Robust optimal contribution selection, accounting for uncertainty

Managing the conversion of genetic variation into genetic gain is crucial for the long-term sustainability of breed-ing programs. Standard optimal contribution selection (OCS) balances short-term genetic gain with long-termgenetic variation within the population by balancing the expected response to selection against the expected groupcoancestry weighted by the parameter λ. However, OCS assumes that the selection criterion is known without er-ror, which is not the case when using phenotypic values or estimated breeding values (EBV). The aim of this work is to evaluate how to take into account the uncertainty and correlation of EBV in OCS. To address this issue, weused robust optimization when solving for OCS. The approach extends the standard OCS into the robust OCS byincluding the square root of the conditional variance of the EBV weighted by the optimised contributions, giving the standard deviation of the estimated genetic gain (’group uncertainty’), and κ parameter that controls emphasison the ’group uncertainty’. To demonstrate the difference between standard and robust OCS, we simulated a selective breeding program in which we obtained the true and estimated breeding value alongside their conditional covariance and genetic variation. We compared how genetic gain, group coancestry and group uncertainty, andcorresponding optimised contributions changed as we varied the parameter κ while λ was fixed. The preliminary<br/>results suggest that there is a non-zero correlations between the group uncertainty and group coancestry. We willdemonstrate how this correlation changes as we vary the κ parameter and how it affects the optimised contributions, genetic gain, and group coancestry in short- and long-term