Search

The US dairy cattle genetic evaluation is currently a multistep process, including multibreed traditional BLUP estimations followed by single-breed SNP effects estimation. Single-step GBLUP (ssGBLUP) combines pedigree and genomic data for all breeds in one analysis. Unknown parent groups (UPG) or metafounders (MF) can be used to address missing pedigree information. Fertility traits are notably difficult to evaluate due to low heritabilities, changing management, and a higher recent emphasis on selection to move in a favorable direction. We assessed bias, dispersion, and accuracy of fertility traits in all-breed US dairy cattle using pedigree-based BLUP (PBLUP) and ssGBLUP with UPG or MF; with 5% or 10% residual polygenic effect. Validation methods included the linear regression method and comparison of early and late deregressed proofs for Holstein and Jersey breeds. By comparing MF or UPG in PBLUP, we observed similar results in terms of bias, dispersion, and correlations between early and recent predictions. When genomics was used, ssGBLUP with MF and 10% residual polygenic effect consistently outperformed other models regarding bias, dispersion, and correlations. Compared with multistep results, ssGBLUP with MF and 10% residual polygenic effect showed less bias and increased correlations but slightly overdispersed estimates. Overall, genomic prediction of fertility traits using ssGBLUP was accurate and unbiased, more so with MF than with UPG., (The Authors. Published by Elsevier Inc. on behalf of the American Dairy Science Association®. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).)

Dairy cattle health traits are paramount from a welfare and economic viewpoint, and modern breeding programs therefore prioritize the genetic improvement of these traits. Estimated breeding values for health traits are published as the probability of animals staying healthy. They are obtained using threshold models, which assume that the observed binary phenotype (i.e., healthy or sick) is dictated by an underlying normally distributed liability exceeding or not exceeding a threshold. This methodology requires significant computing time and faces convergence challenges, as it implies a nonlinear system of equations. Linear models have more straightforward computations and provide a robust approximation to threshold models; thus, they could be used to overcome these challenges. However, linear models yield estimated breeding values on the observed scale, requiring an approximation to the liability scale analogous to that from threshold models to later obtain the estimated breeding values on the probability scale. In addition, the robustness of the approximation of linear to threshold models depends on the amount of information and the incidence of the trait, with extreme incidence (i.e., ≤5%) deviating from optimal approximation. Our objective was to test a transformation from the observed to the liability, and then to the probability scale, in the genetic evaluation of health traits with moderate and very low (extreme) incidence. Data comprised displaced abomasum (5.1 million), ketosis (3.6 million), lameness (5 million), and mastitis (6.3 million) records from a Holstein population with a pedigree of 6 million animals, of which 1.7 million were genotyped. Univariate threshold and linear models were performed to predict breeding values. The agreement between estimated breeding values on the probability scale derived from threshold and linear models was assessed using Spearman rank correlations and comparison of estimated breeding values distributions. Correlations were at least 0.95, and estimated breeding value distributions almost entirely overlapped for all the traits but displaced abomasum, the trait with the lowest incidence (2%). Computing time was ∼3 times longer for threshold than for linear models. In this Holstein population, the approximation was suboptimal for a trait with extreme incidence (2%). However, when the incidence was ≥6%, the approximation was robust, and its use is recommended along with linear models for analyzing categorical traits in large populations to ease the computational burden., (The Authors. Published by Elsevier Inc. on behalf of the American Dairy Science Association®. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).)

The exact accuracy of estimated breeding values can be calculated based on the prediction error variances obtained from the diagonal of the inverse of the left-hand side (LHS) of the mixed model equations (MME). However, inverting the LHS is not computationally feasible for large datasets, especially if genomic information is available. Thus, different algorithms have been proposed to approximate accuracies. This study aimed to: 1) compare the approximated accuracies from 2 algorithms implemented in the BLUPF90 suite of programs, 2) compare the approximated accuracies from the 2 algorithms against the exact accuracy based on the inversion of the LHS of MME, and 3) evaluate the impact of adding genotyped animals with and without phenotypes on the exact and approximated accuracies. Algorithm 1 approximates accuracies based on the diagonal of the genomic relationship matrix (G). In turn, algorithm 2 combines accuracies with and without genomic information through effective record contributions. The data were provided by the American Angus Association and included 3 datasets of growth, carcass, and marbling traits. The genotype file contained 1,235,930 animals, and the pedigree file contained 12,492,581 animals. For the genomic evaluation, a multi-trait model was applied to the datasets. To ensure the feasibility of inverting the LHS of the MME, a subset of data under single-trait models was used to compare approximated and exact accuracies. The correlations between exact and approximated accuracies from algorithms 1 and 2 of genotyped animals ranged from 0.87 to 0.90 and 0.98 to 0.99, respectively. The intercept and slope of the regression of exact on approximated accuracies from algorithm 2 ranged from 0.00 to 0.01 and 0.82 to 0.87, respectively. However, the intercept and the slope for algorithm 1 ranged from -0.10 to 0.05 and 0.98 to 1.10, respectively. In more than 80% of the traits, algorithm 2 exhibited a smaller mean square error than algorithm 1. The correlation between the approximated accuracies obtained from algorithms 1 and 2 ranged from 0.56 to 0.74, 0.38 to 0.71, and 0.71 to 0.97 in the groups of genotyped animals, genotyped animals without phenotype, and proven genotyped sires, respectively. The approximated accuracy from algorithm 2 showed a closer behavior to the exact accuracy when including genotyped animals in the analysis. According to the results, algorithm 2 is recommended for genetic evaluations since it proved more precise., (© The Author(s) 2024. Published by Oxford University Press on behalf of the American Society of Animal Science.)

Threshold models are often used in genetic analysis of categorical data, such as calving ease. Solutions in the liability scale are easily transformed into probabilities; therefore, estimated breeding values are published as the probability of expressing the category of main interest and are the industry's gold standard because they are easy to interpret and use for selection. However, because threshold models involve nonlinear equations and probability functions, implementing such a method is complex. Challenges include long computing time and convergence issues, intensified by including genomic data. Linear models are an alternative to overcome those challenges. Estimated breeding values computed using linear models are highly correlated (≥0.96) with those from threshold models; however, the lack of a transformation from the observed to the probability scale limits the use of linear models. The objective of this study was to propose transformations from observed to probability scale analogous to the transformation from liability to probability scale. We assessed computing time, peak memory use, correlations between estimated breeding values, and estimated genetic trends from linear and threshold models. With 11M animals in the pedigree and almost 965k genotyped animals, linear models were 5× faster to converge than threshold models (32 vs. 145 h), and peak memory use was the same (189 GB). The transformations proposed provided highly correlated probabilities from linear and threshold models. Correlations between direct (maternal) estimated breeding values from linear and threshold models and transformed to probabilities were ≥0.99 (0.97) for all animals in the pedigree, sires with/without progeny records, or animals with phenotypic records; therefore, estimated genetic trends were analogous, suggesting no loss of genetic progress in breeding programs that would adopt linear instead of threshold models. Furthermore, linear models reduced computing time by 5-fold compared to the threshold models; this enables weekly genetic evaluations and opens the possibility of using multi-trait models for categorical traits to improve selection effectiveness., (© The Author(s) 2024. Published by Oxford University Press on behalf of the American Society of Animal Science.)

Mortality is an economically important trait usually handled as a discrete outcome from hatch time until selection in most broiler breeder programs. However, in other species, it has been shown that not only does the genetic component change over time, but also there are maternal genetic effects to be considered when mortality is recorded early in life. This study aimed to investigate alternative trait definitions of mortality with varying models and effects. Three years' worth of data were provided by Cobb-Vantress, Inc. and included 2 mortality traits. The first trait was binary, whether the bird died or not (OM), and the second trait was a categorical weekly mortality trait. After data cleaning, 6 wk of data for the 2 given mortality traits were used to develop 5 additional trait definitions. The definitions were broiler mortality (BM), early and late mortality (EM & LM), and 2 traits with repeated records as cumulative or binary (CM and RM, respectively). Variance components were estimated using linear and threshold models to investigate whether either model had a benefit. Genomic breeding values were predicted using the BLUP90 software suite, and linear regression validation (LR) was used to compare trait definitions and models. Heritability estimates ranged from 0.01 (0.00) to 0.16 (0.01) under linear and 0.04 (0.01) to 0.21 (0.01) under threshold models, indicating genetic variability within the population across these trait definitions. The genetic correlation between EM and LM ranged from 0.48 to 0.81 across the different lines, indicating they have divergent genetic backgrounds and should be considered different traits. The LR accuracies showed that EM and LM used together in a 2-trait model have comparable accuracies to that of OM while giving a more precise picture of mortality. When including the maternal effect, the direct heritability considerably decreased for EM, indicating that the maternal effect plays an important role in early mortality. Therefore, a suitable approach would be a model with EM and LM while considering the maternal effect for EM. Single nucleotide polymorphism effects were estimated, and no individual SNP explained more than 1% of the additive genetic variance. Additionally, the SNP with the largest effect size and variance were inconsistent across trait definitions. Chicken mortality can be defined in different ways, and reviewing these definitions and models may benefit poultry breeding programs., (© The Author(s) 2024. Published by Oxford University Press on behalf of the American Society of Animal Science.)

