Study species

The core portion of this study includes 86 samples of Distichium capillaceum (Appendix 1). Sixty-nine come from Sweden, 16 from mainland Norway and one from Svalbard. The samples cover its phenotypic variation in Scandinavia. To explore Distichium relationships in a wider context I downloaded internal transcribed spacers 1 and 2 (ITS) sequences from GenBank for eight additional D. capillaceum specimens, from mainland Norway (3 samples), Jan Mayen (1), Svalbard (1), Greenland (2) and Antarctica (1), and two sequences of D. hagenii Ryan ex H. Philib. The beginnings of the downloaded ITS sequences were less complete than the newly generated ones and they were therefore not included for the sequence length information in the Results. Two specimens of D. inclinatum were used as outgroup based on its position as sister to the other two species of the genus in the study by Fedosov et al. (2016).

Molecular methods

Total DNA was extracted using the Mag-Bind Plant DNA Plus 96 Kit (Omega Biotek) with the KingFisher Flex and Duo magnetic particle processors. Double stranded DNA templates were prepared by polymerase chain reaction (PCR). PCR was performed using IllustraTM Hot Start Mix RTG (GE Healthcare) in a 25 µl reaction volume according to the manufacturer's instructions.

In all cases, the specified PCR programs were initiated by a denaturation step of 5 min at 95°C and followed by a final extension period of 8 min at 72°C. The PCR programs were, for ITS and for the plastid trnLUAA intron plus trnLUAA-trnFGAA spacer (trnL-trnF), 4 cycles of 30 s at 95°C, 40 s at 57°C and 1 min at 72°C, 4 cycles of 30 s at 95°C, 30 s at 55°C and 1 min at 72°C, 35 cycles of 30 s at 95°C, 30 s at 52°C and 1 min at 72°C. The primers ‘ITSbryoR’ (Hedenäs 2014) and ‘ITS4bryo’ (Stech 1999) were used to amplify ITS and the primers ‘trnC’ and ‘trnF’ (Taberlet et al. 1991) to amplify trnL-trnF. For the plastid rpl16 G2 intron (rpl16) the PCR program was 43 cycles of 30 s at 95°C, 40 s at 58°C and 1 min 15 s at 72°C, with the primers ‘F71’ (Jordan et al. 1996) and ‘rpl16-antR2’ (Hedenäs 2008).

The amplified PCR products were purified from excess primers and nucleotides by adding 1 µl of Exonuclease I (20 U µl^–1) and 4 µl of FastAP Thermosensitive Alkaline Phosphatase (1 U µl^–1) (Thermo Scientific) and incubating at 37°C for 30 min followed by an enzyme inactivation step at 80°C for 15 min. The purified PCR products, together with the same primers used for PCR amplification, were subsequently sent to Macrogen Europe B.V for single-stranded sequencing on an Applied Biosystems 3730XL sequencer.

Sequence editing and analysis

Nucleotide sequence fragments were edited and assembled for each DNA region using PhyDE 0.9971 (< www.phyde.de/index.html>; accessed 16 March 2021). The assembled sequences were aligned manually in PhyDE. Regions of partially incomplete data in the beginning and end of the sequences were identified and were excluded from subsequent analyses. Gaps were coded using the simple indel coding of Simmons and Ochoterena (2000) in SeqState (Müller 2005). Gaps provided additional information, and this was included in the analyses. The sequence alignments used in the analyses are available in the Dryad Digital Repository (Hedenäs 2021). GenBank accession numbers are listed in Appendix 1.

ITS paralogues are occasionally encountered in bryophytes (Košnar et al. 2012, Hedenäs et al. 2019). The ITS chromatograms included in this study did not show ‘messy’ patterns or noise that could suggest paralogy, and the 5.8S gene was invariable among the samples (cf. Shaw et al. 2002, Feliner and Rosselló 2007). Therefore, the revealed ITS variation was interpreted as being among homologous haplotypes.

Potential reticulation was revealed using TCS (Clement et al. 2000) and the phi-test in SplitsTree ver. 4.12.6 (Huson and Bryant 2006) provided statistically significant evidence for recombination (p = 0.01885). Relationships among specimens were therefore evaluated in a network context. The relationships were evaluated in NeighborNet (NN) split networks, produced in SplitsTree and in TCS networks, and potential support for lineages in a tree context was tested by jacknife analyses (1000 replications) performed with the program TNT (Goloboff et al. 2003). Two analyses were performed. The first included all specimens for which ITS was available, incorporating the sequences downloaded from GenBank. The second analysis included all three molecular markers and thus only the Scandinavian specimens for which new sequences were generated. Because visual inspection of jacknife results and NN split networks revealed no conflicts between well-supported structures in the nuclear and plastid NN split networks, all sequence data were combined in the second analysis.

