Genomic evaluation of Brown Swiss dairy cattle with limited national genotype data and integrated external information.

This study demonstrated the feasibility of a genomic evaluation for the dairy cattle population for which the small national training population can be complemented with foreign information from international evaluations. National test-day milk yield data records for the Slovenian Brown Swiss cattle population were analyzed. Genomic evaluation was carried out using the single-step genomic best linear unbiased prediction method (ssGBLUP), resulting in genomic estimated breeding values (GEBV). The predominantly female group of genotyped animals, representing the national training population in the single-step genomic evaluation, was further augmented with 7,024 genotypes of foreign progeny-tested sires from an international Brown Swiss InterGenomics genomic evaluation (https://interbull.org/ib/whole_cop). Additionally, the estimated breeding values for the altogether 7,246 genotyped domestic and foreign sires from the 2019 sire multiple across-country evaluation (MACE), were added to the ssGBLUP as external pseudophenotypic information. The ssGBLUP method, with integration of MACE information by avoiding double counting, was then performed, resulting in MACE-enhanced GEBV (GEBVM). The methods were empirically validated with forward prediction. The validation group consisted of 315 domestic males and 1,041 domestic females born after 2012. Increase, inflation, and bias of the GEBV(M) reliability (REL) were assessed for the validation group with a focus on females. All individuals in the validation benefited from genomic evaluations using both methods, but the GEBV(M) REL increased most for the youngest selection candidates. Up to 35 points of GEBV REL could be assigned to national genomic information, and up to 17 points of GEBVM REL could additionally be attributed to the integration of foreign sire genomic and MACE information. Results indicated that the combined foreign progeny-tested sire genomic and external MACE information can be used in the single-step genomic evaluation as an equivalent replacement for domestic phenotypic information. Thus, an equal or slightly higher genomic breeding value REL was obtained sooner than the pedigree-based breeding value REL for the female selection candidates. When the abundant foreign progeny-tested sire genomic and MACE information was used to complement available national genomic and phenotypic information in single-step genomic evaluation, the genomic breeding value REL for young-female selection candidates increased approximately 10 points. Use of international information provides the possibility to upgrade small national training populations and obtain satisfying reliability of genomic breeding values even for the youngest female selection candidates, which will help to increase selection efficiency in the future.

Efficient and accurate computation of base generation allele frequencies.

Allele frequencies are used for several aspects of genomic prediction, with the assumption that these are equal to the allele frequency in the base generation of the pedigree. The current standard method, however, calculates allele frequencies from the current genotyped population. We compared the current standard method with BLUP and general least squares (GLS) methods explicitly targeting the base population to determine whether there is a more accurate and still efficient method of calculating allele frequencies that better represents the base generation. A data set based on a typical dairy population was simulated for 325,266 animals; the last 100,078 animals in generations 9 to 12 of the population were genotyped, with 1,670 SNP markers. For the BLUP method, several SNP genotypes were analyzed with a multitrait model by assuming a heritability of 0.99 and no genetic correlation among them. This method was limited by the time required for each BLUP to converge (approximately 6 min per BLUP run of 15 SNP). The GLS method had 2 implementations. The first implementation, using imputation on the fly and multiplication of sparse matrices, was very efficient and required just 49 s and 1.3 GB of random access memory. The second implementation, using a dense full A22-1 matrix, was very inefficient and required more than 1 d of wall clock time and more than 118.2 GB of random access memory. When no selection was considered in the simulations, all methods predicted equally well. When selection was introduced, higher correlations between the estimated allele frequency and known base generation allele frequency were observed for BLUP (0.96 ± 0.01) and GLS (0.97 ± 0.01) compared with the current standard method (0.87 ± 0.01). The GLS method decreased in accuracy when introducing incomplete pedigree, with 25% of sires in the first 5 generations randomly replaced as unknown to erroneously identify founder animals (0.93 ± 0.01) and a further decrease for 8 generations (0.91 ± 0.01). There was no change in accuracy when introducing 5% genotyping errors (0.97 ± 0.01), 5% missing genotypes (0.97 ± 0.01), or both 5% genotyping errors and missing genotypes (0.97 ± 0.01). The GLS method provided the most accurate estimates of base generation allele frequency and was only slightly slower compared with the current method. The efficient implementation of the GLS method, therefore, is very well suited for practical application and is recommended for implementation.

