Die vorliegende Arbeit befasst sich mit der Optimierung von Zuchtprogrammen. Zum einen wurde eine neue Methode zur Berücksichtigung der mittleren Inzucht in Zuchtplanungsrechnungen entwickelt. Zum anderen werden zwei gänzlich unterschiedliche Zuchtprogramme modelliert und aktuelle Optimierungsansätze validiert. Dabei werden sowohl der naturale als auch der monetäre Zuchtfortschritt und der diskontierte Züchtungsgewinn berücksichtigt. Im Projekt FUGATO+brain wurde die Zuchtplanungssoftware ZPLAN neu programmiert und mit weiteren zuchtplanerischen Werkzeugen versehen. Als Ergebnis des Projektes entstand die Software ZPLAN+. Diese ermöglicht die Modellierung von komplexen Zuchtprogrammen und kann zur Optimierung von Zuchtprogrammen genutzt werden. Die Software ist anwenderfreundlich und umfasst alle Bereiche der Zuchtplanung. Zur Berechnung der mittleren Inzucht wurde eine neue Methode für die Implementierung in der Zuchtplanung entwickelt. Die Methode basiert auf der mittleren Kinship in einer Zuchtpopulation. Die Kinship ist definiert als die Wahr-scheinlichkeit, dass innerhalb einer Gruppe am gleichen Locus zwei zufällig gewählte Allele herkunftsgleich sind. Die Berechnung der Kinship erfolgt auf Grundlage der Genflussmethode. Zur Validierung der Methode wurde eine früher beschriebene Schafpopulation verwandt, die in unterschiedlichen Weisen modifiziert wurde. Insgesamt wurden drei verschiedene Szenarien modelliert, wovon das erste von einem Populationswachstum ausging. Im zweiten Szenario wurde angenommen, dass die Populationsgröße durch einen Flaschenhals verringert wird und sich dann wieder erhöht. Für die dritte Modellierung wurde die Population über einen Zeitraum getrennt und dann wieder zusammengeführt. Es konnte gezeigt werden, dass sich mit der vorgeschlagenen Methode in sämtlichen komplexen Populationsstrukturen die mittlere Inzucht und die effektive Populationsgröße berechnet lässt. In einer Zuchtplanungsrechnung für Reitpferde sollte der gezielte Einsatz von Embryotransfer in einem Pferdezuchtprogramm validiert werden. Hierfür wurde ein Zuchtprogramm in ZPLAN+ modelliert, welches das aktuelle Zuchtprogramm des Hannoveraner Verbandes e.V. näherungsweise abbildet. In verschiedenen Szenarien wurde eine schärfere Selektion auf der Stutenseite modelliert, wobei die besten Stuten des Zuchtprogramms als Spenderstuten für den Embryotransfer eingesetzt wurden. Es wurde davon ausgegangen, dass die zur Selektion zur Verfügung stehenden Stuten sowohl Ergebnisse in der Eintragung, als auch Ergebnisse einer Leistungsprüfung haben. Die Anzahl der zur Selektion verfügbaren Stuten wurde ebenso variiert wie die Anzahl der selektierten Stuten und die Anzahl der geborenen Fohlen je Spenderstute. Deutlich wurde, dass der Embryotransfer die Möglichkeit bietet den Zuchtfortschritt in einem Pferdezuchtprogramm stark zu steigern, wobei dies mit einer Steigerung der Kosten für die Züchter einhergeht. Mit dem vorgeschlagenen Ansatz zur Inzuchtberechnung konnte gezeigt werden, dass die scharfe Selektion und der starke Einsatz der Spenderstuten eine Erhöhung der mittleren Inzucht und daraus folgend eine geringere effektive Populationsgröße nach sich zieht. Im dritten Abschnitt der Arbeit sollten die Auswirkungen der Einbeziehung von genomischen Informationen in ein Legehennenzuchtprogramm gezeigt werden. Dafür wurde in enger Kooperation mit der Lohmann Tierzucht GmbH ein Zuchtprogramm zur Produktion von 500 Mio. Legehennen in ZPLAN+ nachgebildet. Die Produktion der Elterntiere basiert auf einer Kreuzung von vier Nukleuslinien, die konventionelle Selektion stützt sich vor allem auf die Leistungsprüfung von Hennen in den einzelnen Linien. Zur Nutzung der genomischen Informationen wurde von unterschiedlich großen Kalibrierungsstichproben ausgegangen. In einem ersten Schritt wurden die genomischen Informationen der Hähne zusätzlich zu allen konventionellen Selektionskriterien genutzt. Dabei wurde die Anzahl der