Development and Characterization of an Ecient Mutant Resource of the Mega Variety, Samba Mahsuri for Rice Improvement

Novel variants of rice (Oryza sativa L.) can be obtained from induced mutations. The objective of the present study is to induce mutations in the background of Samba Mahsuri (BPT-5204), a popular and mega rice variety of India for creating novel variants for morphological, physiological and biotic stresses through using Ethyl Methane Sulphonate (EMS). Results: A population derived from 10, 500 M 1 plants and their descendants was phenotyped for a wide range of traits leading to the identication of 124 mutants having variations in key agro-morphological traits, and 106 mutants exhibited variation for physiological traits. Higher yield is the ultimate goal of crop improvement and we identied 574 mutants having higher yield compared to wild type by having better yield attributing traits. Further, a total of fty mutants showed better panicle exsertion phenotypes as compared to Samba Mahsuri leading to enhancement of yield. Upon rigorous screening for three major biotic stresses, nine mutants showed enhanced tolerance for yellow stem borer, and thirteen different mutants each showed enhanced tolerance for sheath blight and bacterial leaf blight, respectively. On the whole, 1406 desired mutant lines identied at M 2 were forwarded to an advanced generation (M 5 ). Whole genome re-sequencing and analyses of 15 stable mutants revealed high level of similarities (99.91 to 99.99%) with the Samba Mahsuri. Conclusions: The lines showing enhanced tolerance to important biotic stresses (YSB, ShB and BLB) as well as several economically important traits are unique genetic resources which can be utilized for the identication of novel genes/alleles for different traits. The lines which have better agronomic features can be used as pre-breeding lines. The entire mutagenic population is maintained as a national resource for genetic improvement of the rice crop.


Background
Rice is a staple food for almost 56% of the world's population and it is the source of 20% of the world's dietary energy supply.
Global rice consumption is projected to increase from 500 million to around 750 million tons by 2050 (Pathak et al. 2018).Hence, rice production has to be enhanced (50% projected growth) to match the increased requirements for the future.Development of genetic stocks suitable to the various agro-climatic zones is being done based on selection among the improved cultivars or through bringing improvements of mega varieties by modifying speci c traits.This is often leading to a narrow genetic base which results in low variability and consequently limits on the possibilities for recombination and genetic segregation, leading to the reduced genetic gains (Fu, 2015).Therefore, in spite of the development and release of numerous rice varieties and hybrids, the genetic gain has been relatively low (Singh et al. 2016).Hence, there is an ample need to enhance available variation amongst the rice germplasm.
Induced mutations are one of the best available options to create variation among the well adapted mega varieties.Mutations induced arti cially using physical agents like fast neutron (Ruengphayak et (Suzuki et al. 2008), sodium azide (SA) (Awan et al. 1980 andTai et al. 2016) have been employed in rice to create mutations.Chemical or physical mutagens have higher mutation e ciencies as compared to insertions created by T-DNA (Ram et al. 2019), transposable elements (Greco et al. 2001, Xuan et al. 2016) or RNAi (Wang et al. 2013), TALEN-based gene editing (Li et al. 2012), CRISPR/Cas9 genome editing (Xie et al. 2015), etc.Among the mutagens, EMS is being used frequently owing to its production of high frequency (2 to10 mutations/ Mb) of single nucleotide changes (point mutations) by alkylation of speci c nucleotides (Till et al. 2007; Viana et al. 2019) and a relatively small population, ca.10,000 plants is su cient to saturate the genome with mutations.According to FAO/IAEA-MVD reports (2020) (www.mvd.iaea.org), a total of 828 mutant varieties were developed in rice among them 60 were derived through chemical mutagenesis.Till date using chemical mutagens, 6 mutant resources generated in diverse Japonica cultivars viz.Nipponbare, Tainung 67, Kitaano, TC65, Yukihikari, Kinmaze, and BRS Querencia and three mutant resources in indica (IR64, Kasalath and SSBM) and one aus (Nagina22) were developed (Li et al. 2017).The International Rice Functional Genomics Consortium announced the public availability of more than 200,000 rice mutant lines, which represent mutations in about half of the known functional genes mapped for rice to date (Krishnan et al. 2009).In India, one EMS mutant resource was developed in the genetic background of Nagina 22 (N22) which is a drought and heat tolerant aus rice variety (Mohapatra et al. 2014).This mutation resource is useful in functional genomics studies.Samba Mashuri (BPT-5204) is a mega variety of India that was released in the year 1984 and is now also grown in other Asian countries.It is a long duration (145-150 days) ne-grain type variety, yielding an average of 5-6 t ha -1 with excellent cooking quality and has wide adaptability.However, this variety is susceptible to biotic stresses like sheath blight, blast, bacterial leaf blight (BLB) and insect pests such as yellow stem borer (YSB), Brown plant hopper (BPH) and gall midge.Furthermore, there are varieties that yield more and are also more adapted to abiotic stresses.In spite of these constraints, it is favored by consumers and farmers in India because of its excellent cooking and eating quality characteristics.Enhancing genetic variability in the Samba Mashuri would be useful to tackle some of the existing constraints of this variety.Furthermore the excellent combining ability of this variety (Sundaram 2018) will help in transferring such traits into other varietal backgrounds.Because of the above, Samba Mashuri is an ideal genotype to develop a comprehensive mutant population.
Currently, mutagenesis coupled with molecular mapping and development of molecular markers has become an important tool with excellent potential for crop improvement (Kharkwal and Shu, 2009).Markers identi ed through next-generation sequencing technologies (NGS) are an e cient and cost-effective approach to characterize the variants in mutant collections (McCallum et al. 2000).Mutant resources are being used not only for creating the variation but also for the novel gene identi cation and deciphering gene function (Xu et al. 2017).To generate variability in the background of Samba Mahsuri, the current research focused on developing mutants using EMS and characterized them for various morphological, physiological, yield traits and biotic stress, so as to be able to use these mutants as donors in rice improvement programmes.

