Investigation of genetic contributions to behaviour in Thoroughbred horses

Millions of years evolution shaped the modern horse, Equus caballus. A key evolutionary trend was an expanded neocortex, part of the brain that processes sensory information. The horse was domesticated after other companion animals which suggests behavioural unsuitability to a captive lifestyle. Nonetheless, the past ~5500 years have seen humans utilise horses for many purposes such as meat, war, and transport. Ancient DNA has revealed that selection for genes contributing to behavioural flexibility have enabled adaption to the domestic setting. Evidence suggests individuals with greater behavioural flexibility may fare better in the domestic setting. The Thoroughbred is a product of >300 years of artificial selection targeting athletic performance traits and only half of foals born each year reach the racetrack. Underlying a percentage of these individuals may be a behavioural unsuitability. Trainers observe behaviour/training progression daily and modify training regimen accordingly. Anecdotally Thoroughbred behaviour is believed to have familial basis. Recent genome-wide approaches identified genes with neurobiological functions associated with racing durability and precocity. Heritability and genetic contributions to athletic performance and health traits have been the focus of many studies and to date limited research has targeted underlying genetic variation contributing to behavioural traits. Behavioural traits are known in other species to be shaped by physiological systems and also environmental stimuli, life experience and genetic makeup of an individual. Behavioural assessment using handler-completed questionnaires is well-established in equine research, and has been used solely, and in conjunction with objective measures. In this thesis objective and subjective measures of stress response were used to test the primary hypothesis that there are underling genetic contributions to behavioural traits in the Thoroughbred. In Chapter 2 saliva samples were collected from n=96 Thoroughbreds before and ~30 mins after early training milestones. Results showed significant differences in mean concentration, concurrent with previous studies that training events are stressful. This demonstrated that this method can be used to non-invasively detect stress response. The same n=96 Thoroughbreds were prospectively assessed for coping. Together with % change cortisol response to first backing for n=34 Thoroughbreds from Chapter 2, Chapter 3 tested the hypothesis that handler perceived behavioural differences correlate with an objective measure of behaviour. No significant association was found between cortisol and perceived coping. These findings indicate that handlers cannot detect cortisol response differences, and therefore the objective measure identifies an invisible behavioural phenotype. In Chapter 4 the saliva response (% change) phenotype was used to investigate selection signals underlying behavioural traits in the Thoroughbred. For the first time a composite selection signals (CSS) approach using genome-wide genetic markers was used with a behavioural phenotype in the horse. Significant selection signals were detected at eleven genomic regions with overrepresentation of behaviour and nervous system development genes identified. In Chapter 5 perceived coping was applied to CSS analysis in a test set of n=96 Thoroughbreds unbiased for racing performance. Selection signals at thirteen regions of interest were identified with overrepresentation of GABA receptor signalling and neuroinflammation genes. For the first time, this study has shown that subjective phenotypes derived from handler and trainers’ observations of stress response are able to detect selection signatures within the genome of the horse. This thesis presents two stress response phenotypes, which when engaged by genome-wide CSS analyses identified divergent gene sets with overlapping neurobiological themes.