Genetic evaluation for stillbirth and preweaning mortality in Australian dairy cattle.

The welfare of calves is important to both farmers and consumers. Practices that increase the proportion of calves born alive and enable them to thrive through to weaning contribute to improved sustainability. Stillbirths (SB) are calvings where the calf dies at birth or within 24 h after birth. Preweaning mortality (PWM) refers to calves that die after the first day of life but before weaning based on termination data. Both SB and PWM are binary traits characterized by low heritability. Data collection for these traits is incomplete, compared with traits such as milk yield in cows. Despite these challenges, genetic variation can be measured and used to produce breeding tools, such as EBVs, to reduce calf mortality over time. The aim of this study was to compare the performance of various linear models to predict SB and PWM traits in Holstein and Jersey cattle and evaluate their applicability for industry-wide use in the Australian dairy industry. Calving records from around 2.25 million Holstein and Jersey dams were obtained from DataGene's Central Data Repository from the year 2000 onward, to calculate genetic parameters. About 7% of calves were recorded as SB in the period from 2000 to 2021 (n = 1.48 million calvings). The prevalence of PWM was much lower than SB during the same period at 2% (n = 0.89 million calves). Genetic parameters were estimated for SB direct, SB maternal, and PWM using bivariate linear models with calving ease (CE) as the second trait in the model. The heritability of these calf traits was low and varied between 1% and 5% depending on the breed, trait, and model. In Holstein cattle, heritabilities were 2% for PWM and SB direct and 1% for SB maternal, whereas in Jersey cattle heritabilities were 5% for PWM, 2% for SB direct, and 1% for SB maternal. The genetic trends for both SB direct and SB maternal in Holstein cattle indicate improvement in both traits, whereas there was no apparent increase or decrease in PWM in the past 2 decades. The coefficient of genetic variation for SB direct and PWM was between 11.7% and 23.0% in Holstein and Jersey cattle, demonstrating considerable genetic variation in calf survival traits as a first step to using genetic selection to increase the proportion of calves born alive and calves weaned. A focus on improved calf and calving recording practices is expected to increase the reliability of genetic predictions.

Genetic parameters and evaluation of mortality and slaughter rate in Holstein and Jersey cows.

The longevity of dairy cattle has economic, animal welfare, and health implications and is influenced by the frequency of mortality on the farm and sale for slaughter. In this study cows removed from the herd due to death or slaughter during the lactation were coded 1 and cows that were not terminated were coded 0. Genetic parameters for mortality rates (MR) and slaughter rates (SR) were estimated for Holstein (H) and Jersey (J) breeds by applying both linear (LM) and threshold (TM) sire models using about 1.2 million H and 286,000 J cows. Estimated breeding values (EBV) for MR and SR were predicted using animal models to assess the opportunity for selection and genetic trends. Cow termination data, recorded between 1990 and 2020 on a voluntary basis by Australian dairy farmers, were analyzed. Cow MR has increased from below 1% in the 1990s to 4.1% and 3.6% in recent years in H and J cows, respectively. Most dead cows (∼36%) left the herd before 120 d of lactation, while cows that were slaughtered left the herd toward the end of the lactation. Using the LM, heritability (h2) estimates for MR were lower (1%) than those for SR (2%-3.5%). When h2 were estimated using a TM, the estimates for both traits varied between 4% and 20%, suggesting that the difference in incidence level is one of the reasons for the difference in the h2 values between MR and SR. Early test-day milk yield (MY) and 305-d MY (305-d MY) have unfavorable genetic correlations (0.32-0.41) with MR in both breeds. The genetic correlations of calving interval with MR were stronger (0.54-0.68) than with SR (0.28-0.45) suggesting that poor fertility can serve as an early indicator of poor cow health that may lead to increased risk of death. High early test-day somatic cell count is genetically associated with increased likelihood of slaughter (0.24-0.46), but not with increased likelihood of death. In H, 305-d protein yield (PY) had the strongest genetic correlation (-0.34 to -0.40) with SR whereas in J, both 305-d PY and fat yield showed high genetic (-0.64 to -0.70) and moderate environmental (-0.35 to -0.37) correlations with SR. The genetic correlation of removal from the herd due to death and slaughter was negative (-0.3) in J and zero in H. Strong selection for improved fertility and survival and less selection emphasis for MY, has led to an improvement in the genetic trend for cow MR in H and the trend in J has stabilized. Although genetic evaluations for cow MR are feasible, the reliabilities of the EBV are low and the level of cow MR in Australia are relatively low compared with similar countries. Therefore, genetic evaluation for survival based on mortality and slaughter data could be sufficient in the current selection circumstances where breeding objectives are broadly defined. Nevertheless, all Australian farmers should be encouraged to continue recording mortality and slaughter data for monitoring of the trends and for future development of genetic evaluations.