Mutagenesis of rice seeds
The rice variety, Samba Mahsuri (Nucleus seed obtained from Baptla Rice Research station, Andhra Pradesh, India) was used for mutagenic treatment.Based on kill curve analysis the seeds were treated in two different concentrations of Ethyl methane sulphonate (EMS) i.e. 1.2% (1-6819) and 0.8 % (6820-10500) by M/s Bench Biotech Company, Gujarat, India.The duration of treatment was about 12 hrs and the treated seeds were washed thoroughly under running tap water for 6 h to leach out the residual chemical.

Development of M 1 and M 2 generations
A total of 10500 M 1 plants were raised from mutagenized (M 0 ) seeds in the elds of M/s Bench Biotech Company, Gujarat.These

Phenotypic characterization of the Samba Mahsuri mutants
A large subset of 10,500 M 2 lines were subjected to various morphological (Plant height, Pink apicules, Albino, Xantha and Coloration), agronomical (panicle types, complete panicle emergence, early owering, grain types and high yield with high grain number), physiological (strong culm, different types of ag leaf, sterile plant, stay green and shattering) and biotic stresses (sheath blight, bacterial leaf blight and yellow stem borer) studies.
Plant height was measured from the base of the plant to the tip of the tallest panicle.High yielding mutants were phenotyped based on the yield parameters like number of tillers, number of panicles, panicle length (cm), grain number per panicle and yield per plant (g) at maturity (Singh et al. 2018).The number of days from seeding to grain ripening (85% of grains on panicle are mature) was recorded and maturity duration of the mutants was observed.The lines that matured earlier to the wild type (Samba Mashuri) were selected as early maturing mutants.At the heading stage, mutants having length of > 2cm of uppermost internode were selected as EUI (Elongated Upper Internode) mutants (Xu et al. 2004).The mutants which showed complete emergence of panicle not having any choking in the ag leaf but not showing the elongated upper internode were considered as complete panicle emergence (CPE) mutants.The grain lengths were measured by using a digital Vernier caliper (M/s Mitutoyo, USA) and classi ed based on length and breadth of grain (Ramkumar et al. 2010).For the identi cation of strong culm mutants, the diameter and physical strength of the culm was taken into account.Culm diameter was measured using a Digital sliding Vernier caliper in the eld from the plants at a height of 30cm above ground level (Yadav et al. 2017), while the physical strength of the culms was measured with a 'prostrate tester' (DIK-7401 Daiki Rika Kogyo co Ltd, Tokyo, Japan) where the plant was pushed at an angle of 45 • to calculate the pushing resistance (Hai et al. 2005).
In each generation, the mutants with the trait of interest were identi ed and advanced to the next generations in ear to row method.The whole mutant population was advanced up to M 4 generation considering the diversity.Later on, from M 5 generation, the advancement of mutants was done based on yield and uniformity.The data for the above-mentioned traits were recorded at three different crop growth stages i.e. seedling, vegetative and reproductive stages following DUS guidelines (Mohapatra et al. 2014).

