Interação entre famílias e ambientes na fase inicial do melhoramento genético da cana-de-açúcar

Abstract

Sugarcane breeding is based on selection and cloning to obtain a superior genotype. The first stage, the seedling assessment trial, is characterized by a large number of genotypes where the population needs to be split into several environmental for logistic issues. Thus, this work aimed to understand the effects of the interaction between families and the environment in the first sugarcane breeding stage. The initial population consisted of 287 crosses evaluated in three locations: Araras-SP (CCA), Valparaíso-SP (EVA), and Ivinhema-MS (MS). The statistical design was Federer's augmented blocks design, in which two varieties, RB92579 and RB966928, were used as checks. Statistical analysis was done via mixed models in the R software, estimating the genetic parameters, such as coefficient of genetic variation (CVg), residual (CVr), heritability (h²), and the genetic correlation between locations for the Brix production (in tons) per Hectare (TBH). In addition, selection intensities (IS) of 10% and 40% were assumed to predict selection gains (Gs). For CCA, the genetic parameters were 0.38 (h²), 0.11 (CVg), and 0.13 (CVr); 0.37 (h²), 0.23 (CVg), and 0.30 (CVr) for MS; and 0.33 (h²), 0.16 (CVg) and 0.30 (CVr) for EVA. The genetic correlations were 0.22 (CCA-MS) and 0.34 (EVA-CCA and EVA-MS). Thus, the MS environment presented the greatest selection potential, and CCA showed the best experimental precision. Assuming the CCA family rank and the IS of 10%, the Gs was 0.99 TBH for CCA, but when extrapolating the classification for MS and VAS, the GS were 0.09 and 0.57, respectively. Analogously, when considering the classification obtained in the MS, the predicted Gs were 1.24 (MS), 0.40 (EVA), and 0.14 (CCA); for EVA classification, predicted Gs were 2.02 (EVA), 0.21 (MS) and 0.31 (CCA). Therefore, an evident complex interaction between families and environments was observed since the extrapolation of selected families is not efficient as the selection in the place itself. When considering an IS of 40%, the same considerations observed at 10% are observed. The main difference is that the Gs for the places the families have originally selected present a more pronounced reduction.