Phenotyping of the Mapping Population and Statistical Analysis

F₁₄₋₁₉ generations of the 432 RILs together with the parents were grown during the six seasons, namely, rainy 2014 (S1), rainy 2015 (S2), rainy 2016 (S3), rainy 2017 (S4), rainy 2018 (S5), and post-rainy 2018 (S6) at IABT garden (E115) of Main Agricultural Research Station, University of Agricultural Sciences, Dharwad, India (Figure 1). During each season, the RILs were grown in two replications with a spacing of 30 × 10 cm with recommended agronomic practices. The observations were recorded on productivity traits, such as the number of pods per plant (NPPP), pod weight per plant (PWPP), shelling percentage (SP), and test weight (TW), and on quality traits, such as protein content (PC), oil content (OIL), OLE, linoleic acid content (LIN), and oleic to linoleic acid ratio (O/L). The quality parameters were estimated using the near-infrared reflectance spectroscopy (NIRS) (Model XDS RCA, FOSS Analytical AB, Sweden, Denmark) at ICRISAT, Patancheru, India.

Open in a separate window

Figure 1

Flow chart of genotyping, high density genetic map construction, multi-season phenotyping, identification of genomic regions for productivity, and quality traits and their validation.

An ANOVA was performed for each trait observed during each season to test the significant differences among the RILs. A pooled analysis of variance was performed for all the traits across the seasons allowing G × E interactions. Phenotypic coefficient of variation (PCV), genotypic coefficient of variation (GCV), and broad sense heritability (h²b.s) were estimated using the plant breeding package Windostat ver. 8.5 (Indostat Services, Hyderabad, India, https://www.indostat.org/agriculture.html). Pearson's correlation coefficients (r) among the different traits were estimated over the seasons using the 16th version of SPSS (SPSS Inc., Chicago, IL, USA).

DNA Extraction and Genotyping With AhTE and SSR Markers

DNA was isolated from the young leaves of each RIL and the parents following the modified cetyltrimethyl-ammonium bromide (CTAB) method as described by Cuc et al. (2008). DNA quality of each sample was checked on 0.8% agarose gel. Furthermore, DNA quantification was done using Nano-Drop (UV technologies, Wilmington, DE, USA), and DNA concentration was normalized to ~5–10 ng/μl for genotyping the parents and the RIL population using AhTE and SSR markers. In total, 343 AhTE markers (Gayathri et al., 2018) and 91 SSR markers (as shown in Pandey et al., 2012) were screened for parental polymorphism between TMV 2 and TMV 2-NLM. Subsequently, the markers polymorphic between the parents were identified and used to genotype the RILs (Figure 1). PCR and separation of the amplicons and scoring of the alleles were performed as described by Kolekar et al. (2016). Genotypic data on 105 AhTE markers generated by Hake et al. (2017) on these RILs were also employed for genetic mapping.

GBS of the RILs, Sequence Analysis and SNP Calling

Genotyping-by-sequencing was performed for the RILs and their parents as described by Dodia et al. (2019). To perform GBS, 10 ng DNA from each RIL was digested using the restriction endonuclease enzyme ApeKI that recognizes the site G/CWCG. The ligation enzyme, T4 ligase, was used to ligate the digested products with uniquely barcoded adapters. Such digestion and ligation were performed for each RIL, and an equal proportion of the products from each sample was mixed to construct the libraries. These libraries were amplified and purified to remove the excess adapters. They were sequenced on HiSeq 2500 platform (Illumina Inc., San Diego, CA, USA) to generate genome-wide sequence reads.

The sequence reads for the parents and the RILs were obtained as FASTQ files, which were used for SNP discovery using TASSEL version 5.2 (Bradbury et al., 2007) (Supplementary Figure 1). Initially, the perfectly matched barcodes were detected with four bases remnants of the digestion site of the restriction enzyme in the sequencing reads generated for RILs and parental genotypes. Reads were sorted and de-multiplexed using the barcodes. They were trimmed for the first 64 bases starting from the recognition site of the restriction enzyme. Reads containing “N” within the first 64 bases were identified and discarded. Reads passing the quality filtering criteria were mapped onto the reference genome of cultivated peanut A. hypogaea (Bertioli et al., 2019) using the Burrows-Wheeler Alignment (BWA) tool (Li and Durbin, 2009). The mapped reads were exported in the form of Sequence Alignment Map (SAM) file. Furthermore, the alignment file was processed for SNP calling using SNP caller plugin implemented in TASSEL version 5.2.0 GBS v2 pipeline as per the standard instruction (https://bitbucket.org/tasseladmin/tassel-5-source/wiki/Tassel5GBSv2Pipeline). The RILs with <85 Mb data were not processed for further analysis to avoid false-positives. The SNPs with more than 50% missing data and minor allele frequency (MAF) of ≤ 0.3 were filtered out to avoid the noise during genetic map construction. The SNPs with <50% missing data for the RILs were imputed using Beagle version 5.2 (Browning et al., 2018) algorithm. Furthermore, filtering was performed to check the percentage heterozygosity and polymorphic SNPs between the parents (Supplementary Figure 1).

