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.     

How genomic selection has increased rates of genetic gain and inbreeding in the Australian national herd,genomic information nucleus,and bulls

Genomic selection has been commonly used for selection for over a decade. In this time, the rate of genetic gain has more than doubled in some countries, while inbreeding per year has also increased. Inbreeding can result in a loss of genetic diversity, decreased long-term response to selection, reduced animal performance and ultimately, decreased farm profitability. We quantified and compared changes in genetic gain and diversity resulting from genomic selection in Australian Holstein and Jersey cattle populations. To increase the accuracy of genomic selection, Australia has had a female genomic reference population since 2013, specifically designed to be representative of commercial populations and thus including both Holstein and Jersey cows. Herds that kept excellent health and fertility data were invited to join this population and most their animals were genotyped. In both breeds, the rate of genetic gain and inbreeding was greatest in bulls, and then the female genomic reference population, and finally the wider national herd. When comparing pre- and postgenomic selection, the rates of genetic gain for the national economic index has increased by ~160% in Holstein females and ~100% in Jersey females. This has been accompanied by doubling of the rates of inbreeding in female populations, and the rate of inbreeding has increased several fold in Holstein bulls since the widespread use of genomic selection. Where cow genotype data were available to perform a more accurate genomic analysis, greater rates of pedigree and genomic inbreeding were observed, indicating actual inbreeding levels could be underestimated in the national population due to gaps in pedigrees. Based on current rates of genetic gain, the female reference population is progressing ahead of the national herd and could be used to infer and track the future inbreeding and genetic trends of the national herds.     

Methods for genomic evaluation of a relatively small genotyped dairy population and effect of genotyped cow information in multiparity analyses

Methods for genomic prediction were evaluated for an Israeli Holstein dairy population of 713,686 cows and 1,305 progeny-tested bulls with genotypes. Inclusion of genotypes of 343 elite cows in an evaluation method that considers pedigree, phenotypes, and genotypes simultaneously was also evaluated. Two data sets were available: a complete data set with production records from 1985 through 2011, and a reduced data set with records after 2006 deleted. For each production trait, a multitrait animal model was used to compute traditional genetic evaluations for parities 1 through 3 as separate traits. Evaluations were calculated for the reduced and complete data sets. The evaluations from the reduced data set were used to calculate parent average for validation bulls, which was the benchmark for comparing gain in predictive ability from genomics. Genomic predictions for bulls in 2006 were calculated using a Bayesian regression method (BayesC), genomic BLUP (GBLUP), single-step GBLUP (ssGBLUP), and weighted ssGBLUP (WssGBLUP). Predictions using BayesC and GBLUP were calculated either with or without an index that included parent average. Genomic predictions that included elite cow genotypes were calculated using ssGBLUP and WssGBLUP. Predictive ability was assessed by coefficients of determination (R²) and regressions of predictions of 135 validation bulls with no daughters in 2006 on deregressed evaluations of those bulls in 2011. A reduction in R² and regression coefficients was observed from parities 1 through 3. Fat and protein yields had the lowest R² for all the methods. On average, R² was lowest for parent averages, followed by GBLUP, BayesC, ssGBLUP, and WssGBLUP. For some traits, R² for direct genomic values from BayesC and GBLUP were lower than those for parent averages. Genomic estimated breeding values using ssGBLUP were the least biased, and this method appears to be a suitable tool for genomic evaluation of a small genotyped population, as it automatically accounts for parental index, allows for inclusion of female genomic information without preadjustments in evaluations, and uses the same model as in traditional evaluations. Weighted ssGBLUP has the potential for higher evaluation accuracy.     

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.     

