Introduction

Rice is a staple food crop for half of the world population, contributing to nearly 20% of the total calorie intake of humans1. As the global population is predicted to reach almost 10 billion by 20502, the development of novel high-yield and superior-quality rice cultivars is urgently required to meet the future global food demand3.

Rice is one of the most extensively studied crop species because of its social and economic importance as well as its environmental impact4. Many genetic factors related to grain production such as starch biosynthesis and abiotic stress resistance, have captured the interest of researchers5, including the genes betaine aldehyde dehydrogenase (BADH2, Os08g0424500), granule-bound starch synthase 1 (GBSS1, Os06g0133000), granule-bound starch synthase 2 (GBSS2, Os07g0412100), and HPPD (4-hydroxyphenylpyruvate dioxygenase) inhibitor sensitive 1 (HIS1, Os02g0280700). BADH2 encodes for the enzyme that biosynthesizes 2-acetyl-1-pyrroline, which is a potent rice flavor compound6. Fragrant type of rice cultivars have been identified to be determined by allelic variations of the BADH2 gene6. GBSS1 and GBSS2 are responsible for the amylose content of rice by converting ADP-glucose into amylose instead of amylopectin in the starch biosynthesis pathway7. The amylose to amylopectin ratio is the most important physiological indicator of rice grain quality, particularly regarding the cooking and eating qualities8,9. HIS1 imparts resistance to bTH benzobicyclon and other b-Triketone herbicides that are widely applied to weed control in rice paddy fields10. Because weed control of large-scale farming is highly dependent on herbicides, the wide-spectrum tolerance to multiple herbicides are essential traits for increased rice production with reduced labor.

Oryza sativa and O. glaberrima were independently domesticated in Asia and Africa, respectively, from different wild ancestors1. In general, wild species obtain various alleles for resistance or tolerance to environmental stresses as adaption to a broad biogeographical diversity11,12. This rich diversity of wild relatives allows for introgression breeding to be a prominent approach in rice to enhance agricultural traits such as resistance to biotic and abiotic stresses13,14,15. Several previous studies have reported that introgression of alleles from wild germplasms enhanced the productivity and grain quality in soybeans and tomatoes16,17. However, in rice, most introgression breeding strategies have focused on conferring pest and disease resistance to O. sativa species13,14,15.

The genus Oryza comprises of two cultivated species and 22 wild species with 11 representative genome types: six diploids (AA, BB, CC, EE, FF, and GG) and five polyploids (BBCC, CCDD, HHJJ, HHKK, and KKLL)18. Oryza sativa and Oryza glaberrima, cultivated species, belong to the AA group with six wild species (Oryza nivara, Oryza rufipogon, Oryza barthii, Oryza glumaepatula, Oryza longistaminata, and Oryza meridionalis). Wild Oryza species have been considered as potential genetic resources that carry valuable alleles which are not present in the cultivated species19. Regarding abiotic and biotic stress resistance, many rice breeders have characterized favorable alleles in wild Oryza species11,12,20, and numerous alleles conferring stress resistance have been introduced from wild Oryza species into elite cultivars18,21, as demonstrated by the chromosomal introgression of iron resistance from O. meridionalis into O. sativa22. These results propose that the stability and productivity of elite rice cultivars could be improved by gene transfer from wild germplasms. However, the genes related to grain quality and herbicide resistance in rice, such as BADH2, GBSS1, GBSS2, and HIS1, have only been investigated within the O. sativa group.

This study aimed to investigate the evolution and divergence of the BADH2, GBSS1, GBSS2, and HIS1 gene families within the Oryza genus, including both cultivated and wild species. To provide a robust foundation for the evolution analysis of gene families, a phylogenetic tree was constructed among 10 Oryza species and macrosynteny was analyzed in four Asian Oryza species, including O. sativa ssp. japonica and indica, O. nivara, and O. rufipogon. Through a comprehensive phylogenetic and syntenic network approach, the evolutionary events and selection pressures were explored. The findings in this study contribute to a better understanding of the evolution of these target gene families and provide valuable insights for breeders in the selection of beneficial germplasm to develop more adaptive and productive rice cultivars.