getesteten Hähne variiert und in einem weiteren Schritt wurde davon ausgegangen, dass die Hennen ebenfalls genotypisiert sind. In einem weiteren Szenario basierte die Selektion nur auf Pedigreedaten und genomischen Informationen. Deutlich wurde, dass in der zweiten Variante das Generationsintervall massiv gesenkt werden konnte. Der Zuchtfortschritt konnte in allen modellierten Varianten erhöht werden, wobei es Unterschiede in den Einzelmerkmalen gab. Die Einführung der genomischen Informationen in die Legehennenzucht ist verbunden mit einem massiven Kostenanstieg. Inwieweit der gesteigerte Zuchtfortschritt den Kostenanstieg rechtfertigt bedarf einer Marktanalyse seitens der Zuchtunternehmen.

Orientador: Evandro Vagner Tambarussi / Resumo: Parentesco entre indivíduos em populações naturais é uma informação com múltiplos usos e pode ser acessada por meio de estimadores que usam dados moleculares. Todavia, cada estimador possui pressuposições e, muitas vezes, rigorosas para espécies florestais. Assim, a modelagem dos dados é uma etapa importante e, quase em todos os casos, negligenciada. Este trabalho teve como objetivo avaliar a eficácia de diferentes estimadores de parentesco em Acrocomia aculeata (Jacq.) Lodd. ex Mart., Hymenaea stigonocarpa Mart. ex Hayne e Dipteryx alata Vogel., bem como em quatro diferentes populações simuladas. A partir das coancestrias médias ( ������̅ ) estimadas, foi calculado o erro dos estimadores pressupondo que os indivíduos analisados eram meios-irmãos (������̅=0,125). As estimativas de parentesco oscilaram conforme a espécie e o método utilizado, gerando diferentes valores de erro para cada estimador. A correlação foi observada apenas entre estimadores que possuíam o mesmo método de estimativa ou com pressupostos similares. As populações simuladas tiveram melhores valores estimados e menores erros em comparação com dados reais. Os valores de erro dos estimadores encontrados, demonstram que somente aplicação dos estimadores para a inferência de determinado grau de parentesco, pode gerar resultados viesados e corroborar para ineficácia da tomada de decisão, sendo necessário o uso de informações complementares associadas ao parentesco, como análise do sistema de reprodução, estrutura gen... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: Understanding kinship between individuals in natural populations offers useful information that can be assessed based on estimators of molecular data. However, each estimation method is based on assumptions and is often restricted for forest species. Thus, modeling the data is an important step that is almost always neglected. This study aims to evaluate the efficacy of different kinship estimators for Acrocomia aculeata (Jacq.) Lodd, ex Mart., Hymenaea stigonocarpa Mart. Ex Hayne, and Dipteryx alata Vogel., and four different simulated populations. From the estimated mean coancestry (�����̅), the error estimates were calculated assuming that analyzed individuals were perfect half-siblings (�����̅=0,125). Estimates of kinship ranged according to the species and method used, generating different error values for each estimator. A correlation was observed only between estimators that used the same estimation method or similar assumptions. Simulated populations showed more accurate estimates and lower error values compared to actual data. The error values of the estimators demonstrate that the application of estimators to infer a certain degree of kinship can generate biased results and lead to inefficient decision making. Thus, the use of complementary information associated with kinship is necessary, such as analysis of the reproduction system and genetic structure of the population, enabling more precise inferences of the kinship between evaluated individuals. / Mestre

