The psbA/rbcL spacer region sequences could not be aligned across all taxa because they are so variable (see below), and therefore rbcL gene sequences were used in phylogenetic analyses to establish their evolutionary relationships. The phylogenetic analyses showed that the Antarctic filamentous Xanthophyceae were distributed in three deeply diverging clades, which we name Xanthonema 1, Xanthonema 2 and Tribonema 1 (Fig. 1). These algae were only distantly related to Bumilleria sicula and Bumilleriopsis filiformis, which showed that the clades were not misidentified taxa belonging to other genera in the Tribonemataceae (Fig. 1). Within the Xanthonema clade 1, the rbcL sequences for Antarctic strains B4-1 and B8-5 were virtually identical, i.e. they differed in only one position. The two strains differed by only 1–2 bp from strain CCAP 808/3 (isolated from a snow field in Alaska) and from GenBank AJ874710 (from identical Antarctic strains, Ohtani 889 and Ling 906; Maistro et al., 2007). Strains A19 and Broady773 from Antarctica and three non-Antarctic strains (CCAP 836/2, SAG 836-1 and UTEX 353) had identical rbcL sequences. Other closest relatives to the Antarctic strains were non-Antarctic strains (UTEX 155, CCAP 808/2, CCALA 516) and Antarctic strain PAB 421 (Fig. 1). Interestingly, UTEX 155 was a duplicate strain of SAG 836-1, i.e. both represented the same isolate that had been kept at two different culture collections. However, the SAG 836-1 rbcL sequence differed by 15 and 14 positions from an existing GenBank sequence of strain UTEX 155 (AF084612) and our sequence (EF455920). Finally, Antarctic strains Broady 395 and Broady 601 and the temperate strain CCALA 517 had nearly identical sequences (8 and 9 rbcL bp separated the Antarctic strains from the temperate strain).
In Xanthonema clade 2, the Antarctic strains A16-5 and Turner 907 shared almost identical rbcL sequences (1 bp difference). They were related to two temperate freshwater strains, CCALA 518 and SAG 60.94 (Fig. 1). The Antarctic strain Broady 759 was more distantly related (Fig. 1). Its rbcL sequence was identical with that from the authentic strain of Xanthonema sessile (ASIB V98; AJ874329), and Broady and colleagues (1997) found that this Antarctic strain was identical to temperate strain ASIB V98 at the morphological level. Therefore, we assigned strain Broady 759 to X. sessile.
The Xanthonema clade 3 was formed by strain Broady 735, isolated from soil attached to vegetables imported from New Zealand into Antarctica (Broady et al., 1997), and two temperate Ukrainian strains (SAG 2181 and SAG 2182). The three strains varied by one position. Total pairwise rbcL sequence differences among the three Xanthonema clades were 63–112 bp, which was approximately the same magnitude as between representatives of three different genera. That is, there were 69–103 bp differences among Xanthonema debile CCALA517, B. filiformis SAG 809-2 and Tribonema sp. SAG 21.94.
The Tribonema clade 1 contained Antarctic strains A21 and Ohtani 887 (identical sequences) as well as Antarctic strain SAG 2165 (4 bp difference). They were most closely related to two temperate freshwater strains (SAG 21.94, SAG 23.94), which were assigned to two distinct species of Tribonema (Table 1). There were no more than four and eight positions different between the rbcL sequences of these Antarctic and temperate strains.
The psbA/rbcL spacer sequences were determined to investigate the close relationships of strains within a single species or among closely related species. The spacer contained large regions of hypervariable nucleotides that were unalignable in entirety over the four major clades. However, a short region of about 30 positions at the 5′-end (pos. 93–122 of reference sequence EF455930) and a second region of about 212 bp at the 3′-end (pos. 319–531 of reference sequence EF455930) could be aligned (Fig. 2). The Xanthonema clade 1 psbA/rbcL spacer sequences were also aligned about 45 bp further downstream (pos. 169 of EF455930) into the hypervariable region, and five groups (A–E) were revealed (Table 1; Fig. 2). Group A strains (from Antarctica and an Alaskan snow field) were almost identical except for a single position and a single 1 bp indel. Group A did not correspond to any morphological species, and therefore it was designated Xanthonema sp. A.
Figure 2. Alignment of psbA/rbcL spacer sequences of studied strains of Tribonemataceae grouped according to sequence similarity (groups A–I, see Table 1). Parts of the more conserved 5′- and 3′-ends are shown as well as a large part of the hypervariable region in between. Numbers refer to sequence positions of strain B4-1 (Accession No. EF455930).
