Spore morphology and its systematic implication in Pteris (Pteridaceae)

Pteris (Pteridaceae) spores are usually trilete and can be distinguished by the perine ornamentation. The systematic value of spore morphology in Pteris is unclear, especially based on the renewed infrageneric classification of Pteris. In the present study, we used scanning electron microscopy (SEM) to understand spore characters in 57 Pteris species, one Onychium species, and two Astrolepis species; 40 species are reported here for the first time. The observed spore characters combined with published spore data, totaling 100 species from 16 sections of Pteris, were mapped onto a reconstructed phylogenetic tree. Seven characters (five proposed in previous studies), including an equatorial flange, laesural ridges, proximal ridges, distal ridges, tubercula on distal faces, coarse reticula on distal faces, and a row of extervermiculi between the distal face and equatorial flange, were analyzed to investigate spore morphology evolution in Pteris. However, the results showed no synapomorphies with other genera in Pteridaceae. Most of the characters were found to have arisen independently several times in different lineages or were even frequently reversed. Equatorial flanges and tubercula on distal faces are plesiomorphies and present in most Pteris species. Overall, the application of spore morphology in section circumscription is limited. Thus, we suggest combining spore morphology with leaf characters for Pteris infrageneric classification.


Introduction
Phylogenetic studies of the genus Pteris (Pteridaceae) have revealed its major lineages. The monophyly of the genus Pteris has been confirmed using six cpDNA and one nuclear markers, and P. platyzomopsis Christenh. & H. Schneid. (Platyzoma microphyllum R. Br.) is included in this genus [1]. Furthermore, Pteris is classified into three subgenera and 16 sections [2]. However, most leaf characters were shown to have arisen several times in different lineages or were even frequently reversed [3]. Homoplasy results show that the leaf morphological characters cannot circumscribe sections well, even when combined with the geographic distribution.
Previous studies have tried to evaluate taxonomic information on the spore morphology of Pteris, based on perine ornamentation. Tryon & Lugardon [10] determined that most Pteris spores are trilete with an equatorial flange. To our knowledge, this is the first study to classify Pteris spore morphology based on the geographic distributions of 18 Pteris species. Dai et al. [11] generalized the spore morphology of 30 Pteris species and four varieties in China; Pteris spores are classified as four groups: (1) without an equatorial flange; (2) with an equatorial flange but without proximal and distal ridges; (3) with an equatorial flange, proximal ridge, and distal ridge; and (4) with an equatorial flange and a row of extervermiculi between the distal face and equatorial flange. The results are incongruent with the previously circumscribed infrageneric classification in China: sections Campteria, Quadriauricula, and Pteris [12]. Yang et al. [13] examined 40 species and varieties of Pteris in China. They used characters similar to those used by Dai et al. [11], but classified the spores into six types and thought that most species could have several different types. Based on the studies of 25 Pteris species in Mexico and Mesoamerica, Palacios-Rios et al. [14] proposed that the commissural flange (laesural ridges) and ornamentation in Pteris spores were of taxonomic value. Ornamentation, such as tuberculate and reticulate patterns, had not been applied for spore classification in previous studies. These were only used for the description of specific species. In general, previous studies are limited by specific geographic areas or small species sample size. The systematic values of Pteris spore morphologies, especially based on the revised phylogenetic classification, have not been studied. The evolution pattern of spore morphology remains unclear.
Using scanning electron microscopy (SEM), this study was undertaken to investigate the spore morphology of the genus Pteris. To infer the systematics values of the spore characters, we reconstructed a phylogeny based on DNA sequences from previous studies and then mapped spore characters. Here, considering previous spore studies, an overview of worldwide Pteris spore evolution is provided. The taxonomic value of spore morphologies in infrageneric delimitation is discussed.

Phylogenetic analyses
A phylogenetic tree based on the cpDNA dataset, rbcL and matK, was constructed. The sequences of 180 Pteris species and 21 outgroup species from other genera of Pteridaceae were collected (S1 Table). One hundred thirty species have both rbcL and matK sequences, most from Chao et al. [3]; 50 species have only rbcL sequences from GenBank. Acrostichum aureum L. and Ceratopteris thalictroides (L.) Brongn. comprised the most phylogenetically distant outgroups [15].
Alignment was performed with ClustalW [16] as implemented in BioEdit v.7.0.5 [17], manually checked, and revised where necessary. Gaps were treated as missing characters. The resulting sequence data matrices were analyzed to infer phylogenetic relationships, using Bayesian inference. For the combination of rbcL and matK markers, the best-fitting evolutionary model for each partition was GTR+I+Γ [18] under the Akaike Information Criterion [19] in Modeltest v.3.7 [20].
Pteris phylogeny was inferred by Bayesian inference (BI) was conducted using MrBayes v.3.2.1 [21,22]. Ten million generations of four Markov chain Monte Carlo (MCMC) were run. Two independent runs were conducted to avoid getting stuck upon local optima; the temperature of the heated chains was set to 0.2. Trees were sampled every 1,000 generations, and the first 1,000 trees were discarded as burn-in. Convergence was checked by Tracer 1.6 [23]. Then, Bayesian posterior probabilities (BPP) were determined from the 50% majorityrule consensus tree of the retained trees.

