Biogeography of the cosmopolitan terrestrial diatom Hantzschia amphioxys sensu lato based on molecular and morphological data

Until now, the reported diversity of representatives from the genus Hantzschia inhabiting soils from different parts of Eurasia was limited to the few species H. amphioxys, H. elongata and H. vivax and some of their infraspecific taxa. We have studied the morphology, ultrastructure and phylogeny of 25 soil diatom strains, which according to published description would be assigned to “H. amphioxys sensu lato” using 18S rDNA, 28S rDNA and rbcL. We show that strains are made up of seven different species of Hantzschia, including five new for science. Five strains were identified as H. abundans. This species has a slight curvature of the raphe near its external proximal ends. Four of the examined strains were represented by different populations of H. amphioxys and their morphological characteristics fully correspond with its isolectotype and epitype. The main specific features of this species include 21–25 striae in 10 μm, 6–11 fibulae in 10 μm, 40–50 areolae in 10 μm and internal proximal raphe endings bent in opposite directions. H. attractiva sp. nov., H. belgica sp. nov., H. parva sp. nov., H. pseudomongolica sp. nov. and H. stepposa sp. nov. were described based on differences in the shape of the valves, significant differences in dimensions, a lower number of striae and areolae in 10 μm and the degree and direction of deflection of the internal central raphe endings. Based on the study of the morphological variability and phylogeny of soil Hantzschia-species from different geographical locations we conclude that while some species such as H. amphioxys are truly cosmopolitan in their distributions, some sympatric populations of pseudocryptic taxa exist in the Holarctic.

Diatoms are one of the most widespread group of unicellular algae, being reported from all types of water bodies and from all continents 1 . Currently, two hypotheses about freshwater diatom distribution are being discussed; one suggesting that all or most diatom species have a cosmopolitan distribution 2,3 and the second that restricted geographic distributions and even endemicity are more common 4,5 . On the one hand, capacity for dispersal, wide environmental tolerances and interbreeding between morphologically distinct units should result in wide, perhaps ubiquitous distributions [6][7][8] . On the other hand, current careful investigation of taxonomy and distribution of diatoms 5,[9][10][11][12] and other microorganisms 13,14 have shown that unicellular organisms can be specific to restricted areas. Due to the long history of references that Hantzschia amphioxys (Ehrenberg) Grunow has a cosmopolitan distribution [15][16][17] , an investigation of the global distribution of this species is an important tool for our understanding of global diatom biogeography.
Until recently, the genus Hantzschia Grunow included less than 50 species known from different parts of the world and from different water ecosystems like freshwater and saline waterbodies, terrestrial and soil ecosystems 1 . Many new species from this genus were described during the last decade. Over 250 described taxa are currently known from the genus Hantzschia 18 . Main morphological features that are used for species delimitation are

Results and discussion
In most of the forest soil samples used in this survey, specimens belonging to the genus Hantzschia are quite common. Based molecular as well as on light microscopy (LM) and scanning electron microscopy (SEM) observations of 25 strains, seven different taxa were recognized. Figure 1 contains the locations of the strain's habitats. In anticipation of the nomenclatural consequences, we are using the new names already here but will describe them formally later. Molecular data. The obtained phylogenetic tree for representatives of the different strains Hantzschia contains several large clades, some of which are monophyletic, while others contain several different species names (Fig. 2). In the analyzed tree, the largest clade is represented by different strains of H. amphioxys, the structure of which is described in the corresponding molecular analysis section. At the same time, the most significant is that in the same clade there is strain H. amphioxys D27_008, which has been designated as epitype 20 . One of the largest is the clade with H. abundans, which, in addition to our strains, and some that have already been published, includes the group of strains referred to as "Hantzschia sp. 3" (Sterre6)e, (Sterre6)f from Souffreau et al. 16 . We propose to refer to all of these strains as H. abundans. The next clade consists of the new species of H. attractiva and three strains of Hantzschia sp. 2 (Mo1)a, (Mo1)e, (Mo1)m from Souffreau et al. 16 , the latter we propose to merge into the new species named H. pseudomongolica, which is sister to H. attractiva. Given the topology of the tree and the morphological features of the representatives, we can conclude that there is a close relation between H. abundans and H. attractiva plus H. pseudomongolica. A separate group consists of two clades with sufficient statistical support (likelihood bootstrap, LB 76; posterior probability, PP 100), one of which is represented by two strains of H. parva, and the other with strains of H. cf. amphioxys (Sterre1)f, (Sterre1)h. Another large clade represents a set of strains of Hantzschia sp. 1 and Hantzschia sp. 2 (Mo1)h, (Mo1)l from Souffreau et al. 16 16 ) and H. stepposa. Interesting is the position of the H. abundans (Tor3)c strain, which is very distant from other representatives of H. abundans and probably is a cryptic taxon, whose taxonomic status needs to be revised.