Screening of mutant lines for biotic stresses
Screening for Yellow Stem Borer (YSB) resistance/susceptibility was done in a phased manner at both vegetative and reproductive phases by augmenting the natural pest infestation through arti cial releases (Devasena et al. 2018).The susceptibility and resistant level of the mutant line was determined based on the Standard Evaluation Scale (SES) scale, IRRI, 2014.The varieties, Pusa Basmati-1 and Samba Mahsuri were used as susceptible checks, while W1263 was the tolerant check for YSB screening.
Screening for sheath blight tolerance/susceptibility of the mutant population was done at two locations viz., ICAR-IIRR and ICRISAT farms using a highly virulent isolate of rice sheath blight pathogen R. solani (WGL-12 isolate).Arti cial inoculation of the isolate was done in colonized typha stem bits and culture prepared according to the procedure given by Bhaktavatsalam et al. (1978).Varieties including TN1 and Samba Mahsuri (Wild type) were used as susceptible checks, while 'Tetep' was used as a tolerant check for sheath blight screening.Observations were recorded after 20 days of inoculation and scored as per IRRI-SES scale (Standard Evaluation System) (IRRI, 1996).Similarly, the mutagenized population was evaluated for bacterial blight (BB) under arti cial conditions following the clip-inoculation method (Kauffman et al 1973), the lesion length was recorded after 15 days of infection and scored according to SES scale, IRRI (2014) (Yugander et al. 2019).For BB screening, the varieties TN1 and Samba Mahsuri (Wild type) were used as susceptible checks, Improved Samba Mahsuri (a derivative of Samba Mahsuri having Xa21,xa13 and xa5 genes; Sundaram 2008) was used as the resistant check.

Genotyping of the mutant lines with SSR markers
Genomic similarity was carried out among the promising mutants with Samba Mahsuri using 60 SSR markers distributed across 12 chromosomes (Table S1).The DNA extraction was carried out by modi ed CTAB method (Saghai-Maroof et al. 1984).The PCR reaction volume was set to 15 μL and ampli cation was achieved using a Thermo Cycler according to the step-cycle program (Devi et al. 2015).The PCR products were run on a 3% Agarose gel stained with Ethidium bromide for 2h at 120 Volts, and DNA bands were visualized under UV light in a gel documentation system.The genomic similarity was calculated based on polymorphism between Samba Mahsuri and mutant (Table S1).

Whole genome resequencing of mutants
The genomic DNA was isolated from three genetic stocks (which include the nucleus seed of Samba Mahsuri which was used for mutagenesis along with other two stocks of this variety) as well as fteen mutants with different traits using the CTAB method as mentioned above.The isolated genomic DNA was checked for quality on agarose gel and quantity using Qubit 4 uorometer (Thermo Fisher Scienti c, USA).Library preparation and sequencing was outsourced to Nucleome Informatics Private Limited, Hyderabad, India.In short, sequencing libraries were prepared using Truseq Nano DNA HT Sample preparation Kit (Illumina, USA) and quali ed libraries were sequenced on Illumina HiSeq2500 platform to obtain 2x150 bp reads at approximately 40X coverage.
The obtained high-quality reads from the three genetic stocks of Samba Mahsuri were used for constructing the Samba Mahsuri reference genome using the consensus calling approach described previously (Abe et al. 2012).Brie y, the sequenced reads were aligned using bwa mem (v0.7.17; http://bio-bwa.sourceforge.net/bwa.shtml)to the indica reference genome, R498 (Du et al. 2017).Using Samtools (v0.1.20;http://www.htslib.org/doc/samtools.html) the bam les were sorted, and deduplicated.The nal bam les were used for variant calling with reference to R498 genome using freebayes (v1.0.0;Garrison et al. 2012).The variants were ltered using VCF tools (v0.1.15;Danecek et al. 2011) and bash scripts to retain only homozygous SNPs supported by a minimum of 10 reads and with minimum quality score of 30.All the SM sequences were processed in the same way and nally all the SNPs were merged to obtain a single VCF le.Using BCF tools (v1.7; http://samtools.github.io/bcftools/bcftools.html)consensus, the SNPs were replaced in the R498 reference genome to obtain SM reference genome.
The Samba Mahsuri reference genome was used for aligning the mutant sequence reads and variant calling using the same pipeline as discussed above.After obtaining the ltered SNPs, only the GC to AT type transitions were retained for further analyses.Custom scripts were used for identifying the chromosome-wise distribution of the SNPs.Dissimilarity index was calculated by dividing the total number of GC to AT type homozygous SNPs by the total number of bases in the reference genome (i.e.390,983,850).Tassel 5 was used for obtaining the relatedness dendrogram (Bradbury et al. 2007).SnpEff (v4.3t;Cingolani et al. 2012) was used for annotating the SNPs.

