Detecting and managing suspected admixture and genetic drift in domestic livestock : modern Dexter cattle : a case study

• Main Author: Bray, Timothy C.
• Published: Cardiff University 2008
• Subjects: 636.08
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.584336

This study combines a range of contemporary genetic analysis methods to analyse the Dexter cattle breed in conjunction with the development of a novel method of admixture determination. The Dexter was chosen for its heterogeneous genetic composition due to a complex population history. Comparison against other European cattle breeds showed the Dexter to be one of the most diverse breeds and clearly distinguishable from other breed populations. The levels of migrant individuals exchanged between the Dexter and other European breeds was seen to be in the middle of the range for all breeds, as was the conservation value of the Dexter as determined through the Weitzman genetic distance approach. The Dexter was shown to stand out from other European cattle breeds due to high levels of subdivision into different regions of the herd book. The hypothesis that the ancestry of subdivisions was entirely responsible for this genetic divergence could not be proven. The quantification of admixture proportions were made for two putative ancestral representative breeds, Red Devon and Kerry. It was found that a selection of carefully chosen Traditional Dexter individuals were more closely related to the Kerry breed. Admixture contributions for remaining breed populations were inconclusive with the exception of a small sample group representing the breed in America which demonstrated a higher Red Devon contribution. Genetic drift is heavily implicated in the results shown and it is notable that high levels of variance were associated with admixture contributions. An approximate Bayesian computation approach was designed and developed to better model the admixture scenario of interest. A method allowing for two admixture events was constructed in order to calculate parental contributions and compare them to simulated datasets according to a genetic model. Initial testing proved successful using a single admixture event. The addition of a second admixture event reduced the accuracy of the method. Testing scenarios of up to half a million simulations with nine loci were unable to successfully quantify either simulated or real admixture events here. Testing suggests that the effectiveness of the approach is thought to increase with numbers of simulated datasets used. Recommendations for the successful application of the method are made.