Multiple-trait Restricted Maximum Likelihood Research Articles

Genetic parameters for milkability from the first three lactations in Fleckvieh cows

Test-day records for average flow rate (AFR) from the routine dairy recording from Bavarian Fleckvieh cows were analysed. Two data sets with observations on approximately 20 000 cows each were sampled from the total data set. For the estimation of variance parameters, a two-step approach was applied. In a first step multiple-trait restricted maximum likelihood (REML) analyses were carried out. For each of the first three lactations, six time periods with up to 33 days were defined. An algorithm for iterative summing of expanded part matrices was applied in order to combine the estimates. In a second step covariance functions (CF) for additive-genetic variances and non-genetic animal variances were derived using second-order Legendre polynomials plus an exponential term. Estimates of test-day heritability for AFR ranged from 0.21 to 0.40, and were largest in lactation 1. For lactations 1 and 3, heritabilities decreased considerably towards the end of lactation. Genetic correlation estimates within lactation decreased as the distance between days in milk (DIM) increased. Genetic correlations between corresponding DIM in the three lactations were generally large, ranging from 0.80 to 0.99. The largest estimates were found between DIM from lactations 2 and 3. Results from this study suggest that including AFR data from second and third lactations in genetic evaluation systems could the improve accuracy of genetic selection.

Use of Ultrasound to Identify Brangus Cattle with Superior Intramuscular Fat and Other Carcass Traits

The objectives of this research were 1) to determine the effects of sire on the expression of intramuscular fat and 2) to determine the relationship of intramuscular fat with selected performance traits. Purebred Brangus cattle (n = 1,215) were subjected to real-time ultrasound evaluation of body composition; traits recorded were the 12th-13th rib subcutaneous fat thickness (FT), longissimus area (LA), and percentage intramuscular fat (IMF). Performance data provided by the International Brangus Breeder’s Association (IBBA) included birth BW (BBW; n = 323), weaning BW (WBW; n = 328), yearling BW (YBW; n = 1,087), and scrotal circumference (SC; n = 329). Data were analyzed to determine genetic relationships with the animal model using multiple trait restricted maximum likelihood (MTDFREML) procedures. Contemporary group was included in the model as a fixed effect, and age (in days) was included as a covariate. Analyses of covariant models were examined to determine the relative importance of sire and carcass traits to IMF. Genetic correlations of IMF with LA, FT, and YBW were –0.25, 0.36, and 0.31, respectively. Sire, FT, and LA were significant (P<0.05) sources of variation in IMF; whereas BBW, WBW, YBW, SC, and age did not significantly influence IMF. These results suggest that IMF estimated by real-time ultrasound is moderately heritable. Because IMF is an important factor in determining quality grade (QG), ultrasonically determined IMF may be considered in selecting seedstock for increased QG.

Estimates of genetic parameters and genotype by environment interactions for growth traits of rainbow trout ( Oncorhynchusmykiss) as inferred using molecular pedigrees

Genetic correlations between three character sets (early progeny growth; later progeny growth at 8.5 °C; later progeny growth at 15 °C) were determined from within the scope of a working commercial rainbow trout ( Oncorhynchus mykiss) aquaculture facility. Microsatellite multiplex genotyping systems were used to resolve the molecular pedigrees for groups of progeny with high accuracy (91–95%). The heritability estimates obtained from restricted maximum likelihood (REML) were higher than expected, ranging from 0.360 to 0.719 for the growth traits total length, weight and condition factor. Estimates of genetic correlations were obtained through correlation of single trait estimated breeding values (EBV), correlation of multiple trait EBV, and by estimates of genetic correlations coming directly from multiple trait REML. REML determined genetic correlations were high (0.86±0.026), indicating that early progeny growth is a good predictor of later progeny growth and that the magnitude of genotype by environment interactions (G×E) is low. Genetic correlations estimated from multiple trait EBV (0.84±0.048) were not significantly different from the REML estimates. Both were significantly different from the genetic correlations estimated from single trait EBV (0.60±0.049), suggesting that single trait estimates have limited value. Interestingly, the genetic correlations between condition factor and length or weight of all character sets were significantly different from unity, indicating that condition factor is a trait that is genetically independent of both length and weight.

Genetic correlation estimates between ultrasound measurements on yearling bulls and carcass measurements on finished steers.

