Congruence of chloroplast- and nuclear-encoded DNA sequence variations used to assess species boundaries in the soil microalga Heterococcus(Stramenopiles, Xanthophyceae)

Heterococcus is a microalgal genus of Xanthophyceae (Stramenopiles) that is common and widespread in soils, especially from cold regions. Species are characterized by extensively branched filaments produced when grown on agarized culture medium. Despite the large number of species described exclusively using light microscopic morphology, the assessment of species diversity is hampered by extensive morphological plasticity. Two independent types of molecular data, the chloroplast-encoded psbA/rbcL spacer complemented by rbcL gene and the internal transcribed spacer 2 of the nuclear rDNA cistron (ITS2), congruently recovered a robust phylogenetic structure. With ITS2 considerable sequence and secondary structure divergence existed among the eight species, but a combined sequence and secondary structure phylogenetic analysis confined to helix II of ITS2 corroborated relationships as inferred from the rbcL gene phylogeny. Intra-genomic divergence of ITS2 sequences was revealed in many strains. The ‘monophyletic species concept’, appropriate for microalgae without known sexual reproduction, revealed eight different species. Species boundaries established using the molecular-based monophyletic species concept were more conservative than the traditional morphological species concept. Within a species, almost identical chloroplast marker sequences (genotypes) were repeatedly recovered from strains of different origins. At least two species had widespread geographical distributions; however, within a given species, genotypes recovered from Antarctic strains were distinct from those in temperate habitats. Furthermore, the sequence diversity may correspond to adaptation to different types of habitats or climates. We established a method and a reference data base for the unambiguous identification of species of the common soil microalgal genus Heterococcus which uses DNA sequence variation in markers from plastid and nuclear genomes. The molecular data were more reliable and more conservative than morphological data.


Background
Heterococcus is a genus of yellow-green microalgae (Xanthophyceae, Stramenopiles) that is common and widespread in soils of cold regions such as the Alps or Antarctica [1,2]. In addition to soils, three species have been reported from freshwater [3][4][5][6], and Heterococcus is the only xanthophyte known from lichen symbiosis [7,8].
Heterococcus produces extensively branched filaments when grown on agarized culture medium ( Figure 1); however, in field samples it produces unicellular coccoid cells that are weakly connected. Perhaps uniquely for microalgal genera, all species have been described based upon isolates grown in culture and observed with a light microscope [1,2,6]. Without culturing, Heterococcus is often mistaken for other coccoid xanthophytes, eustigmatophytes or green algae. Sixty-one Heterococcus species have been described [9], and 51 species are recognized [10]. Extensive ultrastructural observations were undertaken by Lokhorst [2], but he reluctantly concluded that ultrastructural features were not sufficient to distinguish species.
Sexual reproduction is unknown for Heterococcus, and therefore the biological species concept cannot be employed (e.g. [11]); only the morphological (typological) species concept has been used. That is, Heterococcus species identity is limited to light microscopic morphological characters interpreted within the extensive plasticity that is exhibited during culture studies [1,2,6]. For example, branching patterns are not present in very young or old cultures, and filament formation is suppressed (coccoid cells are produced) when cultures are grown at suboptimal temperature ranges [1] ( Figure 1). Cladistic analysis of these morphological features would be extremely difficult because cell sizes, branching patterns, colony growth, chloroplast number and other features overlap extensively among the species, even when grown under optimum conditions. Molecular phylogenetic analysis is often a reliable alternative for identification of species; however, species diversity of Heterococcus using molecular markers was unstudied and no molecular reference data base existed. From only seven Heterococcus species DNA sequences had previously been reported, and all these sequences were from conserved molecular markers. The sequences revealed the probable monophyletic origin of the genus and its basal position within the Xanthophyceae, which was distinct from other filamentous members (e.g. Tribonema, Vaucheria) [12][13][14][15]. We used molecular phylogenetics, especially within the framework of the monophyletic species concept [16][17][18], to evaluate 33 culture strains identified as Heterococcus (Figure 2). Fourteen strains were originally identified to species level using morphology, and ten of those strains were authentic culture strains, i.e. the culture strains used to describe the species [1,[3][4][5]19]. Unfortunately, the cultures used to describe all other species have been lost. For nine authentic strains, there are extended morphological descriptions with numerous illustrations produced by two independent authors [2][3][4][5]. We added 19 unidentified culture isolates, including twelve cultures recently isolated. Our goals were (1) to test boundaries of Heterococcus species as inferred from morphological features and (2) to establish a reference data base of strains unambiguously distinguished with DNA sequence data. We chose two highly variable molecular markers, i.e. the chloroplast-encoded psbA/rbcL spacer region [20,21] and the nuclear-encoded internal transcribed spacer 2 of the nuclear rDNA cistron [22][23][24], to examine species boundaries. We also determined full plastid-encoded rbcL gene sequences to infer the phylogenetic position of species.

