Assigning unknown parent groups to reduce bias in genomic evaluations of final score in US Holsteins

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.     

Accuracy of genomic prediction for milk production traits in the Chinese Holstein population using a reference population consisting of cows

Genomic selection using dense markers covering the whole genome is a tool for the genetic improvement of livestock and is revolutionizing the breeding system in dairy cattle. Progeny-tested bulls have been used to form reference populations in almost all countries where genomic selection has been implemented. In this study, the accuracy of genomic prediction when cows are used to form the reference population was investigated. The reference population consisted of 3,087 cows. All individuals were genotyped with Illumina BovineSNP50. After genotype imputation and editing, 48,676 single nucleotide polymorphisms were available for analysis. Two methods, genomic BLUP (GBLUP) and BayesB, were used to render genomic estimated breeding values (GEBV) for 5 milk production traits. Accuracies of GEBV were assessed in 3 ways: r_GEBV,EBV (the correlation between GEBV and conventional EBV) in 67 progeny-tested bulls, r_GEBV,EBV from a 5-fold cross validation in the 3,087 cow reference population, and the theoretical accuracy (for GBLUP) calculated in the same way as for conventional BLUP. The results showed that using GBLUP, the r_GEBV,EBV and theoretical accuracy of genomic prediction in Chinese Holstein ranged from 0.59 to 0.76 and 0.70 to 0.80, respectively, which was 0.13 to 0.30 and 0.23 to 0.33 higher than the accuracies of conventional pedigree index, respectively. The results indicate that, as an alternative, genomic selection using cows in the reference population is feasible.     

Genomic predictions for New Zealand dairy bulls and integration with national genetic evaluation

A method is described for the prediction of breeding values incorporating genomic information. The first stage involves the prediction of genomic breeding values for genotyped individuals. A novel component of this is the estimation of the genomic relationship matrix in the context of a multi-breed population. Because not all ancestors of genotyped animals are genotyped, a selection index procedure is used to blend genomic predictions with traditional ancestral information that is lost between the process of deregression of the national breeding values and subsequent re-estimation using the genomic relationship matrix. Finally, the genomically enhanced predictions are filtered through to nongenotyped descendants using a regression procedure.     

Multiple-trait genomic evaluation of linear type traits using genomic and phenotypic data in US Holsteins

Currently, the USDA uses a single-trait (ST) model with several intermediate steps to obtain genomic evaluations for US Holsteins. In this study, genomic evaluations for 18 linear type traits were obtained with a multiple-trait (MT) model using a unified single-step procedure. The phenotypic type data on up to 18 traits were available for 4,813,726 Holsteins, and single nucleotide polymorphism markers from the Illumina BovineSNP50 genotyping Beadchip (Illumina Inc., San Diego, CA) were available on 17,293 bulls. Genomic predictions were computed with several genomic relationship matrices (G) that assumed different allele frequencies: equal, base, current, and current scaled. Computations were carried out with ST and MT models. Procedures were compared by coefficients of determination (R²) and regression of 2004 prediction of bulls with no daughters in 2004 on daughter deviations of those bulls in 2009. Predictions for 2004 also included parent averages without the use of genomic information. The R² for parent averages ranged from 10 to 34% for ST models and from 12 to 35% for MT models. The average R² for all G were 34 and 37% for ST and MT models, respectively. All of the regression coefficients were <1.0, indicating that estimated breeding values in 2009 of 1,307 genotyped young bulls’ parents tended to be biased. The average regression coefficients ranged from 0.74 to 0.79 and from 0.75 to 0.80 for ST and MT models, respectively. When the weight for the inverse of the numerator relationship matrix (A⁻¹) for genotyped animals was reduced from 1 to 0.7, R² remained almost identical while the regression coefficients increased by 0.11-0.26 and 0.12-0.23 for ST and MT models, respectively. The ST models required about 5 s per iteration, whereas MT models required 3 (6) min per iteration for the regular (genomic) model. The MT single-step approach is feasible for 18 linear type traits in US Holstein cattle. Accuracy for genomic evaluation increases when switching ST models to MT models. Inflation of genomic evaluations for young bulls could be reduced by choosing a small weight for the A⁻¹ for genotyped bulls.     

