Introgression of qDTY1.1 Governing Reproductive Stage Drought Tolerance into an Elite Basmati Rice Variety “Pusa Basmati 1” through Marker Assisted Backcross Breeding

: Drought stress at the reproductive stage in rice is one of the most important cause for yield reduction, affecting both productivity and quality. All Basmati rice varieties, including the popular cultivar “Pusa Basmati 1 (PB1)” is highly sensitive to reproductive stage drought stress (RSDS). We report for the ﬁrst time, improvement of a Basmati cultivar for RSDS tolerance, with the introgression of a major quantitative trait locus (QTL), “ qDTY1.1 ” into PB1. The QTL was sourced from an aus variety, Nagina 22 (N22). A microsatellite (simple sequence repeat (SSR)) marker “RM 431” located at telomeric end (38.89 mb) of chromosome 1, and located within a 1.04 mb QTL region was employed for foreground selection for qDTY1.1 in the marker assisted backcross breeding process. A set of 113 SSR markers polymorphic between N22 and PB1 were utilized for background selection to ensure higher recurrent parent genome recovery. After three backcrosses followed by ﬁve generations of selﬁng, eighteen near isogenic lines (NILs) were developed, through combinatory selection for agro-morphological, grain and cooking quality traits. The NILs were evaluated for three consecutive Kharif seasons, 2017, 2018 and 2019 under well-watered and drought stress conditions. RSDS tolerance and yield stability indicated that P1882-12-111-3, P1882-12-111-5, P1882-12-111-6, P1882-12-111-7, P1882-12-111-12, P1882-12-111-15 and P1882-12-111-17 were best in terms of overall agronomic and grain quality under RSDS. Additionally, NILs exhibited high yield potential under normal condition as well. The RSDS tolerant Basmati NILs with high resilience to water stress, is a valuable resource for sustaining Basmati rice production under water limiting production environments.


Introduction
India is gifted with a vast rice varietal diversity spread across diverse ecosystems. The region spanning from the Himalayan foothills traversing through north-western Indo-Gangetic plains is particularly bestowed with rices of incredible quality, popularly known as Basmati [1]. Over the time, Basmati has become pride possession of Indian subcontinent, serving exquisite cuisine and thereby an export commodity for trade in the studies have demonstrated that qDTY1.1 was flanked by markers RM431 on telomeric end and RM11943 on the centromeric end. Additionally, Bernier et al. [17] identified a different major QTL on chromosome 12, qDTY12.1, from the cross between Vandana and Way Rarem explaining approximately 51% of phenotypic variation. There are also other major QTLs reported which were demonstrated effective either under upland or lowland situations [14,18].
Marker assisted introgression of major-effect QTLs could be a proficient and rapid approach for breeding rice varieties tolerance to drought stress [17]. Consequent attempts to introgress/pyramid these QTLs by marker assisted selection, primarily into mega-varieties has found significant advancements towards breeding climate-adaptive cultivars [19,20], such as Sabitri (qDTY3.2 and qDTY12.1) [21], IR 64 (qDTY2.2 and qDTY4.1), Vandana (qDTY12.1) [22] and Pusa 44 [23]. In the past twelve years, there has been about sixtysix RSDS tolerant varieties released around the world, which involve several of these QTLs [14]. Although important, there has not been a previous attempt to improve Basmati cultivars for RSDS tolerance. One of the major reasons for this hiatus was the lack of donors from the Basmati group. Use of donors from the non-Basmati backgrounds for Basmati improvement, present a major challenge of loss of grain quality of the Basmati parent, while introgression [24]. However, marker assisted backcross breeding coupled with phenotypic selection, has been demonstrated to effectively address this problem [25][26][27].
Developed by the Indian Council of Agricultural Research (ICAR)-Indian Agricultural Research Institute (ICAR-IARI), New Delhi and released for commercial cultivation during year 1989, Pusa Basmati 1 (PB1) is the world's first semi-dwarf (105 cm) Basmati variety with high yielding potential. This variety, that showcased a tremendous level of improvement over the traditional Basmati cultivars that were low yielding (~2.3 tons/ha), photosensitive, tall (>150 cm), lodging with weaker stem and of long duration (>150 days), became popular among the farmers in no time. Besides the high yield (5.0 tons/ha), PB1 possesses insensitivity to photoperiodism, resistance to lodging, superior grain quality, semi-dwarf stature and a duration of 135-140 days. The grain quality of PB1 was adorned with strong aroma with explicit cooking qualities such as high kernel elongation ratio of 1.8, an average milled grain length of 7.4 mm and cooked kernel length of 13.7 mm [19]. However, as that of traditional cultivars, PB1 too is highly sensitive to several stress factors such as pests and diseases as well as drought. This study forms the maiden attempt to improve any Basmati cultivar towards climate resilience, by introgressing RSDS tolerance into PB1. The augmented objectives were to recover the Basmati grain quality traits as well as the recovery of all the agronomic traits including yield in the near isogenic lines (NILs). Further, evaluation of the improved NILs across multiple locations and environments to ascertain the stability in yield and drought tolerance to deploy as commercial Basmati rice cultivar.

