Search

Additional file 1: Table S1. Summary statistics of phenotypic data for the mean time of entry (MTE) for nest visits with oviposition for the entire recording period (24–64 weeks of age). Table S2. Genetic ( $${\sigma }_{a}^{2}$$ σ a 2 ), common environment ( $${\sigma }_{c}^{2}$$ σ c 2 ), and residual ( $${\sigma }_{e}^{2}$$ σ e 2 ) variances of laying duration traits. Variance components estimated from the model used to calculate least-square means of laying duration traits. Table S3. Genetic correlations between 28-day periods for mean laying duration. Table S4. Genetic correlations between 28-day periods for mean duration in the nest before laying. Table S5. Genetic correlations between 28-day periods for mean duration in the nest after laying.

International audience; The recent progress and cost reduction of genotyping technologies allows for the implementation of genomic selection (GS) on more and more cultivated species. In this study, we explore the genomic determinism of commercial traits and test the effectiveness and the possible cost reduction of GS in breeding programs of the cupped oyster Crassostrea gigas. Two populations of more than 1000 individuals have been phenotyped and genotyped for 40,000 SNP (Guitierez et al., 2017). Heritability was estimated to be moderate for commercial traits of interest (between 0.19 and 0.34). The accuracy of the genomic prediction models outperformed the classical selection on pedigree by 22 to 55%. A limited linkage disequilibrium (LD) level (less than 0.1) was observed. These results suggest that the use of GS in oyster breeding can improve the selection of breeding candidates to enhance commercial traits but need a specific account and exploration of the very low LD.

Gregarious nesting occurs when a hen chooses a nest already occupied while other nests are vacant, increasing welfare and production problems. This study aims to propose two predictors of gregarious nesting: weighted indegree (WID) and weighted outdegree (WOD), which are the number of times a hen entered an individual nest next to a vacant nest or an occupied nest , respectively. Nesting behaviour was recorded using individual electronic nests for 210 days. Genetic parameters for WID, WOD and nesting behaviour traits were estimated for 1,268 and 1,080 Rhode Island Red and White Leghorn respectively. For both lines, heritability estimates were high for WID (0.40/0.53) and low for WOD (0.12/0.15). Moderate genetic correlations were estimated between WID and WOD (+0.26/-0.35) and strong genetic correlations with some nesting behaviour traits. WID and WOD could help to select against gregarious nesting but the effects for hens using collective nests remain to be evaluated.

Oviposition time is a predictor of the laying rhythm. Drift in oviposition time (i.e. more than 24h between 2 oviposition times) can lead the hen to make a pause day. The recent availability of an individual electronic nest enables high-throughput phenotyping of laying behavior like the record of Daily Oviposition Time (DOT). In this study, the genetic background of DOT along the production time was investigated through the estimation of heritability coefficients for 10 successive periods. For the genetic evaluation purpose, the use of the mean or the variance of DOT per hen, instead of the repeated data, to estimate the breeding values of hens was then considered.Data were continuously recorded on laying hens raised in a barn thanks to individual electronic nests, which use radio-frequency identification of hens. The oviposition time was recorded on 1,180 Rhode Island Red hens (RIR) and 932 White Leghorn hens (WL) between 24 and 64 weeks of age. Heritability coefficients of DOT were estimated for 10 periods of 28-days between 24 and 64 weeks of age, using REML methodology applied to an animal model. Heritability coefficients of DOT, of the mean of DOT per hen (MOT) and the Logarithm of the DOT Variance per hen (LVOT) were estimated for the whole period (24 to 64 weeks of age) and the Estimated Breeding Values (EBVs) of hens for DOT, MOT and LVOT were compared.Earliest recorded DOT was 2h05 and 3h07 after turning on the light for RIR and WL respectively. According to the period of 28-days, low to moderate heritability was estimated for DOT with values ranging from 0.28 ± 0.06 and 0.42 ± 0.07 for RIR, and 0.13 ± 0.03 and 0.24 ± 0.05 for WL. For the whole period, moderate to high heritability was estimated for MOT (0.73 ± 0.08 for RIR and 0.56 ± 0.08 for WL), and LVOT (0.42 ± 0.07 for RIR and 0.45 ± 0.08 for WL). The Spearman correlations between the EBVs of hens for DOT and MOT were 0.99 in both lines. The Spearman correlations between the EBVs of hens for DOT and LVOT were 0.87 and 0.65 for RIR and WL respectively.These results indicate that DOT captures genetic information closer to MOT than LVOT. The rankings of candidates are the same for MOT and DOT. However, the calculation time of EBVs is faster for MOT than for DOT, suggesting that MOT is more interesting for the genetic evaluations. The existence of genetic background for the oviposition time is promising to identify candidate genes implicated in the expression of the laying rhythm.