Bayesian single-step genomic evaluations combining local and foreign information in Walloon Holsteins.

Most dairy cattle populations found in different countries around the world are small to medium sized and use many artificial insemination bulls imported from different foreign countries. The Walloon population in the southern part of Belgium is a good example for such a small-scale population. Wallonia has also a very active community of Holstein breeders requesting high level genetic evaluation services. Single-step Genomic BLUP (ssGBLUP) methods allow the simultaneous use of genomic, pedigree and phenotypic information and could reduce potential biases in the estimation of genomically enhanced breeding values (GEBV). Therefore, in the context of implementing a Walloon genomic evaluation system for Holsteins, it was considered as the best option. However, in contrast to multi-step genomic predictions, natively ssGBLUP will only use local phenotypic information and is unable to use directly important other sources of information coming from abroad, for example Multiple Across Country Evaluation (MACE) results as provided by the Interbull Center (Uppsala, Sweden). Therefore, we developed and implemented single-step Genomic Bayesian Prediction (ssGBayes), as an alternative method for the Walloon genomic evaluations. The ssGBayes method approximated the correct system of equations directly using estimated breeding values (EBV) and associated reliabilities (REL) without any explicit deregression step. In the Walloon genomic evaluation, local information refers to Walloon EBV and REL and foreign information refers to MACE EBV and associated REL. Combining simultaneously all available genotypes, pedigree, local and foreign information in an evaluation can be achieved but adding contributions to left-hand and right-hand sides subtracting double-counted contributions. Correct propagation of external information avoiding double counting of contributions due to relationships and due to records can be achieved. This ssGBayes method computed more accurate predictions for all types of animals. For example, for genotyped animals with low Walloon REL (<0.25) without MACE results but sired by genotyped bulls with MACE results, the average increase of REL for the studied traits was 0.38 points of which 0.08 points could be traced to the inclusion of MACE information. For other categories of genotyped animals, the contribution by MACE information was also high. The Walloon genomic evaluation system passed for the first time the Interbull GEBV tests for several traits in July 2013. Recent experiences reported here refer to its use in April 2016 for the routine genomic evaluations of milk production, udder health and type traits. Results showed that the proposed methodology should also be of interest for other, similar, populations.

Assessing the effect of pregnancy stage on milk composition of dairy cows using mid-infrared spectra.

Changes in milk production traits (i.e., milk yield, fat, and protein contents) with the pregnancy stage are well documented. To our knowledge, the effect of pregnancy on the detailed milk composition has not been studied so far. The mid-infrared (MIR) spectrum reflects the detailed composition of a milk sample and is obtained by a nonexhaustive and widely used method for milk analysis. Therefore, this study aimed to investigate the effect of pregnancy on milk MIR spectrum in addition to milk production traits (milk yield, fat, and protein contents). A model including regression on the number of days pregnant was applied on milk production traits (milk yield, fat, and protein contents) and on 212 spectral points from the MIR spectra of 9,757 primiparous Holstein cows from Walloon herds. Effects of pregnancy stage were expressed on a relative scale (effect divided by the squared root of the phenotypic variance); this allowed comparisons between effects on milk traits and on 212 spectral points. Effect of pregnancy stage on production traits were in line with previous studies indicating that the model accounted well for the pregnancy effect. Trends of the relative effect of the pregnancy stage on the 212 spectral points were consistent with known and observed effect on milk traits. The highest effect of the pregnancy was observed in the MIR spectral region from 968 to 1,577 cm-1. For some specific wavenumbers, the effect was higher than for fat and protein contents in the beginning of the pregnancy (from 30 to 90 or 120 d pregnant). In conclusion, the effect of early pregnancy can be observed in the detailed milk composition through the analysis of the MIR spectrum of bovine milk. Further analyses are warranted to explore deeply the use of MIR spectra of bovine milk for breeding and management of dairy cow pregnancy.