Machine learning approaches for the prediction of lameness in dairy cows.

Lameness is one of the costliest health problems, as well as a welfare concern in dairy cows. However, it is difficult to detect cows with possible lameness, or the ones that are at risk of becoming lame e.g. in the next week or so. In this study, we investigated the ability of three machine learning algorithms, Naïve Bayes (NB), Random Forest (RF) and Multilayer Perceptron (MLP), to predict cases of lameness using milk production and conformation traits. The performance of these algorithms was compared with logistic regression (LR) as the gold standard approach for binary classification. We had a total of 2 535 lameness scores (2 248 sound and 287 unsound) and 29 predictor features from nine dairy herds in Australia to predict lameness incidence. Training was done on 80% of the data within each herd with the remainder used as validation set. Our results indicated that in terms of area under curve of receiver operating characteristics, there were negligible differences between LR (0.67) and NB (0.66) while MLP (0.62) and RF (0.61) underperformed compared to the other two methods. However, the F1-score in NB (27%) outperformed LR (1%), suggesting that NB could potentially be a more reliable method for the prediction of lameness in practice, given enough relevant data are available for proper training, which was a limitation in this study. Considering the small size of our dataset, lack of information about environmental conditions prior to the incidence of lameness, management practices, short time gap between production records and lameness scoring, and farm information, this study proved the concept of using machine learning predictive models to predict the incidence of lameness a priori to its occurrence and thus may become a valuable decision support system for better lameness management in precision dairy farming.

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.

Including milk production, conformation, and functional traits in multivariate models for genetic evaluation of lameness.

Lameness is a serious health and welfare issue that can negatively affect the economic performance of cows, especially on pasture-based dairy farms. However, most genetic predictions (GP) of lameness have low accuracy because lameness data are often incomplete as data are collected voluntarily by farmers in countries such as Australia. The objective of this study was to find routinely measured traits that are correlated with lameness and use them in multivariate evaluation models to improve the accuracy of GP for lameness. We used health events and treatments associated with lameness recorded by Australian farmers from 2002 to early 2019. The lameness incidence rates in Holstein and Jersey cows were 3.3% and 4.6%, respectively. We analyzed the records of 36 other traits (milk production, conformation, fertility, and survival traits) to estimate genetic correlations with lameness. The estimated heritability ± standard error (and repeatability ± standard error) for lameness in both Holstein and Jersey breeds were very low: 0.007 ± 0.002 (and 0.029 ± 0.002) and 0.005 ± 0.003 (and 0.027 ± 0.006), respectively, in univariate sire models. For the GP models, we tested including measurements of overall type to prediction models for Holsteins, stature and body length for Jersey, and milk yield and fertility traits for both breeds. The average accuracy of GP, calculated from prediction error variances, were 0.38 and 0.24 for Holstein and Jersey sires, respectively, when estimated using univariate sire models and both increased to 0.43 using multivariate sire models. In conclusion, we found that the accuracy of GP for lameness could be improved by including genetically correlated traits in a multivariate model. However, to further improve the accuracy of predictions of lameness, precise identification and recording incidences of hoof or leg disorder, or large-scale recording of locomotion and claw scores by trained personnel should be considered.

Genetic parameters for mid-infrared spectroscopy-predicted fertility.

Female fertility is a challenging trait to improve genetically because of its low heritability, its unfavorable genetic correlation with milk yield, and its relatively small number of records. The MFERT trait is the probability of conception to first insemination predicted using mid-infrared (MIR) spectroscopy of a milk sample collected during lactation as part of routine milk recording, age at calving, days in milk, and milk production. As such, MFERT could become available on many more cows than traditional fertility traits. Our objectives were (1) to estimate the heritability of MFERT; (2) to estimate genetic correlations between MFERT, traditional fertility traits, and milk production traits; and (3) to assess the potential of MFERT to be used as an indicator trait for fertility in a selection index. The MFERT trait had a heritability of 0.16, which was higher than that (0.05) obtained for traditional fertility traits. Genetic correlations between MFERT and traditional fertility traits were low to moderate. The weakest and strongest correlations (mean ± standard error) were with pregnancy at the end of the mating season (0.13 ± 0.05) and calving to first service (-0.61 ± 0.03), respectively. Based on our estimates, including MFERT in a fertility index will only substantially increase the accuracy of the index when there are many more records available for MFERT than for the traditional fertility traits. This is likely to be the case because the number of milk samples from commercial machines belonging to milk recording companies in Australia that are capable of generating MIR spectra is growing. Hence, the number of records for MFERT is expected to increase substantially in the near future.

The use of milk mid-infrared spectroscopy to improve genomic prediction accuracy of serum biomarkers.

