Cet article présente l’histoire et le bilan de la sélection génomique bovine en France, avec ses étapes de démarrage dans les grandes races laitières, son extension progressive à toutes les races, le développement du génotypage permis par la diminution de son coût, l’intégration de nouveaux caractères et la redéfinition des objectifs de sélection avec davantage d’importance donnée aux caractères fonctionnels. Il présente les évolutions méthodologiques réalisées au cours du temps, concernant le génotypage, les populations de référence et les méthodes d’évaluation génomique. Les résultats acquis sont présentés, en termes d’évolution génétique sur les différents caractères – toujours favorable – d’évolution de la consanguinité – très variable entre races – et de maîtrise des anomalies génétiques. Une forte synergie a été construite entre recherche et sélection : toutes les évolutions ont été suscitées par la recherche, mais d’un autre côté, les données nouvelles engendrées par la sélection ont nourri la recherche. Enfin, l’article décrit succinctement les fortes évolutions organisationnelles qui ont été pour partie induites par la sélection génomique.

BACKGROUND:Dairy cattle breeds are populations of limited effective size, subject to recurrent outbreaks of recessive defects that are commonly studied using positional cloning. However, this strategy, based on the observation of animals with characteristic features, may overlook a number of conditions, such as immune or metabolic genetic disorders, which may be confused with pathologies of environmental etiology. RESULTS:We present a data mining framework specifically designed to detect recessive defects in livestock that have been previously missed due to a lack of specific signs, incomplete penetrance, or incomplete linkage disequilibrium. This approach leverages the massive data generated by genomic selection. Its basic principle is to compare the observed and expected numbers of homozygotes for sliding haplotypes in animals with different life histories. Within three cattle breeds, we report 33 new loci responsible for increased risk of juvenile mortality and present a series of validations based on large-scale genotyping, clinical examination, and functional studies for candidate variants affecting the NOA1, RFC5, and ITGB7 genes. In particular, we describe disorders associated with NOA1 and RFC5 mutations for the first time in vertebrates. CONCLUSIONS:The discovery of these many new defects will help to characterize the genetic basis of inbreeding depression, while their management will improve animal welfare and reduce losses to the industry.

After calving, high-yielding dairy cows mobilize body reserves for energy, sometimes to the detriment of health and fertility. This study aimed to estimate the genetic correlation between body weight loss until nadir and daily milk production (MY24) in first- (L1) and second-lactation (L2) Holstein cows. The data set included 859,020 MY24 records and 570,651 daily raw body weight (BWr) phenotypes from 3,989 L1 cows, and 665,361 MY24 records and 449,449 BWr phenotypes from 3,060 L2 cows, recorded on 36 French commercial farms equipped with milking robots that included an automatic weighing platform. To avoid any bias due to change in digestive content, BWr was adjusted for variations in feed intake, estimated from milk production and BWr. Adjusted body weight was denoted BW. The genetic parameters of BW and MY24 in L1 and L2 cows were estimated using a 4-trait random regression model. In this model, the random effects were fitted by second-order Legendre polynomials on a weekly basis from wk 1 to 44. Nadir of BW was found to be earlier than reported in the literature, at 29 d in milk, and BW loss from calving to nadir was also lower than generally assumed, close to 29 kg. To estimate genetic correlations between body weight loss and production, we defined BWL5 as the loss of weight between wk 1 and 5 after calving. Genetic correlations between BWL5 and MY24 ranged from -0.26 to 0.05 in L1 and from -0.11 to 0.10 in L2, according to days in milk. These moderate to low values suggest that it may be possible to select for milk production without increasing early body mobilization.

