Bias In Genetic Evaluations Research Articles

A note on dairy cow behavior when measuring enteric methane emissions with the GreenFeed emission monitoring system in tiestalls

Changes in the environment or novel procedures can result in altered cow behavior during data collection; training is often recommended to ensure accurate data is being recorded. Currently, little is known regarding the habituation of dairy cows during methane emission testing with the GreenFeed emission monitoring system (C-Lock Inc., Rapid City, SD), or how behavior relates to enteric methane emission measurements. Methane emissions were estimated from a total of 202 Holstein dairy cows (120–150 d in milk) housed in tiestalls as part of a larger project. Cows were tested on d 0 (training day) and d 1–5 (test day) for approximately 10 min, during which behavior was recorded by a trained observer. While cows spent more time with their head outside of the machine on the training day (d 0) than during the test days (d 1–5), the opposite pattern was observed for the number of leg movements. No differences in estimated methane production were found over the different days, though it was negatively correlated with both behaviors. These results highlight the importance of habituation of dairy cows to the GreenFeed system for methane measurements to minimize changes to cow behavior under tiestall conditions, whereas the methane emissions themselves are less affected. However, further research is needed to determine the impact of cow behavior on the reliability and repeatability of methane emission measurements as it may introduce bias in genetic evaluations for methane efficiency.

Parameter estimation and assessment of bias in genetic evaluation of carcass traits in Hanwoo cattle using real and simulated data.

Most carcass and meat quality traits are moderate to highly heritable, indicating that they can be improved through selection. Genetic evaluation for these types of traits is performed using performance data obtained from commercial and progeny testing evaluation. The performance data from commercial farms are available in large volume, however, some drawbacks have been observed. The drawback of the commercial data is mainly due to sorting of animals based on live weight prior to slaughter, and this could lead to bias in the genetic evaluation of later measured traits such as carcass traits. The current study has two components to address the drawback of the commercial data. The first component of the study aimed to estimate genetic parameters for carcass and meat quality traits in Korean Hanwoo cattle using a large sample size of industry-based carcass performance records (n = 469,002). The second component of the study aimed to describe the impact of sorting animals into different contemporary groups based on an early measured trait and then examine the effect on the genetic evaluation of subsequently measured traits. To demonstrate our objectives, we used real performance data to estimate genetic parameters and simulated data was used to assess the bias in genetic evaluation. The results of our first study showed that commercial data obtained from slaughterhouses is a potential source of carcass performance data and useful for genetic evaluation of carcass traits to improve beef cattle performance. However, we observed some harvesting effect which leads to bias in genetic evaluation of carcass traits. This is mainly due to the selection of animal based on their body weight before arrival to slaughterhouse. Overall, the non-random allocation of animals into a contemporary group leads to a biased estimated breeding value in genetic evaluation, the severity of which increases when the evaluation traits are highly correlated.

Potential effects of hormonal synchronized breeding on genetic evaluations of fertility traits in dairy cattle: A simulation study

About 30% of producers use hormone protocols to synchronize ovulation and perform timed artificial insemination (AI) in Canada. Days from calving to first service (CTFS) and first service to conception (FSTC) become masked phenotypes leading to biased genetic evaluations of cows for these fertility traits. The objectives of this study were to (1) demonstrate and quantify the potential amount of bias in genetic evaluations, and (2) find a procedure that could remove the bias. Simulation was used for both objectives. The proposed solution was to identify cows that have been treated by hormone protocols, make their CTFS and FSTC missing, and perform a multiple trait analysis including traits that have high genetic correlations with CTFS and FSTC, and which are not affected by the hormone protocols themselves. A total of 12 scenarios (S1-S12) were tested, changing the percentage of herds and cows that were randomly selected to be under timed AI. Cows that were given hormone protocols had CTFS of 86 d and FSTC of 0, which were used in genetic evaluation. Four criteria were used to indirectly measure the presence of bias: (1) the correlation between true (TBV) and estimated (EBV) breeding values (accuracy); (2) the differences in the mean EBV of top 25, 50, and 75 sires; (3) changes in correlation between TBV and EBV rankings; and (4) the changes in mean EBV over the simulated generations. All criteria changed unfavorably and proportionally to the increased use of timed AI. The accuracy within each class of animals (cows, dams, or sires) decreased proportionally with increased use of timed AI, varying from 0.32 (S12) to 0.52 (S1) for bull EBV for CTFS. The average EBV of the top sires (best 25, 50, 75, or 100 sires) approached population average EBV values when increasing the number of treated animals. The sire rank correlation between EBV and TBV within simulated scenarios was smaller for scenarios with more synchronized animals, going from 0.38 (S12) to 0.67 (S1). The long-term use of hormonal synchronized cows clearly decreased the mean EBV over generations in the population for CTFS and FSTC. The inclusion of genetically correlated traits in a multiple trait model was effective in removing the bias due to the presence of hormonal synchronized cows. However, given the constraints within the simulation, it is important that further investigation with real data is conducted to determine the true effect of including timed AI records within genetic evaluations of fertility traits in dairy cattle.