The main objective of this study was to quantify genotype-by-environment interactions for egg quality traits and body weight in laying hens between cage and cage -free systems. Records of 18,524 eggs for quality traits and 10,266 hens for body weight from a pedigreed Rhode Island Red line were used. Hens were hatched across 5 batches from 2018 to 2020, siblings are housed in two systems: a cage and a floor system. Random regressions were modelled to estimate variance components. Heritability estimates were moderate to high in both systems(0.29-0.77). Genetic correlations between the cage and the floor system were high (≥0.79), and genetic correlations between the intercept and the slope of the reaction norm ranged from -0.69 to 0.27. The results indicate that there is no need to maintain a dedicated breeding programme per system, if the breeding goals are identical, which is unlikely.

National audience; In laying hens, the importance of cage-free systems (barn, free-range and organic) has increased in recent years. These systems require that laying hens lay their eggs in nests. Indeed, off-nest laid eggs impair the production performance of the flock because they are most of the time down graded or lost. Consequently, select laying hens on nesting behavior would be a good opportunity to reduce off-nest laid eggs and improve egg production. However, little is known about such traits. Our study aimed at estimating genetic parameters of nesting behavior traits and their genetic correlations with a production trait, in order to evaluate their potential inclusion in breeding programs. Data were continuously recorded on laying hens raised in a barn thanks to individual electronic nests, which use radio-frequency identification of hens. Nesting behavior and production traits were recorded between 24 (peak of egg production) and 64 (middle of egg production) weeks of age on two pure lines: 1,430 Rhode Island Red hens (RIR) and 1,008 White Leghorn hens (WL). The nesting behavior traits studied were the mean laying duration (MLD), i.e. the mean time spent in the nests to lay, the mean distance between nests of laying (MDN) and the rhythm of lay, like clutch number and the mean oviposition time (MOT). The egg production trait studied was the laying rate in the nests (LRN). Genetic parameters were estimated by applying there stricted maximum like lihood method to an animal model. High (0.55 and 0.68) and moderate (0.32 and 0.37) heritability estimates were found for MLD and MDN respectively. For these two traits, null or favorable genetic correlations with LRN were estimated. Select laying hens with a low MLD and high MDN, i.e. high exploratory behavior, could help to reduce off-nest laid eggs. In the two lines, high heritability estimateswere found for clutch number and MOT with values between 0.40 and 0.66. For LRN, heritability estimates were 0.09 and 0.25 for RIR and WL respectively. Strong negative genetic correlations (-0.72 and -0.61) were estimated between clutch number and LRN, indicating that select laying hens with a low clutch number could increase LRN. Genetic correlations between MOT and LRN were lower with estimates values from -0.04 and -0.44 for RIR and WL respectively. The existence of a genetic background for nesting behavior traits is promising to identify quantitative traits locus and candidate genes implicated in the expression of such traits.

International audience

International audience; The number of laying hens raised in cage-free systems has increased in Europe. In these breeding systems, hens live freely in large groups and must lay eggs in nests. Breeders need to measure the individual laying performance of hens under these very different cage conditions. To do so, electronic nests have been developed. These nests enable to measure regular selection criteria, i.e. egg production and egg quality traits, but also provide access to new egg-laying traits, related to pre-laying behavior and nesting behavior. Reproductive rhythm traits can also be measured precisely by recording the laying time. Although there is little knowledge of the genetic background of these egg-laying traits, there are already selection opportunities for increasing egg-laying performance in these breeding systems as well as improved adaptation capacity and welfare.; Les effectifs de poules pondeuses élevées dans des systèmes alternatifs à la cage augmentent en Europe. Dans ces systèmes, les poules évoluent librement dans des groupes de grande taille et doivent pondre dans des nids. Les sélectionneurs ont besoin de mesurer les performances de ponte individuelles des poules dans ces conditions très différentes de la cage. Pour cela, des nids électroniques ont été développés. Ces nids permettent de mesurer les critères de sélection classiques de production d’œufs et de qualité d’œuf, mais également de nouveaux caractères de ponte liés au comportement précédant la ponte et au comportement au nid. Les caractères de rythme de ponte des poules peuvent aussi être mesurés précisément grâce à l’enregistrement de l’heure de ponte. Les connaissances sur le déterminisme génétique de ces caractères de ponte, encore partielles, offrent d’ores et déjà des perspectives de sélection pour améliorer les performances de ponte dans ces systèmes d’élevage ainsi que la capacité d’adaptation des animaux et leur bien-être.