Submitted by MARCOS VINICIUS ANTUNES DE LEMOS (marcoslemoszootec@gmail.com) on 2017-06-02T15:32:45Z No. of bitstreams: 1 Tese_Marcos_VA_Lemos1.pdf: 3226396 bytes, checksum: 1f97ad6fd98b95977fbdc4ba573a83bb (MD5) / Approved for entry into archive by Luiz Galeffi (luizgaleffi@gmail.com) on 2017-06-02T17:51:15Z (GMT) No. of bitstreams: 1 lemos_mva_dr_jabo.pdf: 3226396 bytes, checksum: 1f97ad6fd98b95977fbdc4ba573a83bb (MD5) / Made available in DSpace on 2017-06-02T17:51:15Z (GMT). No. of bitstreams: 1 lemos_mva_dr_jabo.pdf: 3226396 bytes, checksum: 1f97ad6fd98b95977fbdc4ba573a83bb (MD5) Previous issue date: 2017-05-05 / Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) / Objetivou-se identificar regiões no genôma de bovinos da raça Nelore que apresentam variações no número de cópias (CNV) e, associar estes CNV com o perfil de ácidos graxos da carne. Além disso, objetivou-se realizar associação genica ampla utilizando os método de single step (GWASss) a fim de detectar regiões genômicas associadas aos ácidos graxos dos grupos saturados, mono e poliinsaturados, assim como os omegas 3, 6 e sua relação. O estudo de caracterização e distribuição dos CNVs ao longo do genoma de bovinos Nelore, foi realizado através do software PennCNV utizando dados genotípicos de 3.794 aniamais, resultando em 399.361 CNVs identificados. Após controle de qualidade, 2.902 foram mantidos nas analises, resultando em 195.873 CNVs, com tamanho medio de 54,744 pb, maximo de 8.7 Mb e minimo 3 kb. As regiões de CNV foram geradas pela sobreposição dos CNVs através do software CNVRuler. Os cromossomos que mostraram maior incidencia de CNVR foram BTA19 (24,26%), BTA23 (18,68%) e BTA25 (18,05%). Ja os que mostraram menor incidencia foram BTA29 (1,63%), BTA13 (9,72%) and BTA8 (9,72%). As 9.805 regiões da CNV estimadas no presente estudo cobrem aproximadamente 13.05% do genoma bovino e sobrepõem-se a 5.495 genes conhecidos que envolvem processos biológicos que poderiam estar envolvidos na adaptação ambiental da subespécie a áreas tropicais. O estudo de GWASss identificou 115 janelas que explicaram mais de 1% da variação genética aditiva para os 22 ácidos graxos estudados. A identificação destas regiões e seus genes genes, tais qual ELOVL5, ESRRG, PCYT1A e os genes do grupo ABC (ABCA5, ABCA6 e ABCA10) são genes que estão relacionados direto e inderamente ao metabolismo lipídico. O GWAS entres os fenótipos de AG e os CNVs resultaram em um total de 186 CNVR siginificantivos para os grupos dos ácidos graxos saturados (43), monosaturados (42), poliinsaturados (66) e omegas (35), nas quais foram identificados 278 genes com funçao descrita. Estes resultados apontaram genes associados a AG de varias saturações, podendo ser destacados os genes SAMD8 e BSCL2, os quais estão relacionados ao metabolism lipidico; e o gene RAPGEF6, relacionado ao metabolism energetico. Assim os inúmeras regiões genômicas encontradas neste estudo, bem como os genes identificados nas mesmas, devem contribuir para a formação de uma base genética do perfil de ácidos graxos da carne de bovino Nelore (Bos indicus), podendo contribuir para uma melhor seleção das características associadas à melhora da saúde humana. saúde. O conhecimento desses CNVs deverá melhorar a compreensão dos mecanismos genéticos e fisiológicos que contribuem para as características produtivas, bem como na seleção de animais mais produtivos e eficientes, contribuindo para o melhoramento genético das características produtivas. / The objective of this study was to identify genomic regions of Nellore cattle that present variations in the number of copies (CNV) and to associate these CNV with the fatty acid profile of the meat. In addition, the objective of this study was to carry out a genome-wid association using the single step method (GWASss) to detect genomic regions associated with saturated, mono and polyunsaturated fatty acids, as well as omega 3, 6 and their relationship. The study of the characterization and distribution of CNVs along the Nellore genome was performed by PennCNV software using genotypic data of 3,794 animals, resulting in 399,361 