Breeding objectives in the dairy industry have shifted from being solely focused on production to including fertility, animal health, and environmental impact. Increased serum concentrations of candidate biomarkers of health and fertility, such as ß-hydroxybutyric acid (BHB), fatty acids, and urea are difficult and costly to measure, and thus limit the number of records. Accurate genomic prediction requires a large reference population. The inclusion of milk mid-infrared (MIR) spectroscopic predictions of biomarkers may increase genomic prediction accuracy of these traits. Our objectives were to (1) estimate the heritability of, and genetic correlations between, selected serum biomarkers and their respective MIR predictions, and (2) evaluate genomic prediction accuracies of either only measured serum traits, or serum traits plus MIR-predicted traits. The MIR-predicted traits were either fitted in a single trait model, assuming the measured trait and predicted trait were the same trait, or in a multitrait model, where measured and predicted trait were assumed to be correlated traits. We performed all analyses using relationship matrices constructed from pedigree (A matrix), genotypes (G matrix), or both pedigree and genotypes (H matrix). Our data set comprised up to 2,198 and 9,657 Holstein cows with records for serum biomarkers and MIR-predicted traits, respectively. Heritabilities of measured serum traits ranged from 0.04 to 0.07 for BHB, from 0.13 to 0.21 for fatty acids, and from 0.10 to 0.12 for urea. Heritabilities for MIR-predicted traits were not significantly different from those for the measured traits. Genetic correlations between measured traits and MIR-predicted traits were close to 1 for urea. For BHB and fatty acids, genetic correlations were lower and had large standard errors. The inclusion of MIR predicted urea substantially increased prediction accuracy for urea. For BHB, including MIR-predicted BHB reduced the genomic prediction accuracy, whereas for fatty acids, prediction accuracies were similar with either measured fatty acids, MIR-predicted fatty acids, or both. The high genetic correlation between urea and MIR-predicted urea, in combination with the increased prediction accuracy, demonstrated the potential of using MIR-predicted urea for genomic prediction of urea. For BHB and fatty acids, further studies with larger data sets are required to obtain more accurate estimates of genetic correlations.

Value of sharing cow reference population between countries on reliability of genomic prediction for milk yield traits.

Increasing the reliability of genomic prediction (GP) of economic traits in the pasture-based dairy production systems of New Zealand (NZ) and Australia (AU) is important to both countries. This study assessed if sharing cow phenotype and genotype data of NZ and AU improves the reliability of GP for NZ bulls. Data from approximately 32,000 NZ genotyped cows and their contemporaries were included in the May 2018 routine genetic evaluation of the Australian Dairy cattle in an attempt to provide consistent phenotypes for both countries. After the genetic evaluation, deregressed proofs of cows were calculated for milk yield traits. The April 2018 multiple across-country evaluation of Interbull was also used to calculate deregressed proofs for bulls on the NZ scale. Approximately 1,178 Jersey (Jer) and 6,422 Holstein (Hol) bulls had genotype and phenotype data. In addition to NZ cows, phenotype data of close to 60,000 genotyped Australian (AU) cows from the same genetic evaluation run as NZ cows were used. All AU and NZ females were genotyped using low-density SNP chips (<10K SNP) and were imputed first to 50K and then to ∼600K (referred to as high density; HD). We used up to 98,000 animals in the reference populations, both by expanding the NZ reference set (cow, bull, single breed to multi-breed set) and by adding AU cows. Reliabilities of GP were calculated for 508 Jer and 1,251 Hol bulls whose sires are not included in the reference set (RS) to ensure that real differences are not masked by close relationships. The GP was tested using 50K or high-density SNP chip using genomic BLUP in bivariate (considering country as a trait) or single trait models. The RS that gave the highest reliability for each breed were also tested using a hybrid GP method that combines expectation maximization with Bayes R. The addition of the AU cows to an NZ RS that included either NZ cows only, or cows and bulls, improved the reliability of GP for both NZ Hol and Jer validation bulls for all traits. Using single breed reference populations also increased reliability when NZ crossbred cows were added to reference populations that included only purebred NZ bulls and cows and AU cows. The full multi-breed RS (all NZ cows and bulls and AU cows) provided similar reliabilities in NZ Hol bulls, when compared with the single breed reference with crossbred NZ cows. For Jer validation bulls, the RS that included Jer cows and bulls and crossbred cows from NZ and Jer cows from AU was marginally better than the all-breed, all-country RS. In terms of reliability, the advantage of the HD SNP chip was small but captured more of the genomic variance than the 50K, particularly for Hol. The expectation maximization Bayes R GP method was slightly (up to 3 percentage points) better than genomic BLUP. We conclude that GP of milk production traits in NZ bulls improves by up to 7 percentage points in reliability by expanding the NZ reference population to include AU cows.