Improved performances are partly due to heterosis effects. One of the basis of heterosis is dominance, which cannot be inherited.However, it can be exploited to boost the total genetic merit of the animals. This has a special interest in avian selection schemeswhere commercial animals are crossbred. In this study, we have estimated additive and dominance genetic variances for severalegg quality traits in pure-line layers.Around 10,500 egg quality performances were used, collected from 1,148 female Rhode Island layers, phenotyped at 70 weeksold and genotyped using a 600K high density SNP chip. Five egg quality traits were analysed: egg weight (EW), egg shell color(ESC), egg shell strength (ESS), albumen height (AH) and egg shell shape (ESShape). Additive and dominance genetic varianceswere estimated via EM-REML with univariate models. That included an inbreeding coefficient and an additive and a dominancerandom effect. Dominance variance explained a small fraction of the phenotypic variance (between 2 to 4 % across all traits).However, it represented a relevant fraction of the total genetic variance for some of the traits (16%, 10%, 35%, 2.4% and 15% ofthe total genetic variance for EW, ESC, ESS, AH, ESShape, respectively).Further research will estimate additive and dominance genetic correlations between the traits to maximize the total genetic gainof these traits simultaneously. In addition, a genomic BLUP with dominance effects is envisaged for the joint analyses of purebredand crossbred performances, to evaluate the potential to generate superior crossbred performances.

Use of data from crossbred animals for a genomic evaluation of pure line layer. XIth European symposium on poultry genetics (ESPG)

To reduce the cost of genomic selection, low density SNP chip can be used in combination with imputation for genotyping the selection candidates instead of using high density (HD) SNP chip. Concurrently, next-generation sequencing (NGS) techniques has been increasingly used in livestock species but remain expensive to be routinely used in selection. An alternative and costefficient solution is the Genotyping by Genome Reducing and Sequencing (GGRS) techniques to sequence only a fraction of the genome by using restriction enzymes. This approach was simulated from sequences of 1027 individuals in a pure layer line, using four enzymes (EcoRI, TaqI, AvaII and PstI). Imputation accuracy on HD genotypes was assessed as the mean correlation between true and imputed genotypes. Egg weight, egg shell color, egg shell strength and albumen height were evaluated with single-step GBLUP methodology. The impact of imputation errors on the genomic estimated breeding values (GEBV) was also investigated.AvaII or PstI led to the detection of more than 10K SNPs in common with the HD SNP chip resulting in imputation accuracy higher than 0.97. The impact of imputation errors on the ranking of the selection candidates was reduced with Spearman correlation (between GEBV calculated on true and imputed genotypes) higher than 0.97 for AvaII and PstI. Finally, the GGRS approach can be an interesting alternative to low density SNP chip for genomic selection. However, with real data, heterogeneity between individuals with missing data has to be taken into account.

Over the past several years, high-density genotyping has led to major advances in genomic selection and understanding of genomes. This is also the case in laying hens with the availability of the Affymetrix HD 600K chip since 2013. Nowadays, NGS costs are becoming more and more affordable. Based on the HD SNP chip, the use of imputation to go back to the sequence level then makes it possible to obtain information on millions of polymorphisms for a large number of individuals. In this study, we analyzed the quality of the imputation of the HD chip genotypes on sequence genotypes for 90 individuals of a layer line presenting the 2 types of information. After quality control, these individuals presented genotypes for 290K and more than 6 million SNPs respectively with 600K genotyping and 20X sequencing. We carried out a cross-validation with 8 draws of 5 individuals. The correlation between true and imputed genotypes was between 0.96 and 0.99 depending on the individual.Correlations were greater than 0.90 for all chromosomes except chromosomes 16 and 25. After this validation work, we imputed an additional set of 95 individuals to perform GWAS. These analyses confirmed some QTLs of interest as a QTL of egg shell color on chromosome 7 and a QTL of egg shell strength on chromosome 1.

