Morphological, Biochemical and Molecular Characterization of Indigenous Aromatic Rice of Assam

We carried out the morphological, biochemical and molecular characterization of 20 indigenous Joha (aromatic) rice cultivars of Assam. Distinctiveness, Uniformity and Stability (DUS) characterization of the cultivars revealed polymorphism in thirty-seven traits, establishing distinctiveness for their utilization in breeding programmes. Unweighted Neighbour Joining (UNJ) clustering based on usual Euclidean distances for the polymorphic markers grouped the cultivars into three multi-genotypic clusters. The Joha rice cultivars showed highly signicant differences for all the quantitative traits except for panicle length. The genotypic and phenotypic coecients of variability (GCV & PCV) were high for grain yield ha − 1 (24.62 & 24.85%) and lled grains panicle − 1 (23.69 & 25.02%). All the traits except days to owering and maturity, ag leaf breadth and spikelet fertility exhibited high heritability along with high to moderate genetic advance, indicating the predominant role of additive gene action. Mahalanobis D 2 analysis revealed three multi-genotypic and four mono-genotypic clusters of the cultivars. The cultivars' average polyunsaturated fatty acids were 37.9% oleic acid, 39.22% linoleic acid and 0.5% linolenic acid. The fatty acid prole of Local Joha was superior to the other cultivars as it showed a high level of linoleic and linolenic acid and low saturated fatty acid content. Kon Joha 4 and Ronga Joha contained the highest iron (82.88 mg kg − 1 ) and zinc (47.39 mg kg − 1 ), respectively, while protein content of Kon Joha-1 and amylose content of Harinarayan were the highest. Joha-Bihpuria showed the highest gel consistency of 140.50 mm. Kalijeera, Kunkuni Joha, Kon Joha-5, Manimuni Joha and Kon Joha-2 accorded a strong aroma. PCR amplied 174 alleles with a mean value of 2.64 across the 66 polymorphic SSR markers. PIC values ranged from 0.091 to 0.698, with an average of 0.326. The highly informative (PIC > 0.50) markers were RM316, RM283, RM585, RM1388, RM3562, RM171, R1M30, RM118, RM11and RM29 for identication of the twenty aromatic rice cultivars. The UNJ clustering based on Jaccard's coecients classied the 20 cultivars into three distinct clusters with eight, ten and two entries. and represented through cluster analysis using the algorithm of Unweighted Neighbour-joining (UNJ) method by feeding the distance matrix as input data. Genetic relatedness among the genotypes was computed by using the Jaccard’s coecient of similarity (Jaccard, 1908) and a dendrogram was constructed illustrating the genetic relationship among the rice genotypes using UNJ method as proposed by Gascuel (1997), which uses a criterion of weighted average, in DARwin 6 (Perrier and Jacquemoud-Collet, 2006). The number of different alleles per locus (N a ), major allele frequency (MAF), number of effective allele (N e ), Shannon’s information index (I), observed heterozygosity (H o ), expected heterozygosity (H e ) and polymorphism information content (PIC) values were calculated using GenAlEx version 6.5 (Peakall and Smouse, 2012). by unweighted neighbour-joining (UNJ) of usual Euclidean distance based on thirty-seven polymorphic traits as dissimilarity measure the twenty aromatic rice cultivars three multi-genotypic clusters viz., G1, G2 and G3 The clusters further subdivided into sub-clusters and had an unequal distribution of the cultivars. The cultivars viz., Kon Joha-Bongaigaon, Kon 1, Joha-Golaghat, Manimuni Joha, Kola Joha, Joha-Bihpuria, Soru Joha-Tinsukia, Kon Joha 4 and Kalijeera belonged to G1. At the same time, G2 had Kon Joha 5, Kon Joha-Moran, Harinarayan, Jeera Joha, Local Joha, Keteki Joha, Kon Joha 3, Kon Joha 2 and Ronga Joha. These clusters had the most signicant number of nine cultivars, followed by G3, including Kon Joha-Teok and Kunkuni Joha. The group constellation pattern proved the existence of a substantial amount of variability among the indigenous Joha rice cultivars. ndings Sripunitha Sivasubramaniam rice length-breadth ratio (3.32) grain yield per (3012 (54.21 (32.42 panicle (27.26 (36.73 stem thickness (4.62 grains (252), grain rice breadth (2.82 2.33 length-breadth ratio (3.89) the shortest plant height (111 on fatty acid prole, Fe and Zn content, crude protein, amylose, gel consistency and aroma, a cultivar's mean value was considered desirable for all other biochemical traits except for polyunsaturated fatty acids when it exceeded the cultivars' mean plus the standard deviation. A low mean less than the cultivars' mean minus the standard deviation was desirable for polyunsaturated fatty acids. programme, based on the extent of genetic variation for alleles.

