Pyrrosia s.l. comprises ca. 60 species with a disjunct Africa/Asia and Australia distribution. The infrageneric classification of Pyrrosia s.l. is controversial based on the phylogenetic analyses of chloroplast markers and morphology. Based on the expanded taxon sampling of Pyrrosia s.l. (51 species), we investigated its phylogeny, biogeography, character evolution and environmental adaptation by employing five chloroplastid markers (rbcL, matK, psbA-trnH, and rps4 + rps4-trnS) and one single (low)-copy nuclear gene, LEAFY. Pyrrosia s.l. was divided into six major clades and eight subclades. Reticulate evolution was revealed both among clades and among species in Pyrrosia s.l. Ancestral character state optimization revealed high levels of homoplastic evolution of the diagnostic characters in Pyrrosia s.l., while the crassulacean acid metabolism pathway seems to have an independent origin. Molecular dating and biogeographic diversification analyses suggested that Pyrrosia s.l. originated no later than the Oligocene and the main clades diversified during the Oligocene and Miocene, with southern Asia, the Indo-China Peninsula and southwestern and southern China as the most likely ancestral areas. Transoceanic long-distance dispersal, rather than vicariance, contributed to the intercontinental disjunction. Diversification scenarios of Pyrrosia s.l. under geological movements and climate fluctuation are also discussed.
Pyrrosia Mirbel (Polypodiaceae) is a terrestrial fern genus that constitutes subfamily Platycerioideae B.K. Nayar along with Platycerium Desv1. This genus contains ca. 51–100 species and is widely distributed in Asia, ranging from Australia and New Zealand to Siberia and from Africa to various south Pacific islands2,3,4,5,6,7. Pyrrosia is well circumscribed by stellate hairs and characteristic connective venation pattern, which are two key characters to understanding the evolution of Polypodiaceae4,8,9,10. Most species of Pyrrosia are drought tolerant, and five species have been reported to use crassulacean acid metabolism (CAM) pathway11,12,13,14,15. The occurrence of CAM is considered an effective adaptation to arid environments, although CAM in lycophytes and ferns is rare16,17,18.
The uniform appearance of species of Pyrrosia leads to difficulties in species-level classification. Several authors have conducted taxonomic revisions on regional or worldwide scales2,3,4,6,8,19,20,21,22,23,24,25,26. Giesenhagen27 was the first to systematically describe the morphology and classification of 50 species. Ching26 studied species from mainland Asia and neighboring islands and treated Pyrrosia as a natural genus; he transferred 49 species into Pyrrosia and described five new species of this genus. Shing3, Shing and Iwatsuki2,20 considered more than 100 species in Pyrrosia and recognized 64 species in Asia and the adjacent Oceania. Hovenkamp4 completed a monograph of Pyrrosia from a global perspective and recognized ca. 51 species with a wide species concept. Finally, it is widely accepted that Pyrrosia contains ca. 60 species8. However, the infrageneric classification of Pyrrosia is controversial (Supplementary Table S1). Nayar and Chandra divided 14 species from India into six groups28. Shing3 divided Pyrrosia, excluding Drymoglossum C. Presl and Saxiglossum C. Presl, into two subgenera: subg. Pyrrosia and subg. Niphopsis. subg. Niphopsis includes only P. samarensis and P. angustata, and subg. Pyrrosia was divided into two sections and five series. Hovenkamp recognized ten groups based on cladistics analyses using 24 morphological characters, while seven species were not included in any groups4. Furthermore, he considered five groups therein were not well-established monophyletic group due to the lack of autapomorphies4. This consideration maybe related to the absence of trait evaluation based on a robust phylogenetic tree. Nearly all the groups established by Hovenkamp are irreconcilable with those recognized in Shing’s classification, except that the P. angustata-group of Hovenkamp is equivalent to the subg. Niphopsis of Shing. Yang completed a taxonomic revision of Pyrrosia, excluding Drymoglossum C. Presl and Saxiglossum C. Presl, in China and proposed two subgenera and six sections25. Unlike Hovenkamp, Yang grouped P. stigmosa, which previously belonged to the P. costata-group, with sect. Drakeanae, which was equivalent to the P. sheareri-group of Hovenkamp based on the fronds shape and indumenta characters. Yang also grouped P. subfurfuracea and P. calvata, belonging to the P. sheareri-group of Hovenkamp, with sect. Costatae, which is equivalent to the P. costata-group, based on the scales-bearing types25. In addition, infrageneric classification of Pyrrosia was always closely linked to the recognition of the segregate genera Drymoglossum and Saxiglossum 5,8,29,30,31,32. Drymoglossum and Saxiglossum are currently accepted as members of Pyrrosia, which has been confirmed by phylogenetic work33,34,35.
Several molecular phylogenetic studies involved Pyrrosia, some of which revealed that Pyrrosia was monophyletic and Platycerium was a sister group33,36,37,38,39,40,41,42,43,44,45. In the most recently study, Vasques et al.46 established a subgeneric classification of Pyrrosia based on three chloroplast (cp) markers of 38 species. Six subgenera were proposed: subg. Lune, subg. Neoniphopsis, subg. Niphobolus, subg. Niphopsis, subg. Pyrrosia and subg. Solis. Testo and Sundue studied the evolution of ferns based on six chloroplast markers of a 4000-species dataset suggested that Pyrrosia was paraphyletic and that P. liebuschii (Hieron) Schelpe was nested in Platycerium 47. Zhou et al.35 segregated the P. africana-group as a new genus Hovenkampia Li Bing Zhang & X.M. Zhou, and recognized four clades in Pyrrosia based on five cp markers in the recently study. Zhao preliminarily studied the infrageneric relationship of Pyrrosia based on 26 species, mainly in Asia, by using one cpDNA, rps4-trnS, and one nrDNA, LEAFY 48. Four main clades were recognized in Pyrrosia by rps4-trnS and a potential hybrid origin of P. piloselloides was suggested. Nevertheless, the sampling is far from completion, and the P. africana-group was not included. Evidences from both single parent genetic chloroplast markers and parental genetic nuclear gene are urgent needed to test the monophyletic of Pyrrosia and further understand the phylogenetic relationship within Pyrrosia based on more taxa.
The integration of phylogenetics, historical biogeography, paleogeography and climatology has provided a new perspective to understand the origin and diversification of biotas, which are of great interest to evolutionary biologists47,49,50,51,52,53,54,55,56,57,58,59. Pyrrosia is exclusively distributed in paleotropic regions, while Platycerium, the mostly related group of Pyrrosia, is distributed in paleotropic regions and South America4. Based on morphological analyses, Hovenkamp4 proposed that Pyrrosia originated from Africa in the Jurassic and that the present distribution of the genus was reached via ‘rafting’ to the Indian subcontinent and Australia. Schneider et al.56 and Schuettpelz and Pryer60 dated the original time of Pyrrosia to the late Eocene (ca. 35 Ma) based on molecular dating, while Testo and Sundue47 established the original time of the main lineage of Pyrrosia in early Paleocene (ca. 63.4 Ma). Kreier and Schneider43 discussed the phylogeny and biogeography of Platycerium and reestablished two lineages (Africa + Madagascar + South America and Australia + Asia) in Platycerium. However, the ancestral distributions for the basal nodes are poorly resolved. Therefore, elucidating the historical biogeography of Pyrrosia is perplexing due to differences in the distribution and species diversity centre of Pyrrosia from those of Platycerium.
The present study, based on comprehensive taxa sampling, employs nucleotide sequences of five chloroplast DNA markers (rbcL, matK, psbA-trnH, and rps4 + rps4-trnS) and one single (low)- copy nuclear gene, LEAFY, to reconstruct the phylogeny of Pyrrosia and to explore its historical biogeography. In addition, evolution of the morphological diagnostic characters and environmental adaptations related to drought resistance are investigated.
Five cpDNA gene regions—matK, rbcL, psbA-trnH and rps4 + rps4-trnS— were amplified in all 109 Pyrrosia individuals and the related taxa in Polypodiaceae and Davalliaceae. The sequence characters and parsimony-informative sites for individual gene markers are summarized in Table 1. The combined data matrix of the cpDNA fragments included up to 3760 nucleotides, of which 1325 (32.58%) were variable and 1054 (28.03%) were potentially parsimony informative. A unique insertion/deletion (up to 169 bp at site 131 to 300 bp) was present in the psbA-trnH gene of Pyrrosia. Maximum parsimony (MP) analyses of the combinded cpDNA data set resulted in 14,809 equally most parsimonious trees with a length of 4,007 steps. The consistency and retention index were relatively high (CI = 0.60, RI = 0.86). The optimal maximum likelihood (ML) phylogram for the combinded cpDNA data was -InL = 26,674.4552.
Polymerase chain reaction (PCR) amplification of LEAFY was successfully performed in 81 individuals representing 39 species of Pyrrosia and the related taxa in Polypodiaceae and Davalliaceae. The amplified LEAFY gene fragment included intron 1, exon2, intron 2 and the flanking exon 1 and exon 3 sequences. The aligned data matrix had 1081 characters, including 606 variable characters and 515 potentially parsimony-informative characters (Table 1). The MP analyses were stopped with 13,200 equally most parsimonious trees of 1539 steps sampled (CI = 0.61, RI = 0.88). The optimal ML phylogram had an –InL value of 9,969.4164.