Download figure to PowerPoint
Five strains comprised group B (Table 1; Fig. 2). Two strains were from Antarctica (A19 and Broady 773) and their sequences were almost identical, i.e. they differed in a single sequence position (Fig. 2) and by two indels of 1 bp each. Both Antarctic strains were distinct from the three temperate strains (which had identical spacer regions) by the presence/absence of a short indel (7–9 bp) at the 5′-end of the psbA/rbcL spacer as well as at two other sequence positions. Two of the three temperate strains with identical spacer sequences (CCAP 836/2, UTEX 353) were duplicate authentic strains for Xanthonema hormidioides. The third strain (SAG 836-1) was an authentic strain for X. debile. The authentic strains were (putatively) established by Vischer when he described both X. hormidioides and X. debile (Vischer, 1936; 1945). However, a duplicate authentic strain of X. debile, UTEX 155, was distinctly different (group D), indicating that SAG 836-1 became mislabelled during its history. Therefore, all five group B strains were named X. hormidioides.
Group E (Antarctic strains Broady 395 and 601), whose rbcL sequences were identical, also had identical psbA/rbcL spacer sequences (Fig. 2). They differed from their closest relative (CCALA 517) by 15 sequence positions and an insertion of 65 bp. The CCALA 517 strain was identified as X. exile; it was isolated from a temperate locality, as was the culture used by Klebs (1896) when he described that species, and therefore we used that name for group E. Xanthonema exile was the type species for the genus Xanthonema, and therefore species in clade 1 must be named Xanthonema.
Group C and D sequences from strains CCALA 516 and UTEX 155 were rather distinct as well as strain CCAP 808/2, i.e. they could not be meaningfully aligned (Table 1; Fig. 2). Group C (CCALA 516) was identified as Xanthonema bristolianum and we used that name. Group D contained the authentic strain (UTEX 155) for X. debile and thus the group was assigned this name. Strain CCAP 808/2, which was identified as Bumilleria exilis, was not closely related to strain CCALA 517 (Fig. 2). Because we designated CCALA 517 as the epitype for X. exile (see below), CCAP 808/2 had to be assigned a different species that we designated Xanthonema sp. B. Note that strain PAB 421, used in the rbcL phylogenetic analyses, was not included because the strain became extinct before the psbA/rbcL spacer sequence could be determined.
Xanthonema clade 2 was deeply divergent from Xanthonema clade 1, and the clade represented a new genus distinct from Xanthonema. Clade 2 had two groups, G and H (Table 1; Fig. 2). Within group G, the Antarctic strains A16-5 and Turner 907 had identical spacer sequences. They differed from the temperate strain of that group, CCALA 518, by 18 sequence positions, an indel of one nucleotide and another indel of two nucleotides. The other temperate strain of group G, SAG 60.94, appeared to be even more distant from the two Antarctic strains based upon the spacer region, i.e. with differences in 15 sequence positions (Fig. 2) and five indels of 1–17 bp long. However, the hypervariable regions of all group G strains were still easily aligned (Fig. 2). Group G contained strain CCALA 518 that represented Heterothrix mucicola; therefore, we temporarily assigned the group G strains to this species until a new genus can be described (see taxonomic discussion below). In group H, Antarctic strain Broady 759 (assigned X. sessile; see above) was more distantly related because its psbA/rbcL spacer was shorter and the spacer could not be aligned with those from group G (Fig. 2).
Xanthonema clade 3 (group F) sequences aligned well (Fig. 2) but the clade was deeply divergent, and strains appeared to belonging to a second new genus (Fig. 1). Broady 735, from New Zealand soil material, had no more than three and eight sequence differences when compared with the two Ukrainian temperate strains (SAG 2181 and SAG 2182; Fig. 2). The strains in group F did not correspond to any described morphological species, and therefore we temporarily named the strains ‘Xanthonema’ sp. F until a new genus and species is described.
Tribonema clade 1 had a total of 16 variable spacer sequence positions. Isolate A21 was identical with strain Ohtani 887; however, only a 5′ partial sequence, 356 bp, could be obtained for the latter strain (Fig. 2). Antarctic strain SAG 2165 differed from strain A21 by nine sequence positions and two indels of 1 bp each. Both had only 9/12 and 6/5 sequence differences with their next closest temperate relatives, strains SAG 21.94 and SAG 23.94. Strains in group I could not be assigned a single species name because group I contained two Tribonema species and therefore we temporarily named the group Tribonema sp. I until an unambiguous species name can be found.