Genotype-by-environment (temperature-humidity) interaction of milk production traits in Australian Holstein cattle.

Dairying in Australia is practiced in highly diverse climatic conditions and production systems, which means that re-ranking of genotypes could occur across environments that vary in temperature and humidity-that is, genotype-by-environment interactions (G × E) may exist. The objective of this study was to investigate G × E for heat tolerance with respect to milk production traits in Australian Holsteins. A total of 6.7 million test-day milk yield records for first, second, and third lactations from 491,562 cows and 6,410 sires that had progeny in different climatic environments were included in the analysis. The environmental gradient used was the temperature-humidity index (THI) calculated from climate data from 163 Australian public weather stations between 2003 and 2017. Data were analyzed using univariate reaction norm (RM) sire model, and the results were compared with multi-trait model (MT). The MT analysis treated test-day yields at 5th percentile (THI = 61; i.e., thermoneutral conditions), 50th percentile (THI = 67; i.e., moderate heat stress conditions), and 95th percentile (THI = 73; i.e., high heat stress conditions) of the trajectory of THI as correlated traits. A THI series of 61, 67, and 73, for example, is equivalent to average temperature and relative humidity of approximately 20°C and 45%, 25°C and 45%, and 31°C and 50%, respectively. We observed some degree of heterogeneity of additive (AG) and permanent environmental (PE) variance over the trajectory THI from RM analysis, with estimates decreasing at higher THI values more steeply for PE than for AG variance. The genetic correlations of the tests between the 5th and 95th percentiles of THI for milk, protein, and fat yield from RM were 0.88 ± 0.01 (standard error), 0.79 ± 0.01, and 0.86 ± 0.01, respectively, whereas the corresponding estimates from MT were 0.86 ± 0.02, 0.84 ± 0.03, and 0.87 ± 0.03. We observed lower genetic correlations between the 5th and 95th percentiles of THI for milk tests from recent years (i.e., 2009 and 2017) compared with earlier years (i.e., 2003 and 2008), which suggests that the level of G × E is increasing in the studied population and should be monitored especially in anticipation of future expected increase in daily average temperature and frequency of heat events. Overall, our results indicate presence of G × E at the upper extreme of the trajectory of THI, but the current extent of sire re-ranking may not justify providing separate genetic evaluations for different levels of heat stress. However, variations observed in the sire sensitivity to heat stress suggest that dairy herds in high heat load conditions could benefit more from using heat-tolerant or resilient sires.

Genetic evaluation of gestation length and its use in managing calving patterns.

Data collected from Australian Holstein cows that calved between 1995 and 2016 were used for estimating genetic parameters and genetic evaluation of gestation length (GL). Genetic parameters were estimated using a sire maternal grandsire model considering GL in heifers as correlated, but a different trait to adult cows and animal repeatability model. The key objective of this study was to assess if selective mating of bulls with short GL estimated breeding values (EBV) can help to modify calving patterns in predominantly pasture-based production systems and thereby contribute to the reduction of calving induction. The mean GL in heifers and adult cows was 280 and 281 d, respectively. The heritability of direct GL was 0.28 in heifers, which is slightly lower than in adult cows (0.36). The maternal heritability of GL was close to 0.04 in both heifer and adult cows. The genetic correlation between direct effects in heifers and adult cows was lower (0.88) than between maternal effects (0.94). A genetic evaluation model that considered heifer and adult cow data as the same trait in a repeatability animal model showed adequate variability in EBV for both direct and maternal GL. The genetic trend in direct GL EBV declined from 2005 in bulls and from 2006 in cows. Selective mating of bulls with short direct GL EBV showed that the GL and calving interval of their mates can be modified by up to 3.5 d for GL and 2.0 d for calving interval. Relative to parent average, the use of genotype data to calculate genomic EBV increased the reliability of EBV by up to 11% in validation bulls when daughter trait deviation of bulls with trait deviation of cows (11%) and deregressed breeding values (8%) were used as response variables. On average, for animals with only genotype data, the GL EBV can be predicted with about 50 to 60% reliability (expected) depending on the response variable (deregressed breeding values or daughter trait deviation and trait deviation) and the size of reference set. Overall, the results of this study show that calving patterns can be made tighter by selectively mating short GL EBV bulls to cows that do not become pregnant early in the mating season. Additionally, better reproductive management and the use of bulls with high female fertility EBV are still crucial for the success of a pasture-based dairy production system.

Components of the covariances between reproductive performance traits and milk protein concentration and milk yield in dairy cows.