The three phylogenetic analyses (MP, ML and Bayesian inference (BI)) revealed congruent topologies based on the combined data set of five cpDNA markers. Platycerioideae was recognized as a well-supported monophyly. The traditional Pyrrosia was paraphyletic, six main monophyletic clades (labelled as clades I–VI) were resolved (Fig. 1). Clade I clustered with Platycerium with weak supported, and this clade was the endemic African species, P. schimperiana (Mett. ex Kuhn) Alston (=Hovenkampia schimperiana (Mett. ex Kuhn) Li Bing Zhang & X. M. Zhou) which belonged to the P. africana-group of Hovenkamp4. Herein, we use Pyrrosia s.l. to represent Pyrrosia including P. africana-group, and Pyrrosia s.s. to represent Pyrrosia without P. africana-group. Clade II comprised P. costata and P. stigmosa, this clade corresponds to the P. costata-group described by Hovenkamp4, and it was resolved as the basal clade of Pyrrosia s.s. Clades III only contained P. niphoboloides. Clades IV to VI consisted of species distributed in Asia and Oceania, except the widespread P. lanceolata, which occupied Asia, Oceania and Africa. Clade IV contained four subclades (subclades A, B, C and D): P. angustissima, a species that previously belonged to the separated monotype genera Saxiglossum, was resolved as subclade A; P. nummulariifolia, P. rasamalae and P. kinabaluensis comprised subclade B with high support; subclade C consisted of species of the P. angustata-group; subclade D included the P. lingua-group and two undecided species, P. laevis and P. ensata. Clade V was strongly supported as the sister to clade VI, and both contained two subclades (subclades E and F; subclades G and H). Clade V contained P. piloselloides (subclade E) and species from the P. confluens-group, P. lanceolata-group and P. foveolata (subclade F). Clade VI was separated into two well-supported subclades: subclade G contained some of the species of the P. sheareri-group, and subclade H consisted of some of the species of the P. sheareri-group and species of P. porosa-group.
Clade I, II, V and VI were recognized as monophyletic groups. Clade III was the basal clade of Pyrrosia s.l. Phylogenetic topologies of cpDNA and LEAFY showed significant conflicts of the relationship among clades I, IV and V, and subclades within clade IV (Fig. 2). (1) Clade I and V embed in clade IV, and resulted the monophyly clade IV in cpDNA trees splited into three parts in the tree generated from LEAFY: the first part, P. angustata (subclade C), was resolved as the basal group of clade IV; the second included subclades A, B and a species, P. laevis, which belonged to subclade D; the third included species from subclade D. (2) The phylogenetic position of P. schimperiana (clade I, = H. schimperiana, P. africana-group) was contradictory between datasets: P. schimperiana was resolved as an independent clade in the cpDNA trees (Fig. 1), while in the tree generated from LEAFY, P. schimperiana was nested in the second part of clade IV. (3) Clade V was the sister group of clade VI in the cpDNA trees, while it nested into clade IV and clustered with subclade D in the phylogenetic tree of LEAFY.
Furthermore, an individual of P. lingua had two sequence types of LEAFY: the sequences of one type clustered with other samples of P. lingua, and the sequences of another one clustered with P. petiolosa. Similarly, one individual of P. porosa also had two sequences types, one type clustered with P. nudicaulis, and the sequences of another type nested with P. assimilis.
The mean ages and 95% highest posterior density (HPD) intervals of all labelled nodes within Platycerioideae were indicated in the chronogram (Fig. 3b) and in Table 2. The crown age of Platycerioideae was in the late Eocene (node 1, 37.98 Ma, HPD = 25.75–46.71 Ma). Clade I was separated from Platycerium at ca. 26.33 Ma (node 5, HPD = 16.78–39.39). The estimated age of the Pyrrosia s.s. (clades II–VI) was approximately the boundary of the Eocene and Oligocene (node 2, 33.71 Ma, HPD = 22.64–42.43). The most recent common ancestor (MRCA) of the main Asian and Oceanian species was dated backed 30.07 Ma (node 3, HPD = 20.80–38.75), Clade IV separated from clades V and VI at ca. 27.40 Ma (node 4, HPD = 19.56–35.90), and the latter two clades separated at ca. 24.19 Ma (node 6, HPD = 16.71–31.16). The crown node of clade IV was dated back to 22.82 Ma (node 7, HPD = 14.06–30.73). Estimated crown ages for clades V and VI are very similar, at 17.98 Ma and 18.14 Ma, respectively (nodes 11 and 10, Fig. 3b and Table 2).
Ancestral area reconstruction
The ancestral distributions obtained from the dispersal-extinction-cladogenesis (DEC) model for the major clades are shown in Fig. 3b and Table 2. The DEC model suggested that Platycerioideae most likely originated in areas E (Fig. 3b, node 1). The origin for the MRCA of Pyrrosia s.s. was unclear; two areas including southern Asia, Indo-China Peninsula and southwestern and southern China (area B), and Africa (area E), were all the supposed origin. The MRCA of the most Asian and Oceanian species originated in areas B (node 4). Clade IV may have originated in areas B and expanded towards Eastern Asia (area A) and Malesia (area C). The sister clades V and VI may have originated in area B and then expanded towards Eastern Asia (area A) and Australasia (area D), respectively. Clade I and Platycerium were suggested that originated in areas E. Several historical dispersal events were inferred, including the dispersal from area B to A (26.5 times), C (26.5 times), D (17.5 times) and E (17 time), and from area E to C (4.5 times), B (4 times), A (0.5 time), D (0.5 time) and D to B (3 time), C (3 times). The split of Pyrrosia s.s. from clade I and Platycerium and the split of clades II, V and VI, the diversification of the crown group of clade VI, V and VI and the infraspecific range expansions were suggested as dispersal events. The split of clade III from clades IV–VI was estimated as vicariance (node 3). Several independent dispersal events were detected within clades IV and V, respectively.
Infrageneric relationships within Pyrrosia s.l
Six main clades and eight subclades were recognized according to the cpDNA phylogenetic reconstructions of Pyrrosia s.l. in this study. We suggested 11 groups for the infrageneric delimitation of Pyrrosia s.l.
P. schimperiana was resolved as clade I which clustered with Platycerium. This results was similar to the studies of Testo and Sundue47 and Zhou et al.35, but different from the most studies that treat P. schimperiana as the basal group of Pyrrosia 44,46, Zhou et al.35 established a new genus Hovenkampia, and described its diagnostic characters different from Pyrrosia s.s. as: rhizomes completely parenchymatous; stomata polocytic; perispore thin, tightly adhering to the exospore surfaces. As Hovenkamp4, Hennipman10 and Uffelen and Hennipman24 pointed that the characters above are all not synapomorphy of P. africana-group. Rhizomes of the P. rhodesiana (P. porosa-group) are parenchymatous. Stomata of P. mannii and P. penangiana are polocytic. P. africana-group sheared the finely granulose epispore ornamentation with the species in clade VI of our study. Furthermore, scales of this species are pseudopeltate; the fronds are monomorphic, oblanceolate and estipitate; the indumenta are monomorphic and indumenta rays are narrow boat-shaped, these characters are all shared by more than one group of Pyrrosia s.s.
Clade II is monophyletic and as the basal group of Pyrrosia s.s. lineage. Species of this clade are all monomorphic fronds, short rhizome, basifixed scales. Smooth spores are the homogeneous characters of this clade. This clade is in the equivalent of P. costata-group of Hovenkamp4, Pyrrosia Clade of Zhou et al.35, and subg. Pyrrosia of Vasques et al.46.
Clade III contained only P. niphoboloides, which was assigned to the P. piloselloides-group by Hovenkamp4 (or members of previous segregated genus Drymoglossum) along with P. piloselloides. However, these species did not cluster together in the present cpDNA phylogenetic trees, with P. piloselloides clustered in clade V, which indicates that the P. piloselloides-group described by Hovenkamp is polyphyletic, although the members have similar morphological characters (e.g. dimorphic fronds, venation, and sori arrangement). While, P. niphoboloides was included in Galeoglossa subclade, which is in the equivalent of subclade B of clade IV of our study in Zhou et al.35. Vasques et al.46 treated this species in subg. Solis, which was the basal group of Pyrrosia except subg. Lunae.The sequences of P. niphoboloides we used were downloaded from GenBank. The phylogenetic position of P. niphoboloides still need more evidences.