CNVs identified. After quality control, 2,902 were maintained in the analyzes, resulting in 195,873 CNVs, with an average size of 54,744 pb, maximum of 8.7 Mb and minimum 3 kb. The CNV regions were generated by the overlap of the CNVs by CNVRuler software. The chromosomes that showed the highest incidence of CNVR were BTA19 (24.26%), BTA23 (18.68%) and BTA25 (18.05%). Those with the lowest incidence were BTA29 (1.63%), BTA13 (9.72%) and BTA8 (9.72%). The 9,805 CNV regions estimated in the present study covered approximately 13.05% of the bovine genome and overlaped 5,495 genes known to envolve in biological processes that could be involved in the environmental adaptation of the subspecies to tropical areas. The GWASss study showed 115 windows that explained more than 1% of the additive genetic variation for the 22 fatty acids studied. The identification of these regions and their genes, such as ELOVL5, ESRRG, PCYT1A and the ABC group genes (ABCA5, ABCA6 and ABCA10) are genes directly and indirectly related to lipid metabolism. The GWAS between AG phenotypes and CNVs resulted in a total of 186 CNVRs that were significant for the saturated (43), monosaturated (42), polyunsaturated (66) and omegas (35) fatty acids groups, in which 278 genes with described function. These results pointed genes associated to AG of various saturations, and the genes SAMD8 and BSCL2, which are related to lipid metabolism; and the RAPGEF6 gene, related to energetic metabolism. Thus, the numerous genomic regions found in this study, as well as the genes identified in them, should contribute to the formation of a genetic basis of the Nellore beef (Bos indicus) fatty acid profile, contributing to a better selection of the traits associated with improvement of human health. The knowledge of these CNVs could improve the understanding of the genetic and physiological mechanisms that contribute to the productive traits, as well as the selection of more productive and efficient animals. / FAPESP: 2013/11853-4

A seleção genômica é a mais moderna tecnologia no que tange a utilização de marcadores genéticos para a seleção e melhoramento genético de animais domésticos. Em síntese a metodologia consiste em uma seleção assistida por marcadores em uma escala ampla do Genoma e foi proposta por Meuwissen et al. (2001). O desequilíbrio de ligação (DL) entre os marcadores e loci de características quantitativas (QTL) e a composição da população referência são pontos chave para a confiabilidade da seleção genômica. À medida que os indivíduos se distanciam geneticamente, o DL entre marcadores e QTL diminui, o que dificulta a aplicação da seleção genômica em populações de animais multirraciais e explica o fato de até hoje a maior parte das pesquisas se concentrarem na estimação de valores genômicos apenas para animais de raças puras. A aplicação dessa metodologia em populações multirraciais de bovinos constitui uma possibilidade real de grandes avanços no melhoramento genético do rebanho brasileiro de bovinos de corte. / Genomic Selection is the newest technology into the use of genetic markers at the animal breeding. In sinthesys the metodology consists in a marker assisted selection in a genome wide scale, it was pourpused by Mewissent et al. (2001). The linkage disequilibrium between the marker and quantitative trait loci (QTL) and the composition of the reference population are key points to the reablility of the genomic selection. Once the individuals get genetically distants, the DL between markers and QTLs decays, what turns hard the aplication of genomic selection in multirracial populations of animals and explains the fact of untill today the major part of researchs are dedicated to the estimation of genomic breeding values only to purebreed animals. The aplication of this methodology in multibreed cattle populations consists in a real possibility to reach greats advances in genetic improvement of the brazilian beef cattle heard.