Reproductive performance in dairy cows can be improved through genetic selection and herd management. Milk protein concentration is strongly associated with various measures of reproductive performance, but the relative importance of genetic and environmental components of these associations have not been defined. The primary objective of this study was to estimate the magnitudes of correlations and covariances between 9 reproductive performance traits in dairy cows and each of milk protein concentration and milk yield at 4 levels: genetic, permanent environmental effects of cow, herd-year-season, and residual levels. A retrospective single cohort study was conducted using data collected from seasonally and split calving dairy herds. We used animal models to partition covariances for the relationships between 9 fertility traits and each of milk protein concentration and milk yield at lactation level, with up to 80,203 lactations from 27,244 cows that were 780 herd-year-seasons in 65 herds. For the fertility traits, of the explained covariance with milk protein concentration, between 33 and 79% (median 53%) was genetic and 21 to 67% (median 47%) was nongenetic. We concluded that research should be conducted to identify management strategies that capture the nongenetic components of relationships between milk protein concentration and reproductive performance. Genetic correlations with milk protein concentration were generally similar to genetic correlations with milk yield, but the correlation with milk protein concentration was closer (i.e., the absolute value of the correlation coefficient was nearer to 1) for pregnant by wk 6, a key trait for seasonally and split calving dairy herds (correlation coefficient ± standard error = 0.28 ± 0.05 and -0.17 ± 0.07 for milk protein concentration and milk yield, respectively). As the associations also have substantial genetic components, it is possible that reliabilities of estimated breeding values for fertility may be improved by including milk protein concentration in multitrait genetic evaluation models for fertility traits. From our preliminary analyses, reliabilities were only slightly higher when pregnancy by wk 6 of the breeding period was analyzed with milk protein concentration rather than alone or with milk yield, but further research should be considered to assess this question. Importantly, the benefits of these strong relationships can only be fully harnessed through joint use of both management strategies and genetic strategies.

Genetic parameters for health traits using data collected from genomic information nucleus herds.

As with many other countries, data availability has been a limitation in Australia for developing breeding values for health traits. A genomic information nucleus of approximately 100 herds across the country, selected on the basis of their record keeping, has recently been established in Australia and is known as Ginfo. The objective of this study was to evaluate the feasibility of developing genomic breeding values for health traits using farmer-collected data from Ginfo herds. Having this genotyped population opens up opportunities to develop new genomic breeding values, such as dairy health traits. In these herds, the 4 most recorded groups of health diseases identified were mastitis, reproductive problems, lameness, and metabolic disorders with incidence levels of 16, 9, 2, and 1.5%, respectively. Heritability estimates from pedigree and genomic analysis ranged from 0.01 to 0.03 for mastitis, 0.005 to 0.02 for reproductive disorders, 0 to 0.02 for lameness, and 0.00 to 0.06 for metabolic disorders. However, although heritability is low, there is sufficient genetic variation to make genetic progress feasible (coefficient of variation ranging from 3 to 26.3%). Genetic correlations between health and milk production traits (0.08 to 0.48) and overall type (-0.00 to 0.57) are unfavorable, but favorable with other economical traits, such as fertility (0.10 to 0.51), survival (-0.16 to -0.37), and somatic cell count (0.07 to 0.55). For a genomic reference population of 11,458 cows, the reliability of genomic predictions is comparatively low (<30%), but is promising for health traits, such as mastitis (∼30%) and a broader-based all disease trait (∼28%), when bulls have daughters in the reference population, but not when they only have genomic information (0 to 18%). Further improvement of the reliabilities of health breeding values continues to be an important objective. The study has provided a good foundation for future research to develop breeding values for health traits.

Genetic parameters for lactose and its correlation with other milk production traits and fitness traits in pasture-based production systems.

Lactose is a major component of milk (typically around 5% of composition) that is not usually directly considered in national genetic improvement programs of dairy cattle. Daily test-day lactose yields and percentage data from pasture-based seasonal calving herds in Australia were analyzed to assess if lactose content can be used for predicting fitness traits and if an additional benefit is achieved by including lactose yield in selecting for milk yield traits. Data on lactose percentage collected from 2007 to 2014, from about 600 herds, were used to estimated genetic parameters for lactose percentage and lactose yield and correlations with other milk yield traits, somatic cell count (SCC), calving interval (CIV), and survival. Daily test-day data were analyzed using bivariate random regression models. In addition, multi-trait models were also performed mainly to assess the value of lactose to predict fitness traits. The heritability of lactose percentage (0.25 to 0.37) was higher than lactose yield (0.11 to 0.20) in the first parity. Genetically, the correlation of lactose percentage with protein percentage varied from 0.3 at the beginning of lactation to -0.24 at the end of the lactation in the first parity. Similar patterns in genetic correlations were also observed in the second and third parity. At all levels (i.e., genetic, permanent environmental, and residual), the correlation between milk yield and lactose yield was close to 1. The genetic and permanent environmental correlations between lactose percentage and SCC were stronger in the second and third parity and toward the end of the lactation (-0.35 to -0.50) when SCC levels are at their maximum. The genetic correlation between lactose percentage in the first 120 d and CIV (-0.23) was similar to correlation of CIV with protein percentage (-0.28), another component trait with the potential to predict fertility. Furthermore, the correlations of estimated breeding values of lactose percentage and estimated breeding values of traits such as survival, fertility, SCC, and angularity suggest that the value of lactose percentage as a predictor of fitness traits is weak. The results also suggest that including lactose yield as a trait into the breeding objective is of limited value due to the high positive genetic correlation between lactose yield and protein yield, the trait highly emphasized in Australia. However, recording lactose percentage as part of the routine milk recording system will enable the Australian dairy industry to respond quickly to any future changes and market signals.