Introduction applied as per Sali rice recommendation for Assam. The entire quantity of P 2 O 5 and K 2 O and a half dose of N were applied as basal dose at the time of nal eld preparation; the remaining nitrogen applied in two equal splits at maximum tillering and booting stage. The standard agronomic practices recommended for the state of Assam were adopted in both experiments. Observations were recorded according to the National Test Guidelines for DUS test in rice developed by the Directorate of Rice Research, Hyderabad (Shobha Rani et al., 2004). The yield attributing traits were based on ve random plants per replication, while days to owering and maturity recorded on a plot basis. The characteristics observed were days to rst owering, days to 50% owering, days to maturity, number of productive tillers plant − 1 , plant height (cm), panicle length (cm), spikelet fertility (%), 1000-grain weights (g), grain yield plant − 1 (g), biological yield plant − 1 (g), harvest index (%), protein content (Kjeldahl's method), iron (Fe) and zinc (Zn) content (AAS), fatty acid pro ling in rice bran.
Genomic DNA extraction Seeds of cultivars were germinated and grown in a growth chamber, maintaining a temperature of 30 o C, 10 hours of light and 85% relative humidity. At the three-leaf stages, the leaves were harvested by liquid nitrogen and taken for DNA isolation. The genomic DNA then was isolated by the CTAB (Cetyl Trimethyl Ammonium Bromide) method (Murray and Thompson, 1980).

DNA quanti cation
The concentration and quality of genomic DNA samples were estimated on a 260, 280 nm, and 230 nm spectrophotometer, the samples with a 260/280 ratio exceeding 1.8 were considered good quality DNA. Quality of DNA fragment was also con rmed by 0.8% agarose gel electrophoresis using 1XTBE buffer at 100V for 90 min.

Molecular marker selection
Seventy-one SSR markers were used for the genotyping (Supplementary Table S1). The markers were selected from various published literatures as well as from Gramene database (www.gramene.org) maintaining the genome wide distribution of markers.
PCR ampli cation using SSR primers PCR reaction was performed in 25 µL mixture containing 1 µL (25 g/ µL) template DNA, 2.5 µL of 10x PCR buffer with 25 mM MgCl 2 , 1.0 µL 5 mM of each forward and reverse SSR primers, 1.0 µL of 10 mM dNTPs and 0.2 µL of Taq DNA polymerase. PCR reactions were performed in a thermal-cycler (Eppendorf, Hamburg, Germany). The ampli cation pro le consisted of initial denaturation for 2 min at 95 o C; 35-40 cycles of denaturation at 95 o C, annealing at 50-60 o C and extension at 72 o C. After that, nal extension was carried out at 72 o C for 7 min.
Gel electrophoresis PCR products of SSR markers were resolved on capillary electrophoresis system (Qiagen Pvt. Ltd., Hamburg, Germany).

Scoring of SSR data
The molecular weight of the PCR products for each SSR primer was determined from a ladder of known molecular weights. During band scoring, faint bands and bands with smeared background were avoided, and only intense bands were scored on the basis of their product size. The presence of a product in a certain genotype was designated as '1', and the absence was designated as '0'. Only the speci c PCR products showing consistency in the successive ampli cations were selected to minimize the possibility of mis-scoring markers.