Modeling missing pedigree in single-step genomic BLUP

The objective was to compare methods of modeling missing pedigree in single-step genomic BLUP (ssGBLUP). Options for modeling missing pedigree included ignoring the missing pedigree, unknown parent groups (UPG) based on A (the numerator relationship matrix) or H (the unified pedigree and genomic relationship matrix), and metafounders. The assumptions for the distribution of estimated breeding values changed with the different models. We simulated data with heritabilities of 0.3 and 0.1 for dairy cattle populations that had more missing pedigrees for animals of lesser genetic merit. Predictions for the youngest generation and UPG solutions were compared with the true values for validation. For both traits, ssGBLUP with metafounders provided accurate and unbiased predictions for young animals while also appropriately accounting for genetic trend. Accuracy was least and bias was greatest for ssGBLUP with UPG for H for the trait with heritability of 0.3 and with UPG for A for the trait with heritability of 0.1. For the trait with heritability of 0.1 and UPG for H, the UPG accuracy (SD) was ?0.49 (0.12), suggesting poor estimates of genetic trend despite having little bias for validations on young, genotyped animals. Problems with UPG estimates were likely caused by the lesser amount of information available for the lower heritability trait. Hence, UPG need to be defined differently based on the trait and amount of information. More research is needed to investigate accounting for UPG in A₂₂ to better account for missing pedigrees for genotyped animals.     

Short communication: Validation of genomic breeding value predictions for feed intake and feed efficiency traits

Validating genomic prediction equations in independent populations is an important part of evaluating genomic selection. Published genomic predictions from 2 studies on (1) residual feed intake and (2) dry matter intake (DMI) were validated in a cohort of 78 multiparous Holsteins from Australia. The mean realized accuracy of genomic prediction for residual feed intake was 0.27 when the reference population included phenotypes from 939 New Zealand and 843 Australian growing heifers (aged 5–8 mo) genotyped on high density (770k) single nucleotide polymorphism chips. The 90% bootstrapped confidence interval of this estimate was between 0.16 and 0.36. The mean realized accuracy was slightly lower (0.25) when the reference population comprised only Australian growing heifers. Higher realized accuracies were achieved for DMI in the same validation population and using a multicountry model that included 958 lactating cows from the Netherlands and United Kingdom in addition to 843 growing heifers from Australia. The multicountry analysis for DMI generated 3 sets of genomic predictions for validation animals, one on each country scale. The highest mean accuracy (0.72) was obtained when the genomic breeding values were expressed on the Dutch scale. Although the validation population used in this study was small (n = 78), the results illustrate that genomic selection for DMI and residual feed intake is feasible. Multicountry collaboration in the area of dairy cow feed efficiency is the evident pathway to achieving reasonable genomic prediction accuracies for these valuable traits.     

Genomic and pedigree-based genetic parameters for scarcely recorded traits when some animals are genotyped

Genetic parameters were estimated using relationships between animals that were based either on pedigree, 43,011 single nucleotide polymorphisms, or a combination of these, considering genotyped and non-genotyped animals. The standard error of the estimates and a parametric bootstrapping procedure was used to investigate sampling properties of the estimated variance components. The data set contained milk yield, dry matter intake and body weight for 517 first-lactation heifers with genotypes and phenotypes, and another 112 heifers with phenotypes only. Multivariate models were fitted using the different relationships in ASReml software. Estimates of genetic variance were lower based on genomic relationships than using pedigree relationships. Genetic variances from genomic and pedigree relationships were, however, not directly comparable because they apply to different base populations. Standard errors indicated that using the genomic relationships gave more accurate estimates of heritability but equally accurate estimates of genetic correlation. However, the estimates of standard errors were affected by the differences in scale between the 2 relationship matrices, causing differences in values of the genetic parameters. The bootstrapping results (with genetic parameters at the same level), confirmed that both heritability and genetic correlations were estimated more accurately with genomic relationships in comparison with using the pedigree relationships. Animals without genotype were included in the analysis by merging genomic and pedigree relationships. This allowed all phenotypes to be used, including those from non-genotyped animals. This combination of genomic and pedigree relationships gave the most accurate estimates of genetic variance. When a small data set is available it might be more advantageous for the estimation of genetic parameters to genotype existing animals, rather than collecting more phenotypes.     