Economically important traits are usually complex traits influenced by genes, environment and genotype-by-environment (G × E) interactions. Ignoring G × E interaction could lead to bias in the estimation of breeding values and selection decisions. A total of 1,778 pigs were genotyped using the PorcineSNP80 BeadChip. The existence of G × E interactions was investigated using a single-step reaction norm model for growth traits of days to 100 kg (AGE) and backfat thickness adjusted to 100 kg (BFT), based on a pedigree-based relationship matrix (A) or a genomic-pedigree joint relationship matrix (H). In the reaction norm model, the herd-year-season effect was measured as the environmental variable (EV). Our results showed no G × E interactions for AGE, but for BFT. For both AGE and BFT, the genomic reaction norm model (H) produced more accurate predictions than the conventional reaction norm model (A). For BFT, the accuracies were greater based on the reaction norm model than those based on the reduced model without exploiting G × E interaction, with EV ranging from 0.5 to 1, and accuracy increasing by 3.9% and 4.6% in the reaction norm model based on A and H matrices, respectively, while reaction norm model yielded approximately 8.4% and 7.9% lower accuracy for EVs ranging from 0 to 0.4, based on A and H matrices, respectively. In addition, for BFT, the highest accuracy was obtained in the BJLM6 farm for realizing directional selection. This study will help to apply G × E interactions to practical genomic selection., (© 2020 Wiley-VCH GmbH.)

Test-day traits are important for genetic evaluation in dairy cattle and are better modeled by multiple-trait random regression models (RRM). The reliability and bias of genomic estimated breeding values (GEBV) predicted using multiple-trait RRM via single-step genomic best linear unbiased prediction (ssGBLUP) were investigated in the 3 major dairy cattle breeds in Canada (i.e., Ayrshire, Holstein, and Jersey). Individual additive genomic random regression coefficients for the test-day traits were predicted using 2 multiple-trait RRM: (1) one for milk, fat, and protein yields in the first, second, and third lactations, and (2) one for somatic cell score in the first, second, and third lactations. The predicted coefficients were used to derive GEBV for each lactation day and, subsequently, the daily GEBV were compared with traditional daily parent averages obtained by BLUP. To ensure compatibility between pedigree and genomic information for genotyped animals, different scaling factors for combining the inverse of genomic (G -1 ) and pedigree (A -1 ) relationship matrices were tested. In addition, the inclusion of only genotypes from animals with accurate breeding values (defined in preliminary analysis) was compared with the inclusion of all available genotypes in the analyzes. The ssGBLUP model led to considerably larger validation reliabilities than the BLUP model without genomic information. In general, scaling factors used to combine the G 22 ) relationship matrices were tested. In addition, the inclusion of only genotypes from animals with accurate breeding values (defined in preliminary analysis) was compared with the inclusion of all available genotypes in the analyzes. The ssGBLUP model led to considerably larger validation reliabilities than the BLUP model without genomic information. In general, scaling factors used to combine the G -1 and A -1 22 matrices. Similar results were observed when including either all available genotypes or only genotypes from animals with accurate breeding values. These findings indicate that ssGBLUP using multiple-trait RRM increases reliability and reduces bias of breeding values of young animals when compared with parent average from traditional BLUP in the Canadian Ayrshire, Holstein, and Jersey breeds.-1 and A -1 22 matrices. Similar results were observed when including either all available genotypes or only genotypes from animals with accurate breeding values. These findings indicate that ssGBLUP using multiple-trait RRM increases reliability and reduces bias of breeding values of young animals when compared with parent average from traditional BLUP in the Canadian Ayrshire, Holstein, and Jersey breeds., (Copyright © 2019 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

The purpose of this study was to determine whether multi-country genomic evaluation can be accomplished by multiple-trait genomic best linear unbiased predictor (GBLUP) without sharing genotypes of important animals. Phenotypes and genotypes with 40k SNP were simulated for 25,000 animals, each with 4 traits assuming the same genetic variance and 0.8 genetic correlations. The population was split into 4 subpopulations corresponding to 4 countries, one for each trait. Additionally, a prediction population was created from genotyped animals that were not present in the individual countries but were related to each country's population. Genomic estimated breeding values were computed for each country and subsequently converted to SNP effects. Phenotypes were reconstructed for the prediction population based on the SNP effects of a country and the prediction animals' genotypes. The prediction population was used as the basis for the international evaluation, enabling bull comparisons without sharing genotypes and only sharing SNP effects. The computations were such that SNP effects computed within-country or in the prediction population were the same. Genomic estimated breeding values were calculated by single-trait GBLUP for within-country and multiple-trait GBLUP for multi-country predictions. The true accuracy for the prediction population with reconstructed phenotypes was at most 0.02 less than the accuracy with the original data. The differences increased when countries were assumed unequally sized. However, accuracies by multiple-trait GBLUP with the prediction population were always greater than accuracies from any single within-country prediction. Multi-country genomic evaluations by multiple-trait GBLUP are possible without using original genotypes at a cost of lower accuracy compared with explicitly combining countries' data., (Copyright © 2019 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

The objective of this study was to model differences in pedigree accuracy caused by selective genotyping. As genotypes are used to correct pedigree errors, some pedigree relationships are more accurate than others. These accuracy differences can be modeled with uncertain parentage models that distribute the paternal (maternal) contribution across multiple sires (dams). In our case, the parents were the parent on record and an unknown parent group to account for pedigree relationships that were not confirmed through genotypes. Pedigree accuracy was addressed through simulation and through North American Holstein data. Data were simulated to be representative of the dairy industry with heterogeneous pedigree depth, pedigree accuracy, and genotyping. Holstein data were obtained from the official evaluation for milk, fat, and protein. Two models were compared: the traditional approach, assuming accurate pedigrees, and uncertain parentage, assuming variable pedigree accuracy. The uncertain parentage model was used to add pedigree relationships for alternative parents when pedigree relationships were not certain. The uncertain parentage model included 2 possible sires (dams) when the sire (dam) could not be confirmed with genotypes. The 2 sires (dams) were the sire (dam) on record with probability 0.90 (0.95) and the unknown parent group for the birth year of the sire (dam) with probability 0.10 (0.05). An additional set of assumptions was tested in simulation to mimic an extensive dairy production system by using a sire probability of 0.75, a dam probability of 0.85, and the remainder attributed to the unknown parent groups. In the simulation, small bias differences occurred between models based on pedigree accuracy and genotype status. Rank correlations were strong between traditional and uncertain parentage models in simulation (≥0.99) and in Holstein (≥0.99). For Holsteins, the estimated breeding value differences between models were small for most animals. Thus, traditional models can continue to be used for dairy genomic prediction despite using genotypes to improve pedigree accuracy. Those genotypes can also be used to discover maternal parentage, specifically maternal grandsires and great grandsires when the dam is not known. More research is needed to understand how to use discovered maternal pedigrees in genetic prediction., (Copyright © 2019 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

The success and sustainability of a breeding program incorporating genomic information is largely dependent on the accuracy of predictions. For low heritability traits, large training populations are required to achieve high accuracies of genomic estimated breeding values (GEBV). By including genotyped and nongenotyped animals simultaneously in the evaluation, the single-step genomic BLUP (ssGBLUP) approach has the potential to deliver more accurate and less biased genomic evaluations. The aim of this study was to compare the accuracy and bias of genomic predictions for various traits in Canadian Holstein cattle using ssGBLUP and multi-step genomic BLUP (msGBLUP) under different strategies, such as (1) adding genomic information of cows in the analysis, (2) testing different adjustments of the genomic relationship matrix, and (3) using a blending approach to obtain GEBV from msGBLUP. The following genomic predictions were evaluated regarding accuracy and bias: (1) GEBV estimated by ssGBLUP; (2) direct genomic value estimated by msGBLUP with polygenic effects of 5 and 20%; and (3) GEBV calculated by a blending approach of direct genomic value with estimated breeding values using polygenic effects of 5 and 20%. The effect of adding genomic information of cows in the evaluation was also assessed for each approach. When genomic information was included in the analyses, the average improvement in observed reliability of predictions was observed to be 7 and 13 percentage points for reproductive and workability traits, respectively, compared with traditional BLUP. Absolute deviation from 1 of the regression coefficient of the linear regression of de-regressed estimated breeding values on genomic predictions went from 0.19 when using traditional BLUP to 0.22 when using the msGBLUP method, and to 0.14 when using the ssGBLUP method. The use of polygenic weight of 20% in the msGBLUP slightly improved the reliability of predictions, while reducing the bias. A similar trend was observed when a blending approach was used. Adding genomic information of cows increased reliabilities, while decreasing bias of genomic predictions when using the ssGBLUP method. Differences between using a training population with cows and bulls or with only bulls for the msGBLUP method were small, likely due to the small number of cows included in the analysis. Predictions for lowly heritable traits benefit greatly from genomic information, especially when all phenotypes, pedigrees, and genotypes are used in a single-step approach., (Copyright © 2018 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

The objective of this study was to compare genetic trends from single-step genomic BLUP (ssGBLUP) and traditional BLUP models for milk production traits of US Holsteins. Phenotypes were 305-d milk, fat, and protein yields from 21,527,040 cows recorded between January 1990 and August 2015. The pedigree file included 29,651,623 animals and was limited to 3 generations back from recorded or genotyped animals. Genotypes for 764,029 animals were used, and analyses were by a 3-trait repeatability model as used in the US official genetic evaluation. Unknown-parent groups were incorporated into the inverse of a relationship matrix (H -1 in ssGBLUP and A -1 in BLUP) with the QP transformation. For ssGBLUP, 18,359 genotyped animals were randomly chosen as core animals to calculate the inverse of the genomic relationship matrix with the APY algorithm. Computations took 6.5 h and 1.4 GB of memory for BLUP, and 13 h and 115 GB of memory for ssGBLUP. For genotyped sires with at least 10 daughters, the average genetic levels for predicted transmitting ability (PTA) and genomic PTA were similar up to 2008, with a higher level for ssGBLUP later (approximately by 36 kg for milk, 2.1 kg for fat, and 1.1 kg for protein for bulls born in 2010). For genotyped cows, the average genetic levels were similar up to 2006, with a higher level for ssGBLUP (approximately by 91 kg for milk, 3.6 kg for fat, and 2.7 kg for protein for cows born in 2012). For all cows, the average levels were slightly higher for ssGBLUP, with much smaller differences than for genotyped cows. Trends for BLUP indicate bias due to genomic preselection for genotyped sires and cows. For official evaluations released in December 2016, traditional PTA had the same trend as multiple-step genomic PTA for both genotyped bulls and cows except for the youngest bulls, who had traditional PTA slightly lower than genomic PTA. For genotyped bulls born in recent years, genetic gain for official traditional and genomic evaluations was similar in contrast to ssGBLUP and BLUP differences. Official PTA for cows were adjusted so that the Mendelian sampling variance was comparable with that for bulls, and those adjustments likely removed bias due to genomic preselection from traditional PTA, especially for genotyped cows. The ssGBLUP method seems to account partially for that bias and is computationally suitable for national evaluations., (The Authors. Published by FASS Inc. and Elsevier Inc. on behalf of the American Dairy Science Association®. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).)