National single-step genomic method that integrates multi-national genomic information.

The aim of this paper was to develop a national single-step genomic BLUP that integrates multi-national genomic estimated breeding values (EBV) and associated reliabilities without double counting dependent data contributions from the different evaluations. Simultaneous use of all data, including phenotypes, pedigree, and genotypes, is a condition to obtain unbiased EBV. However, this condition is not always fully met, mainly due to unavailability of foreign raw data for imported animals. In dairy cattle genetic evaluations, this issue is traditionally tackled through the multiple across-country evaluation (MACE) of sires, performed by Interbull Centre (Uppsala, Sweden). Multiple across-country evaluation regresses all the available national information onto a joint pedigree to obtain country-specific rankings of all sires without sharing the raw data. In the context of genomic selection, the issue is handled by exchanging sire genotypes and by using MACE information (i.e., MACE EBV and reliabilities), as a valuable source of "phenotypic" data. Although all the available data are considered, these "multi-national" genomic evaluations use multi-step methods assuming independence of various sources of information, which is not met in all situations. We developed a method that handles this by single-step genomic evaluation that jointly (1) uses national phenotypic, genomic, and pedigree data; (2) uses multi-national genomic information; and (3) avoids double counting dependent data contributions from an animal's own records and relatives' records. The method was demonstrated by integrating multi-national genomic EBV and reliabilities of Brown Swiss sires, included in the InterGenomics consortium at Interbull Centre, into the national evaluation in Slovenia. The results showed that the method could (1) increase reliability of a national (genomic) evaluation; (2) provide consistent ranking of all animals: bulls, cows, and young animals; and (3) increase the size of a genomic training population. These features provide more efficient and transparent selection throughout a breeding program.

Changes throughout lactation in phenotypic and genetic correlations between methane emissions and milk fatty acid contents predicted from milk mid-infrared spectra.

The aim of this study was to estimate phenotypic and genetic correlations between methane production (Mp) and milk fatty acid contents of first-parity Walloon Holstein cows throughout lactation. Calibration equations predicting daily Mp (g/d) and milk fatty acid contents (g/100 dL of milk) were applied on milk mid-infrared spectra related to Walloon milk recording. A total of 241,236 predictions of Mp and milk fatty acids were used. These data were collected between 5 and 305 d in milk in 33,555 first-parity Holstein cows from 626 herds. Pedigree data included 109,975 animals. Bivariate (i.e., Mp and a fatty acid trait) random regression test-day models were developed to estimate phenotypic and genetic parameters of Mp and milk fatty acids. Individual short-chain fatty acids (SCFA) and groups of saturated fatty acids, SCFA, and medium-chain fatty acids showed positive phenotypic and genetic correlations with Mp (from 0.10 to 0.16 and from 0.23 to 0.30 for phenotypic and genetic correlations, respectively), whereas individual long-chain fatty acids (LCFA), and groups of LCFA, monounsaturated fatty acids, and unsaturated fatty acids showed null to positive phenotypic and genetic correlations with Mp (from -0.03 to 0.13 and from -0.02 to 0.32 for phenotypic and genetic correlations, respectively). However, these correlations changed throughout lactation. First, de novo individual and group fatty acids (i.e., C4:0, C6:0, C8:0, C10:0, C12:0, C14:0, SCFA group) showed low phenotypic or genetic correlations (or both) in early lactation and higher at the end of lactation. In contrast, phenotypic and genetic correlations between Mp and C16:0, which could be de novo synthetized or derived from blood lipids, were more stable during lactation. This fatty acid is the most abundant fatty acid of the saturated fatty acid and medium-chain fatty acid groups of which correlations with Mp showed the same pattern across lactation. Phenotypic and genetic correlations between Mp and C17:0 and C18:0 were low in early lactation and increased afterward. Phenotypic and genetic correlations between Mp and C18:1 cis-9 originating from the blood lipids were negative in early lactation and increased afterward to become null from 18 wk until the end of lactation. Correlations between Mp and groups of LCFA, monounsaturated fatty acids, and unsaturated fatty acids showed a similar or intermediate pattern across lactation compared with fatty acids that compose them. Finally, these results indicate that correlations between Mp and milk fatty acids vary following lactation stage of the cow, a fact still often ignored when trying to predict Mp from milk fatty acid profile.