Relationships between reproductive traits of heifers and cows and yield traits for Holsteins in Japan

The objective of this study was to investigate relationships between reproductive traits in heifers and cows and yield traits for Holsteins in Japan. Insemination and lactation records for cows calved between 1990 and 2003 in Hokkaido region were obtained. Age at first service, age at conception, and conception rate for first service were calculated for heifers. Days from calving to first service, days open, and conception rate for first service were calculated for first- and second-parity cows. The yield traits used were 305-d milk, fat, and protein yields. A threshold animal model was applied for the conception rate for first service, and a linear animal model was applied for the other traits. Single-trait and 2-trait genetic analyses were performed by the Bayesian method using Gibbs sampling. Heritability estimates ranged from 0.027 to 0.051 for conception rate for first service, and from 0.074 to 0.128 for the other reproductive traits. If the relationships of other traits were not considered, days from calving to first service was favorable to genetic selection for reproductive traits because of relatively high heritability and because it can be available earlier than the days open. Genetic correlations among reproductive traits were high, especially in cows. The genetic correlations between reproductive traits for heifers and those for cows were lower than the genetic correlations between reproductive traits for first parity and those of second parity, suggesting that reproductive traits for heifers should be evaluated separately from reproductive traits for cows. Genetic correlations between yield and reproductive traits in cows were antagonistic. In contrast, genetic correlations between reproductive traits for heifers and yield traits were slightly desirable. Depending on the reporting rate of insemination records for heifers and the results of investigations for relationships with productive maturity, selection by reproductive traits for heifers will enable the improvement of reproductive performance without a loss in genetic progress for yield traits.     

Comparison of genomic predictions using genomic relationship matrices built with different weighting factors to account for locus-specific variances

Various models have been used for genomic prediction. Bayesian variable selection models often predict more accurate genomic breeding values than genomic BLUP (GBLUP), but GBLUP is generally preferred for routine genomic evaluations because of low computational demand. The objective of this study was to achieve the benefits of both models using results from Bayesian models and genome-wide association studies as weights on single nucleotide polymorphism (SNP) markers when constructing the genomic matrix (G-matrix) for genomic prediction. The data comprised 5,221 progeny-tested bulls from the Nordic Holstein population. The animals were genotyped using the Illumina Bovine SNP50 BeadChip (Illumina Inc., San Diego, CA). Weighting factors in this investigation were the posterior SNP variance, the square of the posterior SNP effect, and the corresponding minus base-10 logarithm of the marker association P-value [−log₁₀(P)] of a t-test obtained from the analysis using a Bayesian mixture model with 4 normal distributions, the square of the estimated SNP effect, and the corresponding −log₁₀(P) of a t-test obtained from the analysis using a classical genome-wide association study model (linear regression model). The weights were derived from the analysis based on data sets that were 0, 1, 3, or 5 yr before performing genomic prediction. In building a G-matrix, the weights were assigned either to each marker (single-marker weighting) or to each group of approximately 5 to 150 markers (group-marker weighting). The analysis was carried out for milk yield, fat yield, protein yield, fertility, and mastitis. Deregressed proofs (DRP) were used as response variables to predict genomic estimated breeding values (GEBV). Averaging over the 5 traits, the Bayesian model led to 2.0% higher reliability of GEBV than the GBLUP model with an original unweighted G-matrix. The superiority of using a GBLUP with weighted G-matrix over GBLUP with an original unweighted G-matrix was the largest when using a weighting factor of posterior variance, resulting in 1.7 percentage points higher reliability. The second best weighting factors were −log₁₀ (P-value) of a t-test corresponding to the square of the posterior SNP effect from the Bayesian model and −log₁₀ (P-value) of a t-test corresponding to the square of the estimated SNP effect from the linear regression model, followed by the square of estimated SNP effect and the square of the posterior SNP effect. In addition, group-marker weighting performed better than single-marker weighting in terms of reducing bias of GEBV, and also slightly increased prediction reliability. The differences between weighting factors and scenarios were larger in prediction bias than in prediction accuracy. Finally, weights derived from a data set having a lag up to 3 yr did not reduce reliability of GEBV. The results indicate that posterior SNP variance estimated from a Bayesian mixture model is a good alternative weighting factor, and common weights on group markers with a size of 30 markers is a good strategy when using markers of the 50,000-marker (50K) chip. In a population with gradually increasing reference data, the weights can be updated once every 3 yr.     