Effects of genomic selection on genetic improvement, inbreeding, and merit of young versus proven bulls

Genomic selection has the potential to revolutionize dairy cattle breeding because young animals can be accurately selected as parents, leading to a much shorter generation interval and higher rates of genetic gain. The aims of this study were to assess the effects of genomic selection and reduction of the generation interval on the rate of genetic gain and rate of inbreeding. Furthermore, the merit of proven bulls relative to young bulls was studied. This is important for breeding organizations as it determines the relative importance of progeny testing. A closed nucleus breeding scheme was simulated in which 1,000 males and 1,000 females were born annually, 200 bulls were progeny tested, and 20 sires and 200 dams were selected to produce the next generation. In the “proven” (PROV) scenario, only cows with own performance records and progeny-tested bulls were selected as parents. The proportion of the genetic variance that was explained by simulated marker information (M) was varied from 0 to 100%. When M increased from 0 to 100%, the rate of genetic gain increased from 0.238 to 0.309 genetic standard deviations (σ) per year (+30%), whereas the rate of inbreeding reduced from 1.00 to 0.42% per generation. Alternatively, when young cows and bulls were selected as parents (YNG scenario), the rate of genetic gain for M = 0% was 0.292 σ/yr but the corresponding rate of inbreeding increased substantially to 3.15% per generation. A realistic genomic selection scheme (YNG with M = 40%) gave 108% higher rate of genetic gain (0.495 σ/yr) and approximately the same rate of inbreeding per generation as the conventional system without genomic selection (PROV with M = 0%). The rate of inbreeding per year, however, increased from 0.18 to 0.52% because the generation interval in the YNG scheme was much shorter. Progeny-testing fewer bulls reduced the rate of genetic gain and increased the rate of inbreeding for PROV, but had negligible effects for YNG because almost all sires were young bulls. In scenario YNG with M = 40%, the best young bulls were superior to the best proven bulls by 1.27 σ difference in genomic estimated breeding value. This superiority increased even further when fewer bulls were progeny tested. This stochastic simulation study shows that genomic selection in combination with a severe reduction in the generation interval can double the rate of genetic gain at the same rate of inbreeding per generation, but with a higher rate of inbreeding per year. The number of progeny-tested bulls can be greatly reduced, although this will slightly affect the quality of the proven bull team. Therefore, it is important for breeding organizations to predict the future demand for proven bull semen in light of the increasing superiority of young bulls.     

Genomic predictions based on a joint reference population for the Nordic Red cattle breeds

The main aim of this study was to compare accuracies of imputation and genomic predictions based on single and joint reference populations for Norwegian Red (NRF) and a composite breed (DFS) consisting of Danish Red, Finnish Ayrshire, and Swedish Red. The single nucleotide polymorphism (SNP) data for NRF consisted of 2 data sets: one including 25,000 markers (NRF25K) and the other including 50,000 markers (NRF50K). The NRF25K data set had 2,572 bulls, and the NRF50K data set had 1,128 bulls. Four hundred forty-two bulls were genotyped in both data sets (double-genotyped bulls). The DFS data set (DSF50K) included 50,000 markers of 13,472 individuals, of which around 4,700 were progeny-tested bulls. The NRF25K data set was imputed to 50,000 density using the software Beagle. The average error rate for the imputation of NRF25K decreased slightly from 0.023 to 0.021, and the correlation between observed and imputed genotypes changed from 0.935 to 0.936 when comparing the NRF50K reference and the NRF50K–DFS50K joint reference imputations. A genomic BLUP (GBLUP) model and a Bayesian 4-component mixture model were used to predict genomic breeding values for the NRF and DFS bulls based on the single and joint NRF and DFS reference populations. In the multiple population predictions, accuracies of genomic breeding values increased for the 3 production traits (milk, fat, and protein yields) for both NRF and DFS. Accuracies increased by 6 and 1.3 percentage points, on average, for the NRF and DFS bulls, respectively, using the GBLUP model, and by 9.3 and 1.3 percentage points, on average, using the Bayesian 4-component mixture model. However, accuracies for health or reproduction traits did not increase from the multiple population predictions. Among the 3 DFS populations, Swedish Red gained most in accuracies from the multiple population predictions, presumably because Swedish Red has a closer genetic relationship with NRF than Danish Red and Finnish Ayrshire. The Bayesian 4-component mixture model performed better than the GBLUP model for most production traits for both NRF and DFS, whereas no advantage was found for health or reproduction traits. In general, combining NRF and DFS reference populations was useful in genomic predictions for both the NRF and DFS bulls.     

