Introduction
The quest to unravel the vibrant history of rice-centric agriculture in the Indian subcontinent has inspired many early researchers. Initial phases of agricultural origin and successive appearance of domestication syndrome has drawn the attention of researchers since long (Tewari et al., Reference Tewari, Srivastava, Saraswat, Singh and Singh2006; Fuller et al., Reference Fuller, Sato, Castillo, Qin, Weisskopf, Kingwell-Banham, Song, Ahn and Van Etten2010; Fuller, Reference Fuller2011a, 2008/Reference Tewari, Srivastava, Saraswat, Singh and Singh2009). Yet an understanding of the spatio-temporal expansion of agriculture, diversification of agro-ecosystems and growth of indigenous rice culture producing locally appreciated landraces remained at a rudimentary stage largely owing to a dearth of research.
The trajectory of rice agricultural involution has been greatly shaped by cultural, social and economic history of the agrarian societies (Geertz, Reference Geertz1970; Chang, Reference Chang1976; Barker, Reference Barker2011; Bray et al., Reference Bray, Coclanis, Fields-Black and Dagmar2015; Fuller and Castillo, Reference Fuller, Castillo, Lee-Thorp and Katzenberg2015). On one hand, rice fulfilled the calorific need of various human societies of varied cultural background; one the other, it was also assimilated into local traditions occupying a special place to be used in rituals, ceremonies, communal offerings. Moreover, wide spatio-temporal expansion enabled the plant to adapt to local climatic conditions. Hence, human socio-cultural attributes in concert with novel abiotic and abiotic elements have driven evolution of rice (Kumar, Reference Kumar1988; Chang, Reference Chang, Kiple and Ornelas2000; Fuller and Lucas, Reference Fuller, Lucas, Boivin, Crassard and Petraglia2017; Liu et al., Reference Liu, Lister, Zhao, Petrie, Zeng, Jones, Staff, Pokharia, Bates, Singh and Weber2017; Ray and Chakraborty, Reference Ray and Chakraborty2018; Ray and Ray, Reference Ray and Ray2018; Hufford et al., Reference Hufford, Berny Mier y Teran and Gepts2019). The process of diversification of cultivated rice stimulated the formation of many region-specific local germplasms that conferred cultural identity to the landraces (Chang, Reference Chang1976, Reference Chang, Kiple and Ornelas2000). These landraces are often closely knitted with unique environmental condition as well as agricultural systems of multitude of rice-dependent ethnic groups (Bennett, Reference Bennett, Frankel and Bennett1970; Frankel et al., Reference Frankel, Brown and Burdon1998; Brown, Reference Brown and Brush1999). The history of most of the landraces is unknown; however, ancient texts like Charaka Samhita (c. 700 B.C.), Susruta Samhita (c. 400 B.C.), Krishi-Parashara (c. 400 A.D.), Kashyapiakrishisukti (9th century) and many others are often replete with their names, features, cultural preference, culinary qualities and cultivation (Ahuja et al., Reference Ahuja, Ahuja, Thakrar and Singh2008a, Reference Ahuja, Ahuja, Thakrar and Ranib).