Indirect genomic predictions for milk yield in crossbred Holstein-Jersey dairy cattle

The objective of this study was to predict genomic breeding values for milk yield of crossbred dairy cattle under different scenarios using single-step genomic BLUP (ssGBLUP). The data set included 13,880,217 milk yield measurements on 6,830,415 cows. Genotypes of 89,558 Holstein, 40,769 Jersey, and 22,373 Holstein-Jersey crossbred animals were used, of which all Holstein, 9,313 Jersey, and 1,667 crossbred animals had phenotypic records. Genotypes were imputed to 45K SNP markers. The SNP effects were estimated from single-breed evaluations for Jersey (JE), Holstein (HO) and crossbreds (CROSS), and multibreed evaluations including all Jersey and Holstein (JE_HO) or approximately equal proportions of Jersey, Holstein, and crossbred animals (MIX). Indirect predictions (IP) of the validation animals (358 crossbred animals with phenotypes excluded from evaluations) were calculated using the resulting SNP effects. Additionally, breed proportions (BP) of crossbred animals were applied as a weight when IP were estimated based on each pure breed. The predictive ability of IP was calculated as the Pearson correlation between IP and phenotypes of the validation animals adjusted for fixed effects in the model. Regression of adjusted phenotypes on IP was used to assess the inflation of IP. The predictive ability of IP for CROSS, JE, HO, JE_HO, and MIX scenario was 0.50, 0.50, 0.47, 0.50, and 0.46, respectively. Using BP was the least successful, with a predictive ability of 0.32. The inflation of the IP for crossbred animals using CROSS, JE, HO, JE_HO, MIX, and BP scenarios were 1.17, 0.65, 0.55, 0.78, 1.00, and 0.85, respectively. The IP of crossbred animals can be predicted using single-step GBLUP under a scenario that includes purebred genotypes.     

Validation of genomic predictions for wellness traits in US Holstein cows

The objective of this study was to evaluate the efficacy of wellness trait genetic predictions in commercial herds of US Holstein cows from herds that do not contribute phenotypic information to the evaluation. Tissue samples for DNA extraction were collected from more than 3,400 randomly selected pregnant Holstein females in 11 herds and 2 age groups (69% nulliparous, 31% primiparous) approximately 30 to 60 d before their expected calving date. Lactation records from cows that calved between September 1, 2015, and December 31, 2015, were included in the analysis. Genomically enhanced predicted transmitting abilities for the wellness traits of retained placenta, metritis, ketosis, displaced abomasum, mastitis, and lameness were estimated by the Zoetis genetic evaluation and converted into standardized transmitting abilities. Mean reliabilities of the animals in the study ranged between 45 and 47% for each of the 6 traits. Animals were ranked by their standardized transmitting abilities within herd and age group then assigned to 1 of 4 groups of percentile-based genetic groups of equal size. Adverse health events, including retained placenta, metritis, ketosis, displaced abomasum, mastitis, and lameness, were collected from on-farm herd management software, and animal phenotype was coded as either healthy (0), diseased (1), or excluded for each of the 6 outcomes of interest. Statistical analysis was performed using a generalized linear mixed model with genetic group, age group, and lactation as fixed effects, whereas herd and animal nested within herd were set as random effects. Results of the analysis indicated that the wellness trait predictions were associated with differences in phenotypic disease incidence between the worst and best genetic groups. The difference between the worst and best genetic groups in recorded disease incidence was 2.9% for retained placenta, 10.8% for metritis, 1.1% for displaced abomasum, 1.7% for ketosis, 7.4% for mastitis, and 3.9% for lameness. Odds ratio estimates between the highest and lowest genetic groups ranged from 1.6 (lameness) to 17.1 (displaced abomasum) for the 6 traits analyzed. These results indicate that wellness trait information of young calves and heifers can be used to effectively predict meaningful differences in future health performance. Improving wellness traits through direct genetic selection presents a compelling opportunity for dairy producers to help reduce disease incidence and improve profitability when coupled with sound management practices.     

