Identification of quantitative trait loci associated with iron deficiency chlorosis resistance in groundnut ( Arachis hypogaea )

Iron deficiency chlorosis is an important abiotic stress affecting groundnut production worldwide in calcareous and alkaline soils with a pH of 7.5–8.5. To identify genomic regions controlling iron deficiency chlorosis resistance in groundnut, the recombinant inbred line population from the cross TAG 24 × ICGV 86031 was evaluated for associated traits like visual chlorosis rating and SPAD chlorophyll meter reading across three crop growth stages for two consecutive years. Thirty-two QTLs were identified for visual chlorosis rating (3.9%–31.8% phenotypic variance explained [PVE]) and SPAD chlorophyll meter reading [3.8%–11% PVE] across three stages over 2 years. This is the first report of identification of QTLs for iron deficiency chlorosis resist-ance-associated traits in groundnut. Three major QTLs (>10% PVE) were identified at severe stage, while majority of other QTLs were having small effects. Interestingly, two major QTLs for visual chlorosis rating at 60 days (2013) and 90 days (2014) were located at same position on LG AhXIII. The identified QTLs/markers after validation across diverse genetic could be used in genomics-assisted breeding.

dicots and monocots, except graminaceous species which adopts strategy-II. Groundnut adopts strategy-I and found susceptible to iron deficiency (Gholizadeh, Baghban, Kohnehrouz, & Hekmatshoar, 2007). Strategy-I involves steps like acidification of the rhizosphere through H(+)-ATPase-mediated extrusion of protons to increase the solubility of Fe 3+ (Santi & Schmidt, 2009), reduction of Fe 3+ to Fe 2+ through the ferric-chelate reductase (e.g. AhFRO1; Ding et al., 2009), and transport of Fe 2+ throughout the root plasma membrane through metal transporter (e.g. AhIRT1; Ding et al., 2010). Strategy-II involves production and release of phytosiderophores from roots, solubilization of iron by forming Fe 3+ -phytosiderophore complex and uptake through high-affinity transport system in plasma membrane of root cells. The Fe chelated by microbial siderophores might also be acquired (Prasad & Djanaguiraman, 2017).
Although iron is abundant in nature, it is often unavailable because it forms insoluble ferric hydroxide complexes in the presence of oxygen at neutral or basic pH as in calcareous soils (Guerinot & Yi, 1994). Iron deficiency chlorosis (IDC) is common worldwide among crops grown in calcareous and alkaline soils due to lower levels of available Fe (Fe 2+ ) for uptake. Calcareous soils are widespread with an estimated 800 m ha worldwide, mainly concentrated in areas with arid or Mediterranean climates (Land, FAO, & Plant Nutrition Management, 2000). In India, more than one-third of the soils are calcareous and spread mostly in the low rainfall areas of the western and central parts of the country, where groundnut is a major crop. The IDC is more prevalent in the Saurashtra region of Gujarat, Marathwada region of Maharashtra, and parts of Rajasthan, Tamil Nadu and Karnataka states in India causing significant reduction in pod yield (15%-32%; Singh, 2001;Singh, Basu, & Singh, 2003;Singh, Chaudhari, Koradia, & Zala, 1995). IDC is also a common problem in groundnut-producing areas with calcareous soils in northern China (Li & Yan-Xi, 2007) and Pakistan (Akhtar, Shahzad, Arshad, & Fayyaz-Ul-Hassan, 2013;Imtiaz, Rashid, Khan, Memon, & Aslam, 2010) causing significant reduction in yield.
Severity of IDC will be usually quite high after excessive rainfall and also for groundnuts grown under irrigation due to high bicarbonate ion concentration in the rhizosphere (Singh et al., 1995;Zuo, Ren, Zhang, & Jiang, 2007).
Iron deficiency in groundnut initially appears as chlorosis on young rapidly expanding leaves which is characterized by interveinal chlorosis. Iron deficiency has been found to decline net photosynthetic rate resulting from the reduction of photosynthetic pigment contents and inhibition of PSII photochemistry (Su et al., 2015). During severe deficiency, veins also become chlorotic, leaves become white and papery and later turn brown and necrotic, while the plants show stunted growth resulting in reduced yield, seed Fe content and fodder. Acute iron deficiency leads to death of plants and complete crop failure. Although application of Fe-containing fertilizers into soil or as foliar spray has been suggested (Frenkel, Hadar, & Yona, 2004;Irmak, Çıl, Yücel, & Kaya, 2012), it is often associated with problems like conversion into unavailable form (Fe 3+ ) or poor translocation within the plant (Hüve, Remus, Lüttschwager, & Merbach, 2003). Though foliar application of Fe chelates can overcome this problem, it is not economical as groundnut is predominantly grown as a rainfed subsistence crop by the resource-poor farmers in semi-arid tropics. Hence, the development of Fe-efficient genotypes can be a successive tool to overcome the Fe deficiency in soil and also for the improvement in human health (Imtiaz et al., 2010).