Prior to the green revolution (1965), India boasted around 200,000 indigenous landraces (Richharia and Govindaswami, Reference Richharia and Govindaswami1990). Unfortunately, thousands of these indigenous germplasms were lost owing to persistent advocacy of high yielding varieties and dominant market forces (Pingali, Reference Pingali, Hunter, Guarino, Spillane and McKeown2017). The North-Eastern states of India, a home to many ethnic groups of people, are also a bio-culturally diverse region. It has been estimated that more than 10,000 heirloom landraces are grown in a range of agro-ecosystems (Hore, Reference Hore2005). Many of them possessed various economically and culturally useful traits, e.g. disease and pest resistance, yield in marginal condition, high grain yield, aroma, early maturation, taste, stickiness, high anthocyanin content (Hore, Reference Hore2005; Das and Das, Reference Das and Das2014; Asem et al., Reference Asem, Imotomba, Mazumder and Laishram2015; Neog et al., Reference Neog, Sarma, Chary, Dutta, Rajbongshi, Sarmah, Baruah, Sarma, Sarma, Borah and Rao2016) and won cultural identities owing to their distinctive properties, such as Joha, the iconic aromatic landraces of Assam. Apart from Basmati-type long grain rice, the short and medium grain landraces from North-East India also contribute significantly to the diversity of aromatic rice. Despite their potential, the past researchers have broadly evaluated the genetic diversity (Choudhury et al., Reference Choudhury, Khan and Dayanandan2013; Das et al., Reference Das, Sengupta, Parida, Roy, Ghosh, Prasad and Ghose2013; Choudhury et al., Reference Choudhury, Singh, Singh, Kumar, Srinivasan, Tyagi, Ahmad, Singh and Singh2014; Roy et al., Reference Roy, Banerjee, Mawkhlieng, Misra, Pattanayak, Harish, Singh, Ngachan and Bansal2015, Reference Roy, Samal, Rao, Patnaik, Jambhulkar, Patnaik and Mohapatra2016a, Reference Roy, Marndi, Mawkhlieng, Banerjee, Yadav, Misra and Bansalb), but their history or the role of heirloom rice in agro-biodiversity conservation and management remained elusive, while both have a lot of bearing on the diversity. That we propose to address in this article employing a panel of standard rice micro-satellite markers.
Materials and methods
Samples and dataset
In this study, we have considered the genotype data of the North-East Indian landraces from two published reports by Roy et al. (Reference Roy, Banerjee, Mawkhlieng, Misra, Pattanayak, Harish, Singh, Ngachan and Bansal2015, Reference Roy, Samal, Rao, Patnaik, Jambhulkar, Patnaik and Mohapatra2016a, Reference Roy, Marndi, Mawkhlieng, Banerjee, Yadav, Misra and Bansalb). These reports described a total of 171 landraces from the six states, namely Assam (Joha = 54), Arunachal Pradesh (66), Manipur (Chakhao = 16), Mizoram (Tai = 9), Nagaland (21) and Sikkim (5); of which, 64 Arunachal Pradesh accessions were non-aromatic, so our dataset consisted of 107 aromatic and 64 non-aromatic. This dataset was used to analyse genetic diversity and population differentiation.
In order to determine population structure and subpopulation identity, we combined the data from Roy et al. (Reference Roy, Banerjee, Mawkhlieng, Misra, Pattanayak, Harish, Singh, Ngachan and Bansal2015, Reference Roy, Samal, Rao, Patnaik, Jambhulkar, Patnaik and Mohapatra2016a, Reference Roy, Marndi, Mawkhlieng, Banerjee, Yadav, Misra and Bansalb) with the global dataset from Garris et al. (Reference Garris, Tai, Coburn, Kresovich and McCouch2005). It enabled us to detect subpopulation type of indigenous rice in comparison to the global accessions of known subpopulation identity. So, the final dataset consisted of genotypes for 27 common microsatellites for all 405 accessions. We have employed a common panel of 27 microsatellite markers used in Garris et al. (Reference Garris, Tai, Coburn, Kresovich and McCouch2005) that reduced the originally used 169 markers to 27. But, this has not largely changed the assignment of rice accessions used in Garris et al. (Reference Garris, Tai, Coburn, Kresovich and McCouch2005) that justified our selection.