Results
Variation in yield contributing traits and biotic stress A total of 10,500 mutants (M 1 ) were generated upon treatment of Samba Mahsuri with EMS mutagen in two different concentrations.Phenotypic observations at M 2 generation for various morphological, physiological and yield parameters identi ed 1231 mutants (~ 12% of population) showing promising variation for the traits of interest (Table 1; Table S2 to S7).These mutants were categorized for different traits, among those, the mutants showing variation in the yield contributing traits occupied the major proportion (43%), followed by physiological and panicle emergence mutants (15% each) and the panicle type and morphological trait mutants (Fig. 1).The selected mutants were subsequently advanced to M 3 and M 4 generations wherein 418 and 276 mutants were showing stable inheritance of variations respectively (Table S2 to S7).Considering the traits of interest, 180 mutants were selected and advanced up to M 5 generation for further studies (Table S2 to S7).Upon screening of mutagenic population of M 2 for three biotic stresses (YSB, ShB and BLB), 1453 promising mutant lines were identi ed, of which ShB mutants occupied the major proportion (55%).These mutants were advanced to the subsequent generations and a total of 35 stable mutants were recovered in M 5 generation for the three biotic stresses (Table S8 to S10).
Variations in the morphological traits Around (0.79%, 83 mutants) of the total population showed variation in various morphological traits (viz., tall, dwarf, grassy bonsai and pink apiculus) (Fig. 2A & 2B).Among these, mutants having tallness (130-155cm) were the highest proportion (51%, 43 mutants) of the total morphological traits, followed by dwarf mutants (43.3%; 36 mutants) which have a length <110cm.We also identi ed one extreme dwarf mutant (bonsai type having a shoot length of 30-45cm).This mutant had less number of internodes (3) and very narrow internodal length with grassy appearance.The details of the traits and the number of mutants observed in each generation are given in Table 1 and Table S2.

Variations in the physiological traits
The M 2 mutants were characterized for various physiological traits viz., ag leaf variation, leaf variations, albino, xantha, coloration, strong culm, stay green, early maturity, shattering and sterility.Of the total population, 1.8% mutants (191) showed variations in the physiological traits.Among these, Strong culm that imparts lodging resistance is an important trait and we identi ed 26 mutants (0.24%) with culm strength ranging between 26-32 N/m 2 , while the wild type showed culm strength of 25N/m 2. (Fig. 2C).Early maturing (108-136 days) mutants were observed at a frequency of 0.69% (73 mutants) (Fig. 2D).Staygreen in the post-anthesis period is found to be an e cient drought tolerance trait and we obtained 4 stay green mutants (0.03%) in the M 2 generation.We also obtained a good frequency (0.26%) of sterile plants; however these could not be maintained further due to non-uniform segregation (Table S3).

Variations in panicle emergence
Complete panicle emergence (CPE) and Elongated Upper Internode (EUI) are two major traits that were studied under the category of panicle emergence.In this category we identi ed 182 mutants (1.72% of the population) of which CPE mutants were more numerous (1.47% of the population) and the EUI mutants were less frequent (0.25% of the population).The CPE mutants exhibited 2 to 2.5cms of panicle exsertion and thus no panicle choking (Fig. 2G) whereas 10-20% panicle chocking was observed in wild type.The EUI mutants exhibited a length of 7-26cm for the upper most internode (Fig. 2H) (Table S5).