Technical note: adjustment of traditional cow evaluations to improve accuracy of genomic predictions

Genomic evaluations are calculated using deregressed predicted transmitting abilities (PTA) from traditional evaluations to estimate effects of single nucleotide polymorphisms. The direct genomic value (sum of an animal's marker effects) should be consistent with traditional PTA, which is the case for bulls. However, traditional PTA of yield traits (milk, fat, and protein) for genotyped cows are higher than their direct genomic values. To ensure that characteristics of cow PTA for yield traits were more similar to those for bull PTA, mean and variance of cow Mendelian sampling (PTA minus parent average) were adjusted to be similar to those of bulls. The same adjustments were used for all genotyped cows in a breed. To determine gains in reliabilities, predictions were made for bulls with August 2010 evaluations that did not have traditional evaluations in August 2006. By adjusting cow PTA and parent averages of genotyped animals, Holstein and Jersey regressions of August 2010 deregressed PTA on genomic evaluations based on August 2006 data became closer to 1 for the adjusted predictor population compared with the unadjusted predictor population. Evaluation bias was decreased for Holsteins when the predictor population was adjusted. Mean gain in reliability over parent average increased 3.5 percentage points across yield traits for Holsteins and 0.9 percentage points for Jerseys when the predictor population was adjusted. The accuracy of genomic evaluations for Holsteins and Jerseys was increased through better use of information from cows.     

Optimization of a genomic breeding program for a moderately sized dairy cattle population

Although it now standard practice to genotype thousands of female calves, genotyping of bull calves is generally limited to progeny of elite cows. In addition to genotyping costs, increasing the pool of candidate sires requires purchase, isolation, and identification of calves until selection decisions are made. We economically optimized via simulation a genomic breeding program for a population of approximately 120,000 milk-recorded cows, corresponding to the Israeli Holstein population. All 30,000 heifers and 60,000 older cows of parities 1 to 3 were potential bull dams. Animals were assumed to have genetic evaluations for a trait with heritability of 0.25 derived by an animal model evaluation of the population. Only bull calves were assumed to be genotyped. A pseudo-phenotype corresponding to each animal's genetic evaluation was generated, consisting of the animal's genetic value plus a residual with variance set to obtain the assumed reliability for each group of animals. Between 4 and 15 bulls and between 200 and 27,000 cows with the highest pseudo-phenotypes were selected as candidate bull parents. For all progeny of the founder animals, genetic values were simulated as the mean of the parental values plus a Mendelian sampling effect with variance of 0.5. A probability of 0.3 for a healthy bull calf per mating, and a genomic reliability of 0.43 were assumed. The 40 bull calves with the highest genomic evaluations were selected for general service for 1 yr. Costs included genotyping of candidate bulls and their dams, purchase of the calves from the farmers, and identification. Costs of raising culled calves were partially recovered by resale for beef. Annual costs were estimated as $10,922 + $305 × candidate bulls. Nominal profit per cow per genetic standard deviation was $106. Economic optimum with a discount rate of 5%, first returns after 4 yr, and a profit horizon of 15 yr were obtained with genotyping 1,620 to 1,750 calves for all numbers of bull sires. However, 95% of the optimal profit can be achieved with only 240 to 300 calves. The higher reliabilities achieved through addition of genomic information to the selection process contribute not only in obtaining higher genetic gain, but also in obtaining higher absolute profits. In addition, the optimal profits are obtained for a lower number of calves born in each generation. Inbreeding, as allowed within genomic selection for the Israeli herd, had virtually no effect on genetic gain or on profits, when compared with the case of exclusion of all matings that generate inbreeding. Annual response to selection ranged from 0.35 to 0.4 genetic standard deviation for 4 to 15 bull sires, as compared with 0.25 to 0.3 for a comparable half-sib design without genomic selection.     

