A software application for parental selection in crossing stage at Chacra Experimental sugarcane breeding station.

UCHINO, D.; ULLIVARRI, R.; AGUATE, F.; BALZARINI, M.; CASTELLANOS, R.

Saint-Gilles Réunion Island

Workshop; 11th Germplasm & breeding 8th molecular biology ISSCT workshop; 2015

Parental selection in a sugarcane breeding program should take into account multiple traits. In order to improve sugar and alcohol production, parentals of high yield varieties with good disease ratings are preferred to make new crossings. Particularly in subtropical areas with frosts, precocity is also an important trait to evaluate new hybrids. Using data from CHACRA Breeding Program, was developed an easy-to-use software application for parental selection to identify crossings with high expected genetic value for a given trait. To set up the algorithm we used CHACRA data for the 1997-2014 period corresponding to the first two clonal stages and late variety trials. The statistical protocol involved calculation of the Best Linear Unbiased Predictor (BLUP) for male and female effects of each parental from historical data of the derived progeny. For this purpose, a mixed linear model with fixed effect of crossing series and random effects of female, male and their interactions was run for each clonal stage and BLUP of parental effects were obtained. The BLUP of kg.Brix.ha-1 (TBxH), tons.cane.ha-1 (TCH), early Brix (EBx), average Brix (Bx) of consecutive samplings taken before the period with frosts in the clonal stages and four times during the crop cycle in the variety trials, were calculated. It was also performed a variance component analysis to weigth information from the different breeding stages for parental selection. According to these analysis were selected parentals with high EBx (> 95th percentile) in first clonal, high TBxH, TCH and Bx in the second clonal stage; and those that produced varieties with good performance in the variety trials in advanced breeding stage. BLUP values of the selected parentals were then used to obtain a Predictor of the Crossing Value (PCV) for all potential crosses between the selected parentals. Correlation between PCV and historical crossing data adjusted for series effect were significant and high. The parental selection algorithm and PCV calculations were compiled in an interactive web application using package with R [1]. By using this software application, combining historical information was straightforward.