Genomic selection strategies in a small dairy cattle population evaluated for genetic gain and profit

The objective of this study was to evaluate a genomic breeding scheme in a small dairy cattle population that was intermediate in terms of using both young bulls (YB) and progeny-tested bulls (PB). This scheme was compared with a conventional progeny testing program without use of genomic information and, as the extreme case, a juvenile scheme with genomic information, where all bulls were used before progeny information was available. The population structure, cost, and breeding plan parameters were chosen to reflect the Danish Jersey cattle population, being representative for a small dairy cattle population. The population consisted of 68,000 registered cows. Annually, 1,500 bull dams were screened to produce the 500 genotyped bull calves from which 60 YB were selected to be progeny tested. Two unfavorably correlated traits were included in the breeding goal, a production trait (h² = 0.30) and a functional trait (h² = 0.04). An increase in reliability of 5 percentage points for each trait was used in the default genomic scenario. A deterministic approach was used to model the different breeding programs, where the primary evaluation criterion was annual monetary genetic gain (AMGG). Discounted profit was used as an indicator of the economic outcome. We investigated the effect of varying the following parameters: (1) increase in reliability due to genomic information, (2) number of genotyped bull calves, (3) proportion of bull dam sires that are young bulls, and (4) proportion of cow sires that are young bulls. The genomic breeding scheme was both genetically and economically superior to the conventional breeding scheme, even in a small dairy cattle population where genomic information causes a relatively low increase in reliability of breeding values. Assuming low reliabilities of genomic predictions, the optimal breeding scheme according to AMGG was characterized by mixed use of YB and PB as bull sires. Exclusive use of YB for production cows increased AMGG up to 3 percentage points. The results from this study supported our hypothesis that strong interaction effects exist. The strongest interaction effects were obtained between increased reliabilities of genomic estimated breeding values and more intensive use of YB. The juvenile scheme was genetically inferior when the increase in reliability was low (5 percentage points), but became genetically superior at higher reliabilities of genomic estimated breeding values. The juvenile scheme was always superior according to discounted profit because of the shorter generation interval and minimizing costs for housing and feeding waiting bulls.     

Alternative somatic cell count traits exploitable in genetic selection for mastitis resistance in Italian Holsteins

The aim of the present study was to characterize alternative somatic cell count (SCC) traits that could be exploited in genetic selection for mastitis resistance. Data were from 66,407 first-parity Holsteins in 404 herds. Novel SCC traits included average somatic cell score (SCS, log-transformation of SCC) in early lactation (SCS_150), standard deviation of SCS of the entire lactation (SCS_SD), the presence of at least one test-day (TD) SCC >400,000 cells/mL in the lactation, and the ratio of number of TD SCC >400,000 cells/mL to total number of TD in the lactation. Novel traits and lactation-mean SCS (SCS_LM) were analyzed using linear mixed or logistic regression models, including month of calving, year of calving, number of TD, and milk yield as fixed effects, and herd and residual as random terms. A multitrait linear animal model was applied to a random subset of 152 herds (n = 22,695 cows) to assess heritability of and genetic correlations between SCC traits. Alternative SCC traits were affected by the environmental factors included in the model; in particular, results suggested a seasonal effect and a tendency toward an improvement of the udder health status in the last years. Association was also found between novel SCC traits and milk production. Alternative SCC traits exhibited coefficients of additive genetic variation that were similar to or larger than that of traditional SCS_LM. Heritability of novel SCC traits was smaller than heritability of SCS_LM (0.126 ± 0.014), ranging from 0.044 ± 0.008 (SCS_SD) to 0.087 ± 0.010 (SCS_150). Genetic correlations between SCC traits ranged from 0.217 ± 0.096 (SCS_150 and SCS_SD) to 0.969 ± 0.010 (SCS_LM and SCS_150). Alternative SCC traits exhibited additive genetic variation that is potentially exploitable in breeding programs of Italian Holstein population to improve resistance to mastitis.     

