Multiple-trait Prediction Research Articles

A predictive assessment of genetic correlations between traits in chickens using markers

BackgroundGenomic selection has been successfully implemented in plant and animal breeding programs to shorten generation intervals and accelerate genetic progress per unit of time. In practice, genomic selection can be used to improve several correlated traits simultaneously via multiple-trait prediction, which exploits correlations between traits. However, few studies have explored multiple-trait genomic selection. Our aim was to infer genetic correlations between three traits measured in broiler chickens by exploring kinship matrices based on a linear combination of measures of pedigree and marker-based relatedness. A predictive assessment was used to gauge genetic correlations.MethodsA multivariate genomic best linear unbiased prediction model was designed to combine information from pedigree and genome-wide markers in order to assess genetic correlations between three complex traits in chickens, i.e. body weight at 35 days of age (BW), ultrasound area of breast meat (BM) and hen-house egg production (HHP). A dataset with 1351 birds that were genotyped with the 600 K Affymetrix platform was used. A kinship kernel (K) was constructed as K = λG + (1 − λ)A, where A is the numerator relationship matrix, measuring pedigree-based relatedness, and G is a genomic relationship matrix. The weight (λ) assigned to each source of information varied over the grid λ = (0, 0.2, 0.4, 0.6, 0.8, 1). Maximum likelihood estimates of heritability and genetic correlations were obtained at each λ, and the “optimum” λ was determined using cross-validation.ResultsEstimates of genetic correlations were affected by the weight placed on the source of information used to build K. For example, the genetic correlation between BW–HHP and BM–HHP changed markedly when λ varied from 0 (only A used for measuring relatedness) to 1 (only genomic information used). As λ increased, predictive correlations (correlation between observed phenotypes and predicted breeding values) increased and mean-squared predictive error decreased. However, the improvement in predictive ability was not monotonic, with an optimum found at some 0 < λ < 1, i.e., when both sources of information were used together.ConclusionsOur findings indicate that multiple-trait prediction may benefit from combining pedigree and marker information. Also, it appeared that expected correlated responses to selection computed from standard theory may differ from realized responses. The predictive assessment provided a metric for performance evaluation as well as a means for expressing uncertainty of outcomes of multiple-trait selection.

Genetic relationship between lactation persistency and conformation traits in Polish Holstein-Friesian cow population

The objective of this study was to find genetic relationships between lactation persistency and conformation traits of Polish Holstein-Friesian cows. The data were derived from SYMLEK - the Polish National Database. The analysis used 22 conformation traits and 3 persistency measures of 18 216 first lactation cows. Persistency was defined as milk yield in the second 100 days in milk (DIM) divided by the yield in the first 100 DIM, milk yield in the third 100 DIM divided by the yield in the first 100 DIM, and milk yield at 280 DIM divided by milk yield at 60 DIM. The lactation curve was modelled with fourth-order Legendre polynomials using a multiple-trait prediction method. The multiple-trait REML method was applied for (co)variance component estimation. The linear model for persistency included fixed effects of herd-year-season of calving, age of calving class, and random animal effect; the linear model for type traits included fixed effects of herd-year-season of calving-classifier, age of calving class, lactation stage, and random animal effect. Stature and composite type traits (except type and conformation) were moderately or strongly genetically correlated with each of the three persistency measures. Genetic correlations of lactation persistency with overall feet and leg score and overall udder score were height and positive, whereas the correlations with stature and size were height but negative. All linearly scored traits had little genetic relationship with persistency measures except for rear udder height, udder width, and foot angle. In these three cases the genetic correlations with persistency measures were relatively high, positive, and preferable. The obtained genetic correlations suggest that increased weights of composite traits like size, overall feet and leg score, and overall udder score in the selection index, as well as selection for better rear udder height and foot angle, might cause a favourable correlated response in persistency.

Genetic analysis of twinning rate and milk yield using a threshold‐linear model in Japanese Holsteins

The objective of this study was to estimate genetic parameters and breeding values for the twinning rate of the first three parities (T1, T2 and T3) and 305-day milk yield in first lactation (MY), using a four-trait threshold-linear animal model in Japanese Holsteins. Data contained 1 323 946 cows calving between 1990 and 2007. Twinning was treated as a binary character: 'single' or 'twin or more'. Reported T1, T2 and T3 were 0.70%, 2.87%, and 3.73%, respectively. Individual 305-day milk yield was computed using a multiple trait prediction for cows with at least eight test-day records. (Co)variance components were estimated via Gibbs sampling for randomly sampled subsets. Posterior means of heritabilities for T1, T2 and T3 were 0.11, 0.16 and 0.14, respectively. Genetic correlations between parities were 0.92 or greater. Genetic correlations of MY with twinning rate were not 'significant' (i.e. their 95% highest probability density intervals contained zeros). Multiple births at different parities were considered as the same genetic trait. The average evaluations of T1 (T2) for sires born before 1991 was 0.48% (2.25%) compared with a mean of 0.76% (3.37%) for sires born after 1992. A recent increase in the reported twinning rate reflects the positive genetic trend for sires in Japanese Holsteins.

