T. Yin

Inferring relationships between clinical mastitis, productivity and fertility: a recursive model application including genetics, farm associated herd management, and cow-specific antibiotic treatments.

A dataset of test-day records, fertility traits, and one health trait including 1275 Brown Swiss cows kept in 46 small-scale organic farms was used to infer relationships among these traits based on recursive Gaussian-threshold models. Test-day records included milk yield (MY), protein percentage (PROT-%), fat percentage (FAT-%), somatic cell score (SCS), the ratio of FAT-% to PROT-% (FPR), lactose percentage (LAC-%), and milk urea nitrogen (MUN). Female fertility traits were defined as the interval from calving to first insemination (CTFS) and success of a first insemination (SFI), and the health trait was clinical mastitis (CM). First, a tri-trait model was used which postulated the recursive effect of a test-day observation in the early period of lactation on liability to CM (LCM), and further the recursive effect of LCM on the following test-day observation. For CM and female fertility traits, a bi-trait recursive Gaussian-threshold model was employed to estimate the effects from CM to CTFS and from CM on SFI. The recursive effects from CTFS and SFI onto CM were not relevant, because CM was recorded prior to the measurements for CTFS and SFI. Results show that the posterior heritability for LCM was 0.05, and for all other traits, heritability estimates were in reasonable ranges, each with a small posterior SD. Lowest heritability estimates were obtained for female reproduction traits, i.e. h(2)=0.02 for SFI, and h(2)≈0 for CTFS. Posterior estimates of genetic correlations between LCM and production traits (MY and MUN), and between LCM and somatic cell score (SCS), were large and positive (0.56-0.68). Results confirm the genetic antagonism between MY and LCM, and the suitability of SCS as an indicator trait for CM. Structural equation coefficients describe the impact of one trait on a second trait on the phenotypic pathway. Higher values for FAT-% and FPR were associated with a higher LCM. The rate of change in FAT-% and in FPR in the ongoing lactation with respect to the previous LCM was close to zero. Estimated recursive effects between SCS and CM were positive, implying strong phenotypic impacts between both traits. Structural equation coefficients explained a detrimental impact of CM on female fertility traits CTFS and SFI. The cow-specific CM treatment had no significant impact on performance traits in the ongoing lactation. For most treatments, beta-lactam-antibiotics were used, but test-day SCS and production traits after the beta-lactam-treatment were comparable to those after other antibiotic as well as homeopathic treatments.

Strategy for the simulation and analysis of longitudinal phenotypic and genomic data in the context of a temperature × humidity-dependent covariate

A simulation study was conducted to evaluate the performance of genomic random regression models for the continuous environmental descriptor temperature-humidity index (THI). Statistically innovative aspects of the study included the combined simulation of both longitudinal phenotypic data representing the same trait in the course of THI and genomic data. The longitudinal trait was simulated (phenotypically expressed) at 5 different values of THI. For a moderate heritability trait, heritabilities were 0.30, 0.35, 0.40, 0.40, and 0.35 for THI of 15, 30, 45, 60 and 75, respectively. In a consecutive run, low heritabilities of 0.05, 0.1, 0.15, 0.15, and 0.10 were simulated, respectively. On the genomic level, simulation combined high and low linkage disequilibrium with 5,000-, 15,000-, and 50,000-SNP chip applications to simulate different scenarios of genomic architecture. With regard to data analyses, 2 strategies were applied to evaluate the accuracy of genomic predictions across THI, with special focus on the extreme ends of the environmental scale. In the first strategy, 100, 80, 50, or 20% of phenotypes at THI 75 were deleted randomly and the remaining data set was used to predict the breeding value at THI 75 for non-phenotyped, but genotyped cows. In the second strategy, 1,600 cows had complete information (genotypes and phenotypes) and 400 cows were genotyped, but with missing phenotypes for all THI. For the first strategy and without phenotypic observations at THI 75, accuracies of genomic predictions were lower than 0.34. When only 20% of cows had phenotypic records at THI 75, accuracies increased (~0.60). Such a small proportion of phenotyped cows was sufficient to predict reliable genomic breeding values for cows without phenotypes for extreme THI. For the second strategy, also for low linkage disequilibrium combined with a low density 5,000-SNP chip, the average accuracy of genomic predictions was 0.52, which is substantially higher than accuracies based on pedigree relationships. From a practical perspective, genomic random regression models can be used to predict genomic breeding values for scarce phenotypes (e.g., novel traits) traits measured in extreme environments, or traits measured late in life, such as longevity.