Four performance-related traits [growth trait (GROW), feed efficiency trait 1 (FE1) and trait 2 (FE2), and dissection trait (DT)] and 4 categorical traits [mortality (MORT) and 3 disorder traits (DIS1, DIS2, and DIS3)] were analyzed using linear and threshold single- and multi-trait models. Field data included 186,596 records of commercial broilers from Cobb-Vantress, Inc. Average-information restricted maximum likelihood and Gibbs sampling-based methods were used to obtain estimates of the (co)variance components, heritabilities, and genetic correlations in a traditional approach using best linear unbiased prediction (BLUP). The ability to predict future breeding values (measured as realized accuracy) was checked in the last generation when traditional BLUP and single-step genomic BLUP were used. Heritability estimates for GROW, FE1, and FE2 in single- and multi-trait models were similar and moderate (0.22 to 0.26) but high for DT (0.48 to 0.50). For MORT, DIS1, and DIS2, heritabilities were 0.13, 0.24, and 0.34, respectively. Estimates from single- and multi-trait models were also very similar. However, heritability for DIS3 was higher from the single-trait threshold model than for the multi-trait linear-threshold model (0.29 vs. 0.19). Genetic correlations between growth traits and MORT were weak, except for maternal GROW, which had a moderate negative correlation (-0.50) with MORT. The genetic correlation between MORT and DIS1 was strong and positive (0.77). Feed efficiency 1, which was moderately heritable (0.25) and is highly selected for, was not genetically related to MORT of broilers and other disorders. Broiler MORT also had moderate heritability (0.13), which suggests that MORT and FE1 can be improved through selection without negatively impacting other important traits. Selection of heavier maternal GROW also may decrease offspring MORT.

The objective of this study was to investigate the feasibility of genomic evaluation for cow mortality and milk production using a single-step methodology. Genomic relationships between cow mortality and milk production were also analyzed. Data included 883,887 (866,700) first-parity, 733,904 (711,211) second-parity, and 516,256 (492,026) third-parity records on cow mortality (305-d milk yields) of Holsteins from Northeast states in the United States. The pedigree consisted of up to 1,690,481 animals including 34,481 bulls genotyped with 36,951 SNP markers. Analyses were conducted with a bivariate threshold-linear model for each parity separately. Genomic information was incorporated as a genomic relationship matrix in the single-step BLUP. Traditional and genomic estimated breeding values (GEBV) were obtained with Gibbs sampling using fixed variances, whereas reliabilities were calculated from variances of GEBV samples. Genomic EBV were then converted into single nucleotide polymorphism (SNP) marker effects. Those SNP effects were categorized according to values corresponding to 1 to 4 standard deviations. Moving averages and variances of SNP effects were calculated for windows of 30 adjacent SNP, and Manhattan plots were created for SNP variances with the same window size. Using Gibbs sampling, the reliability for genotyped bulls for cow mortality was 28 to 30% in EBV and 70 to 72% in GEBV. The reliability for genotyped bulls for 305-d milk yields was 53 to 65% to 81 to 85% in GEBV. Correlations of SNP effects between mortality and 305-d milk yields within categories were the highest with the largest SNP effects and reached >0.7 at 4 standard deviations. All SNP regions explained less than 0.6% of the genetic variance for both traits, except regions close to the DGAT1 gene, which explained up to 2.5% for cow mortality and 4% for 305-d milk yields. Reliability for GEBV with a moderate number of genotyped animals can be calculated by Gibbs samples. Genomic information can greatly increase the reliability of predictions not only for milk but also for mortality. The existence of a common region on Bos taurus autosome 14 affecting both traits may indicate a major gene with a pleiotropic effect on milk and mortality., (Copyright © 2017 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

Background: Much effort is put into identifying causative quantitative trait nucleotides (QTN) in animal breeding, empowered by the availability of dense single nucleotide polymorphism (SNP) information. Genomic selection using traditional SNP information is easily implemented for any number of genotyped individuals using single-step genomic best linear unbiased predictor (ssGBLUP) with the algorithm for proven and young (APY). Our aim was to investigate whether ssGBLUP is useful for genomic prediction when some or all QTN are known., Methods: Simulations included 180,000 animals across 11 generations. Phenotypes were available for all animals in generations 6 to 10. Genotypes for 60,000 SNPs across 10 chromosomes were available for 29,000 individuals. The genetic variance was fully accounted for by 100 or 1000 biallelic QTN. Raw genomic relationship matrices (GRM) were computed from (a) unweighted SNPs, (b) unweighted SNPs and causative QTN, (c) SNPs and causative QTN weighted with results obtained with genome-wide association studies, (d) unweighted SNPs and causative QTN with simulated weights, (e) only unweighted causative QTN, (f-h) as in (b-d) but using only the top 10% causative QTN, and (i) using only causative QTN with simulated weight. Predictions were computed by pedigree-based BLUP (PBLUP) and ssGBLUP. Raw GRM were blended with 1 or 5% of the numerator relationship matrix, or 1% of the identity matrix. Inverses of GRM were obtained directly or with APY., Results: Accuracy of breeding values for 5000 genotyped animals in the last generation with PBLUP was 0.32, and for ssGBLUP it increased to 0.49 with an unweighted GRM, 0.53 after adding unweighted QTN, 0.63 when QTN weights were estimated, and 0.89 when QTN weights were based on true effects known from the simulation. When the GRM was constructed from causative QTN only, accuracy was 0.95 and 0.99 with blending at 5 and 1%, respectively. Accuracies simulating 1000 QTN were generally lower, with a similar trend. Accuracies using the APY inverse were equal or higher than those with a regular inverse., Conclusions: Single-step GBLUP can account for causative QTN via a weighted GRM. Accuracy gains are maximum when variances of causative QTN are known and blending is at 1%.

The purpose of this study is review and evaluation of computing methods used in genomic selection for animal breeding. Commonly used models include SNP BLUP with extensions (BayesA, etc), genomic BLUP (GBLUP) and single-step GBLUP (ssGBLUP). These models are applied for genomewide association studies (GWAS), genomic prediction and parameter estimation. Solving methods include finite Cholesky decomposition possibly with a sparse implementation, and iterative Gauss-Seidel (GS) or preconditioned conjugate gradient (PCG), the last two methods possibly with iteration on data. Details are provided that can drastically decrease some computations. For SNP BLUP especially with sampling and large number of SNP, the only choice is GS with iteration on data and adjustment of residuals. If only solutions are required, PCG by iteration on data is a clear choice. A genomic relationship matrix (GRM) has limited dimensionality due to small effective population size, resulting in infinite number of generalized inverses of GRM for large genotyped populations. A specific inverse called APY requires only a small fraction of GRM, is sparse and can be computed and stored at a low cost for millions of animals. With APY inverse and PCG iteration, GBLUP and ssGBLUP can be applied to any population. Both tools can be applied to GWAS. When the system of equations is sparse but contains dense blocks, a recently developed package for sparse Cholesky decomposition and sparse inversion called YAMS has greatly improved performance over packages where such blocks were treated as sparse. With YAMS, GREML and possibly single-step GREML can be applied to populations with >50 000 genotyped animals. From a computational perspective, genomic selection is becoming a mature methodology.

Background: Metafounders are pseudo-individuals that encapsulate genetic heterozygosity and relationships within and across base pedigree populations, i.e. ancestral populations. This work addresses the estimation and usefulness of metafounder relationships in single-step genomic best linear unbiased prediction (ssGBLUP)., Results: We show that ancestral relationship parameters are proportional to standardized covariances of base allelic frequencies across populations, such as [Formula: see text] fixation indexes. These covariances of base allelic frequencies can be estimated from marker genotypes of related recent individuals and pedigree. Simple methods for their estimation include naïve computation of allele frequencies from marker genotypes or a method of moments that equates average pedigree-based and marker-based relationships. Complex methods include generalized least squares (best linear unbiased estimator (BLUE)) or maximum likelihood based on pedigree relationships. To our knowledge, methods to infer [Formula: see text] coefficients from marker data have not been developed for related individuals. We derived a genomic relationship matrix, compatible with pedigree relationships, that is constructed as a cross-product of {-1,0,1} codes and that is equivalent (apart from scale factors) to an identity-by-state relationship matrix at genome-wide markers. Using a simulation with a single population under selection in which only males and youngest animals are genotyped, we observed that generalized least squares or maximum likelihood gave accurate and unbiased estimates of the ancestral relationship parameter (true value: 0.40) whereas the naïve method and the method of moments were biased (average estimates of 0.43 and 0.35). We also observed that genomic evaluation by ssGBLUP using metafounders was less biased in terms of estimates of genetic trend (bias of 0.01 instead of 0.12), resulted in less overdispersed (0.94 instead of 0.99) and as accurate (0.74) estimates of breeding values than ssGBLUP without metafounders and provided consistent estimates of heritability., Conclusions: Estimation of metafounder relationships can be achieved using BLUP-like methods with pedigree and markers. Inclusion of metafounder relationships reduces bias of genomic predictions with no loss in accuracy.