Analysis of milk production traits in early lactation using a reaction norm model with unknown covariates

The reaction norm model is becoming a popular approach to study genotype × environment interaction (G×E), especially when there is a continuum of environmental effects. These effects are typically unknown, and an approximation that is used in the literature is to replace them by the phenotypic means of each environment. It has been shown that this method results in poor inferences and that a more satisfactory alternative is to infer environmental effects jointly with the other parameters of the model. Such a reaction norm model with unknown covariates and heterogeneous residual variances across herds was fitted to milk, protein, and fat yield of first-lactation Danish Holstein cows to investigate the presence of G×E. Data included 188,502 first test-day records from 299 herds and 3,775 herd-years in a time period ranging from 1991 to 2003. Variance components and breeding values were estimated with a Bayesian approach implemented using Markov chain Monte Carlo. The posterior distribution of the variance of genetic slopes was markedly shifted away from zero for all traits under study, supporting the presence of G×E. The ratio of the genetic slope variance to the genetic level variance was highest for fat yield, followed by protein and milk yields. Genetic correlations between environments that differ by plus and minus 1 standard deviation from the mean environmental effect were 0.93, 0.91, and 0.89 for milk, protein, and fat yield, respectively. Genetic variances and heritabilities increased with increasing level of environmental effects. The rank correlations between predicted breeding values at the 5th and 95th percentiles of the distribution of environmental effects were, respectively, equal to 0.91, 0.90, and 0.76, for milk, protein, and fat yield. Thus in this study, although G×E was detected, it has a small effect on reranking of candidates for selection.     

Genetic parameters of mid-infrared methane predictions and their relationships with milk production traits in Holstein cattle

Many countries have pledged to reduce greenhouse gases. In this context, the dairy sector is one of the identified sectors to adapt production circumstances to address socio-environmental constraints due to its large carbon footprint related to CH₄ emission. This study aimed mainly to estimate (1) the genetic parameters of 2 milk mid-infrared-based CH₄ proxies [predicted daily CH₄ emission (PME, g/d), and log-transformed predicted CH₄ intensity (LMI)] and (2) their genetic correlations with milk production traits [milk (MY), fat (FY), and protein (PY) yields] from first- and second-parity Holstein cows. A total of 336,126 and 231,400 mid-infrared CH₄ phenotypes were collected from 56,957 and 34,992 first- and second-parity cows, respectively. The PME increased from the first to the second lactation (433 vs. 453 g/d) and the LMI decreased (2.93 vs. 2.86). We used 20 bivariate random regression test-day models to estimate the variance components. Moderate heritability values were observed for both CH₄ traits, and those values decreased slightly from the first to the second lactation (0.25 ± 0.01 and 0.22 ± 0.01 for PME; 0.18 ± 0.01 and 0.17 ± 0.02 for LMI). Lactation phenotypic and genetic correlations were negative between PME and MY in both first and second lactations (?0.07 vs. ?0.07 and ?0.19 vs. ?0.24, respectively). More close scrutiny revealed that relative increase of PME was lower with high MY levels even reverting to decrease, and therefore explaining the negative correlations, indicating that higher producing cows could be a mitigation option for CH₄ emission. The PME phenotypic correlations were almost equal to 0 with FY and PY for both lactations. However, the genetic correlations between PME and FY were slightly positive (0.11 and 0.12), whereas with PY the correlations were slightly negative (?0.05 and ?0.04). Both phenotypic and genetic correlations between LMI and MY or PY or FY were always relatively highly negative (from ?0.21 to ?0.88). As the genetic correlations between PME and LMI were strong (0.71 and 0.72 in first and second lactation), the selection of one trait would also strongly influence the other trait. However, in animal breeding context, PME, as a direct quantity CH₄ proxy, would be preferred to LMI, which is a ratio trait of PME with a trait already in the index. The range of PME sire estimated breeding values were 22.1 and 29.41 kg per lactation in first and second parity, respectively. Further studies must be conducted to evaluate the effect of the introduction of PME in a selection index on the other traits already included in this index, such as, for instance, fertility or longevity.     