Genetic Map Construction

High-quality SNPs obtained after filtering were further considered for genetic analysis. A chi-square (χ²) test was applied on polymorphic AhTE, SSR, and SNP markers with a null hypothesis that two alleles from both parents of RIL population at a particular locus segregate in a 1:1 ratio. The markers showing high segregation distortion (χ²-test, P < 0.001) were filtered out and not considered for the linkage map construction. The genetic map was constructed using JoinMap (version 4.0) (Van Ooijen, 2006) with logarithm of odds (LOD) threshold ranging from 3.0 to 10.0 and a minimum recombination threshold of 45%. Grouping and ordering of the markers were performed using the regression mapping algorithm. Kosambi map function (Kosambi, 1943) was used for genetic map construction, and to convert the recombination frequencies into map distances in centiMorgans (cM). Chromosome-wise marker positions with their respective names were used to draw the final genetic map using MapChart (Voorrips, 2002). The mapped markers were also analyzed for their genic and non-genic location and functional annotation (especially for the SNP and AhTE markers).

Main-Effect and Epistatic Quantitative Trait Locus Analysis

The main-effect quantitative trait locus (QTL) analysis was carried out using a “composite interval mapping (CIM)” approach (Zeng, 1994) with Model 6 and scanning distance of 1.0 cM between markers and moving window size of 10.0 cM using Windows QTL Cartographer version 2.5 (Wang et al., 2007). A forward–backward stepwise regression method was used to set the marker cofactors for the background selection. The highest peak was considered to locate QTL where the distance between the peak and the QTL was <5.0 cM. Permutation (1,000) test was performed to work out the threshold and identify the significant QTL. The QTLs with >3.0 LOD and phenotypic variance explained (PVE) >10% were considered as major effect QTLs for a particular trait. Those with PVE <10% were considered as minor effect QTLs. Based on the trait name and chromosome number, the QTLs were named, where the first letter “q” indicated the QTL and the abbreviated capital letters indicated the trait followed by chromosome number and the numerical number indicating the serial number of the QTL for a trait. For instance, qPC-Ah16-1 was the first QTL for PC detected on chromosome Ah16.

Analysis for the epistatic QTL (epiQTL) (Q × Q) was conducted using the function “two-dimensional scanning ICIM-EPI” implemented in inclusive composite interval mapping (ICIM) software version 4.1 (Wang et al., 2014) with 5 cM step and 0.001 probability mapping parameters in stepwise regression. The minimum threshold LOD value for significant epiQTL was set at 3.0.

Single Marker Analysis

Association of the markers with the productivity and quality traits was tested by SMA using the lm() function (linear regression) of the R program.

Putative Gene Discovery From Major QTLs/SMA

Putative genes were identified for the major QTLs or QTL clusters. The region between the flanking markers of a particular QTL on the physical map was considered for candidate gene discovery. Where the physical distance between the two flanking markers was more, the marker closer to the peak was selected, and a physical distance of 5 Mb (toward the QTL peak) was searched for the candidate genes. Also, those markers which were identified to be significantly associated with the traits by SMA were checked for their location (genic and non-genic) and effect at PeanutBase (www.peanutbase.com).

GBS Based High-Density Genetic Map

A total of 1,067.58 million raw reads (100.87 GB) were obtained for the 403 RILs and the two parents. On average, 2.42 million reads (0.23 GB) data were generated for each sample. After filtering, a total of 978.96 million reads (82.9 GB) were mapped onto the tetraploid reference genome of cultivated tetraploid peanut “Tifrunner” (Bertioli et al., 2019). In total, 47,584 raw SNPs (mean read depth of 73.7) were extracted for the downstream analysis. Out of these SNPs, 1,205 polymorphic SNPs were identified between the parental genotypes (TMV 2 and TMV 2-NLM). Further filtering based on missing data and segregation distortion identified 713 SNPs high-quality polymorphic SNPs. The missing data for these SNPs ranging from 0.01 (1%) to 0.489 (48.9%) (Supplementary Table 4).