Results

Phylogenetic analysis

We constructed two phylogenetic trees using peptide and nucleotide sequences, respectively, using 50 true orthologs from 10 Oryza species with A. thaliana and G. max as outgroups (Fig. 1). There were no significant differences in topology between two phylogenetic trees (Fig. 1 and Fig. S2). The eight AA-genome Oryza species were clustered together, thereby separating from O. punctata (BB-genome) and O. brachyantha (FF-genome) (Fig. 1). Within the AA-genome clade, the African and Asian species were grouped separately and the Asian species were further separated into two groups, namely the cultivated (O. sativa ssp. japonica and O. sativa ssp. indica) and wild groups (O. rufipogon and O. nivara) (Fig. 1). Using the divergence time between O. brachyantha and the other Oryza species at 15 million years ago (mya) as the calibration point1, the divergence times for the branching points was calculated in the phylogenetic tree (Fig. 1). The estimated divergence time between the AA- and BB-genomes was 14.8 mya, 9.2 mya between the African and Asian clades, 3.6 mya between the Asian wilds, and 0.4 mya between the Asian cultivars (Fig. 1).

Figure 1
figure 1

Phylogeny of 10 Oryza species and two dicot plants. The phylogenetic tree was constructed using the amino acid sequences of 50 true orthologs among 10 Oryza species and two dicot species as outgroups. The number of each node is the bootstrap value between nodes which is obtained from 1000 bootstrap replications. Divergence times within the genus Oryza were estimated by assuming 15 million years ago for the divergence point between O. brachyantha and the other Oryza species. Mya million years ago.

Full size image

Synteny analysis

To further understand the evolutionary process during the domestication of cultivated rice, O. sativa ssp. indica and O. sativa ssp. japonica, we further investigated the macrosynteny among the four Asian species at the chromosomal-level (Fig. 2A). O. sativa species had a more conserved synteny with O. rufipogon than O. nivara (Fig. 2B). O. sativa ssp. japonica and O. rufipogon showed the most conserved synteny among the four species and its breakpoint was identified only once on chromosome 3 (Fig. 2). The conservation of synteny between O. sativa ssp. indica and O. sativa ssp. japonica was slightly lower than that of synteny between O. sativa ssp. indica and O. rufipogon (Fig. 2B).

Figure 2
figure 2

Synteny analysis between Asian rice groups. (A) Macrosynteny visualization of four Asian species ordered by O. indica, O. japonica, O. rufipogon, and O. nivara. (B) The distribution of the synteny block size of six species pairs show the degree of syntenic conservation. The x-axis indicates the number of genes required to call a synteny block and the y-axis indicates the number of synteny blocks between each species pair.

Full size image

Identification of the BADH2, GBSS1, GBSS2, and HIS1 gene families

Orthologous genes of BADH2, GBSS1, GBSS2, and HIS1 were identified in the 10 Oryza species using clustering analysis based on the annotation information of the O. sativa ssp. japonica reference genomic sequence23. A total of 8, 10, 8, and 43 orthologous genes were identified as BADH2, GBSS1, GBSS2, and HIS1 gene family members, respectively (Table 1). For BADH2, GBSS1, and GBSS2, in general, single-copy genes remained across the Oryza species (Table 1). BADH2 was lost in O. meridionalis and O. punctata and GBSS2 was lost in O. brachyantha and O. meridionalis (Table 1). While GBSS1 was uniformly identified in every Oryza species without any loss events (Table 1). The HIS1 family showed high variation in their copy number ranging from two (O. brachatia and O. glumaetuala) to seven (O. sativa ssp. japonica and O. puctata) (Table 1).

Table 1 Number of orthologs (HIS1, BADH2, GBSS2, and GBSS2) in the 10 Oryza species.
Full size table

Gene synteny analysis of HIS1

The phylogenetic tree was constructed using 43 HIS1 orthologous genes and was further clustered into five subclasses consisting of HIS1 and HSL 1–4 based on the annotation data of O. sativa ssp. japonica (Fig. 3). In each clade, 9, 4, 5, 9, and 16 genes were grouped as HIS1, HSL1, HSL2, HSL3, and HLS4, respectively (Fig. 3). HIS1 and HSL genes are tandemly located on chromosomes 2, 3, and 6 (Fig. 4A). HIS1 and HSL3 genes are located on chromosome 2 in all Oryza species except for O. brachyantha (Fig. 4A). HSL1, HSL2, and HSL4 are mostly located on chromosome 6 in Oryza species and additional two HSL4 are located in chromosome 3 in O. punctata (Fig. 4A). In O. sativa ssp. japonica, one copy of HSL1 and HSL4 were identified in chromosome 6, and the HSL1 gene was only detected in O. glaberrima, O. barthii, and O. sativa ssp. japonica (Fig. 4A). The synteny analysis showed that all HIS1 and HSL3 orthologs had syntenic relationships in a pair-wise manner, while the synteny of HSL1, HSL2, and HSL4 were not maintained among most of the Oryza species (Fig. 4B).