Results
Four of the strains, identified as Heterococcus, were green algae ( Figure 2). These were not included in the rest of the study. The rbcL gene sequences were used to assess the phylogenetic relationships of the remaining 29 strains (Figure 3, Additional file 1). For 25 strains, PCR amplification was successful for the whole region from psbA (downstream), through the rbcL, through the rbcL/ rbcS spacer and to the rbcS gene; therefore the full rbcL gene, 1467 base pairs long, was determined (Additional file 2). We failed to obtain full rbcL sequences for three authentic strains, Heterococcus fuornensis Vischer strain SAG 835-5, H. caespitosus Vischer strain SAG 835-2a, and H. protonematoides Vischer strain SAG 835-9, but we used available sequences (AM421004, AM421002 and AJ579575) for these three strains. Also, for strains DB14-15 and MZ1-6 the full rbcL failed to amplify. Fifteen different rbcL sequences were recovered among the 29 strains, which implies that the rbcL gene was identical among many strains (Additional file 3). Only the 15 different rbcL sequences were used for phylogenetic analyses ( Figure 3, Additional file 1). Monophyly of Heterococcus was highly supported with all methods except maximum likelihood, and this confirmed the generic identity of the 29 strains. The analyses resolved two well supported clades, named "H. caespitosus clade" and "H.

psbA/rbcL spacer
To further examine the relationships, the psbA/rbcL spacer sequences were determined for 28 strains (H. fuornensis strain SAG 835-5 failed to amplify). The spacers varied greatly in length and primary sequences; the sequences could not be aligned across all strains. Nevertheless, two short sequence stretches were aligned across all strains. The first was 23 nucleotides at the 5'-end (pos. 78-99 of reference sequence H. viridis Chodat strain SAG 835-3, JX681220) and the second was 36 nucleotides at the 3'-end (pos. 312 -347, same reference sequence). The 33 strains identified as Heterococcus used in this study. The Heterococcus strains are listed with their species names (where provided) from previous morphological analyses, their assignment to clades and lineages in the rbcL phylogeny (boxed with thick lines; see Figure 3), their assignments to a certain species recognized in this study (boxed with thin lines), their new species designations (see Discussion) and their geographic origin. Highlighted in green are genotypes, i.e. groups of strains exhibiting high sequence similarities (see text). Strains in bold letters represent cryopreserved epitypes (reference strains) designated for each species (see Discussion). An asterisk marks an authentic reference strain (see text). a marks those strains that have recently been isolated by us or were provided to us for this study; n.a. 1 , not applicable because the psbA/rbcL spacer sequence could not be determined (see text); n.a. 2 , not applicable because strains were identified as green algae (see text).
In most Heterococcus strains the nucleotide length of the psbA/rbcL spacer ranged from 275 nucleotides (H. caespitosus strain SAG 835-2a,) to 289 nucleotides (H. sp. strain DB14-15). The sequence for H. crassulus strain SAG 835-4 was 1762 nucleotides, and the identical sequences for two strains, DB14-1-1 and DB14-5-1, were 2143 nucleotides. Sequence similarities further downstream grouped the strains into seven "spacer groups", A -G, within which the psbA/rbcL spacers were identical or displayed only very few differences (Figure 2, Additional file 3). When mapped on the rbcL phylogeny, the strains of spacer groups A, B and C were included in the H. viridis clade, strains of spacer groups D and E fell in the H. caespitosus clade, and spacer groups F and G represented the lineages "H. sp." and H. crassulus ( Figure 3). Between closely related groups or within a group, also other regions of the psbA/rbcL spacer sequences could be aligned. For example, strains of the Heterococcus viridis clade (groups A-C) had sequence regions that aligned well, but there were up to 28 nucleotide differences among them. In addition, there was a hypervariable region of different lengths (20-31 nucleotides, between pos. 172 and 193 of the reference sequence H. viridis SAG 835-3, JX681220) that was not alignable among the three groups, but clearly distinguished them from each other. In the H. caespitosus clade, i.e. between groups D and E, the psbA/ rbcL spacers also aligned well over the entire lengths, but differed at 14 sequence positions and a single indel. Similarly, there was a maximum of 13 psbA/rbcL spacer sequence differences between strains of group A. In group A there were nine strains isolated from Antarctica ( Figure 2). There were no more than two nucleotides difference among them when Antarctic strain MZ3-7 was not considered and the previously unidentified strain SAG 56.94, isolated from Germany, had just one to three sequence differences with the eight Antarctic isolates. Conversely, strain MZ3-7 was with seven to nine spacer differences more distant to the other eight Antarctic strains. Strain H. brevicellularis Vischer SAG 835-1 was the closest neighboring strain of strain MZ3-7; there were just 4 sequence positions different between both strains. Group B contained two Antarctic strains (MZ1-3, MZ1-6) that had identical spacers; Group B also contained two German strains (DB14-15, DB15-5) with identical spacers; however, the Antarctic strains differed at 4 positions when compared to the German strains. Finally, group D had three strains that had only one nucleotide difference, while two strains in group F had only two sequence differences.