Statistical analysis
A pooled ANOVA for the traits over the two years was done considering replication, genotype and environment as xed effects (Singh and Chaudhary, 1985) in MS Excel 2007. Genetic parameters were estimated by the formula given by Burton (1952) for GCV and PCV, Hanson, Robinson and Comstock (1956) for heritability and Allard (1960) for expected genetic advance in MS Excel 2007. Mahalanobis D 2 analysis was done in Windostat version 9.2 (http://www.windostat.org). Usual Euclidian distances between the cultivars were worked out from the standardized data matrix in DARwin version 6.0.021 (Perrier and Jacquemoud-Collet, 2006) and represented through cluster analysis using the algorithm of Unweighted Neighbour-joining (UNJ) method by feeding the distance matrix as input data. Genetic relatedness among the genotypes was computed by using the Jaccard's coe cient of similarity (Jaccard, 1908) and a dendrogram was constructed illustrating the genetic relationship among the rice genotypes using UNJ method as proposed by Gascuel (1997), which uses a criterion of weighted average, in DARwin 6 (Perrier and Jacquemoud-Collet, 2006). The number of different alleles per locus (N a ), major allele frequency (MAF), number of effective allele (N e ), Shannon's information index (I), observed heterozygosity (H o ), expected heterozygosity (H e ) and polymorphism information content (PIC) values were calculated using GenAlEx version 6.5 (Peakall and Smouse, 2012).

Results And Discussion
Morphological characterization of the cultivars A total of sixty-two qualitative and quantitative traits characterized the distinctiveness of the twenty Joha rice cultivars. Qualitative characteristics are considered morphological markers for identifying rice landraces because environmental changes least in uence these traits (Raut, 2003). The stable morphological traits serve as reliable morphological markers for the identi cation of a cultivar. Each cultivar must have certain novel diagnostic features which will distinguish a variety from the others. Such diagnostic characters should be uniformly present in the population and inherit in the next generation, and then only the character is supposed to be stable and can be used as morphological marker traits to distinguish that variety. Among the traits, the spectrum of variability in the twenty cultivars revealed 22 monomorphic and 37 polymorphic characteristics (Tables 2 & 3), suggesting their potential for cultivar characterization and distinctiveness. Earlier studies of Bisne and Sarawgi (2008), Mathure et al. (2011), Subudhi et al. (2012, Subba Rao et al. (2013), Sarawgi et al. (2014), Parikh et al. (2012 and Sinha et al. (2015) corroborated the above ndings.

Pooled analysis of variance
The pooled ANOVA over the two years (Supplementary Table S2) revealed that the mean squares due to years were signi cant for fteen traits viz., days to rst owering, ag leaf length, ag leaf breadth, ag leaf area, days to 50% owering, days to maturity, plant height, productive tillers plant − 1 , lled grains panicle − 1 , spikelet fertility, 1000-grain weights, biological yield plant − 1 , grain yield plant − 1 , harvest index and grain yield kg ha − 1 , suggesting a signi cant in uence of the years on the phenotypic expression of these traits. The yearly variation was mainly due to the difference in the 2018 and 2019 crops' sowing time. The crops were sown on 11th July and 17th June in 2018 and 2019, respectively. The June planted crop in 2019 exhibited higher mean performances for most of the traits mentioned above. Delaying the sowing time decreased the number of days to owering and maturity for most of the cultivars. A similar observation was reported by Song et al. (1996) for days to heading reduced in different rice cultivars due to delayed sowing. Nahar et al. (2009) observed a signi cant decrease in lled grain production consequent upon delayed transplanting, attributed to low temperature at anthesis and spikelet primordial formation. Khalifa (2009) noted the early date of sowing the best time for maximizing morpho-physiological traits such as tillering, panicle initiation, chlorophyll content, leaf area index, sink capacity, panicle length, panicle number and grain yield. Delayed sowing signi cantly reduces the number of lled grains, panicles and test weight, nally lowering rice cultivars' grain yield (Patel et al., 2019). The cultivar differences for all the traits except for panicle length were also evident from highly signi cant mean squares. The Years x Cultivars interaction component was substantial for days to rst owering, days to 50% owering, days to maturity, lled grains per panicle and spikelet fertility, suggesting differential behaviour of the cultivars in the early and late sown crops. The  Gangashetty et al. (2013) and Verma et al. (2014) further supported the presence of signi cant cultivar differences. Mahapatra et al. (1996), Reddy et al. (1998) and Panwar et al. (2008) also reported substantial Years x Cultivars interaction for days to 50% owering and days to maturity.