The absence of nucleotide sequences of the 18S rDNA gene for the Hantzschia strains discussed in Souffreau et al. 16 did not allow us to carry out phylogenetic analysis based on three genes (18S rDNA, 28S rDNA and rbcL). However, in GenBank, 18S rDNA sequences are available for other Hantzschia strains, including strain D27_008, for which the nucleotide sequence of the 18S V4 rDNA region is known (NCBI Accession FR873247). Comparison of this sequence with similar ones in our strains shows a similarity of 98.4% (368 bp) with H. amphioxys MZ-BH9, MZ-BH14 and MZ-BH15. At the same time, the 18S V4 rDNA regions of these three strains are identical and they differ from the epitype by four transitions and one transversion. The similarity with the MZ-BH8 strain is somewhat less with 98.1%, due to one transition (transitions), in which it differs both from the rest of our strains and from the epitype. Also, for the 18S V4 rDNA region of the H. amphioxys D27_008 strain, compared with all of our H. amphioxys strains, a specific feature is characteristic-the position of one nucleotide (insertion of thymine T). The level of similarity of 98.1-98.4% significantly exceeds the previously calculated rate of 97% for 399/387 bp 20 between the strain H. amphioxys D27_008 and the strain H. amphioxys var. major A4 (NCBI Accession HQ912404) from Ruck and Theriot 42 .
The studied Hantzschia strains presented low similarities between the clades indicated in the Fig. 2

Morphological comparisons.
In order to obtain an overview of the diversity and distribution of the diatoms with Hantzschia amphioxys-like morphology, we took soil samples for diatom analysis from steppe, meadow and forest biogeocenoses of Eurasia. In total, we studied 20 soil samples and from these thirteen strains of Hantzschia were isolated. H. attractiva sp. nov., H. belgica sp. nov., H. parva sp. nov., H. pseudomongolica sp. nov. and H. stepposa sp. nov. were described based on differences in the shape of the valves, significant differ-   (15)(16)(17)(18)(19)(20) and one-sided deviation of the internal proximal raphe ends (Table 1). A soil isolate from a wet meadow dominated by Phalaridetum, which was located 30 m from the river Nidder, Windecken, Hessen, Germany, was chosen as holotype. In general, our strains correspond to the diagnosis of the H. abundans species according to all major morphological characteristics. An exception, however, is the slightly smaller valve length in strain MZ-BH6 (37.5-39 µm, in diagnosis-not less than 40 µm), a smaller valve width in strains MZ-BH1 and MZ-BH6 (6-7 µm, in diagnosis-7-10 µm). We also noted an increase in areolae density up to 50 in 10 µm (40 in 10 µm in diagnosis), although close values (45 in 10 µm) were noted by Zidarova et al. 19 for the population of H. abundans from Byers Peninsula, South Shetland Islands. We have not noted the formation of the keel near the raphe from the outer side of the valve in our strains, which, according to Lange-Bertalot 28 , could indicate an excess of silica in the substrate. From the outer side of the valve in all the studied strains, we noted a slight curvature of the raphe near its central endings, which was not previously noted (Fig. 3m). This www.nature.com/scientificreports/ feature is more characteristic for different strains of H. amphioxys. According to morphological characteristics such as the width and length of the valve, the number of striae and fibulae 10 μm (supplementary Table S2 in Souffreau et al. 16 ), as well as the position in the phylogenetic tree rbcL-28S rDNA (Fig. 2), we suggest to include the Hantzschia sp. 3 strains (Sterre6)e and (Sterre6)f from Souffreau et al. 16 to H. abundans. H. attractiva sp. nov. is visually and parametrically similar with some Hantzschia species but differs in a number of ways. In terms of the ratio of the length and width of the valve, the shape of its ends and concavity in the center of the ventral side of the valve, H. attractiva is most similar to H. subrupestris Lange-Bertalot 28 . However, these species can be distinguished from one another by a smaller number of striae (14-16 in 10 μm) and areolae (24 in 10 μm) in H. subrupestris (Table 