Development of introgression lines in high yielding, semi-dwarf genetic backgrounds to enable improvement of modern rice varieties for tolerance to multiple abiotic stresses free from undesirable linkage drag

Occurrence of multiple abiotic stresses in a single crop season has become more frequent than before. Most of the traditional donors possessing tolerance to abiotic stresses are tall, low-yielding with poor grain quality. To facilitate efficient use of complex polygenic traits in rice molecular breeding research, we undertook development of introgression lines in background of high-yielding, semi-dwarf varieties with good grain quality. The study reports the development and evaluations of over 25,000 introgression lines in eleven elite rice genetic backgrounds for improvement of yield under multiple abiotic-stresses such as drought, flood, high/low temperature. The developed introgression lines within each genetic background are near isogenic/recombinant inbred lines to their recipient recurrent parent with 50 to 98% background recovery and additionally carry QTLs/genes for abiotic stresses. The multiple-stress tolerant pyramided breeding lines combining high yield under normal situation and good yield under moderate to severe reproductive-stage drought, semi-dwarf plant type with good grain quality traits have been developed. The introgression lines in dwarf backgrounds open new opportunity to improve other varieties without any linkage drag as well as facilitate cloning of QTLs, identification and functional characterization of candidate genes, mechanisms associated with targeted QTLs and the genetic networks underlying complex polygenic traits.

occurrence of multiple abiotic stresses in a single crop season has become more frequent than before. Most of the traditional donors possessing tolerance to abiotic stresses are tall, low-yielding with poor grain quality. To facilitate efficient use of complex polygenic traits in rice molecular breeding research, we undertook development of introgression lines in background of high-yielding, semidwarf varieties with good grain quality. the study reports the development and evaluations of over 25,000 introgression lines in eleven elite rice genetic backgrounds for improvement of yield under multiple abiotic-stresses such as drought, flood, high/low temperature. The developed introgression lines within each genetic background are near isogenic/recombinant inbred lines to their recipient recurrent parent with 50 to 98% background recovery and additionally carry QTLs/genes for abiotic stresses. the multiple-stress tolerant pyramided breeding lines combining high yield under normal situation and good yield under moderate to severe reproductive-stage drought, semi-dwarf plant type with good grain quality traits have been developed. the introgression lines in dwarf backgrounds open new opportunity to improve other varieties without any linkage drag as well as facilitate cloning of QTLs, identification and functional characterization of candidate genes, mechanisms associated with targeted QTLs and the genetic networks underlying complex polygenic traits.

Genotypic characterization of introgression lines in different backgrounds.
The genotypic selection (foreground and background selection) across generation advancements in all backgrounds were carried out to identify the high yielding stress tolerant promising lines with qDTY and qDTY combinations with high recurrent parent genome recovery (RPG) and free from linkage drag. The foreground/background selections were carried out and only the lines with desired phenotype and genotype (qDTY/qDTY combinations) were advanced to the next generation. The detailed information on the number of introgressed lines that were characterized genotypically in all the studied background in each season is presented in  Fig. S2). The recurrent parent genome recovery ranged from 89 to 93% (Table 1, Fig. S2). The grain yield advantage of introgression lines in TDK1-Sub1 background varied from 302 kg ha -1 (qDTY 6.1 + qDTY 6.2 ) to 465 kg ha -1 (qDTY 3.1 + qDTY 6.1 + qDTY 6 www.nature.com/scientificreports/ 2656 kg ha -1 (qDTY 3.1 + qDTY 6.1 + Sub1) under moderate stress and 178 kg ha -1 (qDTY 6.1 + qDTY 6.2 ) to 317 kg ha -1 (qDTY 3.1 + qDTY 6.1 + qDTY 6.2 ) under non-stress condition compared to the recurrent parent (  Fig. S6). The large reduction in grain yield under severe stress compared to non-stress was observed in the introgression lines in TDK1-Sub1 ( (Fig. S8) backgrounds. Under moderate stress grain yield improvement of 252 kg ha -1 (qDTY 3.1 + qDTY 6.1 + qDTY 6.2 ) over MTU1010 was observed in the drought tolerant introgression lines in MTU1010 background. The grain yield advantage under severe drought stress, moderate drought stress and control non-stress conditions was 495, 991 and 1621 kg ha -1 , respectively in Swarna * 3/Moroberekan introgression lines (qDTY 3.2 ) compared to the recurrent parent (Table 1, (Fig. S7).
