Figure 8. Divergence time estimated from the MCC tree in BEAST. Node numbers indicate posterior probabilities (below) and mean divergence times (above). Node bars represent the 95% HPD interval (blue bar). Background colors of gray, green, blue, and orange indicate outgroups, autotrophs, hemiparasites, and holoparasites, respectively. The stars (★) indicate the two newly sequenced Cymbaria species.
4. | Discussion
4.1. | Pseudogenization/loss events of ndh genes and the unique rbc L-mat K inversion
It is acknowledged that the lifestyle transition from autotrophy to heterotrophy triggers the degradation of chloroplast genomes (Wicke & Naumann, 2018). Contrasting with the hypervariability of holoparasites,Cymbaria species and other Orobanchaceae hemiparasites exhibit high similarity to autotrophs in length, GC content, and intact genes. It has been confirmed that holoparasites have the characteristic of chloroplast genome reduction (Wicke et al., 2013). The high variability of holoparasites is explained by increases in pseudogenization and gene loss. However, patterns of variation in hemiparasites are diverse. The chloroplast genomes of hemiparasites in the family Orobanchaceae were more similar to those of autotrophs, which is in contrast to the reductions in the genome sizes of hemiparasites in the order Santalales (X. Guo, Zhang, Fan, Liu, & Ji, 2021; Y. Li et al., 2017; Shin & Lee, 2018). This might be attributed in part to GC-biased gene conversion and mutational biases, which suggests that sophisticated mechanisms contribute to the stability (Niu et al., 2017).
Angiosperms typically possess 113 plastid genes, consisting of 79 functional PCGs, 30 tRNA, and four rRNA genes (Wicke et al., 2011). Within the Orobanchaceae family, pseudogenes and gene losses were largely absent in autotrophs, occasionally observed in most hemiparasites, and common in nearly all holoparasites. This can be explained by the tendency for the chloroplast genomes of parasites to be reduced in size (Naumann et al., 2016; Wicke & Naumann, 2018). The chloroplast NAD(P)H-dehydrogenase complex comprises 11 ndh genes (Ma, Liu, Bai, & Yong, 2021), and the pseudogenization or loss of these genes represents the initial stage of reductive evolution (Wicke & Naumann, 2018). The results of our study indicated that C. mongolica lost genes (ndh I, ndh J) and that C. daurica contained pseudogenes (ndh F,ndh H) and lost genes (ndh A, ndh C, ndh E,ndh G, ndh I), indicating that these two species are in the initial phase of the autotroph-to-heterotroph transition. A previous study has confirmed that a hemiparasitic lifestyle can lead to an increase in the pseudogenization/loss of ndh genes (Xin Li et al., 2021). This can be explained to some extent by the facultative root hemiparasitic lifestyle of the two Cymbaria species. The degradation of ndh genes affects several morphological and physiological traits and enhances the adaptation of plants to environmental stress (Sabater, 2021).
The unique inversion including rbc L-mat K in the LSC region, which most likely stems from a palindromic repeat-mediated rearrangement. Inversions of the LSC fragments have also been observed in Schwalbea americana (Wicke et al., 2013) andSiphonostegia chinensis(Jiang et al., 2022); this distinct evolutionary mechanism among Orobanchaceae members might explain the unique phylogenetic position of the tribe Cymbarieae. This inversion has also been observed inCodonopsis pilosula subsp. tangshen (Yue et al., 2022) andAvena sativa (Q. Liu et al., 2020).
The codon usage bias of the two Cymbaria species might be affected by natural selection and mutation, as has been observed in several other angiosperms (Q.-F. Lu, Luo, & Huang, 2020). The number of repeats plays a key role in maintaining the stability of in several angiosperms chloroplast genome (Jansen, Saski, Lee, Hansen, & Daniell, 2010). The chloroplast genome of C. daurica has more repetitive sequences than that of C. mongolica , suggesting that the stability of the former might be higher than that of the latter. Our findings suggest that A/T mononucleotide SSRs were dominant, and this is consistent with the high prevalence of AT richness (Xia Liu, Li, Yang, & Zhou, 2018).
4.2. | Specific DNA barcodes for distinguishing herb C. daurica from its adulterant C. mongolica .
Traditional Mongolian medicine continues to receive much clinical attention because of its distinctive properties and herb resources (Z. F. Liu et al., 2022). C. daurica  has been used to treat pruritus, psoriasis, fetotoxicity, impetigo, and diabetes. The incidence ofC. daurica adulterated with its sister species C. mongolica is increasing, and this poses a threat to the clinical efficacy of the herb. The high similarity in morphology between C. daurica and C. mongolica is the root cause of this problem. Distinguishing between these two Cymbaria species is exceedingly difficult because they only differ in anther morphology (Zhang et al., 2013).
Both morphological and microscopic characteristics are currently used to identify Cymbaria species (Z.-W. Wang, Domg, Wu, & Li, 2012). However, both of these methods rely on the features of the anther locule, which are only visible during the short flowering period, and specialists are required to obtain accurate identifications. Several divergence hotspot regions were first identified through sequence divergence and nucleotide variability. We then developed and validated four pairs of specific DNA barcodes to distinguish between these two species. Each DNA barcode had at least one Indel locus and several SNP loci. Some previous studies have suggested that specific barcodes are superior to universal barcodes for identifying morphologically similar species (Fang, Dai, Liao, Zhou, & Liu, 2022; G. Y. Lu et al., 2022). Overall, these four pairs of specific DNA barcodes could be used to accurately and rapidly distinguish between C. daurica andC. mongolica without the need to evaluate the morphological characteristics of the anther during flowering nor specialized training.