In plant breeding, genomic prediction (GP) may be an efficient tool to increase the accuracy of selecting genotypes, mainly, under multi-environments trials. This approach has the advantage to increase genetic gains of complex traits and reduce costs. However, strategies are needed to increase the accuracy and reduce the bias of genomic estimated breeding values. In this context, the objectives were: i) to compare two strategies to obtain markers subsets based on marker effect regarding their impact on the prediction accuracy of genome selection; and, ii) to compare the accuracy of four GP methods including genotype × environment interaction and two kernels (GBLUP and Gaussian). We used a rice diversity panel (RICE) and two maize datasets (HEL and USP). These were evaluated for grain yield and plant height. Overall, the prediction accuracy and relative efficiency of genomic selection were increased using markers subsets, which has the potential for build fixed arrays and reduce costs with genotyping. Furthermore, using Gaussian kernel and the including G×E effect, there is an increase in the accuracy of the genomic prediction models. / No melhoramento de plantas, a predição genômica (PG) é uma eficiente ferramenta para aumentar a eficiência seletiva de genótipos, principalmente, considerando múltiplos ambientes. Esta técnica tem como vantagem incrementar o ganho genético para características complexas e reduzir os custos. Entretanto, ainda são necessárias estratégias que aumentem a acurácia e reduzam o viés dos valores genéticos genotípicos. Nesse contexto, os objetivos foram: i) comparar duas estratégias para obtenção de subconjuntos de marcadores baseado em seus efeitos em relação ao seu impacto na acurácia da seleção genômica; ii) comparar a acurácia seletiva de quatro modelos de PG incluindo o efeito de interação genótipo × ambiente (G×A) e dois kernels (GBLUP e Gaussiano). Para isso, foram usados dados de um painel de diversidade de arroz (RICE) e dois conjuntos de dados de milho (HEL e USP). Estes foram avaliados para produtividade de grãos e altura de plantas. Em geral, houve incremento da acurácia de predição e na eficiência da seleção genômica usando subconjuntos de marcadores. Estes poderiam ser utilizados para construção de arrays e, consequentemente, reduzir os custos com genotipagem. Além disso, utilizando o kernel Gaussiano e incluindo o efeito de interação G×A há aumento na acurácia dos modelos de predição genômica.