Mean performance of the cultivars
The variation among the various traits under study revealed free variability in different cultivars populations, re ecting the unforeseen impact of potential variability on yield. The cultivar mean performance for the observed traits (Supplementary Table S3) identi ed the top-ranking cultivars Kon Joha 2 for stem thickness; Joha-Bihpuria, Joha-Golaghat and Kola Joha for the earliest days to rst owering; Kon Joha-Teok for ag leaf length; Joha-Bihpuria, Ronga Joha and Soru Joha-Tinsukia for ag leaf breadth; Ronga Joha for ag leaf area; Kola Joha for the earliest days to 50% owering and maturity; Keteki Joha for the shortest height and the highest number of productive tillers; Manimuni Joha, Kon Joha-Moran, Kon Joha 2, Jeera Joha and Kon Joha-Teok for lled grains per panicle; Jeera Joha and Kon Joha-Teok for spikelet fertility; Kola Joha for thousand-grain weights; Soru Joha-Tinsukia for the most extended rice length and the highest rice length/breadth ratio; Kon Joha 2 for the broadest rice breadth; Ronga Joha and Kola Joha for biological yield; Ronga Joha for grain yield per plant; Joha-Bihpuria for harvest index; and Soru Joha-Tinsukia for grain yield per hectare (Table 4). The majority of the Joha rice cultivars showed low to medium tillering, whereas none of the cultivars showed high tillering ability, as also supported by Ogunbayo et al. (2005) Semwal et al. (2014) and Sarhadi et al. (2015). The ndings of Koutroubas et al. (2004), Vanaja and Babu (2006), Bajpai and Singh (2010) and Srivastava and Jaiswal (2013) further corroborated the present results on grain physical quality characteristics.