Identifying and developing IDC resistant genotypes is challenging due to high level of temporal and spatial variability of chlorosis expression in the field (King, 2011). Inconsistency in expression of iron deficiency symptoms could be due to various factors such as soil heterogeneity, bicarbonate ion concentration, soil moisture, temperature and relative humidity. The IDC response is usually assessed by visual chlorosis rating (VCR), chlorophyll content and SPAD chlorophyll meter reading (SCMR) in groundnut (Li & Yan-Xi, 2007;Mann et al., 2018;Samdur et al., 1999Samdur et al., , 2000. Higher SCMR is an indicator of lesser incidence of leaf chlorosis. Higher VCR and lower SCMR indicate "susceptibility", while lower VCR and higher SCMR indicate "resistance" to IDC. Growing of IDC resistant groundnut cultivars under calcareous soils has shown significantly higher pod yield compared to susceptible cultivars (Li & Yan-Xi, 2007;Mann et al., 2018;Prasad, Satyanarayana, Potdar, & Craufurd, 2000;Samdur et al., 1999). For effective development of IDC resistant groundnut genotypes, it is necessary to understand the genetic basis and also identify the specific genomic regions associated with IDC resistance. Earlier, inheritance study by Gowda, Kulkarni, Nadaf, and Habib (1993) indicated recessive nature of IDC resistance in groundnut showing trigenic (21:43) and pentagenic (525:499) ratios in F 2 population. On the contrary, genetic investigations assessed by six generation mean analysis indicated dominant nature of IDC resistance and also presence of non-allelic interactions for related characters like chlorophyll and carotenoid content (Samdur, Manivel, & Mathur, 2005). Our recent investigations on inheritance of IDC resistance among four crosses of groundnut based on F 2 and F 3 behaviour indicated duplicate dominant genes governing this trait (Pattanashetti, Naidu, Prakyath Kumar, Singh, & Biradar, 2018). In groundnut, three genes involved in iron acquisition have been identified, that is AhFRO1, encoding an Fe(III)-chelate reductase involved in reduction of Fe 3+ into Fe 2+ (Ding et al., 2009), and two iron transporters, AhIRT1 (Ding et al., 2010) and AhNRAMP1 (Xiong et al., 2012). Recently, AhIRT1 and AhNRAMP1 expression have also been correlated with cadmium (Cd) uptake in groundnut under iron deficiency (Chen, Xia, Deng, Liu, & Shi, 2017).
The integration of available genomic resources together with modern genomics approaches, high throughput phenomics and simulation modelling will help in achieving higher genetic gains (Varshney et al., 2018). Breeding for IDC resistance using genomics-assisted breeding (GAB) can increase the selection efficiency and cost-effectiveness, reduce the duration of breeding cycle, but it requires identification of linked QTLs/markers Varshney et al., 2013Varshney et al., , 2019. The deployment of linked markers in soybean has successfully demonstrated 2.6-fold increase in selection efficiency relative to phenotypic selection, wherein 73% of lines developed were with superior IDC resistance, and there was 70% reduction in cost of IDC evaluation compared to traditional breeding schemes (Charlson, Cianzio, & Shoemaker, 2003). Hence, the present study undertook the extensive phenotyping of IDC resistance-associated traits and analysed these data together with available genetic maps data for identification of QTLs and linked markers for IDC resistance in groundnut.

| Plant materials
The RIL population (318 lines) of the cross TAG 24 × ICGV 86031 was developed earlier at ICRISAT, Patancheru to map drought tolerance traits in groundnut (Faye et al., 2015;Gautami et al., 2012;Ravi et al., 2011;Varshney et al., 2009). As per field screening at College of Agriculture, Vijayapur, India during rainy season 2009, the parent TAG 24 was found IDC susceptible (VCR 4.0), while ICGV 86031 as IDC resistant (VCR 1.0; Figure 1a). Keeping the IDC response of parents in view, phenotyping of this RIL population for IDC resistance-associated traits like VCR and SCMR was undertaken towards identifying genomic regions associated with IDC resistance in groundnut.