Exploiting biological priors and sequence variants enhances QTL discovery and genomic prediction of complex traits.

BACKGROUND: Dense SNP genotypes are often combined with complex trait phenotypes to map causal variants, study genetic architecture and provide genomic predictions for individuals with genotypes but no phenotype. A single method of analysis that jointly fits all genotypes in a Bayesian mixture model (BayesR) has been shown to competitively address all 3 purposes simultaneously. However, BayesR and other similar methods ignore prior biological knowledge and assume all genotypes are equally likely to affect the trait. While this assumption is reasonable for SNP array genotypes, it is less sensible if genotypes are whole-genome sequence variants which should include causal variants. RESULTS: We introduce a new method (BayesRC) based on BayesR that incorporates prior biological information in the analysis by defining classes of variants likely to be enriched for causal mutations. The information can be derived from a range of sources, including variant annotation, candidate gene lists and known causal variants. This information is then incorporated objectively in the analysis based on evidence of enrichment in the data. We demonstrate the increased power of BayesRC compared to BayesR using real dairy cattle genotypes with simulated phenotypes. The genotypes were imputed whole-genome sequence variants in coding regions combined with dense SNP markers. BayesRC increased the power to detect causal variants and increased the accuracy of genomic prediction. The relative improvement for genomic prediction was most apparent in validation populations that were not closely related to the reference population. We also applied BayesRC to real milk production phenotypes in dairy cattle using independent biological priors from gene expression analyses. Although current biological knowledge of which genes and variants affect milk production is still very incomplete, our results suggest that the new BayesRC method was equal to or more powerful than BayesR for detecting candidate causal variants and for genomic prediction of milk traits. CONCLUSIONS: BayesRC provides a novel and flexible approach to simultaneously improving the accuracy of QTL discovery and genomic prediction by taking advantage of prior biological knowledge. Approaches such as BayesRC will become increasing useful as biological knowledge accumulates regarding functional regions of the genome for a range of traits and species.

Improving the reliability of female fertility breeding values using type and milk yield traits that predict energy status in Australian Holstein cattle.

The objectives of this study were (1) to propose changing the selection criteria trait for evaluating fertility in Australia from calving interval to conception rate at d 42 after the beginning of the mating season and (2) to use type traits as early fertility predictors, to increase the reliability of estimated breeding values for fertility. The breeding goal in Australia is conception within 6 wk of the start of the mating season. Currently, the Australian model to predict fertility breeding values (expressed as a linear transformation of calving interval) is a multitrait model that includes calving interval (CVI), lactation length (LL), calving to first service (CFS), first nonreturn rate (FNRR), and conception rate. However, CVI has a lower genetic correlation with the breeding goal (conception within 6 wk of the start of the mating season) than conception rate. Milk yield, type, and fertility data from 164,318 cow sired by 4,766 bulls were used. Principal component analysis and genetic correlation estimates between type and fertility traits were used to select type traits that could subsequently be used in a multitrait analysis. Angularity, foot angle, and pin set were chosen as type traits to include in an index with the traits that are included in the multitrait fertility model: CVI, LL, CFS, FNRR, and conception rate at d 42 (CR42). An index with these 8 traits is expected to achieve an average bull first proof reliability of 0.60 on the breeding objective (conception within 6 wk of the start of the mating season) compared with reliabilities of 0.39 and 0.45 for CR42 only or the current 5-trait Australian model. Subsequently, we used the first eigenvector of a principal component analysis with udder texture, bone quality, angularity, and body condition score to calculate an energy status indicator trait. The inclusion of the energy status indicator trait composite in a multitrait index with CVI, LL, CFS, FNRR, and CR42 achieved a 12-point increase in fertility breeding value reliability (i.e., increased by 30%; up to 0.72 points of reliability), whereas a lower increase in reliability (4 points, i.e., increased by 10%) was obtained by including angularity, foot angle, and pin set in the index. In situations when a limited number of daughters have been phenotyped for CR42, including type data for sires increased reliabilities compared with when type data were omitted. However, sires with more than 80 daughters with CR42 records achieved reliability estimates close to 80% on average, and there did not appear to be a benefit from having daughters with type records. The cost of phenotyping to obtain such reliabilities (assuming a cost of AU$14 per cow with type data and AU$5 per cow with pregnancy diagnosed) is lower if more pregnancy data are collected in preference to type data. That is, efforts to increase the reliability of fertility EBV are most cost effective when directed at obtaining a larger number of pregnancy tests.