Plant Material and Experimental Sites
PB1, the elite Basmati cultivar was selected as the recurrent parent. An aus cultivar, N22, was used as donor parent for the QTL, qDTY1.1. N22 is a pure line selection from a landrace, Rajbhog. Well-known as a universal donor for high temperature tolerance, drought tolerance and grain dormancy [28], N22 is a tall (~120 cm), short duration (90-95 days) cultivar, with short bold non-aromatic grains. The entire study was conducted between two locations, New Delhi and Aduthurai. Experiments at New Delhi was conducted in the research fields of the Genetics Division, ICAR-IARI, situated at 28 • 35 N latitude, 77 • 12 E longitude with an altitude of 228.16 m above mean sea level. The other experimental site at Aduthurai was the research farm of IARI-Rice Genetics and Breeding Research Centre situated at 11 • 00 N latitude, 79 • 28 E longitude and 20 m above MSL. The whole generation advancement between New Delhi and Aduthurai was carried out through shuttle breeding approach. Final drought stress evaluation of NILs was done at New Delhi for three subsequent crop seasons.

Development of NILs through Marker Assisted Backcrossing
Hybridization was taken up between PB1 and N22 at New Delhi, using PB1 as recipient (female) and N22 as donor (male) in Kharif 2013. Hybrid seeds (F 1 s) were harvested and propagated in the late Rabi season of 2013-2014 at Aduthurai, Tamil Nadu. F 1 s were backcrossed to PB1 as recurrent parent (RP), to generate BC 1 F 1 . During subsequent seasons, two successive backcrosses were done to generate BC 2 F 1 and BC 3 F 1 during 2014 and 2014-2015. In the BC 3 generation, from Kharif 2015, four successive selfings were carried out to generate BC 3 F 5 NILs by 2016-2017 late Rabi season at Aduthurai. All through during different generations, selections were carried out to recover maximum recurrent parent genome recovery (RPGR) among the progenies. The selected NILs, beginning from the BC 3 F 4 generation, were subjected to three consecutive drought screenings at New Delhi during Kharif seasons 2017 to 2019.
For molecular marker analyses, the genomic DNA was isolated from fresh leaf samples using CTAB method with minor modifications [29]. Polymerase chain reaction was performed with the volume of 10 µL containing 20-30 ng template DNA, 5 pmol of each primer, 0.05 mM dNTPs (MBI, Fermentas, Lithuania, USA), 10× PCR buffer (10 mM Tris, pH 8.4, 50 mM KCl and 1.8 mM MgCl 2 ) and 0.5 U of Taq DNA polymerase (Bangalore Genei Pvt. Ltd., Bangalore, India), running a amplification profile consisting of one cycle of initial denaturation at 94 • C for 5 min; followed by 35 cycles containing denaturation with at 94 • C for 30 s, annealing at 55 • C for 30 s, extension at 72 • C for 1 min; and a final extension at 72 • C for 7 min. The amplified PCR products were resolved by electrophoresis having 3.5% agarose gel and fluoro-stained with ethidium bromide. The amplicon resolution was pictographed using a Gel Doc XR+ ® (BioRad Laboratories, Hercules, CA, USA) gel documentation system.
For the marker analyses, marker sequences were downloaded from the marker database at Gramene (http://www.gramene.org) and the primers were custom synthesized from Sigma-Aldrich (Bengaluru, India). The qDTY1.1 linked simple sequence repeat (SSR) marker RM431 (F: tcctgcgaactgaagagttg; R: agagcaaaaccctggttcac) was polymorphic between the parents and was thus used for foreground selection [15]. For background analysis, the genome wide polymorphism between PB1 and N22 was determined using a genome wide survey employing 651 SSR markers, which resulted in identification of 113 markers that could distinguish both the genotypes. These polymorphic SSR markers spanned across the rice genome and were used for background selection in various backcross generations (Supplementary Table S1), using the reductive screening approach [30]. The recovery of PB1 type alleles among the backcross progenies during each generation was used for computing the RPGR [31]. For the graphical comparison of the genomes of NILs and parents, graphical genotypes were drawn using GGT v.2.0 [32].