Variations in yield and yield related traits
The yield-related traits included more number of tillers, number of productive tillers and better grain lling, grain number and phenotypic acceptability (PA).In the M 2 generation, the mutants in yield traits were the most frequent; i.e. a total of 532 mutants representing 5.06% of the total mutant population (Fig. 2I).Among these traits, number of productive tillers and better grain lling occupied the major proportion with 4.06% (484 mutants), while more number of tillers (15-26 tillers) occupied the least with a proportion of 0.12%.Overall acceptability of the mutant lines was also recorded and we obtained a frequency of 0.19% (20 mutants) in the M 2 generation (Table S6).

Variations in the Grain types:
The wild type has medium slender grain type, however upon treatment with EMS we observed a wide range of grain types (Table 1).The grain type mutants occupied 1.5% (158) of the total M 2 population, of which medium bold grain without awns were the highest frequency (0.36%), followed by long slender grain without awns (0.34%).Interestingly, we also observed seven (0.06%) sickle shaped seed mutants.The types of variants obtained are indicated in Fig. 2J.All these mutants were forwarded to M 5 generation, however in subsequent generations we focused on medium slender and long slender grain types, as these are the commercially preferred grain types (Table S7).

Biotic stress variations among the mutants
The mutant population that had been generated was large and screening such a large population at one time for various biotic stresses in the same season was not practical.For ease of operations, 2500 M 2 families each were screened in two seasons for all three biotic stresses.While going for the second season screening, the selected tolerant mutants from the rst season were carried forwarded for rescreening in the next generation along with the base mutant population.A high frequency (13.83%) of the total population showed tolerance for at least one of the three biotic stresses at M 2 generation.But rescreening was continued up to M 5 generation, the frequency of tolerance was reduced from 13.83% to 0.33%.Among those, Sheath blight tolerant mutants were the most frequent (7.62%) followed by the bacterial leaf blight tolerant (4.59%) while the YSB tolerant mutants were the least (1.61%) at the M 2 generation (Table S8 to S10).The reduction in frequency of tolerant mutants following rescreening indicates that there were a large number of false positives possibly due to lines that escaped infection.The details of the tolerant mutants obtained in each generation for various biotic stresses are given in Table 2 and the visual symptoms are given in g. 2K, 2L and 2M.

Mutants with multiple trait variations
Most of the mutants showed trait-speci c variations in comparison with Samba Mashuri, but a few mutant lines (8) showed variation in multiple traits.For instance, the mutant line TI -24 showed different trait combinations like strong culm, early owering, medium bold, pink apiculus, complete panicle emergence, broad ag leaf and dense panicle and stay green.Similarly, another mutant TI-21 also exhibited variations in multiple traits viz., Strong culm, broad ag leaf and early owering yet showed medium slender grain like wild type.The details of the mutants which showed multiple trait variations are given in the Table S11.

Genotyping of mutant lines:
To know the genomic similarity between mutants and wild type, a set of 60 SSR markers distributed uniformly across the genome were used.A total of 49 mutants were selected randomly for the estimation of genomic similarity, which revealed a similarity range of 85.0-98.3% in comparison to the wild type (Table S12).
We further performed whole genome resequencing of the wildtype (three genetic stocks) and fteen mutant lines showing different traits to study the extent of mutagenesis.We obtained high quality sequences with an average depth of 41X (Table 3).Greater than 97% of the reads from the mutant lines mapped to the Samba Mahsuri reference genome (Table 3).Analyses of the EMS-induced SNPs revealed that the dissimilarity between Samba Mahsuri and the mutant lines ranges from 0.008% to 0.089% (Table 4).The number of EMS-induced homozygous SNPs in the mutant lines was between 29850 and 348872.Annotation of the SNPs showed that among the total number of SNPs, 16 to 20% of the SNPs were in the coding region of the mutant lines (Table S13) and relatedness analysis revealed clustering of the mutant lines (Figure S1).