La sélection génomique (SG) des animaux et des plantes a été rendue possible grâce aux avancées des biotechnologies, notamment des puces à ADN de haute densité et de faible coût. Son efficacité et sa profitabilité a été clairement démontrée chez les bovins laitiers, où elle a été très rapidement mise en pratique. En revanche, son application pour les petits ruminants est encore limitée, et, notamment, n’a pas démarré en France. Ses potentialités sont toutefois à l’étude dans quelques programmes concernant les ovins et caprins laitiers, et les responsables des filières correspondantes désirent connaitre l’efficacité de cet outil dans leur situation. Cependant, la prudence est de règle, compte tenu des différences entre les schémas de sélection des bovins laitiers et des petits ruminants. Cette étude fait partie d’un programme entrepris pour évaluer l’utilisation et la gestion de l’information génomique dans les schémas de sélection ovin et caprin. Au cours de cette thèse ont été examinés (1) l’impact de la SG sur le gain génétique dans des schémas de sélection de petits ruminants, (2) l’efficacité économique de la SG en petits ruminants, en prenant l’exemple d’un programme de sélection ovin-viande; (3) l’importance d’une optimisation de certaines décisions (quantifiées par des variables dans un modèle décrivant les schémas) pour maximiser le progrès génétique et (4) une piste contribuant à l’optimisation de la population de référence. Les modèles utilisés appartiennent au champ des méthodes déterministes et les exemples ont porté sur les schémas de sélection existants (ovins laitiers, ovins viande et caprins laitiers). Les résultats de cette étude suggèrent que la sélection génomique peut être plus rentable que la sélection classique en terme de gain génétique, à condition qu’une population de référence de taille moyenne soit disponible (environ 2000 individus). Ils montrent, en particulier dans les schémas laitiers, que le potentiel de la SG de réduire l’intervalle de génération pourrait fortement augmenter le gain génétique. Dans le schéma ovin allaitant modélisé, combiner l’information génomique et les phénotypes de caractères bouchers donne plus de gain génétique que la sélection classique ou la SG sans phénotype sur les candidats. En termes d’impacts économiques, les résultats du schéma ovin allaitant modélisé montrent que toutes les stratégies de sélection génomiques sont plus onéreuses que la sélection classique. Cependant, les gains marginaux (recettes totales moins coûts variables) de certains scénarii de SG s’avèrent légèrement plus élevés que pour la sélection classique. L’étude montre également, dans tous les schémas et stratégies de sélection, que l’optimisation de l’utilisation de variables de décision pourrait grandement augmenter le gain génétique et l’efficacité économique, par rapport aux situations actuelles. Avec cette étude, on peut conclure que la mise en place de la sélection génomique dans les programmes de sélection des petits ruminants est possible et pourrait être plus bénéfique que la sélection classique dans certains cas. Cependant, il y a plus d’obstacles par rapport aux bovins laitiers, en particulier, la construction d’une population de référence fiable et des coûts élevés de génotypages par rapport à la valeur des candidats à la sélection. Ces obstacles pourraient freiner sa mise en œuvre, voire l’empêcher dans certaines races. / Implementing genomic selection (GS) in small ruminant breeding programs is still at the research and development level. This new way of selection in animals and plants was made possible thanks to the development of low costs, high density SNP chips. It proved to be highly beneficial in dairy cattle breeding programs. The French small ruminant industries are strongly interested in evaluating the efficiency of this tool in their situation. However, they are also very cautious given the inherent differences in terms of capacity and functionalities between dairy cattle and small ruminant breeding programs. This study is part of bigger efforts mobilized to evaluate the use and management of genomic information in sheep and goats breeding programs. The PhD work examined (1) the impact of genomic selection on genetic gain of small ruminant breeding programs; (2) the economic efficiency of genomic selection in small ruminant, through an example of a meat sheep breeding program; (3) the benefits of optimizing the use of decision variables on genetic gain; and (4) contributed some ideas on how to optimize the choice of individuals in the reference population. The modeling parts were done by deterministic methods and the examples focused on the existing breeding programs (dairy sheep, meat sheep and dairy goats) with medium to small size breeding units. The results of this study suggest that adopting genomic selection can be more profitable than classic selection in terms of genetic gain, provided that, at least, a medium size reference population is available (around 2,000 individuals). They show, especially in dairy breeds, that the GS potentials of reducing generation interval could greatly increase the genetic gain. In meat sheep breeding program, exploring the possibility of combining genomic information and meat phenotypes gave higher genetic gain than classic or pure genomic selection. In terms of economic impacts, results of the meat sheep breeding program we modeled show that all genomic selection strategies are more expensive than classic selection. However, the contribution margins (total revenues minus total variable costs) of some GS variants were slightly higher than benefits from classic selection. The study also shows, across breeds and selection strategies, that optimizing the use of decision variables could greatly increase the genetic gain and benefits, compared to the current situation. With this thesis we can conclude that adopting genomic selection in small ruminant breeding programs is possible and could be more beneficial than classic selection in some cases. However, there are more obstacles compared to dairy cattle, especially, construction of reliable reference populations and high costs of genotypes relative to the value of selection candidates. These might delay implementation in general or prevent it in some breeds.