The possible existence of molecularly defined groups was also tested for the specimen set with all three markers in the online assemble species by automatic partitioning (ASAP) tool (Puillandre et al. 2021; < https://bioinfo.mnhn.fr/abi/public/asap/>; accessed 17 September 2021), using the default settings. For this analysis, the two samples lacking ITS information (P739, P751) were excluded.

Morphological study and analysis of measurements

After the molecular relationships among the studied D. capillaceum specimens had been clarified, the morphology of 3–10 selected specimens from each lineage or grade (from here onwards informally called ‘groups’) were studied in detail, in total 39 specimens that are indicated with an asterisk (*) in Appendix 1. Both standard comparisons of qualitative and quantitative characters and detailed measurements of selected gametophyte and sporophyte features were performed, employing dissecting and compound microscopes.

For each specimen, detailed gametophyte characters were measured in two stems (to avoid sampling all leaves from an untypical stem). The lengths (mm) and maximal widths (mm) of the basal sheathing and the apical lamina were measured in five leaves from each stem. For three of these leaves (two leaves from one stem and one from the other), length (µm), width (µm) and length to width ratio of 20 cells was measured in the lower portion of the apical lamina and in the basal sheathing lamina. When available, the length (mm) of 20 capsules, length (µm), width (µm) and length to width ratio of 20 exothecial cells from an arbitrarily selected capsule, and the diameter of 20 spores (µm) were measured. An Olympus SC50 digital camera and the Olympus cellSens Standard ver. 1.13 software for automatic and continuous image stacking were used to produce temporary images of leaves and cells. Measurements were taken from such leaf and cell images, using the Olympus cellSens Standard 1.13 software.

To compare the detailed measurements between the groups within D. capillaceum the measurements were first compared in two principal component analyses (PCA) based on 1) the leaf sizes and mean values of the 20 measured cells from the three leaves which cells were measured, in total 117 leaves, and 2) the mean values of the remaining measurements. These analyses show whether the combined information in the sets of ten and ten characters, respectively, correspond with the molecularly identified groups. Corresponding PCA results with the cell length/width ratios excluded were compared with the mentioned ones to explore if these ratios may have put additional weight to the cell size characters. Secondly, the individual characters were compared between the molecularly identified groups. Both the Levene and Brown-Forsythe tests of homogeneity of variance were significant for most characters and plots of residuals in preliminary Anovas showed many deviations from normality. Thus, the nonparametric Kruskal–Wallis Anova by Ranks for multiple comparisons was used to compare the measurements among or between the groups, respectively. All statistical calculations were made in STATISTICA 13.3 (TIBCO-Software-Inc. 2017). Bonferroni corrections were applied in cases of multiple statistical comparisons.

Distribution maps for the identified molecular groups were produced in QGIS ver. 3.16 (< https://qgis.org/en/site/>; accessed 16 March 2021).

Figure 1.

(a) NN split networks based on ITS for Distichium capillaceum, using D. hagenii and D. inclinatum as outgroups. DC and DH numbers in grey represent sequences downloaded from GenBank, see Appendix 1. (b) NN split networks based on ITS, rpl16 and trnL-trnF for D. capillaceum, using D. inclinatum as outgroup. Grey numbers indicate specimens for which ITS could not be generated. The letters A–E indicate lineages or grades of the network that are discussed in the text (cf. Fig. 2).

Molecular relationships

The total number of aligned ITS sites in the 92 studied Distichium capillaceum, two D. hagenii and two D. inclinatum specimens for which ITS sequences could be generated, after deletion of regions at the beginnings and ends that were incomplete for some specimens, was 762. Of these, 59 sites were variable (32 in D. capillaceum), with 35 (14) parsimony-informative; 47 indels were present (22), with 32 (15) informative. For the 86 D. capillaceum and two D. inclinatum specimens for which rpl16 sequences were generated, the length was 647, 37 (11) sites were variable, and 34 (7) were parsimony-informative; 8 (3) indels with 8 (3) informative. For the 86+2 specimens for which trnL-trnF sequences were generated, the length was 549, 32 (17) sites were variable, and 28 (12) of these were parsimony-informative; 8 (3) indels with 7 (2) informative. Sequence lengths for newly generated sequences of D. capillaceum were, for ITS 655–664 (n = 84), for rpl16 644–645 (86), and for trnL-trnF 496–512 (86) and of D. inclinatum, for ITS 727–744 (n = 2), for rpl16 640 (2) and for trnL-trnF 516 (2).