National audience; Avian breeders are currently putting in place genomic selection in their selection schemes. One key parameter to optimize selection schemes, and so genetic gain of selected populations, is to correctly predict the accuracy of genomic evaluation. In a previous study, it has been shown that genomic selection at birth was a very promising strategy for breeders selection in purebred. As crossbred between pure lines is commonly used in avian selection schemes, raises the question about the use of crossbred offspring in breeders evaluation, and their impact on evaluation accuracy. In this study, we had 2 goals. First, estimation of genetic correlation, for the same egg quality trait, between performance from pure line, and cross line layers. Then, estimation of the differences in evaluation accuracy, by adding crossbred information to pure line evaluation. Results shown that genetic correlation between purebred and crossbred trait is strong, but did not allow to consider that it is the same trait. Adding crossbred performances into evaluations allowed to obtain better results, globally, than in separated evaluations.; Depuis quelques années, la sélection génomique commence à se mettre en place dans la filière avicole. Dans le but d’optimiser les schémas de sélection, et donc le progrès génétique des populations sélectionnées, il est important de connaître la précision de l’évaluation génomique, comparativement à l’évaluation génétique. Dans une étude précédente, il avait été observé que l’usage de la sélection génomique à la naissance était une stratégie très prometteuse pour la sélection des reproducteurs en race pure. Le croisement entre lignées pures étant utilisé à large échelle par les schémas de sélection avicoles, il s’est posé la question de l’utilisation des données des descendantes croisées sur l’évaluation des reproducteurs, et de leur influence sur la précision de l’évaluation. Dans le cadre de cette étude, 2 objectifs ont été poursuivis. Tout d’abord, estimer la corrélation génétique, pour un même caractère de qualité d’œuf, entre la performance exprimée par des poules de lignée pure, et celle des poules croisées. Ensuite, estimer les différences de précision apportées par l’intégration des données des croisées à l’évaluation en lignée pure. Les résultats ont montré que la corrélation génétique entre caractère pur et croisé est forte, mais ne permet pas de considérer qu’il s’agit du même caractère. L’ajout des performances de croisées aux évaluations a permis d’obtenir des résultats globalement meilleurs qu’en évaluations séparées.

La structure du déséquilibre de liaison (DL) au sein des populations en sélection impacte fortement la précision obtenue lors des études de cartographie de QTL ou lors de l’évaluation génomique des reproducteurs. Chez les oiseaux, la structure hétérogène du génome nécessite de décrire précisément le DL pour optimiser la sélection. L’utilisation des puces SNP haute densité pour le génotypage des populations de volailles est une opportunité pour approfondir notre connaissance de la structure du DL de ces populations. L’objectif de cette étude est d’acquérir une connaissance haute résolution de la structure du DL au sein de populations de poules de types ponte et chair. Nous avons analysé les génotypes (puce 600 K Affymetrix® Axiom® HD SNP) de 1541 animaux issus de 3 populations. L’étendue et le niveau du DL ont été estimés par le r2 moyen à distance physique donnée entre SNP. Cette étude met en évidence des différences importantes de structure du DL entre lignées et entre chromosomes. L’étendue et le niveau du DL sont plus importants dans les lignées de type ponte ou pour les macro-chromosomes et le chromosome Z. Ce niveau important de DL peut faciliter la détection de QTL sur ces chromosomes, mais peut également compliquer la localisation fine de polymorphismes causaux. A l’inverse, le faible niveau de DL observé sur les micro-chromosomes nécessite l’utilisation d’une forte densité de SNP pour détecter une association avec un phénotype, mais devrait permettre la cartographie fine d’un polymorphisme causal. Ces différences sont à prendre en considération pour définir une stratégie de génotypage économique et efficace pour la cartographie fine de QTL ou l’évaluation génomique., Knowledge of the linkage disequilibrium (LD) pattern is useful in animal genetic studies as it underlies mapping studies and genomic selection. This is all the more important in birds given the heterogeneous structure of the avian karyotype. Recently, the availability of the high density 600 K Affymetrix® Axiom® HD SNP genotyping array allowed to assess an in-depth knowledge of the LD pattern in chicken genome. The aim of the present study was to assess a higher resolution of the LD pattern in chicken genome in layer and broiler lines. In this study, we analyzed genotypes of 1541 animals from layers and broiler commercial populations to characterize their LD pattern. LD was measured by the average r2 value at a given physical distance between SNP. LD extended over a larger region for layer lines than for broiler line. Most differences between lines appeared at small interval distances (< 0.5Mb). LD extent and decay differed considerably between chromosomes categories. Average r2 values were higher for Z chromosome than for macro, intermediates and microchromosomes. The extent of useful LD observed for autosomal chromosomes was at least ten-fold longer for layer lines than for broiler. Finally, this study shed light on high LD for Z chromosome. The differences in LD pattern observed between chromosomes and chicken lines should be taken into account to define an economically efficient genotyping strategy.