Validation of simultaneous deregression of cow and bull breeding values and derivation of appropriate weights.

Training of genomic prediction in dairy cattle may use deregressed proofs (DRP) as phenotypes. In this case, DRP should be estimated breeding values (EBV) corrected for information of relatives included in the data used for genomic prediction, and adjusted for regression to the mean (i.e., their reliability). Deregression is especially important when combining animals with EBV with low reliability, as commonly the case for cows, and high reliability. The objective of this paper, therefore, was to compare the performance of different deregression procedures for data that include both cow and bull EBV, and to develop and test procedures to obtain the appropriate deregressed weights for the DRP. Considered DRP were EBV: without any adjustment, adjusted for information of parents and regression to the mean, or adjusted for information of all relatives and regression to the mean. Considered deregressed weights were weights of initial EBV: without any adjustment, adjusted for information of parents, or adjusted for information of all relatives. The procedures were compared using simulated data based on an existing pedigree with 1,532 bulls and 13,720 cows that were considered to be included in the data used for genomic prediction. For each cow, 1 to 5 records were simulated. For each bull, an additional 50 to 200 daughters with 1 record each were simulated to generate a source of data that was not used for genomic prediction. The simulated trait had either a heritability of 0.05 or 0.3. The validation involved 3 steps: (1) computation of initial EBV and weights, (2) deregression of those EBV and weights, (3) using deregressed EBV and weights to compute final EBV, (4) comparison of the initial and final EBV and weights. The methods developed to compute appropriate weights for the DRP were either very precise and computationally somewhat demanding for larger data sets, or were less precise but computationally trivial due their approximate nature. Adjusting DRP for all relatives, known as matrix deregression, yields by definition final EBV that are identical to the original EBV. Matrix deregression is therefore preferred over other approaches that only correct for information of parents or not performing any deregression at all. It is important to use appropriate weights for the DRP, properly corrected for information of relatives, especially when individual reliabilities of final EBV are computed based on the prediction error variance of the model.

Integration of external estimated breeding values and associated reliabilities using correlations among traits and effects.

Based on a Bayesian view of linear mixed models, several studies showed the possibilities to integrate estimated breeding values (EBV) and associated reliabilities (REL) provided by genetic evaluations performed outside a given evaluation system into this genetic evaluation. Hereafter, the term "internal" refers to this given genetic evaluation system, and the term "external" refers to all other genetic evaluations performed outside the internal evaluation system. Bayesian approaches integrate external information (i.e., external EBV and associated REL) by altering both the mean and (co)variance of the prior distributions of the additive genetic effects based on the knowledge of this external information. Extensions of the Bayesian approaches to multivariate settings are interesting because external information expressed on other scales, measurement units, or trait definitions, or associated with different heritabilities and genetic parameters than the internal traits, could be integrated into a multivariate genetic evaluation without the need to convert external information to the internal traits. Therefore, the aim of this study was to test the integration of external EBV and associated REL, expressed on a 305-d basis and genetically correlated with a trait of interest, into a multivariate genetic evaluation using a random regression test-day model for the trait of interest. The approach we used was a multivariate Bayesian approach. Results showed that the integration of external information led to a genetic evaluation for the trait of interest for, at least, animals associated with external information, as accurate as a bivariate evaluation including all available phenotypic information. In conclusion, the multivariate Bayesian approaches have the potential to integrate external information correlated with the internal phenotypic traits, and potentially to the different random regressions, into a multivariate genetic evaluation. This allows the use of different scales, heritabilities, variance components, measurement units, or trait definitions for external and internal traits. However, one possible issue for implementing multivariate Bayesian approaches could be the availability or estimation of genetic correlations between external and internal traits.