A comparison of dairy cattle breeding designs that use genomic selection

Different dairy cattle breeding schemes were compared using stochastic simulations, in which the accuracy of the genomic breeding values was dependent on the structure of the breeding scheme, through the availability of new genotyped animals with phenotypic information. Most studies that predict the gain by implementing genomic selection apply a deterministic approach that requires assumptions about the accuracy of the genomic breeding values. The achieved genetic gain, when genomic selection was the only selection method to directly identify elite sires for widespread use and progeny testing was omitted, was compared with using genomic selection for preselection of young bulls for progeny testing and to a conventional progeny test scheme. The rate of inbreeding could be reduced by selecting more sires every year. Selecting 20 sires directly on their genomic breeding values gave a higher genetic gain than any progeny testing scheme, with the same rate of inbreeding as the schemes that used genomic selection for preselection of bulls before progeny testing. The genomic selection breeding schemes could reduce the rate of inbreeding and still increase genetic gain, compared with the conventional breeding scheme. Since progeny testing is expensive, the breeding scheme omitting the progeny test will be the cheapest one. Keeping the progeny test and use of genomic selection for preselection still has some advantages. It gives higher accuracy of breeding values and does not require a complete restructuring of the breeding program. Comparing at the same rate of inbreeding, using genomic selection for elite sire selection only gives a 13% increase in genetic gain, compared with using genomic selection for preselection. One way to reduce the costs of the scheme where genomic selection was used for preselection is to reduce the number of progeny tested bulls. This was here achieved without getting lower genetic gain or a higher rate of inbreeding.     

Deterministic models of breeding scheme designs that incorporate genomic selection

A deterministic model to calculate rates of genetic gain and inbreeding was used to compare a range of breeding scheme designs under genomic selection (GS) for a population of 140,000 cows. For most schemes it was assumed that the reliability of genomic breeding values (GEBV) was 0.6 across 4 pathways of selection. In addition, the effect of varying reliability on the ranking of schemes was also investigated. The schemes considered included intense selection in male pathways and genotyping of 1,000 young bulls (GS-Y). This scheme was extended to include selection in females and to include a “worldwide” scheme similar to GS-Y, but 6 times as large and assuming genotypes were freely exchanged between 6 countries. An additional worldwide scheme was modeled where GEBV were available through international genetic evaluations without exchange of genotypes. Finally, a closed nucleus herd that used juvenile in vitro embryo transfer in heifers was modeled so that the generation interval in female pathways was reduced to 1 or 2 yr. When the breeding schemes were compared using a GEBV reliability of 0.6, the rates of genetic gain were between 59 and 130% greater than the rate of genetic gain achieved in progeny testing. This was mainly through reducing the generation interval and increasing selection intensity. Genomic selection of females resulted in a 50% higher rate of genetic gain compared with restricting GS to young bulls only. The annual rates of inbreeding were, in general, 60% lower than with progeny testing, because more sires of bulls and sires of cows were selected, thus increasing the effective population size. The exception was in nucleus breeding schemes that had very short generation intervals, resulting in higher rates of both gain and inbreeding. It is likely that breeding companies will move rapidly to alter their breeding schemes to make use of genomic selection because benefits to the breeding companies and to the industry are considerable.     

