Optimalisatie van genetische selectie voor springprestaties bij Belgische Warmbloedpaarden

In the last century, the growing popularity of show jumping as a sport has increased the need for horses performing at the highest level in this discipline. As a consequence, this took Warmblood horse breeding to another dimension with the development of studbooks and breeding programs to breed better show jumpers. Despite a high genetic connectedness among Warmblood horse studbooks, they remain highly competitive to each other and developed their own genetic evaluation for show jumping performance using phenotypes applicable to competition data. However, no method to distinguish the intra- from the inter- competition variability has ever been set as the gold standard for genetic evaluation by Warmblood studbooks. Furthermore, the late availability and limited access to all level show jumping performances as well as little pedigree exchanges across studbooks hamper the genetic progress of the population. This PhD thesis aimed to optimize the genetic selection for show jumping performance in (Belgian) Warmblood horses (BWP) by addressing and resolving the previously cited limitations of nowadays selection. In a first study, we developed a novel show jumping performance phenotype applicable to any competition dataset: the adjusted fence height (AFH). This phenotype is based on a new non-arbitrary method that accounts for both the inter- (fence height) and intra-competition (ranking) variability using only the data. We assessed the choice of the phenotype and the importance of the rider effect in order to develop an optimal model to evaluate the genetic merit for show jumping performance. For that, genetic evaluations based on phenotypes often used in Warmblood horse breeding (Blom-transformed ranking and highest level achieved) were compared to genetic evaluations using AFH. Estimated heritabilities of Blom-transformed ranking, AFH and highest level achieved were 0.09, 0.12 and 0.39, respectively. The studied models were also moderately to highly genetically correlated (rg = 0.60-0.99). Moreover, the inclusion of a rider effect proved to substantially improve the models based on elementary performances. This study also showed that the best fitted model for AFH outperformed the other models by resulting in more accurate estimated breeding values available to breeders to make informed breeding decisions. Consequently, these findings suggest that a model based on AFH holds the potential to become the gold standard for show jumping performance and we recommend its implementation in (Belgian) Warmblood horse breeding programs. Warmblood studbooks developed linear scoring systems to assess jumping ability at a young age. In a second study, we assessed the potential of BWP's early life jumping (ELJ) traits scored during young horse contests as proxies for show jumping performance. Considering that the participation to these contests is on a voluntary basis, we also investigated the existence of a pre-selection that can negatively impact the development of reliable genetic evaluations. For that, we studied the association between participation to young horse contests and participation to show jumping competitions. The ELJ traits were lowly to moderately heritable (h²=0.05-0.30) but also moderately to highly correlated with AFH (rg = 0.37-0.63). These findings showed the value of ELJ traits as proxies for show jumping performance. Moreover, estimated heritabilities for start status in young horses contests were moderate to high (h²=0.56-0.65) but more importantly moderately to highly correlated with AFH (rg = 0.30-0.77) which proved the existence of a pre-selection in young horse contests. A stimulation of participations to young horse contests is encouraged to obtain a more representative sample of the population. This would therefore help to better optimize the genetic progress for show jumping performance by including ELJ traits in BWP breeding program. In a third study, ELJ traits were further looked at by developing optimal combinations of them to better predict genetic merit for show jumping performance. The use of proxies to predict genetic merit for a complex trait is indeed of interest in animal breeding. It can address the issues related to late availability, phenotyping costs and measurement difficulties of complex traits. The standard method to implement proxies in a breeding program is to include them in a multi-trait model with the trait of interest. However, in the big data era, multi-trait models require more and more powerful computing resources. In this study, a methodology was developed to identify optimal combinations of ELJ traits to predict genetic merit for show jumping performance in BWP horses to be used in genetic evaluations with a reduced number of traits (a combination of proxies and the trait of interest). For that, we regressed ELJ traits onto estimated breeding values for show jumping performance to obtain summarized early phenotypes using predictive modelling. These traits were lowly to moderately heritable (h²=0.10-0.26) and moderately to highly genetically correlated with AFH (rg=0.36-0.85). Moreover, the comparison between the two-trait models that included summarized ELJ traits and AFH and a multi-trait model showed their comparable efficiency. These findings showed the benefit of the implementation of summarized ELJ traits in BWP's breeding program and make the methodology developed an attractive alternative to large multi-trait models that could benefit animal breeding. It is known that the access to competition data especially at international level is not an easy task for Warmblood studbooks. They often only have access to national competition results which raised the question of whether show jumping performance at national level has a good predictive value for determining genetic merit for show jumping performance or if the use of international data (highest level) would be better to improve show jumping performance. In a fourth study, first the variance components for show jumping performance at various levels in Belgian Warmblood horses were investigated. Second, the benefit of the inclusion of show jumping performances at international level in the genetic evaluation for show jumping performance was assessed. Show jumping performance was lowly heritable and ranged from 0.04 (international level) to 0.14 (recreational level and all levels included). Estimated genetic correlations between performances at each level were high (rg=0.55-0.95). Moreover, this study proved that genetic evaluations including extreme levels only (recreational or international level) led to biased and less accurate estimated breeding values. On the contrary, the use of performance data at all levels (from recreational to international) helps to better predict genetic merit for show jumping performance and would be valuable for breeders to make informed breeding decisions. The high genetic connectedness between Warmblood studbooks due to exchanges of genetic material makes a joint genetic evaluation for show jumping performance possible and potentially beneficial for Warmblood studbooks. However, it has not been implemented in practice, partly due to difficulties to link horses to each other in pedigrees caused by a lack of unique identification numbers. In a fifth study, a generic methodology to integrate pedigrees based on string comparisons of horses' names was developed and applied to two Belgian Warmblood studbooks (BWP and Zangersheide). Genetic evaluations for AFH using single-studbook pedigrees, an uncorrected and concatenated version of the pedigree and the integrated pedigree were compared to assess the impact and benefits of a joint genetic evaluation for show jumping performance. The uncorrected version of the pedigree led to biased estimation of the variance components for show jumping performance but also biased and less accurate estimated breeding values. Moreover, the use of the integrated pedigree in the genetic evaluation compared to single-studbook pedigrees increased the estimated breeding values accuracy enabling for more accurate information available to breeders. We recommend the use of our methodology to integrate Warmblood studbook pedigrees and encourage the development of joint genetic evaluations that would benefit Warmblood horse breeding. In this PhD thesis, scientific evidence is provided to show that an optimization of genetic selection for show jumping performance in (Belgian) Warmblood horse breeding is within reach. An improved phenotype for show jumping performance that solved issues encountered when working with competition results has been developed and would be a good fit as a standard for show jumping performance. This PhD thesis also proposed the use of proxies to optimize the selection and developed a methodology to best combine them in a "summarized" proxy. Furthermore, the inclusion of show jumping performances at international level combined with performances at lower levels in the genetic evaluation for show jumping performance proved to be valuable and could help promote the exchanges of competition data. In addition to this, the methodology to integrate pedigrees that was developed in the last chapter offers a realistic opportunity for building an international evaluation for show jumping performance that would surely benefit all Warmblood studbooks.