The stringent phenotypic selection was employed to select multiple stress tolerance lines possessing tolerance to low temperature, high-temperature and drought stress.  Table 2). The variation in low temperature tolerance at germination and seedling stage was observed. IR 128762-187-1-B-B-B-B was found to be strongly low temperature tolerant at seedling stage and tolerant at germination stage, IR 128762-83-B-B-B-B-B was moderately low temperature tolerant at seedling stage and tolerant at germination stage and IR 128762-155-2-B-B-B-B was strongly low temperature tolerant at seedling stage and tolerant at germination stage ( Table 2). The yield potential of selected lines under non-stress condition ranged from 5329 to 6617 kg ha -1 , under RS from 2111 to 2571 kg ha -1 and under high-temperature stress from 3694 to 4394 kg ha -1 ( Table 2). The percentage reduction in grain yield (61%) under reproductive stage drought stress was more as compared to the yield reduction under reproductive stage high-temperature stress (34%) ( Table 2).  Table 3.
Genomic and environmental interactions. Despite being the large effect QTLs reported for the grain yield under drought stress at reproductive stage, the near isogenic lines with one, two, three or multiple QTLs showed grain yield variability in each of the genetic background. The contrasting performance of near isogenic lines possessing similar qDTY and also qDTY combinations may be due to the epistatic interactions of the introgressed QTLs among themselves and with the background marker loci under variable stresses level. The digenic interactions between marker loci and with the background were detected in low yielding nearly isogenic lines in IR64-Sub1, TDK1-Sub1, Savitri, MR219 and Samba Mahsuri backgrounds 17,22 . An epistatic interaction of qDTY 12 www.nature.com/scientificreports/

Discussion
Development of introgression lines have been extensively employed in the identification and validation of the useful effects of genomic regions from the donor parents. Majority of the QTLs that have been detected so far in many studies 20,61-72 and very few of them have been exploited and validated further in breeding programs. Some of the candidate genes that have already been cloned 73 could be exploited in future breeding programs.
The genetic backgrounds such as Swarna, Swarna-Sub1, IR64, Samba Mahsuri, Vandana, Anjali, Kalinga III, MTU1010, Savitri, MR219 and TDK1-Sub1 that were used in this study to develop introgression lines are widely cultivated and accepted by the farmer in South and South-East Asian countries such as India, Nepal, Malaysia and Lao PDR. All these have production bottleneck due to their susceptibility towards drought, heat and low temperature stresses. Rice breeding at IRRI has made a significant progress in finding the key genomic regions associated with drought as well as submergence tolerance. Majority of the large-effect grain yield drought QTLs and Sub1 for submergence have shown consistency in performances and explained a large part of phenotypic variances 18,[23][24][25]29,30 .
The loci contributing grain yield improvement under drought stress reported to be collocated with the loci governing early days to flowering, plant height and reduced yield under irrigated conditions. The large segregating populations have been developed in Swarna-Sub1 and IR64 backgrounds to break these linkages and semi-dwarf medium duration lines have been successfully developed 74 . Apart from this, multilocational evaluations of the promising breeding lines offer opportunities to understand the QTL x environment interactions.
Combining multiple stress tolerances in single background of high yielding breeding line/variety are the breeder's goals that have been targeted for a long time in the past. This is primarily because, in the years with well-characterised climate, the multiple stress-tolerant varieties should provide yield comparable to the popular www.nature.com/scientificreports/ high-yielding varieties in that area but shall also make the introgression of multiple lines easy, effective and without any undesirable effect. Rice drought molecular breeding at IRRI has witnessed some of the successful example of marker assisted introgression of drought QTLs in the elite backgrounds 14,15,17,18 . The developed multiple stress tolerant introgression lines offer opportunities to the rice community to dissect the genetic variations existing among the introgression lines and to identify potentially useful pre-breeding lines to be serve as donors. In addition, very limited efforts have been performed to identify the QTL × background interactions influencing the overall expression of the target trait. There is a strong need to find such kind of unknown background noises for the effective deployment of the QTLs in marker-assisted breeding program. The developed introgression lines in different genetic backgrounds in the present study would be an important genetic material to predict such genetic interactions which complicates the genotype-phenotype relationship of complex quantitative traits 75,76 such as drought, low temperature and high temperature tolerance. The developed introgression lines in different genetic backgrounds with same QTLs combination showed contrasting yield under different level of drought stress. The developed introgression lines in background of TDK1 Sub1, Samba Mahsuri, Savitri, MR219, MTU1010, Vandana, Anjali and Kalinga III can further be used for the genetic dissection of epistatic interactions contributing to the grain yield variability in rice under drought.