Overall genotypic data available for mapping included 865 markers; comprising of 713 SNPs, 143 AhTEs (105 from Hake et al., 2017 and 38 from this study), and 9 SSRs (Supplementary Table 5). The polymorphism percentage for the SNP, AhTE, and SSR markers were 1.49% (713/47,584), 20.08% (143/712), and 9.89% (9/91), respectively. Of these 865 markers, a total of 700 (including 553 SNPs, 136 AhTEs, and 8 SSRs) were mapped to construct a new genetic map spanning a map distance of 2,438.2 cM for this mapping population (Figure 2A and Supplementary Table 6). The genetic map with 20 linkage groups showed a marker density of 3.48 cM/locus. The number of mapped loci ranged from 20 (Ah07 with a density of 2.65 cM/locus) to 66 (Ah02 with a density of 1.60 cM/locus). The length of the chromosomes ranged from 53.06 cM (Ah07) to 180.52 cM (Ah13) (Figure 2B and Supplementary Table 5). Overall, the genetic map showed good marker collinearity with a physical map having a few exceptions (Figure 2C).

Open in a separate window

Figure 2

High density genetic map of RIL population of TMV 2 and TMV 2-NLM. (A) Map chart of high density genetic map. (B) Summary of genetic map with number of mapped loci, map distance (cM), and map density (cM/loci). (C) Collinearity of the genetic map with the reference genome (Arachis hypogaea L.). Prefix G and P stands for genetic map and physical map, respectively.

Main-Effect QTL Discovery for Yield and Quality Traits

A QTL analysis was conducted for the productivity (NPPP, PWPP, SP, and TW) and quality (PC, OIL, OLE, LIN, and O/L) traits across six seasons using composite interval mapping at 1,000 permutations. In total, 33 QTLs were identified for the four productivity traits that included nine QTLs for NPPP, two for PWPP, 10 for SP, and 12 for TW (Figure 3 and Supplementary Table 7). Among the nine QTLs for NPPP, three were identified as major QTLs. Among them, the first one (qNPPP-Ah02-1) identified on chromosome Ah02 with the LOD score of 9.6 showed 23.6% PVE during S4 season. The second QTL qNPPP_Ah04-1 detected on Ah04 had a PVE of 22.9% with an LOD score of 3.8 during S1 season. The third QTL qNPPP_Ah14-3 reported on chromosome Ah14 had a PVE of 17.3% with the LOD score of 7.7 during the S3 season (Figure 3 and Table 1). Among the two QTLs detected for PWPP, the first QTL on Ah02 (qPWPP-Ah02-1) was a major QTL with the highest LOD score of 10.6 and PVE of 20.9% (Figure 3 and Table 1), and it was stable over four seasons (S1, S2, S3, and S4). The other QTL on Ah01 (qPWPP-Ah01-1) was a minor QTL with a LOD value of 3.3 and a PVE of 5.1%. Of the 10 QTLs for SP, three were major-effect QTLs. Of them, the QTL (qSP-Ah13-1) identified on Ah13 had the highest PVE of 52.8% with the LOD score of 35.5 and stability over four seasons (S2, S3, S4, and S5). The second major effect QTL (qSP-Ah16-1) on Ah16 identified over three seasons (S1, S2, and S6) with the highest LOD score of 6.6 showed the highest PVE of 19.9% (Figure 3 and Table 1). The third major QTL (qSP-Ah02-1) on Ah02 had a PVE of 12.6% with a LOD score of 5.7 was detected only during S3. The remaining seven QTLs identified on Ah01, Ah05, Ah09, Ah10, Ah11, and Ah20 were minor QTLs that appeared only during one or two seasons. The favorable alleles for NPPP and PWPP were contributed by TMV 2-NLM, while TMV 2 contributed the favorable allele at two major QTLs for SP. No major-effect QTL was detected for TW; however, 12 minor-effect QTLs were identified with the highest PVE of 8.5% (qTW-Ah02-4).