Genetic variability parameters
Grain yield per hectare showed the highest range of variation followed by lled grains per panicle, thousand-grain weights, grain length/breadth ratio, grain yield per plant, productive tillers per plant, rice length, stem thickness and biological yield per plant. Flag leaf breadth showed minor variability among the cultivars.
The magnitude of the genotypic and phenotypic coe cient of variation (Table 5)  ). The phenotypic variations for all the above traits except productive tillers per plant were determined largely by the genotypes. Therefore, phenotype-based selection would be effective for these traits. These ndings were in tune with Singh and Choudhary (1996), DebChoudhary and Das (1998), Kavitha and Reddy (2002), Karim et al. (2007) Heritability estimates ranged from 37.36 per cent for grain yield per plant to 99.99 per cent for ag leaf length ( Table 5). Most of the traits exhibited high heritability values (82.64 to 99.99%); the exceptions were productive tillers (58.00%), biological yield (41.19%) and harvest index (45.26%) with average estimates and grain yield were low heritable. The genetic advances as per cent of mean were high (> 20%) for grain yield kg per ha (50.24%), lled grain per panicle (46.22%), rice length/breadth ratio (40.33%), thousand-grain weights (39.85%), grain length/breadth ratio (38.77%), rice length (34.98%), grain length (34.06%), ag leaf area (21.79%), stem thickness (20.79% and productive tillers per plant (20.64%). The traits excluding days to rst/50 per cent owering and maturity, ag leaf breadth and spikelet fertility exhibited high heritability in conjunction with high to moderate genetic advance, indicating the most likely role of additive gene action and effectiveness of simple selection for the traits. Similar results on days to 50 per cent owering were in agreement with Bihari et al. Sankar et al. (2006) and Karthikeyan et al. (2009). High heritability and low genetic advance for days to 50% owering agreed with Chaurasia et al. (2012). Plant height registered high heritability coupled with the moderate genetic advancement in conformity with the ndings of Mishra and Verma (2002) and Chaurasia et al. (2012). Expression of moderate heritability coupled with high genetic advance for productive tillers plant − 1 was in tune with Mishra and Verma (2002), Kumari et al. (2003), Jaiswal et al. (2007) and Nandan et al. (2010). For lled grains panicle − 1 , a high heritability concomitant with high genetic advance was in agreement with the results of Hasib et al. (2004) and Panwar (2005). High heritability in concurrence with high genetic advance for 1000-grain weights was in accord with the ndings of Reddy et al. (1997), Murthy et al. (1999), Rao (2000), Mishra and Verma (2002) and Nandan et al. (2010). The grain quality traits viz., grain length, breadth and length-breadth ratio, rice length, breadth, and length-breadth ratio registered high heritability coupled with the high genetic advance in consonance with the ndings of Jaiswal et al. (2007) Genetic divergence among the twenty Joha rice cultivars The twenty aromatic Joha rice cultivars were assessed for the nature and magnitude of genetic divergence (Singh and Chaudhary, 1985) based on the twentythree quantitative traits following Mahalanobis D 2 statistics. The V statistics and the analysis of dispersion (Supplementary Table S4) showed that the mean differences for the pooled effect of p characters between the cultivars were highly signi cant. Based on the D 2 values, the twenty Joha rice cultivars belonged to seven clusters, which showed three multi-genotypic and four mono-genotypic clusters. Table 6 summarizes the cluster-wise distribution of the cultivars. The average intra-and inter-cluster distances (Table 7) showed cluster I to have the maximum intra-cluster distance (331.08) followed by cluster IV (307.84) and cluster II (223.91). The rest clusters were mono-genotypic, consisting of one cultivar each. The inter-cluster D 2 values were the least between the clusters V and VI (353.94), while it was the maximum between clusters IV and VI (7303.31). Cluster I, the largest group comprising eight cultivars, was distantly apart from cluster IV and cluster VI by 2281.86 and 1815.30 units, respectively, and was nearest to cluster III (490.87), followed by cluster V (854.97). Cluster II, with three cultivars, had the lowest inter-cluster distance from cluster III (564.34), followed by cluster V (892.56), while the highest inter-cluster distance was from cluster IV (4709.44). Cluster III consisted of one aromatic Joha rice cultivar and was nearest to cluster VII (434.10) whereas distantly placed from cluster IV (2323.71). Cluster IV was a multi-genotypic cluster having four aromatic Joha rice cultivars and was nearest to cluster VII (1256.85) while most distantly related to cluster VI (7303.31). Mono-genotypic cluster V was closest to cluster VI (353.94) and separated from cluster IV by 5187.67 units. Cluster VII having one aromatic Joha rice cultivar, was distantly placed from cluster VI (4073.37). Hybridization between parents from the widely divergent cluster pairs viz., IV-VI, IV-V, II-IV, VI-VII, V-VI and III-IV would be likely to produce a broad spectrum of variability and transgressive segregations with high heterotic effects as also suggested by The cluster mean performances for the various traits showed a wide range of variations among the clusters (Supplementary Table S5). Cluster III registered the earliest days to rst and 50 per cent owering (108 & 114) and maturity (147) and the highest mean performance for ag leaf breadth (0.84 cm), grain yield per plant (14.83 g) and harvest index (36.40%). Cluster VII showed superiority for spikelet fertility (90.92%), thousand-grain weights (22.50 g), grain and rice length (9.04 &6.97 mm), rice length-breadth ratio (3.32) and grain yield per hectare (3012 kg). The highest mean performances for ag leaf length (54.21 cm), ag leaf area (32.42 cm2), panicle length (27.26 cm) and biological yield per plant (36.73 g) characterized cluster IV while cluster VI exhibited superiority for stem thickness (4.62 mm), lled grains per panicle (252), grain and rice breadth (2.82 & 2.33 mm). Cluster II was superior to other clusters in having the highest grain length-breadth ratio (3.89) and the shortest plant height (111 cm).
Of the twenty-three traits observed, only eight contributed to the genetic divergence. The contribution towards the total variation was the maximum for ag leaf length (72.11%) followed by rice length (13.68%), grain length (6.84%), rice breadth (3.16%), grain yield per hectare (1.58%), thousand-grain weights (1.05%), grain breadth (1.05%) and rice length-breadth ratio (0.53%). These results agreed with the ndings of Allam et al. The clustering pattern of the twenty Joha rice cultivars also con rmed the quantum of diversity present in the indigenous aromatic rice of Assam and offer scope for its exploitation through breeding for yield improvement. Previous studies reported different numbers of clusters in fragrant rice, e.g., two clusters by Das et al. (2012), four clusters by Kole (2000), ve clusters by Patel et al. (2015), six clusters by Singh et al. (1996)