4.3. | Climate aridification and increasing host accelerate the diversification of Cymbaria
Traditional Orobanchaceae has been merged with all hemiparasitic genera as well as a few holoparasitic genera formerly placed in Scrophulariaceae (Bennett & Mathews, 2006; dePamphilis, Young, & Wolfe, 1997; Fischer, 2004; McNeal et al., 2013; Wolfe, Randle, Liu, & Steiner, 2005; Young, Steiner, & dePamphilis, 1999) and the three autotrophic genera previously considered as sister taxa to Orobanchaceae (Schneeweiss, 2013; Xia, Li, Wen, & Wang, 2021). Moreover, the two new tribes Brandisieae and Pterygielleae have been proposed (Jiang et al., 2022; Yu et al., 2018). To date, Orobanchaceae includes nine well-supported clades corresponding to nine tribes, i.e., the two autotrophic tribes Rehmannieae and Lindenbergieae, and the seven parasitic tribes Cymbarieae, Buchnereae, Orobancheae, Brandisieae, Pedicularideae, Rhinantheae, and Pterygielleae. The topology of the major clades and autotroph–parasite sister relationships revealed by our phylogenetic analyses (Fig. 5) were generally consistent with previous findings.
The hemiparasitic tribe Cymbarieae is distinguished from other tribes in the family Orobanchaceae by the presence of bracteoles, a tubular calyx that is weakly dorsiventral, a highly two-lipped corolla, and anthers with two mostly rounded and equal thecae (Fischer, 2004). Cymbarieae has traditionally been considered sister to all other parasitic lineages (Bennett & Mathews, 2006; McNeal et al., 2013). However, a recent study has challenged this classification after finding that the holoparasitic tribe Orobancheae was sister to all other parasitic members (Xi Li et al., 2019). Our findings are consistent with the traditional classification of the Cymbarieae. The Cymbarieae was found to be sister to the remaining parasitic lineages, and the hemiparasites evolved earlier than the holoparasites; this is consistent with the progressive nature of the evolution of increased host dependence (Xu et al., 2022). This inconsistency in the placement of Cymbarieae might stem from phylogenetic sources of error, such as incomplete lineage sorting or deep coalescence. Resolving these early nodes will require a coalescent approach that involves many genes with different histories.
Within parasitic plants, the hemiparasitic tribe Cymbarieae diverged from the remaining lineages in the mid-Oligocene (31.44 Mya), which was most likely induced by global climate cooling and the retreat of the Tethys Sea during the Eocene–Oligocene Transition at 34 Mya (Abels, Dupont-Nivet, Xiao, Bosboom, & Krijgsman, 2011). The emergence and divergence of hemiparasites might be attributed to the rapid expansion of grasslands during the Oligocene (Torsvik & Cocks, 2016), which would have provided them with opportunities to exploit host plants. Cymbaria species diversified in the late Miocene (6.72 Mya), which was driven by the final uplift of the Qinghai–Tibetan Plateau, the onset of the East Asian monsoon, and the large accumulation of dust in the Loess Plateau from 10 to 7 Mya [77–79]. Both climate aridification and the increase in host steppe vegetation (Hurka et al., 2019) likely accelerated the adaptive evolution of Cymbaria species in the Mongol–Chinese steppe region.
5. | Conclusions
We characterized two Cymbaria chloroplast genome and conducted a comparative analysis using chloroplast genomes across 54 Orobanchaceae species. The chloroplast genomes of C. mongolica and C. daurica had a typical quadripartite structure, and their total lengths were 149,431 bp and 151,545 bp, respectively. Although the chloroplast genomes of holoparasites are hypervariable, the genome size, GC content, and intact genes of Cymbaria species, other hemiparasites, and autotrophs are highly similar. The pseudogenization/loss of ndhgenes might be associated with the facultative root hemiparasitic habits of Cymbaria . The rbc L-mat K inversion in the LSC region most likely stemmed from a palindromic repeat-mediated rearrangement. Specific DNA barcodes were developed using four pairs of primers (CymN1, CymN2, CymY, and CymR) that amplified sequences from the divergent hotspot regions to distinguish the traditional Mongolian herbC. daurica from its adulterant C. mongolica . The genusCymbaria and the Schwalbea -Siphonostegia clade were clustered in the tribe Cymbarieae. This tribe comprised an independent clade sister to the remaining parasitic lineages, which is in contrast to the relationships hypothesized in a recent study. The monophyletic genus Cymbaria diversified during the late Miocene period (6.72 Mya). The aridification of the climate and increases in host steppe vegetation likely promoted the adaptive evolution of Cymbariaspecies in the Mongol-Chinese steppe region. Our results provide key information for clarifying the taxonomic identification, phylogenetic placement, and reductive evolution of Cymbaria ; our findings will also help assessments of the authenticity of the traditional Mongolian medicine “Xinba.”