1). In the case of H. subrupestris, the central raphe ends are straight or barely noticeably deflected in one direction, whereas in H. attractiva they deviate in one direction very clearly. Compared with H. abundans, the diagnostic features of H. attractiva include: less pronounced concavity of the middle part of the valve from the ventral side, vertical position of the valve ends, which are also less pronounced; www.nature.com/scientificreports/ a larger number of areolae in 10 μm and one-sided deviation of the central raphe ends on both sides of the valve (they are located straight in H. abundans). Our species differs from H. bardii Lange-Bertalot, Cavacini, Tagliaventi et Alfinito in the shape of the valve ends (in H. bardii, they are obliquely cuneate and finally protracted narrow-rostrate to subcapitate), a smaller number of striae in 10 µm and a larger number of areolae 40 . Of nearly similar size to H. attractiva is H. compactoides Lange-Bertalot, Cavacini, Tagliaventi et Alfinito, which can be distinguished by a more pronounced bend of the valve from the ventral side, the direction of the valve ends in the same direction and their rostrate shape 40  Li, which differs from H. parva, in addition to the narrower shape of the valve and tapering valve ends, by a smaller width (up to 4.5 μm), more dense striation (34-35 striae in 10 μm) and a higher fibular density (11-14 in 10 μm) 45 . In spite of the close dimensions of the valve, described from Uruguay H. delicatula Metzeltin, Lange-Bertalot et García-Rodríguez 43 has an almost flat ventral side of the valve and its ends are in almost the same plane as the cuneate, as well as lower density of areolae (30 to 10 μm), which distinguishes it from H. parva (Table 1). The greater width of the valve (9-10 µm), as well as its flattened form on the ventral side, distinguish H. parva from another Uruguayan species H. valdeventricosa Metzeltin, Lange-Bertalot et García-Rodríguez 43 .
According to its morphometric features, H. stepposa sp. nov. is similar to H. amphioxys. Also, these two species are similar in the deflection of the central raphe ends on the inner side of valve-they curve in opposite directions. Despite this, the two species are easily distinguished by the shape of the valves: H. amphioxys is characterized by a pronounced concavity of the ventral side of the valve in the middle, whereas in H. stepposa, the ventral side is almost straight; in H. amphioxys, the valve ends are clearly directed to the dorsal side of the valve, while in H. stepposa they are arranged vertically. The presence of central raphe ends at the inner side that are deflected in opposite directions distinguishes H. stepposa from most of the known Hantzschia species, whose valve length do not exceed 100 μm (including H. attractiva and H. parva). The only species with a similar position of the central ends of the raphe and up to 100 µm in length is H. giessiana 28 . However, along with the large size of the valve (50-100 μm), it is also characterized by a completely different shape of the valves-a strongly concave ventral side and ends curved towards it (Table 1).
Originally strain (Sterre3)a was presented in Souffreau et al. 16 as "H. cf. amphioxys", however, the authors presented only the morphometric characteristics and the light micrograph of one valve without clarifying its taxonomic status and explaining its isolated position in the phylogenetic tree in relation to the Hantzschia amphioxys clade (Fig. 4 in Souffreau et al. 16 ). We present light and scanning electron micrographs based on a study of the original material with description of a new species H. belgica sp. nov. In comparison with other Hantzschia species that have internal proximal raphe ends that are deflected in opposite direction (H. amphioxys and H. stepposa), H. belgica differs in shape and direction of the valve ends. If H. amphioxys and H. stepposa are characterized by extended and capitated ends, which are directed to the dorsal side of the valve in H. amphioxys and arranged vertically in H. stepposa, in H. belgica the ends of the valves are not distinctly protracted and not directed to either the dorsal or ventral margins.