ITS2
Nuclear-encoded ITS2 sequences were determined for 28 strains as an independent assessment of the plastidencoded sequences. Heterococcus fuornensis strain SAG . The names next to the tree represent clades and lineages recovered in the phylogenetic analyses. Thick lines indicate internal branches resolved by maximum likelihood, maximum parsimony, minimum evolution distance and Bayesian analyses and with significant statistical support (bootstrap >95%, posterior probability = 1.0). Black filled circle marks the branch indicating the monophyletic origin of Heterococcus that was significantly supported (bootstrap >95%, posterior probability = 1.0) except for the maximum likelihood analyses. The phylogeny shown is part of a larger ML phylogeny (calculated with GARLI v0.96 [25,26]) based on a rbcL data set (1325 bp long, 517/418 variable/parsimony informative sites) consisting of 15 Heterococcus sequences and 32 other Xanthophyceae sequences corresponding to clades C, B, T, and V as defined in [14] (see Additional file 1) as well as two outgroup taxa. Scale bar, substitutions per site.  [27] sequencestructure NJ tree (unrooted) as derived from the multiple sequence-structure alignment of ITS2 helix II. Bootstrap values (100 pseudo-replicates) are mapped to the appropriate internodes. Branch lengths are drawn proportional to inferred changes. The template ITS2 variant used in B) is highlighted in bold. (B) ITS2 secondary structure of ITS2 variant DB14-1-1_acl11 (group F, H. ramosissimus) used for homology modeling of helix II (shaded) for all strains of Heterococcus. The secondary structure was visualized with VARNA [28]. Helices are numbered I-IV. Typical ITS2 motifs are highlighted by filled arrowheads. Open arrowheads mark positions of two CBCs that distinguish groups D (= H. leptosiroides) and E (= H. caespitosus). An additional conserved region throughout all strains of Heterococcus is indicated by a cloud (see text). In contrast to the template structure the region d1 is deleted in four strains (group D, see Additional file 4). The region d2 is deleted in all other strains not classified in group F. (C) Subtree as obtained by using the complete sequence-structure information from helices I-IV (template highlighted in bold). Further subtrees as derived using clade specific structural templates (helices I-IV) are provided as Additional files 4, 5, 6. (D) Visualization of the complete sequence-structure alignment used to generate the tree as shown in A). Consensus structure (51%) of helix II for all ITS2-sequences obtained from the complete multiple sequence-structure alignment without gaps. Sequence conservation is indicated from red (not conserved) to green (conserved). Nucleotides which are 100% conserved in all sequences are written as A, U, G or C. Nucleotide bonds which are 100% conserved throughout the alignment are marked in yellow. Note the U-U mismatch. The figure was generated with 4SALE [29]. 835-5 was successfully amplified and included; however, amplification failed for strain MZ1-6 and this strain was not included in the ITS2 analyses. Based upon alignment similarity, the ITS2 sequences formed the same groups that were recovered in the psbA/rbcL spacer analysis; therefore, we used the same group notation for both datasets. Within a spacer group, the ITS2 sequences and their secondary structures were easily aligned and rather similar; conversely, between spacer groups, the sequences and secondary structures were highly variable, i.e. they could be aligned with confidence only for a few short segments. The ITS2 sequences exhibited a considerable length variation of up to about 130 nucleotides between spacer groups. The shortest ITS2 had 285 nucleotides (strain SAG 2163 from group C; strain EIF 399/372 from group D); the longest sequence had 416 nucleotides (strains DB14-1-1, DB 14-5-1 from group F). Within each spacer group, the ITS2 sequences were relatively constant in length (variation < 10 nucleotides), except for group D where sequences were either short (285-287 nucleotides) or long (315-319 nucleotides), and the difference was due to an indel at the terminal end of helix III in the secondary structure model (see below; Additional file 4). The ITS2 sequence from Heterococcus fuornensis, which had a distinctive rbcL gene but could not be amplified for the psbA/rbcL spacer, showed little similarity to other spacer groups.
The inferred RNA secondary structures folded into the common core structure known for eukaryotes [23] which consisted of four helices with the third being the longest and helix IV the shortest ( Figure 4, Additional files 4, 5, 6). Because of the high sequence length variation there was not a single ITS2 secondary structure from which the secondary structure models of the remaining sequences could be deduced using homology modeling. Only helix II could be modeled throughout the set of sequences independent of the used sequencestructure pair. However, within each group complete secondary structures could be obtained by homology modeling (Figure 4, Additional files 4, 5, 6). Throughout the set of sequences, conserved regions were restricted to the entire helix II (pos. 86-125 of reference sequence H. viridis SAG 835-3, JX681147), which had a constant length of 40 nucleotides, and a segment of about 50 nucleotides (pos. 165-189 and 205-228 of the same reference sequence) located at or close to the distal end of helix III ( Figure 4, Additional files 4, 5, 6). It was followed by an extended terminal end of the helix III of 45 and 133 nucleotides in spacer groups D and F, whereas the corresponding sequence region in other spacer groups comprised of six (H. fuornensis strain SAG 835-5, no assigned group) to 18 nucleotides (spacer group E). That means there was a continuous lengthening/ shortening of the ITS2 helix III within Heterococcus ( Figure 4, Additional files 4, 5, 6). Another conserved ITS2 region useful to distinguish groups among Heterococcus strains was an unpaired sequence segment (~12 nucleotides) adjacent to helix II (Figure 4; pos. 126-137 of reference sequence H. viridis SAG 835-3, JX681147). It separated H. crassulus SAG 835-4, H. fuornensis SAG 835-5, and two clusters of strains from each other. The one cluster comprised the strains from groups A-C, the other the strains from groups D-F. Within each cluster the sequence segments were invariant.
Multiple copies of ITS2 were recovered in eight strains (from groups A, B, D, F and G; Additional file 2), i.e. there were no clear sequence reads possible without cloning. Four to 12 clones per strain were sequenced and this revealed up to seven ITS2 variants per strain (Additional file 2). Differences between ITS2 variants consisted of one to seven sequence positions and a few small indels (< 5 nucleotides); they were mostly located in helices I, IV and the basal part of helix III preceding the conserved segment. In groups B and D differences between ITS2 variants were also located in the conserved helix II. In group D three out of the ten detected ITS2 variants were lacking the extended 45 nucleotides long terminal end of helix III. These shorter variants were present in all three strains of group D or in about half (10) out of the sequenced 21 clones, while the longer ITS2 variants were retrieved only from two strains, EIF 423/A790-45 and EIF 128/A788-70.