Field Management under Stressed and Unstressed Conditions
In order to assess the efficiency of qDTY1.1 introgression into PB1, the selected NILs along with their parents and checks were evaluated for three consecutive Kharif seasons of 2017 to 2019 period at ICAR-IARI, New Delhi, under an irrigated ecology with drought stress imposed at reproductive stage. The drought tolerant check was IR86918-B-B-305, a backcross inbred line carrying qDTY1.1 in the background of IR64 developed from the cross IR64/N22. The sensitive check used in this experiment was IR64. Two treatments were maintained, stressed and unstressed. Experiments were laid out using a randomized complete block design (RCBD) with two replications for each stress treatment. The plot size was 6.5 m 2 with a spacing of 20 by 15 cm. Approximately 25 g of seeds were nursery sown to raise the population. In the plots designated for stress treatment, tensiometers were installed at every six NILs to monitor and characterize the soil moisture status. Initially, for both the treatments, seedlings were raised in a wet-bed nursery and after 21 days transplanted into a flooded field with 5 cm standing water. To ensure uniform establishment of the transplanted seedlings, all the plants were maintained under irrigation for 30 days post transplanting. At the 31st day after transplanting, water from the stress treatment plots were drained to initiate the stress. The stressed plots were left un-irrigated until the soil moisture tension reached −70 kPa at 30 cm depth. Severe leaf rolling and leaf drying were observed at this soil moisture level. At this severe stress, a flash lifesaving irrigation was provided, and the excess water was drained out approximately after 24 h. This cycle was constantly repeated until harvest [15,33]. The unstressed plots were maintained with normal irrigation and the plants were maintained in the standing water. Altogether, three irrigations were given into stressed plots, while the unstressed plots were irrigated six times. During the crop duration, from June to November, a total of 839.8, 913.4 and 608.1 mm of rainfall was received in the years 2017, 2018 and 2019, respectively.

Phenotypic Data Collection
From each of the NILs under both treatments, phenotypic data were recorded from five randomly tagged plants on days to 50% flowering (DF)-the number of days from sowing to flowering in 50% plants/tillers was recorded, plant height at maturity (PH)-was measured from ground level to the tip of main panicle, panicle length (PL)-average length of the primary panicles was taken from peduncle base to the tip, number of reproductive tillers (NT)-average number of tillers containing grain filled panicles per plant, spikelet fertility % (SF)-the proportion of filled grains to the total number grains per panicle and grain yield (GY)-the harvested grain from the plant were dried to optimum moisture level of approximately 14% after which they were weighed. The NILs were also characterized for kernel length and breadth before and after cooking. From these measurements, length/breadth ratio (LBR) and cooking quality characteristics, such as kernel length elongation ratio (KER) were computed. Further, alkali spreading value (AS) and aroma were determined using standard protocols [34].

Statistical Analyses
Initially, independent analyses of variance (ANOVA) was carried for each season to identify significant responses among the NILs under stressed and unstressed situations and to compare them with the checks. A subsequent combined ANOVA was carried out using linear mixed model approach with genotypes as fixed factor and seasons and stress as random factors. The trait predictions obtained from the model was saved as best linear unbiased predictors (BLUPs). The data were analyzed using STAR package version 2.0.1. (http://bbi.irri.org/products). To identify the genotypes with stable performance across the environments, seasons under drought condition, an additive main effective and multiplicative interaction (AMMI) model was constructed using BLUPs and AMMI stability value (AST) [35] and yield stability index (YSI) [36] were generated, as follows: Yield stability index, where, SS IPC1 and SS IPC2 are the sum of squares of interaction principal component axes (IPCA) 1 and 2, respectively, and IPC1 and IPC2 are the respective IPCA scores. Similarly, R AST and R Y are the respective genotypes ranks based on AST and yield (Y).
To compare the effect of drought on different traits, a forward stepwise regression analysis was carried out for delineating the traits which contributed significantly to yield. Based on the BLUPs for grain yield under stressed (S) and unstressed (NS) treatments, different indices namely, drought yield index [37], stress tolerance index [38], stress susceptibility index [39], and percent reduction in yield were worked out as follows: where, Y i represents the mean yield of a genotype i on the untransformed scale, Y G refers to geometric mean across genotypes, the suffices, NS and S represent unstressed and stressed conditions, respectively and Y refers to the arithmetic mean yield across genotypes.

