The cpDNA-based (trnL-F, matK, rbcL, and trnH-psbA) phylogeny of the genus Kengyilia, especially with regard to the origin of maternal donor during hexaploid polyploidization events, were largely unresolved due to the occurrence of many polytomies and incongruence among published gene tree [14, 15]. Ma et al. [19] pointed out that despite missing samples, phylogenetic analysis of plastome sequences can offer the greatest phylogenetic resolution. In this study, a resolved tree with highly statistic support was inferred from the plastome sequences of Kengyilia and those of its relatives in Triticeae, allowing the relationship regarding to the maternal lineages of Kengyilia to be clarified.
In phylogenomic tree, ten species of Kengyilia (K. alatavica, K. hirsuta, K. laxiflora, K. batalinii, K. kokonorica, K. thoroldiana, K. grandiglumis, K. mutica, K. stenachyra, and K. rigidula), Roegneria, and Pseudoroegneria were in one group with consistent support, indicating that Pseudoroegneria is likely to be the maternal donor of these ten StYP genome Kengyilia species and the sampled StY genome Roegneria species. Since Kengyilia species arose from two hybridization events followed by three genome doublings (the St, Y and P genomes), with one firstly generating the StY genome Roegneria and the other forming the StYP genome Kengyilia [11, 13], Roegneria severed as the maternal donor during the speciation of the ten Kengyilia species.
Analysis of trnL-F suggested that four species of Kengyilia (K. kokonorica, K. melanthera, Kengyilia mutica, and Kengyilia thoroldiana) were closely related to species of Agropyron [14]. A similar deep-level relationships regarding to maternal lineages is also presented by Luo et al. [15], although molecular characters (including matK, rbcL, and trnH-psbA) and more taxa were sampled from Kengyilia. In this study, only K. melanthera was grouped with the species of Agropyron, and the remaining three species (K. kokonorica, K. mutica, and K. thoroldiana) were placed into the clade including St-containing species. Moreover, the plastome sequence of K. melanthera and Agropyron are obviously distinct from those of the St-containing species. Thus, the molecular phylogenies based on published cpDNA fragments and the present plastome sequence data in resolution of the placement of K. kokonorica, K. mutica, and K. thoroldiana led to apparently contradictory results. Discordances among phylogenetic trees result from methodological artifacts (e.g., sampling error and/or a failure of molecular characters) and the complex dynamics of the evolutionary processes in organisms (e.g., hybridization and/or ancestral polymorphisms) [6, 20]. Sampling error is likely to be the candidate for the current incongruences because our samples for the comparative phylogenies with Kengyilia species included nearly all of the monogenomic genera accepted in genome-based classifications of the Triticeae, and most monogenomic genera were not covered in previous study [14, 15]. It is well known that molecular characters can affect the accuracy of phylogenetic estimates [19]. Incongruences would also be the result of lack of molecular characters. Less molecular characters in cpDNA regions, as indicated by our estimate for the variable features of each chloroplast protein-coding genes (Table S2), together with its slowly evolving rates in chloroplast genome, would not only provide a few variable information for the accuracy of phylogenetic reconstruction but also result in the occurrence of polytomies in phylogenetic tree. On the contrary, the plastome data offer enough molecular characters for the accuracy of phylogenetic estimates with well-supported topology. Both hybridization and ancestral polymorphisms acting alone or in concert can generate discordance and therefore are the principal processes to explain the phylogenetic incongruence in Triticeae species [6, 21]. Analysis of genetic distance matrix based on the 52 protein-coding genes suggested that Lophopyrum and Thinopyrum are closely related to the St-containing species. In phylogenomic tree inferred from complete chloroplast genome, Lophopyrum, Thinopyrum, Dasypyrum, and two species of Pseudoroegneria (Pse. stipifolia and Pse. congnata) form a monophyletic group. These results indicated Lophopyrum, Thinopyrum, Dasypyrum, Pseudoroegneria (most likely Pse. stipifolia and Pse. congnata) shared ancestral polymorphisms due to incomplete diversification of common maternal ancestry. Such ancestral polymorphisms could be genetically transmitted to some polyploid species (e.g.: StP, StY, StYP) via the hybridization between Pseudoroegneria as female parent and the donors with Y and/or P genomes. The hypothesis of hybridization is also a likely candidate to explain the conflict because different polyploid species with the same genotypes could derive from different parental donors via independent hybridization events, generating a diverse array of polyploid genotypes in Triticeae [5, 22]. The present plastome data also provides support for the independent origin some polyploid species, which can be shown by different Kengyilia species that was grouped with different Roegneria species in a phylogenetic tree. For example, in the clade I of phylogenomic tree, three Kengyilia species (K. hirsuta, K. laxiflora, and K. batalinii) were clustered with R. grandis with strongly statistic support (100% UFboot, 100% SH-aLRT, and 1.0 PP), and five Kengyilia species (K. thoroldiana, K. grandiglumis, K. mutica, K. stenachyra, and K. rigidula) were grouped with R. longearistata (100% UFboot, 100% SH-aLRT, and 1.0 PP). Analysis of genetic distances based on 52 protein-coding sequences also presented similar results. Sympatric distribution among R. grandis, R. longearistata and Agropyron species have provide an opportunity in physical proximity for hybridization events. It is thus suggested that the different Kengyilia species derived their StY genome from different Roegneria species. Our data also indicated that Agropyron species severed as the maternal donor during the speciation of K. melanthera, providing additional support for the independent origin of different Kengyilia species. However, it seems unlikely that the maternal Agropyron lineage in K. melanthera resulted from hybridization between high ploidy Roegneria species with StY genomes (served as paternal donor) and diploid P genome Agropyron species. One possible explanation is that the P genome of K. melanthera originated from the tetraploid Agropyron lineage as the female parent. Given the present data, multiple origins of polyploid species result in a maternal haplotype polymorphism and could explain the rich diversity and wide adaptation of polyploid species in the genus Kengyilia [11].