Members of clade IV are characterized by long-creeping rhizomes, and peltate scales, but they have various frond shapes, indumenta and venation types, and spore ornamentation. This clade has four relatively independent subclades. Species of this clade are widely distributed in East and South Asia, and extend to the Indonesia Archipelago. The basal subclade A, which includes only P. angustissima, had been previously described as the segregated genus Saxiglossum 26. We recognized that P. angustissima belonged to the basal member of clade IV.P. angustissima has some autapomorphy characters, including linear-triangular rhizome scales, linear laminae with involute margins and a special type of drynarioid venation26,31,61. These features of P. angustissima clearly distinguish it from other members of clade IV, which makes it a separate subgroup. Subclade B contains three species, P. nummulariifolia, P. kinabaluensis and P. rasamalae, all of which belonged to the P. albicans-group of Hovenkamp. These species have succulent lamina with distinct water-tissues, dimorphic indumenta (acicular and woolly rays), and spores with longitudinal ridges and finely granulose, but they lack hydathodes. Subclade C contains P. angustata and P. samarensis, which belong to P. angustata-group of Hovenkamp. This subclade has similar morphology, including dimorphic fronds, entire margin scales, pseudo-drynarioid venation, and spores with longitudinal ridges (which occurs in only this subgroup). Subclade D, which was recognized as monophyletic, includes P. laevis, P. ensata and the P. lingua-group. The close relationship between the P. lingua-group and P. laevis and P. ensata revealed here agrees with the findings of Yang25, who assigned them to section Pyrrosia. Hovenkamp did not consider the P. lingua-group to be a well-established monophyletic group because the morphology of the spore was heterogeneous and there are no autapomorphies in this group. We found that species of subclade D all featured long-creeping rhizomes and fronds from monomorphic to moderately dimorphic with distinct stipitate and lamina ovate to lanceolate. In this subclade, persistent boat-shaped ray hairs exist in all species and coarsely sparse tubercules are the main spore ornamentation. The presence of common features in subclade D suggests that this group is monophyletic. Clade IV was mostly similar to subg. Niphopsis of Vasques et al.46 and Niphopsis clade of Zhou et al.35 with P. niphoboloides an exception. Nevertheless, the phylogenetic relationships among species, especially the species of subclade D were still uncertain.
Clade V includes subclades E and F. Subclade E contains only P. piloselloides, which is a member of the P. piloselloides-group. Subclade F contains species of the P. confluens-group, the P. lanceolata-group and an undecided species, P. foveolata. Both of the former two groups were considered as monophyletic by Hovenkamp4, on the basis of restricted hydathodes, monomorphic indumenta, large sori with short paraphyses (in the P. confluens-group), and sunken sori with centrally situated paraphyses (in the P. lanceolata-group). Nearly all species of these two groups and P. foveolata have common features, such as: long-creeping rhizomes, fronds that are dimorphic in various ways (with the exception of P. longifolia), lamina that are elliptic or elongated to strap-shaped with decurrent base and indistinct stipes, sori that are sunken with distinct stellate paraphyses, indumenta that are monomorphic and persistent, indumenta rays that are short boat-shaped, and perispore that are bilateral with tuberculate and irregularly verrucated protuberances. Evidence was sufficient to support the P. confluens-group, P. lanceolata-group and P. foveolata as monophyletic. Clade V of our study was equivalent to subg. Niphobolus of Vasques et al.46 and Niphobolus clade of Zhou et al.35. P. rasamalae was included in Niphobolus clade inexplicably in Zhou et al.35, while, this species was treated as a member of Clade IV in our study with highly support.
Clade VI occupies the most obvious common characters, including thick and short rhizomes, monomorphic fronds, and distinct hydathodes. Subclades G and H are resolved as separate monophylies with highly supported values. Subclade G contains P. bonii, P. calvata, P. subtruncata, P. fengiana, P. sheareri, P. drakeana and P. subfurfuracea. Three species of the P. sheareri-group together with P. polydactyla, P. hastata, and P. flocculosa are nested in subclade H, which contains most species of the P. porosa-group. Our results are quite different from Yang’s treatment, in which P. subfurfuracea and P. calvata were added to the P. costata-group based on similar thick and short rhizomes, un-peltate (basifixed and pseudopeltate) scales and other features25. Subclade G shares the following common features: pseudopeltate or basifixed scales, monomorphic fronds with a distinct stipe, densely granulated perispores, and dimorphic indumenta with acicular rays appressed to a layer with woolly rays (indumenta of P. sheareri are appressed and boat-shaped). Subclade H contains species of the P. porosa-group, three species belonging to the P. sheareri-group and three undecided species - P. transmorrisonensis, P. mannii, and P. penangiana. Two separate species, P. mannii and P. penangiana, were resolved as the basal lineages of subclade H. P. polydactyla and P. hastata formed the sister group of the lineages including the P. porosa-group and P. flocculosa. The frond shape of P. hastata and P. polydactyla is most unique in Pyrrosia s.l., and both are pedately divided to 4/5 depth of the frond into 3–5 to 6–8 divisions with distinct stipes. The indumenta are boat-shaped and persistent. The other species of subclade H share many common features, for example, linear to narrow lanceolate and oblanceolate fronds with gradually narrowed and decurrent fronds base, and fronds estipitate; the indumenta are persistent and dense with acicular rays, and in most species, acicular rays are appressed to a layer with mainly woolly rays (woolly ray stellate hairs are not obvious in P. gralla, P. assimilis, and P. penangiana). Although P. polydactyla and P. hastata shared some morphological characters with subclade G, they were polyphyly in our study, and paraphyly in Vasques et al.46 and Zhou et al.35. Zhou et al. combined them as a P. sheareri group35.
Incongruence between the cpDNA and LEAFY phylogenetic trees and potential reticulate evolution
The chloroplast genome is maternally inherited in ferns, and the nuclear genome exhibits amphilepsis. Consequently, comparative studies of these two genomes might uncover potential reticulate evolution, including introgression and/or hybrid speciation62. LEAFY is a well-studied single-copy gene in ferns, and it has been successfully used to resolve reticulate evolution at low taxonomic levels63,64. Considering the high proportion of hybridization in ferns and the advantages of LEAFY, both cpDNA and LEAFY were included in the phylogenetic analyses.
Each of four clades I, II, V and VI was resolved as monophyletic with high support values, both in the cpDNA and LEAFY gene trees. CpDNA and LEAFY gene trees showed incongruent phylogenetic positions for clade I and clade V. Both of clade I and V nested in clade IV in the nuclear tree. Hybridization and incomplete lineage sorting (ILS) are two important factors that lead to phylogenetic incongruence65,66,67. According to the result of ancestral area reconstruction, the potential ancestral area of clade I was area E, ancestral areas of clade V are area B or E, and the ancestral area of clades IV and VI was B. Dispersal events have been inferred between area B and E. By contrast, ILS is resulted from the ancestral alleles being sorted into some lineages randomly. In our study, both alleles of clade I and V nested in clade IV, thus, hypothesis of ILS seems less plausible although it is difficult to distinguish from ancient hybridization, especially without genomic data.
We suggested that clade V might be an ancient hybrid origin and reached the current distribution areas during species dispersal in history. Ancestral species of clade IV might be the male parent of clade V and ancestral species of clade VI might be the female parent of clade V. We suggested three potential scenarios of the origin of clade I: (1) the new established genus Hovenkampia was an ancient hybrid origin, ancestral species of Platycerium were the female parent and Pyrrosia s.s. were the male parent; (2) P. schimperiana is hybrid species with other species in Hovenkampia and species in clade IV as its parents; (3) three individuals of P. schimperiana we used in this study are hybrid individuals.
In subclade H, P. porosa had two divergent homoeologous copies of LEAFY, one copy of which clustered with P. nudicaulis, while another copy clustered with P. assimilis and then nested with P. tonkinensis. In the cpDNA tree, individuals of P. porosa, P. nudicaulis, and P. assimilis were clustered together and were distinctly separated from P. tonkinensis. In addition, the basal chromosome number within Pyrrosia s.l. was 36 or 37; most of the reported data of P. porosa are tetraploid and hexaploid4, only one diploids of P. porosa have been reported in India68. It revealed that some plants of the P. porosa might be allopolyploid. Furthermore, one individual of P. gralla clustered with P. davidii, and the copies of another individual nested in P. tonkinensis in the phylogenetic tree of LEAFY, while it was distinctly separated from P. tonkinensis in the cpDNA tree. The short branches of cpDNA trees (Fig. 1) and molecule dating (Fig. 3) showed that P. porosa, P. nudicaulis, P. assimilis, P. tonkinensis, and P. gralla were differentiated recently. Based on the results of cpDNA and nrDNA, a recent hybrid or reticulate evolution were revealed in P. porosa-group. In subclade D, P. lingua also had two divergent homoeologous copies of LEAFY, a copy of which clustered with P. martini, while another one clustered with P. petiolosa in the phylogenetic tree of LEAFY. However, individuals of P. lingua and P. petiolosa were distinctly separated in the cpDNA tree. Only diploids were reported in P. lingua and P. petiolosa. Incongruence between the cpDNA and LEAFY trees suggested a recent hybrid might exist in P. lingua-group. Nevertheless, owing to the limited information on chromosome numbers of Pyrrosia s.l., the acquisition of more exact results will require more samples and further comprehensive cytological studies.
Morphological characters assessment and evolution
Several features, including the presence of specialized sterile fronds, rhizome growth-form, the distance between adjacent phyllopodia, the insertion type of rhizome scales, scale margin morphology, stomata type, venation type, sori arrangement, indumenta shape and epispore ornamentation, were all treated as diagnostic characters for infrageneric classification and species delimitation of Pyrrosia s.l. in previous studies2,3,4,6,8,20,22,23,24,31,61,69. If we only use morphologic characters to evaluate the infrageneric classification in Pyrrosia s.l., most groups or sections are not monophyletic4,25. Ancestral character state optimization based on a stable molecular phylogenetic tree in this study enabled a synthetic evaluation of all diagnostic characters (Fig. 4).