The ITS2 phylogenetic analyses were confined to helix II (with adjacent unpaired conserved region, pos. 85-134 of reference sequence H. viridis SAG 835-3, JX681147) for assessing relationships among all studied strains. The sequence alignment was with 50 positions relatively short; it contained no more than 14/9 variable/ parsimony informative sites and just nine sequences were not identical with others. However, a well-resolved phylogeny was obtained when secondary structure was considered in addition to primary structure information ( Figure 4). The resolved helix II sequence groups were congruent with the groups recovered in the rbcL phylogeny (see spacer group letters on Figure 4). A common origin of H. crassulus with H. fuornensis was well supported in the unrooted ITS2 (helix II) phylogeny, and this contrasted with the rbcL phylogeny where the relationships of both were unresolved ( Figure 3). Also an unrooted (maximum likelihood) phylogeny of only Heterococcus rbcL gene sequences did not support the common origin of both species (not shown). The helix II phylogenetic tree resolved a close relationship of groups D and E (as the rbcL phylogeny, H. caespitosus clade in Figure 3), but at the same time both groups were clearly separated species because there were two CBCs [23,24] in helices II and III (Figure 4, Additional file 4); also their helices I and IV could not be aligned. No resolution was provided within the H. viridis clade, i.e. among spacer groups A, B and C ( Figure 4). The complete ITS2 sequence was used to produce phylogenetic trees for individual spacer groups or rbcL clades. For example, then within the rbcL H. viridis clade the spacer groups A-C were resolved (Additional file 5). Within group A, both variants of strain MZ3-7 shared a common origin and were separated from other strains of the group. The three authentic strains, Heterococcus viridis SAG 835-3, H. mainxii SAG 835-6, and H. marietanii SAG 835-7, shared identical ITS2 sequences with each other (Additional file 7). Similarly, the ITS2 sequences of the two Antarctic strains EIF 398 and EIF PAB 398/473 were identical ( Figure 2, Additional file 3). Two authentic strains, H. brevicellularis SAG 835-1 and H. moniliformis SAG 835-8, and one unidentified strain (SAG 56.94) shared identical ITS2 sequences except for a short indel (4 nucleotides) in helix IV. Another congruence with the chloroplast-encoded data was within group E where the ITS2 sequences of two authentic strains, H. caespitosus SAG 835-2a and H. protonematoides SAG 835-9, were identical. Conversely, within group B no differentiation among strains was possible due to the extensive radiation of multiple ITS2 variants of strain MZ1-3 (Additional file 5). Similarly, group D had extensive radiation of ITS2 variants and no relationships among strains were resolved (Additional file 4). Here the shorter variants of both strains EIF 423/A790-45 and EIF PAB 399/372 were intermixed among each other; they formed two independent lineages distinct from a clade comprising the variants with extended terminal end of helix III. Within group F no clear distinction of the two strains DB14-1-1 (with multiple variants) and DB14-5-1 was provided ( Figure 4).

Monophyletic species concept
Our results show that morphological features do not characterize species; for example, we found that five authentic culture strainsused in the original descriptions for the five specieshad nearly identical DNA sequences and ITS2 secondary structures. Furthermore, we found other examples where authentic strains or identified strains were synonymous with another species (see below). Almost all Heterococcus species have been described using the same morphological approach, we have examined all existing authentic culture strains, and we find that morphological species descriptions are inadequate for this asexual genus. We conclude that morphological features characterize only individuals, not species. Therefore, we must apply a different species concept for Heterococcus.
The 'monophyletic species concept' of Johansen and Casamatta [18], which is derived from the 'phylogenetic (autapomorphic) species concept' of Mishler and Theriot [16,17], is easily applied to asexual species when molecular data are available. In our study, the DNA sequences and ITS2 secondary structure comparisons recovered a clear and robust phylogenetic structure for the 29 Heterococcus strains. Eight groups of sequences were repeatedly recovered using three different molecular markers; sequences within each group were very similar or identical while those between groups were highly variable. Using the monophyletic species concept, we recognize these groups as eight distinct species, and we identify previously unidentified strains and environmental clones to species level.
In a previous study, the rbcL gene and psbA/rbcL spacer were used, in conjunction with the monophyletic species concept, to define species in the Tribonemataceae, another asexual lineage of filamentous Xanthophyceae [21]. In that study, strains of the same species formed a monophyletic clade in the maximum likelihood rbcL gene phylogeny, and strains within the same species differed by less than 10 nucleotides. Within each species, the psbA/rbcL spacer was easily aligned, and within species variation was limited to single nucleotide differences and short indels. As with our study, the entire spacer could not be aligned between species. Therefore, the molecular-based monophyletic species concept identifies species in the same way for both studies.
The original iconotypes used to nomenclaturally anchor all Heterococcus names consist of ink drawings of various morphological features. We have shown that these morphological features are not reliable for species identity, and ink drawings are very limited for reference. In some cases, neotype material was dried and deposited in a herbarium [2], but this too is ambiguous because in at least one case, the wrong culture was used (see below) and because the material does not clearly separate species (still based upon morphology). Therefore, the names are herein further anchored with epitypes to avoid all ambiguity. The epitypes here designated are cryopreserved culture strains that can be re-investigated. The nomenclatural details are summarized below.