Genetically linked small and large dairy cattle populations were simulated to test the effect of different sources of information from foreign populations on the accuracy of predicting breeding values for young animals in a small population. A large dairy cattle population (P L ) with >20 generations was simulated, and a small subpopulation (P S ) with 3 generations was formed as a related population, including phenotypes and genomic information. Predicted breeding values for young animals in the small population were calculated using BLUP and single-step genomic BLUP (ssGBLUP) in 4 different scenarios: (S1) 3,166 phenotypes, 22,855 pedigree animals, and 1,000 to 6,000 genotypes for P S ; (S2) S1 plus genomic estimated breeding value (GEBV) for 4,475 sires from P L as external information; (S3) S1 plus 221,580 phenotypes, 402,829 pedigree animals, and 53,558 genotypes for P L ; and (S4) single nucleotide polymorphism (SNP) effects calculated based on P L data. The ability to predict true breeding value was assessed in the youngest third of the genotyped animals in the small population. When data only from the small population were used and 1,000 animals were genotyped, the accuracy of GEBV was only 1 point greater than the estimated breeding value accuracy (0.32 vs. 0.31). Adding external GEBV for sires from P L did not considerably increase accuracy (0.33 vs. 0.32 in S1). Combining phenotypes, pedigree, and genotypes for P S and P L was beneficial for predicting accuracy of GEBV in the small population, and the prediction accuracy of GEBV in this scenario was 0.38 compared with 0.31 from estimated breeding values. When SNP effects from P L were used to predict GEBV for young genotyped animals from P S , accuracy was greatest (0.56). With 6,000 genotyped animal in P S , accuracy was greatest (0.61) with the combined populations. In a small population with few genotypes, the highest accuracy of evaluation may be obtained by using SNP effects derived from a related large population., (Copyright © 2017 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

Most breeding companies evaluate economically important traits in males and females as a single trait, assuming genetic correlation of 1 between phenotypes measured in both sexes. This assumption may not be true because genes may be differently expressed in males and females. We estimated genetic correlations between males and females for growth and efficiency traits in broiler chickens, growth traits in American Angus beef cattle, and birth weight and preweaning mortality in purebred pigs; therefore, each trait was treated differently in males and females. Variance components were estimated in single- and multiple-trait models, jointly or separated into both sexes. Furthermore, we calculated traditional and genomic evaluations, and we correlated EBV or genomic EBV (GEBV) from joint and separate evaluations for males and females. For broiler chickens, genetic correlations ranged from 0.86 to 0.94. For Angus cattle, genetic correlations ranged from 0.86 to 0.98 for early growth traits and were less, ranging from 0.68 to 0.84, for postweaning gain. In pigs, genetic correlations ranged from 0.98 to 0.99 for birth weight and from 0.71 to 0.73 for preweaning mortality. For some models in all 3 animal species, the joint and separate analyses had different heritabilities. Despite differences in heritability, the correlations within the sex-specific trait EBV and between the sex-specific and the joint trait EBV were very strong, regardless of the model or inclusion of genomic information. Males and females differed for traits measured late in the animal's life; however, strong traditional EBV correlations and also GEBV correlations indicate that considering the traits equal in males and females may have no negative impact on selection.

The purpose of this study was to examine accuracy of genomic selection via single-step genomic BLUP (ssGBLUP) when the direct inverse of the genomic relationship matrix (G) is replaced by an approximation of G(-1) based on recursions for young genotyped animals conditioned on a subset of proven animals, termed algorithm for proven and young animals (APY). With the efficient implementation, this algorithm has a cubic cost with proven animals and linear with young animals. Ten duplicate data sets mimicking a dairy cattle population were simulated. In a first scenario, genomic information for 20k genotyped bulls, divided in 7k proven and 13k young bulls, was generated for each replicate. In a second scenario, 5k genotyped cows with phenotypes were included in the analysis as young animals. Accuracies (average for the 10 replicates) in regular EBV were 0.72 and 0.34 for proven and young animals, respectively. When genomic information was included, they increased to 0.75 and 0.50. No differences between genomic EBV (GEBV) obtained with the regular G(-1) and the approximated G(-1) via the recursive method were observed. In the second scenario, accuracies in GEBV (0.76, 0.51 and 0.59 for proven bulls, young males and young females, respectively) were also higher than those in EBV (0.72, 0.35 and 0.49). Again, no differences between GEBV with regular G(-1) and with recursions were observed. With the recursive algorithm, the number of iterations to achieve convergence was reduced from 227 to 206 in the first scenario and from 232 to 209 in the second scenario. Cows can be treated as young animals in APY without reducing the accuracy. The proposed algorithm can be implemented to reduce computing costs and to overcome current limitations on the number of genotyped animals in the ssGBLUP method., (© 2015 Blackwell Verlag GmbH.)

The objective of this study was to remove bottlenecks generally found in a computer program for average-information REML. The refinements included improvements to setting-up mixed-model equations on a hash table with a faster hash function as sparse matrix storage, changing sparse structures in calculation of traces, and replacing a sparse matrix package using traditional methods (FSPAK) with a new package using supernodal methods (YAMS); the latter package quickly processed sparse matrices containing large, dense blocks. Comparisons included 23 models with data sets from broiler, swine, beef, and dairy cattle. Models included single-trait, multiple-trait, maternal, and random regression models with phenotypic data; selected models used genomic information in a single-step approach. Setting-up mixed model equations was completed without abnormal termination in all analyses. Calculations in traces were accelerated with a hash format, especially for models with a genomic relationship matrix, and the maximum speed was 67 times faster. Computations with YAMS were, on average, more than 10 times faster than with FSPAK and had greater advantages for large data and more complicated models including multiple traits, random regressions, and genomic effects. These refinements can be applied to general average-information REML programs.

Background: As more and more genotypes become available, accuracy of genomic evaluations can potentially increase. However, the impact of genotype data on accuracy depends on the structure of the genotyped cohort. For populations such as dairy cattle, the greatest benefit has come from genotyping sires with high accuracy, whereas the benefit due to adding genotypes from cows was smaller. In broiler chicken breeding programs, males have less progeny than dairy bulls, females have more progeny than dairy cows, and most production traits are recorded for both sexes. Consequently, genotyping both sexes in broiler chickens may be more advantageous than in dairy cattle., Methods: We studied the contribution of genotypes from males and females using a real dataset with genotypes on 15 723 broiler chickens. Genomic evaluations used three training sets that included only males (4648), only females (8100), and both sexes (12 748). Realized accuracies of genomic estimated breeding values (GEBV) were used to evaluate the benefit of including genotypes for different training populations on genomic predictions of young genotyped chickens., Results: Using genotypes on males, the average increase in accuracy of GEBV over pedigree-based EBV for males and females was 12 and 1 percentage points, respectively. Using female genotypes, this increase was 1 and 18 percentage points, respectively. Using genotypes of both sexes increased accuracies by 19 points for males and 20 points for females. For two traits with similar heritabilities and amounts of information, realized accuracies from cross-validation were lower for the trait that was under strong selection., Conclusions: Overall, genotyping males and females improves predictions of all young genotyped chickens, regardless of sex. Therefore, when males and females both contribute to genetic progress of the population, genotyping both sexes may be the best option.

The purpose of this study was to estimate correlations between purebred and F1 crossbred performance to verify the appropriateness of current models used in multibreed selection. Records on birth weight (WB) and weaning weight (WW) from purebred Limousins (LIM) and Limousin × Angus progeny (F1) were used to estimate genetic parameters using a multiple-trait (purebred and F1 weights were different traits) approach. For WB, there were 148,647 records for LIM and 17,981 for F1, and for WW, there were 81,585 records for LIM and 21,778 for F1. The fixed effect in models for LIM and F1 animals was contemporary group. Random effects for LIM animals were direct genetic, maternal genetic, and maternal permanent environment effects. Random effects for F1 were sire and dam. The pedigree for Angus dams used for crossing was unavailable and therefore these dams were assumed unrelated. The direct h2 estimates (SE) for purebred animals were 0.41 (0.05) and 0.24 (0.02) for WB and WW, respectively. For F1, the same estimates were 0.22 (0.09) and 0.32 (0.05). Genetic correlations estimates between purebreds and crossbreds were 0.84 (0.07) and 0.64 (0.18) for WB and WW, respectively. The genetic correlation for WW estimated in this study suggests that F1 and purebred information for this trait should not be treated, genetically, as the same trait due to different genetic effects molding it. However, the genetic correlation for WB was much higher, indicating that this trait in purebreds and F1 is essentially the same trait.