| Phenotyping of RIL population
The field experiment was conducted for two consecutive years (2013 and 2014) during rainy season at College of Agriculture, Vijayapur, India (16°49'N, 75°43'E, 593 m above mean sea level, and 597 mm average annual rainfall) on calcareous vertisol soils that are alkaline (pH > 8) and deficient in available Fe (DTPAextractable Fe < 4 mg/kg; Table S1). Field screening for IDC response of RIL population along with parents (320 lines) was done using randomized complete block design in two replications. Each

| Evaluation for IDC resistance
Iron deficiency chlorosis resistance-associated traits like VCR and SCMR were assessed across three stages, that is 30, 60 and 90 days after sowing (DAS) for two consecutive years. VCR scoring was done as per the scale proposed by Singh and Chaudhari (1993) (1 to 5 scale: 1-normal green leaves with no chlorosis, 2-green leaves but with slight chlorosis on some leaves, 3-moderate chlorosis on several leaves, 4-moderate chlorosis on most of the leaves, 5-severe chlorosis on all the leaves) ( Figure 1c) on overall line basis. Higher VCR score indicates susceptibility, while lower VCR indicates resistance to IDC. Based on VCR score, lines can be considered as resistant (VCR 1 to 2), moderately resistant (>2 to 3) or susceptible (>3 to 5).
The chlorophyll meter SPAD 502 (Soil Plant Analysis Development meter, Konica Minolta, Japan) was used to measure the absorbance of the leaf in the red (at 650 nm) and near infrared region (at 940 nm). Using these two transmittances, it calculated a numerical SPAD value which is proportional to the chlorophyll present in the leaf and is negatively related to chlorosis of the plants. The SCMR (SPAD values) was recorded in the standard leaf (third leaf from the top on main stem, i.e., fully expanded) of five plants showing most severe symptoms per genotype or plot, and their mean was calculated. Higher SCMR indicates resistance, while lower SCMR indicates susceptibility to IDC. As the SCMR is a continuous variable, it is difficult to make classes for IDC response. However, for better understanding of the distribution for SCMR, we grouped the RILs into six categories with an interval of five, that is ≤20, >20-25, >25-30, >30-35, >35-40 and >40.

| Estimation of chlorophyll and active Fe content
Chlorophyll (a, b and total) and active Fe (Fe 2+ ) content have been shown to be positively correlated with IDC resistance in several crop species. To confirm the IDC resistance/ susceptibility, chlorophyll and active Fe content were estimated at most severe stage (60 DAS) among both parents and selected five RILs each of IDC resistant and susceptible types during 2014. Sample from standard leaf of five plants per genotype was used to estimate the chlorophyll and active Fe content and mean was calculated. The chlorophyll content was estimated using method described by Shoaf and Lium (1976) and expressed as mg/g on fresh weight basis. Active Fe content was estimated using the method described by Katyal and Sharma (1980) using o-phenanthroline extractant and expressed as mg/kg on fresh weight basis.

| Statistical analyses
The phenotypic data of VCR and SCMR at three stages for individual years (2013,2014) were analysed for analysis of variance technique by mixed model procedure of SAS version 9.3 (SAS Institute Inc., 2017) considering replication and genotype as random effect.
Square root transformation has been applied on VCR before analysis.
Best linear unbiased predictor (BLUP) values were estimated for VCR and SCMR at three stages during individual years. To assess genetic variability in the RIL population, components such as genotypic coefficient of variation (GCV), phenotypic coefficient of variation (PCV) (Burton & Devane, 1953), broad sense heritability (H bs ) (Falconer, Mackay, & Frankham, 1996) and genetic advance as per cent mean (GAM) (Johnson, Robinson, & Comstock, 1955) were estimated using genotypic and phenotypic variances, and BLUP mean values.
Association among traits was calculated using Pearson's correlation coefficients.

| Genetic mapping of RIL population
The genetic mapping data generated earlier for the TAG 24 × ICGV 86031 RIL population using 191 SSR marker loci were used for QTL analysis. The detailed information on identification of polymorphic markers on parental genotypes, parental polymorphism, genotyping of mapping population and construction of genetic map has been described in detail in Varshney et al. (2009) and Ravi et al. (2011).