(1) The presence of specialized fertile fronds is essentially different from monomorphic and moderately dimorphic fronds, in which fertile fronds are the same or longer and narrower than sterile fronds. Dimorphic fronds appeared in clade III and scattered in clades IV and V. Hovenkamp4 considered that monomorphic fronds transformed in the presence of specialized sterile fronds; thus, it can be observed that dimorphic fronds have evolved multiple, independent times in Pyrrosia s.l. (Fig. 4a). (2) Rhizome growth-forms and the distance between adjacent phyllopodia are relevant characters. Short and thick rhizomes correspond with contiguous phyllopodia and long-creeping rhizomes correspond with separate phyllopodia. Long-creeping rhizomes are homoplastic characters, occurring in clades III, IV and V (Fig. 4b). (3) Regarding the insertion type of rhizome scales, basifixed scales are considered ancestral, and they occur in clades II and VI. Peltate scales are the predominant type in Pyrrosia s.l., and both basifixed scales and pseudopeltate scales appear to have evolved at least five times (Fig. 4c). (4) Scales with ciliate-dentate margins are the mostly ancestral types, and scales with long and curly cilia margins are apomorphic within clade VI, while scales with entire margins independently evolving at least five times (Fig. 4d). (5) The dominant pericytic stomata type of Pyrrosia s.l. is a deuterogenic feature in Polypodiaceae. Three species of Pyrrosia s.l. (P. schimperiana, P. mannii and P. penangiana) still have polocytic stomata, which are common in other genera of Polypodiaceae. Polocytic stomata in Pyrrosia might be the result of reversion or secondary development (Fig. 4e). (6) Pseudo-drynarioid and campyloneuroid venation types of Pyrrosia s.l. are deuterogenic features in Polypodiaceae, and drynarioid venation is the ancestral feature. Species occupying pseudo-drynarioid and campyloneuroid venation were both polyphyletic, and they interspersed in clade II, IV, V and VI. A high frequency of homoplasy of venation types appears within Pyrrosia s.l. (Fig. 4f). (7) Sori arrangement is an important diagnostic character in Pyrrosia s.l. and the other ferns. Confluent sori and orbicular and discrete sori have been treated as the diagnostic character even at the species and genus level in Polypodiaceae69,70,71,72. Coenosori was be considered as one of the main characters to separate the previous genus Drymoglossum from Pyrrosia s.l., although some authors considered this classification “purely artificial”69. The homoplasy of confluent sori in Pyrrosia s.l. was confirmed in this study. Meanwhile, the other three types of sori arrangement were also homoplasy (Fig. 4g). (8) There are four main shapes of stellate hairs in Pyrrosia s.l.: hairs with straight acicular rays, those with straight boat-shaped rays, those with curly wooly rays, and those with straight boat-shaped and curly wooly rays at one axis. The indumenta of each species may be monomorphic or dimorphic. Monomorphic indumenta contain only one type of stellate hair above, while dimorphic indumenta contain two types of stellate hairs. We recognized five types of indumenta in Pyrrosia s.l. (Fig. 4h) and found that both monomorphic and dimorphic indumenta are homoplastic. Dimorphic indumenta occurred in clades II, IV and VI, while monomorphic indumenta occurred in all clades. In addition, clades I, II and V, and subclades D all have boat-shape or lanceolate hairs, and the other clades (subclades) except P. hastata, P. polydactyla, and P. sheareri have acicular hairs. Overall, indumenta of Pyrrosia s.l. are homoplasy (Fig. 4h). (9) Epispore ornamentation appears relatively unique for each clade or subclade. Clades III and V share the obviously bisculpate spores with dense, small granulate and warty or spike protuberances, and clades I and VI have spores with more or less dense finely granulose protuberances. Smooth spores appear only in clade II, and spores with longitudinal ridges appear only in subclade C. These two types of epispore ornamentations can be treated as synapomorphies of clade II and subclade C, respectively. Epispore ornamentation of subclades A and D appears sparsely granulose and irregularly ridged (with the exception of P. abbreviata), and those of subclade B and P. abbreviata of subclade D are longitudinally ridged with finely granulose protuberances (Fig. 4i).
Ancestral character state optimization revealed high levels of homoplastic evolution in Pyrrosia s.l. Only the epispore ornamentation is synapomorphic in clade II and subclade C, respectively. There is no single character that can be used as an apomorphy to distinguish groups (clades and subclades) in Pyrrosia s.l. from each other; thus, the combination of two or more characters are necessary to identify all groups. For instance, given that rhizome growth-forms are similar between subclades G and H and clade II, combining this character with the indumenta morphology and the fronds shape can identify clade II, while combining the frond shape and insertion type of rhizome scales can identify subclades G and H. Furthermore, some of the anatomical characters, such as the distribution pattern of collenchyma, sclerenchyma and parenchymatous cells in rhizomes, might be helpful in defining traits for monophyletic group recognition and species identification in Pyrrosia s.l.6,22.
Most species of Pyrrosia s.l. are extremely drought tolerant. The morphological characters such as the texture of the lamina that is coriaceous, thick-leathery, succulent, or peronate; the presence of a thick cuticle on the epidermis; sunken stomata; sunken sori; or distinct hydathodes are all xerophytic adaptations. Regarding drought resistance, poikilohydrous and succulent forms are considered two different growth-forms4. Poikilohydrous plants of Pyrrosia s.l. can roll and stretch their fronds in response to drought resistance, which are similar to resurrection plants. Most of these species, such as P. porosa, P. schimperiana, are short rhizomes, the anticlinal walls of the adaxial epidermis appear thin or slightly thickened, hypodermis and water tissue are absent or indistinct, and the indumenta form dense mat. Poikilohydrous plants are adaptable to seasonal climates with long dry periods. By contrast, succulent plants of Pyrrosia s.l. species possessing fronds with thick water tissues, such as P. abbreviata, P. angustissima, P. longifolia and P. nummulariifolia, can survive by storing water during short periods of drought. These species are long-creeping rhizome, the adaxial epidermis are strongly or moderately thick, and stomata are sunken or slightly sunken. In the phylogenetic tress, poikilohydrous plants of Pyrrosia s.l. nested in clades I, II and VI, while, succulent form species nested in clades III, IV and V. These two growth-forms adapting to drought resistance are all independently polyphyletic in Pyrrosia s.l.
In addition to morphological specializations, another key innovation associated with the success of Pyrrosia s.l. in more arid habitats is the form of photosynthesis known as crassulacean acid metabolism (CAM). The CAM cycle has been reported in five species of Pyrrosia s.l., namely, P. confluens, P. dielsii, P. lanceolata, P. longifolia, and P. piloselloides 11,12,15,18, while the C3 pathway is employed in P. rupestris, P. costata, P. flocculosa, P. porosa, P. schimperiana and P. sheareri. CAM is an important adaptation of photosynthetic carbon fixation, and CAM plants restrict gas exchange with the atmosphere during the daytime. The water use efficiency of CAM plants is better than that of C3 plants, and consequently, CAM plants can adapt to many tropical and subtropical environments with intermittent or strong seasonal water supply73.
Chloroplast-containing cells of CAM plants fix CO2 initially at night, and during the daytime, organic acids behind closed stomata create an internal CO2 source that is re-assimilated by rubisco in the chloroplast74. Stomatal behaviour is relate to water pressure and CO2 concentration75. CAM is correlated with various anatomical and morphological features responding to water pressure. We found that stomata of species that utilize the CAM pathway are deep sunken, stoma grooves are strongly contracted above the stomata, and the number of epidermal cells adjoining the stoma groove is 6–10 (12)27 (Supplementary Fig. S2). Furthermore, these species live in exposed trunks or rocks and occupy more exposed microhabitats, even living together with C3 epiphytes15. Based on the available ecological information and the characteristics of stomata and the epidermal cells16,74, we suggested that species utilizing CAM pathway including the five has been reported clustered in clade V. CAM pathway might have a single origin in Pyrrosia s.l. The results of our molecular dating indicate that the divergence time of clade V was dated to ca. 17.98 Ma (node 11), which revealed an early Miocene origin of the CAM pathway in Pyrrosia s.l.
The emergence of CAM photosynthesis in different taxa may be driven by the same external driving forces. Just as the occurrence of CAM pathways in Bromeliaceae was driven73, progressive aridification and declining CO2 concentrations during the Oligocene and Miocene may have gradually favored for the emergence of CAM photosynthesis in Pyrrosia s.l.
Diversification of Pyrrosia s.l.
The poor fossil record of polygrammoid ferns discourages estimations of exact differentiation times and recognition the distribution area of ancestors. Molecular dating by Schuettpelz and Pryer60 showed that Pyrrosia s.l. originated in Oligocene (ca. 35.1 Ma). Testo and Sundue47 estimated the origination time between Pyrrosia s.l. and Platycerium to be 63.41 Ma. The diversification of most of the main clades of Pyrrosia s.s. in Testo and Sundue’s study were also dated back to the Oligocene, with a steady period of ca. 20 My from the early Eocene to the early Oligocene76. Therefore, we believe Pyrrosia s.l. originated no later than Oligocene and underwent diversification during the Oligocene and Miocene.