Parental Polymorphism
The genome wide polymorphism survey using 651 SSR markers, revealed a diversity of 17.4% between the parents PB1 and N22, identifying 113 markers to be polymorphic between them ( Table 1). Out of 101 markers tested on chromosome 1, the carrier chromosome of qDTY1.1, 14 markers were observed polymorphic including the linked marker, RM431. The target chromosome diversity was 13.9%. Of the remaining chromosomes, highest diversity was found for chromosome 8 (40.9%) while the chromosome 6 (9.2%) indicated low diversity.

Introgression of qDTY1.1 into PB 1
The breeding scheme for development of PB1 NILs carrying qDTY1.1 is given in Figure 1. From a total of 23 seeds initially collected of the cross PB1/ N22, five plants were found to be pure hybrids showing heterozygosity for the foreground marker, RM431 linked to qDTY1.1. These F 1 s were backcrossed to the recurrent parent, PB1 to obtain 39 BC 1 F 1 seeds. Out of these, 15 seedlings were confirmed heterozygous for qDTY1.1 by foreground selection. These 15 plants were further analyzed with 113 polymorphic background SSRs ( Table 2). Based on the recovery of PB1 alleles, the RPGR among the BC 1 F 1 plants was estimated to range between 78.0% and 86.7%. The plant with highest recovery (86.7%) had 83 background markers in homozygous state for recurrent parent (RP) allele, while 30 markers were heterozygous. Further, these plants were also tested for their agronomic and grain quality similarities with PB1. The plant showing phenotypic resemblance and high RPGR (86.7%) was used to backcross with PB1 to generate 37 BC 2 F 1 seeds. However, only one BC 2 F 1 plant was found to be heterozygous for RM431. Background analysis on this plant using 30 markers heterozygous from the previous backcross generation, showed the RPGR of 94.7%. Subsequent backcrossing of this plant to PB1 yielded 25 seeds. On raising the BC 3 F 1 generation from these seeds, five plants were found to be heterozygous for qDTY1.1 linked marker, RM431. Further the one F 1 out of five with maximum RPGR and excellent cooking quality was selected for further generation advancement. Background selection with the remaining unrecovered heterozygous markers among the BC 2 F 1 plant, resulted in an RPGR ranging from 95.6% to 97.4%. These BC 3 F 1 plants were further subjected to phenotype matching with PB1, and the plant with maximum genome recovery (97.4%) as well as phenotype similarity was used for further step of selfing. Selfing the selected BC 3 F 1 , yielded 229 BC 3 F 2 seeds. Through foreground followed by background selections of these BC 3 F 2 plants, 55 plants homozygous for qDTY1.1 with an RPGR of 97.8-98.7% were identified. All these 55 plants were advanced to BC 3 F 3 families which underwent a rigorous phenotype selection for agro-morphological and grain quality traits to select 24 families for further advancement to BC 3 F 4 generation. The BC 3 F 4 families were grown under both irrigated and stressed conditions. Subsequent selection among the 24 BC 3 F 4 families was carried out for agronomic performance, drought response and grain quality. This resulted in 18 families of near isogenic lines, having close similarity to PB1 and comparable grain and cooking qualities. Phenotype matching of these 18 NILs with PB1, both in terms of agronomic features and grain quality, identified them closer to PB1. All the 18 NILs had target marker pattern of the donor (250 bp) parent for qDTY1.1 ( Figure 2). The RPGR at BC 3 F 5 generation, ranged between 98.2% and 99.1%. All along the selfing generation, selection was carried out with major emphasis on recovery of grain quality, agro-morphological features, and yield of PB1, among the NILs. The graphical genotype of targeted segment for qDTY1.1 from donor and recovery of background genome on chromosome 1 in 18 NILs are depicted in Figure 3.