Behavior of the Linear Regression method to estimate bias and accuracies with correct and incorrect genetic evaluation models

Bias in genetic evaluations has been a constant concern in animal genetics. The interest in this topic has increased in the last years, since many studies have detected overestimation (bias) in estimated breeding values (EBV). Detecting the existence of bias, and the realized accuracy of predictions, is therefore of importance, yet this is difficult when studying small data sets or breeds. In this study, we tested by simulation the recently presented method Linear Regression (LR) for estimation of bias, slope, and accuracy of pedigree EBV. The LR method computes statistics by comparing EBV from a data set containing old, partial information with EBV from a data set containing all information (old and new, a whole data set) for the same individuals. The method proposes an estimator for bias (Δpˆ), an estimator of slope (bpˆ), and 3 estimators related to accuracies: the ratio between accuracies [Formula: see text] the reliability of the partial data set (accp2ˆ), and the ratio of reliabilities (ρp,w2ˆ). We simulated a dairy scheme for low (0.10) and moderate (0.30) heritabilities. In both cases, we checked the behavior of the estimators for 3 scenarios: (1) when the evaluation model is the same as the model used to simulate the data; (2) when the evaluation model uses an incorrect heritability; and (3) when the data includes an environmental trend. For scenarios in which the evaluation model was correct, the LR method was capable of correctly estimating bias, slope, and accuracies, with better performance for higher heritability [i.e., corr(bp,bpˆ) was 0.45 for h2 = 0.10 and 0.59 for h2 = 0.30]. In cases of the use of incorrect heritabilities in the evaluation model, the bias was correctly estimated in direction but not in magnitude. In the same way, the magnitudes of bias and of slope were underestimated in scenarios with environmental trends in data, except for cases in which contemporary groups were random and greatly shrunken. In general, accuracies were well estimated in all scenarios. The LR method is capable of checking bias and accuracy in all cases, if the evaluation model is reasonably correct or robust, and its estimations are more precise with more information (e.g., high heritability). If the model uses an incorrect heritability or a hidden trend exists in the data, it is still possible to estimate the direction and existence of bias and slope but not always their magnitudes.

Correcting within-family pre-selection in genetic evaluation of growth—A simulation study on rainbow trout

article i nfo Genetic improvement programs for some fish species apply a two-stage selection scheme in which phenotypic selection is first practiced within families based on early body size. Pre-selection improves genetic gain in the breeding objective traits correlated with the pre-selection criteria, and it can also reduce management costs of a program. In this study, stochastic simulation of a rainbow trout, Oncorhynchus mykiss, breeding scheme with 150 full-sib families (2:2 mating design) was utilized to explore how within-family pre-selection and different information on the culled fish affect variance estimates and accuracy of genetic evaluation in grow-out body weights. The bias in genetic parameters and breeding values (EBVs) was quantified for fingerling weight at id- tagging (BW1), used as the criterion for pre-selection, and for two harvest weights recorded at the freshwater nucleus (BW2) and sea test station (BW2sea) in a split-family design. At tagging, fish from each full-sib family were either randomly sampled (R) or pre-selected, and the BW1 records of the culled fish were either individu- ally measured (S + IND), augmented with the replicated family-specific averages of the culled fish (S + AVER), or were treated as missing (S-MIS). These four alternative data treatments were compared using a fixed initial family size of 100 individuals before tagging and two different pre-selection intensities (40% or 21% of fish per family selected). Variance estimates in R and S + IND did not diverge from the simulated a priori values in either of the selection intensities studied. The strategy S + AVER resulted in unbiased genetic variance estimates but decreased the residual variance, especially for BW1 and BW2. The accuracy of EBVs was, nevertheless, equally high for R, S + IND and S + AVER, and these values did not essentially differ between the two selection intensi- ties. For S-MIS, the variance estimates were strongly biased in each trait, and the EBV accuracies were, on aver- age, lower than in the other three treatments. Common environment variances were consistently overestimated and residual variances underestimated, whereas genetic variances were biased in both directions depending on the trait and pre-selection intensity. Further, for S-MIS, frequent convergence problems occurred in the estima- tionofvariance components.For fish breedingschemesapplying within-familypre-selection,dataaugmentation for culled fish by their average values of BW1 will sufficiently control for selection bias in genetic evaluation of growth. For accurate estimation of variance components either random samples from families or individual re- cords from all culled fish are preferable.