| Quantitative trait locus (QTL) analysis
The QTL analysis for VCR and SCMR at three different stages (30, 60, 90 DAS) for two years (2013 and 2014) was performed by using BLUP values along with genotyping data for 191 SSR markers. The QTL analysis was performed by using Windows QTL Cartographer version 2.5. Composite interval mapping (CIM) approach was deployed for identification of location and effect of QTLs. This software uses a dynamic algorithm which considers various gene actions (additive and dominance), QTL-environment interactions and close linkage (Wang, Basten, & Zeng, 2007). Parameters such as model 6, scanning intervals of 1.0 cM between markers and putative QTLs with a window size of 10.0 cM were used for conducting the CIM analysis. In addition, forward-backward stepwise regression was selected for background control set by the number of marker cofactors along with 500 times permutations with 0.05 significance level and "Locate QTLs" option to locate QTLs. The QTL analysis was conducted on phenotyping data of individual years (2013, 2014) separately by using BLUPs.

| Phenotypic variability for IDC resistanceassociated traits
Large variation was observed among parents and RIL population for IDC resistance-associated traits like VCR and SCMR across three stages (30, 60 and 90 DAS) during both the years (2013, 2014) evident from highly significant differences for genotypes based on F test (p < .0001) ( Table 1). For better clarity on variability for the IDC associated traits among parents as well as RILs, mean values of the genotypes are presented in this section. In both the years, ICGV 86031 remained resistant to IDC across all three stages evident from lowest VCR scores (1.0) and higher SCMR values (39.1-45.3), while TAG 24 was found susceptible to IDC at all three stages evident from higher VCR scores (3.0-4.0) and lower SCMR values (10.95-23.05; Table 2). Wide variation was observed among the RILs for both VCR and SCMR across three stages for two years, wherein some values were beyond the parents (Table 2). Variability observed for IDC response among parents and RILs is depicted in Figure 1.
Frequency distribution of RILs (number) with VCR scores from 1 to 4 across three stages for two years is given in Figure 2a    2013 and 2014, respectively, at the same QTL region between the markers Seq2C11-GM2259 on AhXIII.

| Identification of QTLS for IDC resistance
The QTL analysis for SCMR in 2013 rainy season revealed seven QTLs across three stages (2 for VCR 30 DAS, one for VCR 60 DAS and four for VCR 90 DAS; Table 5, Figure 3). The LOD score for these QTLs ranged from 3.0 to 6.2, PVE 3.8%-7.3% and the additive effect of individual QTLs from −1.46 to 1.05. Three QTLs were identified on LG AhVIII, two QTLs on AhXIV, while one QTL each on LG AhIV and AhXIII. The QTL analysis for SCMR in 2014 rainy season identified six QTLs (4 for SCMR 60 DAS, two for SCMR 90 DAS), while no QTL was detected for SCMR 30 DAS (Table 5, Figure 3). The LOD value for these QTLs ranged from 3.0 to 5.3, PVE 3.9%-11% and additive effect 1.39 to 2.04. Of the six QTLs identified, two QTLs each were located on LG AhXIV and AhXV, while one QTL each was on LG AhIX and AhXII. One major QTL was identified for SCMR 60 DAS located between marker interval GM2603-Seq16G08 on LG AhXV explaining 11% PVE.