Discussion
Samba Mahsuri (BPT-5204) is a popular rice variety in South and Eastern India which is highly susceptible to many biotic and abiotic stresses.The everlasting demand for this variety has encouraged various researchers to select this variety for marker assisted selection (MAS) to improve tolerance for some of the stresses, where the donors are available in gene pool.To enhance the scope of MAS and to develop the donors for traits where the donors are not available, creation of variation through mutagenesis was sought.Among various physical and chemical mutagens, EMS induced mutations creates more point mutations and less chromosomal segment breaks as well demonstrated in rice (Till et al 2007;Greene et al. 2003).The screening of mutants generated by T-DNA or transposons suffer from restrictions for eld evaluation as they are considered as genetically modi ed (GM) but this restriction does not apply for the EMS induced mutants.Owing to these advantages, in the present study EMS was opted for induction of mutagenesis.Till date, more than 22 mutant resources were developed in diverse genetic backgrounds of rice using different mutagens across the globe (Sevanthi et al. 2018), but no studies were reported on the development of mutagenic population in Samba Mahsuri till date.
Identi cation of suitable mutagenic dose is the key component in developing mutagenic populations since it determines the mutation densities.To generate such population, Samba Mashuri seed were treated with EMS in two different concentrations (1.2% and 0.8% for 12 hrs) and the 1.2% treatment yielded a greater frequency of mutants with favored agro-morphological traits.Similar mutation induction studies were conducted by Mohapatra et al. (2014) in aus cultivar Nagina 22 and con rmed that 1.5% for 12 hrs of EMS treatment generated useful mutants after using a wide range of EMS concentrations (0.2-2%).Wu et al. (2005) established a mutant population in IR64 using ve different EMS concentrations (0.4, 0.6, 0.8, 1.0 and 1.6%) for kill curve analysis and observed that the 1.6% concentration exhibited the high frequency rate of mutations.Talebi et al. (2012) used six different concentrations (0.25%, 0.50%, 0.75%, 1%, 1.25%, 1.5% and 2%) of EMS to treat MR219, a Malyasian indica variety and xed 0.25% and 0.50% for 12 hrs as LD values for generating variation.Several mutagenesis studies on plants have reported that the optimal concentration of the chemical mutagen can vary with the variety used, which might be due to possible in uence of the genome on the effects of the mutagen (Rao and Reddi, 1986).
The progeny from a set of 10,500 Samba Mahsuri M 1 mutant lines were characterized for agro-morphological traits and biotic stresses from M 2 to M 5 generations and mutants were identi ed that have variations in different morphological characters like height (tall, dwarf, bonsai) and pink apiculus.Similar plant height variations were also identi ed by Mohapatra et al. 2014.Our observations also suggested that some mutants might have mutations with large phenotypic effects which led to fall well outside the phenotype of the wild type.
The ag leaf and top leaf blades play a major role in photosynthesis and affect the grain yield of rice (Takai et al. 2013).
Therefore we classi ed mutants having broad and long ag leaf variations in the category of physiological traits.These mutants might carry mutations among alleles/genes responsible for ag leaf development.Earlier studies demonstrated that a mutation in the Flag Leaf gene (NAL 1) affects various yield related traits in rice (Taguchi-Shiobara et al. 2015).Similarly, Sakamoto et al. (2005) explained that the brassino-steroid de cient mutant having erect leaves showed increase in biomass and grain yield in rice.Phenotypes such as albino, xantha and coloration mutants were observed in M 1 generation but they were not observed in adult plants of M 2 generation indicating that they were lethal.
Improvement of lodging resistance by incorporation of the semi-dwarf trait alone is not su cient and other traits such as strong culm are needed for reinforcement of the phenotype (Ookawa et al. 2016).According to Ookawa et al.(2010), the diameter and wall thickness will in uence the parameter of the physical strength of the culm.In the present study the identi ed 26 strong culm mutants will be used for the identi cation of genes/alleles that governs the trait.A QTL associated with strong culm was earlier identi ed by Ookawa et al. 2010.Using chromosomal segment substitution lines they have reported that Strong Culm 2 (SCM 2), a gain-of -function mutant of APO1, was responsible for the strong culm phenotype.Maturity duration is one of the important growth parameters that determine crop productivity under a given climatic condition.We recovered thirty-six early maturing mutants (maturing between 108-135 days).Mutants that are similar in yield with wild type but having an early maturing nature are very useful for the rice varietal improvement programme.Similarly, Chakrabarti et al. (1995) developed early maturing PNR series mutants utilizing gamma radiations in Basmati 370 PNR-17-3 and Mustikarini et al. ( 2017), developed a gamma-ray induced red rice (Celak Madu accession) mutant which matures earlier than the wild type (i.e. in 115 days) and is drought tolerant .
The stay green or delayed leaf senescence phenotype of the plant is another attractive and desired characteristic of rice.Very few genetic sources are available for this trait and hence the stay green mutants identi ed in this study can be used in improving rice production and biomass under stressful environment conditions.In this connection it is pertinent to note that Ramkumar et al.
(2019) have identi ed a stay green mutant (SGM-3) of Nagina22 which has better harvest index under drought stress.
Similarly, elongation of upper internode (EUI) is an important component of plant architecture that can be used to overcome the sheathed panicle of the rice male sterile line (Shen et al. 1987) by enhancing pollen dispersal and facilitating hybrid rice seed production (Yang et al. 2002).In our study, the identi ed EUI mutants were observed to have better upper node elongation than the wild type and their incorporation into the CMS line can help enhance pollen dispersal.Okuno and Kawai (1978) rst reported EUI mutants, derived from the Japanese rice cultivar Norin 8 by gamma-ray treatment.Similar EUI studies were conducted by Rutger and Carnahan (1981) and reported enhanced upper internode elongation during the heading stage.Similarly, incomplete panicle emergence leads to yield reduction (Yuan et al. 1988) which is generally observed in Samba Mahsuri resulting in 10-15% reduction in yield.Hence, the identi ed mutants that have complete panicle emergence may play an important role in rice breeding programmes to increase yield of Samba Mahsuri as well as other rice varieties whose yield may be affected due to incomplete panicle emergence.