Recovery of RP Alleles in Carrier Chromosome
The polymorphism survey between PB1 and N22 on chromosome 1, the carrier chromosome of qDTY1.1, was performed using 102 markers, of which 14 were found polymorphic including the foreground marker, RM431. There were two flanking markers for qDTY1.1 in the marker array, RM431 and RM11943. Among these, only RM431, the telomeric end marker was found polymorphic between the parents. RM11943 was found monomorphic. Further, among the downstream markers towards the telomeric end, only one marker RM6840 was polymorphic. However, RM6840 was at the distal end, 4.27 Mb away from RM431. On the centromeric end, however, the next upstream polymorphic marker was RM3825. RM3825 was used for recombinant selection. These markers, except for RM431, were included in the whole genome background recovery analysis. All the 14 polymorphic markers in chromosome 1 showed complete recovery of RP alleles among the 18 NILs by BC 3 F 5 generation (Supplementary Figure S1).

Per se Performance of the NILs under Reproductive Stage Drought Stress and Unstress Treatments
Agronomic evaluation of NILs along with parents and checks over the three Kharif seasons under field-imposed drought stress as well as under normal conditions, indicated significant variation for several traits (Table 3). These evaluations belonged to BC 3 F 4 , BC 3 F 5 and BC 3 F 6 generations. ANOVA over individual years revealed significant variation for genotypes, treatment and genotype × treatment components, for most of the agronomic characters studied. However, only two traits, SF and GY showed consistently significant variation for genotype × treatment component in all the seasons. For the remaining traits, during 2018, genotype effects for NT and genotype × treatment effects for DF, NT and PL were found non-significant, as that of the genotype × treatment effect of NT in 2017. Combined ANOVA, indicated a uniform pattern across the traits, particularly for GY and SF. NT had non-significant effects for various sources of variation except genotype component. Traits that showed significant variation for all the sources except for one component, included DF and PH where in year × treatment interaction was non-significant, while PL had non-significant year × genotype effect. Except for PL and NT, year × treatment × genotype interaction was significant for all the other traits. Lease significant difference (LSD) values generated from the individual ANOVA in different years were utilized for mean comparison of NILs with checks.
Considering the individual traits (Supplementary Table S2), DF exhibited uniform pattern across three experimental years, but a general delay was observed in flowering under reproductive stage drought stress. The delay was conspicuous in IR64 and was particularly apparent during 2019 period. Among the NILs, P1882-12-111-9 and P1882-12-111-10 showed significant delay in DF under stress during year 2017 than rest of the NILs and PB1. In the same year, there were five NILs that showed significant late flowering than PB1. During 2018, both under stressed as well unstressed conditions, DF of all the NILs were at par with PB1. Similarly, during 2019 season too, DF of all the NILs were similar to PB1, except for P1882-12-111-1 under stress, that showed delayed flowering. On an average, the flowering delay under stress was between one to five days among the NILs. Overall, it was observed that drought stress increased the vegetative period in most of the NILs and checks across all the three years of evaluation (Table 4). Table 3. Analysis of variance (ANOVA) for agronomic performance of NILs in individual years and combined under two stress treatments, stressed and unstressed.

Traits
Seasons

Drought Tolerance Level of NILs Judged through Percent Reduction of Yield and Stress Indices
In order to assess the true tolerance of NILs, percent reduction in yield (%R) and stress indices were calculated. The average values are provided in Table 5. Further, the data for individual lines across different years is provided in Supplementary Table S3. The unstressed yield levels of the 18 NILs were non-significantly different from that of the recurrent parent, PB1 and the donor parent N22 in several cases. However, under drought, the yield level was significantly superior than PB1 in seventeen NILs, and seven NILs showed yield similar to N22. The %R was found to be maximum in PB1 in all the experimental years. The minimum and maximum %R among NILs was found to be 82%

Stepwise Forward Regression Analysis to Identify the Significantly Affected Yield Contributing Trait
To find out the trait which has contributed significantly to yield and hence conditioned primarily by the QTL transferred, a stepwise forward regression was done with BLUP values for individual years and across years by taking the grain yield as the dependent variable (Table 6). SF stood out as the major determinant trait contributing to grain yield consistently across seasons, particularly in 2017, where it alone contributed for 51% of the variation in grain yield under drought conditions. It was also clear that the contribution of SF was more apparent under drought stress than unstressed conditions, from its significant positive coefficients across the seasons. Remaining yield contributing traits did not show strong and stable influence on yield.