Ancestral area reconstruction based on the DEC model in RASP 3.2 revealed that Southern Asia, the Indo-China Peninsula and southwestern and southern China (area B); and Africa (area E) are the probable ancestral areas of Pyrrosia s.l. Janssen et al.44 suggested that Pyrrosia s.l. might be of African origin because the African species P. liebuschii was the basal clade of Pyrrosia s.l. Holtum5 suggested that the African species may have dispersed from Asia because P. sheareri is closer to the primitive conditions, and this species predominantly lives in China with a centre of distribution located in Southeast Asia. Although P. sheareri is neither the ancestor nor the basal species of Pyrrosia s.l. based on our phylogenetic analyses, area B, particularly the Himalayan region, has a considerably higher species diversity. Furthermore, the phylogenetic analyses based on LEAFY determined that P. costata (representing clade II), was the basal clade, and the Africa species P. schimperiana (clade I) was nested into clade IV. The ancestral area of clade II was assumed to be area B or areas B and C, which further demonstrated that area B is most likely the original area of Pyrrosia s.l. Moreover, this area also displays high species diversity in nearly all subfamilies in Polypodiaceae, and it is another diversity center outside of the tropical Americas77. Southeast Asia is the origin area of Thylacopteris 40, Microsorum 78, Lepisorus 79 and drynarioid ferns45, as well as some taxa in Eupolypods II Deparia 53 and some angiosperms49,55. Similar patterns were also found in the closely related microsoroids, which have a diversity centre in southeastern Asia and colonized to African regions several times44.
The distribution of Pyrrosia s.l. presents an Africa-Asia disjunction, which is common in plants and has recently attracted much attention42,44,49,54,55,57,80,81,82,83. Four main driving mechanisms for the disjunctive distribution have been summarized: (1) transoceanic long-distance dispersal, (2) overland migration via land bridges, (3) boreotropical dispersal via Eocene forests, and (4) rafting of India49,54,82,83. Hovenkamp4 proposed that the disjunctive distribution pattern of Pyrrosia s.l. resulted from vicariance, such as the breakup of Africa from Gondwana (120–140 Ma) and the separation of Australia from India (125 Ma), rather than dispersal. The Indian plate became progressively more isolated from eastern Gondwana in the Cretaceous and Paleocene, and it moved northward towards Asia with glancing contact to Asia at ca. 57 Ma84. In our analyses, the divergence time of Africa lineages and Asia-Australasian lineages was estimated to be much younger (Fig. 3b); thus, the split could not be vicariance resulting from the breakup of Gondwana and Laurasia, and the Indian plate as either a raft or stepping stone is therefore too old for the origin of Pyrrosia s.l. In addition, the “boreotropical” floristic connection hypothesis, in which the plants moved across the North Atlantic during late Paleocene to the middle Eocene, presents a time frame that is too early to explain the Africa-Asia disjunction of Pyrrosia s.l.85,86. The closure of the Tethys Sea led to the direct connection between mainland Africa and western Asia during the early Miocene87,88, and the split time of the African lineages and Asia-Australia lineages seems to fit the time frame of the overland migration hypothesis (Fig. 3b). However, no species or fossil records of Pyrrosia s.l. have been reported in northern Africa or the adjacent Asian area, which weaken the support for this hypothesis. Overland migration is therefore an alternative dispersal scenario. Transoceanic long-distance dispersal has been used extensively to explain the intercontinental disjunction of plants49,52,53,54,57,82,83,89,90. Fern spores are minute, and like wind-dispersed seeds and pollens, they may have been transported transoceanic by prevailing monsoon wind systems or ocean currents between Africa and northern India44,53,57,91. Transoceanic long-distance dispersal might be the most plausible hypothesis to clarify the Africa-Asia disjunction of Pyrrosia s.l.
Within Pyrrosia s.s., multiple dispersal events from area B to C and sequentially to area D, and from area B to A can be inferred. Southwest Asia suffered frequent orogenesis, particularly the Himalayan regions, after experiencing many rapid uplifts and unroofing92. Meanwhile, Southeast Asia also suffered a complex interplay of plate movements and grew incrementally by the addition of continental fragments. Australia collided with Southeast Asia ca. 25 Ma years ago93. Frequent and severe geological movements during the Miocene provided the possibility for the dispersal between areas. The global climate fluctuated dramatically from the late Oligocene, as the monsoon system was established and subsequently strengthened by the late Oligocene warming, mid-Miocene climatic optimum and persistent Miocene cooling86,94,95. Dramatic climate fluctuation may have triggered the speciation and diversification of Pyrrosia s.l., and frequent habitat fragmentations and range transition may have led to the accumulation of species diversity.
We infer the following diversification scenario for Pyrrosia s.l.: Pyrrosia s.l. originated from Southern Asia, the Indo-China Peninsula and southwestern and southern China (area B) before the Oligocene, and the global climate subsequently underwent a rapid cooling at the late Eocene and the early Oligocene with the temperature declining ca. 5 °C86. Contemporaneously, Southeast Asia, particularly the Himalaya regions, experienced aridification. The sudden cooling and aridification may have been a devastating blow for Pyrrosia s.l. Some of the ancestral species may have migrated or dispersed to the much warmer southern areas and finally reached Africa. It is undisputed that climate cataclysm and long-distance dispersal can cause species extinction. The species of Pyrrosia s.l. that reached Africa were random, and thus, the African species are polyphyletic rather than monophyletic. A steady time interval of nearly 20 My before 34 Ma in Testo and Sundue’s47 study may also indicate lower speciation rates or/and increased extinction rates of Pyrrosia s.l. in that time. Fossil pollen data indicated that Oligocene climate and vegetation can be characterized as “tropical seasonal” but under fairly humid conditions in East Africa, and rain forests experienced strong successive retreats in the subsequent time period96. Ecosystems began to transform into much drier grassland, and many ecosystems, including widespread grasslands/savannah with many new C4 plants, were established across eastern Africa in the mid-Miocene based on grass macrofossils and pollen data. After 17 My of cooler conditions, the global climate reached a warm phase peak during the late middle Miocene (15–17 Ma), known as the mid-Miocene Climatic Optimum. During the Miocene, speciation of Asia-Australia groups was flourishing, which may have been triggered by frequent geological movements and dramatic climatic fluctuations during that time. In contrast, lower average speciation rates or/and increased extinction rates resulted in less diversity in ferns in Africa97, which further resulted in a disproportional species number between Africa and Asia-Australia.
Materials and Methods
Our taxon sampling strategy was designed to include all six subfamilies of Polypodiaceae: Platycerioideae, Loxogrammoideae, Drynarioideae, Microsoroideae, Polypodioideae and Grammitidoideae and its sister group Davalliaceae based on the latest classification for extant lycophytes and ferns, PPGI1. In Platycerioideae, all groups of Pyrrosia s.l. proposed by Hovenkamp4 as well as the two previously segregated genera Drymoglossum and Saxiglossum were included. In total, 112 individuals representing 51 species of Pyrrosia s.l. were sampled (Supplementary Table S2). One to five individuals from the different regions were collected for each species. 19 individuals were used besides Pyrrosia s.l.: Platycerium bifurcatum (Cav.) C. Chr., Pl. coronarium (O.F. Müll.) Desv. and Pl. wallichii Hook were sampled in Platycerium. Samples of Loxogrammoideae, Drynarioideae, Microsoroideae, Polypodioideae and Grammitidoideae and its sister group Davalliaceae were chosen as outgroups for the phylogenetic analyses.
Fresh healthy leaves were dried immediately in silica gel, and kept dry until DNA extraction. The voucher specimens were deposited at the Herbarium of Institute of Botany, the Chinese Academy of Sciences (PE) and the Herbarium of Institute of Medicinal Plant Development, Chinese Academy of Medical Sciences (IMD). Voucher information is listed in Supplementary Table S2.
DNA extraction, PCR amplification, cloning, and sequencing
Total genomic DNA was extracted from silica gel-dried leaves or herbarium material using the Plant Genomic DNA Kit (Tiangen Biotech, Beijing, China) following the manufacturer’s instructions.
For each individual, five cpDNA regions (matK, rbcL, psbA–trnH and rps4 + rps4–trnS) and the nuclear gene region (LEAFY) were separately amplified with standard PCR. The matK region was amplified using primers and PCR protocols introduced by the CBoL Plant Barcoding Working Group (http://www.barcodinglife.org/index.php/Public_Primer_PrimerSearch). The rbcL region was amplified using primers 1F98 and 1351R99, following the PCR protocol described by Hasebe et al.100. The psbA–trnH region was amplified using primers psbAF and trnHR according to the protocol outlined by Chen et al.101. Amplification primers and protocols to amplify rps4 + rps4–trnS were those described by Nadot et al. Smith and Cranfill102,103. LEAFY was amplified using primers 1dF and 3dR, which were designed by Zhao48.
The PCR products were purified using PEG 8000 or the TIANgel Midi Purification Kit (Tiangen Biotech, Beijing, China) following the manufacturer’s protocol. Then they were directly sequenced in both directions using the amplification primers with an ABI PrismTM BigDye Terminator Cycle Sequencing Ready Reaction kit (Perkin Elmer, Norwalk, CT, USA). Sequences were analysed using at the ABI 3730XL automated sequencer (Applied Biosystems, Foster City, CA, USA). Cloning of samples with allelic variation in LEAFY was conducted with a pEASY-T3 Cloning Kit according to the manufactures’ protocols (Transgen Biotech), and 6 to 12 clones were sequenced for each sample. GenBank accession numbers are listed in Supplementary Table S2.
Sequences were assembled in the ContigExpress program of the Vector NTI Suite v.6.0 (Informax, North Bethesda, Maryland, U.S.A.). New combined sequences were assembled in single-region datasets that were aligned using CLUSTAL X v1.83104 and then manually adjusted in BioEdit v.7.1.11105. Phylogenetic analyses of the combinded cpDNA data set and LEAFY were performed using MP, ML and BI in PAUP * 4.0b10106, RAxMLv7.0.4107, and MrBayes v3.2.5108, respectively.
