Efficacy of early generation selection in a sugarcane improvement program

Family selection is an effective and cost efficient method in sugarcane improvement. Data are now available from a BSES selection program that shows its value. Combined data from families tested as stage 2 (4-sett plots) from 1989 to 1991 in Bundaberg showed that the selection rate of elite clones from stage 3 to stage 4 improved from 2.8%-7.4% in low Net Merit Grade (NMG) families «6) to 13.1%-17.3% in high NMG families. Similar data are available for family selection in original seedlings (stage 1). This showed that selection rate from stages 3 to 4 increased from 3.8% to 13.6% as family mean NMG increased from 9 to >13. Family data have also been used to estimate the breeding value of parents using the statistical technique, Best Linear Unbiased Prediction (BLUP). BLUP estimates based on NMG for parents of families harvested from 1993 to 1996 were correlated with actual family performance in 1995, 1996, and 1997. For 81-172 families with BLUP estimates for both parents, the correlations were 0.62--0.65 compared with 0.45-0.50 for the previous system for estimating breeding value. This shows that BLUPs are an effective tool for predicting family performance. In stage 3 of selection, large numbers of clones are tested in unreplicated trials. While there are large errors associated with this stage, evaluation of NMG data from stage 4 trial series (5 locations) showed a good linear relationship between NMG in stages 3 and 4. The frequency of clones with mean stage 4 NMG 2: 10 increased from 1.6% to 20.0% as stage 3 NMG increased from <10 to 2: 14. While further improvement in the effectiveness of selection at this early stage is paramount, these data demonstrate that current methods are resulting in continued genetic gain.