Rajeev K Varshney FRS

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.

Targeting Induced Local Lesions in Genomes (TILLING) is considered a powerful reverse genetics approach for functional genomics studies. However, because of availability of low-cost and high-throughput sequencing technology, it has become possible to sequence TILLING lines and identify SNPs associated with genes responsible for traits. One TILLING population has been developed in the “Tifrunner” genotype of groundnut, an economically important oilseed crop grown in tropical and warm temperate regions of the world. The TILLING population has shown phenotypic variation for several traits including resistance to leaf spots and the features of prominent main stem. A total of 25 lines comprising of 16 susceptible and 9 resistant lines for leaf spots, and 11 lines with presence and 14 lines with absence of the prominent main stem from the TILLING population were sequenced on Illumina HiSeq 2500 and a total of 745.8 Gb sequencing data has been generated. These sequence data are being analyzed to identify structural variations including SNPs and INDELs across the lines with Tifrunner. In parallel, two mapping populations from these TILLING lines namely T47-7 (resistant to leaf spots) x T33-3 (susceptible to leaf spots) and T90-1 (presence of stem) x T71-2 (absence of stem) are being developed. It is planned to phenotype the segregating progenies and also sequence the extreme bulksA of segregating progenies for these traits. We anticipate identification of candidate genes and SNPs for these important traits by deploying the BSA-Seq approach in groundnut in due course.

Enhancing drought tolerance in chickpea, an important grain legume for people in the semi-arid regions of the world, is crucial for improving its productivity in the context of changing climatic scenarios. Terminal drought is one of the major constraints limiting chickpea production and causes up to 50% yield losses. Extensive genotypic and phenotypic data analyses revealed the presence of a QTL cluster on CaLG04 harbouring robust main-effect QTLs for 12 traits and explaining up to 58.20% phenotypic variation, referred as “QTL-hotspot”. To identify candidate genes related to drought tolerance, we narrowed down the “QTL-hotspot” region from ca. 3 Mb to ~300 kb by using a combination of bin mapping based QTL analysis and gene enrichment analysis. As a result, the “QTL-hotspot” region was split into two sub-regions, namely “QTL-hotspot_a” (139.22 kb; 15 genes) and “QTL-hotspot_b” (153.36 kb; 11 genes). Subsequently, we characterized the fine mapping population derived from ICC4958 (drought tolerant) and ICC1882 (drought sensitive) using different phenotyping platforms like LeasyScan, lysimeter and under field conditions. We identified major QTLs for canopy development, biomass, yield and water-use related traits co-localized in “QTL-hotspot” region, explaining up to 74% phenotypic variation. Our results provided crucial evidence of genetic linkages between traits phenotyped at multiple levels of plant organization, thereby increasing our cognizance of complex traits like drought. Analyses of genetic variation in the refined “QTL-hotspot” region among 3,000 diverse chickpea genomes identified mutations in five promising candidate genes that can be deployed in chickpea breeding programs and future sustainable agriculture.

The Genomic Open-Source Breeding informatics initiative (GOBii) is a Bill and Melinda Gates funded project with the mission to implement genomic and marker assisted selection as part of routine breeding programs for staple crops at CGIAR centers. We believe that much of the gains that have been achieved by major ag-biotech companies can also be achieved in these centers through adopting data management systems and bioinformatics pipelines that aid breeding decisions. Our challenge is to implement genomics data management and connect to breeding data management and analysis tools being developed as part of diverse projects and within different organizational structures. To achieve our goal, we have employed a global team of data curators, developers, molecular breeders and system administrators based at Cornell University, The Boyce Thompson Institute and at each of our collaborating CGIAR centers; CIMMYT, ICRISAT and IRRI. This team have experience and skill sets that cross multiple data management and curation projects with backgrounds in industry and academia and together can collaborate to find best solutions for use cases gathered in their own environments. We are partnering synergistically with adjacent data management and genotyping projects to prevent feature redundancy and promote the use of data management systems, and we are aligned with the Excellence in Breeding Program to ensure united approaches and goals. Our CGIAR center partners have now become the experts in managing and analyzing genomics and genotyping data and are training their own communities in using these systems. Together we have built Genomic Selection pipelines in Galaxy, data QC tools with Diversity Arrays Technologies and marker-assisted backcrossing, forward breeding and pedigree verification with the James Hutton Institute, based on use cases collected by our CGIAR molecular breeders. Together we are building a global community of knowledge surrounding best practices for implementing marker-assisted and genomic selection at CGIAR centers.

Peanut or groundnut (Arachis hypogaea), grown and consumed in several Asian and African countries in addition to Americas, plays an important role in providing daily nutritional requirement for large population of the world. Aflatoxin contamination and allergens are the major quality and food safety concerns across globe which adversely impact the global peanut trade and commerce. On the other hand, high oleic acid is an industry preferred trait for imparting increased shelf life to peanut-based products. Through precise phenotyping, genomics, transcriptomics and molecular breeding approaches, we are developing better understanding of these traits, conducting trait mapping and candidate gene discovery, and deploying molecular breeding for developing improved peanut varieties. For example, transcriptome analysis have identified several important candidate genes and pathways for three different types of resistance mechanisms of aflatoxin contamination namely in vitro seed colonization (IVSC), pre-harvest aflatoxin contamination (PAC), and aflatoxin production (AP). Further, genetic analysis of multi-parent advanced generation intercross (MAGIC) and genome-wide association study (GWAS) on a diverse association mapping panel are likely to provide associated genomic regions and candidate genes for aflatoxin contamination. Development and deployment of precise ELISA-based methods for quantifying five major and important peanut allergens (Ara h 1, Ara h 2, Ara h 3, Ara h 6 and Ara h 8) have led to the identification of several hypoallergenic lines. Subsequently sequence/GWAS analysis is likely to identify the alleles responsible for making peanut, hypo or hyper allergenic. Allele-specific genetic markers were successfully deployed for developing several high oleic molecular breeding lines in multiple genetic backgrounds. Many of these lines are in final year of testing in India and are most likely to get released in 2019 for cultivation. Identification and development of improved peanut lines with combination of these nutritionally important and oil quality traits are likely to enhance the consumption and international trade of peanut.

The wild accessions of chickpea have huge genetic diversity which can be exploited for increasing genetic gains in the crop. Several large scale re-sequencing efforts have been carried out to identify variants in chickpea, but the complete genetic repertoire has not been captured as these efforts involve mapping of reads on single chickpea reference genome. We report the construction of chickpea pangenome by sequencing and de novo assembly of accessions from eight different annual chickpea wild species. These accessions were sequenced at ~180X coverage using Illumina platform, generating a total of 1.14 Tbp data. The de novo assemblies for these species resulted in size varying from 512.3 Mbp to 927.0 Mbp with high N50 values. Comparisons of these assemblies identified genes exhibiting copy number variations and presence absence variations, some of which show evidence of positive selection and might have association with important agronomic traits. The chickpea pangenome will serve as a valuable genomics resource and will have broad implications in chickpea breeding.