Genetic diversity and population differentiation
We calculated basic genetic diversity statistics, such as number of alleles, expected and observed heterozygosity of the North-East Indian landraces. We have also estimated fixation indices (G ST, G'ST and standardized corrected G''ST), pairwise differentiation and performed an analysis of molecular variance (AMOVA). All these analyses were conducted using GENODIVE (Meirmans and Van Tienderen, Reference Meirmans and Van Tienderen2004)
Population structure analyses
We have conducted population structure analysis employing two independent methods, model-based and multivariate-based. In the model-based analysis, we implemented admixture model in ParallelStructure with burn-in = 100,000, iterations = 10,000 (Pritchard et al., Reference Pritchard, Stephens and Donnelly2000; Miller et al., Reference Miller, Pfeiffer and Schwartz2010; Besnier and Glover, Reference Besnier and Glover2013). Since we have a priori knowledge of four standard subpopulations of rice, namely indica, japonica, aus and aromatic (Garris et al., Reference Garris, Tai, Coburn, Kresovich and McCouch2005), the optimal number of clusters was not determined. Instead, we have sought to determine the ancestry of the landraces with respect to the global accessions, i.e. accessions used in Garris et al. (Reference Garris, Tai, Coburn, Kresovich and McCouch2005). Hence, we have run the algorithm from k = 2–4 and considered q value above 0.8 to assign full ancestry to a subpopulation. In multi-variate-based analysis, a principal coordinate analysis (PCoA) was performed in GeneAlex ver.6.5 (Peakall and Smouse, Reference Peakall and Smouse2012) using covariance matrix. The purpose of this was to assess the population structure without employing a model a priori and to test whether the results are robust and invariable across the analyses.
Population assignment
In order to precisely determine the subpopulation identity (namely indica, japonica and aus) of the North-Eastern landraces, we have rerun STRUCTURE employing usepopinfo model. Aromatic accessions were excluded as it failed to form a distinct group in the population structure analysis. While assignment, core groups for indica, japonica, aus were created selecting 12–30 accessions with the highest ancestry (q) coefficient from prior analysis with admixture model. While core indica and core japonica consisted of 30 accessions each, the core aus group comprised 13 accessions with q value falling between 0.8 and 0.99. The rest of the aus accessions have q value below 0.8 and hence discarded from further analyses. The power of the usepopinfo model is that it can assign samples of unknown identity to the pre-defined groups of known identity. Following this, North-Eastern landraces were evaluated by assigning them to any of the core groups.
In a similar manner, PCoA was also performed using the same core indica, japonica and aus group along with the North-East Indian (unknown) samples to compare with the results obtained from usepopinfo model in STRUCTURE.
Results
Population differentiation and genetic diversity
The six populations showed low (Nagaland) to high diversity (Arunachal Pradesh) in terms of the number of alleles, effective number of alleles and heterozygosity (N A = 2–7.4; N EA = 1.3–3.7; H S = 0.22–0.64) (online Supplementary Table S1a). An AMOVA revealed partitioning of variance mostly within population (75%) and among population (25%) with F ST = 0.25 (online Supplementary Table S1b). Whereas G ST, G'ST and standardized corrected G''ST estimates ranged between 0.25 and 0.577 implying populations are highly structured. Pairwise differentiation estimates depicted minimal segregation among Joha, and landraces from Mizoram, Sikkim; but they were at moderate level differentiation with Arunachal Pradesh and Manipur, and highest with Nagaland (Table 1).
Table 1. Pairwise differentiation of six state groups comprising landraces
Significant at P = 0.005.
**Significant at P = 0.001.
Genetic structure analyses
Both the model-based structure analysis and multivariate analyses revealed an almost similar result. In PCoA, PC1 and PC2 explained almost 13.13 and 6.75% of the variance, respectively, and explicitly segregated into four major subpopulations on multivariate space. A major fraction of Joha, Sikkim, and a few of Tai landraces laid closest to aus. Chakhao were mostly spaced near to indica, aus and aromatic. While almost all Nagaland landraces and almost 40% of Arunachal Pradesh were located closest to japonica, the rest of the Arunachal Pradesh landraces were scattered among four groups revealing admixture (Fig. 1(a)).
Fig. 1. (a) PCoA of the North-Eastern landraces with all the global accessions in Garris et al. (Reference Garris, Tai, Coburn, Kresovich and McCouch2005); (b) PCoA of the North-Eastern landraces with core groups created after STRUCTURE assignment of global accessions in Garris et al. (Reference Garris, Tai, Coburn, Kresovich and McCouch2005) [inset box showing the legends: blue diamond – indica, black square – japonica, orange triangle – aus, purple cross – aromatic, red circle – Joha, yellow circle – Chakhao, black plus – Tai, inverted purple triangle – Nagaland landraces, green circle – Arunachal Pradesh landraces, green squares – Sikkim landraces].