Estimation of genetic parameters for preweaning growth traits of Brahman cattle in Southeastern Mexico

The genetic parameters for Brahman cattle under the tropical conditions of Mexico are scarce. Therefore, heritabilities, additive direct and maternal correlations, and genetic correlations for birth weight (BW) and 205days adjusted weaning weight (WW205) were estimated in four Brahman cattle herds in Yucatan, Mexico. Parameters were estimated fitting a bivariate animal model, with 4,531 animals in the relationship matrix, of which 2,905 had BW and 2,264 had WW205. The number of sires and dams identified for both traits were 122 and 962, respectively. Direct heritability estimates for BW and WW205 were 0.41 ± 0.09 and 0.43 ± 0.09, and maternal heritabilities were 0.15 ± 0.07 and 0.38 ± 0.08, respectively. Genetic correlations between direct additive and maternal genetic effects for BW and WW205 were -0.41 ± 0.22 and -0.50 ± 0.15, respectively. The direct genetic, maternal, and phenotypic correlations between BW and WW205 were 0.77 ± 0.09, 0.61 ± 0.18, and 0.35, respectively. The moderate to high genetic parameter estimates suggest that genetic improvement by selection is possible for those traits. The maternal effects and their correlation with direct effects should be taken into account to reduce bias in genetic evaluations.

Estimation of genetic parameters and response to selection for a continuous trait subject to culling before testing

The consequences of assuming a zero environmental covariance between a binary trait 'test-status' and a continuous trait on the estimates of genetic parameters by restricted maximum likelihood and Gibbs sampling and on response from genetic selection when the true environmental covariance deviates from zero were studied. Data were simulated for two traits (one that culling was based on and a continuous trait) using the following true parameters, on the underlying scale: h² = 0.4; r(A) = 0.5; r(E) = 0.5, 0.0 or -0.5. The selection on the continuous trait was applied to five subsequent generations where 25 sires and 500 dams produced 1500 offspring per generation. Mass selection was applied in the analysis of the effect on estimation of genetic parameters. Estimated breeding values were used in the study of the effect of genetic selection on response and accuracy. The culling frequency was either 0.5 or 0.8 within each generation. Each of 10 replicates included 7500 records on 'test-status' and 9600 animals in the pedigree file. Results from bivariate analysis showed unbiased estimates of variance components and genetic parameters when true r(E) = 0.0. For r(E) = 0.5, variance components (13-19% bias) and especially (50-80%) were underestimated for the continuous trait, while heritability estimates were unbiased. For r(E) = -0.5, heritability estimates of test-status were unbiased, while genetic variance and heritability of the continuous trait together with were overestimated (25-50%). The bias was larger for the higher culling frequency. Culling always reduced genetic progress from selection, but the genetic progress was found to be robust to the use of wrong parameter values of the true environmental correlation between test-status and the continuous trait. Use of a bivariate linear-linear model reduced bias in genetic evaluations, when data were subject to culling.

Bias in genetic evaluation using random regression test-day model

The effect of an upgraded Finnish evaluation model on bias in estimated breeding values for protein yield was investigated. Evaluations based on repeatability animal model and on random regression test-day model without and with heterogeneous variance adjustment were compared. Comparisons were based on the average difference between pedigree indices and the future estimated breeding values, based on own or on daughter performance records. This was defined as empirical bias. The pedigree indices were computed from reduced data sets where four years of the most recent data were excluded. Results showed an upward bias in the protein yield pedigree indices for Ayrshire young sires of 2.2 kg, 2.5 kg and 1.8 kg from the repeatability animal model, random regression test-day model and random regression test-day model with heterogeneous variance adjustment, respectively. Pedigree indices for daughters of young sires were upward biased, whereas pedigree indices for daughters of proven sires were slightly underestimated when heterogeneous variance was not accounted. Inclusion of test-day yields from the fourth lactation onwards increased the bias. Moving from repeatability animal model to random regression test-day model did not reduce the bias, whereas adjustment of heterogeneous variance reduced bias.;

Adjusting for heterogeneity of variance for carcass traits affects single and multiple trait selections in genetic evaluation of Japanese Black cattle

