Estimation of Variance Components for Lamb Weights

Estimation of variance components for lamb weights in three sheep populations J. J. Tosh and R. A. Kemp J ANIM SCI 1994, 72:1184-1190.

The online version of this article, along with updated information and services, is located on the World Wide Web at: http://jas.fass.org/content/72/5/1184

www.asas.org

Downloaded from jas.fass.org by guest on July 12, 2011

Estimation of Variance Components for Lamb Weights in Three Sheep Populations1 J. J. Tosh* and R. A. Kemp*,t "Centre for Genetic Improvement of Livestock, Department of Animal and Poultry Science, University of Guelph, Guelph, Ontario, Canada N1G 2W1 and tLivestock Technology, Ontario Ministry of Agriculture and Food, Guelph, Ontario, Canada N1G 3N6

ABSTRACT: Variance components were estimated for lamb weight at birth, 50 d, and 100 d of age. Data from the Canadian flock recording program for lambs born in 1977 to 1991 for Hampshires ( n = 6,395) and Polled Dorsets ( n = 29,204) and 1982 to 1991 for Romanovs ( n = 3,432) were studied. Observed weights were pre-adjusted for the effects of age of dam, sex of lamb, birth-rearing type, month or quarter of year of birth, parity-lambing interval, and age of dam at first lambing, using estimates derived from a fixed effects model including contemporary groups plus these factors. Pre-adjusting for nuisance variables reduced the number of equations in the model for variance component estimation. A single-trait animal model with derivative-free restricted maximum-likelihood procedures was used. Random effects were additive direct and maternal genetic, litter (common environmental), and error. An alternate model excluded maternal genetic effects. Estimates of litter

variance as a proportion of phenotypic variance were of moderate size (.12 to .43) and consistent across breeds and models. The mean correlation between direct and maternal genetic effects, across traits and breeds, weighted by the number of animals, was -.40 (SE = .15). The maternal genetic variance or directmaternal genetic covariance component, or both, was different from zero ( P < .05) for all traits in Hampshires and Polled Dorsets, suggesting that maternal effects were important for weight of lambs even a t 100 d of age. Estimates of direct heritability ranged from .05 to .45, varying across traits, breeds, and models. In Romanovs, with the complex model, no estimate of direct or maternal heritability or directmaternal genetic correlation was different from zero ( P > .lo), which emphasizes differences in these variance components across the breeds and has implications for genetic evaluation programs.

Key Words: Birth Weight, Genetic Parameters, Heritability, Lamb Weight, Maternal Effects, Sheep ~~~~~~~

~

~

J. Anim. Sci. 1994. 72:1184-1190

Introduction Profitability of sheep production for meat depends to a great extent on lamb weight, so selection objectives concentrate on this trait. Genetic parameters are needed to estimate breeding values and compare responses from different selection programs. Estimates of parameters for early weight traits in other species indicate values to expect in sheep, which is helpful because very few studies with sheep have used large sets of field data or considered maternal effects. Consequently, parameter estimates

for sheep are not very reliable and a model appropriate for explaining variation in lamb weight has not been established. Work on large data sets has been for only a small sample of sheep breeds, such as Suffolk (Shrestha et al., 1985) and Dorset (Shrestha et al., 1986), or for breeds not commonly raised in North America (Eikje, 1974). The objective of this work was to estimate variance components for three weight traits in lambs from three breeds of diverse biological types using an animal model with direct and maternal genetic effects as well as litter effects.

Materials and Methods 'Appreciation is extended to H. Song of Agriculture Canada for providing the data and for helpful discussion. The authors gratefully acknowledge K. Meyer for providing the DFREML programs. This research was financially supported by the Red Meat I1 Program of the Ontario Ministry of Agriculture and Food. Received September 24, 1993. Accepted January 8, 1994.