Quality Assessment of NILs under Stressed and Unstressed Situations
The recovery grain and cooking quality traits is a major criterion in improvement of Basmati rice. Analysis of the grain quality recovered among the NILs under both stressed as well as unstressed conditions during 2019 season indicated that all the improved lines were as aromatic as PB1, the recurrent parent having a panel score of 2.0 for aroma (Table 7). Visually, the milled grains of the NILs appeared similar to that of PB1. However, some of the NILs showed non-significant variation for kernel length before cooking (KLBC) with respect to PB 1. By and large, similar pattern was observed for all the raw grain quality traits. After cooking, no significant variation was observed for kernel length after cooking (KLAC) than PB1 (Figure 4), among several of the NILs under unstressed conditions. However, under stress there were six NILs that showed significantly superior elongation than PB1, such as P1882-12-111-3, P1882-12-111-5, P1882-12-111-6, P1882-12-111-7, P1882-12-111-8 and P1882-12-111-11. Interestingly, under drought, the elongation ratio was found significantly good in all NILs except in seven NILs, that were on par with PB1. Furthermore, the alkali spreading value (ASV) of all the NILs were similar to PB1 (score of 7) as against the score of 5 recorded among the other checks.

Discussion
Climate change is predicted to affect rice cultivation worldwide adversely, through various impacts, such as drought, excess rainfall, temperature fluctuations and also predisposing it to several biotic stresses. Any adverse effect on rice production would threaten world food security, because more than half of the world population is rice dependent. This is particularly important to rice consuming countries like India, where about 90% of the total rice production is internally consumed [40]. Among these, biotic threats due to diseases and pests, as well as abiotic stresses such as drought, submergence and salinity are particularly relevant under climate change scenario. Predominantly in the rainfed environment of South and Southeast Asia, drought and submergence are more frequently encountered during the crop growing season [41]. Most of the prominent and popular rice varieties are vulnerable to these abiotic stresses [20]. Since climatic vagaries occur spontaneously and intermittently, growing tolerant rice varieties for wide range of stresses is the only economically viable option to manage abiotic stresses. This option is particularly important in the case of drought, because it remains as the most frequently occurring stress that bears the potential to fail a rice crop. Judging drought tolerance of genotypes and its transfer into elite backgrounds is quite laborious due to complex nature of tolerance. It is this complexity that renders the improvement for drought resilience through conventional breeding tardy. However, efforts can be remarkably hastened by leveraging molecular markers that are linked to various traits associated with drought tolerance. Moreover, use of markers can aid a relatively cost-effective and environment neutral selection, while improving the accuracy and reducing the turnover time. Molecular marker assisted introgression of traits provides various advantages such as easy recovery of an otherwise difficult phenotypic trait, easy selection, increased accuracy and shortening the breeding time. Therefore, marker assisted backcross breeding (MABB) was proven to be an efficient strategy for incorporation of desired trait associated genes/QTLs in numerous prominent rice varieties [21,42,43]. MABB along with stringent phenotypic selection was successfully employed for Basmati rice improvement for different biotic stresses [26,30,[44][45][46][47][48]. MABB is also proven to a superior method for QTLs transfer into desired cultivar/variety for different abiotic stresses as well, such as drought. Out of the several QTLs reported for reproductive stage drought tolerance in rice, qDTY1.1 [15,16], qDTY2.1 [33], qDTY3.1 [33,49] and qDTY12.1 [50,51], showed consistent grain yield under drought across different genetic backgrounds and has been used in breeding applications. Recently, Dwivedi et al. [23] reported successful pyramiding of two QTLs, qDTY2.1 and qDTY3.1 into the megavariety, Pusa 44 significantly enhancing yield under drought. Despite the proven advantage, however, only few success stories in improving rice varieties for drought tolerance using MABB has been reported rice.