ABSTRACTHeterogeneity of variance among subclasses of an effect is a potential source of bias in genetic evaluation. Degrees of the heterogeneity of variance among farm‐market‐year‐sex (FMYS) subclasses for carcass weight, beef marbling standard number, rib‐eye area, rib thickness and subcutaneous fat thickness were investigated in Japanese Black cattle. Consequences of adjusting for the heterogeneity on the predicted breeding values (PBVs) or on the genetic indexes derived from the PBVs of the five carcass traits were assessed. A total of 57 461 records were collected between 1997 and 2002 from steers and heifers fattened at farms across Japan. These records were grouped into 1591 FMYS subclasses. Bartlett's test showed that the degree of the heterogeneity of variance among the FMYS subclasses was sizeable in all traits (P < 0.0001). By applying a two‐step adjustment procedure it was possible to reduce the standard deviation, the coefficient of variation and the Gini coefficient of the phenotypic variances by 67.5% to 75.0% in the different traits. The applied adjustment caused a substantial re‐ranking of elite dams in the PBV for each trait as well as in the genetic index. This study provided evidence that the applied adjustment reduces the bias in the PBVs due to heterogeneous variances and increases the accuracy of bull‐dam selection.

Estimation of accuracy and bias in genetic evaluations with genetic groups using sampling1

Accuracy and bias of EBV are important measures of the quality of genetic evaluations. A sampling method that accounts for the uncertainty in the estimation of genetic group effects was used to calculate accuracy and bias of estimated effects. The method works by repeatedly simulating phenotypes for multiple traits for a defined data and pedigree structure. These simulated values are analyzed using BLUP with genetic groups in the relationship matrix. Accuracies and biases are then calculated as correlations among and differences between true and estimated values across all replicates, respectively. The method was applied to the Irish beef production data set for 15 traits and with 15 genetic groups to account for differences in breed means. Accuracy and bias of estimated genetic group effects, estimated comparisons between genetic group effects, EBV within genetic group, and EBV across genetic group were calculated. Small biases were detected for most estimated genetic group effects and most estimated comparisons between genetic group effects. Most of these were not important relative to the phenotypic SD of the traits involved. For example, a bias of 0.78% of the phenotypic SD was detected for carcass conformation in Aberdeen Angus. However, one trait, calf quality, which had few performance records in the data set, displayed larger bias, ranging from -10.31 to 5.85% of the phenotypic SD across the different estimated genetic group effects. Large differences were observed in the accuracies of genetic group effects, ranging from 0.02 for feed intake in Holstein, which had no data recorded, to >0.97 for carcass conformation, a trait with large amounts of data recorded in the different genetic groups. Large differences were also observed in the accuracies of the comparisons among genetic group effects. The accuracies of the EBV within genetic group and EBV across genetic group were sometimes different; for example, carcass conformation in Belgian Blue had an average accuracy within genetic group of 0.69 compared with an average accuracy across genetic group of 0.89. This suggests that the accuracy of genetic groups should be taken into account when publishing EBV across genetic groups.

Effectiveness of adjusting for heterogeneity of variance in genetic evaluation of Japanese Black cattle1

Heterogeneity of variance among subclasses of an effect is a potential source of bias in genetic evaluation. The objectives of this study were to quantify the heterogeneity of variance in carcass weight in Japanese Black cattle, to develop an adjustment method to account for the heterogeneity, and to evaluate the effectiveness of the method. A total of 96,950 records were collected from steers and heifers slaughtered from 1997 to 2005. These records were grouped into 2,767 farm-market-year-sex subclasses. Fourteen log-linear models for the variances were set up to estimate the heterogeneous phenotypic variances within subclasses. Schwarz's Bayesian information criterion was used for model selection. The preadjustment of records to a baseline variance was based on maximum likelihood estimates obtained from the selected model. As a result of adjustment, the SD, the CV, and the Gini coefficient for the phenotypic variance decreased by 68.6, 69.8, and 70.1%, respectively. When the top 5% of sires and top 1% of dams were selected, Spearman's rank correlation coefficients between the adjusted and unadjusted data were 0.95 for the selected sires and 0.78 for the selected dams. The effectiveness of the adjustment was evaluated in terms of the ability to predict breeding values, using the results of the successive genetic evaluations. Mean squared error between the parent averages and actual predicted values of the genetic merit for the sires whose progeny had a carcass record only from 2003 to 2005 was significantly reduced by the adjustment (P < 0.05). The results suggest that the genetic evaluation becomes more accurate by adjusting the data using the procedure developed in this study.