National audience; Meeting the issues related to sustainable development of livestock farming requires innovations in genetics. In every sector, this is all about improving the robustness of animals by enhancing their abilities to adapt by including numerous efficiency traits in the breeding goals of selection programs particularly regarding health and adaptability to less standardized feed resources and of fluctuating nutritional value, changing and varied environments, and climate change. This leads to consider genotype-by-environment interactions within genetic evaluation procedures and to assess the performances of animals in low input systems, especially regarding feed and drug use. Since the end of the 2000s, the method that has been progressively used in all livestock breeding sectors is genomic selection, which, by disconnecting the computation of genetic values of selection candidates from phenotype recordings, makes it easier to select new traits and to take into account genotype-by-environment interactions. In the relatively near future, it will be technically possible to use strategies of targeted modification and selection of genomes that could substantially increase genetic gains, in comparison with those obtained with current genomic selection programs. However, the use of these new breeding techniques raises numerous issues on scientific, operational, ethical, and political levels.; Les objectifs et méthodes d’amélioration génétique doivent évoluer pour répondre aux enjeux du développement durable des filières d’élevage. Dans toutes les espèces, il s’agit désormais d’améliorer la robustesse des animaux en promouvant leurs capacités d’adaptation, en intégrant dans les objectifs des programmes de sélection de nombreux caractères d’efficience, en particulier de santé et d’adaptation à des ressources alimentaires moins standardisées et de qualité fluctuante, à des environnements changeants et variés, et au changement climatique. Cela amène à considérer les interactions génotype x environnement dans la prédiction des valeurs génétiques et à évaluer les performances des animaux dans des systèmes à moindres intrants, notamment alimentaires et médicamenteux. Depuis la fin des années 2000, la méthode mise en oeuvre progressivement dans toutes les filières est la sélection génomique qui, en déconnectant le calcul des valeurs génétiques des candidats à la sélection de l’obtention de phénotypes, facilite la sélection de nouveaux caractères et la prise en compte des interactions génotype x environnement. Dans un futur relativement proche, il sera techniquement possible de mettre en oeuvre des stratégies de modification et sélection ciblées des génomes qui pourraient substantiellement accroitre les gains de progrès génétique en comparaison avec ceux obtenus dans les programmes actuels de sélection génomique. Toutefois, l’utilisation de ces nouvelles techniques de sélection animale pose de nombreuses questions, tant sur le plan scientifique, qu’opérationnel, éthique ou politique.

Avian breeders are actually putting in place genomic selection in their selection schemes. One key parameter to optimize selection schemes is to correctly predict the accuracy of genomic evaluation. This paper presents results obtained by comparing accuracy of genetic and genomic evaluations on multiple traits, using several reference populations with different amount of phenotypic informations. Genetic parameters (heritability and genetic correlations) were observed, and found very stable. EBVs and GEBVs accuracies showed that genomic evaluation was each time more accurate than genetic evaluation, especially when phenotypic information was restreint. It was also found that the use of sexual chromosome genotype had a negative impact on traits evaluation for dams (no effect found on sires). In any case, genomic evaluation of breeding birds at birth seems to be a promising strategy for layers.

The main goal of selection is to choose breeders of the next generation among a set of selection candidates. In genomic selection, the choice of breeders is based on the use of information on DNA polymorphisms, in particular SNP, in addition of performance measures. Since 2013, a commercial high density genotyping chip (600,000 markers) for chicken allowed the implementation of genomic selection in layer and broiler breeding. However, genotyping costs with this chip still remain high for a routine use on a large number of selection candidates. Consequently, it is interesting to develop, at a lower cost, low density genotyping chips. To do so, a set of SNP markers has to be selected to enable an imputation (prediction) of missing genotypes with high accuracy on a high density chip (HD chip). In this perspective, we conducted various simulation studies to choose the optimal strategy for low density genotyping of two different lines of laying hen. Different low density genotyping chips were designed according to two methodologies: a choice of SNP depending on a clustering of SNP based on linkage disequilibrium threshold or a choice of SNP at regular intervals (kb) along each chromosome. Imputation accuracy was assessed as the mean correlation between true and imputed genotypes. Results showed that correlations were more sensitive to false imputation of SNPs with low Minor Allele Frequency (MAF) with the equidistant methodology. Imputation accuracy improved with SNP density and when a higher LD threshold is used for SNP selection. Given the particular structure of the avian genome with chromosomes of very heterogeneous sizes and extents of LD, imputation accuracy differed according to the type of chromosome. All the simulation studies showed that linkage disequilibrium methodology enabled to get better results of imputation than with equidistant methodology.