Invited review: Genomic selection in dairy cattle: Progress and challenges

A new technology called genomic selection is revolutionizing dairy cattle breeding. Genomic selection refers to selection decisions based on genomic breeding values (GEBV). The GEBV are calculated as the sum of the effects of dense genetic markers, or haplotypes of these markers, across the entire genome, thereby potentially capturing all the quantitative trait loci (QTL) that contribute to variation in a trait. The QTL effects, inferred from either haplotypes or individual single nucleotide polymorphism markers, are first estimated in a large reference population with phenotypic information. In subsequent generations, only marker information is required to calculate GEBV. The reliability of GEBV predicted in this way has already been evaluated in experiments in the United States, New Zealand, Australia, and the Netherlands. These experiments used reference populations of between 650 and 4,500 progeny-tested Holstein-Friesian bulls, genotyped for approximately 50,000 genome-wide markers. Reliabilities of GEBV for young bulls without progeny test results in the reference population were between 20 and 67%. The reliability achieved depended on the heritability of the trait evaluated, the number of bulls in the reference population, the statistical method used to estimate the single nucleotide polymorphism effects in the reference population, and the method used to calculate the reliability. A common finding in 3 countries (United States, New Zealand, and Australia) was that a straightforward BLUP method for estimating the marker effects gave reliabilities of GEBV almost as high as more complex methods. The BLUP method is attractive because the only prior information required is the additive genetic variance of the trait. All countries included a polygenic effect (parent average breeding value) in their GEBV calculation. This inclusion is recommended to capture any genetic variance not associated with the markers, and to put some selection pressure on low-frequency QTL that may not be captured by the markers. The reliabilities of GEBV achieved were significantly greater than the reliability of parental average breeding values, the current criteria for selection of bull calves to enter progeny test teams. The increase in reliability is sufficiently high that at least 2 dairy breeding companies are already marketing bull teams for commercial use based on their GEBV only, at 2 yr of age. This strategy should at least double the rate of genetic gain in the dairy industry. Many challenges with genomic selection and its implementation remain, including increasing the accuracy of GEBV, integrating genomic information into national and international genetic evaluations, and managing long-term genetic gain.     

Strategies for implementing genomic selection for feed efficiency in dairy cattle breeding schemes

Alternative genomic selection and traditional BLUP breeding schemes were compared for the genetic improvement of feed efficiency in simulated Norwegian Red dairy cattle populations. The change in genetic gain over time and achievable selection accuracy were studied for milk yield and residual feed intake, as a measure of feed efficiency. When including feed efficiency in genomic BLUP schemes, it was possible to achieve high selection accuracies for genomic selection, and all genomic BLUP schemes gave better genetic gain for feed efficiency than BLUP using a pedigree relationship matrix. However, introducing a second trait in the breeding goal caused a reduction in the genetic gain for milk yield. When using contracted test herds with genotyped and feed efficiency recorded cows as a reference population, adding an additional 4,000 new heifers per year to the reference population gave accuracies that were comparable to a male reference population that used progeny testing with 250 daughters per sire. When the test herd consisted of 500 or 1,000 cows, lower genetic gain was found than using progeny test records to update the reference population. It was concluded that to improve difficult to record traits, the use of contracted test herds that had additional recording (e.g., measurements required to calculate feed efficiency) is a viable option, possibly through international collaborations.     

Invited Review: Reliability of genomic predictions for North American Holstein bulls

Genetic progress will increase when breeders examine genotypes in addition to pedigrees and phenotypes. Genotypes for 38,416 markers and August 2003 genetic evaluations for 3,576 Holstein bulls born before 1999 were used to predict January 2008 daughter deviations for 1,759 bulls born from 1999 through 2002. Genotypes were generated using the Illumina BovineSNP50 BeadChip and DNA from semen contributed by US and Canadian artificial-insemination organizations to the Cooperative Dairy DNA Repository. Genomic predictions for 5 yield traits, 5 fitness traits, 16 conformation traits, and net merit were computed using a linear model with an assumed normal distribution for marker effects and also using a nonlinear model with a heavier tailed prior distribution to account for major genes. The official parent average from 2003 and a 2003 parent average computed from only the subset of genotyped ancestors were combined with genomic predictions using a selection index. Combined predictions were more accurate than official parent averages for all 27 traits. The coefficients of determination (R²) were 0.05 to 0.38 greater with nonlinear genomic predictions included compared with those from parent average alone. Linear genomic predictions had R² values similar to those from nonlinear predictions but averaged just 0.01 lower. The greatest benefits of genomic prediction were for fat percentage because of a known gene with a large effect. The R² values were converted to realized reliabilities by dividing by mean reliability of 2008 daughter deviations and then adding the difference between published and observed reliabilities of 2003 parent averages. When averaged across all traits, combined genomic predictions had realized reliabilities that were 23% greater than reliabilities of parent averages (50 vs. 27%), and gains in information were equivalent to 11 additional daughter records. Reliability increased more by doubling the number of bulls genotyped than the number of markers genotyped. Genomic prediction improves reliability by tracing the inheritance of genes even with small effects.     