| D ISCUSS I ON
Iron deficiency chlorosis is a major production constraint in groundnut growing areas with calcareous soils that are deficient in available Fe which causes significant yield loss in India (Singh, 2001), northern China (Li & Yan-Xi, 2007) and Pakistan (Akhtar et al., 2013;Imtiaz et al., 2010). In a recent study at Vijayapur (India), IDC response assessed among 26 released varieties, 13 advanced breeding lines and 4 germplasm lines indicated that majority of the released cultivars were found either susceptible or moderately resistant to IDC, while one each of released cultivar (ICGV 86031), advanced breeding line (ICGV 06146) and germplasm (A30b, an interspecific derivative) were found IDC resistant (Boodi, 2014 Phenotyping for IDC resistance-associated traits like VCR and SCMR across three stages during both years revealed significant differences among parents and large variation in the RIL population (Tables 1 and 2). Earlier, huge variation has been noted among parents and RIL populations while mapping for IDC resistance-related traits like visual scoring and SPAD values (SCMR) in soybean (Butenhoff, 2015;Lin, Cianzio, & Shoemaker, 1997) and mungbean (Prathet, Somta, & Srinives, 2012), and also for zinc efficiency score in wheat (Genc et al., 2009). Though visual scoring is a fast and convenient method to evaluate for IDC in groundnut, due to the complexity associated with field screening, it is essential to confirm IDC resistance through biochemical parameters like chlorophyll and active Fe content. Since significant correlations have been established between SCMR, chlorophyll and active Fe content for parents and selected RILs in the present study (Table 3) and also by earlier researchers in groundnut (Akhtar et al., 2013;Li & Yan-Xi, 2007;Samdur et al., 2000), SCMR is found ideal for confirmation of IDC resistance since it is reliable, faster and convenient. Further, higher values of genetic components such as GCV, PCV, H bs and GAM and correlations recorded for VCR and SCMR (Tables 2 and 4) also indicate their utility as important traits in breeding for IDC resistance in groundnut.
Iron deficiency chlorosis severity across two years indicated higher severity at 30 DAS during 2013, while much higher severity at 60 and 90 DAS during 2014. Earlier reports in groundnut indicate beginning of iron deficiency at 10-15 days after emergence, while attaining of maximum intensity at 30-70 days (Singh & Chaudhari, 1993) or 50-65 days after emergence (Li, Yan-Xi, & Jian-min, 2009 IDC resistance has been reported to be under the control of one or few dominant genes among several legumes including groundnut (Pattanashetti et al., 2018;Samdur et al., 2005), while some reports suggest polygenic inheritance with additive effect in soybean (Cianzio & Fehr, 1982) and tomato (Dasgan, Abak, Cakmak, Romheld, & Sensoy, 2004).
The BLUP values are more robust and better indicators of the phenotypic performance; hence, the QTLs identified based on BLUPs can be more precise and reliable. The QTL analysis using genotyping and phenotyping data identified a total of 32 QTLs located on eight LGs with PVE ranging from 3.9% to 31.8% (  (Butenhoff, 2015;Charlson, Bailey, Cianzio, & Shoemaker, 2005;Charlson et al., 2003) and mungbean (Sommanus, 2000;Srinives, Kitsanachandee, Chalee, Sommanas, & Chanprame, 2010). Mapping studies could detect QTLs associated with IDC resistance/ iron efficiency in soybean (Butenhoff, 2015;Lin et al., 1997) and mungbean (Prathet et al., 2012). According to "soybase website" (https://soyba se.org), there are 40 QTLs identified for Fe efficiency. Some QTLs associated with seed mineral content for Fe and/or Zn, B, Mn and Cu have been reported in soybean (Bellaloui et al., 2015;King et al., 2013), Andean common beans (Blair, Astudillo, Rengifo, Beebe, & Graham, 2011), lentil (Aldemir et al., 2014) and wheat (Genc et al., 2009). In Andean common bean (Phaseolus vulgaris), genes conditioning iron reductase activity in iron-sufficient plants appear to be associated with genes contributing to seed iron accumulation (Blair et al., 2011). Similarly in soybean, seeds with higher iron content were found to show higher degree of IDC resistance (King et al., 2013). Hence, seed Fe content and iron efficiency go together as significant correlations have been noted between them.
Several candidate genes associated with IDC resistance have been identified in soybean (Mamidi et al., 2011;Peiffer et al., 2012). For instance, a 12-bp deletion within the predicted dimerization domain in Glyma03g28610 (bHLH gene) was found to hinder the FIT/bHLH heterodimer that induces iron acquisition genes like FRO2 and IRT1 thereby resulting in susceptibility to IDC in soybean (Peiffer et al., 2012). QTL analysis using high-density SNP genetic map in soybean identified seven major effect QTLs on seven chromosomes, wherein 12 candidate genes associated with iron metabolism were mapped near these QTLs supporting poly- high-density genotyping SNP array "Axiom_Arachis" (Pandey et al., 2017) with 58K highly informative genome-wide SNPs have provided now further opportunities for fine mapping and candidate gene discovery for IDC resistance in groundnut.

| CON CLUS IONS
This study reports 32 QTLs for IDC resistance-associated traits VCR and SCMR across three stages over two years. Three major QTLs explaining 11%-31.8% PVE were identified on AhXIII and AhXV, while other QTLs of small effects having less than 10% PVE were identified on eight LGs. The LG AhXIV alone harboured 12 QTLs in the genomic region GM2308-GM626 and seems to be important for IDC resistance in groundnut. Further, the saturated linkage maps could be used to fine map these QTLs and also to find candidate genes controlling IDC resistance in groundnut. The identified markers associated with IDC resistance can be deployed in molecular breeding after validation across diverse genotypes for improving IDC resistance in groundnut.

ACK N OWLED G EM ENTS
Authors would like to thank Dr Vincent Vadez, ICRISAT, Patancheru (India) for supplying the seeds of the RIL population for phenotyping and also for critically reviewing this paper. This work was undertaken as part of the post-graduate research of Omprakash and

CO N FLI C T S O F I NTE R E S T
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.