Figure 3
figure 3

Phylogenetic tree of the HIS1 and HSL gene families. The phylogenetic neighbor-joining tree was constructed using the amino acid sequences of HIS1 and HSL orthologs in 10 Oryza species. In the HSL family, four subclasses (HSL1, HSL2, HSL3, and HSL4) were identified. Gene classes are indicated by the label colors, where 9 HIS1 were labeled in yellow, 4 HSL1 in orange, 5 HSL2 in red, 9 HSL3 in blue, and 16 HSL4 in green.

Full size image
Figure 4
figure 4

Chromosomal distribution and syntenic network of HIS1 and HSL gene families. (A) The location of HIS1 and HSL gene families at rice chromosomes 2, 3, and 6. Rectangles represented genes and their colors indicate the gene classes. The dotted and solid lines indicate the gene duplication and transfer events, respectively. (B) Syntenic network of HIS1 and HSL3 (upper) and HSL1, HSL2, and HSL4 (lower). The lines represent the syntenic connection between the two genes and the circle color indicates the subclass of the gene family.

Full size image

Gene synteny analysis of GBSS1, GBSS2, and BADH2

Among the four Asian species, synteny analysis was conducted for GBSS1, GBSS2, and BADH2, which had a low copy number variation (Table 1). Among all the Oryza species with orthologous genes, including orthologs that were identified based on their sequence similarity, the synteny was highly conserved between the target genes (Fig. 5A, Table S1). The synteny blocks harboring GBSS1, GBSS2, and BADH2, were identified on chromosomes 6, 7, and 8, respectively (Fig. 5B). In the synteny blocks of GBSS1, there are 671 genes when comparing O. sativa ssp. indica with O. sativa ssp. japonica, 1766 genes when comparing O. sativa ssp. japonica with O. rufipogon, and 133 genes when comparing O. rufipogon with O. nivara (Fig. 5B). The GBSS2 block contained 54, 1634, and 235 gene numbers. The block of BADH2 had 758, 1493, and 184 gene numbers (Fig. 5B).

Figure 5
figure 5

Syntenic conservation of GBSS1, GBSS2, and BADH2. (A) Syntenic network of the three genes across the Oryza species. The circles represent the orthologous genes detected by sequence similarity, dashed circles indicate the absence of the gene, and the lines represent a syntenic connection between the two genes. The line colors indicate whether the two genes are connected, and gray or red indicates the presence or absence thereof. (B) The feature of synteny blocks harboring the three genes among the Asian species including O. sativa ssp. indica, O. sativa ssp. japonica, O. rufipogon, and O. nivara. The red lines show the location and syntenic linkage of the orthologs. The numbers beside each synteny represent the number of genes in the synteny block of the focal genes.

Full size image

Selection pressure

The Ka/Ks ratio of the homologs was calculated to determine whether BADH2, GBSS1, GBSS2, and HIS1 underwent negative or positive selection (Fig. 6). The mean Ka/Ks values for HIS1, GBSS1, GBSS2, and BADH2 were 0.92, 0.1, 0.69, and 0.16, respectively, indicating that these genes evolved under purifying selection (Fig. 6). Moreover, the mean Ka/Ks value of the GBSS1 and BADH2 gene pair were lower than those of the means other two families, which suggests that the GBSS1 and BADH2 duplicates evolved at a slower rate (Fig. 6). Additionally, we calculated the Ka/Ks values by comparing target orthologs from two monocot plants, Sorghum bicolor (S. bicolor) and Lessia perrie (L. perrie), with those in Oryza species. The comparisons between S. bicolor and Oryza species showed Ka/Ks values of 0.2, 0.13, and 0.14 for GBSS1, GBSS2, and BADH2, respectively (Table S2). When comparing L. perrie with Oryza species, the Ka/Ks values were 0.12 and 0.11 for GBSS1 and BADH2, respectively (Table S2). These results indicate that the target genes had been undergone purifying selection within cereal plants.