Superiority of QTL-assisted selection in dairy cattle breeding schemes

The superiority of selection schemes employing information about a known quantitative trait locus (QTL) over conventional schemes is examined for dairy cattle breeding schemes. Stochastic simulation of a dairy cattle population with selection practices, structures, and parameters similar to the US Holstein population was implemented. Additive genetic effects were estimated by an animal model. Two schemes were compared: a QTL-assisted selection scheme in which the genotype of a known QTL was accounted for in the animal model as a fixed factor, and a QTL-free selection scheme in which the QTL was simulated but was not fit separately in the animal model. Under the QTL-assisted selection scheme, all animals in the mixed model were assumed to be genotyped for the QTL. The effect of using QTL information on the genetic response, the frequency of the favorable QTL allele, and the accuracy of evaluation were examined. Moreover, the effect was studied in four distinct paths of selection: active sires, proven young bulls, bull dams, and first-lactation cows. Average superiority values of 4.6, 7.6, 11.7, and 1.1% for genetic response were observed over 16 yr of selection for active sires, young bulls, bull dams, and first-lactation cows, respectively. Frequency of the favorable QTL allele changed faster in bull dams than males, and was the slowest in first-lactation cows. Finally, accuracy of evaluation under the QTL-assisted selection scheme was higher than under the QTL-free selection scheme. Young bulls ofthe QTL-assisted selection scheme on average had 0.049 higher accuracy, and first-lactation cows had on average 0.185 higher accuracy than corresponding animals of the QTL-free selection scheme.     

Effect of enlarging the reference population with (un)genotyped animals on the accuracy of genomic selection in dairy cattle

Genomic selection (GS) permits accurate breeding values to be obtained for young animals, shortening the generation interval and accelerating the genetic gain, thereby leading to reduced costs for proven bulls. Genotyping a large number of animals using high-density single nucleotide polymorphism marker arrays is nevertheless expensive, and therefore, a method to reduce the costs of GS is desired. The aim of this study was to investigate an influence of enlarging the reference population, with either genotyped animals or individuals with predicted genotypes, on the accuracy of genomic estimated breeding values. A dairy cattle population was simulated in which proven bulls with 100 daughters were used as a reference population for GS. Phenotypic records were simulated for bulls with heritability equal to the reliability of daughter yield deviations based on 100 daughters. The simulated traits represented heritabilities at the level of individual daughter performance of 0.3, 0.05, and 0.01. Three scenarios were considered in which (1) the reference population consisted of 1,000 genotyped animals, (2) 1,000 ungenotyped animals were added to the reference population, and (3) the 1,000 animals added in scenario 2 were genotyped in addition to the 1,000 animals from scenario 1. Genotypes for ungenotyped animals were predicted with an average accuracy of 0.58. Additionally, an adjustment of the diagonal elements of the G matrix was proposed for animals with predicted genotypes. The accuracy of genomic estimated breeding values for juvenile animals was the highest for the scenario with 2,000 genotyped animals, being 0.90, 0.79, and 0.60 for the heritabilities of 0.3, 0.05, and 0.01, respectively. Accuracies did not differ significantly between the scenario with 1,000 genotyped animals only and the scenario in which 1,000 ungenotyped animals were added and the adjustment of the G matrix was applied. The absence of significant increase in the accuracy of genomic estimated breeding values was attributed to the low accuracy of predicted genotypes. Although the differences were not significant, the difference between scenario 1 and 2 increased with decreasing heritability. Without the adjustment of the diagonal elements of the G matrix, accuracy decreased. Results suggest that inclusion of ungenotyped animals is only expected to enhance the accuracy of GS when the unknown genotypes can be predicted with high accuracy.     

Differing genetic trend estimates from traditional and genomic evaluations of genotyped animals as evidence of preselection bias in US Holsteins

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.     