The selection of a subset of 27 markers from a total of 169 used in Garris et al. (Reference Garris, Tai, Coburn, Kresovich and McCouch2005) has not grossly affected the assignment of global accessions. It strengthened our inference on assignment with the selected subset of microsatellites. Model-based analyses yielded similar break-up of subpopulations, indica, aus and japonica. The multivariate space was occupied by three distinct subpopulations with high ancestry (q > 0.8) coefficient. But, aromatics failed to form a distinct group and demonstrated admixture even when the value of K increased from 3 to 4. Two of 18 aromatic accessions were assigned to aus (with q > 0.8) and a set of eight accessions were admix (0.6 < q < 0.8). Almost all the Joha (49 of 54 for K = 3) are assigned to aus group with high q value. Similarly, eight of nine Tai and four of five Sikkim landraces were also categorized as aus (Table 2a, online Supplementary Table S2). Whereas, 20 out of 21 landraces from Nagaland assigned to japonica, one-third of Arunachal Pradesh were similarly assigned to japonica, the rest were admixed. Likewise, five landraces of Chakhao were indica and the rest were admixed with shared ancestry. The assignments remained the same when K = 4 was chosen as the optimum number of clusters, except a majority (77.2%) of Arunachal Pradesh landraces formed a fourth cluster and were not assigned to either of the subpopulations (Table 2b, online Supplementary Table S3).
Table 2. The assignment of landraces into sub-populations using the admixture model (K = 3), (K = 4) and usepopinfo model (K = 3)
Determination of subpopulation identity using core group
Usepopinfo model precisely assigned most of the North-Eastern landraces to specific subpopulations. The assignment was almost similar to the one obtained from the admixture model except for four Joha landraces. These were grouped with aus employing admixture model initially but resembled admixed type in usepopinfo model. The other major point of difference was the assignment of Chakhao, half of which were grouped in indica but admixture model revealed their admixed origin (Table 2c, online Supplementary Table S4). The PCoA with core groups mostly yielded comparable results that mirrored model-based analyses. In PCoA, Chakhao and Arunachal Pradesh landraces were mostly located between indica and japonica in the gene space and appeared as admixed (Fig. 1(b)).
Aromatic rice in aus: short and medium grain aromatic
More than half of the aromatic landraces were assigned to aus (55.1%), the remaining were japonica (18.6%), indica (4.6%) and admix types (21.7%). A majority (47 out of 59) of aromatic rice varieties in aus group were short to medium grain aromatic type (Table 3) where short (S) is ⩽5.50 mm, medium (M) is 5.51–6.60 mm, long (L) is 6.61–7.50 mm, extra-long (EL) is >7.5 mm (Juliano and Villareal, Reference Juliano and Villareal1993). On the other hand, aromatic landraces in japonica were mostly dominated by long (7) and extra-long grain (10) types. All three types, short (1), medium (2) and long (2) grains, were represented in indica aromatic (Table 3).
Table 3. A table showing grain size of aromatic landraces and their corresponding sub-populations (S, short, M, medium, L, long, EL, extra-long; admix not shown)
Discussion
The spread and diversification of rice culture had been very complex and often associated with various evolutionary events, such as bottleneck, expansion, introgression (Sweeney et al., Reference Sweeney, Thomson, Cho, Park, Williamson, Bustamante and McCouch2007). Humans, the most predominant disseminator, carried rice grains as food, seed, merchandise and gifts to places covering great distances (Chang, Reference Chang, Kiple and Ornelas2000; Barker, Reference Barker2011; Fuller and Castillo, Reference Fuller, Castillo, Lee-Thorp and Katzenberg2015). In addition, selective regimes owing to cultural preference of multitude of agrarian groups had played a crucial role in crop evolution, thus shaping its phenotype and the underlying genome (Zeven, Reference Zeven1998; Chang, Reference Chang, Kiple and Ornelas2000; Fuller and Lucas, Reference Fuller, Lucas, Boivin, Crassard and Petraglia2017). The collective effect was a wider agro-ecological range and adaptive amplitude of Asian cultivated rice; unlike relatively narrow ecological spectrum of its wild ancestors.