Carcass and growth measurements of finished crossbred steers (n = 843) and yearling ultrasound and growth measurements of purebred bulls (n = 5,654) of 11 breeds were analyzed to estimate genetic parameters. Multiple-trait restricted maximum likelihood (REML) was used to estimate heritabilities and genetic correlations between finished steer carcass measurements and yearling bull ultrasound measurements. Separate analyses were conducted to examine the effect of adjustment to three different end points: age, backfat thickness, and weight at measurement. Age-constant heritability estimates from finished steer measurements of hot carcass weight, carcass longissimus muscle area, carcass marbling score, carcass backfat, and average daily feedlot gain were 0.47, 0.45, 0.35, 0.41, and 0.30, respectively. Age-constant heritability estimates from yearling bull measurements of ultrasound longissimus muscle area, ultrasound percentage of intramuscular fat, ultrasound backfat, and average daily postweaning gain were 0.48, 0.23, 0.52, and 0.46, respectively. Similar estimates were found for backfat and weight-constant traits. Age-constant genetic correlation estimates between steer carcass longissimus muscle area and bull ultrasound longissimus muscle area, steer carcass backfat and bull ultrasound backfat, steer carcass marbling and bull ultrasound intramuscular fat, and steer average daily gain and bull average daily gain were 0.66, 0.88, 0.80, and 0.72, respectively. The strong, positive genetic correlation estimates between bull ultrasound measurements and corresponding steer carcass measurements suggest that genetic improvement for steer carcass traits can be achieved by using yearling bull ultrasound measurements as selection criteria.

Characteristics of Line 1 Hereford females resulting from selection by independent culling levels for below-average birth weight and high yearling weight or by mass selection for high yearling weight.

Simultaneous selection for low birth weight and high yearling weight has been advocated to improve efficiency of beef production. Two sublines of Line 1 Hereford cattle were established by selection either for below-average birth weight and high yearling weight (YB) or for high yearling weight alone (YW). Direct effects on birth weight and yearling weight diverged between sublines with approximately four generations of selection. The objective of this study was to estimate genetic trends for traits of the cows. A three-parameter growth curve [Wt = A(1 - b0e(-kt))] was fitted to age (t, d)-weight (W, kg) data for cows surviving past 4.5 yr of age (n = 738). The resulting parameter estimates were analyzed simultaneously with birth weight and yearling weight using multiple-trait restricted maximum likelihood methods. To estimate maternal additive effects on calf gain from birth to weaning (MILK) the two-trait model previously used to analyze birth weight and yearling weight was transformed to the equivalent three-trait model with birth weight, gain from birth to weaning, and gain from weaning to yearling as dependent variables. Heritability estimates were 0.32, 0.27, 0.10, and 0.20 for A, b0, k, and MILK, respectively. Genetic correlations with direct effects on birth weight were 0.34, -0.11, and 0.55 and with direct effects on yearling weight were 0.65, -0.17, and 0.11 for A, b0, and k, respectively. Genetic trends for YB and YW, respectively, were as follows: A (kg/generation), 8.0+/-0.2 and 10.1+/-0.2; b0 (x 1,000), -1.34+/-0.07 and -1.16+/-0.07; k (x 1,000), -14.3+/-0.1 and 4.3+/-0.1; and MILK (kg), 1.25+/-0.05 and 1.89+/-0.05. Beef cows resulting from simultaneous selection for below-average birth weight and increased yearling weight had different growth curves and reduced genetic trend in maternal gain from birth to weaning relative to cows resulting from selection for increased yearling weight.

Approximation of Estimates of (Co)variance Components with Multiple-Trait Restricted Maximum Likelihood by Multiple Diagonalization for More Than One Random Effect

Canonical transformation for REML can be applied to models with several random effects by simultaneously diagonalizing (co)variance matrices for all random effects. This procedure is an approximation when matrices cannot be diagonalized completely. The level of the approximation was studied with simulated data by comparing multiple diagonalization and exact REML estimates. For a range of diagonalization levels, the error of the estimates was 2 to 10 times lower than the fraction of non-diagonalizable (co)variances because 57 to 91% of these variances were recovered after the backtransformation step. To determine whether REML by multiple diagonalization is successful with real data, a study used 98,113 records of 44,765 Holstein cows for 14 conformation traits. Effects included in the model were herd classification, animal with unknown parent groups, and permanent environmental effects. Estimates of (co)variance components were on average 5% higher for type compared with estimates from an earlier analysis using only first records, but the estimate for udder cleft increased 2.4 times. Correlations for the permanent environmental effect were within .1 of genetic correlations. Multitrait REML by multiple diagonalization provides accurate multiple-trait estimates for repeatability models more efficiently than general model REML.