Spore morphologies
In order to analysis all sections of Pteris, we examined the spores of as many species as possible, including 57 Pteris species, one Onychium species, and two Astrolepis species (60 species). For each species, one or two specimens and 30-100 spores were examined. Most of the sampled spores were taken from the DNA vouchers cited by previous samples for phylogenetic studies. Other sampled spores were from another specimen with fertile leaves because the cited DNA vouchers were unavailable or had no fertile leaves. Voucher information is provided in S1 Table. To observe detailed spore morphologies, spores were air-dried, scattered, and then mounted onto aluminum scanning electron microscope stubs. The stubs were then coated with gold in a sputter coater for 3 min. Samples were visualized via SEM; the accelerating voltage was 15 kV (TM3000; Hitachi, Tokyo, Japan). The proximal and distal views of each species were examined and recorded. The morphological terminology of spores followed Tryon & Lugardon [10] and Huang [24].

Ancestral character state reconstruction
To infer the systematics values of the spore characters, the key characters applied to the classification of Pteris spores in previous studies were investigated [10,11,13,14]. A distal ridge means a distinct prominent platform on distal face, parallel to an equatorial flange. Three laesurae are usually flanked by conspicuous ridges (laesural ridges). Because ornamentation of spores is different on proximal ridges and distal faces and most diverse on distal faces, tubercles and coarse reticula were also recorded. The various characters were defined as absent or present as follows: (1) an equatorial flange, (2) laesural ridges, (3) proximal ridges (peripheral ridges paralleled to equatorial flange), (4) distal ridges, (5) tubercula on distal faces, (6) coarse reticula on distal faces, and (7) a row of extervermiculi between the distal face and equatorial flange. Some intermediate characters were coded as ambiguous.
To reconstruct the ancestral character states of spore morphologies, previous research about Pteris morphologies from SEM were consulted [4,10,11,13,14,[25][26][27]. Sixty species from this study and 40 species from other studies described the spore characters of a total of 100 species, which cover all sections of Pteris, except section Dentatae. These multiple characters were summarized to build a character matrix containing the seven morphological characters.
Character tracing was performed using the most parsimonious reconstruction of ancestral states in Mesquite v3.40 [28]. Characters were treated as unordered, categorical variables and mapped onto the phylogenetic trees obtained from the cpDNA dataset to infer patterns of evolution.

Pteris phylogeny
Phylogenetic analyses (Bayesian inference) were performed based on the combined datasets of rbcL and matK markers. The topology obtained was similar to that in previous studies [3]; the 180 Pteris species were grouped into three major clades with strong support (BPP = 1; Figs 1 and 2), corresponding to three subgenera, Campteria, Platyzoma, and Pteris [2]. The tree topology further allowed for analysis of spore character evolution (Figs 3-5). However, the relationship implied for the three subgenera was not well supported, and the Actiniopteris +Onychium clade was included. Subgenus Campteria clusters with the Actiniopteris+Onychium clade first (BPP = 0.54), then with subgenus Pteris (BPP = 0.53), and finally with  subgenus Platyzoma (BPP = 0.6). In the studies of Zhang et al. [1,2], the Actiniopteris+Onychium clade is the closest taxa of genus Pteris (discussed below).

Spore morphology and specific character of infrageneric classification
Using SEM, spore morphologies of 57 Pteris species, one Onychium species, and two Astrolepis species were examined; 40 species of them are reported here for the first time. Proximal and distal views of each species were presented (Figs 6-13) based on Zhang & Zhang's infrageneric classification [2]. The section Campteria has the most species in this study, and they were divided into four parts for further analysis of spore morphology. Spores of most Pteris species are tetrahedral and trilete. The proximal view is usually triangular with rounded corners. Some abnormal spores are found, such as monolete spores of P. biaurita L. (Fig 8K and 8L), and tetralete spores of P. biaurita var. walkeriana Fraser-Jenk. & Rajkumar (Fig 7I and 7J). In total, the spores of 100 species were analyzed for infrageneric delimitation, including 60 species reported in the present study and 40 species from previous studies (Figs 1 and 2, S1 Table).
Spore morphologies of species within a section are more or less similar but not consistent for all species within a section. A summarized overview of common morphology of each taxon mapped the phylogenetic tree is given in Fig  The spore sizes in Pteris, based on equatorial diameter, range from 35-65 μm. The smallest spores are about 35-40 μm, such as in P. longipinna Hayata (Fig 10C and 10D) and P. deltodon Baker (Fig 11L and 11M). Most of spores are smaller than 60 μm in equatorial diameter. Few species have spores larger than 60 μm in equatorial diameter, only P. normalis D. Don (Fig 9L  and 9M) and P. kathmanduensis Fraser-Jenk. & T.G. Walker (Fig 9N and 9O).