In our study, we unravelled a divergent genetic history of various rice landraces of North-East India entangled with their rich cultural legacy. It perhaps led to differentiation across cultures and enabled the conservation of rice biodiversity. Our interpretation also offered an additional insight into the independent origin of aroma in the aus group, not in group V as commonly held notion emphasized.
Agrobiodiversity management and conservation
Many ethnic tribal groups of the North-Eastern India artificially selected, cultivated and conserved their signature landraces over centuries that eventually have become their cultural flag. Joha epitomizes Assamese culture, Tai has been an identifier of Mizoram and so on. The diversity of rice genetic resource from this region has drawn the attention of the scientists for decades. The endeavour has characterized a large number of landraces that provided some preliminary insight on genetic diversity and its distribution (Vairavan et al., Reference Vairavan, Siddiq, Arunachalam and Swaminathan1973; Choudhury et al., Reference Choudhury, Khan and Dayanandan2013; Das et al., Reference Das, Sengupta, Parida, Roy, Ghosh, Prasad and Ghose2013; Choudhury et al., Reference Choudhury, Singh, Singh, Kumar, Srinivasan, Tyagi, Ahmad, Singh and Singh2014; Roy et al., Reference Roy, Banerjee, Mawkhlieng, Misra, Pattanayak, Harish, Singh, Ngachan and Bansal2015; Roy et al., Reference Roy, Samal, Rao, Patnaik, Jambhulkar, Patnaik and Mohapatra2016a, Reference Roy, Marndi, Mawkhlieng, Banerjee, Yadav, Misra and Bansalb). A similar pattern of genetic diversity in these groups has also been detected in our analyses; however, in some cases (e.g. Sikkim), relatively smaller sample size constrained the correct assessment.
The choice of diverse cultural groups over centuries or millennia has tapped and curated the diversity in these heirloom crop landraces through a long process of experimentation replete with trials and errors (Chang, Reference Chang, Smith and Dilday2002; Salick, Reference Salick, Gepts, Famula, Bettinger, Brush, Damania, McGuire and Qualset2012). Apart from cultivation, most or many cultural groups also maintained their own seed stock valued for ritual uses (Ahuja and Ahuja, Reference Ahuja and Ahuja2006). They perhaps rarely exchanged seeds with different neighbouring groups, this resulted into genetic isolation and differentiation largely owing to divergent ancestry, selection and drift (Zeven, Reference Zeven1998; Villa et al., Reference Villa, Maxted, Scholten and Ford-Lloyd2005). The fact was reflected in the high values of fixation indices, implying a presence of high population structure. Likewise, the pairwise indices revealed that six state-wise groups are at various levels of divergence, ranging from minimal to high degree. While groups falling in aus (Joha, Tai, Sikkim accessions) are in little to moderate differentiation among themselves (0.09–0.142), they were in moderate to high with the others (0.167–0.272); on the other hand, Nagaland accessions included in japonica was highly differentiated from the rest (0.32–0.58).