In the NN split network based on ITS, a high jacknife support (90–100) was provided for D. capillaceum plus D. hagenii, and for a branch within group (grade) E (Fig. 1a). The same branches got a high support in the analysis based on all three molecular markers (Fig. 1b), and in addition four other branches within D. capillaceum got a moderate support, groups (lineages) C, D, and less inclusive branches within groups B and E. The latter network corresponds with the configuration of the TCS network (Fig. 2). From the TCS network, it is also evident that within group E and especially groups A and D reticulation seems to occur.

The ASAP analysis based on all three markers suggested that only two statistically supported groups of specimens exist, one including the two D. inclinatum samples and one including all D. capillaceum samples.

Morphological evaluation

The PCAs based on the detailed measurements of 1) the three selected leaves per specimen and 2) other measured features of D. capillaceum suggest considerable morphological overlap between the different groups identified in the NN split networks (Fig. 3a–b). In the first PCA, cell length and cell length/width in the leaf acumen mainly explain the variation along the y-axis, whereas the other characters explain the variation along the x-axis (Fig. 3c). In the second PCA, spore size and exothecial cell width mainly explain the variation along the y-axis, whereas the other characters explain the variation along the x-axis (Fig. 3d). In the corresponding PCAs with the ratios between cell lengths and widths excluded similarly large overlaps between the different groups as in Fig. 3a and b were found (not shown).

Figure 2.

TCS haplotype network on ITS, rpl16 and trnL-trnF for D. capillaceum, using D. inclinatum as outgroup. Grey numbers indicate specimens for which ITS could not be generated. The letters A–E indicate lineages or grades of the network that are discussed in the text (cf. Fig. 1).

Figure 3.

PCAs based on ten specimens (9 in c) of group A, seven (7) of group B, nine (7) of C, ten (6) of D and three of group E (cf. Fig. 1). (a) The positions of three leaves from each specimen based on leaf and leaf lamina cell sizes, along the first two axes in a PCA. Factors 1 and 2 explain 29.25% and 22.11% of the variation, respectively. (c) Explanatory factors in the plane of factors 1 and 2 in (a). This PCA is based on the following, for each leaf: length and width of the leaf acumen (Lfl-ac, Lfw-ac), length and width of the leaf base (Lfl-ba, Lfw-ba), mean cell length, width and length/width ratio for 20 cells in the leaf acumen (Cl-ac, Cw-ac, Cl/w-ac), and mean cell length, width and length/width ratio for 20 cells in the leaf base (Cl-ba, Cw-ba, Cl/w-ba). (c) The positions of specimens based on leaf and sporophyte features, along the first two axes in a PCA. Factors 1 and 2 explain 29.01% and 22.21% of the variation, respectively. (d) Explanatory factors in the plane of factors 1 and 2 in (b). This PCA is based on the following, for each specimen: mean length and width of acumen and base of ten leaves from two shoots (Lfl-ac, Lfw-ac, Lfl-ba, Lfw-ba), mean capsule length based on 20 capsules (Cap-l), and mean exothecial cell length, width and length/width ratio from one arbitrarily selected capsule (Exc-l, Exc-w, Exc-l/w), median length of peristome teeth in two capsules (Per-l) and mean spore size based on 20 spores (Spo).

When individual measured characters are compared between the species, the means differ between some groups in most cases (Table 1, Fig. 4), but again the overlap is great and no pattern that consistently distinguishes one or several groups from the other ones exist. No other characters were found that distinguished either of the groups. Instead, also characters like costa excurrency and roughness, how strongly the ends of meeting leaf lamina cells project, and exostome teeth splits and ornamentation varied strongly within the groups.

Habitat and geographical distribution

No habitat differentiation between the molecular groups of Scandinavian D. capillaceum was evident from the label information of the specimens, whereas different geographical distributions were observed (Fig. 5). The mostly southern group A is absent from the northern third of Scandinavia and is rare at higher elevations, except for one GenBank specimen (DC7) collected in northern Norway, whereas the northern group B is absent in the southern third. Groups C and E are primarily found in the mountain range or close to the mountains and group D, finally, was found in the mountains and far north plus the Baltic Sea region.

Molecular and morphological patterns

The ITS-based NN split network included Distichium specimens from Scandinavia as well as a few from areas outside this region (Fig. 1a). It shows that even if D. hagenii is morphologically most similar to D. inclinatum (Hagen 1899–1904, Mönkemeyer 1927, Nyholm 1987, Hallingbäck et al. 2006, Hassel et al. 2013), it is more closely related to D. capillaceum. The available molecular information, including that of Fedosov et al. (2016), therefore suggests that D. inclinatum is molecularly strongly differentiated from both the other species. Interestingly, four out of eight GenBank specimens of D. capillaceum, from Svalbard, Jan Mayen and southern Norway, were situated closest to D. hagenii in the network. It seems possible that when every second GenBank specimen of D. capillaceum was found in this position, and GenBank specimen DC5 sits on a long branch, this could possibly be due to an artefact. Finally, the northern Norwegian D. inclinatum specimen included by Hassel et al. (2013), i.e. DC7 in the present investigation, actually belongs to D. capillaceum group A. Thus, the Maximum likelihood tree based on atpF-atpH in the study by Hassel et al. (2013) includes only D. capillaceum and D. hagenii and does not provide information regarding the relationships between the three species of the genus.