This work studied differences between expected (calculated from pedigree) and realized (genomic, from markers) relationships in a real population, the influence of quality control on these differences, and their fit to current theory. Data included 4940 pure line chickens across five generations genotyped for 57,636 SNP. Pedigrees (5762 animals) were available for the five generations, pedigree starting on the first one. Three levels of quality control were used. With no quality control, mean difference between realized and expected relationships for different type of relationships was ≤ 0.04 with standard deviation ≤ 0.10. With strong quality control (call rate ≥ 0.9, parent-progeny conflicts, minor allele frequency and use of only autosomal chromosomes), these numbers reduced to ≤ 0.02 and ≤ 0.04, respectively. While the maximum difference was 1.02 with the complete data, it was only 0.18 with the latest three generations of genotypes (but including all pedigrees). Variation of expected minus realized relationships agreed with theoretical developments and suggests an effective number of loci of 70 for this population. When the pedigree is complete and as deep as the genotypes, the standard deviation of difference between the expected and realized relationships is around 0.04, all categories confounded. Standard deviation of differences larger than 0.10 suggests bad quality control, mistakes in pedigree recording or genotype labelling, or insufficient depth of pedigree., (© 2014 Blackwell Verlag GmbH.)

Assigning unknown parent groups (UPG) in mixed-model equations using single-step genomic BLUP was investigated to reduce bias and to increase accuracy in genomic estimated breeding values (GEBV). The original UPG were defined based on the animal's birth year and the sex of the animal's unknown parents. Combining the last 2 UPG for the animals' birth years and separating the UPG for US and non-US Holsteins were considered in the redefinition. A full data set in the 2011 national genetic evaluation of final score in US Holsteins was used to calculate estimated breeding values (EBV) for validation, and a subset of the 2011 data, which excluded phenotypes recorded after 2007, was used to calculate GEBV for all animals, including 34,500 genotyped bulls. The EBV and GEBV in 2007 were compared with EBV in the 2011 full data. The last group effects for unknown sires and dams were overestimated with the GEBV model using the reduced 2007 data. The genetic trends from EBV in 2011 and GEBV in 2007 with the original UPG in the last few years demonstrated inflation, whereas GEBV with the redefined UPG by combining the last 2 groups showed deflation. On the other hand, the redefined UPG by separating for US and non-US Holsteins reduced the bias in GEBV. Regression coefficients smaller than unity for GEBV for young genotyped bulls with no daughters in 2007 on progeny deviations in 2011 also indicated inflation. The redefining of UPG reduced bias and slightly increased accuracy in GEBV for both US and non-US genotyped bulls. Rank correlations between GEBV in 2007 and in 2011 with the redefined UPG were higher than those with no UPG and the original UPG, especially for non-US bulls. Redefining of UPG in genomic evaluation could improve reliability of GEBV and provide correct genetic trends., (Copyright © 2014 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

Computing the inverse of the genomic relationship matrix using recursion was investigated. A traditional algorithm to invert the numerator relationship matrix is based on the observation that the conditional expectation for an additive effect of 1 animal given the effects of all other animals depends on the effects of its sire and dam only, each with a coefficient of 0.5. With genomic relationships, such an expectation depends on all other genotyped animals, and the coefficients do not have any set value. For each animal, the coefficients plus the conditional variance can be called a genomic recursion. If such recursions are known, the mixed model equations can be solved without explicitly creating the inverse of the genomic relationship matrix. Several algorithms were developed to create genomic recursions. In an algorithm with sequential updates, genomic recursions are created animal by animal. That algorithm can also be used to update a known inverse of a genomic relationship matrix for additional genotypes. In an algorithm with forward updates, a newly computed recursion is immediately applied to update recursions for remaining animals. The computing costs for both algorithms depend on the sparsity pattern of the genomic recursions, but are lower or equal than for regular inversion. An algorithm for proven and young animals assumes that the genomic recursions for young animals contain coefficients only for proven animals. Such an algorithm generates exact genomic EBV in genomic BLUP and is an approximation in single-step genomic BLUP. That algorithm has a cubic cost for the number of proven animals and a linear cost for the number of young animals. The genomic recursions can provide new insight into genomic evaluation and possibly reduce costs of genetic predictions with extremely large numbers of genotypes., (Copyright © 2014 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

Data sets of US Holsteins, Israeli Holsteins, and pigs from PIC (a Genus company, Hendersonville, TN) were used to evaluate the effect of different numbers of generations on ability to predict genomic breeding values of young genotyped animals. The influence of including only 2 generations of ancestors (A2) or all ancestors (Af) was also investigated. A total of 34,506 US Holsteins, 1,305 Israeli Holsteins, and 5,236 pigs were genotyped. The evaluations were computed by traditional BLUP and single-step genomic BLUP, and computing performance was assessed for the latter method. For the 2 Holstein data sets, coefficients of determination (R(2)) and regression (δ) of deregressed evaluations from a full data set with records up to 2011 on estimated breeding values and genomic estimated breeding values from the truncated data sets were computed. The thresholds for data deletion were set by intervals of 5 yr, based on the average generation interval in dairy cattle. For the PIC data set, correlations between corrected phenotypes and estimated or genomic estimated breeding values were used to evaluate predictive ability on young animals born in 2010 and 2011. The reduced data set contained data up to 2009, and the thresholds were set based on an average generation interval of 3 yr. The number of generations that could be deleted without a reduction in accuracy depended on data structure and trait. For US Holsteins, removing 3 and 4 generations of data did not reduce accuracy of evaluations for final score in Af and A2 scenarios, respectively. For Israeli Holsteins, the accuracies for milk, fat, and protein yields were the highest when only phenotypes recorded in 2000 and later were included and full pedigrees were applied. Of the 135 Israeli bulls with genotypes (validation set) and daughter records only in the complete data set, 38 and 97 were sons of Israeli and foreign bulls, respectively. Although more phenotypic data increased the prediction accuracy for sons of Israeli bulls, the reverse was true for sons of foreign bulls. Also, more phenotypic data caused large inflation of genomic estimated breeding values for sons of foreign bulls, whereas the opposite was true with the deletion of all but the most recent phenotypic data. Results for protein and fat percentage were different from those for milk, fat, and protein yields; however, relatively, the changes in coefficients of determination and regression were smaller for percentage traits. For PIC data set, removing data from up to 5 generations did not erode predictive ability for genotyped animals for the 2 reproductive traits used in validation. Given the data used in this study, truncating old data reduces computation requirements but does not decrease the accuracy. For small populations that include local and imported animals, truncation may be beneficial for one group of animals and detrimental to another group., (Copyright © 2014 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

Currently, the US Department of Agriculture Animal Improvement Programs Laboratory utilizes a multi-step procedure in genomic evaluations for US Holstein bulls and cows, with adjustments for cows. We used a single-step procedure to investigate whether adding cows' genotypes could increase reliability in genomic breeding values for bulls while minimizing bias. The first data set to 2007 was used to calculate genomic estimated breeding values (GEBV) for animals, including young genotyped bulls with no daughters and young cows (heifers) with no records in 2007. The second data set to 2011 was used to calculate GEBV for the same animals, including those young bulls with daughters and young cows with records in 2011. Genotypes (42,503 single nucleotide polymorphism markers) for 34,506 bulls and 5,235 cows from 356,413 bulls and 9,245,619 cows in pedigree were used to calculate single-step GEBV (ssGEBV) and multi-step GEBV (msGEBV). Regression coefficients of 2007 GEBV on 2011 progeny deviations and coefficients of determination were used as indicators of bias and reliability in 2007 GEBV for bulls with no daughters and for cows with no records in 2007, using bull genotypes only and using bull and cow genotypes. Parent averages were also calculated from estimated breeding values of parents to compare with GEBV. For genotyped bulls, inflation was larger for ssGEBV than for msGEBV, whereas reliability was higher for ssGEBV. Using all genotyped bulls and cows, reliabilities were increased by 2 to 3%. Use of genotypes of high-profile cows improves reliability in ssGEBV and msGEBV for bulls., (Copyright © 2013 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

Effects on prediction of analysing a multi-line chicken population as one line were evaluated. Body weight records were provided by Cobb-Vantress for two lines of broiler chickens. Phenotypic records for 183 695 and 164 149 broilers and genotypic records for 3195 and 3001 broilers were available for each line. Lines were combined to create a multi-line population and analysed using a single-step procedure combining the additive relationship matrix and the genomic relationship matrix (G). G was scaled using allele frequencies from each line, the multi-line population, or 0.5. When allele frequencies were calculated from each line, distributions of diagonal elements were bimodal. When allele frequencies were calculated from the multi-line population, the distribution of diagonal elements had one peak. When allele frequency 0.5 was used, the distribution was bimodal. Genomic estimated breeding values (GEBVs) were predicted using each allele frequency. GEBVs differed with allele frequency but had ≥ 0.99 correlations with GEBVs predicted with correct allele frequencies. Means of each line and differences in mean between the lines differed based on allele frequencies. Assumed allele frequencies have little impact on ranking within line but larger impact on ranking across lines. G may be used to evaluate multiple populations simultaneously but must be adjusted to obtain properly scaled estimates when population structure is unknown., (© 2011 Blackwell Verlag GmbH.)

The purpose of this study was to find ways of reducing changes of sire predicted transmitting ability for type's final scores (PTATs) from the first to second crop of daughters. The PTATs were estimated from two datasets: D01 (scores recorded up to 2001) and D05 (scores recorded up to 2005). The PTAT changes were calculated as the difference between the evaluations based on D01 and D05. The PTATs were adjusted to a common genetic base of all evaluated cows born in 1995. The single-trait (ST) animal model included the fixed effects of the herd-year-season-classifier, age by year group at classification, stage of lactation at classification, registry status of animals, and additive genetic and permanent environment random effects. Unknown parent groups (UPGs) were defined based on every other birth year starting from 1972. Modifications to the ST model included the usage of a single record per cow, separate UPGs for first and second crop daughters, separate UPGs for sires and dams, and deepened pedigrees for dams with missing phenotypic records. Also, the multiple-trait (MT) model treated records of registered and grade cows as correlated traits. The mean PTAT change, for all of the sires, was close to zero in all of the models analyzed. The estimated mean PTAT change for 145 sires with 40 to 100 first crop and ≥ 200 second crop daughters was -0.33, -0.20, -0.13, -0.28, and -0.12 with ST, only first records, only last records, updated pedigrees, and allowing separate parent groups (PGs) for sires and dams after updating the pedigrees, respectively. The percentages of sires showing PTAT decline were reduced from 74.5 (with ST) to 57.3 by using only the last records of cows, and to 56.4 by allowing separate UPGs for sires and dams after updating the pedigrees. Though updating of the pedigrees alone was not effective, separate UPGs for sires together with additional pedigree was helpful in reducing the bias.