Evolutionary and cultural history
Knowledge of history is often crucial to gain an insight into the evolutionary origin of the local landraces that confer cultural identity to a region, the genes underlying certain culturally or economically important traits, and traits responsible for local adaptation (Barker, Reference Barker2011). Origin and evolution of the indigenous landraces have been shaped by an interplay between genetic and socio-cultural history, and is often difficult to trace due to lack of sufficient information. Here, certain anecdotes on Joha and Chakhao are available from the ancient texts describing past agriculture, such as, the Sukhapa and Ahom dynasties (1228 A.D.) (Guha, Reference Guha1967, Reference Guha1984; Bhuyan, Reference Bhuyan1974) had introduced late maturing but high yielding sali rice varieties (Guha, Reference Guha, Raychoudhury and Habih1982; Hamilton and Bhuyan, Reference Hamilton and Bhuyan1987). Joha is also classified traditionally under sali rice category (Das et al., Reference Das, Kesari and Rangan2010). The earliest mention of Joha rice is from 14th century Assamese version of epic Ramayana (Saptokandt Ramayana) by poet Kaviraja Madhava Kandali (Ahuja, Reference Ahuja, Ahuja, Thakrar and Rani2008b). Similarly, the iconic Chakhaopoireiton derived its name from the legendary hero Chingkhong poireiton of 34–18 B.C. (Ahuja, Reference Ahuja, Ahuja, Thakrar and Rani2008b). The name Cha-hao-mubi or amubi (Chakhao amubi) was borrowed from three Manipuri words cha, hao and mubi meaning rice, tasty and black, respectively (Sanajaoba, Reference Sanajaoba1988). On the other hand, the origin of the other heirloom rice is almost unknown.
Here, we uncovered a complex history of rice diversification in the frontier provinces with insights from population genetics data. Despite the marginal differences, our analyses with PCoA and model-based assignment were largely in agreement. Prior research has detected an existence of japonica and indica-type landraces in the North-Eastern region (Roy et al., Reference Roy, Banerjee, Mawkhlieng, Misra, Pattanayak, Harish, Singh, Ngachan and Bansal2015), but a preponderance of aus in states of Assam, Sikkim, Mizoram is an important finding (Fig. 2). Roy et al. (Reference Roy, Banerjee, Mawkhlieng, Misra, Pattanayak, Harish, Singh, Ngachan and Bansal2015) have assigned Joha, Manipur, Sikkim landraces either to indica or close to aus and indica but they failed to append any explanation. Moreover, their assignment suffered from methodological weakness owing to an undefined selection of 67 global accessions from a total of 234 accessions in Garris et al. (Reference Garris, Tai, Coburn, Kresovich and McCouch2005). Jain et al. (Reference Jain, Jain and McCouch2004) classified Joha as aromatic indica derived from the hybridization between indica and basmati-type aromatic. Prevalence of aus in the North-east suggests an expansion of aus cultivation beyond its core areas of eastern Indian states and Bangladesh. The intrusion, acceptance and assimilation of aus into the local culture of Assam and other neighbouring states perhaps had not been difficult given their geographic proximity with the primary aus growing region (Fig. 3). The current understanding of the aus places its origin in Jeypore tract of Odisha; but the foci of cultivation stretched to a larger part of undivided Bengal which nurtured a rich tradition of aus or aus-boro cultivation (Civan et al., Reference Civan, Craig, Cox and Brown2015; Civan and Brown, Reference Civan and Brown2018; Ray and Ray, Reference Ray and Ray2018). Intermittent migrations from Nepal, Bhutan, Myanmar and Indian states to Assam had taken place throughout history (Sarma, Reference Sarma2015). The Bangladeshi migrants have introduced new agrarian practices in fertile flooded fields of Assam that diversified land usage, e.g. multiple cropping, double cropping (winter and summer cropping) (Guha, Reference Guha1980; Sarma, Reference Sarma2015).
Fig. 2. State-wise assignment of landraces into three subpopulations of rice, indica, japonica, aus, landraces with shared ancestry was grouped as admix. The circles are roughly proportional to the sample size [yellow – aus, green – indica, red – japonica, blue – admix].
Fig. 3. The putative pathways for spread and mixing of rice culture (including Southern and Southwestern Silk route but excluding maritime Silk route).
On the other hand, the assignment of Manipur (Chakhao) accessions to indica and admix, and Nagaland accession to japonica received support from Roy et al. (Reference Roy, Banerjee, Mawkhlieng, Misra, Pattanayak, Harish, Singh, Ngachan and Bansal2015). We also found a clear signal of japonica invasion into Arunachal Pradesh. However, a sizable fraction of Arunachal Pradesh accessions did not completely fall in any particular group; they displayed shared ancestry or even formed a separate cluster (online Supplementary Table S3). This may represent a genetically distinct group locally evolved like many other regionally domesticated landraces (such as Rayada, Ashwina) in South Asia (Rahman and Zhang, Reference Rahman and Zhang2013).