The five D. capillaceum groups identified in the network based on all three markers included lineages A–D that apparently evolved from the grade E (Fig. 1b). Neither lineage received strong jacknife support. Together with the revealed recombination, the suggested reticulation in the TCS network (Fig. 2), and the ASAP results this suggests that the lineages are best understood as within a single species. This was borne out also by the morphological data, which suggested that groups A–E differ only slightly and with strong overlaps between most groups. The PCAs suggested that groups A and D could be distinguished from each other when sporophytes are present, but only sometimes based on the measured leaf characters. The weakly supported molecular lineages and the weak and mostly overlapping morphological differentiation between the five groups suggest early stages in the speciation process, that is, incipient speciation (cf. de Queiroz 2007).

Unlike the situation in other recently investigated and morphologically variable Scandinavian mosses, such as Meesia uliginosa Hedw., Oncophorus wahlenbergii Brid., Plagiopus oederi (Brid.) Limpr., Racomitrium lanuginosum (Hedw.) Brid. and Tomentypnum nitens (Hedw.) Loeske (Hedenäs 2018, 2020a, b, c, Hedenäs et al. 2020), neither additional cryptic nor morphologically recognizable species are present. Further evidence supporting this is that plants corresponding with the morphological concepts of D. capillaceum var. compactum and var. curvatum were found in more than one of the five groups.

Habitat and geographical distribution

All groups of D. capillaceum occur in a variety of base-rich to calcareous habitats in Scandinavia. The plants are highly modifiable, depending on habitat exposure and humidity, and especially small plants in compact tufts from the most exposed sites are striking.

The samples for which new sequences were generated suggest that group A is absent from the northern third of Scandinavia. However, one GenBank specimen (DC7) collected in northern Norway and included in the ITS network showed that group A occurs at least rarely also further north and underlines that ample sampling is required to correctly interpret geographical ranges. This lends further support to the results of Collart et al. (2021), emphasizing that small samples can be problematic when interpreting and model-ling distributions. Among Scandinavian bryophytes this is illustrated also by Scorpidium cossonii (Schimp.) Hedenäs. Hedenäs (2009) found its basal haplotypes only in the farthest north of Scandinavia, in the study by Hedenäs (2019) such haplotypes were shown to occur south to southern Lapland in Sweden, and further sampling (Hedenäs, unpubl.) revealed a few occurrences in the Scandinavian mountain range south to Oppland in Norway. Based on GenBank samples, D. capillaceum group A occurs also in Antarctica (DC8), thus displaying a bipolar distribution pattern. Whether this group is the only one occurring in the relatively wide Southern Hemisphere distribution area (see map in Ochyra et al. 2008) remains to be investigated. Presently, it seems like groups C and E occur primarily in or near the mountain range and group D, finally, occurs in the mountains and far north plus the Baltic Sea region. The distribution of group D reminds of that found for one plastid lineage of Syntrichia norvegica F. Weber (Hedenäs et al. 2019) and for species like Buckia vaucheri (Lesq.) D. Ríos, M.T. Gallego & J. Guerra and Campylium bambergeri (Schimp.) Hedenäs, Schlesak & D. Quandt (Hedenäs et al. 2014).

Figure 4.

Boxplots with median values, quartiles and whiskers from maximum to minimum values, for measured characters in Distichium capillaceum groups A–E (cf. Fig. 1). Only characters where overall significant differences were found among the groups are included (Table 1). Groups with different letters under the lower whiskers differ significantly from each other in pairwise comparisons (Kruskal–Wallis Anova by Ranks). For n, see Table 1.

The different geographical distributions of the D. capillaceum grade and lineages suggest that they survived the last glacial period in different refugia and entered Scandinavia along different post-glacial routes, or that groups C and E are restricted to the mountains and the far north due to lack of adaptations to thrive under milder climates. The distribution seen for group D and genotypes or species with similar distribution patterns could also be due to poor competitive abilities combined with relaxed competition from larger plants both in the mountains and the limestone habitats of the Baltic region (Hedenäs 2014, 2015). As discussed above for GenBank specimen DC7, considering that relatively few specimens were sampled these interpretations are necessarily preliminary.

Figure 5.

The Scandinavian distributions of specimens belonging to groups A–E (a–e), based on all three molecular markers (Fig. 1). Grey dots indicate the sampling locations, and the larger coloured dots superimposed on these indicate the geographical and elevational distributions of the respective group.