Body mobilization of high-yielding dairy cows is intense after calving and can generate health and fertility troubles. This study aimed to estimate the genetic correlation between body weight loss during the five first weeks of lactation (BWL) and daily milk production (MY24a) in first (L1) and second (L2) lactations Holstein cows. The dataset included 859,020 MY24a and 570,651 body weight (BW) daily phenotypes from 3,989 L1 cows, and 665,361 MY24a and 449,449 BW daily phenotypes from 3,060 L2 cows, recorded in 36 French commercial farms equipped with milking robots including an automatic weighing platform. The genetic parameters of BW and MY24a in L1 and L2 were estimated using a 4-trait random regression model. Genetic correlations between BWL and MY24a ranged from -0.26 to 0.05 in L1, and varied from -0.11 to 0.10 in L2, suggesting that selecting to reduce early body mobilization while maintaining milk production is possible.

Les aptitudes du lait à la transformation en fromage sont étroitement liées à sa composition. Ces caractères, difficiles à mesurer directement, ont été prédits à partir des spectres dans le moyen infrarouge (MIR) du lait en race Montbéliarde (projet From’MIR). Cet article rassemble les résultats de l’analyse du déterminisme génétique des aptitudes fromagères et de la composition fine du lait, prédites à partir de six millions de spectres MIR de 400 000 vaches. Ces caractères sont modérément à fortement héritables et les corrélations génétiques entre caractères fromagers (rendements et coagulation) et avec la composition du lait (protéines, acides gras et minéraux) sont élevées et favorables. Des analyses d’association (GWAS) et de réseaux de gènes, réalisées à partir des génotypes imputés pour l’ensemble des variations du génome de 20 000 vaches, permettent d’identifier des gènes et des variants candidats ainsi qu’un réseau de 736 gènes impliqués dans des voies métaboliques et des gènes régulateurs fonctionnellement liés à la composition du lait. Enfin, l’estimation de la précision d’une évaluation génomique montre qu’un modèle de type contrôles élémentaires, incluant les variants détectés par les GWAS et présumés causaux, permet de prédire des valeurs génomiques précises. Nous avons par ailleurs simulé une sélection incluant les aptitudes fromagères qui montre les possibilités de sélectionner efficacement les vaches pour qu’elles produisent un lait plus « fromageable », avec un impact limité sur le gain génétique des caractères actuellement sélectionnés. Ces résultats ont conduit à la mise en place d’un prototype d’évaluation génomique en race Montbéliarde dans la zone AOP Comté en 2019

In a previous study, we identified candidate causative variants located in 24 functional candidate genes for milk protein and fatty acid composition in Montbéliarde, Normande, and Holstein cows. We designed these variants on the custom part of the EuroG10K BeadChip (Illumina Inc., San Diego, CA), which is routinely used for genomic selection analyses in French dairy cattle. To validate the effects of these candidate variants on milk composition and to estimate their effects on cheesemaking properties, a genome-wide association study was performed on milk protein, fatty acid and mineral composition, as well as on 9 cheesemaking traits (3 laboratory cheese yields, 5 coagulation traits, and milk pH). All the traits were predicted from midinfrared spectra in the Montbéliarde cow population of the Franche-Comté region. A total of 194 candidate variants located in 24 genes and 17 genomic regions were imputed on 19,862 cows with phenotypes and genotyped with either the BovineSNP50 (Illumina Inc.) or the EuroG10K BeadChip. We then tested the effect of each SNP in a mixed linear model including random polygenic effects estimated with a genomic relationship matrix. We confirm here the effects of candidate causative variants located in 17 functional candidate genes on both cheesemaking properties and milk composition traits. In each candidate gene, we identified the most plausible causative variant: 4 are missense in the ALPL, SLC26A4, CSN3, and SCD genes, 7 are located in 5′UTR (AGPAT6), 3′ untranslated region (GPT), or upstream (CSN1S1, CSN1S2, PAEP, DGAT1, and PICALM) regions, and 6 are located in introns of the SLC37A1, MGST1, CSN2, BRI3BP, FASN, and ANKH genes.