An invasion of japonica-type in Nagaland, Arunachal Pradesh and an occurrence of admix types in Manipur could have been expedited by recurrent trades (Fig. 3). The proximate pathway may be the established Southern and Southwestern Silk routes facilitating exchange of tea, spices, spring salt, hemp, silk, horses, etc. (Pemberton, Reference Pemberton1839; Lahiri, Reference Lahiri1991; Yang, Reference Yang2008). Southern Silk route connecting China's Sichuan and Yunnan extending up to Tibet (ChaMadao), Nepal and finally to India was functional since 150 B.C. (Pemberton, Reference Pemberton1839; Bhattasali, Reference Bhattasali1946; Fuquan, Reference Fuquan2004; Lu et al., Reference Lu, Zhang, Yang, Yang, Xu, Yang, Tong, Jin, Shen, Rao and Li2016). A part of Southern Silk route was Southwestern Silk route, connecting China (Southern Yunnan), upper Myanmar to eastern India existed as early as 400 B.C. (Yan et al., Reference Yan, Yifu, Zuan, Heyi, Suichu and Sun1989; Lahiri, Reference Lahiri1991; Yang, Reference Yang2008). Moreover, the occurrence of japonica was common across South Asia by 200 B.C. which provides additional support to the premise of long-term trade relations involving grain exchange (Kingwell-Banham et al., Reference Kingwell-Banham, Bohingamuwa, Perera, Adikari, Crowther, Fuller and Boivin2018). Archeological records in the middle Ganges region around 1400 B.C. demonstrated the earliest agricultural diffusion into the Indian plains via Assam (Saraswat, Reference Saraswat and Singh2004, Reference Saraswat2005; Fuller, Reference Fuller2011b).
Aroma evolution in aus landraces
The aromatic rice has been deep rooted in the socio-cultural life of Indian subcontinent (Singh et al., Reference Singh, Singh and Khush2000; Ahuja et al., Reference Ahuja, Ahuja, Thakrar and Rani2008b). Amongst all, iconic Basmati enjoyed consumers̕ coveted preference from historical past and has become almost synonymous with the aromatic rice of the Indian subcontinent. Currently, japonica origin of the causal polymorphism followed by introgressions offers an explanation for the origin of aroma in Basmati-type long grains prevalent in South Asia (Kovach et al., Reference Kovach, Calingacion, Fitzgerald and McCouch2009). On the other hand, an alternative hypothesis, though less dominant, does not preclude the probability of ancestral origin of aroma in perennial Oryza rufipogon (Prathepha, Reference Prathepha2009). Here, we would like to emphasize on the fact that Basmati-type long grains represent only a small fraction of aromatic rice diversity of India. A large number of short and medium grain aromatic landraces is still cultivated, cooked and relished in various parts of India (Ray et al., Reference Ray, Deb, Ray and Chattopadhayay2013; Chakraborty et al., Reference Chakraborty, Deb and Ray2016; Roy et al., Reference Roy, Samal, Rao, Patnaik, Jambhulkar, Patnaik and Mohapatra2016a, Reference Roy, Marndi, Mawkhlieng, Banerjee, Yadav, Misra and Bansalb). History of these under-explored landraces is not precisely known; yet, a prior research claimed an independent origin (Ray et al., Reference Ray, Deb, Ray and Chattopadhayay2013; Chakraborty et al., Reference Chakraborty, Deb and Ray2016). Khush (Reference Khush, Singh, Singh and Khush2000), in an early effort, categorized Indian aromatic rice varieties as indica. However, later studies have grouped most of the aromatic rice cultivars of the Indian sub-continent, including Basmati types, into a genetically distinct cluster (Glaszmann, Reference Glaszmann1987; Aggarwal et al., Reference Aggarwal, Shenoy, Ramadevi, Rajkumar and Singh2002; Nagaraju et al., Reference Nagaraju, Kathirvel, Ramesh Kumar, Siddiq and Hasnain2002; Garris et al., Reference Garris, Tai, Coburn, Kresovich and McCouch2005), their focus mostly dwelt on long-grain Basmati-types. Jain