L'efficacité de la sélection génomique étant principalement dépendante de la taille de la population de référence, seules les principales races laitières françaises bénéficient aujourd'hui d'évaluations génomiques. Pour contourner cette contrainte et développer des évaluations génomiques pour les races régionales, il a été proposé de créer une population de référence commune entre races.Cependant, utiliser une population de référence multiraciale nécessite que le lien entre QTL et marqueurs soit conservé entre races, ce qui implique, sauf cas particulier, l'utilisation d'une puce haute densité. Les taux d'erreur d'imputation des génotypes haute densité à partir de génotypes moyenne densité (classiquement utilisés en bovins) ont été étudiés. La précision d'imputation étant supérieure à 99% dans la majorité des races laitières, l'imputation des génotypes des animaux des populations de référence des principales races laitières a été réalisée. Cette population de référence « haute densité » a ensuite été utilisée pour le développement d'évaluations génomiques multiraciales. Plusieurs stratégies d'évaluations génomiques (intra-race ou multi-race, à partir de génotypes moyenne ou haute densité) ont été comparées pour différentes tailles de populations de référence. Les évaluations multiraciales basées sur la puce haute densité permettent d'améliorer la précision des évaluations génomiques dans le cas d'une population de référence de 500 taureaux ou moins. L'efficacité d'une troisième stratégie utilisant des évaluations génomiques multiraciales à partir de la puce moyenne densité pour un groupe de races proches a donc été étudiée. L'augmentation de la précision des évaluations génomiques observée dans ce cas était trois fois supérieure à celle qui avait été observée dans le cas des principales races laitières. Par ailleurs, dans les deux cas étudiés ici, la précision des évaluations génomiques intra-races est relativement élevée, même dans le cas d'une population de référence réduite, lorsque les pères des candidats à la sélection sont inclus dans la population de référence.Les résultats obtenus suggèrent donc que la mise en place d'évaluations génomiques dans les races régionales est envisageable. Il faudra toutefois poursuivre les travaux pour déterminer quelle stratégie optimale peut/doit être utilisée dans chacune des races. / Within-breed genomic selection is now implemented in a number of large cattle breeds. However, building reference populations large enough remains a major challenge for smaller breeds. Combining reference populations and implementing a multi-breed approach appears to be an appealing alternative for small breeds.Such an approach requires conserved linkage disequilibrium across breeds to maintain the association between QTL and markers. Therefore, the use of a high density chip is generally needed. Error rates for high density imputation from medium density genotypes, classically used in cattle, were estimated. The mean error rate was below 1% in most dairy cattle breed which implies that a large high density imputed reference populations can be available for genomic selection at low cost. Reference populations from the three major French dairy breeds were imputed to high density and used to develop a multi-breed genomic evaluation.Several alternative genomic selection approaches (within-breed or multi-breed, based on medium or high density genotypes) were compared for breeds with different sizes of reference population. Improvement of genomic prediction accuracy due to the multi-breed evaluation was observed for breeds with 500 animals or less in their reference population. Accuracy of a third alternative using multi-breed evaluation based on medium density genotypes of closely related breeds was investigated. The benefit of multi-breed genomic evaluation was then three times higher in this situation than in the one where populations from major dairy cattle breeds were pooled. It can be noted that in both situations, using within-breed genomic information allowed a significant gain in accuracy compared to pedigree-based evaluations even for a small breed, when all sires of selection candidates belong to the reference population.These results suggest that implementation of genomic selection is feasible in small dairy cattle breeds. However further work is required to determine which optimal strategy can/must be implemented in a given breed.

No description available.

No description available.

No description available.