Genetic analysis of heat stress effects on yield traits, udder health, and fatty acids of Walloon Holstein cows.

Genetic parameters that considered tolerance for heat stress were estimated for production, udder health, and milk composition traits. Data included 202,733 test-day records for milk, fat, and protein yields, fat and protein percentages, somatic cell score (SCS), 10 individual milk fatty acids (FA) predicted by mid-infrared spectrometry, and 7 FA groups. Data were from 34,468 first-lactation Holstein cows in 862 herds in the Walloon region of Belgium and were collected between 2007 and 2010. Test-day records were merged with daily temperature-humidity index (THI) values based on meteorological records from public weather stations. The maximum distance between each farm and its corresponding weather station was 21km. Linear reaction norm models were used to estimate the intercept and slope responses of 23 traits to increasing THI values. Most yield and FA traits had phenotypic and genetic declines as THI increased, whereas SCS, C18:0, C18:1 cis-9, and 4 FA groups (unsaturated FA, monounsaturated FA, polyunsaturated FA, and long-chain FA) increased with THI. Moreover, the latter traits had the largest slope-to-intercept genetic variance ratios, which indicate that they are more affected by heat stress at high THI levels. Estimates of genetic correlations within trait between cold and hot environments were generally high (>0.80). However, lower estimates (<=0.67) were found for SCS, fat yield, and C18:1 cis-9, indicating that animals with the highest genetic merit for those traits in cold environments do not necessarily have the highest genetic merit for the same traits in hot environments. Among all traits, C18:1 cis-9 was the most sensitive to heat stress. As this trait is known to reflect body reserve mobilization, using its variations under hot conditions could be a very affordable milk biomarker of heat stress for dairy cattle expressing the equilibrium between intake and mobilization under warm conditions.

Short communication: Novel method to predict body weight of primiparous dairy cows throughout the lactation.

Body weight (BW) of dairy cows can be estimated using linear conformation traits (calculated BW; CBW), which are generally recorded only once during a lactation. However, predicted BW (PBW) throughout the lactation would be useful, e.g., at milk-recording dates allowing feed-intake prediction for advisory purposes. Therefore, a 2-step approach was developed to obtain PBW for each milk-recording date. In the first step, a random-regression test-day model was used with CBW as observations to predict PBW. The second step consisted in changing means and (co)variances of prior distributions for the additive genetic random effects of the test-day model by using priors derived from results of the first step to predict again PBW. A total of 25,061 CBW from 24,919 primiparous Holstein cows were computed using equations from literature. Using CBW as observations, PBW was then predicted over the whole lactation for 232,436 dates corresponding to 207,375milk-recording dates and 25,061 classification dates. Results showed that using both steps (the 2-step approach) provided more accurate predictions than using only the first step (the one-step approach). Based on the results of this preliminary study, BW of dairy cows could be predicted throughout the lactation using this procedure. These predictions could be useful in milk-recording systems to compute traits of interest (e.g., feed-intake prediction). The developed novel method is also flexible because actual direct measurements of BW can also be used together with CBW, the prediction model being able to accommodate different levels of accuracies of used BW phenotypes.

Short communication: alteration of priors for random effects in Gaussian linear mixed models.