Open in a separate window

Figure 3

Circos plot illustrating main effect and epistatic (QTL × QTL) QTLs identified for productivity and quality traits in ML population of TMV 2 and TMV 2-NLM of peanut. The tracks from outside to inside indicates (1) 20 chromosomes of tetraploid genome Arachis hypogaea, (2) main effect QTLs for number of pods per plant (NPPP) and linoleic acid (IAN) qNPPP–Ah02-1, qNPPP–Ah04-1, qNPPP–Ah08-1, qNPPP–Ah10-1, qNPPP–Ah12-1, qNPPP–Ah13-1, qNPPP–Ah14-1,qNPPP–Ah14-2, qNPPP–Ah14-3, qLIN–Ah19-1, qLIN–Ah19-2, qLIN–Ah19-3, qLIN–Ah19-4, qLIN–Ah19-5, qLIN–Ah19-6, qLIN–Ah19-7, qLIN–Ah19-8, qLIN–Ah19-9, qLIN–Ah19–10, qLIN–Ah19-11, qLIN–Ah19-12, qLIN–Ah19-13, (3) main effect QTLs for test weight (TW), protein content (PC), and oleic acid content (OLE) qTW–Ah02-1, qTW–Ah02-2, qTW–Ah02-3, qTW–Ah02-4, qTW–Ah03-1, qPC–Ah05-1, qPC–Ah05-2, qPC–Ah05-3, qPC–Ah10-1, qOLE–Ah10-1, qTW–Ah12-1, qTW–Ah12-2, qTW–Ah12-3, qTW–Ah12-4, qPC–Ah16-1, qPC–Ah16-2, qPC–Ah16-3, qOLE–Ah16-1, qOLE–Ah19-1, qOLE–Ah19-2, qOLE–Ah19-3, qOLE–Ah19-4, qOLE–Ah19-5, qOLE–Ah19-6, qOLE–Ah19-7, qOLE–Ah19-8, qOLE–Ah19-9, qOLE–Ah19-10, qOLE–Ah19-11, qOLE–Ah19-12, qTW–Ah20-1, qTW–Ah20-2, qTFV–Ah20-3, (4) main effect QTLs for shelling percentage (SP) and oleic to linoleic ratio (0/L) qSP–Ah01-1, qSP–Ah02-1, qSP–Ah05-1, q5P–Ah09-1, qSP–Ah10-1, qSP–Ah10-2, qSP–Ah11-1, OP–Ah13-1, OP-A/216-1, qO/L–Ah19-1, qO/L–Ah19-2, qO/L–Ah19-3, qO/L–Ah19-4, q0/L–Ah19-5, qO/L–Ah19-6, qO/L–Ah19-7, qO/L–Ah19-8, qO/L–Ah19-9, qO/L–Ah19-10, qO/L–Ah19-11, qO/L–Ah19-12, qO/L–Ah19-13, qSP–Ah20-1, (5) main effect QTLs for oil content (OIL) and pod weight per plant (PWPP). qPWPP–Ah01-1, qPWPP–Ah02-1, qOIL–Ah03-1, qOIL–Ah03-2, qOIL–Ah03-3, qOIL–Ah03-4, qOIL–Ah05-1, qOIL–Ah05-2, qOIL–Ah10-1 qOIL–Ah11-1, qOIL–Ah13-1, qOIL–Ah20-1, qOIL–Ah20-2. Innermost links connecting between the loci indicates the epistatic QTLs for NPPP, PWPP, TW, and SP.

For OIL, 11 QTLs were identified, of which two were major, and located on chromosome Ah03 (qOIL-Ah03-3) and Ah05 (qOIL-Ah05-1). qOIL-Ah03-3 was detected over four (S1, S2, S4, and S5) seasons with the highest LOD score of 9.5 and PVE of 13.7%. While qOIL-Ah05-1 was detected over the two seasons (S2 and S4) with the highest LOD score of 4.6 and PVE of 10.7%. The favorable alleles at both these QTLs were contributed by TMV 2-NLM (Figure 3 and Table 1).

A total of 47 QTLs were identified for quality traits (PC, OLE, LIN, and O/L) (Figure 3 and Supplementary Table 7). For PC, out of seven QTLs, two QTLs on Ah16 were major and stable across four seasons. Of them, qPC-Ah16-1 had the highest PVE of 13.3% with a LOD score of 15.3, and qPC-Ah16-2 had the highest PVE of 13.3% with a LOD score of 13.5. The favorable alleles for both these QTLs were contributed by TMV 2-NLM (Figure 3 and Table 1). A total of 14 QTLs were identified for OLE along with 13 QTLs each for LIN and O/L. Of them, 12 QTLs for OLE were major and stable with the highest PVE of 21.3%. The remaining two minor QTLs were mapped on Ah10 and Ah16 (Figure 3, Table 1, and Supplementary Table 7). For LIN, all the 13 QTLs were major and stable with the highest PVE of 17.1% (Figure 3, Table 1, and Supplementary Table 7). However, for O/L out of 13 QTLs, 11 were major and stable with the highest PVE of 18.4%. All the major and stable QTLs for OLE, LIN, and O/L clustered on a 26.5 cM region (61.1–87.6 cM) on Ah19 (Figure 3, Table 1, and Supplementary Table 7). It was inferred that TMV 2 contributed to the decreased level of OLE, and increased level of LIN.

