1. Introduction
The yield advantage of hybrid progenies over their parents, the heterosis, has contributed immensely to boosting the productivity of rice. In self-pollinated crops, heterosis is relatively weak compared to cross-pollinated crops, due to the dominion of homozygous balance [
1]. Therefore, parental selection to promote cross-fertilization is of paramount importance in breeding hybrid rice. A significant breakthrough in hybrid rice breeding came with the discovery of wild-abortive (WA) cytoplasmic male sterility (CMS), which promoted a great deal of controlled cross-fertilization [
2]. However, the WA system was found to work better within the
indica subgroup than within the
japonica, due to poor fertility restoration in the latter [
3]. Using CMS systems, a significant effort has been placed into developing several rice hybrids in the last four decades worldwide. In India alone, 127 rice hybrids have been commercially released during the last three decades all in the
indica background. Among the heterotic systems in rice, however, intersubspecific crosses are more heterotic than intrasubspecific crosses [
4]. Yield heterosis in rice can be arranged in the descending order of sub-specific combinations as
indica/temperate
japonica >
indica/tropical
japonica > temperate
japonica/tropical
japonica >
indica/
indica >
japonica/
japonica [
5]. The development of intersubspecific hybrids, particularly the
indica/
japonica combination, remains poorly resolved due to the major limitation of hybrid sterility. Although the singular or dual origin of rice is still debated [
6,
7], the subspecific differentiation is prominently associated with the reproductive barrier between
indica and
japonica subtypes [
8]. Further divergence of phenotypes and adaptations among
indica and
japonica subspecies resulted in their genetic differentiation. Despite the heterotic potential,
indica–
japonica hybrids were unsuccessful due to several hindrances
viz., low seed setting, poor spikelet filling, and transgression of plant height and growth duration [
9]. Therefore, it is important to break the crossability barrier between the two sub-species and harness the heterotic potential of the
indica/
japonica hybrids.
The major hindrance in producing viable hybrids from
indica/
japonica crosses is hybrid sterility, which occurs to varying degrees [
10]. Sterility can occur either due to prezygotic or postzygotic mechanisms. The prezygotic mechanism prevents the formation of hybrid seed, while the postzygotic mechanisms occur after fertilization and seed formation and encompass hybrid necrosis, hybrid weakness, hybrid sterility/semi sterility and lethality. Hybrid sterility is one of the key forms of postzygotic reproductive isolation between
indica and
japonica, caused due to the failure of chromosome pairing at the meiosis stage abetted by the structural differences between parental chromosomes [
11]. One of the major genetic systems regulating the reproductive isolation of
indica and
japonica is the triallelic system of the
S5 locus [
12]. Popularly known as the wide compatibility (WC) system, the
S5 locus has three alleles,
S5-i,
S5-j and
S5-n [
13]. Localized to chromosome 6 are two specific alleles,
S5-i to
indica sub-species and
S5-j to
japonica sub-species. The third allele, the
S5-n allele, is neutral and does not show any subspecies allegiance. Among the alleles, the hybridity of
S5-i and
S5-j alone leads to sterility, while all the other combinations result in fertile grains. Plants with the
S5-n/
S5-j (or
S5-i) genotype are fully fertile, while plants with the
S5-i/
S5-j genotype have poor fertility. Specifically, the genotypes carrying
S5-n alleles can cross with both
indica and
japonica lines effectively, and are therefore called wide compatible varieties (WCV). The
S5-n allele has a 136 bp deletion in the
S5 gene, causing the deranged sub-cellular localization of the aberrant S5 protein and rendering the gene non-functional. Therefore, whenever a WCV (
S5-n) was crossed with
indica (
S5-i) or
japonica (
S5-j) genotype, fertile hybrids would be produced [
14]. Because of this ability, WCVs can act as the bridge for
indica–
japonica hybridization. WCVs have been known for a long time to the rice breeders as genotypes that cross freely with both
indica and
japonica lines. Therefore, WC has been recognised as one of the key requirements to overcome the fertility barrier among intersubspecific hybrids [
15]. Moreover, WCVs can act as potential donors for
S5-n alleles, which can replace either
S5-i or
S5-j alleles in the parental stocks to generate potentially viable hybrids. Integrating the portability of the
S5-n allele into WA-CMS restorers that are locally well adapted and agronomically desirable can generate wide compatible restorer (WCR) lines. Conventionally, WCVs could be identified through test crossing and analysing the spikelet fertility of the F
1 hybrids, which is tedious and time consuming. Moreover, identification on the basis of spikelet fertility data or by using other morphological traits associated with the WC gene may be inconclusive due to interactions from the external environment. The recent development of molecular markers linked to the
S5 locus could aid to overcome these limitations. A PCR-based co-dominant
S5 functional multiplex marker system (
S5-MMS) could identify
indica and
japonica alleles based on the SNP they are carrying and also the
S5-n allele based on the deletion [
14].
