Genome Based Selection of Traits with Different Complexity in Chickpea

Marker-assisted selection (MAS) for relatively simple traits such as herbicide resistance and growth habit has played an important role in chickpea breeding at the Crop Development Centre, University of Saskatchewan. This MAS program allows selection for desirable breeding lines at early generation, and, thus, only selected lines will continue for agronomic evaluation at later generations. The use of MAS for improving more complex traits, however, had very limited success.  Current trend indicates progress in plant breeding programs will be limited by phenotyping capacity, not by genotyping capacity, thus increasing the effectiveness of selection strategy is increasingly critical. Genomic selection (GS) has become an important method of applying molecular markers for selection of complex traits in plant breeding programs. We examined the prospect of genome selection (GS) in chickpea breeding and evaluate the effect of marker selection on genomic prediction accuracy. A panel of 281 breeding lines were genotyped by GBS (genotyping by sequencing) generating a total of 4,524 SNPs with minor allele frequency (MAF) of >0.05. Best linear unbiased estimates (BLUP) were calculated and input into the genomic prediction model with only additive effects. The genotype effect was treated as random. The effect of training population size on prediction accuracy was evaluated using a validation set comprised of randomly selected breeding lines and training sets of variable sizes (10-90% of total population). The validation set was created by randomly sampling breeding lines without replacement. Prediction accuracy was found to increase with the increase in size of training population. For seed weight as an example, the accuracy increased from 70% to 80% when the size of the training population increased from 28 to 84 lines; further increase in the training population size did not significantly increase the accuracy. The correlation accuracy with varying training population sizes for traits with different complexity (days to flowering, days to maturity, resistance to ascochyta blight, seed iron concentration and grain yield) ranged from 0.20 for seed iron concentration to 0.80 for 1000 seed weight. Comparison of genetic gain from targeted assisted selection and GS for early flowering will be discussed.