The increase in the number of farms using Automatic Milking System (AMS) and the practical difficulties in milk recording, lead to examine ways to adapt and simplify protocols to the realities on the ground. One way consists by using an ICAR Peeters and Galesloot method of estimating 24hour fat (percentage and yields) with robots based on one single sample (Peeters and Galesloot, 2002). The aim of this study is to analyze the accuracy of this multiple regression model with six factors at two level: on test day record and on lactation. The method was tested for fat percentage and yields from one single sample (the first) unadjusted and adjusted in comparison with a reference 24-hour. The validation study of regression coefficients of the model was done on an independent data set. The accuracy of the model was estimated from a large dataset. The analysis of the results shows that the correlations are improved using one single sample adjusted instead of one single sample unadjusted compared with a reference 24-hour, respectively 0.786 instead 0.714 for fat percentage and 0.912 instead of 0.852 for fat yields. Given these results, French Genetics Breeding decided to allow the use of the Peeters and Galesloot method’s on robots protocols for all cows with one single sample during the sampling period and use it as estimated 24-hour fat for genetic evaluation.

Cheese-making properties of pressed cooked cheeses (PCC) and soft cheeses (SC) were predicted from mid-infrared (MIR) spectra. The traits that were best predicted by MIR spectra (as determined by comparison with reference measurements) were 3 measures of laboratory cheese yield, 5 coagulation traits, and 1 acidification trait for PCC (initial pH; pH0PPC). Coefficients of determination of these traits ranged between 0.54 and 0.89. These 9 traits as well as milk composition traits (fatty acid, protein, mineral, lactose, and citrate content) were then predicted from 1,100,238 MIR spectra from 126,873 primiparous Montbéliarde cows. Using this data set, we estimated the corresponding genetic parameters of these traits by REML procedures. A univariate or bivariate repeatability animal model was used that included the fixed effects of herd × test day × spectrometer, stage of lactation, and year × month of calving as well as the random additive genetic, permanent environmental, and residual effects. Heritability estimates varied between 0.37 and 0.48 for the 9 cheese-making property traits analyzed. Coagulation traits were the ones with the highest heritability (0.42 to 0.48), whereas cheese yields and pH0 PPC had the lowest heritability (0.37 to 0.39). Strong favorable genetic correlations, with absolute values between 0.64 and 0.97, were found between different measures of cheese yield, between coagulation traits, between cheese yields and coagulation traits, and between coagulation traits measured for PCC and SC. In contrast, the genetic correlations between milk pH0 PPC and CY or coagulation traits were weak (-0.08 to 0.09). The genetic relationships between cheese-making property traits and milk composition were moderate to high. In particular, high levels of proteins, fatty acids, Ca, P, and Mg in milk were associated with better cheese yields and improved coagulation. Proteins in milk were strongly genetically correlated with coagulation traits and, to a lesser extent, with cheese yields, whereas fatty acids in milk were more genetically correlated with cheese yields than with coagulation traits. This study, carried out on a large scale in Montbéliarde cows, shows that MIR predictions of cheese yields and milk coagulation properties are sufficiently accurate to be used for genetic analyses. Cheese-making traits, as predicted from MIR spectra, are moderately heritable and could be integrated into breeding objectives without additional phenotyping cost, thus creating an opportunity for efficient improvement via selection.

Current knowledge and tools have made it possible to generate animals that are better genetically adapted to local environments and livestock systems. Over the past several years, artificial selection efforts have focused on functional traits and trait balance. An important outstanding question is whether it makes sense to classify animals according to their genetic value given that variability exists among local environments and livestock systems. At present, there is a major focus on selecting for feeding efficiency. Efforts aimed at identifying key phenotypes have revealed a preference for animals demonstrating trait balance (i.e., an overall ability to deal with variable conditions and available resources), which can be estimated using summary indices. In efforts to obtain animals adapted to specific livestock systems, genetic strategies must be customised to a certain degree, a challenge that involves making the best possible use of the genetic diversity exhibited by breeding livestock.