Common QTL Clusters for the Productivity and Quality Traits

Three clusters were identified for the productivity and quality traits. Cluster 1 of 3.4 cM (46.7–50.1 cM on Ah02) was common for NPPP, PWPP, and SP. It showed the maximum PVE of 23.6, 20.9, and 12.6% for NPPP, PWPP, and SP, respectively (Figure 3 and Table 1). This region was highly stable for PWPP as it was detected over four seasons (S1, S2, S3, and S4). Also, the additive effects were high for this region, and the favorable alleles for NPPP, PWPP, and SP were contributed by TMV 2-NLM. Cluster 2 of 10.9 cM (7.1–8 cM on Ah16) carried the major QTL for PC and SP with a PVE of 7.9–13.3% and 11.9–19.9%, respectively (Figure 3 and Table 1). Cluster 3 of 26.5 cM (61.1–87.6 cM on Ah19) controlled OLE, LIN, and O/L with the PVE of 5.0–21.3%, 5.5–17.1%, and 6.0–18.4%, respectively, and this region was consistently stable over all the six seasons (Figure 3 and Table 1).

Single Marker Analysis

Single marker analysis revealed that a total of six markers were significantly associated with OLE along with five each for LIN and O/L with PVE ≥ 10. Of them, four markers (Ah19_155127364, Ah19_155135344, Ah19_155135353, and Ah19_155172354) and one (Ah19_155165240) marker located, respectively, on Ah19 and Ah09 were common for OLE, LIN, and O/L. These markers were also identified to be the flanking markers by CIM. These associations were consistent over all the six seasons (Table 2). In addition, Ah10_36971572 located on Ah10 showed association with OLE only during the S4 season.

A total of 10 markers were significantly associated with PC with a PVE of ≥10. Out of which AhTE0242 and AhTE0060 located at 0–7.16 cM were also identified by CIM. Three markers (Ah12_118126407, AhTE0242, and AhTE0060) showed association in S1 and S4 season, while seven markers (AhTE0281, Ah03_127278448, AhTE0087, AhTE0275, AhTE0120, AhTE1110, and AhTE1451) showed association only during the S1 season (Table 2). There were a few other markers associated with PC; however, they either showed relatively low PVE or appeared only during the specific seasons. SMA revealed that four markers (AhTE0281, AhTE0087, AhTE0120, and AhTE0242) were significantly associated with SP during the S3 season (Table 2). However, none of them was in the main effect QTL region detected for SP. For the remaining traits (OIL, NPPP, PWPP, and TW), none of the markers were detected as significant by SMA.

Epistatic QTL Discovery for Productivity Traits

Epistatic QTL analysis for the complex productivity traits, namely, NPPP, PWPP, TW, and SP identified a total of 94 epiQTLs, such as 87 major epiQTLs. In total 9, 12, 14, and 52 major epiQTLs were identified for NPPP, PWPP, TW, and SP, respectively. It was found that four major main effect QTLs of NPPP and SP were involved in epistatic interaction with PVE more than 10% (Figure 3 and Table 3). The rest of the epiQTLs involved either the main effect QTLs with minor effects or new genomic regions (Figure 3 and Supplementary Table 8). Of the four epistatic interactions for NPPP involving the major and main effect QTLs, three emerged from a genomic region on Ah02 showing significant interaction with the regions on Ah04, Ah06, and Ah12 with the PVE of 15.8, 17.7, and 20.0%, respectively during S4 season. Also, the major QTL for NPPP on Ah04 showed epistatic interactions with its own proximal region (10 cM) during S2, S3, and S4 seasons with a maximum PVE of 28.5% (Figure 3 and Table 3). Out of the remaining five epiQTLs for NPPP, those on Ah06, Ah18, and Ah19 showed interactions with their own close proximal regions (20, 5, and 15 cM) with the highest PVE of 39.6, 13.6, and 35.8%, respectively. Furthermore, the epiQTLs on Ah03 and Ah10 showed significant interactions with genomic regions on Ah16 and Ah19 with maximum PVE of 13.4 and 11%, respectively for NPPP (Figure 3 and Supplementary Table 8).