Mating programs including genomic relationships and dominance effects

Computerized mating programs using genomic information are needed by breed associations, artificial-insemination organizations, and on-farm software providers, but such software is already challenged by the size of the relationship matrix. As of October 2012, over 230,000 Holsteins obtained genomic predictions in North America. Efficient methods of storing, computing, and transferring genomic relationships from a central database to customers via a web query were developed for approximately 165,000 genotyped cows and the subset of 1,518 bulls whose semen was available for purchase at that time. This study, utilizing 3 breeds, investigated differences in sire selection, methods of assigning mates, the use of genomic or pedigree relationships, and the effect of including dominance effects in a mating program. For both Jerseys and Holsteins, selection and mating programs were tested using the top 50 marketed bulls for genomic and traditional lifetime net merit as well as 50 randomly selected bulls. The 500 youngest genotyped cows in the largest herd in each breed were assigned mates of the same breed with limits of 10 cows per bull and 1 bull per cow (only 79 cows and 8 bulls for Brown Swiss). A dominance variance of 4.1 and 3.7% was estimated for Holsteins and Jerseys using 45,187 markers and management group deviation for milk yield. Sire selection was identified as the most important component of improving expected progeny value, followed by managing inbreeding and then inclusion of dominance. The respective percentage gains for milk yield in this study were 64, 27, and 9, for Holsteins and 73, 20, and 7 for Jerseys. The linear programming method of assigning a mate outperformed sequential selection by reducing genomic or pedigree inbreeding by 0.86 to 1.06 and 0.93 to 1.41, respectively. Use of genomic over pedigree relationship information provided a larger decrease in expected progeny inbreeding and thus greater expected progeny value. Based on lifetime net merit, the economic value of using genomic relationships was >$3 million per year for Holsteins when applied to all genotyped females, assuming that each will provide 1 replacement. Previous mating programs required transferring only a pedigree file to customers, but better service is possible by incorporating genomic relationships, more precise mate allocation, and dominance effects. Economic benefits will continue to grow as more females are genotyped.     

Genomic prediction for Nordic Red Cattle using one-step and selection index blending

This study investigated the accuracy of direct genomic breeding values (DGV) using a genomic BLUP model, genomic enhanced breeding values (GEBV) using a one-step blending approach, and GEBV using a selection index blending approach for 15 traits of Nordic Red Cattle. The data comprised 6,631 bulls of which 4,408 bulls were genotyped using Illumina Bovine SNP50 BeadChip (Illumina, San Diego, CA). To validate reliability of genomic predictions, about 20% of the youngest genotyped bulls were taken as test data set. Deregressed proofs (DRP) were used as response variables for genomic predictions. Reliabilities of genomic predictions in the validation analyses were measured as squared correlations between DRP and genomic predictions corrected for reliability of DRP, based on the bulls in the test data sets. A set of weighting (scaling) factors was used to construct the combined relationship matrix among genotyped and nongenotyped bulls for one-step blending, and to scale DGV and its expected reliability in the selection index blending. Weighting (scaling) factors had a small influence on reliabilities of GEBV, but a large influence on the variation of GEBV. Based on the validation analyses, averaged over the 15 traits, the reliability of DGV for bulls without daughter records was 11.0 percentage points higher than the reliability of conventional pedigree index. Further gain of 0.9 percentage points was achieved by combining information from conventional pedigree index using the selection index blending, and gain of 1.3 percentage points was achieved by combining information of genotyped and nongenotyped bulls simultaneously applying the one-step blending. These results indicate that genomic selection can greatly improve the accuracy of preselection for young bulls in Nordic Red population, and the one-step blending approach is a good alternative to predict GEBV in practical genetic evaluation program.     