Taxonomy and nomenclature
Group A strains differed by no more than five sequence positions (one nonsynonymous substitution) in their rbcL genes, and the psbA/rbcL spacer regions aligned well over their entire lengths, with no more than 11 sequence differences. Their ITS2 sequences also aligned well over their entire lengths and there were no more than eight ITS2 sequence positions different. Therefore, we regard group A as a single species, Heterococcus viridis, which is the type species for the genus. It is noteworthy that we used Chodat's [19] authentic strain, SAG 835-3 [3,30]. Group A also contained four additional species that were based upon authentic strains, H. brevicellularis, H. mainxii, H. marietanii and H. moniliformis [4,5]. We conclude that interpretations of largely overlapping morphological features, which were used to establish these as separate species, are not taxonomically sound; therefore, we consider these to be heterotypic synonyms of H. viridis (Figure 2,  Additional files 3 and 7). Previously, Lokhorst [2] found that three of these strains were morphologically almost indistinguishable and he considered them as varieties. Group A also includes two strains previously identified as H. pleurococcoides Pitschmann [1]. However, the two strains were not authentic strains, and we cannot completely conclude that H. pleurococcoides is a heterotypic synonym of H. viridis. In addition, eight unidentified strains are now identified as H. viridis based on our study ( Figure 2 [3], and therefore has nomenclatorial priority over H. protonematoides [5], which becomes a heterotypic synonym. Heterococcus For group G, H. crassulus was represented by an authentic strain, and we accept this as a recognized species. Similarly, for an unnamed group (see Figure 2), H. fuornensis was represented by an authentic strain, and therefore we recognize this as a distinct species.
Heterococcus In group B, there were no more than nine different sequence positions (one nonsynonymous substitution) among the complete rbcL sequences and only four nucleotide differences among the psbA/rbcL spacers. Strains of group B formed a well-supported monophyletic clade independent of other groups/species in the rbcL phylogeny (Figure 3, Additional file 1) as well as phylogenetic analysis of the whole ITS2 region (Additional file 5). Therefore, we recognize group B as a distinct species. Placing a scientific name on group B (species B) is problematic because our study included all existing authentic cultures. Our molecular data, which were rigorously analyzed with phylogenetic methods, contradict species distinctions based upon non-rigorous intuition using highly variable morphological features, and we conclude that our rigorous analyses are more scientifically sound. Nonetheless, there are 61 named species, and perhaps group (species) B belongs to one of those species. If we simply propose a new name, then we are defying the intent of the International Code of Botanical Nomenclature (or any other Code). Therefore, we simply apply four of the oldest names used in [6] for group (species) B and the three other groups (C, D, F) which contained no authentic strains. We assume that none of these names is in contradiction with the morphology of the strains we designate to represent the four species. We argue that establishing axenic cultures and examining filaments at a certain age of a culture time (as it has been done to define species of Heterococcus previously [2][3][4][5][6]) is a poor way to identify species and this does not allow field samples to be identified to species. With Heterococcus growth in culture is a measure of meaningless differences and there is no hope that morphology will ever be useful when trying to put a name on these four groups (species). We suggest that close phylogenetic relationship with defined reference (epitype) strains as well as genetic distance from corresponding strains of other species, evidenced by rbcL gene phylogenies and differences in the psbA/rbcL spacers are appropriate to identify the species. Secondary structure of ITS2 constitutes an additional autapomorphic feature to define species of Heterococcus. We use Heterococcus conicus Pitschmann as name for group Group C consisted of a single strain, SAG 2163 (Figure 2), which formed a distinct lineage in the rbcL and full ITS2 phylogenies (Figure 3, Additional files 1  and 4). It was also distinct in its psbA/rbcL spacer from H. viridis and H. conicus which were the closest relatives with SAG 2163. Therefore, we recognize group C as a distinct species and we use Heterococcus virginis Pitschmann as name. Two unidentified lichen photobionts share identical partial rbcL sequences (JN573801 and JN573802; [8]) and these differed by only one nucleotide from SAG 2163. Therefore, we assign these lichen photobionts to H. virginis as well.
Heterococcus Group D comprised three strains with no nucleotide difference in the rbcL and a single in the psbA/rbcL spacer. In the ITS2 phylogeny, the three strains could not be distinguished due to different ITS2 variants that are intermixed (Additional file 4). Strains of group D exhibit a unique ITS2 secondary structure with a rather long helix III with considerable length variation at its terminal end (Additional file 4). Despite being closely related to H. caespitosus (group E) in the rbcL phylogeny ( Figure 3 and Additional file 1) there are two CBCs in ITS2 that separate group D strains from the latter species. Consequently, we recognize group D as a distinct species, Heterococcus leptosiroides Pitschmann. One environmental clone sequence from Antarctic soils (AJ580925) shared full sequence identity in rbcL gene with group D strains, and therefore, we conclude that the environmental clone belongs to H. leptosiroides. Finally, group F contained two strains with fully identical complete rbcL sequences and two differences in their psbA/rbcL spacers. In the ITS2 phylogeny the two strains could not be distinguished due to the variation of multiple copies ( Figure 4C). The group F strains had a unique ITS2 secondary structure with a particularly long helix III ( Figure 4B). Group F forms an independent lineage within the Heterococcus clade in the rbcL phylogeny (Figure 3, Additional file 1). Consequently, we recognize group F as a distinct species and use Heterococcus ramosissimus Pitschmann as name. The authentic strain of H. endolithicus was described by Darling and coworkers [1], 195/A790-35 (accessioned as strain SAG 63.90 by the SAG culture collection), but our study revealed that it represents a green alga, i.e. a close relative of Desmococcus species (Trebouxiophyceae) (Figure 2). Our microscopic investigation of SAG 63.90 revealed the same morphology as described previously [1]. Significantly, this morphology is somewhat similar to the morphology of Desmococcus [10], and this makes us confident that SAG 63.90 still represents the original isolate. Despite Darling and coworkers [1] having reported a "typical xanthophycean plastid structure" based on electron microscopy, they already considered H. endolithicus distinct from all other Heterococcus species because it did not form long filaments. In addition, three more strains from Antarctic soils were also identified as Desmococcus-like green algae ( Figure 2). Therefore, we exclude H. endolithicus from the genus Heterococcus and propose a new nomenclatural combination for this authentic strain, but unfortunately we cannot apply the specific epithet (endolithicus) because the name Desmococcus endolithicus Broady & Ingerfeld already exists [31]. Therefore, we propose an avowed substitute name: Desmococcus

Infraspecific Variation and Geographical Distribution
Our relatively small sample of 29 Heterococcus strains already showed eight groups (= eight species). Within the five species for which multiple strains were available, the psbA/rbcL spacer sequences even resolved groups of strains with nearly identical sequences (genotypes; Figure 2). Strains with identical, or nearly identical, sequences were repeatedly found in our relatively small sample of Heterococcus strains and, importantly, established at different times from geographically distant localities. This implies that the number of species within Heterococcus might be rather limited. The same genotypes were confined to certain habitats (soil or freshwater) and geographical regions (Europe or Antarctica).
