds, with M6 fitting the data better than M5P. Again, there was some cross-substitution between parameters, i.e fitting p and not CA,+,(M5p) resulted in sizable, negative estimates for p, while fitting p over and above HAM (M6p vs. M6) gave small, negative estimates for p (-0.03 to -0.04). Augmenting the model of analysis by a sire x herd-year effect resulted in a dramatic increase in log L for these data sets, substantially more than due to allowing for a non-zero GA,&, p. Estimates of 0; were 4% to 9 % or of 0;. Considering 5 Angus (AUS) herds, Robinson (1996) obtained estimates of c$, of 11% ( of c$), and of h*, m*and rAM of 0.1 I, 0.25 and 0.014, respectively, compared to estimates of 0.29 (h*), 0.14 (m*) and -0.52 (TAM)under M6. CONCLUSIONS Results identified clear breed differences in maternal effects. For Herefords, there appears to be a substantial direct-maternal environmental covariance, so that fitting a regression on maternal phenotype alleviated the problem of inexplicably large (absolute value), negative estimates of OAAM a large to extent. Fitting a sire x herd-year interaction as an additional effect dramatically increase log L in all cases. This was accompanied by reduced estimates of GAAM (absolute value), h2 and m*, suggesting that inflated values of HAM might have been caused by unaccounted sources of variation, such as paddocks or management groups, particularly if confounded with paternal half-sib groups. REFERENCES Falconer, D.S (1965) Genetics Today, S.J. Geerts (ed.), Pergamon Press, New York. Koch, R.M. (1972)J. Anim. Sci. 35: 1316. Koerhuis, A.N.M. and Thompson, R. (1996) Genet. Select. Evol. : (submitted). Meyer, K. (1993)Anim. Prod. 57 : 37. Meyer, K. (1995) Aust. .I Agric. Res. 46 : 1219. Meyer, K. (1996) Livest. Prod. Sci. : (submitted). Robinson, D.L. (1996) Livest. Prod. Sci. 45 : 1. 481 Proc. Assoc. Advmt. Anim. Breed. Genet. Voll2 Table 2. Estimates of genetic parametersa Model 5 5p 6 6p 7 7p 5 5p 6 6p 7 7p 5 5p 6 6p 7 7p 5 5p 6 6p 7 7p 5 5p 6 6p 7 7p -0.171 -0.157 -0.155 P h2 0.155 0.192 0.251 0.240 0.163 0.162 0.165 0.228 0.283 0.272 0.176 0.172 0.239 0.272 0.387 0.384 0.220 0.222 0.180 0.207 0.256 0.256 0.144 0.147 0.252 0.286 0.400 0.392 0.225 0.224 m2 0.112 0.142 0.225 0.193 0.190 0.168 0.105 0.138 0.210 0.168 0.170 0.138 0.092 0.102 0.181 0.171 0.130 0.123 0.082 0.095 0.145 0.138 0.111 0.109 0.117 0.131 0.264 0.231 0.201 0.172 -0.153 -0.077 -0.102 -0.038 CAM TAM C2 s2 Polled Herefords 0.266 -0.642 0.218 0.263 0, 2 0, 2 log L 327.6 270.2 291.4 262.7 311.9 281.9 358.0 279.6 313.0 268.8 338.2 294.3 272.5 250.4 232.8 230.5 262.2 258.6 333.9 315.1 311.1 308.5 335.7 332.5 290.7 243.3 241.4 226.9 280.2 262.3 701.8 736.8 703.7 728.5 701.1 725.8 718.2 772.8 720.0 767.6 717.2 763.8 523.1 543.2 525.5 529.9 523.7 528.7 561.9 577.6 563.2 567.2 560.3 564.6 615.5 635.4 624.1 632.5 632.6 642.7 -360.41 -196.22 -247.43 -155.85 -92.89 0 -564.66 -268.39 -419.95 -249.04 -182.79 0 -673.31 -598.30 -496.52 -490.18 -7.83 0 -168.48 -139.51 -114.25 -111.57 -3.23 0 -74.53 -57.57 -47.72 -43.19 -5.07 0 -0.195 -0.186 , -0.188 -0.089 -0.035 -0.037 -0.076 -0.032 -0.033 -0.357 0.223 -0.582 0.265 0.040 -0.230 0.224 0.038 Herefords 0.232 0.170 -0.159 -0.650 0.23 1 -0.051 -0.241 0.173 -0.098 -0.564 0.233 0.045 -0.003 -0.022 0.175 0.045 Angus - Australia 0.148 0.125 -0.165 -0.625 0.154 -0.144 -0.562 0.146 -0.077 -0.456 0.156 0.071 -0.059 -0.359 0.147 0.070 Angus - New Zealand 0.144 0.125 -0.100 -0.519 0.147 -0.084 -0.042 -0.029 -0.445 0.139 -0.329 0.150 -0.230 0.140 Limousin 0.159 0.132 -0.647 0.160 -0.509 0.141 -0.532 0.159 -0.304 0.137 0.038 0.037 -0.130 -0.093 -0.099 -0.210 -0.153 -0.113 -0.060 0.086 0.085 'p : regression on maternal phenotype, h*: awe as proportion of the phenotypic variance, variance, s*: variance due to sire x year effects variance, and log L: log likelihood, as deviation direct heritability, m*: maternal heritability, HAM:direct-maternal genetic covaric*: permanent environmental maternal variance as proportion of the phenotypic as proportion of the phenotypic variance, 0;: residual variance, 0:: phenotypic from model 7p 482