Genetic line comparisons and genetic parameters for endoparasite infections and test-day milk production traits

Keeping dairy cows in grassland systems relies on detailed analyses of genetic resistance against endoparasite infections, including between- and within-breed genetic evaluations. The objectives of this study were (1) to compare different Black and White dairy cattle selection lines for endoparasite infections and (2) the estimation of genetic (co)variance components for endoparasite and test-day milk production traits within the Black and White cattle population. A total of 2,006 fecal samples were taken during 2 farm visits in summer and autumn 2015 from 1,166 cows kept in 17 small- and medium-scale organic and conventional German grassland farms. Fecal egg counts were determined for gastrointestinal nematodes (FEC-GIN) and flukes (FEC-FLU), and fecal larvae counts for the bovine lungworm Dictyocaulus viviparus (FLC-DV). The lowest values for gastrointestinal nematode infections were identified for genetic lines adopted to pasture-based production systems, especially selection lines from New Zealand. Heritabilities were low for FEC-GIN (0.05-0.06 ± 0.04) and FLC-DV (0.05 ± 0.04), but moderate for FEC-FLU (0.33 ± 0.06). Almost identical heritabilities were estimated for different endoparasite trait transformations (log-transformation, square root). The genetic correlation between FEC-GIN and FLC-DV was 1.00 ± 0.60, slightly negative between FEC-GIN and FEC-FLU (-0.10 ± 0.27), and close to zero between FLC-DV and FEC-FLU (0.03 ± 0.30). Random regression test-day models on a continuous time scale [days in milk (DIM)] were applied to estimate genetic relationships between endoparasite and longitudinal test-day production traits. Genetic correlations were negative between FEC-GIN and milk yield (MY) until DIM 85, and between FEC-FLU and MY until DIM 215. Genetic correlations between FLC-DV and MY were negative throughout lactation, indicating improved disease resistance for high-productivity cows. Genetic relationships between FEC-GIN and FEC-FLU with milk protein content were negative for all DIM. Apart from the very early and very late lactation stage, genetic correlations between FEC-GIN and milk fat content were negative, whereas they were positive for FEC-FLU. Genetic correlations between FEC-GIN and somatic cell score were positive, indicating similar genetic mechanisms for susceptibility to udder and endoparasite infections. The moderate heritabilities for FEC-FLU suggest inclusion of FEC-FLU into overall organic dairy cattle breeding goals to achieve long-term selection response for disease resistance.

Phenotypic, genetic, and single nucleotide polymorphism marker associations between calf diseases and subsequent performance and disease occurrences of first-lactation German Holstein cows