For example, H. viridis strains SAG 835-3, SAG 835-6 and SAG835-7 were collected from freshwater habitats in Europe while all other strains of the species were from soil in Europe or Antarctica ( Figure 2); they represent a distinct subgroup (genotype) within the species. Similarly, two strains of H. conicus were collected from freshwater in Europe (DB14-15, DB15-5) whereas the other two H. conicus strains were collected from Antarctic soil (MZ1-3, MZ1-6; Figure 2). We draw two conclusions. First, the two species are geographically widespread and will grow where suitable habitats exist. Second, genotypes of those growing in freshwater are distinct from those growing in soil. The sample size is exceedingly small, but there is a suggestion that our molecular data are separating populations within both species that have distinctly different habitats.
We also note that half of the Heterococcus genotypes in our sample originated from Antarctica but not a single genotype was shared between Antarctic and European strains, i.e. none of the Antarctic Heterococcus strains shared identical psbA/rbcL spacer sequences with the European strains. A previous study showed that Antarctic strains within a single species of the xanthophyte Xanthonema were distinguished from their temperate counterparts by only few nucleotides for the highly variable psbA/rbcL spacers [21]. Therefore, our findings for Heterococcus support the view that the Antarctic and temperate strains of xanthophyte species represent different populations of a single species. That is, the Antarctic strains of a given species share their own common evolutionary histories, implying that there was only one (relatively recent) colonization event in Antarctica for each species. Alternatively, if multiple colonization events occurred, then the invasions were too recent to produce significant divergence [32,33].

ITS2 sequence features
Our ITS2 sequences are, to our knowledge, the first ITS2 sequences available for Xanthophyceae. Given that available ITS2 sequence information for stramenopile algae is still limited, two aspects of the Heterococcus ITS2 sequences appear unusual, but might be useful for taxonomy. First, in Heterococcus ITS2 lengths were approximately 300 nucleotides long in most strains; group F (H. ramosissimus) sequences were almost 400 bps. The average length of ITS2 across all eukaryotes is about 210 bps as inferred from the ITS2 database IV [34]. In group D, two size classes occurred, i.e. either~250 or 300 bps, due to a large indel at the terminal end of helix III. Other stramenopile algal groups, the Bacillariophyceae and Phaeophyceae, show a bimodal distribution of their ITS2 sequence lengths, i.e. around 250/290 bps and around 250/350 bps, respectively. Second, the ITS2 sequences were rather variable, i.e. only few and rather short sequence segments were alignable with confidence across the eight Heterococcus species. Such a high sequence variation among species of a single genus is unusual, at least as compared to genera and species of green algae where ITS2 has been revealed as a reliable molecular marker already many times, e.g. [35][36][37][38]. Finally, because the ITS2 rDNA sequences were so variable in Heterococcus, it is not possible to safely define compensatory base changes (CBCs), which can be deduced only from well aligned sequences. CBCs in conserved regions of the helices of ITS2 have been proposed for distinguishing microalgal species when sexual reproduction is unknown [23,24]. However, the concept of CBCs does not imply that two strains lacking CBCs must belong to the same species. That is, there may be other criteria that define microalgal species.
The high ITS2 sequence variability is in line with our maximum likelihood (GARLI and RAxML) analyses that had weak support for the monophyletic origin of the genus (Additional file 1). The monophyletic origin of Heterococcus was also weakly supported by a multiple gene phylogenetic analyses of photosynthetic stramenopiles that included three of our Heterococcus species [15]. Therefore, our results may suggest that more data (and better taxon sampling) are required to firmly demonstrate the monophyly of Heterococcus, or they may suggest that some of the species defined in our study belong to a separate, and sister, genus.