Data. Data obtained from the Canadian Sheep Genetic Evaluation Program contained lamb weights at birth (BWT), 50 d (50WT), and 100 d ( 1OOWT). Observed weights had been previously adjusted for age to 50 and 100 d (Ontario Ministry of Agriculture

1184 Downloaded from jas.fass.org by guest on July 12, 2011

VARIANCE COMPONENTS FOR LAMB WEIGHTS

1185

Table 1. Number of records, animals, contemporary groups, and litters represented in the data for weight at birth (BWT), 50 days (50WT),and 100 days ( 1 O O W T ) for the three breeds

Breed and trait Hampshire BWT 50WT

loow Polled Dorset BWT 50WT l0OWT Romanov BWT

50WT

loow

Records

Animals

Contemporary groups

Litters

2,659 6,395 5,601

3,832 8,883 7,916

237 578 530

1,862 4,539 4,046

13,977 29,204 25,052

19,020 39,103 33,594

913 1,896 1,685

9,628 20,356 17,675

2,436 3,432 2,968

2,987 4,127 3,623

200 266 233

1,090 1,538 1,388

and Food, 1987). Several edits had also been done, primarily for obvious errors, small contemporary groups, and lack of genetic connections to the national base of animals, to conduct national genetic evaluations. Subsequently, extreme weights (more than four standard deviations from the mean) and contemporary groups with only one 50WT record were eliminated. Pedigree information was retained during editing, for use in an animal model. From over 30 breeds, lamb records of Hampshire, Polled Dorset, and Romanov sheep were chosen for use in this study. Table 1 shows the number of observations for the three breeds. These breeds had large numbers of observations and represented diverse biological types. Hampshires are a meat breed, Polled Dorsets are a dual meat and maternal breed, and Romanovs are a prolific breed. Table 2 illustrates the lamb weights observed in the three breeds. For Hampshires and Polled Dorsets, the data included lambings from 1977 to 1991; for Romanovs, data included lambings from 1982 to 1991. Adjustment for Nuisance Variables. To simplify the model for variance component estimation, the data were pre-adjusted for nuisance variables estimated using a fxed effects model and general linear model procedures. Analyses were conducted separately for each weight trait considering the same model in each case. The model included an overall mean and the fixed effects of contemporary group (i.e., management group within flock-year), age of dam (1, 2, 3, . . ., or 10+ yr), sex of lamb (female, male, or wether), birthrearing type (nine classes where the number of lambs born was 1, 2, or 3+ and the number of lambs raised was 1, 2, or 3+; plus one class of bottle-fed lambs), month or quarter of year of lambing (12 or 4 classes), parity-lambing interval (first parity of dam, or later parities following a lambing interval of 6 to 9, 10 t o 14, or 2 15 mo), and age of dam at first lambing ( I 13, 14 to 17, 18 to 26, or 2 27 mo). Age of dam at first lambing was determined at the first recorded lambing,

which may differ from actual first lambing, especially for ewes in the oldest class. One of the parity-lambing interval classes represented first-parity effects, and the other classes represented the combined effects of a later parity and lambing interval, assuming all later parities had the same effect. Hampshire ewes rarely lambed during summer months, and those records tended to be nested within contemporary group; therefore, quarter rather than month of year of lambing was used for that breed. Because Romanovs were recently introduced into Canada, there were few old dams, so those aged 6 yr or more were pooled. Estimates of contrasts of the fixed effects were used as adjustment factors. Observed weights were adjusted for all the fixed factors, except contemporary group, to the basis of a lamb born to a 5-yr-old dam, that was male, born and raised as a single, born in March ( o r the first quarter of the year) after a lambing interval of 1 yr, and whose dam first lambed at 1.5 to 2 yr of age. Although pre-adjusting for

Table 2. Means, standard deviations, and coefficients of variation for weight (kg)at birth (BWT), 50 days (50WT), and 100 days (100WT) for the three breeds Breed and trait