Variances and correlations of milk production, fertility, longevity, and type traits over time in Australian Holstein cattle.

When using historical data, it is often assumed that the genetic correlation of the same trait recorded at different time points is reasonably close to 1. However, selection and possible changes in trait definitions means that this may not necessarily be the case. Regularly monitoring genetic parameters over time is important, as changes could reduce the accuracy of genetic evaluations. About 20 yr (1993 to 2012) of data on milk yield as well as functional and type traits from Australian Holstein dairy cattle were analyzed to assess changes in genetic correlations within and among traits over time by considering 2 traits at a time using linear random regression (RR) and multitrait (MT) models. Both residual and genetic variances for milk yield traits and calving interval (CI) increased over time, with the highest increase observed for protein yield. For most type traits some fluctuations over time were noted in both the residual and additive genetic variances. Genetic correlations among survival (i.e., from first to second lactation), milk yield traits, CI, and some type traits varied over time. The genetic correlation of the same trait (e.g., protein yield, fat yield, and some type traits) measured in different years was also less than 1.0 (0.1-0.9), which is likely to be due to selection or changes in trait definitions. Estimates of parameters from the RR model were generally similar to those from MT models that considered the same trait recorded in different year groups as different traits. However, in the case of survival and CI (i.e., lowly heritable traits), the genetic correlations over time obtained from the MT model were lower (0.21 to 0.75) than those from the RR models (0.9-1.0). Genetic correlations of survival with milk, fat, and protein yields declined from ~0.4 to 0.5 at the beginning of the study period (1993/94) to zero or negative at the end (2009/10), whereas the correlation between CI and milk yield became more unfavorable and increased from 0.3 to 0.5 over the same time period. The same pattern was observed for the genetic correlation between survival and CI, which also became more unfavorable over time and increased from 0.67 to 0.87 in absolute value. The genetic correlations of survival with type traits, such as angularity and body depth, decreased from near zero to negative (-0.3 to -0.4). But genetic correlations between pin set and survival showed less variation (0.2-0.3) over time. Similarly the genetic correlation of CI with body depth and angularity became more antagonistic over time. Over time the importance of traits such as milk yield and overall type as criteria for culling decreased, whereas the importance of fertility and possible disease incidence increased, implying that there has been a switch from voluntary to involuntary reasons dominating culling decisions. Changes in genetic correlations of the same trait and among traits over time have important implications on the accuracy of prediction of traits, such as survival and CI, which often rely on other traits as predictors and ultimately on the accuracy of genetic evaluations (traditional and genomic), and also the prediction of response to selection.

Including overseas performance information in genomic evaluations of Australian dairy cattle.

In dairy cattle, the rate of genetic gain from genomic selection depends on reliability of direct genomic values (DGV). One option to increase reliabilities could be to increase the size of the reference set used for prediction, by using genotyped bulls with daughter information in countries other than the evaluating country. The increase in reliabilities of DGV from using this information will depend on the extent of genotype by environment interaction between the evaluating country and countries contributing information, and whether this is correctly accounted for in the prediction method. As the genotype by environment interaction between Australia and Europe or North America is greater than between Europe and North America for most dairy traits, ways of including information from other countries in Australian genomic evaluations were examined. Thus, alternative approaches for including information from other countries and their effect on the reliability and bias of DGV of selection candidates were assessed. We also investigated the effect of including overseas (OS) information on reliabilities of DGV for selection candidates that had weaker relationships to the current Australian reference set. The DGV were predicted either using daughter trait deviations (DTD) for the bulls with daughters in Australia, or using this information as well as OS information by including deregressed proofs (DRP) from Interbull for bulls with only OS daughters in either single trait or bivariate models. In the bivariate models, DTD and DRP were considered as different traits. Analyses were performed for Holstein and Jersey bulls for milk yield traits, fertility, cell count, survival, and some type traits. For Holsteins, the data used included up to 3,580 bulls with DTD and up to 5,720 bulls with only DRP. For Jersey, about 900 bulls with DTD and 1,820 bulls with DRP were used. Bulls born after 2003 and genotyped cows that were not dams of genotyped bulls were used for validation. The results showed that the combined use of DRP on bulls with OS daughters only and DTD for Australian bulls in either the single trait or bivariate model increased the coefficient of determination [(R(2)) (DGV,DTD)] in the validation set, averaged across 6 main traits, by 3% in Holstein and by 5% in Jersey validation bulls relative to the use of DTD only. Gains in reliability and unbiasedness of DGV were similar for the single trait and bivariate models for production traits, whereas the bivariate model performed slightly better for somatic cell count in Holstein. The increase in R(2) (DGV,DTD) as a result of using bulls with OS daughters was relatively higher for those bulls and cows in the validation sets that were less related to the current reference set. For example, in Holstein, the average increase in R(2) for milk yield traits when DTD and DRP were used in a single trait model was 23% in the least-related cow group, but only 3% in the most-related cow group. In general, for both breeds the use of DTD from domestic sources and DRP from Interbull in a single trait or bivariate model can increase reliability of DGV for selection candidates.

