Murdoch University Research Repository

Welcome to the Murdoch University Research Repository

The Murdoch University Research Repository is an open access digital collection of research
created by Murdoch University staff, researchers and postgraduate students.

Learn more

Genomic prediction to accelerate the rate of genetic gain in Chickpea for providing nutritional food security

Roorkiwal, M., Chellapilla, B., Santantonio, N., Jarquin, D., Samineni, S., Jones, E., Robbins, K., Singh, N.P., Crossa, J. and Varshney, R.K.ORCID: 0000-0002-4562-9131 (2020) Genomic prediction to accelerate the rate of genetic gain in Chickpea for providing nutritional food security. In: Plant & Animal Genome Conference XXVIII, 11 - 15 January 2020, San Diego, CA


Chickpea (Cicer arietinum) is the second most important food legume globally, which plays a key role in ensuring the nutritional food security. Average chickpea productivity has been restricted to ~ 1 t h-1 due to several biotic and abiotic stresses. Prolonged use of conventional breeding approaches have started to fall short of meeting the yield and nutrition demands. To address the issues related to complex traits such as yield which is controlled by multiple QTLs, genomic selection (GS) approach can be very useful in crop breeding to capture several genes with minor additive effects. GS offers breeders to select lines prior to field phenotyping using genotyping data, resulting in reduced cost and shortening of selection cycles. Initial results on GS in chickpea using 320 elite breeding lines suggested high prediction accuracies for diverse yield and yield related traits. Inclusion of G x E effects in GS models has shown significant improvement of prediction accuracies in breeding programs. In order to assess the potential of GS in chickpea breeding program 6000 F5 lines form 12 different crosses from ICRISAT and IARI breeding programs were selected and genotyped using DArTseq platform. After merging the markers from the training and prediction sets, which were run on different DArT marker platforms (DArTseq and LD DArT), about a thousand markers were used to run the prediction models. The cross validation prediction accuracy were run with a 10-fold consolidation scheme. Each cross validation was repeated 10 times with new random folds, and the mean of the prediction accuracies was calculated. To compare the potential of GS models, two set of ~200 lines each were identified based on visual selection by breeder and based on genomic prediction based GEBVs. Both of these set were evaluated in the field conditions during 2018-19. Selection efficiency of GS over visual phenotypic selection was found significantly better. Genomic prediction based line selection over visual selection saves time and cost involved in large scale screening of populations.

Item Type: Conference Paper
Conference Website:
Item Control Page Item Control Page