Global and mid-west U.S. agriculture requires diversification and new sources of protein for sustainable crop production. Mungbean as a legume crop has a long cultivation history in Asia; however, its potential lays hitherto underexplored in the U.S. As a first step towards exploring mungbean for crop diversification in northern latitudes, crop germplasm centers that harbor worldwide crop resources need to be utilized. This study explores the potential of mungbean in the U.S. northern latitudes through the utilization of the United States Department of Agriculture (USDA) germplasm resources. Complete collection of USDA mungbean germplasm was screened under Iowa field conditions in 2017, to shortlist 482 accessions to create an Iowa mungbean panel. The Iowa mungbean panel was further characterized for field adaptability traits in 2018 and 2019 and genotyped using genotype-by-sequencing (GBS) to conduct association mapping of important traits. Genetic markers were identified for both quantitative traits - days to flowering (DTF), plant height (PHT), leaf-drop at maturity (LDMS), 100-seed weight (SDWT) and Fusarium wilt score (WS), and qualitative traits; seed color (SC), seed-coat texture (ST), hypocotyl color (HC), and pod-color (PC). We report FERONIA, a known flowering-pathway gene as the candidate gene for the QTL with largest effect on DTF. In addition, important epistatic interactions were also uncovered for WS and SDWT. Further, accessions with desirable magnitude of traits were identified as potential parents. Diversity analyses and field phenotypic data indicates potential for mungbean improvement to suit mid-western U.S. cultivation.

Plant material and planting: Mungbean accessions were obtained in 2017 from the United States Department of Agriculture, Plant Genetic Resources Conservation Unit, Griffin Georgia. These accessions were sourced from at least 42 countries of the world (Supplementary Table S1). In 2017, ~3000 accessions of mungbean were planted in an evaluation trial near Ames, IA, to begin germplasm evaluation with the intent to make targeted hybridizations to initiate a grain legume mungbean breeding program. Ten seeds were hand planted on June 6th, 2017 in, single-row plots with 30" between adjacent rows, 1"1-inch deep furrows with seed-to-seed distance at 3". Mungbean plots were observed throughout the season, and data was collected on traits such as, emergence, stand count, days to flowering (DTF), disease susceptibility and shattering tendency. Due to large variations in maturity, plants were hand-harvested as plots matured over a period of 2-3 weeks in the Fall season. Soybean single-plant belt thresher (ALMACO, Nevada, US) was used to thresh pods. Based on DTF and other critical agronomic traits data, 482 mung lines were selected for further evaluation in the subsequent year. Supplementary Figure S1 shows the geographic distribution of these selected accessions.

In 2018, due to storage issues, seeds were only available for 445 accessions. In 2018, these 445 mungbean accessions, were grown at two locations in Iowa near Ames (42.0308° N, 93.6319° W) and near Boone (42.0597° N, 93.8802° W) in a randomized complete block design (RCBD) with two blocks/reps per location. Locations were picked for their proximity to Iowa State University and centralized location in the state. Genotypes were mechanically planted using ALMACO four-row planter (ALMACO, Nevada, US) at depth of 1", in two-row plots (7' long) with 50 seeds per row. Row to row distance was 30". In Fall 2018, all accessions were genetically purified in the greenhouse using single-seed descent method and sent to winter-nursery for seed increase. Increased seeds were brought back in the spring. In 2019, 482 accessions were grown in Iowa near Ames (42.0308° N, 93.6319° W), and Muscatine (41.4245° N, 91.0432° W) in randomized complete block design (RCBD) with two blocks/reps per location. Genotypes were mechanically planted using ALMACO four-row planter (ALMACO, Nevada, US) at depth of 1", in four-row plots (7' long) with 25 seeds per. Row to row distance was 15". Row to row distance was reduced for weed suppression under low-input systems particularly hand weeding to avoid any chemical injury due to herbicide on juvenile plants. In-season application of Select Max, Basagran, and Ultra-Blazer was used to control latter flushes of weeds. No adverse effect of herbicide was noted on the crop plants.

DNA marker development: A pilot study was used to determine appropriate strategy for library creation. In the pilot study, we compared double- (Btg1-Taq1) and single-digest (ApeK1) restriction fragmentation for GBS library creation using 8 mungbean accessions selected on the basis of phenotypic diversity. For an average of two million reads-per-sample sequencing depth, double-digest libraries yielded 9824 polymorphic loci, as compared to 16675 loci yielded by single-digest. Results of pilot experiment showed benefits of using ApeK1 for full-scale, therefore ApeK1 was selected for library generation with an aim to generate 2.5 million reads per sample. Plant leaf tissues for all samples were collected by growing genotypes in the growth chamber. In total, 49 duals indexed GBS libraries were generated using ApeKI. This number included 482 USDA accessions and 17 additional non-USDA mungbean genotypes (commercial varieties and improved germplasm obtained from Asian Vegetable Research and Development Center, Taiwan). All libraries were combined into a single pool and sequenced using 2 lanes of a NovaSeq 1x100-bp run. Approx. ≥1,500M reads were generated for the pool. Variants were called using freebayes approach (Garrison and Marth, 2012). One genotype was dropped from further analyses due to large proportion of missing marker data.

Phenotypic data collection:

Days to flowering (DTF): First day of flowering was recorded as days-to-flowering.

Plant height (PHT): Plant height was measured (in cm) from base of the main stem representing the plot to the top-most node at end-of-the-season.

Leaf drop at maturity score (LDMS): LDMS was scored at the end-of-the-season. LDMS was scored on a 1 to 5 scale. Using visual estimation, a score of “1” denoted > 90% leaf-drop; “2” denoted 60-90% leaf-drop; “3” denoted 30-60% leaf-drop; “4” denoted 5-30% leaf-drop; score of “5” denoted that < 5 % leaves had dropped.

Fusarium Wilt score (WS): Field data was recorded as wilt score (WS) 1 to 5, ranging from highly resistant (1), resistant (2), moderately resistant (3), susceptible (4), highly susceptible (5). The 1-5 scores represented the aggregate of incidence and severity. Plots were scored as “1” if no plants were found wilted, dead and when observed no stem lesions were found at the soil level; “2” if stem lesions were observed but no wilted or dead plants were present; “3” if stem lesions were observed with few (1-10 %) mildly wilted plants; “4” if both stem lesions and up to 20 % plants showed severe wilting and/or death; “5”, when 20 to 100 % of the plants were dead.

100-seed wt. (SDWT): measured as sample weight of 100 harvested seeds in grams.

Qualitative traits scored on the Iowa mungbean panel were: pod color (PC), hypocotyl color (HC), seed color (SC) and seed texture (ST). The seed color was rated on a scale of 1-5, where 1 = green, 2 = yellow, 3= brown, 4 = black, and 5 = speckled. Seed texture was rated on a binary scale of 1 = matt and 2 = shiny, and hypocotyl color was rated as 1= purple and 2= green.