In MP analyses, all characters were equally weighted and gaps were treated as missing data. A heuristic search was performed with 1000 random addition replicates, tree bisection-reconnection (TBR) swapping and the MulTrees option in the analysis program. Bootstrap support values (BSMP), based on 1000 replicates with 10 random additions per replicate, and were used to estimate the confidence of the clades. We employed jModelTest v2.1.7109 to identify the best fitting models for ML and BI analyses. In the ML analyses, GTR + G was determined to be the best-fit model according to the Akaike information criterion (AIC) implemented in jModelTest and the BI analyses used the TIM1 + G model determined by the Bayesian information criterion (BIC) for cpDNA datasets. TrN + I + G was determined to be the best model by AIC and BIC for LEAFY. RAxML was conducted with the fast bootstrap option, using 1000 replicates. For BI analyses, four Markov chain Monte Carlo (MCMC) chains were run for 1,000,000 generations each, and were sampled every 1000 generations, starting with a random tree. The convergence of runs and estimation of burn-in were assessed using Tracer v.1.6110 and Bayesian posterior probabilities (PPBI) were calculated for the majority consensus tree of all sampled trees after discarding those sampled within the burn-in phase in MrBayes.
Divergence time estimation
Bayesian approaches were employed of cpDNA data set to estimate the divergence times of Pyrrosia s.l. in BEAST v 1.8.0111 with an uncorrelated lognormal distributed (UCLD) relaxed clock model, the GTR + G substitution model and a Yule process tree prior. The MCMC chains were run for 100,000,000 generations with sampling every 1000 generations and at least 10% burn-in phase. The tree was calibrated at the most basal node (Polypodiaceae and Davalliaceae, 60.4 Ma) obtained from divergence time estimate carried out with leptosporangiate ferns60. The effective sample size (ESS) was estimated in Tracer v.1.6 to be >200 for each parameter. The maximum clade credibility tree with median branch lengths and a 95% highest HPD interval on nodes was compiled using TreeAnnotator v.1.8.0.
Ancestral area reconstruction
Five biogeographical regions were defined: (A) eastern Asia (including central, eastern and northeastern China, Korea Peninsula, southern Japan and Far East); (B) southern Asia, the Indo-China Peninsula and southwestern and southern China; (C) Malesia (including Malaysia, Indonesia, and the Philippines); (D) Australasia (including Australia, New Guinea, New Zealand, and the South Pacific islands); and (E) Africa (including Madagascar) (Fig. 3a). The geographical distribution of Pyrrosia s.l. mostly followed Hovenkamp4 and referred to other literatures3,8,32. We carried out biogeography analyses using the DEC model112,113 implemented in RASP 3.2114. We inferred possible biogeographical scenarios across 1000 trees obtained from the BEAST analysis in the DEC analysis. A composite Akaike weight was used to summarize biogeographic reconstructions across trees115. The maximum area number was set to four because only P. lanceolata occupied four biogeographical regions.
Morphological character evolution
The data employed for the reconstruction of the evolution of morphological characters were obtained and our own observations of morphological character variation using herbarium specimens (Supplementary Table S3) and referred to the previous publications3,4,5,6,7,8,19,20,21,22,23,24,25,26. In addition, we considered observations made during fieldwork.
The evolution of morphological characters was reconstructed with likelihood using Mesquite v3.04116. We input the tree set obtained from the ML analysis based on a simple cpDNA data set (supplementary Fig. S1). In order to exhibit the morphological characters clearly in the phylogenetic tree at species level, this data set includes one individual for each species. Due to the intraspecific variation of the sequences are small, we chose one individual randomly for each species. All characters were scored as discrete binary or multistate characters and treated as unordered and equally weighted (Supplementary Table S4).
The Pteridophyte Phylogeny Group. A community-derived classification for extant lycophytes and ferns. J. Syst. Evol. 54, 563–603 (2016).
Shing, K. H. & Iwatsuki, K. On the fern genus Pyrrosia Mirbel (Polypodiaceae) in Asia and adjacentOceania: 2. J. Jpn. Bot. 72, 72–88 (1997).
Shing, K. H. A Reclassification of the FernGenus Pyrrosia. Am. Fern J. 73, 73–78 (1983).
Hovenkamp, P. H. A Monograph of the Fern Genus Pyrrosia (Polypodiaceae). (E. J. Brill/Leiden University Press, 1986).
Holttum, R. E. Flora of Malaya. Vol. II. Ferns of Malaya. (ed. Holttum, R. E.) (Government Printing Office, Singapore, 1954).
Nayar, B. K. & Chandra, S. Morphological Series within the Genus Pyrrosia, and Their Phylogenetic Interpretation. Can. J. Bot. 45, 615–634 (1967).
Nayar, B. K. Studies in Polypodiaceae IV. Drymoglossum Presl. J. India Bot. Soc. 66, 169–179 (1957).
Lin, Y. X., Zhang, X. C. & Hovenkamp, P. H. In Flora of China Vol. 3 (eds Wu, Z.Y., Raven, P H. & Hong, D. Y.) 786–786 (Science Press; Missouri Botanical Garden Press, 2013).
Wu S. H. & Ching R. C. Fern families and genera of China. 485–549 (Science Press, 1991).
Hennipman, E. The systematics of the Polypodiaceae. (eds Baas, P., Kalkman, K. & Geesink, R.) 105–120 (Springer Netherlands, 1990).
Winter, K., Osmond, C. & Hubick, K. Crassulacean acid metabolism in the shade. Studies on an epiphytic fern, Pyrrosia longifolia, and other rainforest species from Australia. Oecologia 68, 224–230 (1986).
Ong, B., Kluge, M. & Friemert, V. Crassulacean acid metabolism in the epiphytic ferns Drymoglossum piloselloides and Pyrrosia longifolia: studies on responses to environmental signals. Plant Cell Environ. 9, 547–557 (1986).
Griffiths, H., Ong, B., Avadhani, P. & Goh, C. Recycling of respiratory CO2 during crassulacean acid metabolism: alleviation of photoinhibition in Pyrrosia piloselloides. Planta 179, 115–122 (1989).
Wong, S. C. & Hew, C. S. Diffusive Resistance, Titratable Acidity, and CO2 Fixation in Two Tropical Epiphytic Ferns. Am. Fern J. 66, 121–124 (1976).
Winter, K., Wallace, B. J., Stocker, G. C. & Roksandic, Z. Crassulacean Acid Metabolism in Australian Vascular Epiphytes and Some Related Species. Oecologia 57, 129–141 (1983).
Minardi, B. D., Voytena, A. P., Santos, M. & Randi, A. M. The epiphytic fern Elaphoglossum luridum (Fee) Christ. (Dryopteridaceae) from Central and South America: morphological and physiological responses to water stress. Scientific. World J. 2014, 817892 (2014).
Griffiths, H. In Vascular Plants as Epiphytes Vol. 76 (ed. Lüttge U.) 42–86 (Springer Berlin Heidelberg, 1989).
Holtum, J. A. M. & Winter, K. Degrees of crassulacean acid metabolism in tropical epiphytic and lithophytic ferns. Aust. J. Plant Physiol. 26, 749–757 (1999).
Satijia, C. K., Bir, S. S. & Bhardwaj, A. K. Taxonomic revision of the Polypodiaceous genera of India Pyrrosia Mirbel. Bull. Bot. Surv. India 25, 62–89 (1983).
Shing, K. H. & Iwatsuki, K. On the fern genus Pyrrosia Mirbel (Polypodiaceae) in Asia and adjacentOceania: 1. J. Jpn. Bot. 72, 19–35 (1997).
Nayar, B. K. Morphology of spores and prothalli of some species of Polypodiaceae. Botanical Gazette 123, 223–232 (1962).
Kotrnon, K., Thammathaworn, A. & Chantaranothai, P. Comparative Anatomy of theGenus Pyrrosia Mirbel (Polypodiaceae) in Thailand. Nat. Hist. J. Chulalongkorn Univ. 7, 75–85 (2007).
Uffelen, G. Av Synaptospory in the fern genus Pyrrosia (Polypodiaceae). Blumea 31, 57–64 (1985).
Uffelen, G. A. V. & Hennipman, E. The Spores of Pyrrosia Mirbel (Polypodiaceae), a SEM study. Pollen Spores 27, 155–198 (1985).
Yang, L. H. A Systematic Study on the Fern Genus Pyrrosia Mirbel. Ph. D thesis, (Yunnan University, 2012).
Ching, R. C. On the genus Pyrrosia Mirbel from the Mainland of Asia including Japan and Formosa. Acta Phytotaxon. Geobot. 4, 180–181 (1935).
Giesenhagen, K. F. G. Die Farngattung Niphobolus (Verlag von Gustav Fischer 1901).
Nayar, B. K. & Chandra, S. Ferns of India, XV, Pyrrosia Mirbel. Bull. Natl. Bot. Gard. Lucknow 117, 1–98 (1965).
Copeland, E. Genera Filicum-the Genera of Ferns., 189 (The Chronica Botanica Co., Waltham, Mass., 1947).
Pichi-Sermolli, R. E. G. Tentamen Pteridophytorum genera in taxonomicum ordinem redigendi. Webbia 31, 313–512 (1977).