A total of 31,396 females born from 2010 to 2013 in 43 large-scale Holstein-Friesian herds were phenotyped for calf and cow disease traits using a veterinarian diagnosis key. Calf diseases were general disease status (cGDS), calf diarrhea (cDIA), and calf respiratory disease (cRD) recorded from birth to 2 mo of age. Incidences were 0.48 for cGDS, 0.28 for cRD, and 0.21 for cDIA. Cow disease trait recording focused on the early period directly after calving in first parity, including the interval from 10 d before calving to 200 d in lactation. For cows, at least one entry for the respective disease implied a score = 1 (sick); otherwise, score = 0 (healthy). Corresponding cow diseases were first-lactation general disease status (flGDS), first-lactation diarrhea (flDIA), and first-lactation respiratory disease (flRD). Additional cow disease categories included mastitis (flMAST), claw disorders (flCLAW), female fertility disorders (flFF), and metabolic disorders (flMET). A further cow trait category considered first-lactation test-day production traits from official test-days 1 and 2 after calving. The genotype data set included 41,256 single nucleotide polymorphisms (SNP) from 9,388 females with phenotypes. Linear and generalized linear mixed models with a logit link-function were applied to Gaussian and categorical cow traits, respectively, considering the calf disease as a fixed effect. Most of the calf diseases were not significantly associated with the occurrence of any cow disease. By trend, increasing risks for the occurrence of cow diseases were observed for healthy calves, indicating mechanisms of disease resistance with aging. Also by trend, occurrence of calf diseases was associated with decreasing milk, protein, and fat yields. Univariate linear and threshold animal models were used to estimate heritabilities and breeding values (EBV) for all calf and cow traits. Heritabilities for cGDS and cRD were 0.06 and 0.07 for cDIA. Genetic correlations among all traits were estimated using linear-linear animal models in a series of bivariate runs. The genetic correlation between cDIA and cRD was 0.29. Apart from the genetic correlation between flRD with cGDS (-0.38), EBV correlations and genetic correlations between calf diseases with all cow traits were close to zero. Genome-wide association studies were applied to estimate SNP effects for cRD and cDIA, and for the corresponding traits observed in cows (flRD and flDIA). Different significant SNP markers contributed to cDIA and flDIA, or to cRD and flRD. The average correlation coefficient between cRD and flRD considering SNP effects from all chromosomes was 0.01, and between cDIA and flDIA was -0.04. In conclusion, calf diseases are not appropriate early predictors for cow traits during the early lactation stage in parity 1.

Evaluation of breeding strategies for polledness in dairy cattle using a newly developed simulation framework for quantitative and Mendelian traits

BackgroundIntensified selection of polled individuals has recently gained importance in predominantly horned dairy cattle breeds as an alternative to routine dehorning. The status quo of the current polled breeding pool of genetically-closely related artificial insemination sires with lower breeding values for performance traits raises questions regarding the effects of intensified selection based on this founder pool.MethodsWe developed a stochastic simulation framework that combines the stochastic simulation software QMSim and a self-designed R program named QUALsim that acts as an external extension. Two traits were simulated in a dairy cattle population for 25 generations: one quantitative (QMSim) and one qualitative trait with Mendelian inheritance (i.e. polledness, QUALsim). The assignment scheme for qualitative trait genotypes initiated realistic initial breeding situations regarding allele frequencies, true breeding values for the quantitative trait and genetic relatedness. Intensified selection for polled cattle was achieved using an approach that weights estimated breeding values in the animal best linear unbiased prediction model for the quantitative trait depending on genotypes or phenotypes for the polled trait with a user-defined weighting factor.ResultsSelection response for the polled trait was highest in the selection scheme based on genotypes. Selection based on phenotypes led to significantly lower allele frequencies for polled. The male selection path played a significantly greater role for a fast dissemination of polled alleles compared to female selection strategies. Fixation of the polled allele implies selection based on polled genotypes among males. In comparison to a base breeding scenario that does not take polledness into account, intensive selection for polled substantially reduced genetic gain for this quantitative trait after 25 generations. Reducing selection intensity for polled males while maintaining strong selection intensity among females, simultaneously decreased losses in genetic gain and achieved a final allele frequency of 0.93 for polled.ConclusionsA fast transition to a completely polled population through intensified selection for polled was in contradiction to the preservation of high genetic gain for the quantitative trait. Selection on male polled genotypes with moderate weighting, and selection on female polled phenotypes with high weighting, could be a suitable compromise regarding all important breeding aspects.

