Table 1
Principle Component Analysis (PCA) revealed three distinct clusters with a slight overlap of individuals from different groups (Fig. 2). The wild species, O. rufipogon formed a separate cluster (Cluster I) scattered along with the PCA plot, indicating higher levels of genetic diversity. Oryza nivara cluster (Cluster II) was well differentiated from other Oryza types, but a considerable portion of landraces, feral and weedy individuals overlapped. The large cluster (Cluster III) reflects the close genetic relationship among inbred cultivars, landraces, feral and weedy types. The feral populations were grouped with inbred rice varieties as they are currently undergoing the process of de-domestication (Ellstrand et al., 2010). A considerable number of feral individuals were grouped with the wildO. nivara , potentially a high level of reverse introgression from the wild species. Many inbred rice cultivars overlapped with the feral and landraces. In contrast, weedy rice displayed a diffused distribution in the PCA scatter plot, reflecting a close genetic relationship with inbred cultivars, feral rice, and O. nivara . Most of the weedy rice showed a close evolutionary relationship with the inbred and landraces and few displayed close genetic similarities to O. nivara .
Fig. 2. Scatter plot of the first and second principal components (PC) based on the variation of 33 SSR loci for 1340 individuals of 20 weedy rice populations, five O. rufipogonpopulations, six O. nivara populations, 42 inbred rice varieties, seven feral rice populations and 31 landraces from Sri Lanka showing the considerable overlap of genotypes in different Oryza types at 33 SSR loci.
The STRUCTURE analysis of DWWC demonstrated a distinct ∆K peak at K=2 levels (Fig. S1), indicating that wild Oryza was differentiated from other Oryza groups. Within the analysis, two populations of weedy rice (W4 and W6) and three populations of feral rice (F1, F2, and F5) were classified as out-groups (Fig. 3) in relation to their respective Oryza groups, leading to the formation of two separate groups for both weedy and feral rice. This observation can likely be attributed to the adaptation of O. nivara to the rice ecosystems, a process driven by persistent habitat disturbances, continued selective pressures, or hybridization events between cultivated (inbred and landraces) plants and their reproductively compatible wild counterparts (Cao et al ., 2006). Consequently, K=2 was determined to be the most biologically realistic population number. Further, ∆K peaks at K=9 and K=5 (Fig. S1) were considered for further examination of population models. The K=9 model is highly intricate, suggesting that the physical classification of Oryza groups presents a challenge due to the genetic relationships among populations and individuals based on prevailing patterns of genetic admixture. However, the K=5 model is generally consistent with the Principal Component Analysis (PCA) results (Figs. 3 and 4).
The two wild Oryza species, O. nivara and O. rufipogon , differentiated from others. Moreover, these two wild species exhibited genetic differences from one another, i.e. some populations showing admixture, while most individuals can be reliably assigned to a specific population. The O. rufipogon group comprised a most distinct group with a more heterogeneous genetic background. Two weedy and three feral populations were highly admixed with the wild O. nivara . Widely cultivated types, inbred rice varieties, and landraces shared a common ancestry and were evident as largely close groups to weedy rice with the admixture nature. However, some individuals of landraces were admixed with the wild Oryza and weedy rice. The evolution of the weedy type reflects the complex process of genetic incorporation from the crop (inbred or landraces), non-cultivated wildOryza . This illustrates the multi-way genetic transfer to the evolution of weedy types. The STRUCTURE results suggest that there is a complex integration of multi-way gene flow among all members of the DWWC in the rice ecosystem in Sri Lanka. However, when examining K values ranging from 5 to 10, it becomes clear that there is an admixed genetic background for individuals in some populations/cultivars. While most individuals could be assigned to a single population or cultivar, the occurrence of further groupings is uncommon (Fig. 3).
Fig. 3. STRUCTURE graph showing genotype clustering of 20 weedy rice populations, five O. rufipogon populations, six O. nivara populations, 42 inbred rice varieties, seven feral rice population and 31 landraces by model‐based population assignment at K from 2 to 10. Each vertical bar represents an individual, with its assignment probability to genetic clusters represented by different colors. Codes for the weedy rice, O. rufipogon , O. nivaraand feral rice populations are presented in Table S1.
Fig. S1
The UPGMA tree shows similar results, with all Oryza types genetically structured into two well-separated major groups (O. rufipogon and all other Oryza types) and further divided into respective populations and cultivars (Fig. 4). The O. rufipogongroup forms a separate cluster (Cluster I). Besides, the large cluster was further divided into two sub-clusters (Cluster II and Cluster III). Moreover, all O. nivara , three feral rice (F1, F2, and F5) and two weedy (W4 and W6) rice populations, and a few landraces formed a distinct cluster (Cluster III), as shown in the PCA. The large cluster (Cluster III) consisted of inbred varieties, landraces, unmanaged abandoned feral populations, and weedy rice. Most weedy rice populations were grouped with landraces, however, W2 and W3 populations displayed closer relationships with landraces. Furthermore, W1 was distinct from the weedy rice cluster, while the rest of the populations were subdivided into two clusters. Feral rice populations were grouped into a large cluster, but they were also clustered with the inbred rice group. Inbred rice varieties and landraces showed complex clustering patterns due to the relatively high genetic distances among their respective cultivars.
Fig. 4. The UPGMA cladogram is based on Nei (1972) genetic distance. Dendrogram (UPGMA) was constructed based on polymorphisms of 33 SSR loci in six Oryza types (20 weedy rice populations, fiveO. rufipogon populations, six O. nivara populations, 42 inbred rice varieties, seven feral rice populations, and 31 landraces), using Nei’s unbiased genetic distance (Nei, 1972). The bar represents genetic distance, with the same color sharing the same source of collection. Population or varieties/cultivars codes for the sixOryza types of populations are presented in Table S1.
The AMOVA analysis revealed a significant portion of the total genetic diversity present within populations (56%) and among populations (28%), while relatively low (16%) genetic differentiation was observed among Oryza types (Table S7; P <0.01). In specific Oryza types, the total variation was partitioned into among populations/cultivars and within populations/cultivars. A notably larger variation was observed within populations compared to among populations (Table S7).