COMPARATIVE ANALYSIS OF PHENOTYPIC AND GENOMIC SELECTION STRATEGIES IN EARLY GENERATIONS OF MAIZE BREEDING PROGRAMS: A STOCHATIC SIMULATION APROACH

Plant breeding programs aim to achieve higher yields and genetic gains with each generation. New technologies and cost-effective strategies are vital for refining these methods. Stochastic simulations are valuable tools, enabling informed decision-making while reducing financial risks and time loss. This study used R software to simulate maize line development over 40 years, divided into 20 years of burn-in and 20 years of breeding. We evaluated phenotypic selection (PS) and genomic selection (GS) under two heritability levels (30% and 70%), creating four scenarios. The objective was to identify the scenario with the highest gains per unit time by analyzing selection accuracy, mean genetic value, and genetic variance over time. A maize founder population with 30,000 SNPs and 300 QTLs per chromosome was used, incorporating additive, dominance, and environmental effects. Both PS and GS were tested through reciprocal recurrent selection, using the doubled haploid strategy to obtain inbred lines. In the GS scenario, a mixed model selected lines by estimated breeding value, while PS relied on phenotype. Results showed GS outperformed PS in genetic gain (32.32% for H2=0.3 and 30.96% for H2=0.7) but led to a greater reduction in genetic variance. Simulations proved effective for identifying trends and refining strategies efficiently.