Predicting lactation yields in dairy buffaloes by interpolation and multiple trait prediction

Milk recording systems for dairy buffaloes in the Philippines suffer from missing or irregular fat and protein test intervals; hence a suitable method to predict lactation yield is needed. The test interval method was compared with methods that incorporate information on standard lactation curves using Wilmink's (Wil) or Wood's (Wood) function. The latter two methods were also evaluated using a multiple trait procedure with covariances between milk, fat and protein yields, MTP-Wil and MTP-Wood, for 305D yield prediction. For this purpose, 1700 lactation records of dairy buffaloes from seven herds located in the north and south of the Philippines were used. All lactations have daily milk yields and most also have monthly fat and protein test day yields. Accuracy of predicting 305D milk, fat and protein yields were evaluated using ≥9 test day yields (≥9TD) of complete lactations, part lactation of varying lengths and lactations with missing tests with varying intervals between tests. The five methods were compared based on mean squared error and correlation between the predicted and true 305D yield. Prediction of 305D milk and protein yields was equally precise among the five methods when ≥9 test day yields were used. In general, MTP-Wil had slightly better precision compared with TIM and Wil methods in extending part lactations to 305D yields for all 3 traits. The use of Wood's function to model the lactation curve resulted in large prediction error when extending yields to 305D from short lactations. The methods, MTP-Wil and MTP-Wood, which utilize information on covariances between traits had slightly better performance and are able to handle missing tests and longer intervals between tests better than other methods. Fat and protein 305D yields can be predicted using a single test with almost similar accuracy using ≥9TD yields provided that complete milk test day information is available. Overall, MTP-Wil performed better than the other methods and is more applicable in situations where missing tests or irregular test intervals are frequently encountered in milk recording system.

Selecting the desirable method for predicting 305‐day lactation yields in Mongolia

In Mongolia, milk production must be improved both quantitatively and qualitatively, and we must find the most desirable method for predicting 305-day milk yields for cows in Mongolia. Therefore, the predicted 305-day milk yield from the test interval method (TIM), multiple-trait prediction (MTP) and the random regression model (RRM) were compared. The daily milk production records during 305 days for cows calving from 1986 to 2007 from the National Livestock Breeding Center in Japan, and the test day records every month for cows calving from 1985 to 2005 from the Hokkaido Dairy Milk Recording and Testing Association were used. Wilmink's model for the average of lactation curve in MTP was adopted. A cubic Legendre polynomials and the parameters of Wilmink's function were used for RRM. The results suggested that RRM was the most desirable method for predicting 305-day milk yields in Mongolia because estimates by RRM were the most accurate when using over four records. Further analyses will be required to examine the properties of RRM when predicting 305-day milk yields using test day records in Mongolia.

多形質予測法による乳期途中経過記録からの３０５日乳生産量の予測

多形質予測法による乳期途中経過記録からの３０５日乳生産量の予測

Agreement of Predicted 305-Day Milk Yields Relative to Actual 305-Day Milk Weight Yields

The objectives of this study were to compare the multiple trait prediction (MTP) model estimate of 305-d lactation yield with the 305-d daily milk yield data from on-farm automated meters and software and to examine the accuracy of electronic identification (ID). Twenty-four-hour milk and component yields are calculated by using milk weights and samples collected 8 to 10 times/yr by Dairy Herd Improvement (DHI) organizations. Daily milk weights were collected from cows on 20 Canadian farms that used parlor milking systems with electronic ID and that were enrolled in a regular DHI program. A total of 10,175 DHI test days from 1,103 cows with complete 305-d lactation yields were entered into the MTP model, and lactation yields were predicted. Test days were grouped into first, second, and third and greater lactations and within each lactation group, days in milk were categorized in 3 stages (5 to 60, 61 to 120, and 120 to 305 d in milk) for a total of 9 classes. Agreement analysis was used to compare the 305-d sum of daily milk to the MTP 305-d lactation yield predictions by using inputs from test days throughout the lactations. Results indicated that the MTP model overestimated lactation yields across all parity groups, ranging from 310 to 1,552kg in parity 1, 640 to 2,000kg in parity 2, and 567 to 1,476kg in parity 3 and greater. A preliminary examination of electronic ID accuracy was conducted on 4 farms. Two electronic ID systems were examined for cow ID accuracy by verifying the ID number appearing in the parlor with the corresponding ear tag number. There were no ID errors on 3 of 4 farms tested and only a very small number of errors (3/80) on the fourth farm, indicating that the electronic ID systems used in milking parlors identify cows accurately.