Mean

SD

cv

4.54 20.35 36.85

1.23 5.61 9.74

.27 .28 .26

4.06 17.59 30.20

.93 4.70 7.83

.23 .27 .26

2.81 12.86 24.72

.69 3.31 6.06

.25 .26 .25

Hampshire

BWT 50WT lOOWT Polled Dorset BWT 50WT lOOWT Romanov BWT 50WT

loow

Downloaded from jas.fass.org by guest on July 12, 2011

1186

TOSH AND KEMP

nuisance variables with estimates from a fixed effects model is less accurate than adjusting while simultaneously estimating variance components with a more complex mixed model, standard errors of variance components should be minimally affected. Estimates of variance components have large sampling variances, which are often only approximated. Estimation of Variance Components. Variance components were estimated by a single-trait animal model with derivative-free restricted maximum-likelihood (DFREML) procedures (Graser et al., 1987). The computer programs of Meyer (1989, 1991) were used. The basic linear model was y = Xb

+

Zdad

ad am C

e where A is the additive genetic relationship matrix, I, is an identity matrix of order equal to the number of litters, and I, is an identity matrix of order equal to the number of records. Let denote the total, phenotypic variance. The log-likelihood function was maximized with respect t o direct heritability ( h i = maternal heritability


. l o ) in any other case tested. Assuming that the three breeds and three traits represent a sample of all breeds and lamb weights, a general estimate of the direct-maternal genetic correlation was derived from the individual estimates with

standard errors. The mean ?d,, weighted by the number of animals involved, was -.40 with standard error of .15, which differs from zero ( P < .01). Maria et al. (1993) found extreme genetic correlations ( I -.97) for lamb weights in Romanovs, probably with large standard errors. Few studies have investigated this correlation in sheep. In beef cattle, reviews of the literature (Baker, 1980; Koots et al., 1991; Meyer, 1992) report average correlations between -.72 to -.16 for weights at birth or weaning. In swine, Ferraz and Johnson ( 1993 estimated the genetic correlation t o be -.34 for average daily gain. Antagonism between the effects of an individual's genes for growth and those of its dam for a maternal contribution may be due to natural selection for an intermediate optimum. Estimates of maternal heritability were low to moderate in size (Table 3 ) with standard errors of .04 to .lo. This component was greater than zero ( P < .05) in all cases tested except for traits in the Romanovs. Like c2, within breed, estimates of h i tended to decline from birth to 50 d to 100 d. Maternal genetic effects expressed during gestation and lactation were expected to have a diminishing influence on weight as lambs became older. In Romanovs, Maria et al. (1993) obtained hk estimates of .22 for BWT and .25 for weight at 40 d, which are higher than those found in this study for Romanovs but similar to those of the other breeds. Maria et al. (1993) also reported a h i value of essentially zero for weight at 90 d. Direct heritabilities fluctuated across traits, and especially across breeds, taking a wide range of values (Table 3 ) . The standard errors were .06 to .15, and, of those tested, only the estimates for Hampshires