The complexity of genetic control of complex traits lead to wide variations in the performance of introgression lines. It has been observed that the same QTL showed variable performance across different backgrounds and environments. For a successful marker-assisted introgression program, it is very important to choose QTLs with large genetic effect and stability across wide range of environmental conditions and drought intensities 25,77 . Selection of the lines with higher yield based on phenotypic selection provided an opportunity to develop lines with positive interactions.
The marker-assisted breeding programme involving introgression of at least 2-3 major QTLs in different genetic backgrounds has permitted the development of many high-yielding multiple stress-tolerant lines. The introgression of 2-3 major QTLs in the background of widely cultivated popular high-yielding rice varieties lead an economic grain yield improvement of 1.0-1.2 t ha -1 under severe drought stress in farmers' fields. The yield superiority of multiple stress tolerant lines across backgrounds and over generations clearly indicated that the tolerance of drought, submergence, low temperature and high temperature can be successfully combined even though these traits are controlled by different physiological mechanisms. The introgression lines possessed better grain quality than the traditional donors and so the development of multiple stress tolerant variety now with high quality has never been as easy as now. The development and release of marker-assisted breeding product for the rainfed lowland areas in IR64 and Swarna-Sub1 78 backgrounds are the successful examples that should instigate the rice breeders to deploy the identified QTLs in the marker-assisted breeding programs targeting grain yield improvement under multiple abiotic stresses.

conclusions
The near isogenic lines with one, two, three and more QTLs have been developed in -eleven different elite backgrounds by the integration of phenotypic-genotypic selection approach. These near isogenic lines are free from linkage drag and hence we are able to combine increased yield under drought with high yield potential, dwarf plant type and good grain quality. Development of high-yielding and multiple stress tolerant introgression lines in wide range of backgrounds and for two diverse ecosystems validate the effectiveness of QTLs and marker-assisted introgression program. Development of such multiple stress tolerant rice varieties will provide opportunities to use these introgressed lines across different rice ecosystems to reduce yield under abiotic stresses and have an impact on increase in rice production and overall increase in the farm output in stress prone areas of South and South East Asia. The near isogenic lines with combination of QTLs brings a great advantage to breeding by providing opportunities to breeders to introgress QTLs in less time by just attempting one or two crosses as against three to four years that were spend earlier just to bring multiple QTLs together. The study clearly indicated that it is necessary to identify and break such linkage drags for effective use of QTLs in the trait development or breeding programs. The study also validated that once linkage drags are removed, it is possible to combine tolerance to multiple traits controlled by diverse physiological mechanisms-low temperature and high temperature; drought and low temperature, high yield and drought tolerance and low temperature. The findings from the current study can revolutionize the marker assisted breeding as very few of the large number of QTLs identified have been used in the past. The developed introgression lines in different genetic backgrounds are unique set of germplasm to the interaction studies to break the grain yield improvement barrier under multiple stresses. In addition, these lines may serve as an important material to study the functional characterization of candidate genes and genetic networks utilizing advances in functional genomics.

Materials and methods
The present study was conducted at International Rice Research Institute (IRRI), Los Baños, Laguna (Philippines) and IRRI South Asia breeding hub (SAH), Hyderabad, India during years from 2012 to 2017.
Breeding for drought tolerance at iRRi, philippines. Plant material. The first step of breeding program is to identify the appropriate donor exploiting the genetic variations in the existing germplasm. To avoid undesirable linkages associated with the traditional donors, improved breeding lines with genomic regions associated with different abiotic stresses were used in the present study 79  Phenotyping screening. The screening of marker-assisted derived breeding lines was conducted under lowland and/or upland reproductive stage drought stress (RS) and irrigated control/non-stress (NS) conditions at IRRI (International Rice Research Institute). The experiments were planted in two to three replications in single or two or four-rows plots with row length of 5 m and 2.0 to 3.0 m in lowland and upland condition, respectively in an alpha lattice design. The lowland experiments involved transplanting that involved the transplantation of 21-days-old seedlings in the field with a single seedling per hill. The long duration introgression lines were planted 15-20 days earlier than the medium duration varieties. The field management activities of the lowland experiments were carried out in similar way as described earlier by Venuprasad et al. 24 . The detailed information on number of entries tested, generation, experimental design, treatments and number of replications is presented in Table S1.