Genetic Relationships Between Persistency and Reproductive Performance in First-Lactation Canadian Holsteins

The main objective of this study was to estimate genetic relationships between lactation persistency and reproductive performance in first lactation. Relationships with day in milk at peak milk yield and estimated 305-d milk yield were also investigated. The data set contained 33,312 first-lactation Canadian Holsteins with first-parity reproductive, persistency, and productive information. Reproductive performance traits included age at first insemination, nonreturn rate at 56 d after first insemination as a virgin heifer and as a first-lactation cow, calving difficulty at first calving and calving interval between first and second calving. Lactation persistency was defined as the Wilmink b parameter for milk yield and was calculated by fitting lactation curves to test day records using a multiple-trait prediction procedure. An 8-trait genetic analysis was performed using the Variance Component Estimation package (VCE 5) via Gibbs sampling to estimate genetic parameters for all traits. Heritabilities of persistency, day in milk at peak milk yield and estimated 305-d milk yield were 0.18, 0.09 and 0.45, respectively. Heritabilities of reproduction were low and ranged from 0.03 to 0.19. The highest heritability was for age at first insemination. Heifer reproductive traits were lowly genetically correlated, whereas cow reproductive traits were moderately correlated. Heifers younger than average when first inseminated and/or conceived successfully at first insemination tended to have a more persistent first lactation. First lactation was more persistent if heifers had difficulty calving (rg=0.43), or conceived successfully at first insemination in first lactation (rg=0.32) or had a longer interval between first and second calving (rg=0.17). Estimates of genetic correlations of reproductive performance with estimated 305-d milk yield were different in magnitude, but similar in sign to those with persistency (0.02 to 0.51).

Multiple-Trait Prediction of Transmitting Abilities for Herd Life and Estimation of Economic Weights Using Relative Net Income Adjusted for Opportunity Cost

Genetic and phenotypic (co)variances among linear type traits, final score, first lactation milk and fat yield, and 84-mo totals for longevity, relative net income, and relative net income adjusted for opportunity cost of postponed replacement were estimated with a multiple-trait sire model. Data were from 433,116 cows in herds participating in the classification program for conformation traits of the Holstein Association of America. Yield information from all cows in classified herds indicated that classified cows are not a random sample.Heritability of net income adjusted for opportunity cost was higher, .17, than unadjusted net income, .12, but the genetic correlation between the estimates of net income was high, .97. Adjusted net income also had high genetic correlations with first lactation milk yield, .80; fat yield, .60; and dairy form, .48. Heritability of longevity (months in milk to 84 mo) was .06. Adjustment of net income for opportunity cost lowered the genetic correlation with longevity from .84 to .70.Evaluation of lifetime merit using traits measured during first lactation with economic weights developed using adjusted net income was more accurate than indirect prediction of longevity; the approximate reliability of a first-crop AI sire for lifetime merit was .65 compared with .42 for longevity.

Prediction of Transmitting Abilities for Holstein Type Traits

Heritabilities and genetic and phenotypic correlations among 14 linear type traits were estimated from Holstein Association data by multiple trait REML. Data used for parameter estimation were records of 779,391 daughters of 871 sires included in the January 1988 sire evaluation. Each daughter was represented by her appraisal closest to 30 mo of age. Highest heritability was .37 for stature, and lowest was .10 for foot angle. Gains in reliability from using correlated traits in multiple trait prediction were large for some traits (up to 60% for foot angle for cows). Final score variance parameters were estimated from 953,596 records, which were 43% of records included in the national sire evaluation. Sire models that adjusted or did not adjust for merit of mates were compared. Heritability of final score was .27 with adjustment for merit of mates by subtraction of predicted transmitting ability of dam from daughter's record compared with .29 if mate was ignored. Evaluations for type for several popular older sires were reduced moderately by adjustment for merit of mates, but estimated genetic trend increased slightly. An improved genetic grouping procedure that considers group effects as inherited was adapted for use in sire models. Parameter estimates and models presented were implemented by the Holstein Association for computing July 1988 genetic evaluations for linear traits and final score.

Multiple trait prediction for a type of model with heterogeneous genetic and residual covariance structures.

A restricted set of models is defined that allows for heterogeneous genetic and residual covariance structures. Multiple trait models and models with multiple random factors are included. The restriction on the model is that the correlations among genetic effects in different classes are the same. Equivalently, the genetic covariance matrices are assumed to differ between classes due to scaling. This assumption greatly reduces the number of parameters that must be specified and does not adversely affect the computational burden of a mixed model analysis. An application of the model for genetic evaluation of beef cattle is described and illustrated numerically.