In the present study, MABB was utilized to introgress a RSDS tolerance QTL qDTY1.1 from N22 into PB1, one of the popular Basmati rice varieties of India. Popular for its high yield, better-quality grain, and excellent cooking quality and pleasing aroma, PB1 is the first ever semi-dwarf Basmati rice variety developed. Bred through conventional convergent breeding procedures, PB1 was derived from a cross between Pusa 150/Karnal local. Pusa 150 has Basmati 370 in its lineage, while Karnal local was a landrace with better aroma and cooking quality. Karnal local was a selection from Haryana Basmati collection 19 (HBC19), that was later released as Taraori Basmati [52]. Despite its excellent yield gain over the conventional Basmati cultivars, PB1 was semi-dwarf with excellent plant architecture, photo insensitivity, high yielding and with unparalleled grain quality [53]. However, like its congeners, PB1 is also sensitive to drought as well as to many biotic stresses. Released during 1989, PB1 remains popular even today among the Basmati farmers of India, despite above limitations. Furthermore, till date there is no report on improvement any Basmati rice variety for drought tolerance. Therefore, the present was carried out with the objective to incorporate drought tolerance in PB1. The donor parent, N22 is a tall upland rice variety belonging to aus group. N22 is a pure line selection from a landrace, Rajbhog [54] that possesses deeper root system, shorter duration and tolerant to heat and drought [55]. This variety has short bold non-aromatic grains with very less elongation on cooking.
The genome wide average polymorphism between the parents, PB1 and N22 was 17.35%, which was more than that of the carrier chromosome of qDTY1.1. The molecular analyses on the background recovery has revealed maximum recovery of chromosome 1, thereby eliminating the chances for linkage drag. The fact that maximum background markers (101 markers) were surveyed on chromosome 1, substantiate our claim for the elimination of linkage drag. Furthermore, the agronomic recovery of the NILs in general suggests that no undesirable trait has been incorporated by chance from other donor fragments, that might have escaped marker-based background selection. In this study, qDTY1.1 was found providing increased level of drought tolerance in the NILs vis-à-vis PB1. It is also known that QTLs from a particular genetic background usually show minor effects or may remain completely silent in diverse genetic backgrounds [56]. In the case of drought, interaction between QTL × genetic background has been a major bottleneck limiting the use of QTLs for MABB in rice [50,57,58]. Nevertheless, qDTY1.1 derived from N22 has been demonstrated effective in multiple non-Basmati backgrounds such as MTU1010, IR64 and Swarna, which showed its role as a major effective QTL governing grain yield under reproductive stage water stress [15]. This justifies the use of qDTY1.1 as a most desirable candidate for MABB to improve prominent high-yielding varieties with augmented drought tolerance.
P1882-12-111-17 had the maximum grain yield among the NILs followed by P1882-12-111-3, P1882-12-111-5, P1882-12-111-6, P1882-12-111-7 and P1882-12-111-12 under both stressed, and unstressed conditions during all the years. This signified the effectiveness of qDTY1.1 in enhancing yield under drought stress. However, the effect of the introgressed QTL among the PB1 NILs was more conspicuous in the 2019 than in the previous years. In the recent report from World Meteorological Organization [59], it has been reported that climate change is accelerating by the last five-year period between 2015-2019. Although 2016, has been identified as the most erratic year, 2019 also showed significantly high incidence of drought than 2017 and 2018. Therefore, 2019 season data were more explicit to showcase the drought tolerance response of the NILs. This was evident from performance of NILs vis-à-vis PB1, which showed significant deviation for traits such as SF, GY, DF and PH particularly during 2019 and 2017 seasons. In all the seasons, however, the donor (N22) and the positive control (IR86918-B-B-305) remained the best performers under drought possibly indicating the necessity to transfer additional QTLs from multiple donors to further push the tolerance threshold of PB1. Similar observations were made by Vikram et al. [15] that qDTY1.1 had a considerable effect on enhancing the GY under both stressed and unstressed conditions. Similar effect of qDTY1.1 on improving the yield was earlier reported, conditioned through several adventitious drought-linked traits such as root characteristics, relative water content, biomass, and osmotic adjustment [17,60,61]. Meta QTLs have also been documented for maximum root length and GY within the qDTY1.1 region [62,63]. In the PB1 background, qDTY1.1 seems to enhance the yield under drought indirectly by modulating spikelet fertility as inferred from stepwise regression analysis. This implies on the importance of maintaining higher spikelet filling in achieving higher yields. Therefore, we hypothesize that grain filling could be the putative trait, qDTY1.1 is associated with, that require further validation.
Stability of yield performance under stress situations across the seasons is also important for a NIL to be considered for varietal evaluation prior to release as a cultivar across different rice growing areas of the country. The AMMI stability analysis carried out towards assessing the stable yielding potential, revealed that the NILs showed differential stability pattern. Two concomitant stability indices AST and YSI were used for assessing the yield stability, while AST alone was used for SF. Considering the stability indices and mean performance, one of the NILs, P1882-12-111-12 was adjudged stable for both GY and SF. This line can be a potential candidate for varietal evaluation pipeline in the future.