Genetic parameters for daily feed intake (DFI, g/day) and daily gain (DG, g/day) were estimated using records of 1916 Duroc boars from electronic feeder stations. Management was limited and resulted in varied ranges of age and weight on test. Boars were housed in 102 pens, each equipped with one feeder, and allowed ad libitum feeding. Weekly averages of DFI and DG were used due to large variation in daily records. Six traits were defined as DFI and DG during 85-106 (period 1), 107-128 (period 2) and 129-150 days of age (period 3). A six-trait model included age as a linear and a quadratic covariate for DFI and a linear covariate for DG with a fixed effect of year-week-pen and random effects of litter, additive genetic animal and permanent environmental animal. Variance components were estimated by a Bayesian approach using Gibbs sampling algorithm. Estimates of heritability for respective periods were 18%, 12% and 10% for DFI and 21%, 11% and 10% for DG. Genetic correlations between DFI and DG in the same period were 0.70, 0.73 and 0.32 for the respective periods. DFI and DG obtained from automatic feeders can be analysed to reveal variation across testing periods by using weekly averages when many monthly averages are incomplete.

The first national single-step, full-information (phenotype, pedigree, and marker genotype) genetic evaluation was developed for final score of US Holsteins. Data included final scores recorded from 1955 to 2009 for 6,232,548 Holsteins cows. BovineSNP50 (Illumina, San Diego, CA) genotypes from the Cooperative Dairy DNA Repository (Beltsville, MD) were available for 6,508 bulls. Three analyses used a repeatability animal model as currently used for the national US evaluation. The first 2 analyses used final scores recorded up to 2004. The first analysis used only a pedigree-based relationship matrix. The second analysis used a relationship matrix based on both pedigree and genomic information (single-step approach). The third analysis used the complete data set and only the pedigree-based relationship matrix. The fourth analysis used predictions from the first analysis (final scores up to 2004 and only a pedigree-based relationship matrix) and prediction using a genomic based matrix to obtain genetic evaluation (multiple-step approach). Different allele frequencies were tested in construction of the genomic relationship matrix. Coefficients of determination between predictions of young bulls from parent average, single-step, and multiple-step approaches and their 2009 daughter deviations were 0.24, 0.37 to 0.41, and 0.40, respectively. The highest coefficient of determination for a single-step approach was observed when using a genomic relationship matrix with assumed allele frequencies of 0.5. Coefficients for regression of 2009 daughter deviations on parent-average, single-step, and multiple-step predictions were 0.76, 0.68 to 0.79, and 0.86, respectively, which indicated some inflation of predictions. The single-step regression coefficient could be increased up to 0.92 by scaling differences between the genomic and pedigree-based relationship matrices with little loss in accuracy of prediction. One complete evaluation took about 2h of computing time and 2.7 gigabytes of memory. Computing times for single-step analyses were slightly longer (2%) than for pedigree-based analysis. A national single-step genetic evaluation with the pedigree relationship matrix augmented with genomic information provided genomic predictions with accuracy and bias comparable to multiple-step procedures and could account for any population or data structure. Advantages of single-step evaluations should increase in the future when animals are pre-selected on genotypes., (Copyright 2010 American Dairy Science Association. Published by Elsevier Inc. All rights reserved.)

The purpose of this study is to present guidelines in selection of statistical and computing algorithms for variance components estimation when computing involves software packages. For this purpose two major methods are to be considered: residual maximal likelihood (REML) and Bayesian via Gibbs sampling. Expectation-Maximization (EM) REML is regarded as a very stable algorithm that is able to converge when covariance matrices are close to singular, however it is slow. However, convergence problems can occur with random regression models, especially if the starting values are much lower than those at convergence. Average Information (AI) REML is much faster for common problems but it relies on heuristics for convergence, and it may be very slow or even diverge for complex models. REML algorithms for general models become unstable with larger number of traits. REML by canonical transformation is stable in such cases but can support only a limited class of models. In general, REML algorithms are difficult to program. Bayesian methods via Gibbs sampling are much easier to program than REML, especially for complex models, and they can support much larger datasets; however, the termination criterion can be hard to determine, and the quality of estimates depends on a number of details. Computing speed varies with computing optimizations, with which some large data sets and complex models can be supported in a reasonable time; however, optimizations increase complexity of programming and restrict the types of models applicable. Several examples from past research are discussed to illustrate the fact that different problems required different methods.

A method used to approximate reliabilities for animal models was modified to estimate reliabilities for sire-maternal grandsire (MGS) models. Accuracy of the approximation was tested on a calving-ease data set for 2,968 bulls for which the inverse of the coefficient matrix could be obtained. Correlations between estimated and true reliabilities ranged from 0.984 to 0.998 for first- and later-parity calving ease for sire and MGS effects. With no modification of the animal-model procedure, MGS identification was treated as if it were dam identification, which resulted in overestimated reliability. When pedigree information was ignored, reliability was underestimated. Correlations with true values were lower for both of those cases when compared with correct processing of MGS information. The modification provided a slight improvement over assuming MGS to be unknown and will be used for routine USDA evaluation of calving traits.

The objective of this study was to determine the suitability of 2 methods for computing approximate accuracies of predicted breeding values, in which accuracy was defined as the squared correlation between the predicted and true breeding value, when modeling growth traits in beef cattle using random regression (RR) models. The first method (Strabel et al., S-M-B) was designed for use with multitrait models; thus, its use with RR models requires the clustering of measurements into different traits. The second method (Tier and Meyer, T-M) was more general, because it accounted for random coefficients other than zeros and ones and thus it could be used directly when fitting RR models. To investigate the performance of both methods, their results were compared with the true accuracies using a balanced simulated data set. The largest difference between approximate and true average accuracies for direct effects was observed at 205 d when S-M-B was used (4.6% males and 8.8% females). With regard to maternal effects, the largest differences in average accuracies were observed at 205 d in males when S-M-B was used (31.8%) and at the same age in females but when using T-M (33.3%). In general, bias increased for direct effect accuracies in males at the tails of the accuracy range, but for females and for maternal effect accuracies in both sexes, bias increased as accuracy increased. When a population was simulated to create large numbers of progeny for base females that did not have individual records, much greater errors were observed in the regression of approximate values on the true ones. When both approximate methods were compared using a real beef cattle data set, a good agreement was observed, particularly for direct effect accuracies in sires [i.e., at 205 d, the regressions were 0.98 (direct) and 0.95 (maternal) with r(2) over 0.99]. The largest discrepancies for sires between the methods were observed at 205 d for direct (2.7%) and maternal (16.3%) effect accuracies. For dams, the largest differences between methods were also observed at 205 d, 9.3% (direct), and 15.2% (maternal). The differences between methods for nonparent cattle were greater than for dams for maternal effect accuracies but intermediate between sires and dams for direct effect accuracies. In spite of the less biased results provided by T-M, its use could be problematic when employed in evaluations of large populations due to its greater memory and computation requirements (e.g., 170 and 478% more than S-M-B for a population of 11 million).

The objective of this study was to examine the feasibility of using random regression-spline (RR-spline) models for fitting growth traits in a multibreed beef cattle population. To meet the objective, the results from the RR-spline model were compared with the widely used multitrait (MT) model when both were fit to a data set (1.8 million records and 1.1 million animals) provided by the American Gelbvieh Association. The effect of prior information on the EBV of sires was also investigated. In both RR-spline and MT models, the following effects were considered: individual direct and maternal additive genetic effects, contemporary group, age of the animal at measurement, direct and maternal heterosis, and direct and maternal additive genetic mean effect of the breed. Additionally, the RR-spline model included an individual direct permanent environmental effect. When both MT and RR-spline models were applied to a data set containing records for weaning weight (WWT) and yearling weight (YWT) within specified age ranges, the rankings of bulls' direct EBV (as measured via Pearson correlations) provided by both models were comparable, with slightly greater differences in the reranking of bulls observed for YWT evaluations (>or=0.99 for BWT and WWT and >or=0.98 for YWT); also, some bulls dropped from the top 100 list when these lists were compared across methods. For maternal effects, the estimated correlations were slightly smaller, particularly for YWT; again, some drops from the top 100 animals were observed. As in regular MT multibreed genetic evaluations, the heterosis effects and the additive genetic effects of the breed could not be estimated from field data, because there were not enough contemporary groups with the proper composition of purebred and crossbred animals; thus, prior information based on literature values had to be included. The inclusion of prior information had a negligible effect in the overall ranking for bulls with greater than 20 birth weight progeny records; however, the effect of prior information for breeds or groups poorly represented in the data was important. The Pearson correlations for direct and maternal WWT and YWT ranged from 0.95 to 0.98 when comparing evaluations of data sets for which the out-of-range age records were removed or retained. Random regression allows for avoiding the discarding of records that are outside the usual age ranges of measurement; thus, greater accuracies are achieved, and greater genetic progress could be expected.