A number of high yielding mutant lines were identi ed that have more tillers number and high grain number.Similarly, Roy et al.
(2018) identi ed eight high yielding aromatic mutants in the background of Tulpanji using gamma irradiation.Soomro et al. (2002) identi ed two high yielding mutant lines derived from of IR6, through gamma rays.Several high yielding mutants were also popular in many countries such as Vietnam (55 mutant varieties), Bangladesh (44 mutant varieties) and Thailand (two aromatic mutant varieties (RD6, RD15) where they have been cultivated over millions of hectares (Oladosu et al. 2016).In India, the PNR-381 and PNR-102 aromatic rice varieties are early maturing rice mutants that are derived from the Basmati 370 PNR-17-3 cultivar and have become popular in Haryana and Uttar Pradesh States (https://mvd.iaea.org;Chakrabarti, 1995).
In the present study, different grain type mutants were isolated.They could have importance in varietal improvement since the visual characteristics of rice grains is an important attribute that affects consumer's preference.Azad et al. (2012) developed different grain type mutants by irradiating IR8 with gamma rays.Different grain types (short, long, bold) were also reported from the Nagina22 mutant population (Mohapatra et al. 2014).A mutant line derived from Kitake rice variety using fast-neutron (FN) mutagenesis carried a new grain shape 9-1 (gs9-1) allele that also affects grain size (Jiang et al. 2019).
In this study, mutants having variations in multiple traits were observed and such kinds of mutants were also reported earlier in Nagina 22 background (Mohapatra et al. 2014).These mutants may either carry point mutations in multiple genes that independently affect the observed traits or carry a mutation that has a pleiotropic phenotype (Patil, 1966;Till et al. 2007).
In the present study mutation frequency of many mutant types showed a decreasing in the number as we progressed from M 2 to M 5 .We are unable to explain this apparent reversion phenomenon as the phenotype was quite apparent in the M 2 generation.
One possibility is that there are some physiological changes, such as epigenetic changes, that arise due to mutagenesis and which revert back to the wild type phenotype/physiological state in subsequent generations.Brooks et al. 2008 reported that rice bran gene mutation restoration from white rice pericarp to red like wild type.In our study, one example of such a reversion was the sickle shape seed phenotype in M 2 generation which had reverted back to wild type seed shape by the M5 generation.
Biotic stresses (insect pests and diseases) cause more than 42% of crop yield reduction in the world (Pimentel, 1997).Among the biotic stresses, Yellow stem borer, Sheath blight and BLB count amongst important biotic stresses in rice.The resistance breeding for yellow stem borer has not gained any momentum due to the lack of suitable resistance/tolerance sources (Makkar and Bentur, 2017).Sheath Blight is another challenging disease due to unavailability of a reliable resistance source (Chen et al. 2019).
Whereas, for management of BB, there is an emergent need to widen the repertoire of resistance genes to enhance the spectrum and durability of resistance to Xanthomonas.oryzae.Mutagenesis in a gene may result in loss or gain in function with varying levels of expression and such type of mutations have prime importance while dealing with biotic stresses.Therefore, we attempted to identify resistance/tolerance stocks for YSB, Sheath blight and BB from the mutagenized population.In a similar mutation study conducted by Wu et al. (2005), by irradiating IR64 with gamma rays, rice mutants that are resistant to Blast, BB and tungro disease were identi ed.Similarly, Mohapatra et al. (2014) identi ed BB resistance mutants in an EMS induced Nagina 22 population.Our study is consistent with the above mentioned studies which indicate that mutagenized populations harbor a great amount of variability that can be recovered if the mutant population is subjected to appropriate screening techniques.
We have also evaluated the similarity of the mutant lines in comparison to the wild type line using a set of 60 SSR markers that are considered to be very polymorphic and which are fairly well distributed throughout the genome as well as by resequencing.SSR analysis, the mutants exhibit a wide range of diversity ranging from about 1% to 14% in comparison to the wild type line.
Earlier studies reported that the average SSR marker polymorphism between closely related cultivars was 12-15% due to conserved nature of SSR markers (Shanmugavadivel et al. 2013;Yoon et al. 2006;Govindaraj et al. 2005).Genomic similarity analysis with SSRs may depend on the selection markers and the genome wide distribution of markers.However, whole genome analyses of mutants gave the clear picture of diversity which ranged from 0.008% to 0.089%.These results indicated that generated mutants were similar to the wild type line.
Recent advances in high throughput sequencing technologies and computational approaches have facilitated the mapping of causal mutations in mutant lines (Austin et al. 2011;Schneeberger, 2014).In addition to gene identi cation, an allelic series comprised of INDELs can be useful in assessing gene function and for practical purposes of crop improvement.Therefore, comparison among the allelic series in SNP and INDELs across mutants and natural germplasm helps to infer the functional polymorphism as related to phenotypes observed in mutants and germplasm.
The mutants identi ed in this screen, especially those that have agronomically important phenotypes with signi cant similarity to the wild type could be useful to use in MutMap and QTL Seq for the identi cation of linked SNPs.The SNPs that are linked to the phenotype can be readily used in introgression programmes.Such studies will generate new knowledge and also have the potential for application in the crop improvement.