Biochemical characterization of the Joha rice cultivars
In Table 8 showing the biochemical characterization of the twenty Joha rice cultivars based on fatty acid pro le, Fe and Zn content, crude protein, amylose, gel consistency and aroma, a cultivar's mean value was considered desirable for all other biochemical traits except for polyunsaturated fatty acids when it exceeded the cultivars' mean plus the standard deviation. A low mean less than the cultivars' mean minus the standard deviation was desirable for polyunsaturated fatty acids.
Fatty acid pro le Palmitic acid content ranged from 17.98% (Local Joha) to 20.57% (Kon Joha-Teok), with an average of 19.01%. Stearic acid content was the lowest (0.90%) in Harinarayan and the highest (1.86%) in Kalijeera; the average was 1.40 per cent. The range of oleic acid was from 33.53% in Kunkuni Joha to 41.32% in Joha-Golaghat, having an average of 37.90 per cent. The linoleic acid content varied from 36.02% in Joha-Golaghat to 44.61% in Local Joha, with an average of 39.22 per cent. Local Joha recorded the highest linolenic acid content (2.14%), whereas the Kola Joha had the lowest estimate (1.06%). Arachidic acid content ranged from 0.28% (Local Joha) to 0.61% (Kon Joha 3), with an average of 0.48 per cent. Fatty acids are vital components of food and human health. Fatty acids are the major constituents of cell membrane structure and play important biological, structural and functional roles in the human body (Nagy and Tiuca 2017). They act as modulators of gene transcription, cytokine precursors, and energy sources in complex interconnected systems (Glick and Fischer 2013) by producing a vast ATP quantity during their metabolism (Nagy and Tiuca 2017). The role of dietary fatty acids in human health is strongly evident for their in uence on cardiovascular disease and mental health (Glick and Fischer 2013). Besides, rice is a dietary consumption; rice fats have unique health bene ts (Jennings and Akoh, 2009). In the present investigation, the contents of oleic, linoleic and palmitic acids were the primary fatty acids, and those of stearic, linolenic and arachidic acids were minor in the aromatic rice cultivars. Palmitic, stearic and arachidic acids were the saturated fatty acids present in rice bran, increasing the health risk such as atherosclerosis, a disease associated with a heart attack (Oluremi et al. 2013). Linoleic acid gets absorbed as a predominant unsaturated fatty acid, followed by oleic and linolenic acid. The high contents of polyunsaturated fatty acids are desirable for human health as their consumption minimizes the risk of heart-related diseases (Law, 2000). The mean polyunsaturated fatty acid (PUFA) contents of the aromatic cultivars were 37.9% for oleic acid, 39.22% for linoleic acid and 0.5% for linolenic acid, whereas the contents of saturated fatty acids (SFA) accorded 1.40% for stearic acid and 19.01% for palmitic acid. These estimates were comparable or even better than the values of 38.4% oleic acid, 34.4% linoleic acid and 2.2% α-linolenic acid of PUFA; 2.9% stearic acid and 21.5% palmitic acid of SFA (Sayre and Saunders, 1990). The present results were also comparable with those reported by Resurreccion and Juliano (1975) and Taira and Itani (1988).
Similarly, the variations in the fatty acid pro le of the present study proved better in having lower maximum limits for SFA and higher maximum limits for linoleic and linolenic acid than those reported by Goffman et al. (2003), who obtained 13.9-22.1% for palmitic, 1.5-2.7% for stearic, 35.9-49.2% for oleic, 27.3-41.0% for linoleic and 1.0-1.9% for linolenic acid in rice bran. Stearic acid and arachidic acid were present in trace amount in all the studied aromatic rice cultivars. Comparatively, the fatty acid pro le of Local Joha was better than the other cultivars as it showed a high level of linoleic and linolenic acid and low saturated fatty acid content. In general, the Joha rice cultivars' fatty acid pro le quali ed for extraction of quality bran oil for consumption.