Among the subspecies of
Oryza sativa, there is a third group of genotypes that were characteristically distinct from the major subtypes. Known previously as
javanica types, and later as tropical
japonica, these groups of genotypes show an intermediary behaviour between
indica and
japonica [
15]. Most of these types are also characteristically cross compatible with both
indica and
japonica genotypes and are known to harbour WC alleles. However, utilization of tropical
japonica in hybrid rice breeding has been limited due to their lower productivity and other undesirable traits [
16]. Further, the information on their natural diversity among these group of genotypes is restricted, and they are less well characterized at the molecular level [
17]. Notwithstanding, the natural diversity and distribution of the
S5-n allele are not limited to tropical
japonica. Two earlier known WCVs, Dular and Nagina 22, come from the
aus type of rice. To utilize the WC system in hybrid diversification in rice, it is necessary to use genotypes with suitable allele combinations. It may be difficult to utilize WCVs for developing hybrids directly, because of the undesirable traits they may carry. This requires pre-breeding to incorporate desirable alleles including WC into suitable backgrounds and then to improve parental lines for further hybrid breeding. Therefore, recruiting the
S5-n allele into the parental stock should be considered a preliminary step for the development of heterotic
indica/
japonica hybrids. Moving towards this, in the present study, we have surveyed a diverse set of 950 rice genotypes, including tropical
japonica lines and several other germplasm lines, for WC system diversity. Since not much information is available, other than for Dular and Nagina 22, we have further characterized the WCVs for their agronomic potential and hybrid fertility.
3. Discussion
Currently, the extent of rice cultivation in India is ~44 million hectares (mha), which is the highest in the world. With an annual production of ~118 million tonnes (mt), India ranks second after China in production. China produces 211.1 mt from 30.44 mha, a productivity of 69.3 q/ha as compared to India’s productivity of 23.9 q/ha [
18]. The major reasons for China’s productivity can be attributed to the predominant cultivation of
japonica rice, as well as to the formidable success in extensive cultivation of rice hybrids. India almost totally grows
indica rice, which is less productive than
japonica. Further, the extent of cultivation of hybrid rice is marginal (6.8%) in India, extending ~3.0 mha, as compared to ~48% in China [
19]. The demand for rice in India is poised to take a quantum leap in the future, 16% by 2030, 41% by 2040, and 67% by 2050 from the current production level [
20]. To bridge the ensuing gap in food demand, rice production needs to be boosted, and one of the probable approaches is extending hybrid rice cultivation. However, the popularity of rice hybrids in India has been very slow, primarily due to factors such as low marginal yield gain over pure line varieties, increased seed cost and the need for seed renewal at every cropping cycle [
21]. To meet this challenge, heterosis in rice hybrids needs to be improved.
Rice hybrids are to be produced independently for different utilisation segments and ecologies to achieve sustained spread and extensive cultivation. Therefore, the diversification of parental lines for hybrid rice needs to be context specific. This needs a holistic approach with respect to the requirements of the target segment, including grain quality as well as general genetic systems that improve yield heterosis. If the indica–
japonica hybrids can be realised with the same ease of producing three-line or two-line hybrids as currently practised in rice, the challenge of heterotic boosting can be met rather easily. Judicious deployment of the WC system is hence required to meet this task. However, the availability of WCVs across different rice segments is rather limited and comprehensive information is seldom available on the diversity of the WC genetic system. Further, only a limited number of WCVs have been utilized in the intersubspecific hybridization program [
22]. This requires an extensive survey across various rice types and source WCVs within each group to develop/improve the parental stocks with WC. The availability of marker systems targeting the
S5-n allele can simplify this task by integrating the allele through marker-assisted breeding.