Linear mixed models, for which the prior multivariate normal distributions of random effects are assumed to have a mean equal to 0, are commonly used in animal breeding. However, some statistical analyses (e.g., the consideration of a population under selection into a genomic scheme breeding, multiple-trait predictions of lactation yields, and Bayesian approaches integrating external information into genetic evaluations) need to alter both the mean and (co)variance of the prior distributions and, to our knowledge, most software packages available in the animal breeding community do not permit such alterations. Therefore, the aim of this study was to propose a method to alter both the mean and (co)variance of the prior multivariate normal distributions of random effects of linear mixed models while using currently available software packages. The proposed method was tested on simulated examples with 3 different software packages available in animal breeding. The examples showed the possibility of the proposed method to alter both the mean and (co)variance of the prior distributions with currently available software packages through the use of an extended data file and a user-supplied (co)variance matrix.

Genetic variance in micro-environmental sensitivity for milk and milk quality in Walloon Holstein cattle.

Animals that are robust to environmental changes are desirable in the current dairy industry. Genetic differences in micro-environmental sensitivity can be studied through heterogeneity of residual variance between animals. However, residual variance between animals is usually assumed to be homogeneous in traditional genetic evaluations. The aim of this study was to investigate genetic heterogeneity of residual variance by estimating variance components in residual variance for milk yield, somatic cell score, contents in milk (g/dL) of 2 groups of milk fatty acids (i.e., saturated and unsaturated fatty acids), and the content in milk of one individual fatty acid (i.e., oleic acid, C18:1 cis-9), for first-parity Holstein cows in the Walloon Region of Belgium. A total of 146,027 test-day records from 26,887 cows in 747 herds were available. All cows had at least 3 records and a known sire. These sires had at least 10 cows with records and each herd × test-day had at least 5 cows. The 5 traits were analyzed separately based on fixed lactation curve and random regression test-day models for the mean. Estimation of variance components was performed by running iteratively expectation maximization-REML algorithm by the implementation of double hierarchical generalized linear models. Based on fixed lactation curve test-day mean models, heritability for residual variances ranged between 1.01×10(-3) and 4.17×10(-3) for all traits. The genetic standard deviation in residual variance (i.e., approximately the genetic coefficient of variation of residual variance) ranged between 0.12 and 0.17. Therefore, some genetic variance in micro-environmental sensitivity existed in the Walloon Holstein dairy cattle for the 5 studied traits. The standard deviations due to herd × test-day and permanent environment in residual variance ranged between 0.36 and 0.45 for herd × test-day effect and between 0.55 and 0.97 for permanent environmental effect. Therefore, nongenetic effects also contributed substantially to micro-environmental sensitivity. Addition of random regressions to the mean model did not reduce heterogeneity in residual variance and that genetic heterogeneity of residual variance was not simply an effect of an incomplete mean model.

An integration of external information for foreign stallions into the Belgian genetic evaluation for jumping horses.

The aim of this study was to test the integration of external information, i.e. foreign estimated breeding values (EBV) and the associated reliabilities (REL), for stallions into the Belgian genetic evaluation for jumping horses. The Belgian model is a bivariate repeatability Best Linear Unbiased Prediction animal model only based on Belgian performances, while Belgian breeders import horses from neighbouring countries. Hence, use of external information is needed as prior to achieve more accurate EBV. Pedigree and performance data contained 101382 horses and 712212 performances, respectively. After conversion to the Belgian trait, external information of 98 French and 67 Dutch stallions was integrated into the Belgian evaluation. Resulting Belgian rankings of the foreign stallions were more similar to foreign rankings according to the increase of the rank correlations of at least 12%. REL of their EBV were improved of at least 2% on average. External information was partially to totally equivalent to 4 years of contemporary horses' performances or to all the stallions' own performances. All these results showed the interest to integrate external information into the Belgian evaluation.

Mid-infrared prediction of lactoferrin content in bovine milk: potential indicator of mastitis.