Random forest estimation of genomic breeding values for disease susceptibility over different disease incidences and genomic architectures in simulated cow calibration groups

A simulation study was conducted to investigate the performance of random forest (RF) and genomic BLUP (GBLUP) for genomic predictions of binary disease traits based on cow calibration groups. Training and testing sets were modified in different scenarios according to disease incidence, the quantitative-genetic background of the trait (h(2)=0.30 and h(2)=0.10), and the genomic architecture [725 quantitative trait loci (QTL) and 290 QTL, populations with high and low levels of linkage disequilibrium (LD)]. For all scenarios, 10,005 SNP (depicting a low-density 10K SNP chip) and 50,025 SNP (depicting a 50K SNP chip) were evenly spaced along 29 chromosomes. Training and testing sets included 20,000 cows (4,000 sick, 16,000 healthy, disease incidence 20%) from the last 2 generations. Initially, 4,000 sick cows were assigned to the testing set, and the remaining 16,000 healthy cows represented the training set. In the ongoing allocation schemes, the number of sick cows in the training set increased stepwise by moving 10% of the sick animals from the testing set to the training set, and vice versa. The size of the training and testing sets was kept constant. Evaluation criteria for both GBLUP and RF were the correlations between genomic breeding values and true breeding values (prediction accuracy), and the area under the receiving operating characteristic curve (AUROC). Prediction accuracy and AUROC increased for both methods and all scenarios as increasing percentages of sick cows were allocated to the training set. Highest prediction accuracies were observed for disease incidences in training sets that reflected the population disease incidence of 0.20. For this allocation scheme, the largest prediction accuracies of 0.53 for RF and of 0.51 for GBLUP, and the largest AUROC of 0.66 for RF and of 0.64 for GBLUP, were achieved using 50,025 SNP, a heritability of 0.30, and 725 QTL. Heritability decreases from 0.30 to 0.10 and QTL reduction from 725 to 290 were associated with decreasing prediction accuracy and decreasing AUROC for all scenarios. This decrease was more pronounced for RF. Also, the increase of LD had stronger effect on RF results than on GBLUP results. The highest prediction accuracy from the low LD scenario was 0.30 from RF and 0.36 from GBLUP, and increased to 0.39 for both methods in the high LD population. Random forest successfully identified important SNP in close map distance to QTL explaining a high proportion of the phenotypic trait variations.

Prediction accuracies and genetic parameters for test-day traits from genomic and pedigree-based random regression models with or without heat stress interactions

The aim of this study was to compare genetic (co)variance components and prediction accuracies of breeding values from genomic random regression models (gRRM) and pedigree-based random regression models (pRRM), both defined with or without an additional environmental gradient. The used gradient was a temperature-humidity index (THI), considered in statistical models to investigate possible genotype by environment (G×E) interactions. Data included 106,505 test-day records for milk yield (MY) and 106,274 test-day records for somatic cell score (SCS) from 12,331 genotyped Holstein Friesian daughters of 522 genotyped sires. After single nucleotide polymorphism quality control, all genotyped animals had 40,468 single nucleotide polymorphism markers. Test-day traits from recording years 2010 to 2015 were merged with temperature and humidity data from the nearest weather station. In this regard, 58 large-scale farms from the German federal states of Berlin-Brandenburg and Mecklenburg-West Pomerania were allocated to 31 weather stations. For models with a THI gradient, additive genetic variances and heritabilities for MY showed larger fluctuations in dependency of DIM and THI than for SCS. For both traits, heritabilities were smaller from the gRRM compared with estimates from the pRRM. Milk yield showed considerably larger G×E interactions than SCS. In genomic models including a THI gradient, genetic correlations between different DIM × THI combinations ranged from 0.26 to 0.94 for MY. For SCS, the lowest genetic correlation was 0.78, estimated between SCS from the last DIM class and the highest THI class. In addition, for THI × THI combinations, genetic correlations were smaller for MY compared with SCS. A 5-fold cross-validation was used to assess prediction accuracies from 4 different models. The 4 different models were gRRM and pRRM, both modeled with or without G×E interactions. Prediction accuracy was the correlation between breeding values for the prediction data set (i.e., excluding the phenotypic records from this data set) with respective breeding values considering all phenotypic information. Prediction accuracies for sires and for their daughters were largest for the gRRM considering G×E interactions. Such modeling with 2 covariates, DIM and THI, also allowed accurate predictions of genetic values at specific DIM. In comparison with a pRRM, the effect of a gRRM with G×E interactions on gain in prediction accuracies was stronger for daughters than for sires. In conclusion, we found stronger effect of THI alterations on genetic parameter estimates for MY than for SCS. Hence, gRRM considering THI especially contributed to gain in prediction accuracies for MY.