Controlling bias in genomic breeding values for young genotyped bulls

The objectives of this study were to investigate bias in genomic predictions for dairy cattle and to find a practical approach to reduce the bias. The simulated data included phenotypes, pedigrees, and genotypes, mimicking a dairy cattle population (i.e., cows with phenotypes and bulls with no phenotypes) and assuming selection by breeding values or no selection. With the simulated data, genomic estimated breeding values (GEBV) were calculated with a single-step genomic BLUP and compared with true breeding values. Phenotypes and genotypes were simulated in 10 generations and in the last 4 generations, respectively. Phenotypes in the last generation were removed to predict breeding values for those individuals using only genomic and pedigree information. Complete pedigrees and incomplete pedigrees with 50% missing dams were created to construct the pedigree-based relationship matrix with and without inbreeding. With missing dams, unknown parent groups (UPG) were assigned in relationship matrices. Regression coefficients (b₁) and coefficients of determination (R²) of true breeding values on (G)EBV were calculated to investigate inflation and accuracy in GEBV for genotyped animals, respectively. In addition to the simulation study, 18 linear type traits of US Holsteins were examined. For the 18 type traits, b₁ and R² of GEBV with full data sets on GEBV with partial data sets for young genotyped bulls were calculated. The results from the simulation study indicated inflation in GEBV for genotyped males that were evaluated with only pedigree and genomic information under BLUP selection. However, when UPG for only pedigree-based relationships were included, the inflation was reduced, accuracy was highest, and genetic trends had no bias. For the linear type traits, when UPG for only pedigree-based relationships were included, the results were generally in agreement with those from the simulation study, implying less bias in genetic trends. However, when including no UPG, UPG in pedigree-based relationships, or UPG in genomic relationships, inflation and accuracy in GEBV were similar. The results from the simulation and type traits suggest that UPG must be defined accurately to be estimable and inbreeding should be included in pedigree-based relationships. In dairy cattle, known pedigree information with inbreeding and estimable UPG plays an important role in improving compatibility between pedigree-based and genomic relationship matrices, resulting in more reliable genomic predictions.     

The effect of using cow genomic information on accuracy and bias of genomic breeding values in a simulated Holstein dairy cattle population

Using cow data in the training population is attractive as a way to mitigate bias due to highly selected training bulls and to implement genomic selection for countries with no or limited proven bull data. However, one potential issue with cow data is a bias due to the preferential treatment. The objectives of this study were to (1) investigate the effect of including cow genotype and phenotype data into the training population on accuracy and bias of genomic predictions and (2) assess the effect of preferential treatment for different proportions of elite cows. First, a 4-pathway Holstein dairy cattle population was simulated for 2 traits with low (0.05) and moderate (0.3) heritability. Then different numbers of cows (0, 2,500, 5,000, 10,000, 15,000, or 20,000) were randomly selected and added to the training group composed of different numbers of top bulls (0, 2,500, 5,000, 10,000, or 15,000). Reliability levels of de-regressed estimated breeding values for training cows and bulls were 30 and 75% for traits with low heritability and were 60 and 90% for traits with moderate heritability, respectively. Preferential treatment was simulated by introducing upward bias equal to 35% of phenotypic variance to 5, 10, and 20% of elite bull dams in each scenario. Two different validation data sets were considered: (1) all animals in the last generation of both elite and commercial tiers (n = 42,000) and (2) only animals in the last generation of the elite tier (n = 12,000). Adding cow data into the training population led to an increase in accuracy (r) and decrease in bias of genomic predictions in all considered scenarios without preferential treatment. The gain in r was higher for the low heritable trait (from 0.004 to 0.166 r points) compared with the moderate heritable trait (from 0.004 to 0.116 r points). The gain in accuracy in scenarios with a lower number of training bulls was relatively higher (from 0.093 to 0.166 r points) than with a higher number of training bulls (from 0.004 to 0.09 r points). In this study, as expected, the bull-only reference population resulted in higher accuracy compared with the cow-only reference population of the same size. However, the cow reference population might be an option for countries with a small-scale progeny testing scheme or for minor breeds in large counties, and for traits measured only on a small fraction of the population. The inclusion of preferential treatment to 5 to 20% of the elite cows led to an adverse effect on both accuracy and bias of predictions. When preferential treatment was present, random selection of cows did not reduce the effect of preferential treatment.