Conclusion
Exploiting genetic variability induced through mutations can be a powerful strategy in rice improvement where genetic gains may be diminishing due to a narrow gene pool.To generate variability, we have produced a large Samba Mahsuri mutant population that can be further employed in crop improvement.The huge demand for this variety and its wider adaptability and combining ability make this mutant population a very useful resource for rice improvement.The value of these stocks will be enhanced with increasing usage and extensive screening under a wide range of conditions.Systematic screens and mapping studies will not only help in generating new knowledge but will also help in translating the knowledge for crop improvement.

M 1
plants were protected from out crossing through bagging and harvested individually to obtain the M 2 seeds from main panicles.The collected M 2 seeds were distributed among the participating centers (Centre for Cellular and Molecular Biology (CCMB) and ICAR-Indian Institute of Rice Research (ICAR-IIRR) for further studies.The M 2 seeds were raised at ICAR-IIRR and International Crop Research Institute for Semi Arid Tropics(ICRISAT), Patancheru, Hyderabad during kharif (June-July to October-November) 2013.

Table 1 :
Number of mutants that exhibit agronomically important traits in each generation from M 2 to M 5 generation

Table 2 :
Number mutants that exhibit resistance against YSB, BLB and Sheath Blight from M 2 to M 5 generations

Table 3 :
Sequence statistics of Samba Mahsuri (wild type) and mutant lines with different traits

Table 4 :
Number of EMS-induced homozygous SNPs (GC to AT type), dissimilarity index, and the frequency of SNPs in the mutant lines with respect to SM