Lactoferrin (LTF) is a milk glycoprotein favorably associated with the immune system of dairy cows. Somatic cell count is often used as an indicator of mastitis in dairy cows, but knowledge on the milk LTF content could aid in mastitis detection. An inexpensive, rapid and robust method to predict milk LTF is required. The aim of this study was to develop an equation to quantify the LTF content in bovine milk using mid-infrared (MIR) spectrometry. LTF was quantified by enzyme-linked immunosorbent assay (ELISA), and all milk samples were analyzed by MIR. After discarding samples with a coefficient of variation between 2 ELISA measurements of more than 5% and the spectral outliers, the calibration set consisted of 2499 samples from Belgium (n = 110), Ireland (n = 1658) and Scotland (n = 731). Six statistical methods were evaluated to develop the LTF equation. The best method yielded a cross-validation coefficient of determination for LTF of 0.71 and a cross-validation standard error of 50.55 mg/l of milk. An external validation was undertaken using an additional dataset containing 274 Walloon samples. The validation coefficient of determination was 0.60. To assess the usefulness of the MIR predicted LTF, four logistic regressions using somatic cell score (SCS) and MIR LTF were developed to predict the presence of mastitis. The dataset used to build the logistic regressions consisted of 275 mastitis records and 13 507 MIR data collected in 18 Walloon herds. The LTF and the interaction SCS × LTF effects were significant (P < 0.001 and P = 0.02, respectively). When only the predicted LTF was included in the model, the prediction of the presence of mastitis was not accurate despite a moderate correlation between SCS and LTF (r = 0.54). The specificity and the sensitivity of models were assessed using Walloon data (i.e. internal validation) and data collected from a research herd at the University of Wisconsin - Madison (i.e. 5886 Wisconsin MIR records related to 93 mastistis events - external validation). Model specificity was better when LTF was included in the regression along with SCS when compared with SCS alone. Correct classification of non-mastitis records was 95.44% and 92.05% from Wisconsin and Walloon data, respectively. The same conclusion was formulated from the Hosmer and Lemeshow test. In conclusion, this study confirms the possibility to quantify an LTF indicator from milk MIR spectra. It suggests the usefulness of this indicator associated to SCS to detect the presence of mastitis. Moreover, the knowledge of milk LTF could also improve the milk nutritional quality.

Comparison and improvements of different Bayesian procedures to integrate external information into genetic evaluations.

The aim of this research was to compare different Bayesian procedures to integrate information from outside a given evaluation system, hereafter called external information, and in this context estimated breeding values (EBV), into this genetic evaluation, hereafter called internal evaluation, and to improve the Bayesian procedures to assess their potential to combine information from diverse sources. The 2 improvements were based on approximations of prior mean and variance. The first version of modified Bayesian evaluation considers all animals as animals associated with external information. For animals that have no external information (i.e., internal animals), external information is predicted from available external information. Thereby, propagation of this external information through the whole pedigree is allowed. Furthermore, the prediction of external information for internal animals allows large simplifications of the computational burden during setup and solving of mixed model equations. However, double counting among external animals (i.e., animals associated with available external information) is not avoided. Double counting concerns multiple considerations of contributions due to relationships by integration of external EBV for related external animals and is taken into account by the second version of modified Bayesian evaluation. This version includes the estimation of double counting before integration of external information. To test the improvements, 2 dairy cattle populations were simulated across 5 generations. Milk production for the first lactation for each female was simulated in both populations. Internal females were randomly mated with internal males and 50 external males. Results for 100 replicates showed that rank correlations among Bayesian EBV and EBV based on the joint use of external and internal data were very close to 1 for both external and internal animals if all internal and external animals were associated with external information. The respective correlations for the internal evaluation were equal to 0.54 and 0.95 if no external information was integrated. If double counting was avoided, mean squared error, expressed as a percentage of the internal mean squared error, was close to zero for both external and internal animals. However, computational demands increased when double counting was avoided. Finally, the improved Bayesian procedures have the potential to be applied for integrating external EBV, or even genomic breeding values following some additional assumptions, into routine genetic evaluations to evaluate animals more reliably.