Downloaded from jas.fass.org by guest on July 12, 2011

1188

TOSH AND KEMP

differed from zero ( P < .05). Heritability estimates of lamb weight also varied substantially in other studies: Wolf et al. (1981) found values of -.02 to .06 for weights at birth to 12 wk, which were believed to be low due to prior selection; Eikje (1974) reported values of .05 to .18 for weights at ages from 20 to 170 d; Maria et al. (19931, using an animal model, obtained values of .04 to .34 for weights at birth to 90 d; Martin et al. (1980) found values of .05 to .41 for weights at birth to 16 wk, which is practically the same range as those in Table 3; Razungles et al. (1985) obtained values of .14 to .34 for weights at birth, 90 d, and four ages in between; Olson et al. (1976) found values of .18 to .35 for weights of ram lambs at birth to 18 wk; Dzakuma et al. (1978) obtained values ranging from -.09 to .50 for weights at birth and 70 d; and for 50WT and 100WT, Shrestha et al. (1985) found values of .24 to .53 in Suffolks, whereas Shrestha et al. (1986) reported values of .14 to .79 in Dorsets. The h: estimates presented from this study are in the middle of the range of those from the literature. Estimates of heritability of the total additive genetic effects, both direct and maternal, were less variable than estimates of either the direct or maternal heritability (Table 3 ) . The standard errors were large (.14 to .31) because of large sampling variances for the direct-maternal genetic correlation. Consequently, none of the seven estimates of h$ that could be tested was different from zero ( P > . l o ) . For the Romanovs, none of the parameter estimates except those for c2 were different from zero ( P > ,101. Because maternal genetic and litter effects, the latter including maternal permanent environmental effects, were associated with the biological dam, data from cross-fostered or bottle-fed lambs hindered estimation of these components for 50WT and 1OOWT. Romanov lambs were cross-fostered or bottle-fed more frequently than were lambs of the other two breeds (6.3 vs .01). Maternal genetic effects were important to lamb weight at 100 d in the Hampshires and Polled Dorsets, a point beyond the usual weaning age. This is not surprising considering that weight at 100 d has a part-whole relationship with weight a t 50 d and at birth, and carryover effects of the ewe’s maternal genotype can persist into postweaning periods, like the maternal environmental effects of age of dam and birth-rearing type do (Olson et al., 1976; Atkins, 1986). Further, a preweaning handicap due t o low levels of milk production reflecting maternal genotype may result in compensatory growth following weaning, as happens when milk is limited in young dams or large birth-rearing groups (Ch’ang and Rae, 1970). Therefore, maternal genetic effects could be necessary in models that describe weights of lambs of diverse breeds. Estimates of direct heritability and litter variance as a proportion of the phenotypic variance obtained with a model that excluded maternal genetic effects are shown in Table 4 and can be compared to those from the complex model. The estimates of c2 were very similar across models but slightly higher with the model lacking maternal genetic effects, indicating that

Downloaded from jas.fass.org by guest on July 12, 2011

VARIANCE COMPONENTS FOR LAMB WEIGHTS

Table 4. Estimates of phenotypic variance (u:), heritability of direct effects (hj), and litter variance as a proportion of phenotypic variance (c2) from analyses of weight (kg) at birth (BWT), 50 days (50WT), and 100 days (100WT) for the three breeds when the model excluded maternal genetic effects

2P

h:

C2

.6 15.1 39.5

.36"** ,21**" 28'""

43*** .30*** .23***

loow

.5 9.9 25.5

,45*** 35*** 30***

,22"** .21**"

Romanov BWT 50WT lOOWT

.3 6.0 17.1

.26** .lo* 22***

.35*** .23*** 13**:$

Breed and trait Hampshire BWT 50WT lOOWT Polled Dorset BWT 50WT

36***

*P < .05. **P < .01. ***P < ,001.

some of those effects were then attributed to the maternal environment (litter). Conversely, estimates of h: were quite different under the two models. Larger h i values without maternal genetic effects in the model indicate that some of those maternal genetic effects were then attributed t o the direct genetic component, whereas smaller values indicate that some of those effects were then attributed t o random error. Variance components from the simpler model are misleading and should not be used for predicting response to selection or estimating breeding values. When maternal genetic effects are neglected, heritability of the total additive genetic effects is equal to direct heritability. Estimates of h$ thereby obtained from the simpler model (Table 4 ) were consistently larger than those from the complex model (Table 3 ) . The simple model overestimated additive genetic variation and potential for response to selection primarily by ignoring the negative correlation between direct and maternal genetic effects.

Implications Heritability estimates were quite variable, suggesting that parameters specific to the lamb weight trait and breed are necessary when estimating breeding values and predicting response to selection. Models that describe weight of lambs of diverse biological types should include maternal genetic effects even at 100 d. The substantial negative correlation between direct and maternal genetic effects, which can be

1189

explained biologically, indicates that improvements in one of these parts hinders performance in the other and slows the overall rate of progress. Differences between breeds such as the low proportions of genetic variation, both direct and maternal, that were observed in Romanov lambs imply that heterogeneous variances need to be used for across-breed genetic evaluation. Litter variance, due to environmental effects common to littermates, accounts for a major portion of the variation in weight of lambs.