Iron and Zn content (mg kg − 1 ) of the cultivars
The cultivars' iron content varied from 21.33 (Keteki Joha) to 82.88 mg kg − 1 in Kon Joha 4 with an average of 43.57 mg kg − 1 . The zinc content ranged from 12.26 (Keteki Joha) to 47.39 mg kg − 1 in Ronga Joha, showing an average of 28.43 mg kg − 1 . Iron and Zinc de ciency is severe nutritional problems for humans and is particularly prevalent among children and pregnant women, especially in developing countries. The Joha rice cultivars with very high iron and zinc contents in brown rice could play an essential role in the developing world's nutritional upliftment. Therefore, increasing the iron and zinc content and bioavailability in rice grains is an essential target for breeders and offers potential bene ts to a large proportion of the human population. Substantial

Molecular characterization
Among these 71 SSR markers, 66 markers showed polymorphism. The analysis excluded the markers with monomorphic banding patterns. Tables 9 summarize the results obtained on the analysis of 20 aromatic rice cultivars using the polymorphic SSR loci. Figure 1 depicts representative gel pictures of the PCR products.
The 66 polymorphic SSR loci ampli ed a total of 174 alleles. The allelic richness per locus ranged from 2 to 4 alleles, with an average of 2.64 alleles. Among the polymorphic markers, 30 markers produced two alleles each, 30 produced three alleles each, and 6 generated four alleles. The markers RM 283, RM 118, RM316, RM29, RM 585 and RM 26063 ampli ed the maximum number of alleles. The result revealed that all the markers showed distinct polymorphisms among the cultivars studied, indicating the robust nature of microsatellites revealing polymorphism. The number of alleles per locus (2.64) obtained in the present study was comparable with the earlier reports by Joshi and Behera (2007), Shah et al. (2013), and Venkatesan and Bhat (2015), who reported 2.6, 2.75 and 2.3 alleles per locus, respectively. The mean allele number (2.64) obtained in the present study was higher than the result of Meti et al. (2013), who detected 2.08 alleles per locus using 48 traditional indigenous aromatic rice germplasm grown under the Eastern part of India through 12 polymorphic SSR loci. In contrast, the mean alleles (2.64) detected was markedly lower than the average number of alleles reported in previous diversity studies by Pervaiz et al. (2010) and Rahman et al. (2012), who reported an average of 4.4 and 4.18 alleles per locus, respectively. The variability in the number of alleles detected per locus might be due to the use of diverse genotypes and the selection of different SSR primers with scorable alleles. Among the sixty-six markers, the highest major allele frequency was 0.950 in 4 markers (RM 237, RM 215, RM6641, RM3481 & RM21), followed by 0.900 in 9 markers and 0.875 in 2 markers. However, RM 316 recorded the lowest major allele frequency of 0.300. The mean value of major allele frequency among the markers was 0.727, and it ranged from 0.300 to 0.950 among the sixty-six markers scored against a set of twenty aromatic Joha rice cultivars. Similarly, Sajib et al. (2012) reported major allele frequency ranging from 0.41 to 0.91; Shah et al. (2013)  (0.534), RM118 (0.531), RM11 (0.513) and RM29 (0.509), suggesting that these ten markers were highly informative (PIC > 0.50) for identi cation of the twenty aromatic rice cultivars. Among the remaining polymorphic SSR markers, 30 markers were informative (0.50 < PIC > 0.25) and 26 markers were slightly informative (PIC < 0.25). The level of polymorphism detected in the present study was consistent with the reported mean PIC values in previous works (Pal et al. 2004;Hossain et al. 2007;Sajib et al. 2012). However, Nadia et al. (2014) reported an average PIC value of 0.84, markedly higher than the present average PIC value. The presence of su cient polymorphism by the 66 SSR markers among the twenty indigenous Joha rice cultivars justi es their proper classi cation and use in the genetic improvement programme, based on the extent of genetic variation for desirable alleles.

