Y. de Haas

Genetic parameters for predicted methane production and potential for reducing enteric emissions through genomic selection.

Mitigation of enteric methane (CH₄) emission in ruminants has become an important area of research because accumulation of CH₄ is linked to global warming. Nutritional and microbial opportunities to reduce CH₄ emissions have been extensively researched, but little is known about using natural variation to breed animals with lower CH₄ yield. Measuring CH₄ emission rates directly from animals is difficult and hinders direct selection on reduced CH₄ emission. However, improvements can be made through selection on associated traits (e.g., residual feed intake, RFI) or through selection on CH₄ predicted from feed intake and diet composition. The objective was to establish phenotypic and genetic variation in predicted CH₄ output, and to determine the potential of genetics to reduce methane emissions in dairy cattle. Experimental data were used and records on daily feed intake, weekly body weights, and weekly milk production were available from 548 heifers. Residual feed intake (MJ/d) is the difference between net energy intake and calculated net energy requirements for maintenance as a function of body weight and for fat- and protein-corrected milk production. Predicted methane emission (PME; g/d) is 6% of gross energy intake (Intergovernmental Panel on Climate Change methodology) corrected for energy content of methane (55.65 kJ/g). The estimated heritabilities for PME and RFI were 0.35 and 0.40, respectively. The positive genetic correlation between RFI and PME indicated that cows with lower RFI have lower PME (estimates ranging from 0.18 to 0.84). Hence, it is possible to decrease the methane production of a cow by selecting more-efficient cows, and the genetic variation suggests that reductions in the order of 11 to 26% in 10 yr are theoretically possible, and could be even higher in a genomic selection program. However, several uncertainties are discussed; for example, the lack of true methane measurements (and the key assumption that methane produced per unit feed is not affected by RFI level), as well as the limitations of predicting the biological consequences of selection. To overcome these limitations, an international effort is required to bring together data on feed intake and methane emissions of dairy cows.

Alternative Somatic Cell Count Traits as Mastitis Indicators for Genetic Selection

The aim of this study was to define alternative traits of somatic cell count (SCC) that can be used to decrease genetic susceptibility to clinical and subclinical mastitis (CM and SCM, respectively). Three kinds of SCC traits were evaluated: 1) lactation-averages of SCC, 2) traits derived from the proportion of test-day SCC above 150,000 cells/mL, and 3) patterns of peaks in SCC. Genetic parameters for these SCC traits and their genetic correlation with CM and SCM were estimated; CM and SCM were scored as binary traits. Two data sets (A and B) depending on CM recording were available. After editing, subset A contained 28,688 lactations from 21,673 cows in 394 herds. Subset B contained 56,726 lactations of 30,145 cows in 272 herds. Variance components for sire and permanent animal effects were estimated. Estimated heritabilities for all mastitis traits were around 0.03. Heritabilities for SCC traits ranged from 0.01 for patterns of peaks in SCC to 0.13 for lactation-average SCC. Genetic correlations between SCC traits and CM or SCM ranged from 0.55 to 0.93 for CM and from 0.55 to 0.98 for SCM. High genetic correlations were estimated between CM and SCC averaged over 250 d in milk (0.87), and between SCM and presence of test-day SCC >150,000 cells/mL (0.98) in subset A. In subset B, a high genetic correlation was estimated between CM and an SCC peak with a quick recovery (0.93) and between SCM and SCC averaged between 151 and 400 d (0.95). Partial genetic correlations were calculated to investigate the additional information of the alternative SCC traits, compared with lactation-average SCC. They showed that some traits remain informative for CM and others for SCM. Therefore, use of information from a combination of different SCC traits may be more successful in improving overall udder health than the traditional single SCC measure.

Improved accuracy of genomic prediction for dry matter intake of dairy cattle from combined European and Australian data sets