Ravensberg, W. & Hennipman, E. The Pyrrosia species formerly referred to Drymoglossum and Saxiglossum (Filicales, Polypodiaceae). Leiden Botanical Series 9, 281–310 (E.J. Brill/Leiden University Press, 1986).
Lin, Y. X. In Flora Reipublicae Popularis Sinicae Vol. 6(2) (ed. Delectis Florae Reipublicae Popularis Sinicae Agendae Academiae Sinicae Edita) 116–155 (Science Press, 2000).
Liu, H. M. Embracing the pteridophyte classification of Ren-Chang Ching using a generic phylogeny of Chinese ferns and lycophytes. J. Syst. Evol. 54, 307–335 (2016).
Smith, A. R. et al. A classification for extant ferns. Taxon 55, 705–731 (2006).
Zhou, X. M. et al. A plastid phylogeny and character evolution of the Old World fern genus Pyrrosia (Polypodiaceae) with the description of a new genus: Hovenkampia (Polypodiaceae). Mol. Phylogenet. Evol. 114, 271–294 (2017).
Christenhusz, M. J. & Chase, M. W. Trends and concepts in fern classification. Ann. Bot. 113, 571–594 (2014).
Kim, C., Zha, H. G., Deng, T., Sun, H. & Wu, S. G. Phylogenetic position of Kontumia (Polypodiaceae) inferred from four chloroplast DNA regions. J. Syst. Evol. 51, 154–163 (2013).
Schneider, H. et al. Gymnogrammitis dareiformis is a polygrammoid fern (Polypodiaceae) – Resolving an apparent conflict between morphological and molecular data. Plant Syst. Evol. 234, 121–136 (2002).
Haufler, C. H. et al. Systematics of the ant-fern genus Lecanopteris (Polypodiaceae): Testing phylogenetic hypotheses with DNA sequences. Syst. Bot. 28, 217–227 (2003).
Schneider, H. et al. Phylogenetic relationships of the enigmatic malesian fern Thylacopteris (Polypodiaceae, Polypodiidae). Int. J. Plant. Sci. 165, 1077–1087 (2004).
Schneider, H. et al. Unraveling the phylogeny of polygrammoid ferns (Polypodiaceae and Grammitidaceae): exploring aspects of the diversification of epiphytic plants. Mol. Phylogenet. Evol. 31, 1041–1063 (2004).
Kreier, H. P. & Schneider, H. Phylogeny and biogeography of the staghorn fern genus Platycerium (Polypodiaceae, Polypodiidae). Am. J. Bot. 93, 217–225 (2006).
Kreier, H. P. & Schneider, H. Reinstatement of Loxogramme dictyopteris, based on phylogenetic evidence, for the New Zealand endemic fern, Anarthropteris lanceolata (Polypodiaceae, Polypodiidae). Aust. Syst. Bot. 19, 309–314 (2006).
Janssen, T., Kreier, H. P. & Schneider, H. Origin and diversification of African ferns with special emphasis on Polypodiaceae. Brittonia 59, 159–181 (2007).
Schneider, H., Kreier, H. P., Hovenkamp, P. & Janssen, T. Phylogenetic relationships of the fern genus Christiopteris shed new light onto the classification and biogeography of drynarioid ferns. Bot. J. Linn. Soc. 157, 645–656 (2008).
Vasques, D. T., Ebihara, A. & Ito, M. The felt fern genus Pyrrosia Mirbel (Polypodiaceae): a new subgeneric classification with a molecular phylogenetic analysis based on three plastid markers. Acta Phytotax. Geobot. 68(2), 65–82 (2017).
Testo, W. & Sundue, M. A 4000-species dataset provides new insight into the evolution of ferns. Mol. Phylogenet. Evol. 105, 200–211 (2016).
Zhao, H. J. A Preliminary Study of Interspecific Relationships in Pyrrosia Mirbel, with Additional Reference to the Hybrid Origin of Pyr. piloselloides. Master thesis (The Institute of Botany, Chinese Academy of Sciences, 2012).
Yao, G. et al. Phylogenetic relationships, character evolution and biogeographic diversification of Pogostemon s.l. (Lamiaceae). Mol. Phylogenet. Evol. 98, 184–200 (2016).
Wang, J. J. et al. The biogeographic south-north divide of Polygonatum (Asparagaceae tribe Polygonateae) within Eastern Asia and its recent dispersals in the northern hemisphere. PloS ONE 11, e0166134 (2016).
Liu, X. Q. et al. Phylogeny of the Ampelocissus-Vitis clade in Vitaceae supports the New World origin of the grape genus. Mol. Phylogenet. Evol. 95, 217–228 (2016).
Wei, R. et al. Eurasian origin, boreotropical migration and transoceanic dispersal in the pantropical fern genus Diplazium (Athyriaceae). J. Biogeogr. 42, 1809–1819 (2015).
Kuo, L. Y. et al. Historical biogeography of the fern genus Deparia (Athyriaceae) and its relation with polyploidy. Mol. Phylogenet. Evol. 104, 123–134 (2016).
Yu, X. Q. et al. Phylogeny and historical biogeography of Isodon (Lamiaceae): rapid radiation in south-west China and Miocene overland dispersal into Africa. Mol. Phylogenet. Evol. 77, 183–194 (2014).
Nie, Z. L., Deng, T., Meng, Y., Sun, H. & Wen, J. Post-Boreotropical dispersals explain the pantropical disjunction in Paederia (Rubiaceae). Ann. Bot. 111, 873–886 (2013).
Schneider, H. et al. Ferns diversified in the shadow of angiosperms. Nature 428, 553–557 (2004).
Renner, S. S. Multiple Miocene Melastomataceae dispersal between Madagascar, Africa and India. P. Roy. Soc. Lond. B. Bio. 359, 1485–1494 (2004).
Thomas, D. C. et al. West to east dispersal and subsequent rapid diversification of the mega-diverse genus Begonia (Begoniaceae) in the Malesian archipelago. J. Biogeogr. 39, 98–113 (2012).
Janssens, S. B. et al. Evolutionary dynamics and biogeography of Musaceae reveal a correlation between the diversification of the banana family and the geological and climatic history of Southeast Asia. New Phytol. 210, 1453–1465 (2016).
Schuettpelz, E. & Pryer, K. M. Evidence for a Cenozoic radiation of ferns in an angiosperm-dominated canopy. P. Natl. Acad. Sci. USA 106, 11200–11205 (2009).
Shao, W., Yang, L. H. & Zhou, X. L. Taxonomic significance of venation pattern in Pyrrosia. Plant Sci. J. 34, 191–199 (2016).
Schneider, H. Plant morphology as the cornerstone to the integration of fossils and extant taxa in phylogenetic analyses. Species Phylo. Evol. 1, 65–71 (2006).
Chen, C. W., Kuo, L. Y., Wang, C. N. & Chiou, W. L. Development of PCR primer sets for intron 1 of the low-copy gene LEAFY in Davalliaceae. Am. J. Bot. 99, e223–225 (2012).
Wang, L. et al. Indehiscent sporangia enable the accumulation of local fern diversity at the Qinghai-Tibetan Plateau. BMC Evol. Biol. 12, 210 (2012).
Wen, D., Yu, Y., Hahn, M. W. & Nakhleh, L. Reticulate evolutionary history and extensive introgression in mosquito species revealed by phylogenetic network analysis. Mol. Ecol. 25, 2361–2372 (2016).
Lihova, J., Shimizu, K. K. & Marhold, K. Allopolyploid origin of Cardamine asarifolia (Brassicaceae): incongruence between plastid and nuclear ribosomal DNA sequences solved by a single-copy nuclear gene. Mol. Phylogenet. Evol. 39, 759–786 (2006).
Xu, B., Wu, N., Gao, X. F. & Zhang, L. B. Analysis of DNA sequences of six chloroplast and nuclear genes suggests incongruence, introgression, and incomplete lineage sorting in the evolution of Lespedeza (Fabaceae). Mol. Phylogenet. Evol. 62, 346–358 (2012).
Vijayakanth, P. & Sahaya, S. S. Pyrrosia porosa (C. Presl) Hovenkamp – A new diploid cytotype of south India from Kolli Hills of eastern Ghats, Tamil Nadu, India. Int. J. Plant. Biol. Res. 5, 1057–1058 (2017).
Pal, S. & Pal, N. Spore morphology and taxonomy of Polypodiaceae. Grana 10, 141–148 (2009).
Christensen, C. Revision of the polypodioid genera with longitudinal coenosori (Cochlidiinae and “Drymoglossinae”); with a discussion of their phylogeny. Dansk Botanisk Arkiv. 6, 3–93 (1929).
Wang, L. et al. A molecular phylogeny and a revised classification of tribe Lepisoreae (Polypodiaceae) based on an analysis of four plastid DNA regions. Bot. J. Linn. Soc. 162, 28–38 (2010).
Wei, X. P. & Zhang, X. C. Species delimitation in the fern genus Lemmaphyllum (Polypodiaceae) based on multivariate analysis of morphological variation. J. Syst. Evol. 51, 485–496 (2013).
Crayn, D. M., Winter, K. & Smith, J. A. Multiple origins of crassulacean acid metabolism and the epiphytic habit in the Neotropical family Bromeliaceae. P. Natl. Acad. Sci. USA 101, 3703–3708 (2004).
Cushman, J. C. Crassulacean acid metabolism. A plastic photosynthetic adaptation to arid environments. Plant Physiol. 127, 1439–1448 (2001).