A comparative analysis of the main morphological characteristics (length and width of the valves, the number of striae, the number of fibulae and areolae in 10 μm) in the four H. amphioxys strains isolated by us with the proposed diagnoses of Jahn et al. 20 of the lectotype and epitype show a coincidence in most data. Small differences can be attributed to the slightly larger maximum length of the valves in strain MZ-BH14 (46.0 μm with a maximum of 44.6 μm in the epitype). The density of the areolae for epitype is in the range of 48-53 areolae in 10 µm (submitted in this study). Some studies indicate the number of areolae in 10 µm for different populations of H. amphioxys in the range of 40-50 areolae in 10 µm 19,28 . In this regard, the excess of this range is observed only in the strain MZ-BH14, whose density of areolae in the valve can reach 60 in 10 µm. www.nature.com/scientificreports/ Based on a detailed study of morphology and ultrastructure, as well as the use of phylogenetic analysis using the 28S rDNA and rbcL genes, our results found 25 soil strains that represent geographically distant populations and formally could be classified as H. amphioxys sensu lato, containing seven different Hantzschia species, including five new for science. Five strains were identified as H. abundans and are fully consistent with the diagnosis of all major morphological characteristics for that taxon. For these strains, we noted areolar densities up to 50 in 10 μm (the original diagnosis reported 40 areolae in 10 μm). Externally, in all the studied strains, we noted a slight curvature of the raphe near its proximal endings, which was not previously reported. In addition, another four of the studied strains represent different populations of H. amphioxys and their morphological characteristics fully correspond to the adopted isolectotype and epitype 20 . The main specific features of these types include: the presence of 21-25 striae in 10 µm, 6-11 fibula in 10 µm, 40-50 areolae in 10 µm and internal proximal raphe ends curved in opposite directions (Table 1). Five new species were identified on the basis of differences in the shape of valves, the size characteristics of the valves, the number of striae and areolae in 10 μm and the position of the central raphe ends compared to the already known Hantzschia species.
Phylogenetic analysis using the 28S rDNA and rbcL genes showed that adding our strains does not change the tree topology obtained by C. Souffreaua et al. 16 for different lines of H. amphioxys. At the same time, the presence of a detailed morphological description of our strains using SEM results made it possible to distinguish morphological differences between the available clades. The connection of the topology of the obtained tree with the morphological features of the isolated strains shows that morphological features reflect phylogenetic relatedness. These morphological features that also reflect relationships based on genomic data are shape of the valve, the raphe structure (presence or absence of the central nodule and direction of the central branches relative to each other in it), the number of striae and areolae in 10 µm. These morphological features were correlated with the clades in the ML (Maximum Likelihood) and BI (Bayesian inference) trees constructed from fragments of the 28S rDNA and rbcL genes. Ultrastructure. In outside view central raphe ends are straight or slightly curved to the same side (Fig. 3m). In inside view central raphe ends are curved to dorsal side (Fig. 3n). Striae are uniseriate, 40-50 in 10 µm. Helictoglossae are small. Molecular analysis. In the phylogenetic tree based on the 28S rDNA and rbcL genes, H. abundans strains form a separate clade (Fig. 2). In addition to our 5 strains, we also included two other strains of H. abundans, (Wes1)b and (Wes3)b, that were isolated from De Panne, Belgium (Souffreau et al. 16 ). This clade also includes soil strains (Sterre6)e and (Sterre6)f, named by Souffreau et al. 16 as "Hantzschia sp. 3", isolated from Gent, Belgium.
Etymology. The species is named according to the region where it was found and indicating superficial similarities with H. mongolica.
Hantzschia parva Maltsev et Kulikovskiy sp. nov. (Fig. 6). Holotype. Slide no. 02509/MZ-BH4 (Holotype here represented by Fig. 6a-g)     Ecology and distribution. In the course of this research, two H. parva soil strains were isolated from geographically distant biogeocenoses (Fig. 1, Table 2). The strain H. parva MZ-BH4 (holotype) was isolated from the soil in the broadleaf forest biogeocenosis. The strain H. parva MZ-BH3 (paratype) BH2 was isolated from the soil of the steppe biogeocenosis at the city cemetery. The dominant grass cover consisted of Elytrigia repens (L.) Desv. ex Nevski, Hordeum leporinum Link and other grasses.
Etymology. The species named according to the region where it was found.
Molecular analysis. In the phylogenetic tree based on the 28S rDNA and rbcL genes, H. belgica (Sterre3)a occupies a separate position, sister to H. amphioxys plus H. stepposa clade (Fig. 2). These three taxa are closely related, with high statistical support (LB 100; PP 100) and represent a group of Hantzschia species, in which the central raphe ends on the inner side of the valve and are deflected in opposite directions.
Ultrastructure. Raphe not continuous. In outside view central raphe ends slightly curved to the same side (Fig. 9r). In inside view central raphe endings distinctly curved in opposite directions (Fig. 9s). Number of areolae are variable in different strains: from 40-50 in MZ-BH9 to 50-60 in MZ-BH14. Areolae range from 40 to 60 in 10 µm between the strains investigated.
