Evolutionary Algorithms (EA) are excellent at solving many types of problems but are inherently ill-suited to solving constrained problems. Previously there has been four ways to adapt these algorithms to solve constrained problems - pareto optimal strategies, modified representation and operators, penalty functions and repair strategies. This thesis makes significant contributions to the topic of genetic repair and introduces a non-Mendelian repair operator that has been inspired by a naturally occurring genetic repair mechanism in the Arabidopsis thaliana plant. Thus, the analogy between EA and natural evolution is extended to incorporate this (still highly controversial) biological repair process. The first and main objective focuses on Evolutionary Algorithms. This thesis adapts this novel genetic repair strategy to an EA to solve two benchmark constraint based problems - specifically permutation problems as this category of problem are often recognised as the most problematic problems for the canonical EA to deal with. The second objective was more biological, relating to Evolutionary Algorithms. A number of algorithmic and parametric interventions were made to the EA, to examine the repair algorithm’s performance under more biologically inspired conditions. This thesis illustrates that non-Mendelian ancestral repair templates outperform their Mendelian counterparts under a wide variety of conditions and also shows that under biologically inspired conditions, the non-Mendelian repair strategy continues to outperform its Mendelian counterpart.