Cow genotyping strategies for genomic selection in a small dairy cattle population

This study compares how different cow genotyping strategies increase the accuracy of genomic estimated breeding values (EBV) in dairy cattle breeds with low numbers. In these breeds, few sires have progeny records, and genotyping cows can improve the accuracy of genomic EBV. The Guernsey breed is a small dairy cattle breed with approximately 14,000 recorded individuals worldwide. Predictions of phenotypes of milk yield, fat yield, protein yield, and calving interval were made for Guernsey cows from England and Guernsey Island using genomic EBV, with training sets including 197 de-regressed proofs of genotyped bulls, with cows selected from among 1,440 genotyped cows using different genotyping strategies. Accuracies of predictions were tested using 10-fold cross-validation among the cows. Genomic EBV were predicted using 4 different methods: (1) pedigree BLUP, (2) genomic BLUP using only bulls, (3) univariate genomic BLUP using bulls and cows, and (4) bivariate genomic BLUP. Genotyping cows with phenotypes and using their data for the prediction of single nucleotide polymorphism effects increased the correlation between genomic EBV and phenotypes compared with using only bulls by 0.163 ± 0.022 for milk yield, 0.111 ± 0.021 for fat yield, and 0.113 ± 0.018 for protein yield; a decrease of 0.014 ± 0.010 for calving interval from a low base was the only exception. Genetic correlation between phenotypes from bulls and cows were approximately 0.6 for all yield traits and significantly different from 1. Only a very small change occurred in correlation between genomic EBV and phenotypes when using the bivariate model. It was always better to genotype all the cows, but when only half of the cows were genotyped, a divergent selection strategy was better compared with the random or directional selection approach. Divergent selection of 30% of the cows remained superior for the yield traits in 8 of 10 folds.     

Genomic information allows for more accurate breeding values for milkability in dual-purpose Italian Simmental cattle

Milkability is a trait related to the milking efficiency of an animal, and it is a component of the herd profitability. Due to its economic importance, milkability is currently included in the selection index of the Italian Simmental cattle breed with a weight of 7.5%. This lowly heritable trait is measured on a subjective scale from 1 to 3 (1 = slow, 3 = fast), and genetic evaluations are performed by pedigree-based BLUP. Genomic information is now available for some animals in the Italian Simmental population, and its inclusion in the genetic evaluation system could increase accuracy of breeding values and genetic progress for milkability. The aim of this study was to test the feasibility and advantages of having a genomic evaluation for this trait in the Italian Simmental population. Phenotypes were available for 131,308 cows. A total of 9,526 animals had genotypes for 42,152 loci; among the genotyped animals, 2,455 were cows with phenotypes, and the other were their relatives. The youngest cows with both phenotypes and genotypes (n = 900) were identified as selection candidates. Variance components and heritability were estimated using pedigree information, whereas genetic and genomic evaluations were carried out using BLUP and single-step genomic BLUP (ssGBLUP), respectively. In addition, a weighted ssGBLUP was assessed using genomic regions from a genome-wide association study. Evaluation models were validated using theoretical and realized accuracies. The estimated heritability for milkability was 0.12 ± 0.01. The mean theoretical accuracies for selection candidates were 0.43 ± 0.08 (BLUP) and 0.53 ± 0.06 (ssGBLUP). The mean realized accuracies based on linear regression statistics were 0.29 (BLUP) and 0.40 (ssGBLUP). No genomic regions were significantly associated with milkability, thus no improvements in accuracy were observed when using weighted ssGBLUP. Results indicated that genomic information could improve the accuracy of breeding values and increase genetic progress for milkability in Italian Simmental.     