With the aim of increasing the accuracy of genomic estimated breeding values for dry matter intake (DMI) in dairy cattle, data from Australia (AU), the United Kingdom (UK), and the Netherlands (NL) were combined using both single-trait and multi-trait models. In total, DMI records were available on 1,801 animals, including 843 AU growing heifers with records on DMI measured over 60 to 70 d at approximately 200 d of age, and 359 UK and 599 NL lactating heifers with records on DMI during the first 100 d in milk. The genotypes used in this study were obtained from the Illumina Bovine 50K chip (Illumina Inc., San Diego, CA). The AU, UK, and NL genomic data were matched using the single nucleotide polymorphism (SNP) name. Quality controls were applied by carefully comparing the genotypes of 40 bulls that were available in each data set. This resulted in 30,949 SNP being used in the analyses. Genomic predictions were estimated with genomic REML, using ASReml software. The accuracy of genomic prediction was evaluated in 11 validation sets; that is, at least 3 validation sets per country were defined. The reference set (in which animals had both DMI phenotypes and genotypes) was either AU or Europe (UK and NL) or a multi-country reference set consisting of all data except the validation set. When DMI for each country was treated as the same trait, use of a multi-country reference set increased the accuracy of genomic prediction for DMI in UK, but not in AU and NL. Extending the model to a bivariate (AU-EU) or trivariate (AU-UK-NL) model increased the accuracy of genomic prediction for DMI in all countries. The highest accuracies were estimated for all countries when data were analyzed with a trivariate model, with increases of up to 5.5% compared with univariate models within countries.

Genomic selection for feed efficiency in dairy cattle.

Feed is a major component of variable costs associated with dairy systems and is therefore an important consideration for breeding objectives. As a result, measures of feed efficiency are becoming popular traits for genetic analyses. Already, several countries account for feed efficiency in their breeding objectives by approximating the amount of energy required for milk production, maintenance, etc. However, variation in actual feed intake is currently not captured in dairy selection objectives, although this could be possible by evaluating traits such as residual feed intake (RFI), defined as the difference between actual and predicted feed (or energy) intake. As feed intake is expensive to accurately measure on large numbers of cows, phenotypes derived from it are obvious candidates for genomic selection provided that: (1) the trait is heritable; (2) the reliability of genomic predictions are acceptable to those using the breeding values; and (3) if breeding values are estimated for heifers, rather than cows then the heifer and cow traits need to be correlated. The accuracy of genomic prediction of dry matter intake (DMI) and RFI has been estimated to be around 0.4 in beef and dairy cattle studies. There are opportunities to increase the accuracy of prediction, for example, pooling data from three research herds (in Australia and Europe) has been shown to increase the accuracy of genomic prediction of DMI from 0.33 within country to 0.35 using a three-country reference population. Before including RFI as a selection objective, genetic correlations with other traits need to be estimated. Weak unfavourable genetic correlations between RFI and fertility have been published. This could be because RFI is mathematically similar to the calculation of energy balance and failure to account for mobilisation of body reserves correctly may result in selection for a trait that is similar to selecting for reduced (or negative) energy balance. So, if RFI is to become a selection objective, then including it in an overall multi-trait selection index where the breeding objective is net profit is sensible, as this would allow genetic correlations with other traits to be properly accounted for. If genetic parameters are accurately estimated then RFI is a logical breeding objective. If there is uncertainty in these, then DMI may be preferable.

International genetic evaluations for feed intake in dairy cattle through the collation of data from multiple sources

Feed represents a large proportion of the variable costs in dairy production systems. The omission of feed intake measures explicitly from national dairy cow breeding objectives is predominantly due to a lack of information from which to make selection decisions. However, individual cow feed intake data are available in different countries, mostly from research or nucleus herds. None of these data sets are sufficiently large enough on their own to generate accurate genetic evaluations. In the current study, we collate data from 10 populations in 9 countries and estimate genetic parameters for dry matter intake (DMI). A total of 224,174 test-day records from 10,068 parity 1 to 5 records of 6,957 cows were available, as well as records from 1,784 growing heifers. Random regression models were fit to the lactating cow test-day records and predicted feed intake at 70 d postcalving was extracted from these fitted profiles. The random regression model included a fixed polynomial regression for each lactation separately, as well as herd-year-season of calving and experimental treatment as fixed effects; random effects fit in the model included individual animal deviation from the fixed regression for each parity as well as mean herd-specific deviations from the fixed regression. Predicted DMI at 70 d postcalving was used as the phenotype for the subsequent genetic analyses undertaken using an animal repeatability model. Heritability estimates of predicted cow feed intake 70 d postcalving was 0.34 across the entire data set and varied, within population, from 0.08 to 0.52. Repeatability of feed intake across lactations was 0.66. Heritability of feed intake in the growing heifers was 0.20 to 0.34 in the 2 populations with heifer data. The genetic correlation between feed intake in lactating cows and growing heifers was 0.67. A combined pedigree and genomic relationship matrix was used to improve linkages between populations for the estimation of genetic correlations of DMI in lactating cows; genotype information was available on 5,429 of the animals. Populations were categorized as North America, grazing, other low input, and high input European Union. Albeit associated with large standard errors, genetic correlation estimates for DMI between populations varied from 0.14 to 0.84 but were stronger (0.76 to 0.84) between the populations representative of high-input production systems. Genetic correlations with the grazing populations were weak to moderate, varying from 0.14 to 0.57. Genetic evaluations for DMI can be undertaken using data collated from international populations; however, genotype-by-environment interactions with grazing production systems need to be considered.

