Comparison of different selection methods in the seedling stage of sugarcane breeding

Abstract

Family selection is an important procedure to be considered in the early stage of sugarcane (Saccharum spp.) breeding. Different approaches are available, but few comparative studies are performed in practice. The aim of this study was to evaluate the potential genetic gain when different selection strategies at early sugarcane breeding stages are considered. Two experiments involving the first and second selection stages of the Sugarcane Breeding Program of RIDESA/UFSCar were performed. In the first stage, three selection methods based on the concept of selection between and within families were applied to predict the highest genetic gain, that is, BLUPi: simultaneously contemplates family and individual information for selection; BLUPis: promotes the dynamic allocation of individuals to be selected in each family; BLUP AUS : identifies high potential families and establishes differentiated selection intensities; additionally, mass and random selection methods were also performed. In the second stage, the selected clones were evaluated to compare the realized genetic gain. In the first stage, BLUP AUS had the highest predicted gain from selection (P GS; 12.7%) in tonnes of Pol per hectare (TPH). The BLUPis was highly correlated with BLUP AUS and was efficient. Moreover, BLUPi proved to be economically impracticable since phenotypic evaluations must be performed at the individual level. Family selection via BLUP AUS was equivalent to mass selection probably due to the low coefficient of genetic variation (CV g ≤ 15) among the families. However, the family selection strategy provides extra information for breeders that cannot be ignored; the possibility of studying the combining ability of genotypes for identifying promising parents for future cross combinations.

Novel Tools for Adjusting Spatial Variability in the Early Sugarcane Breeding Stage

“…The microenvironment, or the residual, can be controlled by experimental design and statistical methodologies. For example, linear mixed models increase the analysis precision due to the flexibility in modeling the field trial source of variations (Cursi et al, 2020(Cursi et al, , 2021Hoarau et al, 2021). Additionally, residuals can be controlled by collecting data from the field (electroconductivity, physical and chemical variables) to include in the statistical model.…”

“…These statistics are essential to effectively determine the genotypic value of the genetic material resulting from phenotypic evaluations ( Resende and Duarte, 2007 ). Considering EA1, only models 1.3 and 1.4 showed CVg above 10%, indicating genetic variability in the population for exploitation ( Cursi et al, 2020 ). Additionally, the models that account for the soil spatial variability pattern (models 1.3 and 1.4) showed high magnitude values ( H 2 above 0.60) ( Barreto et al, 2021 ).…”

“…In both areas, the experiments were implemented considering the family structure, which is widely adopted by different sugarcane breeding programs worldwide ( Jackson et al, 1995 ; Barbosa et al, 2004 ; Zhou et al, 2013 ; Cursi et al, 2020 ). Briefly, this structure consists of groups of related individuals from the same crossing (family) who share close genetic information in the same plot, i.e., families are our treatments during data analysis.…”