Development of genetic and genomic evaluation for wellness traits in US Holstein cows

In March 2016, Zoetis Genetics offered the first commercially available evaluation for wellness traits of Holstein dairy cattle. Phenotypic data on health events, pedigree, and genotypes were collected directly from producers upon obtaining their permission. Among all recorded health events, 6 traits were chosen to be included in the evaluation: mastitis, metritis, retained placenta, displaced abomasum, ketosis, and lameness. Each trait was defined as a binary event, having a value of 1 if a cow has been recorded with a disorder at any point during the lactation and zero otherwise. The number of phenotypic records ranged from 1.8 million for ketosis to 4.1 million for mastitis. Over 14 million pedigree records and 114,216 genotypes were included in the evaluation. All traits were analyzed using univariate threshold animal model with repeated observations, including fixed effect of parity and random effects of herd by year by season of calving, animal, and permanent environment. A total of 45,425 single nucleotide polymorphisms were used in the genomic analyses. Animals genotyped with low-density chips were imputed to the required number of single nucleotide polymorphisms. All analyses were based on the single-step genomic BLUP, a method that combines phenotype, pedigree, and genotype information. Predicted transmitting abilities were expressed in percentage points as a difference from the average estimated probability of a disorder in the base population. Reliabilities of breeding values were obtained by approximation based on partitioning of a function of reliability into contributions from records, pedigree, and genotypes. Reliabilities of genomic predicted transmitting abilities for young genotyped and pedigreed females without recorded health events had average values between 50.2% (displaced abomasum) and 51.9% (mastitis). Genomic predictions for wellness traits can provide new information about an animal’s genetic potential for health and new selection tools for dairy wellness improvement.     

Alternative SNP weighting for single-step genomic best linear unbiased predictor evaluation of stature in US Holsteins in the presence of selected sequence variants

Causal variants inferred from sequence data analysis are expected to increase accuracy of genomic selection. In this work we evaluated the gain in reliability of genomic predictions, for stature in US Holsteins, when adding selected sequence variants to a pre-existent SNP chip. Two prediction methods were tested: de-regressed proofs assuming heterogeneous (genomic BLUP; GBLUP) residual variances and by single-step GBLUP (ssGBLUP) using actual phenotypes. Phenotypic data included 3,999,631 records for stature on 3,027,304 Holstein cows. Genotypes on 54,087 SNP markers (54k) were available for 26,877 bulls. Additionally, 16,648 selected sequence variants were combined with the 54k markers, for a total of 70,735 (70k) markers. In all methods, SNP in the genomic relationship matrix (G) were unweighted or weighted iteratively, with weights derived either by SNP effects squared or by a nonlinear method that resembles BayesA (nonlinear A). Reliability of genomic predictions were obtained by cross validation. With unweighted G derived from 54k markers, the reliabilities (× 100) were 72.4 for GBLUP and 75.3 for ssGBLUP. With unweighted G derived from 70k markers, the reliabilities were 73.4 and 76.0, respectively. Weighting by nonlinear A changed reliabilities to 73.3, and 75.9, respectively. Addition of selected sequence variants had a small effect on reliabilities. Weighting by quadratic functions reduced reliabilities. Weighting by nonlinear A increased reliabilities for GBLUP but had only a small effect in ssGBLUP. Reliabilities for direct genomic values extracted from ssGBLUP using unweighted G with 54k were higher than reliabilities by any GBLUP. Thus, ssGBLUP seems to capture more information than GBLUP and there is less room for extra reliability. Improvements in GBLUP may be because the weights in G change the covariance structure, which can explain a proportion of the variance that is accounted for when a heterogeneous residual variance is assumed by considering a different number of daughters per bull.