More than 50 loci have been identified associated with hybrid sterility in rice [
23]. Among all loci causing embryo sac abortion, the
S5 locus has been recognised as prominent in controlling indica–
japonica hybrid compatibility. Therefore, the
S5 alleles pertaining to the two major subspecies ensure their reproductive isolation. However, the neutral allele,
S5-n is found distributed in both
indica and
japonica types with varying frequencies. This study, despite being dominated by the
indica genotypes, indicated an overall distribution of the
S5-n allele with a frequency of 0.17. The allelic frequencies among the 950 genotypes tested were 0.69 for the
S5-I allele and 0.14 for the
S5-j allele. The
indica genotypes possessed the
S5-n alleles with a frequency of 14%, whereas the
japonica genotypes indicated a larger proportion of
S5-n, with a frequency of 27%. It may be pertinent here to emphasize that most of the
japonica types used in this study came from the tropical
japonica group, which is already known to harbour a relatively higher proportion of the
S5-n allele [
24]. Apart from this, the Basmati/aromatic group did not carry the WC allele. Almost all of the
indica types contained
S5-i, as expected, except for two IRG lines, IRG-17 and IRG-20. In the case of
japonica types,
S5-j alleles predominated. There were three
aus genotypes that carried the
S5-n allele. In an earlier study, Revathi et al. [
24] found 46% of the genotypes possessing the
S5-n allele using an S5-InDel marker, along with 0.06% of the heterozygotes. However, we could not find any
S5-n heterozygotes in this evaluation. There are also reports of the presence of 19.9% WCVs among a germplasm set of 584 genotypes comprising 154 cultivars, 207 IRRI germplasm accessions, 37 aromatic genotypes, 157 restorer lines, 12 maintainer lines and 17 breeding lines [
14]. All the genotypes that were identified to carry the
S5-n alleles in this study can be construed as novel sources of WC, as they have never been intentionally used for intersubspecific hybridisation. These 150 WCVs can further be utilized in parental diversification and the development of indica/
japonica-based breeding programmes.
For further utilization of the WCVs, it is essential to characterise them agronomically and genetically. Genetic diversity could help classify the genotypes into different target classes, either agronomically, such as duration, yield and plant stature, or genetically, into subpopulations. In order to do this, we further pooled the 160 WCVs identified to draw a random subset for diversity analysis. The subset contained 92 WCVs along with four checks for agronomic and molecular evaluation. Agronomically, the genotypes displayed a three-cluster formation. Irrespective of the genetic grouping based on genomic differences, the agronomy-based grouping helped us to practically utilise the genotypes for breeding based on the similarity/contrast of quantitative and morphological traits [
25]. This aids in formulating crossing schemes considering flowering synchrony, as well as by matching the duration classes. High- and low-yielding types can also be distinguished. Moreover, photo-sensitive and -insensitive lines can be distinguished for choosing the breeding seasons. Therefore, morphology-based grouping can supplement the genetic grouping in designing effective breeding strategies for forming an intersubspecific breeding program [
26]. Using a similar strategy, Kumar et al. [
27] grouped iso-cytoplasmic rice restorers derived from the commercial hybrids into two major groups based on morphological traits. The number of groups based on agronomic traits often varies from the genetic groups because of the extreme phenotypic variation that can occur within a genetically similar class. There are earlier reports in which a small group of 41 rice genotypes were grouped into 6 clusters based on 13 morphological traits [
28].