Genetic parameters of pathogen-specific incidence of clinical mastitis in dairy cows

The objective of this study was to estimate heritabilities for and genetic correlations among different pathogen-specific mastitis traits. The traits were unspecific mastitis, which is all mastitis treatments regardless of the causative pathogen as well as mastitis caused by Streptococcus dysgalactiae, Escherichia coli, coagulase-negative staphylococci (CNS), Staphylococcus aureus and Streptococcus uberis. Also groups of pathogens were investigated, Gram-negative v. Gram-positive and contagious v. environmental pathogens. Data from 168 158 Danish Holstein cows calving first time between 1998 and 2006 were used in the analyses. Variances and covariances were estimated using uni- and bivariate threshold models via Gibbs sampling. Posterior means of heritabilities of pathogen-specific mastitis were lower than the heritability of unspecific mastitis, ranging from 0.035 to 0.076 for S. aureus and S. uberis, respectively. The heritabilities of groups of pathogen ranged from 0.053 to 0.087. Genetic correlations among the pathogen-specific mastitis traits ranged from 0.45 to 0.77. These estimates tended to be lowest for bacteria eliciting very different immune responses, which can be considered as the overall pleiotropic effect of genes affecting resistance to a specific pathogen, and highest for bacteria sharing characteristics regarding immune response. The genetic correlations between the groups of pathogens were high, 0.73 and 0.83. Results showed that the pathogen-specific traits used in this study should be considered as different traits. Genetic evaluation for pathogen-specific mastitis resistance may be beneficial despite lower heritabilities than unspecific mastitis because a pathogen-specific mastitis trait is a direct measure of an udder infection, and because the cost of a mastitis case caused by different pathogens has been shown to differ greatly. Sampling bias may be present because there were not pathogen information on all mastitis treatments and because some farms do not record pathogen information. Therefore, improved recording of pathogen information and mastitis treatments in general is critical for a successful genetic evaluation of udder health. Also, economic values have to be specified for each pathogen-specific trait separately.

Applicability of day-to-day variation in behavior for the automated detection of lameness in dairy cows

Lameness is a major problem in modern dairy husbandry and has welfare implications and other negative consequences. The behavior of dairy cows is influenced by lameness. Automated lameness detection can, among other methods, be based on day-to-day variation in animal behavior. Activity sensors that measure lying time, number of lying bouts, and other parameters were used to record behavior per cow per day. The objective of this research was to develop and validate a lameness detection model based on daily activity data. Besides the activity data, milking data and data from the computerized concentrate feeders were available as input data. Locomotion scores were available as reference data. Data from up to 100 cows collected at an experimental farm during 23 mo in 2010 and 2011 were available for model development. Behavior is cow-dependent, and therefore quadratic trend models were fitted with a dynamic linear model on-line per cow for 7 activity variables and 2 other variables (milk yield per day and concentrate leftovers per day). It is assumed that lameness develops gradually; therefore, a lameness alert was given when the linear trend in 2 or more of the 9 models differed significantly from zero in a direction that corresponded with lameness symptoms. The developed model was validated during the first 4 mo of 2012 with almost 100 cows on the same farm by generating lameness alerts each week. Performance on the model validation data set was comparable with performance on the model development data set. The overall sensitivity (percentage of detected lameness cases) was 85.5% combined with specificity (percentage of nonlame cow-days that were not alerted) of 88.8%. All variables contributed to this performance. These results indicate that automated lameness detection based on day-to-day variation in behavior is a useful tool for dairy management.

Predicting energy balance for dairy cows using high-density single nucleotide polymorphism information.