Among the agronomic traits, it was found that there was a delay in DF found across the NILs under stressed conditions. Delay in the flowering time under drought was also reported earlier in rice [16,64,65], which is due to delay in flower development and slow rate of panicle elongation under stress. The delay was however conspicuous in IR64, the short duration high yielding check variety used in the study. In short duration varieties with little drought adaptation, flower development delay occurs more significantly than longer duration cultivars. Similarly, there was a general decrease in PH among all the lines under stress, the degree of decrease was slightly reduced among the NILs. However, there was no increase in height observed among the NILs, although they have all possessed qDTY1.1, attributable to the tight linkage of qDTY1.1 with SD1 gene, the gene responsible for tallness in N22 [66]. In the present study, we surmise that linkage between the qDTY1.1 and sd1 allele might have broken resulting in semi-dwarf NILs with height similar to that of PB1.
Tolerance indices are better yardsticks for judging the tolerance of genotypes than the yield per se under stress condition. Three popular stress indices were utilized in the present study to identify best performing NILs. DYI is based on the mixed model which accounts for the genotype × stress level interaction across different environments. For drought prone areas, the combination of DYI with deviations in genotype performance under irrigated conditions may facilitate breeders to select genotypes with no yield reduction under favorable environment in comparison to currently cultivated varieties [37]. STI sorts out the entries which perform well under nonstress and fairly well under stress condition, high value of STI for a genotype infers higher tolerance level to the drought stress [39].
Fisher and Maurer [39] proposed SSI that estimates the yield reduction due to unfavorable environment in comparison to favorable environment. Lower the SSI value lesser the yield difference between stress and non-stress condition and hence greater is the tolerance of the genotype [67]. Thus, DYI, STI and SSI favored P1882-12-111-17 as best genotype for drought situation due to its highest grain yield under both stressed and unstressed conditions followed by P1882-12-111-3, P1882-12-111-6, P1882-12-111-7, P1882-12-111-12, P1882-12-111-15 and P1882-12-111-20. Further, P1882-12-111-17 out yielded majority of the remaining NILs. Among various traits measured in the present study, SF stood out as the most significant trait influencing grain yield under drought stress. Successful reproduction and efficient grain filling are the important determinants of ultimate grain yield under stress which is reflected as higher spikelet fertility. Thus, the NILs showed superior yields due to the maintenance of spikelet fertility under drought stress.
PB1 being a Basmati rice variety having specialty grain characteristics, transfer of drought tolerance from a non-aromatic variety with totally different grain architecture was a great challenge. Since the QTL transfer was into a Basmati background from a non-Basmati donor, a critical comprehensive quality check was done under both non-stress and stress situations in 2019. The NILs generated, possessed all the grain and cooking quality attributes of PB1 along with drought tolerance (Figure 4).
The recovery of quality in the NILs was achieved by stringent phenotypic selection integrated to the marker-based selection for the target trait, RSDS. NILs under well-watered treatment has grain quality similar to that of PB1, while an inconspicuous quality variation was noticed under drought situation. In spite of the successful recovery of PB1 genome component governing grain quality among the NILs, there was a marginal reduction in head rice recovery noticed under severe stress, although statistically insignificant, together with a slight increase in chalkiness. It is well known that drought at grain filling interferes with starch packaging in the grains, leading to chalkiness. Besides, the grain quality under stressed treatment was not significantly different from that under unstressed, indicating that drought had little effect of grain quality in the NILs. Grain quality recovery using MABB with augmented phenotypic selection, particularly in Basmati cultivars, was earlier reported from several studies [27,30,68].

Conclusions
In the present investigation, we have developed NILs of the popular Basmati rice variety, PB1 carrying a major QTL, qDTY1.1 for RSDS tolerance. This paper also forms the first ever report of successful transfer of a drought tolerance QTL into a Basmati rice cultivar. The improved drought tolerant NILs of PB1, were also combined with high yield and grain quality. These NILs could be a good alternative for the Basmati growing regions with limited soil moisture regimes. Additionally, these may help farmers to reduce number of irrigations without foregoing yield and grain quality potential of Basmati rice. These NILs are assessed for three consecutive seasons to assess their stability and drought response to identify potential candidates to be deployed for varietal testing and cultivar release. Moreover, these NILs can also serve as improved donor lines for imparting drought tolerance in future Basmati breeding programs.

Conflicts of Interest:
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.