Upland experiments were established under dry direct seeded conditions by maintaining 20 cm × 20 cm between row to row and plant to plant distance. Field management practises under upland was done as described by Bernier et al. 23 . In brief, under upland conditions, combination of Oxadiazon (pre-emergence herbicide) at 6 days after seeding (DAS) @ 0.5 kg ai ha -1 , Bispyribac sodium (early post emergence) at 11 and 22 DAS @ 0.03 kg ai ha -1 in addition to spot weeding at 35 and 55 DAS were used to control the weeds. IPM (Integrated pest management) practices was followed involving ditrac @ 0.05 g kg -1 (0.005%), brodifacoum as rat bait to control rats, and pre-seeding application of the 0.075 kg ai ha -1 of Fipronil along bunds and the plot edges at 7 DAS.
Lowland drought screening. At 30 days after transplanting (DAT), complete drainage of water was carried out and the drought stress at reproductive stage was initiated by holding irrigation. The initiation of reproductive stage drought stress for the long duration lines were adjusted based on their reproductive cycle. When ~ 75% of the population showed symptoms of severe leaf rolling and the water table depth remained > 100 cm for approximately > 2 weeks; a life-saving irrigation was applied through the flash flooding. After 24 h, complete drainage of the water from the field was carried out again to impose the second cycle of drought stress. The 1.1 m PVC (polyvinyl chloride) pipe was installed in the soil to measure the water table depth at regular intervals in the experimental fields. The depletion of the water table level was measured using a meter scale day-to-day after the onset of the drought stress.
Upland drought screening. The upland experiments were irrigated using sprinkler-irrigation up to 45 DAS, with application of irrigation twice a week during seedling establishment stage and in early vegetative developmental stage. Thereafter, stress at reproductive stage was initiated by holding irrigation. The plots were reirrigated when the tensiometer at 30 cm soil depth showed reading below -50 kPa and most of the lines showed wilting and leaf drying. This kind of cyclic stress is effective in drought screening of populations consisting of variable genotypes having broad range of growth duration 80, ensuring all breeding lines receive an adequate level of stress during the reproductive stage of development. All the cultural practices were same in upland non-stress and reproductive stage stress experiments, except that in non-stress experiments the irrigation was continued two times a week up to 10 days before the harvest of the crop.
Biotic stress screening. The introgression lines at F 2 and F 7 /F 8 stages were also evaluated for the disease resistances such as blast which is caused by Magnaporthe oryzae and bacterial blight which is caused by Xanthomonas oryzae pv. Oryzae. The blast screening was done using the mixed inoculum for the races presents in the Philippines at 10 DAS in the blast nursery. The plants possessing genes viz. pik-s, pi2, pi5(t) and pi9 showed resistance against the inoculum. The inoculation and screening for the bacterial blight race 1, PXO61 and race 2, PXO86 was done at maximum tillering stage in field following Kauffman 81 . The resistant or susceptibility of introgression lines against the blast and bacterial blight diseases was observed on IRRI, SES scale 82 .
Phenotypic data collection. The agronomic data on days to 50% flowering (DTF), plant height (PHT), and grain yield (GY) were recorded for all the experiments. DTF was recorded when ~ 50% the plants in a plot showed panicle exertion. At maturity, three random plants per plot were measured from the soil surface to the tip of Scientific RepoRtS | (2020) 10:13073 | https://doi.org/10.1038/s41598-020-70132-9 www.nature.com/scientificreports/ the highest panicle on main tiller to record the PHT (cm). The plant samples at maturity were harvested, dried to 14% of the moisture before weighing, and weights to record the GY (kg ha -1 ). The developed introgression lines were screened under both reproductive-stage drought (RS) and irrigated control non-stress (NS) conditions along with the recipient parent, high-yielding check varieties and drought-tolerant donors. The imposed reproductive stage drought stress was classified into severe and moderate drought stress based on the relative reduction in the grain yield compared to non-stress 79 . The experiments with yield reduction of < 70% were characterize as severe stress (SS) and with yield reduction of 31-70% were categorized as moderate stress (MS phenotypic screening and data collection. The phenotypic screening of breeding lines for the drought stress at reproductive stage was carried under natural conditions as above-mentioned procedure in both dry and wet seasons. The low temperature and high-temperature screening were carried based on the SES (Standard evaluation system for rice) 84 . Germination vigor and the seedling survival rate were two main criteria that were used for the evaluation of low temperature tolerance in rice at germination stage 85 . The high-temperature screening at reproductive stage and low temperature tolerance screening at germination stage was carried out under field condition in dry season while the low temperature tolerance screening at seedling stage was conducted under field in the dry season as well as under controlled condition in the wet seasons. To compare the performance of breeding lines under treatment and non-stress (control) conditions, non-stress experiments were carried out for each drought, heat and low temperature tolerance experiments. Three plants per introgression line were chosen for the measurement of all traits. The phenotypic data on DTF, PHT, number of productive tillers (counted manually), flag leaf length (FLL,cm, panicle length (PL,cm, number of filled grains per panicle (FG/P, spikelet fertility (SF; %, 1000-grain weight (1000 GW; g), grain yield (GY; kg ha -1 ) and grain type was recorded.