As emphasised above, molecular marker-based diversity studies provide the opportunity to classify the population based on genomic similarities. Genotypes based on 71 SSR markers and 92 WCVs fell into two subpopulations in both the graphical and model-based approaches. The estimated genetic diversity of the markers indicated sufficient resolving power suitable for genetic groping of the WCVs. The overall genetic distance revealed that the genetic variability present among these lines was moderate, with 2.80 alleles per marker and an average PIC value of 0.41, with an overall Nei’s gene diversity of 0.49. The average allele number per marker found among the WCVs was similar to that recorded from genotype sets with a narrow genetic base, such as Japanese [
29] and Korean cultivars [
30], but lower than from diverse germplasm sets [
31,
32,
33,
34,
35]. The allele frequency divergence between the subpopulations or subgroups of genotypes indicates the population structure [
36,
37]. The genotypes from different subpopulations are assumed to be descended from different sets of ancestors that have become isolated in the evolutionary process. Therefore, there may be common alleles distributed across the whole population along with specific alleles that are characteristic of subpopulations. Therefore, an admixed model with correlated allelic frequencies is preferred in estimating the population structure. The allelic frequencies either use identical by descent or identical by state probabilities. The assumed subpopulations (K) therefore show a maximum difference in the log probability at each K, Pr(X|K) at the best K [
38]. Accordingly, the two subpopulations identified in this study, one with 31 genotypes and the other with 49 genotypes indicated a large level of admixing between the subpopulations. The same Bayesian model-based approach has been used to study the population structure of plant populations by various researchers. In most of the studies on rice subpopulations, two subpopulations are most commonly detected, pertinently due to indica–
japonica subspecific differentiation [
39]. However, there are reports of up to 7 subpopulations reported among 416 rice accessions collected from China [
33]. Besides, the admixtures are common in rice subpopulations, indicating high admixing between populations. However, there are currently no studies reporting the population structure of WCVs, except for the inclusion of a few genotypes among the large populations studied [
40]. Genetic differentiation among WCV subpopulations indicated that they are moderately divergent and contained admixtures (28.6% in POP1 and 24.6% in POP2). Based on genetic differentiation among the wide compatible varieties identified, within-group variability was high among the subpopulations. The genetic diversity identified herein may not reflect the phenotypic diversity effectively, since the SSR markers are mainly present in non-coding regions [
41]. The use of random markers for assessing genetic diversity might not reflect the functionally useful variations prevalent in the coding regions of the genome [
42], and thus may not classify genotypes based on the quantitative expression of traits. Nevertheless, the overall genetic diversity would be useful in selecting genetically diverse parents for breeding programmes. Phenotypic analysis, therefore, needs to be integrated into genetic variations for ensuring the effective use of functional variants in breeding. The selection of parental lines from distant clusters would help in improved heterosis, which may further improve the yielding potential of hybrids.
The agronomic performance of WCVs indicated that they can be grown in wider ecologies, which is essential for their effective utilisation as WC donors. Donor genotypes with lesser diversity with the target set of recipients can help in improved breeding efficiency due to fewer alterations in the target traits and leverages the recovery of the recipient genome.
Yet, WCVs are to be evaluated for their ability in cross compatibility. We have selected a subset of 41 genotypes to test this, and the selection was made based on their flowering behaviour and plant type. Two
indica parents, Pusa 44 and IR64, were selected along with two
japonica parents, IRGC8146 and IRGC15046, and the 164 hybrids showed average spikelet fertility of 70.5%, implying that the WCVs identified in the study are effective and useable in breeding. Hybrids also showed significant variation for other agronomic traits, as well as varying stability levels. Hybrids and checks showed differential behaviour with changes in the environment. Therefore, the hybrids were evaluated across three locations and it was found that the majority of the hybrids had excellent spikelet fertility (%), with an average of 70.53%. Dular and Nagina 22 have been extensively utilized in inter sub-specific hybridization by Vijayakumar et al. [
22]. Here, we also aimed to identify superior wide compatible varieties with higher spikelet fertility than the existing ones. We have identified a few combinations, namely WCH 41, WCH 73, WCH 120, WCH 36 and WCH 80, which showed higher spikelet fertility (%) than hybrids derived from Dular and Nagina 22. Our results justify the selection of parental lines for the development of intersubspecific hybrids. The average mean value was higher for spikelet fertility when
indica testers were crossed with
indica lines carrying neutral alleles, whereas in the case of the
japonica tester, higher spikelet fertility was observed when crossed with
japonica lines carrying the neutral allele. Additionally, average spikelet fertility was higher (72.3%) when
japonica lines were used as the female parent. In our study, we used all wide compatible varieties as male parents and found higher expression of spikelet fertility. Vijayakumar et al. [
22] observed higher expression of traits in hybrids when the WC gene in a male parent was used.
To study the variation in spikelet fertility between S5-i/S5-n and S5-j/S5-n, F1 hybrids were classified into two groups. In this study, 82 hybrids were derived from crosses between 41 wide compatible varieties with two indica lines, and another 82 hybrids were derived from crosses between 41 wide compatible varieties with two japonica lines, and it was observed that there were no significant differences between the groups with respect to spikelet fertility (%). Therefore, the potentiality of identified WCVs is assumed and the utility of a PCR-based functional marker (InDel S5) is shown for quick and precise identification of wide compatible varieties. These identified superior wide compatible varieties can further be used as a novel source of WC genes and will be utilized for marker-assisted incorporation of neutral alleles into elite rice varieties and the development of inter sub-specific hybrids.