et al. (Reference Jain, Jain and McCouch2004) classified short-grained joha as aromatic indica and proposed an evolutionary origin via hybridization between indica and Basmati rice varieties. Several subsequent studies have analysed and classified various collections of aromatic rice and categorized them as either indica or japonica (Das et al., Reference Das, Sengupta, Parida, Roy, Ghosh, Prasad and Ghose2013; Choudhury et al., Reference Choudhury, Singh, Singh, Kumar, Srinivasan, Tyagi, Ahmad, Singh and Singh2014; Roy et al., Reference Roy, Banerjee, Mawkhlieng, Misra, Pattanayak, Harish, Singh, Ngachan and Bansal2015; Roy et al., Reference Roy, Samal, Rao, Patnaik, Jambhulkar, Patnaik and Mohapatra2016a, Reference Roy, Marndi, Mawkhlieng, Banerjee, Yadav, Misra and Bansalb).
Our study comprised a set of landraces from North-East India, of which a majority (63%) is aromatic. In a two-step examination, we employed both model-based assignment and multivariate analyses to determine the subpopulation identity in a more rigorous manner than previously performed by Roy et al. (Reference Roy, Banerjee, Mawkhlieng, Misra, Pattanayak, Harish, Singh, Ngachan and Bansal2015). It proffered an important insight into the origin of aroma in rice. Almost all Assam, Mizoram and Sikkim landraces were assigned to aus, of which a major fraction (80%) is short and medium grain (Table 3). The existence of many aus-type short and medium grain aromatic rice is a salient finding since it predicts a probable origin of aroma in aus; independently in short grain aromatic landraces, that have not followed a similar trajectory as long-grain basmati type (Ray et al., Reference Ray, Deb, Ray and Chattopadhayay2013). Though we have not examined the history of badh2 gene of the landraces and the wild populations, the probability of independent origin of aroma in aus subpopulation is supported by the facts: (a) separate and independent domestication of aus sub-populations of rice with a little gene flow from japonica and indica (Civan et al., Reference Civan, Craig, Cox and Brown2015; Civan and Brown, Reference Civan and Brown2018), (b) separate origin of many domestication alleles of aus that holds novel gene space (Schatz et al., Reference Schatz, Maron, Stein, Wences, Gurtowski, Biggers, Lee, Kramer, Antoniou, Ghiban and Wright2014), (c) a probability of the presence of aroma in wild ancestral population (Prathepha, Reference Prathepha2009). Adhering to the above reasoning, the origin of aroma in wild ancestors appeared to be a relatively robust conclusion. Following the origin, it might have traced divergent pathways and became fixed over time.
Conclusion
Building on these facts, it seems that human culture had been moulding rice evolution and diversification for a long time. On one hand, it reflected the role of indigenous farmers' knowledge towards conservation and management of traditional rice landraces. On the other hand, it provided a glimpse into the divergent history of various cultural groups coupled with evolutionary processes which had shaped the divergence of rice groups and promoted the dissemination of rice culture. Our study also cast light on the evolution of aroma through an alternate pathway.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1479262119000273
Acknowledgements
Authors would like to thank Dr Utpal Basu of the Department of Molecular Biology and Biotechnology, University of Kalyani, Surajit Ghosh and Debanjan Bhattacharya for their suggestion and assistance at various stages of the study. Language editing by Karen Meehan, UK is acknowledged. The computational help from CIPRES is also acknowledged. The fellowship of the first author was supported by Mr Avik Saha, Co-convenor, Jai Kisan Andolan. This study has not received any fund. The authors declare no conflict of interest.