Lifetime Performance in Dairy Cattle. Genetic Parameters and Expected Improvement from Selection

Abstract The aims of this study were (1) to estimate genetic parameters for some biological and economic measures taken early in life, and subsequent lifetime economic performance traits, and (2) to calculate the expected genetic gain in lifetime performance from selection based on various selection indices. The breeding goal traits were: net income per year of productive life and lifetime income divided by cost. The early economic index traits were: net income per year and income divided by cost, measured during the first lactation or the first 15 months of productive life. Economic traits were calculated on the basis of production, fertility, and disease data on around 10000 cows having an opportunity to produce for at least 5 years. Variance-covariance components were estimated by applying a multiple-trait restricted maximum likelihood procedure to a sire model. The heritability estimates were around 40% for yield in 1st lactation, 20–30% for the early economic measures, and 10–15% for lifetime income/cost. Genetic correlation estimates with lifetime income/cost were 0.75–0.9 for early economic measures, and 0.6–0.7 for 1st and 2nd lactation yields. The annual genetic gain in lifetime economic performance was greater with selection on only 1st lactation yields than when the generation interval was extended to include 120-day yields in 2nd lactation. However, selection on the various early economic measures was up to 12% better than selection on only 1st lactation yields. A potential problem with a positive bias in the estimated correlation between early and lifetime performance exists. This bias is caused by the effect of milk yield on longevity, mediated through the management system.

Multiple-trait restricted maximum likelihood for simulated measures of ovulation rate with underlying multivariate normal distributions.

A data set that was used to estimate covariance components with REML for an animal model with eight measures of ovulation rate treated as separate traits was used as a template to simulate data sets of eight multivariate normal traits that were then truncated to binomial traits. The model for simulation included eight measures on 610 animals with 1,071 animals in the numerator relationship matrix. Heritabilities were equal for the eight measures, and both genetic and phenotypic correlations among the measures were equal. Ten replications for each combination of heritability (.15, .25, and .35) and genetic correlation (.50, .66, and .90) were simulated on the normal scale. For each replicate, estimates of the eight heritabilities and 28 genetic correlations were obtained by multiple-trait REML. The usual transformation of heritability estimated on the binomial scale overestimated heritability on the normal scale. Genetic correlations on the binomial scale seriously underestimated the correlations on the normal scale. Standard errors of the estimates obtained by replication were somewhat larger than the approximate SE from REMLPK (the multi-trait REML program of K. Meyer). A final set of 10 simulated replications with heritability of .25 and genetic correlation of 1.00 resulted in average estimates of .18 for heritability and of .66 for genetic correlation that agree closely with those from the analysis of measures of ovulation at eight estrous cycles used as a template; averages for heritability of .16 and for genetic correlation of .66 were obtained.

Estimation of genetic parameters for milk production in the first three lactations

Variance and covariance components were estimated for milk-production traits in the first 3 lactations. The data set comprised 95 910 first, 74 088 second and 55 371 third lactation records of cows, sired by 603 black and white AI bulls in Lower Saxony (F.R.G.). Estimation was by two multiple trait restricted maximum likelihood procedures, differing by allowance of non-zero residual covariances (MT-O-REML and MT-REML). Further, correlations between traits within lactations were estimated using Henderson's method III (H-III). The results revealed a clearly antagonistic relationship between milk yield and percentages of fat and protein, with correlations in the range -0.40 to -0.50. In comparison with H-III estimates, genetic variances and especially covariances increased markedly when using REML and all available data. Error (co) variances were of the same magnitude for all methods used. Estimated genetic correlations between corresponding traits in Lactations 1 and 2 were greater than 0.90, while for Lactations 2 and 3 the correlation was unity. Results suggest that dairy performance in the first 3 lactations, at least in Lactation 2 and 3, genetically should be regarded as one trait.

Genetic Parameters for Clinical Mastitis, Somatic Cell Counts, and Milk Production Estimated by Multiple-Trait Restricted Maximum Likelihood

Heritabilities and genetic correlations for mastitis, SCC, and milk production in first lactation were calculated using data from the Swedish disease recording system. Genetic parameters were estimated from sire components of variance and covariance obtained from a mulipletrait restriced maximum likelihood procedure. Data were stratified into three subsets according to breed and AI stud. The numbers of cows were 46,431, 25,373, and 25,201, respectively. Heritabilities of mastitis and SCC averaged .014 and .080, but breed differences were seen. Genetic correlations between mastitis and SCC were moderately high, averaging .6. Milk production showed a slightly unfavorable genetic correlation with mastitis and SCC.