Spore character evolution
Seven selected characters were coded (Figs 1 and 2), and ancestral character state evolution combining the phylogeny and the morphology matrix was performed to infer character transitions of Pteris spores (Figs 3-5). Of the seven selected characters, none were synapomorphic for Pteris with respect to other genera in Pteridaceae. All the characters arose and reversed several times in the different lineages.
Laesural ridges present distinctly in some species. However, many species have ambiguous morphology between laesural ridges or just simple prominences on the proximal face and are difficult to code, i.e., some simple prominences could partially fuse into laesural ridges and independent prominences on proximal faces. Those species were coded as polymorphic. Because of those species with polymorphy, the character of laesural ridges does not seem stable in Pteris, and evolved many times (30 steps, Fig 3B). For the species with distinctly laesural ridges, the character is a shared character of species in the section Hypsopodium (Fig 11A, 11C and 11E), section Litobrochia (Fig 12E and 12G), and section Tripedipteris (Fig 12J, 12L, 12N and 12P).
Only the spores with equatorial flanges could have both proximal ridges as well as distal ridges. The species with proximal ridges parallel to the equatorial flange are scattered in different clades (19 steps, Fig 3C). However, the presence of a proximal ridge is a synapomorphy for section Pteris (P. longifolia L., Fig 13C) and the Actiniopteris+Onychium clade (Fig 13I). A distal ridge can be found in most of the sections in the genus Pteris (23 steps, Fig 4D). It is a synapomorphy for section Campteria, section Hypsopodium (Fig 11B, 11D and 11F), and section Litobrochia (Fig 12F, 12H and 12I).
The presence of coarse reticula on distal faces is the character most like a synapomorphy in genus Pteris, while the most phylogenetically close genera Actiniopteris and Onychium are included (15 steps, Fig 4F). The characters occur in basal groups, section Moluccanae, section Pteris, P. platyzomopsis, and Actiniopteris+Onychium clade; and are absent in subgenus Campteria. The character reverses back to be present in several species, especially in section Creticae. Most spores with coarse reticula also have tubercula, including P. actiniopteroides Ching ( Fig  11J and 11K), P. deltodon (Fig 11L and 11M), P. longifolia (Fig 13C and 13D), P. moluccana (Fig 13E and 13F), P. papuana (Fig 13G and 13H), and Onychium siliculosum (Desv.) C.Chr. (Fig 12I and 12J). The presence or absence of one row of extervermiculi between a distal face and an equatorial flange is difficult to code in some species. It could be varied among different spores within species. For the species with one distinct row of extervermiculi, the character is a synapomorphy for sections Campteria and Hypsopodium (Fig 11B and 11E). Pteris spore morphology and its systematic implication

Evolution of Pteris spore morphology
This study examined spores of most species of Pteris and, to our knowledge, is the only one to compare and examine the phylogeny of Pteris in terms of phylogenetic patterns of spore morphology. Equatorial flanges and tubercula on distal faces are shared characters of the genus Pteris, but are plesiomorphies. Equatorial flanges, tuberculate distal faces, and reticulate distal faces show the fewest transitions among the seven selected characters. Equatorial flanges and tuberculate distal faces (rugate surface) are the primary features for the fossil spores, especially Polypodaceoisporites, to be allied with Pteris [10,29]. Our results revealed that the presence of equatorial flanges and tuberculate distal faces are plesiomorphic for Pteris and supported the inference about fossil spores.
The presence of laesural ridges, proposed by Palacios-Rios et al. [14], is currently relatively underestimated and synapomorphic in some sections. Most characters change frequently in different lineages, including proximal ridges, distal ridges, tubercula on distal faces, coarse reticula on distal faces, and a row of extervermiculi between the distal face. In bolbitidoid ferns (Dryopteridaceae), the microstructure of perines also reversed several times [8]. It has been reported that similar perine morphologies of different species could have different ontogenies [7]. The resemblances of spores from different species could arise from convergence or parallel evolution, but the factors to determine spore morphological traits have not been revealed [30].
Here, we consider possible environmental factors. In Pteris, there is a transition of habitat preference from the terrestrial to lithophytic habitats of section Cretica [3], which is more or less consistent with the presence of the coarse reticula on distal faces in this section. Interestingly, several lithophytic Pteris species in other sections also possess coarse reticula on distal faces, such as P. longifolia and P. vittata. We also compared the geographic distribution [3] with the spore morphology, but no distinct pattern was found. The possible relationship between the coarse reticula on distal faces and lithophytic species needs to be further studied.