Conclusions
Application of the monophyletic species concept using the highly variable chloroplast-encoded psbA/rbcL spacer, the more conserved plastid rbcL gene, and the nuclear-encoded ITS2 provided a reference data base for unambiguous identification of the common cold soil microalga Heterococcus. Eight species were recognized and characterized at the molecular level. Previous taxonomic studies relied entirely on morphological features produced in cultures; our data will facilitate diversity assessments that are independent of culturing. In addition, the PCR amplification approach for the psbA/ rbcL spacer is specific for Xanthophyceae. Using the new reference data base, partial sequences of the psbA/ rbcL spacer and/or ITS2 may already be sufficient for the assignment of a new strain to a certain species. There are some difficulties; amplification of the psbA/ rbcL spacer may be hampered by length variations, and sequence analyses of ITS2 may be complicated by multiple variants per strain. Using the monophyletic species concept, our species are mostly in contrast to those defined by the morphological (typological) species concept. We conclude that the extensive morphological plasticity displayed in culture cannot be interpreted without rigorous methods (e.g. cladistics), and the largely overlapping morphological characteristics make cladistic analysis very difficult or impossible. The identical, but highly variable, sequences that were repeatedly recovered among the species, suggest that the species diversity of Heterococcus is not extensive, especially considering the repetition that occurred in our small sampling from Europe and Antarctica. The observed sequence changes within a species may reflect adaptations to different types of habitats or climates and distinguish geographically widely separated strains.

Culture strains
Twenty three culture strains were received from the SAG culture collection [39,40]; five strains were provided by other workers in the field. Another five isolates (strains MZ1-3, MZ1-6, MZ2-4, MZ2-5, MZ3-7) were newly established using methods described previously [21] [42]. The ten new isolates have been accessioned by the SAG culture collection under strain numbers as given in Additional file 2.
DNA extraction, PCR amplification and sequencing of strains DNA was isolated from fresh cultures as in [21]. For determining sequences of the plastid-encoded psbA/rbcL spacer which lies upstream of the rbcL gene, i.e. between the psbA and rbcL genes, and full-length sequences of the rbcL gene the PCR approach of Andersen and Bailey [20] modified to amplify the target sequence in one piece [21] was used. The 5' primer psbA5 [20] or Xan2F [21], anchored in the psbA gene, and the 3' primer RS3 [20] placed in rbcS (downstream of rbcL) were used. However, for strains with extremely long psbA/rbcL spacers, PCR amplification was in two overlapping fragments, i.e. with primer pairs psbA5 and X5RG (the reverse complement of primer X5FG [21]) and Xan3F [21] and RS3 [20]. For amplification of ITS2, PCR primers Xits2F (5' -GCTACACTCTGACACCTG -3'; which binds at the 5'-end of the 18S rRNA gene, i.e. pos. 1462-1477 of reference sequence AM490822 H. viridis SAG 835-3, and LR1850 [43] were used to amplify a rDNA fragment that expanded from 3'-end of SSU downstream to the 5'-end of the LSU rDNA. The same cycling parameters were used for all PCR reactions as described previously [21]. PCR products were purified using Invisorb Spin PCRapid Kit (Invitek, Berlin, Germany) or MSB Spin PCRapace Kit (Invitek, Berlin, Germany). Sequence determination of the psbA/rbcL spacer was as previously [21], but complemented by nine additional primers to obtain the sequences of the extremely long psbA/rbcL spacers present in some Heterococcus strains, i.e. hetnew_F (5'-GGTACAACTGAY CAATT-3'), het_F (5'-GGTGGTACAATTGGYCATC CAGA-3'), spacer2R (5'-ATTCGAGTACGCTCTTGTA-3'), DB_F (5'-GGCAAGCCTTTCACTCTTGAT-3'), DB_R (5'-CCACCCGGATTTAAAAGAGTT-3'), DB_F2 (5'-TTCGATACGGGAAACAACTT-3'), DB_R2 (5'-G ATCCTTTGGTTCAACTTAGAAGA-3'), SAG_F (5'-C AAGCTTCGACTGAGGCTT-3'), and SAG_R (5'-AT TGCAAGGCAAGCCTTG-3'). The latter two sequencing primers were used only for H. crassulus strain SAG 835-4, the "DB" primers only with the two isolates DB14-1-1 and DB14-5-1. The rbcL sequences were checked against the NCBI gene sequence database using nucleotide BLAST (blastn) [44,45] to confirm that they were Xanthophyceae. For four strains from the SAG culture collections no PCR products of plastid-encoded markers as described above could be obtained and then a portion of the nuclear-encoded 18S rRNA gene was sequenced with primer 895R [46] after PCR amplification with primers preferentially binding to green algal rDNA, primers 20 F [8] and CH1750R [46] and checked against the NCBI gene sequence database. For ITS2 sequence determination, the sequencing primers were 5.8SbF and 5.8SbR [47], 1800 F [43] and ITS4Xan (5'-TCCTCCGCTTAGTTATATGC-3'), which was a modification of primer ITS4 [48]. In several cases no clear sequence reads were obtained, even after repeated PCR and sequencing attempts, due to multiple copies of the ITS2 which varied in primary sequences (see Results). Then cloning of the PCR products was performed with the TOPO TA cloning kit and the pCR2.1-TOPO vector (Invitrogen, Carlsbad, CA, USA). Ligations were transformed into competent E. coli TOP 10 cells as supplied by the manufacturer. In the plasmid screening, white E. coli colonies containing correct DNA insertions were identified by direct amplification of the inserted DNA fragment with a vector-specific primer set M13F/M13R. The ITS fragments were re-amplified from M13F/M13R PCR products with primer pair Xits2F/LR1850 as described above or the clones were cultivated overnight in LidBac reaction tubes (Qiagen, Hilden, Germany) with 1 ml LB medium containing 100 μg ampicillin and plasmid DNA was prepared from the clones with a NucleoSpin-Plasmid kit (Macherey and Nagel, Düren, Germany) following manufacturer's instructions. Sequencing reactions were performed with the Dye Terminator Cycle Sequencing v3.1 kit (Applied Biosystems, Darmstadt, Germany) and separated on an ABI Prism 3100 (Applied Biosystems, Darmstadt, Germany) sequencer. The sequences were assembled using the program SeqAssem [49]. For GenBank accession numbers of newly determined sequences for the 29 Heterococcus strains see Additional file 2; the accession numbers for the four green algal sequences determined in this study are JX681197 -JX681200.