The aim of this study was to investigate the possible superiority of a threshold-linear (TL) approach for calving day (CD) and calving success (CS) analysis in beef cattle over 2 multiple-trait (MT), censored models, considering CD at the first 3 calving opportunities. The CD observations on animals that failed to calve in the latter models were defined as cows being assigned a penalty value of 21 d beyond the last observed CD record within contemporary group (PEN model) or censored CD values that were randomly obtained from a truncated normal distribution (CEN-model). In the TL model, CD records were treated as missing if a cow failed to calve, and parameters were estimated in a TL analysis including CS traits (TLMISS-model). The models included the effects of contemporary group (herd x year of calving x mating management), age at calving, physiological status at mating (lactating or nonlactating cow), animal additive genetic effects, and residual. Field data included 6,763 calving records obtained from first, second, and third parities of 3,442 spring-calving Uruguayan Aberdeen Angus cows. Models were contrasted using a data splitting technique, analyzing correlations between predicted breeding values (PBV) for each pair of subsamples, by rank correlations between PBV obtained with the different models, and by inspecting percentage of sires selected in common using the different approaches at 10 and 25% hypothetical percentages of animals selected. Breeding value correlations of CD between the subsamples for the TLMISS approach were greater (0.67 to 0.68) than correlations for the censored MT models (0.49 to 0.54). Average correlations between PBV of CD in 1 subsample obtained by CEN (PEN, TLMISS) and PBV of CS in the other subsample were -0.53 (-0.55, -0.60) in the first calving opportunity (CO), -0.54 (-0.58, -0.63) in the second CO, and -0.50 (-0.49, -0.58) in the third CO. Rank correlations between PBV for CD in PEN and CEN were high (0.93 to 0.97), but correlations of either method with PBV of CD in TLMISS ranged from 0.50 to 0.71. Common identification of bulls for the top 10% of sires (25% of sires), when selected with PEN/CEN models or the TLMISS model, varied between 50 (44%) and 60 (52%). The use of the TL animal model for genetic evaluation seems attractive for genetic evaluation of fertility traits in beef cattle.

A multi-breed model was presented for the genetic evaluation of growth traits in beef cattle. In addition to the fixed effects, random direct and maternal genetic effects, and random maternal permanent environmental effects are considered; the model also fits direct and maternal heterosis and direct and maternal breed-of-founder (BOF) x generation group effects using a Bayesian approach that weights prior literature estimates relative to information supplied by the dataset to which the model will be applied. The multi-breed evaluation procedures also allow the inclusion of external evaluations for animals of other breeds. The multi-breed model was applied to a dataset provided by the American Gelbvieh Association. Different analyses were conducted by varying the weights given to the prior literature relative to the information provided by the dataset. Large differences were observed for the heterosis estimates, the BOF x generation group effect estimates, and the predicted breeding values across breeds due to the weights posed on prior literature estimates versus estimates derived directly from data. However, the rankings within breed were observed to be relatively robust to the different weights on prior information.

The objective of this study was to assess breed, and breed x region interactions for several longevity-related traits, measured up to 5 yr of age in Brown Swiss, Holstein, and Jersey cows in 7 regions of the United States. Data were analyzed using logistic, poisson, and linear models, and survival analyses. The traits were stayability (yes/no survived to 5 yr of age), number of completed lactations, days lived, herd-life, and days in milk (DIM) to 5 yr of age. Probable lifetime DIM were also estimated using data from the first 5 yr of age of the cows. Herd-life was defined as the days lived up to 5 yr of age minus the age at first calving. Days in milk consisted of herd-life up to 5 yr of age minus the dry periods. Three data files were analyzed: herds with one breed of cows, herds with Holstein and Brown Swiss, and herds with Holstein and Jersey cows. Breed x region interaction was usually significant, with larger effects for the southern regions. Jerseys obtained largest values for the ratio of DIM to days lived, and for the number of completed lactations to 5 yr of age. Brown Swiss had the largest probabilities of surviving to 5 yr of age (stayabilities) in all regions. For the other traits, the results for Brown Swiss were inconsistent, but usually the cows of this breed had shorter herd-life and DIM to 5 yr of age than Holsteins. Brown Swiss cows were expected to have more total DIM in their lifetime in the Southeast than Holsteins. Survival analysis gave the most readily interpretable information, although the linear, poisson, and logistic analyses answered slightly different questions. Adjustment for herd size did not modify the results.

Genetic evaluation of growth in Gelbvieh beef cattle was examined by multiple-trait (MTM) and random regression (RRM) analysis. The data set comprised 541,108 animals with 1,120,086 records. Approximately 15% of the animals in the data set had at least one record measured outside of the accepted MTM age ranges for weaning weight (Wwt) and yearling weight (Ywt). Fourteen percent of Wwt records and 19% of Ywt records were measured outside the accepted ranges for MTM analysis, and thus were excluded from MTM evaluations. Two RRM evaluations were performed using cubic Legendre polynomials (RRML) and linear splines (RRMS) with three knots at 1, 205, and 365 d of age. Data Set 1 (d1) utilized all available records, whereas Data Set 2 (d2) included only records measured within MTM ranges (1 d, 160 to 250 d, and 320 to 410 d). The RRML models did not reach convergence until diagonalization was imposed. After diagonalization, it was found that all longitudinal models required fewer iterations to converge than the MTM. Correlations between the MTM, RRML-d2, and RRMS-d2 evaluations were >or=0.99 for all three traits, indicating that these models were equivalent when predicting breeding values from data within the MTM age ranges. Correlations between MTM, RRML-d1, and RRMS-d1 were >0.99 for Bwt and >0.95 for Wwt and Ywt. The lower correlations for Wwt and Ywt indicate that the added information does affect breeding value prediction. The RRM has the capability to incorporate records measured at all ages into genetic evaluations at a computing cost similar to the MTM.

A simulation study examined issues important for genetic evaluation of growth in beef cattle by random regression models with cubic Legendre polynomials (RRML) and linear splines with three knots (RRMS) compared with multiple-trait models (MTM). Parameters for RRML were obtained by conversion from covariance functions. Parameters for MTM and RRMS were extracted from RRML at 1, 205, and 365 d; parameters for RRMS were the same as MTM for all effects except the permanent environment and the residual. Four data sets were generated assuming RRML included records at 1, 205, and 365 d; at 1, 160 to 250, and 320 to 410 d; at 1, 100, 205, 300, and 365 d; and at 1, 55 to 145, 160 to 250, 275 to 325, and 320 to 410 d. Accuracies were computed as correlations between the true (simulated) and predicted breeding values. With the first data set, excellent agreement in accuracy was obtained for all models. With the second data set, the accuracy of MTM dropped by up to 1.5% compared with the first data set, but accuracy was unchanged for both RRML and RRMS. With the third (fourth) data set, accuracies of RRML were up to 2.4% (2.5%) higher than with the first (second) data set. Small differences in accuracy between RRML and RRMS were found with the third and fourth data sets, which were traced to inflated correlations especially between 1 and 205 d in RRMS; inflation could be decreased by adding one extra knot at 100 d to RRMS. Diagonalization of random coefficients was crucial for RRML but not for RRMS, resulting in approximately six (two) times faster convergence with RRML (RRMS). Reduction of dimensionality in RRML associated with small eigenvalues caused a less accurate evaluation for birth weight. Genetic evaluation of growth by RRM requires careful implementation. The RRMS is simpler to implement than the RRML.

Mating and calving records for 47,533 first-calf heifers in Australian Angus herds were used to examine the relationship between days to calving (DC) and two measures of fertility in AI data: 1) calving to first insemination (CFI) and 2) calving success (CS). Calving to first insemination and calving success were defined as binary traits. A threshold-linear Bayesian model was employed for both analyses: 1) DC and CFI and 2) DC and CS. Posterior means (SD) of additive covariance and corresponding genetic correlation between the DC and CFI were -0.62 d (0.19 d) and -0.66 (0.12), respectively. The corresponding point estimates between the DC and CS were -0.70 d (0.14 d) and -0.73 (0.06), respectively. These genetic correlations indicate a strong, negative relationship between DC and both measures of fertility in AI data. Selecting for animals with shorter DC intervals genetically will lead to correlated increases in both CS and CFI. Posterior means (SD) for additive and residual variance and heritability for DC for the DC-CFI analysis were 23.5 d2 (4.1 d2), 363.2 d2 (4.8 d2), and 0.06 (0.01), respectively. The corresponding parameter estimates for the DC-CS analysis were very similar. Posterior means (SD) for additive, herd-year and service sire variance and heritability for CFI were 0.04 (0.01), 0.06 (0.06), 0.14 (0.16), and 0.03 (0.01), respectively. Posterior means (SD) for additive, herd-year, and service sire variance and heritability for CS were 0.04 (0.01), 0.07 (0.07), 0.14 (0.16), and 0.03 (0.01), respectively. The similarity of the parameter estimates for CFI and CS suggest that either trait could be used as a measure of fertility in AI data. However, the definition of CFI allows the identification of animals that not only record a calving event, but calve to their first insemination, and the value of this trait would be even greater in a more complete dataset than that used in this study. The magnitude of the correlations between DC and CS-CFI suggest that it may be possible to use a multitrait approach in the evaluation of AI and natural service data, and to report one genetic value that could be used for selection purposes.