Of the four epistatic interactions for SP involving the major and main effect QTLs, a main effect QTL region on Ah16 (7.1–18 cM) showed epistatic interactions with Ah03, Ah12, and Ah13 with the maximum PVE of 24.2, 23.5, and 23.9%, respectively. Furthermore, a region (159.3–178.3 cM) on Ah13 with major main effect QTL also showed epistatic interactions with Ah08 recorded maximum PVE of 23.2% (Figure 3 and Table 3). In addition, a minor main effect QTL region on Ah01 for SP showed significant interaction with consecutive regions (at 60 and 80 cM) on Ah03 with the highest PVE of 22.3 and 22.6%. Among the remaining 47 epiQTLs, regions on Ah01, Ah05, Ah13, and Ah19 for SP were also involved in epistatic interaction with their own close proximal regions with PVE of 39.7, 30.4, 32.4, and 31.6%, respectively (Figure 3 and Supplementary Table 8).

For PWPP, none of the main-effect QTL was involved in epistatic interactions. However, the new epiQTL regions on Ah04, Ah05, Ah06, and Ah13 showed significant interactions with their own close proximal regions (5 cM) with the highest PVE of 36.4, 29.5, 43.3, and 26.7%, respectively. Apart from these, eight other epiQTLs appeared in at least two seasons with the major effect (PVE ≥10%) (Figure 3 and Supplementary Table 8).

Of the 14 epiQTLs detected for TW, a main effect minor QTL on Ah12 was involved in epistatic interaction with regions on Ah03, Ah07, and Ah14 with the highest PVE of 24.8, 28.5, and 15.8%, respectively. Among the remaining 11 epiQTLs, regions on Ah01, Ah03, Ah05, and Ah11 were also involved in epistatic interactions with maximum PVE of 17.6, 24.8, 28.4, and 14.3%, respectively. Most of these genomic regions identified in this study were important since they carried stable major QTL(s) which also showed significant epistatic interaction for various traits with PVE ≥ 10% across the seasons (Figure 3 and Supplementary Table 8).

Putative Genes Identified in Major Main-Effect QTL Regions/Clusters

In total, the three clusters and four major and stable QTL regions were subjected to candidate gene discovery. In cluster 1, a 5 Mb region from the left flanking marker (Ah02_100281747) toward the common QTL peak for NPPP, PWPP, and SP was considered for gene discovery, and 360 genes were found (Table 4). In clusters 2 and 3, the region between the left and right flanking markers were considered, and 34 and 3 genes were found in the regions, respectively (Table 4).

Of the four major QTL regions, a 5 Mb region from the left flanking marker to the QTL peak was employed for three; qPC-Ah16-1, qOIL-Ah03-3, and qOIL-Ah05-1, and 249, 421, and 333 genes were found in these regions, respectively (Table 4). The region between the two flanking markers was considered for qSP-Ah13-1, and 259 genes were found (Table 4). However, more studies are required to identify the candidate genes contributing to these traits.

Sixteen markers that were identified to be significantly associated with the traits by single marker analysis were checked for their location (genic and non-genic), effect, and probable function (Supplementary Table 9). Of them, 10 were found to be located in the intergenic regions and two each were located in the exonic, 5′ UTR, and intronic regions (Supplementary Table 9). AhTE0281 being located in the 16th exonic region of Arahy.7A57YA on Ah16 contributed for SP and PC (Supplementary Table 9). In addition, Ah12_118126407 being located in the second exon of Arahy.J5SZ1I (Ah12) governed PC. AhTE1451 being located in the 5′ UTR of Arahy.CH9B83 (Ah18) also governed PC (Supplementary Table 9). Similarly, an SNP at 155172354 bp being located in the 5′ UTR of the gene Arahy.X7PJ8H on Ah19 contributed for OLE, LIN, and O/L (Supplementary Table 9). Also, both Ah19_155135344 and Ah19_155135353 being located in the 11th intron of Arahy.MZJT69 on Ah19 contributed for OLE, LIN, and O/L (Supplementary Table 9).

The authors are thankful to the National Agricultural Science Fund (NASF) of the Indian Council of Agricultural Research, India for the financial support provided to GBS. MJ acknowledges the Senior Research Fellowship received from the Council of Scientific and Industrial Research (CSIR), Government of India.