Genetic gain and yield performance of 42 genotypes of Coffea canephora in Brazil

Abstract

Coffea canephora is a species of major economic and social importance, and the identification of superior genotypes is essential for sustainable crop improvement. This study evaluated the performance of 42 C. canephora genotypes over six consecutive harvests (2016–2021) to estimate genetic parameters and assess the efficiency of multi‐harvest selection. The experiment was conducted in Nova Venécia, Espírito Santo, Brazil, using a randomized complete block design with three replications. Yield data were analyzed using restricted maximum likelihood/best linear unbiased prediction (REML/BLUP) mixed models. Significant genotypic, permanent environmental, and genotype × measurement interaction variance components were detected, supporting the use of repeated evaluations for reliable genetic inference. The broad‐sense heritability of the genotype mean was 0.596, while repeatability (0.409) and the mean genotypic correlation across harvests (0.602) indicated moderate temporal consistency in genotype performance. Temporal selection accuracy exceeded 0.70 from the third harvest onward, demonstrating the feasibility of early selection. Predicted genetic gains reached 19.16 bags ha ⁻¹ when selecting the single best genotype and remained substantial at 15.97, 14.11, and 12.49 bags ha ⁻¹ when selecting the top three, five, and eight genotypes, respectively, confirming the effectiveness of the REML/BLUP approach for identifying superior materials and accelerating genetic progress in C. canephora breeding programs.