Incorporating heifer feed efficiency in the Australian selection index using genomic selection.

The economic benefit of expanding the Australian Profit Ranking (APR) index to include residual feed intake (RFI) was evaluated using a multitrait selection index. This required the estimation of genetic parameters for RFI and genetic correlations using single nucleotide polymorphism data (genomic) correlations with other traits. Heritabilities of RFI, dry matter intake (DMI), and all the traits in the APR (milk, fat, and protein yields; somatic cell count; fertility; survival; milking speed; and temperament), and genomic correlations between these traits were estimated using a Bayesian framework, using data from 843 growing Holstein heifers with phenotypes for DMI and RFI, and bulls with records for the other traits. Heritability estimates of DMI and RFI were 0.44 and 0.33, respectively, and the genomic correlation between them was 0.03 and nonsignificant. The genomic correlations between the feed-efficiency traits and milk yield traits were also close to zero, ranging between -0.11 and 0.10. Positive genomic correlations were found for DMI with stature (0.16) and with overall type (0.14), suggesting that taller cows eat more as heifers. One issue was that the genomic correlation estimates for RFI with calving interval (ClvI) and with body condition score were both unfavorable (-0.13 and 0.71 respectively), suggesting an antagonism between feed efficiency and fertility. However, because of the relatively small numbers of animals in this study, a large 95% probability interval existed for the genomic correlation between RFI and ClvI (-0.66, 0.36). Given these parameters, and a genetic correlation between heifer and lactating cow RFI of 0.67, inclusion of RFI in the APR index would reduce RFI by 1.76 kg/cow per year. Including RFI in the APR would result in the national Australian Holstein herd consuming 1.73 × 10(6) kg less feed, which is worth 0.55 million Australian dollars (A$) per year and is 3% greater than is currently possible to achieve. Other traits contributing to profitability, such as milk production and fertility, will also improve through selection on this index; for example, ClvI would be reduced by 0.53 d/cow per year, which is 96% of the gain for this trait that is achieved without RFI in the APR.

Prediction of liveweight of cows from type traits and its relationship with production and fitness traits.

Liveweight (LWT) data for Australian Holstein cows was predicted from different type traits based on actual LWT and type data of 932 cows collected from 20 different herds over a 3-yr period. In addition to LWT measured using scales, visual estimates of LWT were also available on 90% of the cows with LWT data. The future predictive ability of different models was assessed using 10-fold cross-validation. The relationships between LWT and selected type traits, including body condition score (BCS), were also calculated to assess the usefulness of various traits to predict LWT genetically. The relationships of predicted LWT measures with production and fitness traits were also estimated in an attempt to assess the consequence of continuous selection on an economic index that includes predicted LWT with a negative economic value. The heritability of LWT was 0.4. Among type traits, stature, chest width, bone quality, BCS, udder depth, central ligament, and muzzle width were correlated with LWT both genetically and phenotypically and were used to predict LWT of cows. Predicted LWT measures, using several sets of traits and visually estimated LWT were genetically highly correlated with each other (>0.73). Phenotypically, visually estimated LWT of cows was slightly more correlated with actual LWT than that predicted from type traits, but genetically both approaches gave the same accuracy. The predicted estimates of LWT were also positively genetically correlated with energy-corrected milk yield and had near zero correlation with survival. The correlations of different measures of LWT with fertility traits were unfavorable or near zero, suggesting that selection for reduced LWT may not cause deterioration in fertility traits. However, it may be useful to consider broadening the breeding objective to include traits that are associated with energy balance, particularly if traits such as BCS and bone quality are included in the set of traits used to predict LWT. Based on the results from this study, the inclusion of predicted LWT with negative economic values into the breeding objective would have no negative effect on fitness traits.