The main goal of selection is to choose breeders of the next generation among a set of selection candidates. In genomic selection, the choice of breeders rests on the use of information on DNA polymorphisms, in particular SNP, in addition of performance measures. Since 2013, a commercial high density genotyping chip (600,000 markers) for chicken allowed the implementation of genomic selection in layer and broiler breeding. However, genotyping costs with this chip still remain high for a routine use on a large number of selection candidates. Consequently, it is interesting to develop, at a lower cost, low density genotyping chips. To do so, a set of SNP markers has to be selected to enable an imputation (prediction) of missing genotypes on a high density chip (HD chip). This imputation enables to predict missing genotypes of all selection candidates from high density genotyping of a reference population with phenotypes. In this perspective, according to the reference population, various simulation studies were conducted to choose the best strategy for low density genotyping of laying hen lines. Two different low density genotyping chips of 10K SNP were designed according to two methodologies: a choice of SNP depending on a clustering based on linkage disequilibrium threshold or a choice of SNP at regular intervals (kb) along each chromosome. Imputation accuracy was assessed as the mean correlation between true and imputed genotypes. Focusing on populationnal factors that can influence imputation accuracy, it is shown that imputation accuracy improves with an increase in the size of the reference population. By decreasing the kinship degree between reference and candidate population, it is seen that imputation accuracy decreases. Most importantly, results show that a key point in getting good imputations is to have the direct parents in the reference population. Finally, all different genotyping strategies focused on population factors show that linkage disequilibrium methodology enables to get better results of imputation than with equidistant methodology.

Knowledge of the linkage disequilibrium (LD) pattern is useful in animal genetic studies as it underlies mapping studies and genomic selection. Recently, the availability of the high density 600K Affymetrix® Axiom® HD SNP genotyping array allows to asses a higher resolution of the LD structure in chicken genome. In this study, we analysed genotypes of 1541 animals from layers and broiler commercial populations to characterize their LD pattern. LD was measured by the average r2 value at a given physical distance between SNP. LD extends over a larger region for layers line than for broiler line. Most differences between lines appeared at small interval distance (< 0.5Mb). LD extent and decay differed considerably between chromosome categories. Average r2 values were higher for Z chromosome than for macro, intermediates and microchromosomes. The extent of useful LD observed for autosomal chromosomes is at least tenfold longer for layers line than for broiler. Finally, this study shed light on high LD for the Z chromosome. The differences in LD pattern observed between chromosomes and chicken lines should be taken into account to define an economically efficient genotyping strategy.

International audience; Deterministic modelling of a poultry breeding scheme Deterministic modeling of selection schemes consists of calculating the expected genetic gain after a selection process. Consequences of selection on the genetic trend are calculated using equations based on estimated breeding values and on parameters describing the population and the selection process. Professionals could use this tool to compare different strategies for the improvement of their selection schemes. In this study, the model is based on a layer chicken breeding scheme. Breeding goal includes numerous traits such as laying performances, egg quality and body weight. This is a two steps model combining a familial selection process and an individual one. Within selected families, candidates from the good families are undergoing a mass selection whereas candidates from medium families are undergoing a within family selection of a fixed number of candidates. The model requires a large number of parameters: number of candidates, proportions of selected families and selected candidates per family, accuracy of breeding values, genetic correlations between traits, weights of traits in the selection index. The variability of the parameters composing the model enables comparing a large diversity of scenarios, including the use of genomic evaluations. In order to illustrate the usefulness of the model, we simulated 3 selection strategies using classic breeding values and/or genomic breeding values. The genetic gain predictions obtained were then compared. The model included two traits T1 (h² = 0.13) and T2 (h² = 0.33), with a genetic correlation varying between 0 and 0.70 and a weight in the index of 40% and 60% respectively. Results show an increase of predicted gain for both traits with genomic evaluation. Improvement is constant for T2, whereas for T1 the increase of predicted gain is much better when genetic correlation between T1 and T2 is high. The genomic breeding scheme model seems more sensitive than the classic breeding scheme to parameters like the between traits correlation and their weight in the index.