Uma década de melhoramento do arroz de terras altas em Minas Gerais: do progresso genético aos impactos da interação genótipos × ambientes

Abstract

The work brings together two complementary studies on upland rice breeding, based on a decade of data from the UFLA Genetic Breeding Program. We analyzed Value for Cultivation and Use (VCU) trials conducted between 2014/15 and 2023/24 across different locations in Minas Gerais, encompassing 74 elite lines evaluated in 26 experimental environments. The first study, using mixed models (REML/BLUP), quantified genetic progress in grain yield: the accumulated gain was ~430 kg ha⁻¹, equivalent to a 13% increase over the program’s baseline. Decomposition indicated that 62% of the gain arose from within-program advancement of superior lines, and 38% from the introduction of new genotypes, evidencing a balanced strategy between exploiting variability and renewing germplasm. Heritability of genotype means increased from ~0.20 to ~0.85, and selection accuracy from 0.45–0.67 to 0.92. The genotype × environment (GE) interaction accounted for 21–83% of phenotypic variance. The second study employed 26 quantitative environmental covariates for stratification. Climatic similarity (Environmental Kinship, KE) identified four mega-environments with significant yield differences among groups (p = 0.003; η² = 0.411). Principal component analysis explained 77.3% of environmental variation: PC1 was associated with water availability and PC2 with thermal conditions. In parallel, the GGE biplot arranged environments into three groups oriented by genotypic response, without consolidated statistical differences (p = 0.323). Global concordance between methods was moderate (42.3%), higher in ME2 (71.4%). Taken together, the findings indicate interannual variability as the dominant driver of GE, repositioning locations in environmental space. Lines G40, G47, G57, and G63 combined high yield and stability and remained under evaluation through the 2023/24 season. Practically, the results support integrating envirotyping and GGE biplot to guide strategic trial allocation, maintaining renewal rates around 30–35%, and applying stability indices within each mega-environment. These approaches provide a robust methodological basis for quantifying genetic progress and environmental stratification, with potential extension to other crops and regions.