Mating and calving records for 51,084 first-parity heifers in Australian Angus herds were used to examine the relationship between probability of calving to first insemination (CFI) in artificial insemination and natural service (NS) mating data. Calving to first insemination was defined as a binary trait for both sources of data. Two Bayesian models were employed: 1) a bivariate threshold model with CFI in AI data regarded as a trait separate from CFI in NS data and 2) a univariate threshold model with CFI regarded as the same trait for both sources of data. Posterior means (SD) of additive variance in the bivariate analysis were similar: 0.049 (0.013) and 0.075 (0.021) for CFI in AI and NS data, respectively, indicating lack of heterogeneity for this parameter. A similar trend was observed for heritability in the bivariate analysis, with posterior means (SD) of 0.025 (0.007) and 0.048 (0.012) for AI and NS data, respectively. The posterior means (SD) of the additive covariance and corresponding genetic correlation between the traits were 0.048 (0.006) and 0.821 (0.138), respectively. Differences were observed between posterior means for herd-year variance: 0.843 vs. 0.280 for AI and NS data, respectively, which may reflect the higher incidence of 100% conception rates within a herd-year class (extreme category problem) in AI data. Parameter estimates under the univariate model were close to the weighted average of the corresponding parameters under the bivariate model. Posterior means (SD) for additive, herd-year, and service sire variance and heritability under the univariate model were 0.063 (0.007), 0.56 (0.029), 0.131 (0.013), and 0.036 (0.007), respectively. These results indicate that, genetically, cows with a higher probability of CFI when mated using AI also have a high probability of CFI when mated via NS. The high correlation between the two traits, along with the lack of heterogeneity for the additive variance, implies that a common additive variance could be used for AI and NS data. A single-trait analysis of CFI with heterogeneous variances for herd-year and service sire could be implemented. The low estimates of heritability indicate that response to selection for probability of calving to first insemination would be expected to be low.

The objectives of this study were to establish a pattern for the seasonality of days open (DO) by state and region within the United States and to present statistics on regional trends for DO. Data included 8,676,915 records on DO for Holsteins from 1997 to 2002 covering all regions of the United States. Fixed effects in the model included herd, parity, milk-class, state x month of calving (MOC), year of calving x MOC, and parity x MOC. Least squares means of DO were highest for calvings in March and lowest for calvings in September. The highest mean DO of 155 d was recorded in the Southeast, while the mean DO for the Midwest, Northeast, Northwest, and Southwest were 142, 141, 140, and 137 d, respectively. Variation in monthly averages of DO was highest in Southeast with a range of 51 d, and less than 25 d in all the other regions. Seasonality of calving was defined as the ratio of the fewest to the most calvings in months. The SOC was > or = 60% in Southeast and < or = 23% in the other regions. Selected states: Texas, Oklahoma, and Arizona in the Southwest and Missouri, Kansas, and Kentucky in the Midwest showed patterns of variation in monthly averages and seasonality of calving similar to those of Southeast. Distributions of DO were bimodal for some months of calving due to postponed breeding during the hot season or depressed fertility as a result of thermal stress; the second mode at > 200 d was highest in the Southeast but also could be observed in Texas, Wisconsin, and California. High level of heat stress for DO exists in the Southeast and in selected states of the Midwest and the Southwest; these regions contribute less than 10% of national records. A methodology for analyzing DO especially under heat stress needs to consider effects of intentionally delayed breeding--by using a model that accounts for bimodality, for example.

The objective of this study was to add a maternal grandsire (MGS) effect to the existing sire model for national calving ease genetic evaluations. The Animal Improvement Programs Laboratory (AIPL) of USDA assumed responsibility for conducting the national genetic evaluation for calving ease and maintaining the associated database in 1999. Existing evaluations used a sire threshold model. Adding an MGS effect to the model was expected to improve accuracy by partially accounting for merit of mates and differences in maternal ability of the dams. Dystocia data were migrated to a relational database integrated with the AIPL production database. This database design allowed more rigorous data edits by comparison with the production data and improved MGS identification (ID) rate by utilizing pedigrees from the production records. Integration of dystocia data with production data increased MGS ID rate from 58 to 73%. In addition, nearly 200,000 duplicate records were identified using the new edit system. Sire and sire-MGS models were compared using over 10 million observations available for the August 2002 national genetic evaluation. The sire model included herd-year, season, sex of calf, parity of dam, birth year group of sire, and sire. For the sire-MGS model, MGS and birth year group of MGS were added, year-seasons rather than seasons were used, and sex of calf and parity of dam were combined into a single interaction effect. Herd-year, sire, and MGS were random effects. Variance components used for the sire model were those previously used in the national evaluation and for the sire-MGS model were estimated in a separate study. Correlations between predicted genetic merits for service sire calving ease from the two models was 85%, indicating general agreement, but with some significant differences in evaluations. A sire-MGS model was implemented in August 2002 for the national calving ease genetic evaluation system.

The objective of this study was to identify issues in genetic evaluation of beef cattle for growth by a random regression model (RRM). Genetic evaluation data included 2,946,847 records of up to nine sequential weights of 812,393 Nellore cattle measured at ages ranging from birth to 733 d. Models considered were a five-trait multiple-trait model (MTM) and a cubic RRM. The MTM included the effects of contemporary group, age of dam class, additive direct, additive maternal, and maternal permanent environment. Both additive effects were assumed correlated. The RRM included the same effects as MTM, with the addition of permanent and random error effects. The purpose of the random error effect, which was in addition to a residual effect with constant variance, was to model heterogeneous residual variances. All effects in RRM were modeled as cubic Legendre polynomials. Expected progeny differences (EPD) were obtained iteratively using a preconditioned conjugate gradient algorithm. Numerically accurate solutions with RRM were not obtained until the random regressions were orthogonalized. Computing requirements of RRM were reduced by more than 50%, without affecting the accuracy by removing regressions corresponding to very low eigen-values and by replacing the random error effects with weights. Afterward, the correlations between EPD from RRM and from MTM for EPD on selected weights were between 0.84 and 0.89. For sires with at least 50 progeny, these correlations increased to 0.92 to 0.97. Low correlations were caused by differences in parameters. The RRM applied to growth i s prone to numerical problems. Estimates of EPD with RRM may be more accurate than those with MTM only if accurate parameters are applied.

Conformation final scores obtained by the Holstein Association were used in this study. Five subsets were sampled from the original complete data, which contained about 7 million records, and (co)variance components for additive genetic, permanent environmental, and residual effects were estimated with a multiple-trait (MT) model at different ages of cows and with a random regression model. Transmitting abilities were predicted with the random-regression (RR) model and with the repeatability model for the national evaluation, using the entire data set. The RR model included fixed and random regressions on age at classification for additive genetic and permanent environmental effects. Additive genetic and residual variances estimated with the MT model increased with aging. Genetic correlations between age groups decreased with their distance. Heritability and repeatability estimates with the RR model were 0.27 and 0.88 on average, increasing with cow's age. Correlations between transmitting abilities predicted with the RR model at various ages and with the repeatability model were in the range of 0.89 to 0.99. In the 1980s, predicted transmitting abilities (PTA) of sires increased with their daughters' age, but in 1991 and 1995, the PTA decreased. In general, sires whose daughters were evaluated at mature ages tended to have higher PTA. For sires and cows, genetic trends estimated from the repeatability model and from the RR model were similar. The low genetic trend of cows at 60 mo for recent years was due to many grade cows that had only single records and little pedigree information.

Data included 393,097 calving ease, 129,520 gestation length, and 412,484 birth weight records on 412,484 Gelbvieh cattle. Additionally, pedigrees were available on 72,123 animals. Included in the models were effects of sex and age of dam, treated as fixed, as well as direct, maternal genetic and permanent environmental effects and effects of contemporary group (herd-year-season), treated as random. In all analyses, birth weight and gestation length were treated as continuous traits. Calving ease (CE) was treated either as a continuous trait in a mixed linear model (LM), or as a categorical trait in linear-threshold models (LTM). Solutions in TM obtained by empirical Bayes (TMEB) and Monte Carlo (TMMC) methodologies were compared with those by LM. Due to the computational cost, only 10,000 samples were obtained for TMMC. For calving ease, correlations between LM and TMEB were 0.86 and 0.78 for direct and maternal genetic effects, respectively. The same correlations but between TMEB and TMMC were 1.00 and 0.98, respectively. The correlations between LM and TMMC were 0.85 and 0.75, respectively. The correlations for the linear traits were above.97 between LM and TMEB but as low as 0.91 between LM and TMMC, suggesting insufficient convergence of TMMC. Computing time required was about 2 hrs, 5 hrs, and 6 days for LM, TMEB and TMMC, respectively, and memory requirements were 169, 171, and 445 megabytes, respectively. Bayesian implementation of threshold model is simple, can be extended to multiple categorical traits, and allows easy calculation of accuracies; however, computing time is prohibitively long for large models.

Bayesian analysis via Gibbs sampling, restricted maximum likelihood (REML), and Method R were used to estimate variance components for several models of simulated data. Four simulated data sets that included direct genetic effects and different combinations of maternal, permanent environmental, and dominance effects were used. Parents were selected randomly, on phenotype across or within contemporary groups, or on BLUP of genetic value. Estimates by Bayesian analysis and REML were always empirically unbiased in large data sets. Estimates by Method R were biased only with phenotypic selection across contemporary groups; estimates of the additive variance were biased upward, and all the other estimates were biased downward. No empirical bias was observed for Method R under selection within contemporary groups or in data without contemporary group effects. The bias of Method R estimates in small data sets was evaluated using a simple direct additive model. Method R gave biased estimates in small data sets in all types of selection except BLUP. In populations where the selection is based on BLUP of genetic value or where phenotypic selection is practiced mostly within contemporary groups, estimates by Method R are likely to be unbiased. In this case, Method R is an alternative to single-trait REML and Bayesian analysis for analyses of large data sets when the other methods are too expensive to apply.