Figure 6
figure 6

Comparison of the pair-wise Ka/Ks values of HIS1, GBSS1, GBSS2, and BADH2 genes. The Ka/Ks values of the ortholog pairs were calculated in each of the focal genes to determine the selection pressure. The ortholog genes were obtained based on the synteny block among the 10 Oryza species.

Full size image

Discussion

The evolution and speciation of Asian rice species have remained unclear due to their high ortholog sequence similarity1,24. To date, there are two hypotheses concerning the origin and domestication of the O. sativa species24. The single-domestication hypothesis posits that the O. sativa species originated from a wild ancestor and the differentiation between O. sativa ssp. indica and O. sativa ssp. japonica occurred after domestication of the cultivated species25,26,27,28. This single-domestication hypothesis is mainly supported by the molecular evidence of the identical sequences of the key domestication genes between the O. sativa subspecies, including sh4 that reduces shattering and prog1 which is associated with erect growth25,26,27. In contrast, the multiple-domestication hypothesis postulates that O. sativa ssp. indica and O. sativa ssp. japonica were domesticated separately from different wild ancestors29,30. This multiple-domestication hypothesis has gained support through the phylogenetic analyses which shows that the O. sativa subspecies are separated into distinct clades and are closer to the different wild accessions than each other29,30. The phylogenetic tree that was constructed in this study using the true orthologs based on syntenic relationship was consistent with the single-domestication hypothesis as the O. sativa subspecies were grouped into the same clade (Fig. 1). The synteny analysis revealed that the genomic structure of the O. sativa species was more conserved with O. rufipogon than those of O. nivara (Fig. 2B). This result is in agreement with a previous study which reported that the O. sativa species may originate from O. rufipogon and that O. nivara is one of the ecological varieties of O. rufipogon25,26,27.

Genomic similarity is a key factor that determines genetic compatibility which enables the transfer of desired traits between two species through interspecific crossing31. In rice, several reproductive isolations had been observed in interspecific hybrids, which resulted in inviability, weakness, and sterility32. Several genetic models have been suggested to explain the mechanisms of reproductive isolations in plants and structural variations were identified as one of leading factors of reproductive isolation32. In rice and Arabidopsis, pollen incompatibility and inviability had been reported in their hybrids, respectively, due to a change of the gene locus caused by reciprocal gene loss of duplicated genes20,33. Other studies have reported that chromosomal rearrangements enhance the reproductive isolation by suppressing recombination31,34,35, which results in unbalanced gametes that may be inviable31,34.

Suppressed recombination also increased the extent of linkage disequilibrium (LD) block, thereby restricting gene flow in potentially larger genomic regions34,36,37. When introgression of a favorable gene from an external germplasm into a cultivar occurs, other genes that confer undesirable traits can also be transferred if the genes are located in the same LD block38,39. This phenomenon is also known as the linkage drag problem and it is a major concern in introgression breeding which prevents breeders from introducing desirable traits into elite cultivars40,41,42. Linkage drag that leads to the negative relationships among the yield potential, grain quality, and environmental resistance have been reported in O. sativa species40,41. The synteny analysis in this study identified that O. rufipogon would be a more genetically compatible germplasm for O. sativa breeding with reduced reproductive isolation and linkage drag problems (Fig. 2).

Gene evolution analysis has been widely used to investigate gene expansion, domestication process, genetic background, etc. Copy number variation (CNV) is a structural variation that alters the dosage of genes, which could result in phenotypic changes43,44. In plants, most resistance traits are polygenic and highly affected by CNV43,44. A previous study on durum wheat reported that frost resistance levels were determined by the CNV of the CBF-A14 gene family45. In rice, the CNV of 28 functional genes was identified to be involved in insect resistance and response to salt stress43. In Brassica napus, 563 resistance genes experienced 1137 CNV events including 704 deletions and 433 duplications46. Based on the phylogenetic clustering, a total of 43 HIS1 genes were further clustered into five subclasses including HIS1 (9), HSL1 (4), HSL2 (5), HSL3 (9), and HLS4 (16) (Fig. 3). We identified that the HIS1 and HSL families experienced multiple duplication events (Fig. 4). In O. punctata, a tandemly duplicated HSL4 gene was identified on chromosome 6 and an additional pair of tandemly duplicated HSL4 genes were detected on chromosome 3 (Fig. 4). Because these HSL4 genes of O. punctata were not found in other Oryza species, they might be duplicated after the O. punctata speciation event (Fig. 4). The HSL1 genes were only identified in O. glaberrima, O. barthii, and O. sativa ssp. japonica (Fig. 4). Considering that O. glaberrima was domesticated from O. barthii, the HSL1 gene probably originated from O. barthii, and then moved to O. satvia ssp. japonica via O. glaberrima or directly from O. barthii (Fig. 4). In O. sativa ssp. japonica, duplication events of HSL1 and HSL4 were identified on chromosome 6 (Fig. 4). Because four of the five duplicated HSL1 and HSL4 genes are located close to each other (less than 50 kb interval), we propose that the HIS1 and HSL families were mainly expanded through tandem duplication events in the Oryza species (Fig. 4). These results are in agreement with a previous study where tandem duplication events were frequently identified in CNVs47. Our gene evolution analysis can facilitate the improvement in herbicide resistance of rice cultivars through gene transferring from wild germplasm that has a high copy number of HIS1 and HSL genes such as O. punctata (Fig. 4).