Pteris spore morphology for infrageneric classification
The equatorial flanges, reticulate distal faces, and laesural ridges corresponded more to section classification. However, most characters reversed frequently. Mostly, spore morphology is useful at lower taxonomic levels, i.e., species delimitation, rather than section circumscription. Even combining characters did not help characterize most sections of Pteris, except sections Moluccanae and Pteris. Synapomorphies of section Moluccanae are coarse reticula on distal face and the absence of an equatorial flange. Synapomorphies of section Pteris are proximal ridges and reticulate distal faces. The results in this study showed that the inconsistency of spore combining characters with infrageneric classification reported in a previous study [11] are arisen by multiple origins of spore characters, and not only arisen by the traditionally circumscribed infrageneric classification. In general, each section of Pteris has its own specific spore morphology, but species exceptions were found in many sections.
Spore characters, similar to leaf morphologies, reversed several times, but the combination of both characters could be useful. For example, section Semipinnatae has pectinate basiscopic pinnae and entire acroscopic pinnae, features that separate it from sections Cadierii and Hypsopodium [1]. However, the characters can also be found in P. dimorpha Copel. and P. hainanensis in section Cadierii. If considering the spore characters, the presence of equatorial Pteris spore morphology and its systematic implication flanges in section Semipinnatae can help differentiate P. dimorpha and P. hainanensis, which both lack equatorial flanges.
For the different phylogenetic positions of sections Dentatae, Denticulatae, Litobrochia, and Tripedipteris, the evolution of spore characters is considered. The topology is (((Creticae, Mutilatae), (Dentatae, Denticulatae), Litobrochia), Tripedipteris) in this study, but was (((((Creticae, Mutilatae), Dentatae), Litobrochia), Tripedipteris), Denticulatae) in Zhang and Zhang's study [2]. The spore morphologies of sections Denticulatae, Litobrochia, and Tripedipteris (sections Dentatae unsampled in this study) are different in the morphologies of their laesural (Fig 3B) and distal ridges (Fig 4D). Spores of section Denticulatae have no laesural ridges or distal ridges (Fig 2) and are more similar to sections Mutilatae. If section Denticulatae is the most basal one among the three sections, as in the previous tree [2], the evolutionary transitions of laesural ridges will increase, and the transitions of distal ridges will be constant.

Spore morphology of Pteris and related genera
In this study, the relationships of subgenera Campteria, Platyzoma, and Pteris were not well resolved, including separation of the Actiniopteris+Onychium clade. The relationship of Pteris, Onychium, and Actiniopteris is different from that in the study by Zhang et al. [1,2]; the genus Pteris is monophyletic. Onychium and Actiniopteris are thought to be the closest taxa to the genus Pteris [2,3]. Comparing spore morphologies of P. platyzomopsis (Platyzoma microphyllum in [10,31]), Onychium and Actiniopteris are more similar to Pteris species, especially P. longifolia and P. vittata. P. platyzomopsis is more similar to Ceratopteris richardii [10].
Most of the taxa in the genus Pteris are diploids, triploids, or tetraploids; there are few pentaploids and hexaploids [32]. The spore sizes are similar to that in a previous study of the genus Pteris [10]. Comparing the spore sizes and ploidy levels of some species, the ploidies of different species could be inferred. It is found that sexually diploid species, such as P. longipinna and P. deltodon, have spore sizes of 35-40 μm. However, sexual tetraploids, such as P. multifida and P. umbrosa R. Br., have similar spore sizes to sexual diploids. Furthermore, apogamous species have varied spore sizes, and there is no clear correlation with ploidy levels. Species with 40-50 μm spores could be apogamous diploid, such as P. laurisilvicola and P. oshimensis, or apogamous triploid, such as P. actiniopteroides and P. incurvata Y.S. Chao, H.Y. Liu & W.L.Chiou [33]. The correlation among spore sizes, ploidy levels, and reproductive traits is unclear.

Conclusions
For spores of Pteris, the results of this study not only report the morphology but also clarify the use of spore characters in systematic studies of Pteris. Spore morphology can circumscribe some sections, with some exceptions. However, because all of the characters occur with reversals, specific synapomorphies are difficult to discern. The application of spore morphology for section circumscription is limited.
Supporting information S1 Table. Voucher specimens of spore samples examined for this study and GenBank accession numbers for DNA sequences used for the phylogenetic tree construction in this study. The information presented here includes species, spore morphologies, GenBank numbers for rbcL, and matK, voucher, herbarium, and if the spore sample and DNA sequences are from the same voucher specimen. Spore morphologies from previously published research were recorded and cited. (XLSX)