Heritabilities and genetic correlations in the same traits across different strata of herds created according to continuous genomic, genetic, and phenotypic descriptors

The most common approach in dairy cattle to prove genotype by environment interactions is a multiple-trait model application, and considering the same traits in different environments as different traits. We enhanced such concepts by defining continuous phenotypic, genetic, and genomic herd descriptors, and applying random regression sire models. Traits of interest were test-day traits for milk yield, fat percentage, protein percentage, and somatic cell score, considering 267,393 records from 32,707 first-lactation Holstein cows. Cows were born in the years 2010 to 2013, and kept in 52 large-scale herds from 2 federal states of north-east Germany. The average number of genotyped cows per herd (45,613 single nucleotide polymorphism markers per cow) was 133.5 (range: 45 to 415 genotyped cows). Genomic herd descriptors were (1) the level of linkage disequilibrium (r2) within specific chromosome segments, and (2) the average allele frequency for single nucleotide polymorphisms in close distance to a functional mutation. Genetic herd descriptors were the (1) intra-herd inbreeding coefficient, and (2) the percentage of daughters from foreign sires. Phenotypic herd descriptors were (1) herd size, and (2) the herd mean for nonreturn rate. Most correlations among herd descriptors were close to 0, indicating independence of genomic, genetic, and phenotypic characteristics. Heritabilities for milk yield increased with increasing intra-herd linkage disequilibrium, inbreeding, and herd size. Genetic correlations in same traits between adjacent levels of herd descriptors were close to 1, but declined for descriptor levels in greater distance. Genetic correlation declines were more obvious for somatic cell score, compared with test-day traits with larger heritabilities (fat percentage and protein percentage). Also, for milk yield, alterations of herd descriptor levels had an obvious effect on heritabilities and genetic correlations. By trend, multiple trait model results (based on created discrete herd classes) confirmed the random regression estimates. Identified alterations of breeding values in dependency of herd descriptors suggest utilization of specific sires for specific herd structures, offering new possibilities to improve sire selection strategies. Regarding genomic selection designs and genetic gain transfer into commercial herds, cow herds for the utilization in cow training sets should reflect the genomic, genetic, and phenotypic pattern of the broad population.

Genomic prediction by considering genotype × environment interaction using different genomic architectures

Abstract In this study, accuracies of genomic prediction across various scenarios were compared using single- trait and multiple-trait animal models to detect genotype × environment (G × E) interaction based on REML method. The simulated high and low linkage disequilibrium (HLD and LLD) genome consisted of 15,000 and 50,000 SNP chip applications with 300 and 600 QTLs controlling the trait of interest. The simulation was done to create the genetic correlations between the traits in 4 environments and heritabilities of the traits were 0.20, 0.25, 0.30 and 0.35 in environments 1, 2, 3 and 4, respectively. Two strategies were used to predict the accuracy of genomic selection for cows without phenotypes. In the first strategy, phenotypes for cows in three environments were kept as a training set and breeding values for all animals were estimated using three-trait model. In the second one, only 25, 50 or 75% of records in the fourth environment and all the records in the other three environments were used to predict GBV for non-phenotyped cows in the environment 4. For the first strategy, the highest accuracy of 0.695 was realized in scenario HLD with 600 QTL and 50K SNP chip for the fourth environment and the lowest accuracy of 0.495 was obtained in scenario LLD with 600QTL and 15K SNP chips for the first environment. Generally, the accuracy of prediction increased significantly (P<0.05) with increasing the number of markers, heritability and the genetic correlation between the traits, but no significant difference was observed between scenarios with 300 and 600 QTL. In comparison with models without G × E interaction, accuracies of the GBV for all environments increased when using multi-trait models. The results showed that the level of LD, number of animals in training set and genetic correlation across environments play important roles if G × E interaction exists. In conclusion, G × E interaction contributes to understanding variations of quantitative trait and increasing accuracy of genomic prediction. Therefore, the interaction should be taken into account in conducting selection in various environments or across different genotypes.