While HIS1 has diverse CNVs across 10 Oryza species, GBSS1, GBSS2, and BADH2 genes have at most one copy in the 10 Oryza species and the syntenic relationship of their orthologs was deeply conserved in pair-wise comparison among Oryza species (Table 1 and Fig. 5A). These results suggest that these three genes descended from a common Oryza ancestor to present-day cultivars (Fig. 5). The BADH2 and GBSS2 loss events were identified in O. brachyantha, O. punctata, and O. meridionalis, which suggests that the functional role of the BADH2 and GBSS2 genes were developed after speciation from O. meridionalis (Table 1). Meanwhile, no loss event of GBSS1 was identified in the Oryza species (Table 1). Plants have some critical genes that play essential roles in their survival such as photosynthesis, cell division, and reproduction48,49,50. In general, the genetic diversity of essential genes is highly conserved among related species, because the malfunction of these genes directly affects their fitness in the population48. In rice, a previous study reported that GBSS1 is expressed in the endosperm and pollen grains, while the expression of GBSS2 is limited to the vegetative tissues and pericarp7. The starch content of the endosperm serves as the primary source for seed germination and seedling growth7. Therefore, the prevalence of GBSS1 may be the product of selection pressure for its critical role in seed germination vigor, as wild relatives and landraces with lower germination rates were extinguished in nature or removed from the breeding pool during rice evolution and domestication. This result is consistent with our selection pressure analysis which showed that GBSS1 had the lowest Ka/Ks ratio indicating that the sequence diversity of GBSS1 is highly conserved during evolution (Fig. 6).

Gene exchange is a key evolutionary mechanism that enhances the adaptability of the population against environment stresses51. Natural introgression between the Asian cultivated and wild species is common because they are often sympatric24,51,52. The African cultivated species have a relatively limited gene pool in the wild species compared to Asian rice and several historic introgression events from the Asian species are reported24. GBSS1, GBSS2, and BADH2 genes are located in highly conserved synteny blocks as single-copy genes over the 10 Orzya species, which indicates they are true orthologs in the genus Orzya. Using the true orthologs of GBSS1, GBSS2, and BADH2, reconciled gene trees were constructed and estimated divergence time was calculated between orthologous gene pairs, to investigate gene transfer events of the target genes across the Oryza species (Fig. S1 and Table 2). Our results proposed that three transfer events had occurred in GBSS1 from the Asian groups into the African group (O. glaberrima and O. barthii) and other wild species (O. glumaepatula and O. meridionalis) (Fig. 7). In contrast, the GBSS2 was transferred from O. barthii into the Asian species including O. sativa ssp. japonica and O. nivara (Fig. 7). This gene flow in the opposite direction between Asian and African groups indicates that GBSS1 and GBSS2 gained their subfunctions independently in the Asian and African rice populations, respectively, and they were spread to other regions during the domestication process. For BADH2, a gene from either O. barthii or O. glaberrima was moved into O. rufipogon, which suggests that most BADH2 alleles in japonica rice had been developed after divergence between the Asian and African groups (Fig. 7)6,53.

Table 2 The estimated divergence time for pairs of transferred orthologs.
Full size table
Figure 7
figure 7

Schematic illustration of phylogenetic inference to detect gene transfer events. The species tree is depicted by the thick blue area and the inner black lines represent the phylogeny of each gene. The dotted line indicates the absence of the gene in the node. The red x symbol indicates the loss of the gene in the node. The translucent oval between two species represent that either a species or another species contributes to transfer event. The arrows indicate the gene transfer between two species or node.