A proposal of criteria of robustness analysis in genetic evaluation

An assumption underlying the use of BLUP in the context of genetic evaluation is that the expectations of the true breeding values are null. The aim of this paper is to address the robustness of genetic evaluation, in terms of the prediction of genetic trend and selection responses, when this assumption is violated. Animals that are likely to be genetically different are the animals for whom the accuracy of comparison is low, because they are bred in different environments which are not highly connected. In such a case, the environment and genetic effects are partially confounded and the genetic differences between animals in different environments are underestimated. An analytical criterion of robustness is proposed: the lowest coefficient of determination (CD 1) of a comparison between animals included in the evaluation. Its relationship with bias in genetic evaluation is explained. A numerical application considers two kinds of planned progeny test designs for French AI sire evaluation: the “reference sire design” and the “repeater sire design”. Using Monte-Carlo simulations, prediction of genetic trend is shown to be in close relation with CD 1. CD 1 appears to be a good indicator of the robustness of the design and a measure of the part of genetic trend that can be predicted. According to the number of progeny recorded per sire, the repeater sire design only accounted for 3 to 12% of the genetic trend whereas the reference sire design accounted for 22 to 59% for a trait of h 2=0.40. However, in terms of the selection response, the two designs are equivalent when genetic trend was below 0.5 σ a, which is always the case in animal breeding programs.

Effects of herd origin, AI stud and sire identification on genetic evaluation of Holstein Friesian bulls

The purpose of this study was to estimate the effects of herd origin of bull, AI stud and sire identification number (ID) on official estimated breeding values (EBV) for production traits of Holstein Friesian proven bulls. The data included 1,005 Italian Holstein-Friesian bulls, sons of 76 sires, born in 100 herds and progeny tested by 10 AI studs. Bulls were required to have date of first proof between September 1992 and September 1997, to be born in a herd with at least one other bull and to have sire and dam with official EBV when bull was selected for progeny testing. Records of sires with only one son were also discarded. The dependent variable analyzed was the official genetic evaluation for a “quantity and quality of milk” index (ILQ). The linear model to predict breeding values of bulls included the fixed class effects of herd origin of bull, AI testing organization, birth year of bull, and estimated breeding values of sire and dam, both as linear covariates. The R2 of the model was 45% and a significant effect was found for genetic merit of sire (P < 0.001) and dam (P < 0.014), for herd origin of bull (P < 0.01) and for birth year of bull (P < 0.001). The effect of AI testing organization was not significant. The range of herd origin effect was 872 kg of ILQ. However, in this study, the causes of this result were not clear; it may be due to numerous factors, one of which may be preferential treatment on dams of bulls. Analyses of residuals on breeding value of proven bulls for ILQ showed a non significant effect of sire ID, after adjusting for parent average, herd origin effect and birth year effect. Although the presence of bias in genetic evaluation of dairy bulls is not evident, further research is recommended firstly to understand the reasons of the significant herd origin effect, secondly to monitor and guarantee the greatest accuracy and reliability of genetic evaluation procedures.

Bias in Genetic Evaluations by Records of Cows Treated with Bovine Somatotropin

Records from Dairy Records Management Systems in Raleigh were used to estimate effects of bovine somatotropin (bST) treatment and to predict breeding values for milk production traits. The data comprised 5245 test-day records of bST-treated cows and 126,223 test-day records of untreated cows in first lactation for milk, fat, and protein yields. Fixed effects of bST treatment were estimated from test-day animal models with herd-test-date as another fixed factor. Percentage increases due to bST treatment ranged from 7 to 8%for test-day milk, fat, and protein yields. Random regression coefficients for additive genetic and permanent environmental effects were included in the model. To assess the potential for bias in genetic evaluations when some and not all cows are treated with bST, breeding values predicted by the test-day model with and without effects of bST treatment were compared for cows and sires. Correlations between breeding values predicted from models with and without effects of bST treatment were 0.99. However, relatively large bias was found for individual animals. This result suggests that bias in genetic evaluation caused by ignoring bST treatment may be significant.

Potential Bias in Genetic Evaluations from Differences in Variation Within Herds

Evidence is reviewed that indicates substantial differences among herds in variation for production traits in dairy cattle. Examples are given to illustrate the overevaluation and selection of higher proportions of individuals from more variable herds, which reduces response to selection if greater variability is not due in part to greater additive genetic variance. Differences in heritability according to herd mean and variance are examined. Potential bias is substantial in genetic evaluations of cows and may increase over generations where, if the model is realistic, evaluations of dams are used in evaluations of daughters. Possible methods of adjustment are discussed.