Literature Cited Atkins, K. D. 1986. A genetic analysis of the components of lifetime productivity in Scottish Blackface sheep. Anim. Prod. 43:405. 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. Chang, T. S., and A. L. Rae. 1970. The genetic basis of growth, reproduction, and maternal environment in Romney ewes. I. Genetic variation in hogget characters and fertility of the ewe. Aust. J . Agric. Res. 21:115. Dzakuma, J. M., M. K. Nielsen, and T. H. Doane. 1978. Genetic and phenotypic parameter estimates for growth and wool traits in Hampshire sheep. J. Anim. Sci. 47:1014. Eikje, E. D. 1974. Studies on sheep production records. IV.Genetic, phenotypic, and environmental parameters for weight of lambs. Acta Agric. Scand. 24:291. Fahmy, M. H. 1989. Reproductive performance, growth, and wool production of Romanov sheep in Canada. Small Ruminant Res. 2:253. Ferraz, J.B.S., and R. K. Johnson. 1993. Animal model estimation of genetic parameters and response to selection for litter size and weight, growth, and backfat in closed seedstock populations of Large White and Landrace swine. J. Anim. Sci. 71:850. Graser, H.-U., S. P. Smith, and B. Tier. 1987. A derivative-free approach for estimating variance components in animal models by restricted maximum likellhood. J. Anim. Sci. 64:1362. Koots, K. R., J . P. Gibson, and J. W. Wilton. 1991. Benchmark heritabilities and genetic correlations. Beef Research Update, Ontario Agric. College Publ., Univ. of Guelph. p 7. Maria, G. A,, K. G . Boldman, and L. D. Van Vleck. 1993. Estimates of variances due t o direct and maternal effects for growth traits of Romanov sheep. J. Anim. Sci. 71:845. Martin, T. G., D. I. Sales, C. Smith, and D. Nicholson. 1980. Phenotypic and genetic parameters for lamb weights in a synthetic line of sheep. h i m . Prod. 30:261. Meyer, K. 1989. Restricted maximum likelihood to estimate variance components for animal models with several random effects using a derivative-free algorithm. Genet. Sel. Evol. 21:317. Meyer, K. 1991. DFREML-Programs to estimate variance components by restricted maximum likelihood using a derivative-free algorithm. User Notes, Version 2.0. Univ. of New England, Armidale (Mimeo). Meyer, K. 1992. Variance components due to direct and maternal effects for growth traits of Australian beef cattle. Livest. Prod. Sci. 31:179. Olson, L. W., G. E. Dickerson, and H. A. Glimp. 1976. Selection criteria for intensive market lamb production: Growth traits. J. h i m . Sci. 43:78. Ontario Ministry of Agriculture and Food. 1987. Ontario Sheep Record of Performance Program Breeder's Handbook. Animal Industry Branch, Ontario Ministry of Agriculture and Food, Guelph. Razungles, J., L. Tchamitchian, B. Bibe, C. Lefevre, J. C. Brunel,

Downloaded from jas.fass.org by guest on July 12, 2011

1190

TOSH AND KEMP

and G. Ricordeau. 1985. The performance of Romanov crosses and their merits as a basis for selection. In: R. B. Land and D. W. Robinson (Ed.) Genetics of Reproduction in Sheep. p 39. Buttenvorths, London. Shrestha, J.N.B., J. A. Vesely, and J. P. Chesnais. 1985. Genetic and phenotypic parameters for daily gain and body weights in Suffolk lambs. Can. J. h i m . Sci. 65:575. Shrestha, J.N.B., J . A. Vesely, J. P. Chesnais, and D. Cuthbertson.

1986. Genetic and phenotypic parameters for daily gain and body weights in Dorset lambs. Can. J. Anim. Sci. 66:289. Willham, R. L. 1963. The covariance between relatives for characters composed of components contributed by related individuals. Biometrics 19:18. Wolf, B. T., C. Smith, J.W.B. King, and D. Nicholson. 1981. Genetic parameters of growth and carcass composition in crossbred lambs. Anim. Prod. 32:l.

Downloaded from jas.fass.org by guest on July 12, 2011

Citations

This article has been cited by 3 HighWire-hosted articles: http://jas.fass.org/content/72/5/1184#otherarticles

Downloaded from jas.fass.org by guest on July 12, 2011