The objective of this study was to investigate the genetic basis of energy balance (EB) and the potential use of genomic selection to enable EB to be incorporated into selection programs. Energy balance provides an essential link between production and nonproduction traits because both depend on a common source of energy. A small number (527) of Dutch Holstein-Friesian heifers with phenotypes for EB were genotyped. Direct genomic values were predicted for these heifers using a model that included the genotypic information. A polygenic model was also applied to predict estimated breeding values using only pedigree information. A 10-fold cross-validation approach was employed to assess the accuracies of the 2 sets of predicted breeding values by correlating them with phenotypes. Because of the small number of phenotypes, accuracies were relatively low (0.29 for the direct genomic values and 0.21 for the estimated breeding values), where the maximum possible accuracy was the square root of heritability (0.57). Despite this, the genomic model produced breeding values with reliability double that of the breeding values produced by the polygenic model. To increase the accuracy of the genomic breeding values and make it possible to select for EB, measurement and recording of EB would need to improve. The study suggests that it may be possible to select for minimally recorded traits; for instance, those measured on experimental farms, using genomic selection. Overall, the study demonstrated that genomic selection could be used to select for EB, confirming its genetic background.

Predicted accuracy of and response to genomic selection for new traits in dairy cattle.

Genomic selection relaxes the requirement of traditional selection tools to have phenotypic measurements on close relatives of all selection candidates. This opens up possibilities to select for traits that are difficult or expensive to measure. The objectives of this paper were to predict accuracy of and response to genomic selection for a new trait, considering that only a cow reference population of moderate size was available for the new trait, and that selection simultaneously targeted an index and this new trait. Accuracy for and response to selection were deterministically evaluated for three different breeding goals. Single trait selection for the new trait based only on a limited cow reference population of up to 10 000 cows, showed that maximum genetic responses of 0.20 and 0.28 genetic standard deviation (s.d.) per year can be achieved for traits with a heritability of 0.05 and 0.30, respectively. Adding information from the index based on a reference population of 5000 bulls, and assuming a genetic correlation of 0.5, increased genetic response for both heritability levels by up to 0.14 genetic s.d. per year. The scenario with simultaneous selection for the new trait and the index, yielded a substantially lower response for the new trait, especially when the genetic correlation with the index was negative. Despite the lower response for the index, whenever the new trait had considerable economic value, including the cow reference population considerably improved the genetic response for the new trait. For scenarios with a zero or negative genetic correlation with the index and equal economic value for the index and the new trait, a reference population of 2000 cows increased genetic response for the new trait with at least 0.10 and 0.20 genetic s.d. per year, for heritability levels of 0.05 and 0.30, respectively. We conclude that for new traits with a very small or positive genetic correlation with the index, and a high positive economic value, considerable genetic response can already be achieved based on a cow reference population with only 2000 records, even when the reliability of individual genomic breeding values is much lower than currently accepted in dairy cattle breeding programs. New traits may generally have a negative genetic correlation with the index and a small positive economic value. For such new traits, cow reference populations of at least 10 000 cows may be required to achieve acceptable levels of genetic response for the new trait and for the whole breeding goal.

Combining cow and bull reference populations to increase accuracy of genomic prediction and genome-wide association studies

Genomic selection holds the promise to be particularly beneficial for traits that are difficult or expensive to measure, such that access to phenotypes on large daughter groups of bulls is limited. Instead, cow reference populations can be generated, potentially supplemented with existing information from the same or (highly) correlated traits available on bull reference populations. The objective of this study, therefore, was to develop a model to perform genomic predictions and genome-wide association studies based on a combined cow and bull reference data set, with the accuracy of the phenotypes differing between the cow and bull genomic selection reference populations. The developed bivariate Bayesian stochastic search variable selection model allowed for an unbalanced design by imputing residuals in the residual updating scheme for all missing records. The performance of this model is demonstrated on a real data example, where the analyzed trait, being milk fat or protein yield, was either measured only on a cow or a bull reference population, or recorded on both. Our results were that the developed bivariate Bayesian stochastic search variable selection model was able to analyze 2 traits, even though animals had measurements on only 1 of 2 traits. The Bayesian stochastic search variable selection model yielded consistently higher accuracy for fat yield compared with a model without variable selection, both for the univariate and bivariate analyses, whereas the accuracy of both models was very similar for protein yield. The bivariate model identified several additional quantitative trait loci peaks compared with the single-trait models on either trait. In addition, the bivariate models showed a marginal increase in accuracy of genomic predictions for the cow traits (0.01-0.05), although a greater increase in accuracy is expected as the size of the bull population increases. Our results emphasize that the chosen value of priors in Bayesian genomic prediction models are especially important in small data sets.