Cockburn, W., Ting, I. & Sternberg, L. Relationships between stomatal behavior and internal carbon dioxide concentration in crassulacean acid metabolism plants. Plant physiol. 63, 1029–1032 (1979).
Testo, W. & Sundue, M. Primary hemiepiphytism in Colysis ampla (Polypodiaceae) provides new insight into the evolution of growth habit in ferns. Int. J. Plant. Sci. 175, 526–536 (2014).
Ching, R. C. The Himalaya as a centre of concentration for the south-eastern asiatic Polypodiaceous ferns. Acta Bot. Yunnanica 1, 23–31 (1979).
Kreier, H. P., Zhang, X. C., Muth, H. & Schneider, H. The microsoroid ferns: Inferring the relationships of a highly diverse lineage of Paleotropical epiphytic ferns (Polypodiaceae, Polypodiopsida). Mol. Phylogenet. Evol. 48, 1155–1167 (2008).
Wang, L., Schneider, H., Zhang, X. C. & Xiang, Q. P. The rise of the Himalaya enforced the diversification of SE Asian ferns by altering the monsoon regimes. BMC Plant Biol. 12, 210 (2012).
Morley, R. J. Interplate dispersal paths for megathermal angiosperms. Perspect. Plant Ecol. 6, 5–20 (2003).
Souza, W. M. et al. Phylogeography and evolutionary history of rodent-borne hantaviruses. Infect. Genet. Evol. 21, 198–204 (2014).
Lu, L., Wang, W., Chen, Z. & Wen, J. Phylogeny of the non-monophyletic Cayratia Juss. (Vitaceae) and implications for character evolution and biogeography. Mol. Phylogenet. Evol. 68, 502–515 (2013).
Zhou, L., Su, Y. C. F., Thomas, D. C. & Saunders, R. M. K. ‘Out-of-Africa’ dispersal of tropical floras during the Miocene climatic optimum: evidence from Uvaria (Annonaceae). J. Biogeogr. 39, 322–335 (2012).
Ali, J. R. & Aitchison, J. C. Gondwana to Asia: plate tectonics, paleogeography and the biological connectivity of the Indian sub-continent from the Middle Jurassic through latest Eocene (166–35 Ma). Earth-Sci. Rev. 88, 145–166 (2008).
Davis, C. C., Bell, C. D., Fritsch, P. W. & Mathews, S. Phylogeny of Acridocarpus-Brachylophon (Malpighiaceae): Implications for Tertiary tropical floras and afroasian biogeography. Evolution 56, 2395 (2002).
Zachos, J. C., Pagani, M., Sloan, L., Thomas, E. & Billups, K. Trends, rhythms, and aberrations in global climate 65 Ma to present. Science 292, 686–693 (2001).
Rögl, F. Palaeogeographic Considerations for mediterranean and paratethys seaways (Oligocene to Miocene). Naturhistorisches Museums Wien 99, 279–310 (1998).
Raven, P. H. & Axelrod, D. I. Angiosperm biogeography and past continental movements. Ann. Mo. Bot. Gard. 61, 539–673 (1974).
Yuan, Y. M. et al. Phylogeny and biogeography of Exacum (Gentianaceae): a disjunctive distribution in the Indian ocean basin resulted from long distance dispersal and extensive radiation. Syst. Biol. 54, 21–34 (2005).
Muller, S. et al. Intercontinental long-distance dispersal of Canellaceae from the New to the Old World revealed by a nuclear single copy gene and chloroplast loci. Mol. Phylogenet. Evol. 84, 205–219 (2015).
Warrena, B. H., Strasbergb, D., Bruggemannb, J. H., Prys-Jonesc, R. P. & Thébaud, C. Why does the biota of the Madagascar region have such a strong Asiatic flavour? Cladistics 26, 526–538 (2010).
Harrison, T. M., Copeland, P., Kidd, W. S. F. & An, Y. Raising Tibet. Science 255, 1663–1670 (1992).
Hall, R. Southeast Asia’s changing palaeogeography. Blumea 54, 148–161 (2009).
Zachos, J. C., Dickens, G. R. & Zeebe, R. E. An early Cenozoic perspective on greenhouse warming and carbon-cycle dynamics. Nature 451, 279–283 (2008).
An, Z. S., John, E. K., Warren, L. P. & Stephen, C. P. Evolution of asian monsoons and phased uplift of the Himalaya-Tibetan plateay since late Miocene times. Nature 411, 62–66 (2001).
Bonnefille, R. Cenozoic vegetation, climate changes and hominid evolution in tropical Africa. Global Planet. Change 72, 390–411 (2010).
Liu, H. M. et al. Exploring the pteridophyte flora of the eastern Afromontane biodiversity hotspot. J. Syst. Evol. 54, 691–705 (2016).
Olmstead, R. G., Michaels, H. J., Scott, K. M. & Palmer, J. D. Monophyly of the Asteridae and identification of their major lineages inferred from DNA-Sequences of rbcL. Ann. Mo. Bot. Gard. 79, 249–265 (1992).
Gastony, G. J. & Rollo, D. R. Phylogeny and generic circumscriptions of cheilanthoid ferns (Pteridaceae: Cheilanthoideae) inferred from rbcL nucleotide sequences. Am. Fern J. 85, 341–360 (1995).
Hasebe, M. et al. rbcL gene sequences provide evidence for the evolutionary lineages of leptosporangiate ferns. P. Natl. Acad. Sci. USA 91, 5730–5734 (1994).
Chen, S. L. et al. Principles for molecular identification of traditional Chinese materia medica using DNA barcoding. China J. Chinese Mater. Med. 38, 141–148 (2013).
Smith, A. R. & Cranfill, R. B. Intrafamilial relationships of the Thelypteroid Ferns (Thelypteridaceae). Am. Fern J. 92, 131–149 (2002).
Nadot, S., Bittar, G., Carter, L., Lacroix, R. & Lejeune, B. A phylogenetic analysis of monocotyledons based on the chloroplast generps4, using parsimony and a new numerical phenetics method. Mol. Phylogenet. Evol. 4, 257–282 (1995).
Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F. & Higgins, D. G. The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 25, 4876–4882 (1997).
Hall, T. A. BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucleic Acids Sym.Ser. 41, 95–98 (1999).
Swofford, D. L. PAUP: Phylogenetic Analysis Using Parsimony (*and Other Methods) version 4.0b10. (Sinauer Associates, 2003).
Stamatakis, A., Hoover, P. & Rougemont, J. A rapid bootstrap algorithm for the RAxML Web servers. Syst. Biol. 57, 758–771 (2008).
Ronquist, F. et al. MrBayes 3.2: efficient Bayesian phylogenetic inference and model choice across a large model space. Syst. Biol. 61, 539–542 (2012).
Darriba, D., Taboada, G. L., Doallo, R. & Posada, D. jModelTest 2: more models, new heuristics and parallel computing. Nat. Methods 9, 772 (2012).
Rambaut, A., Suchard, M. A., Xie, D. & Drummond, A. J. Tracer version 1.6. http://beast.bio.ed.ac.uk/Tracer (2014).
Drummond, A. J. & Rambaut, A. BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evol. Biol. 7, 214 (2007).
Ree, R. H. & Smith, S. A. Maximum likelihood inference of geographic range evolution by dispersal, local extinction, and cladogenesis. Syst. Biol. 57, 4–14 (2008).
Ree, R. H., Moore, B. R., Webb, C. O. & Donoghue, M. J. A likelihood framework for inferring the evolution of geographic range on phylogenetic trees. Evolution 59, 2299–2311 (2005).
Yu, Y., Harris, A. J., Blair, C. & He, X. RASP (Reconstruct Ancestral State in Phylogenies): a tool for historical biogeography. Mol. Phylogenet. Evol. 87, 46–49 (2015).
Beaulieu, J. M., Tank, D. C. & Donoghue, M. J. A Southern Hemisphere origin for campanulid angiosperms, with traces of the break-up of Gondwana. BMC Evol. Biol. 13, 80 (2013).
Maddison, W. & Maddison, D. Mesquite: a modular system for evolutionary analysis. Version 3.04. http://mesquiteproject.org (2015).
The authors gratefully acknowledge Dr. Bing Liu (IBCAS), Dr. Yea-Chen Liu (NCYU), Dr. Zhong-Yang Li (GNNU) and Mr. Qi Wei (Roy Garden) for providing biological materials. We are grateful to Ms. Hui-Juan Zhao (IBCAS) for providing two sequences of Platycerium bifurcatum (Cav.) C. Chr. (MF450485, MF450486). We are grateful to Dr. Yea-Chen Liu (NCYU) for his helpful suggestion. We are grateful to the editor Dr. J. Heinrichs and Dr. Xinwei Xu and three anonymous reviewers for their very valuable suggestion in improve our manuscript. This study was financially supported by the National Natural Science Foundation of China (Grant No. 81573531) and the CAMS Initiative for Innovative Medicine (CAMS-I2M) (Grant No. 2016-I2M-2-003).
The authors declare that they have no competing interests.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Wei, X., Qi, Y., Zhang, X. et al. Phylogeny, historical biogeography and characters evolution of the drought resistant fern Pyrrosia Mirbel (Polypodiaceae) inferred from plastid and nuclear markers. Sci Rep 7, 12757 (2017). https://doi.org/10.1038/s41598-017-12839-w
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.