Genotyping of the developed introgression lines in different backgrounds.
In the markerassisted introgression program, rice microsatellite (SSR; simple sequence repeats) markers were extensively used. Earlier reported SSR markers associated with the various grain yield qDTYs 15,17,18,[23][24][25][26][27][28] were used to detect the targeted region in the developed introgression lines in various backgrounds (Table S2). The fresh-young leaves from the 21 days old seedlings were collected from each introgression line and the respective parents. The genomic DNA was extracted using the modified CTAB protocol 86 . Polymerase chain reaction (PCR) amplification was performed with each PCR reaction mixture of 15 μl containing DNA (10 ng), PCR buffer (1 ×), dNTPs (100 μM), oligonucleotide primers (100 μM) and Taq polymerase (1 unit). The PCR products were then resolved on high-resolution PAGE (polyacrylamide gel electrophoresis) (CBS scientific, MGV-202-33) ranged from 6-8% (v/v) depending on the SSR markers product size. The separated fragments of DNA were stained with the SYBER Safe™ before visualizing under UV trans-illuminator (AlphaImager™ System). The strong associations of the reported markers with the introgressed regions and the co-dominance nature of SSRs allowed the easy detection of heterozygotes. As many of the identified qDTYs regions that were used in the introgression program were not fine mapped, all SSRs present within the identified genomic regions/QTLs were used in the genotyping of the introgression lines to avoid a loss of the candidate genes associated with abiotic stresses tolerance (drought, low temperature and high temperature) in the region due to the crossover. To reduce the forward genotyping cost and labor, sequential genotyping approach was used. Firstly, the peak maker for each QTL was used to genotype the F 2 segregants, then the breeding lines harbouring the respective donor allele at peak marker were further genotyped with the flanking markers, followed by genotyping with all the markers underlying the QTL regions.
Background recovery. To study the background recovery of the selected introgression lines, background genotyping was carried out. A total of ~ 600 SSR markers equally distributed on all the 12 chromosomes (https :// www.grame ne.org/) were used to study the polymorphism between the parents. The selected polymorphic SSR markers; 156, 111, 101, 89, 80, and 102 SSR markers were used to estimate the background recovery percentage in Swarna, IR64, TDK1-Sub1, Savitri, Samba Mahsuri, MR219 backgrounds, respectively were chosen. Background study has not been conducted yet in Anjali and Kalinga backgrounds. The background recovery percentage (BC%) of the selected introgression lines in different backgrounds were calculated as Vishwakarma et al. 87 . www.nature.com/scientificreports/ where B is the SSR marker loci homozygous for the genetic background; H is the number of markers loci in heterozygous; N is the total number of polymorphic markers that were used for the background genotyping.
Statistical analysis. The agronomic data collected from all the experiments were analysed statistically.
The experiment means and SED (standard error of difference) were calculated using the statistical software PBTools v1.4 developed at IRRI. The LSD (Least significant difference) at P = 5% and 1% significance were used to make the comparison of the test entries means and to estimate the significant differences of the traits of interest between the parents and the introgression lines. The mixed linear model was used to calculate the analysis of variance: where µ is the overall mean, Gi is the effect of i th genotype, Rj is the effect of j th replicate, BK (Rj) is block effect of the j th replicate and eijk is the error. Genotypes were kept as fixed and the replication and block effects were kept random for estimating entry means.