Molecular analysis. In the phylogenetic tree, all the H. amphioxys strains isolated by us, as well as the sequences taken from BOLDSYSTEMS, form a separate clade (Fig. 2). This clade has several prominent subclades, formed with high statistical support. The first subclade (SCI) is represented by two strains of H. amphioxys (Sterre1)e and www.nature.com/scientificreports/ (Sterre3)c, isolated from soil in the Gent (Belgium) region 16 . The second subclade (SCII), which is sister to SCI, is formed by three strains, also isolated from Gent (Belgium), which are 2-5 µm shorter than the valves compared to the first group, one of our strains, MZ-BH8 and the epitype H. amphioxys D27_008. A separate monophyletic group within H. amphioxys has the other two subclades. The third subclade (SCIII) consists of three strains of H. amphioxys isolated from Schirmacher Oasis (Antarctica) 16   www.nature.com/scientificreports/ includes two Belgian strains (Sterre1)e and (Sterre3)c; SCII subclade is formed by three Belgian strains (Sterre4) a (Sterre4)b, (Sterre6)b, one Berlin strain D27_008 and strain MZ-BH8 from the Moscow Region (Russia); SCIII subclade consists of three Antarctic strains SCHIR S11, SCHIR S15, SCHIR S16; SCIV subclade is represented by our three strains: MZ-BH9 (Kursk region, Russia), MZ-BH14 and MZ-BH15 (Zaporizhia region, Ukraine). Due to the absence of distinct differences in morphology and ultrastructure between representatives of the subclade SCI-SCIV (Fig. 2) we don't distinguish new species which may correspond to the separate monophyletic subclades within H. amphioxys. Analysis of the distribution of the studied strains showed the presence of two trends in the biogeography of Hantzschia species. The first one shows that within the genus there are a number of widely distributed species; the second one shows that there are also species with a probably limited distribution (see also for Planothidium by Jahn et al. 12 ). Neither H. abundans, nor H. amphioxys had strains that are more closely related to local endemics, and given the habitats confirmed in our study, they are species with wide distribution. Also, H. parva can be considered as widely distributed, two strains of which were isolated from habitats distant more than 400 km apart. The second trend may indicate the presence of species with a limited distribution: H. stepposa, H. abundans (Tor3)c, and two strains of H. cf. amphioxys (Sterre1)f, (Sterre1)h. However, the question of endemism of these species has to be rethought after revising the characteristics of the morphology and ultrastructure of   46 . Today, there is no single and universal morphological or ultrastructural feature that would characterize the genus Hantzschia and the most effective way to verify membership to this genus is with a molecular phylogeny 47 . Considering this, the data obtained here for 13 new soil strains of H. amphioxys and 6 related species not only significantly complement the previously obtained material 16 , but also represent a significant step in understanding the diversity and biogeography of cryptic and pseudocryptic taxa in the genus Hantzschia.

Materials and methods
Diatom samples of Hantzschia were collected in 2014-2015 from different soil and forest floor horizons in the territory of steppe, meadow and forest biogeocenoses of Eurasia (Fig. 1, Table 2). The forest floor samples were taken at a distance of 1.0-1.5 m from tree trunks in the areas without large branches and accumulated bark from the A01 (fresh debris), A02 (0-5 cm, decomposed debris with partially preserved initial structure of some  www.nature.com/scientificreports/ components), and A03 (5-10 cm, strongly decomposed debris) sub-horizons. In total, we studied 20 soil samples, and from these 13 strains of Hantzschia amphioxys sensu lato were isolated. Samples and strains for light microscopy and scanning electron microscopy investigations were processed by means of a standard procedure involving treatment with 10% HCl and concentrated H 2 O 2 . The material was washed with distilled water. Permanent diatom preparations were mounted in Naphrax. LM observations of the natural population and monocultures were performed using a Zeiss Axio Scope A1 microscope (Germany) equipped with oil immersion objective (× 100/n.a 1.4, DIC). The ultrastructure of the valve was examined with the JSM-6510LV (Japan) scanning electron microscope 49 . Cell measurements were based on at least 50 individuals and expressed as minimum and maximum values in the taxonomic description.
The sample selected and designated as lectotype was from the Labrador Peninsula (sample Nr. 1780) stored in the Ehrenberg collection, Museum für Naturkunde, Leibniz-Institut für Biodiversitäts-und Evolutionsforschung an der Humboldt Universität zu Berlin (BHUPM) with the mica-reference number 250502b blue. Additionally, in the Algae Collection, Botanischer Garten und Botanisches Museum Berlin-Dahlem, Freie Universität Berlin (B) an isolectotype (B 40 0040891) was selected, and the strain D27_008 was designated as epitype with its V4 domain of 18S rDNA already published 20 ; rbcL and 28S rDNA sequences of Hantzschia amphioxys strain D27_008 have been submitted to NCBI in this study (for methods of extraction see Abarca et al. 50 ).