Full size image

Overall, this study enhances our knowledge of the gene family evolution in rice and offers practical implications for rice breeding efforts, which ultimately supports the development of improved rice varieties with enhanced adaptability and productivity.

Material and methods

Data source

The peptide sequences, nucleotide sequences, and gene location information (GFF format) of three domesticated species (O. satvia ssp. japonica, O. sativa ssp. indica, and O. glaberrima), seven wild species (O. rufipogon, O. nivara, O. barthii, O. brachyantha, O. glumaepatula, O. meridionalis, and O. punctata), and two dicot plants (A. thaliana and G. max) were downloaded from EnsemblPlants (https://plants.ensembl.org, accessed on 17 April 2023) and Rice Genome Hub (https://rice-genome-hub.southgreen.fr, accessed on 17 April 2023) (Table S3). The protein and nucleotide sequences were filtered out and the longest isoform per gene was retained for downstream analysis.

Species phylogenetic tree

Synteny analysis was conducted to obtain true orthologs across the 12 species (see section “Synteny detection” for detail). In our study, single-copy orthologous genes located in synteny blocks among 10 Oryza species were defined as true orthologous gene. A total of 50 syntenic orthologs, which are shared between O. sativa ssp. japonica and the 11 other species, were selected as true orthologs for phylogenetic analysis. Peptide and nucleotide sequences of the 50 true orthologs were aligned using ClustalOmega v1.2.454 and then concatenated. Phylogenetic trees were built using RAxML v8.2.12 with 1000 bootstrap replicates55. Two maximum-likely hood models, JJT and GTRGAMMA, were used for phylogenetic tree of peptide and nucleotide sequences, respectively55. The phylogenetic tree was visualized using Interactive Tree Of Life56. The divergence time between the nodes was estimated by the MCMCtree package embedded in PAML v4.10.657.

Synteny detection

The macrosynteny blocks among the 12 species were identified using BLASTP version 2.9.0+ and software MCScanX58,59. The protein sequences of the homolog pairs were obtained by all-against-all BLASTP search with the standard parameter setting for MCScanX analysis (evalue: 1e−10, num alignments: 5, and outfmt: 6). The BLASTP outputs with the gene location data were imported into the MCScanX to identify the synteny blocks with the default parameters (match score: 50, gap penalty: − 1, match size: 5, evalue: 1e−5, and max gaps: 25)58. The synteny was visualized using SynVisio56.

Identification of BAHD2, GBSS1, GBSS2, and HIS1 orthologs

The protein sequences of the 12 species were clustered into orthologous groups using OrthoFinder v2.5.4 with sequence similarity searches conducted using DIAMOND60. Based on the gene annotation of O. sativa spp. japonica, we obtained the names of the four target genes: BADH2 (Os08g0424500), GBSS1 (Os06g0133000), GBSS2 (Os07g0412100), and HIS1 (Os02g0280700). The four orthologous group containing the focal genes of O. satvia ssp. japonica were selected as orthologs corresponding to each gene. To classify the HIS1 orthologs into subclasses, their protein sequences were aligned and used to construct a neighbor-joining phylogenetic tree using MEGAX software with 1000 bootstrap replications61.

Gene evolutionary analysis

The “add_ka_and_ks_to_collinearity.pl” script in MCScanX was used to determine the non-synonymous (Ka) and synonymous (Ks) substitution values of the homologs pairs58. The selection pressures were determined based on the Ka/Ks values of syntenic orthologs pairs in each focal orthologous group. The gene synteny of the focal orthologous groups was identified based on the macrosynteny analysis using an in-house developed python script. The functional annotation of genes in synteny block was conducted using grameen database (https://www.gramene.org/, accessed on 7 February 2024). To construct gene tree, protein sequences of true orthologs of GBSS1, GBSS2, and BADH2 were aligned using ClustalOmega v1.2.4 with the default parameters54. Gene trees were constructed using RAxML v8.2.12 with 1000 bootstrap replicates55 and the trees were reconciled into the species tree to identify the transfer event using Notung v2.6.1.5 (Fig. S1)62. Divergence time (T) between orthologous pair was calculated based on a synonymous substitutions per year (λ) as T = Ks/2 λ (λ = 6.5 × 10−9 for rice)63.