Molecular diversity among the aromatic rice cultivars
An unweighted neighbour-joining (UNJ) cluster analysis based on Jaccard's similarity coe cients resolved the phylogenetic relationships among the aromatic rice cultivars collected from different parts of Assam. The UNJ cluster diagram showed three major clusters (G1, G2 & G3) with additional sub-clusters (Fig. 2). This dendrogram revealed that the cultivars derived from a genetically similar type clustered together. Cluster I comprised eight cultivars, whereas cluster II had ten cultivars, forming the most signi cant cluster. Cluster III had only two cultivars (Local Joha & Kunkuni Joha). Cluster I included Kon Joha 3, Jeera Joha, Kon Joha 4, Keteki Joha, Joha-Bihpuria, Joha-Golaghat, Kon Joha 5 and Kon Joha 2. Kon Joha-Bongaigaon, Kon Joha-Teok, Soru Joha-Tinsukia, Ronga Joha, Kola Joha, Kon Joha-Moran, Kalijeera, Kon Joha1, Manimuni Joha and Harinarayan belonged to cluster II. The cluster I sub-divided into two clusters, IA and IB, consisting of six and two cultivars. Cluster II had two sub-clusters, IIA and IIB, consisting of nine and one cultivars, respectively. Similar to the present investigation's clustering pattern, Meti et al. (2013) obtained two major clusters for 48 aromatic rice genotypes from Odisha using 12 SSR markers at 49 per cent genetic similarity. Shah et al. (2013) effectively differentiated the basmati cultivars from non-basmati ones based on the cluster analysis using 24 microsatellite loci, classifying 40 rice cultivars into three groups. Pervaiz et al. (2010) reported four signi cant clusters of 75 rice landraces using 35 SSR markers at 0.40 similarity coe cients. Thus, SSR markers provided an adequate resolution to discriminate between aromatic rice accessions, and it could serve as a potential tool in the identi cation and characterization of genetically distant accessions from various sources. The microsatellite assay generated genotype-speci c alleles in some of the cultivars evaluated for DNA ngerprints for cultivar identi cation and differentiation of aromatic rice. DNA ngerprints would be enormous assistance for establishing and defending the proprietary rights and maintaining the cultivar purity.

Conclusion
The present investigation on morpho-molecular and biochemical pro ling of a sample of popular indigenous Joha cultivars of Assam has been a step forward for exploiting variability in this unique rice class to improve its inherently low yield potential through breeding. In light of the results on diversity analyses, it is evident that the Joha rice cultivars are highly diverse regarding yield and quality traits and utilization of these diverse traits speci c genotypes to develop crop varieties with a broad genetic base would prove bene cial for aromatic Joha rice improvement programme. The low to a high degree of dissimilarity among the aromatic rice accessions observed in the present study exempli ed the high level of diversity at the molecular level among the aromatic rice cultivars used and indicated their possible utilization in breeding programmes targeted at developing elite aromatic rice varieties. Besides, marker-based identi cation and differentiation of aromatic rice cultivars might help maintain this high-quality product's integrity to bene t both farmers and consumers. The Joha rice cultivar Soru Joha (Tinsukia) with the highest yield (3012 kg ha − 1 ), high spikelet fertility (90.9%) and high Fe content (61.09 mg kg − 1 ) could be considered for an immediate recommendation in the state of Assam. The Joha rice cultivars' fatty acid pro le quali ed for extraction of quality bran oil for consumption.