Total DNA of monoclonal cultures was extracted using InstaGene Matrix according to the manufacturer's protocol. Fragments of 18S rDNA (380 bp, including V4 domain), partial 28S rDNA gene including the D1-D3 region (753 bp) and partial rbcL plastid gene (1032 bp) were amplified using primers from Zimmerman et al. 51 for 18S rDNA fragments, from Jo et al. 52 for 28S rDNA fragments and from Ruck and Theriot 42 for rbcL fragments. The rbcL and 28S rDNA sequences were used for phylogeny reconstruction. V4 domains were used for the step-by-step comparison of the 18S rDNA sequences of our strains and other Hantzschia strains.
Amplifications of the 18S rDNA fragments, partial 28S rDNA and rbcL genes fragments were carried out using the premade mix ScreenMix (Evrogen, Russia) for the polymerase chain reaction (PCR). The conditions of amplification for 18S rDNA fragments were: an initial denaturation of 5 min at 95 °C, followed by 35 cycles at 94 °C for denaturation (30 s), 52 °C for annealing (30 s) and 72 °C for extension (50 s), and a final extension of 10 min at 72 °C 49 . The conditions of amplification for partial 28S rDNA were: an initial denaturation of 5 min at 95 °C, followed by 45 cycles at 94 °C for denaturation (30 s), 62 °C for annealing (30 s) and 72 °C for extension (80 s), and a final extension of 10 min at 72 °C. The conditions of amplification for partial rbcL were: an initial denaturation of 5 min at 95 °C, followed by 45 cycles at 94 °C for denaturation (30 s), 60 °C for annealing (30 s) and 72 °C for extension (80 s), and a final extension of 10 min at 72 °C 49 .
The resulting amplicons were visualized by horizontal agarose gel electrophoresis (1.5%), colored with SYBR Safe (Life Technologies, United States). Purification of PCR products was carried out with a mixture of FastAP, 10 × FastAP Buffer, Exonuclease I (Thermo Fisher Scientific, USA) and water. 18S rDNA fragments and partial 28S rDNA and rbcL genes were decoded from two sides using forward and reverse PCR primers and the Big Dye system (Applied Biosystems, USA), followed by electrophoresis using a Genetic Analyzer 3500 sequencer (Applied Biosystems, USA) 49 .
Editing and assembling of the consensus sequences were carried out by comparing the direct and reverse chromatograms using the Ridom TraceEdit program (ver. 1.1.0) and Mega7 53 . Newly determined sequences and DNA fragments from 27 other Hantzschia species, which were downloaded from GenBank and BOLD (taxa, strain numbers and Accession Numbers are given in Supplementary Table S3), were included in the alignments.
The nucleotide sequences of the 28S rDNA and rbcL genes were aligned separately using the Mafft v7 software and the E-INS-i model 54 . For the protein-coding sequences of the rbcL gene, we checked that the beginning of the aligned matrix corresponded to the first position of the codon (triplet). The resulting alignments had lengths of 752 (28S rDNA) and 1033 (rbcL) characters and can be found in the Supplementary Alignment S4.
The data set was analyzed using Bayesian inference method with posterior probability implemented in Beast ver. 1.10.1. 55 to construct phylogeny. For each of the alignment partitions, the most appropriate substitution model was estimated using the Bayesian information criterion (BIC) as implemented in jModelTest 2.1.10 56 . This BIC-based model selection procedure selected the following models, shape parameter α and a proportion of invariable sites (pinvar): TrNef + I and pinvar = 0.8330 for 28S rDNA; F81 + I and pinvar = 0.9030 for the first codon position of the rbcL gene; JC for the second codon position of the rbcL gene; TPM1uf + G and α = 0.8100 for the third codon position of the rbcL gene. We used the HKY model of nucleotide substitution instead of TrNef, F81, JC and TPM1uf given that they were the best matching model available for Bayesian inference method. A Yule process tree prior was used as a speciation model. The analysis ran for 10 million generations with chain sampling every 1000 generations. The parameters-estimated convergence, effective sample size (ESS) and burnin period were checked using the software Tracer ver. 1.7.1. 55 . The initial 25% of the trees were removed, the rest retained to reconstruct a final phylogeny. The phylogenetic tree and posterior probabilities of its branching were obtained on the basis of the remaining trees, having stable estimates of the parameter models of nucleotide substitutions and likelihood. ML analysis was performed using the program RAxML 57 . The nonparametric LB analysis