Chloroplast-encoded marker analysis
The chloroplast-encoded marker sequences (from 3'-end of psbA downstream to 5'-end of rbcS) were manually aligned using Bioedit [50] and Seaview [51] editors from which the rbcL sequence alignment used for the phylogenetic analyses was extracted. The rbcL sequence alignment was constructed using 15 of the sequences newly determined for Heterococcus in this study to which 32 other sequences available for the Xanthophyceae clades C, B, T, and V as defined previously [14] were added (Additional file 1). The two phaeophycean sequences Fucus vesiculosus NC016735 and Ectocarpus sp. AY372978 were employed to root the phylogeny. The alignment was subjected to distance, maximum-parsimony (MP) and maximum-likelihood (ML) approaches. ModelTest 3.7 [52] used in conjunction with PAUP* 4b10 [53] determined that the GTR+I+G model [54] provided the best fit to the data according to the AIC criterion with estimations of nucleotide frequencies (A = 0.2859, C = 0.1447, G = 0.1981, T = 0.3714), a rate matrix with six different substitution types, assuming a heterogeneous rate of substitutions with a gamma distribution of variable sites, number of rate categories = 4, shape parameter α = 0.8249 and proportion of invariable sites (pinvar) of 0.4977. This model was used for the minimum evolution distance (ME) approach performed with PAUP* 4b10 (DNA distances set to maximum likelihood) and the maximum likelihood ML (approach) using GARLI v0.96 [25,26]. A complementary ML phylogeny construction was done with the program RAxML [55], using the GTR+Γ+I model and with 100 bootstrap replicates. Neighbor-joining distance (NJ) phylogenies were constructed in connection with the "HKY85 model" [56] with PAUP* 4b10. For ME and maximum parsimony (MP) tree reconstruction (PAUP* 4b10) a heuristic search procedure with 10 random input orders of sequences and TBR were employed to find the best tree. Best scoring trees were held at each step. In MP analyses, the sites were weighted (RI over an interval of 1-1000). Bootstrap resampling was performed on NJ, ME, MP with 1000 replications and 2000 replications on ML GARLI trees. For the Bayesian analysis the program MrBayes version 3.1.2 [57] was used with procedures as described earlier [58].
Nuclear-encoded ITS2 sequence-structure analysis Using hidden Markov models (HMMs) nuclear ITS2 sequences have been annotated according to [59]. One ITS2 sequence from each group A, D, F and H. fuornensis strain SAG 835-5 was used for secondary structure prediction. Based on minimum free energy ITS2 secondary structures were directly folded with the help of the "RNAstructure" software [60,61] and manually corrected. The four sequence-structure pairs were used as templates for homology modeling of the remaining 39 secondary structures [62]. In accordance to [63] the phylogenetic analysis followed the procedure outlined in [23,34,64,65]: automatically, a multiple sequence-structure alignment was generated in 4SALE v1.7 [29,66], i.e. either partial (Figure 4) or full (Additional files 4, 5, 6) sequences and their secondary structures were synchronously aligned, making use of an ITS2 sequence-structure specific scoring matrix [66,67]. Based simultaneously on the primary sequence and the secondary structure information, phylogenetic relationships were reconstructed using NJ through in conjunction with an ITS2 sequence-structure specific general time reversible (GTR) substitution model as implemented in Prof-DistS v0.9.9 [27,67]. Bootstrap support [68] was estimated based on 100 pseudo-replicates ( Figure 4, Additional files 4, 5, 6). Trees were visualized using Treeview [69].

Additional files
Additional file 1: Maximum likelihood (ML) phylogeny of rbcL gene sequences for Heterococcus and other members of Xanthophyceae. The phylogeny was calculated with the programme GARLI v0.96 [25,26] based on a rbcL data set (1325 bp long, 517/418 variable/parsimony informative sites) consisting of 15 Heterococcus and 32 other Xanthophyceae sequences (corresponding to clades C, B, T, and V as defined in [14]) as well as two sequences from Phaeophyceae as outgroup. Scale bar, substitution per site. Numbers mapped to internodes are bootstrap values from 2000 replicates, only values >70% have been recorded. The phylogeny in this Figure includes the phylogeny of 15 Heterococcus strains shown in Figure 3 (highlighted). The inserted table lists bootstrap values mapped to internodes of the Heterococcus clade using six different analysis methods (see text). Scale bar, substitution per site.
Additional file 2: DNA sequences newly determined for 29 Heterococcus strains and their GenBank sequence accession numbers. For the psbA/rbcL spacer and full rbcL gene all determined sequences are listed, for ITS2 only those sequences that were different from each other. (p), only psbA/rbcL spacer and partial full rbcL gene could be determined; (a), already made available previously; n.a., not applicable.
Additional file 3: Groups of Heterococcus strains with fully identical rbcL and/or psbA/rbcL spacer sequences. Strains marked in bold were used for the rbcL phylogeny ( Figure 3, Additional file 1). Species assignment is according to the new species designation as in Figure 2 (see Discussion).