Phylogeographical patterns in the northwestern European moss Scorpidium revolvens (Sw. ex Anonymo) Rubers (Scorpidiaceae, Bryophyta)

ABSTRACT Introduction. The phylogeography of Scorpidium revolvens (Sw. ex Anonymo) Rubers in northwestern Europe, primarily Scandinavia, was explored. Possibly because of human activity and atmospheric pollution, the species is sparser in the south than in the north, therefore its genetic diversity may be lowest in the south. Methods. Relationships within Scorpidium revolvens were evaluated in a network context, based on nuclear ITS and plastid rpl16 data. The regional diversity and haplotype composition of Sc. revolvens, Sc. cossonii (Schimp.) Hedenäs and Sarmentypnum exannulatum (Schimp.) Hedenäs were compared. Key results. When the sister species Scorpidium cossonii was used as an outgroup to determine polarity, the basalmost Sc. revolvens haplotype was found to be arctic, and the closest haplotypes to this were found in northern or montane specimens. From a basal grade, three lineages evolved, with differential north and south distribution. The haplotype diversity of Sc. revolvens was lowest in the south, and the haplotype composition of Sc. revolvens, Sc. cossonii and Sarmentypnum exannulatum in the south differed from that in other regions. Conclusions. A northern origin for Scorpidium revolvens in Scandinavia is suggested, in common with Sc. cossonii. Wider geographical sampling might confirm a northern origin for Scorpidium. Geographical distributions of the three Sc. revolvens lineages suggest partly separate colonisation routes. Compared with the north, Sc. revolvens in the sparsely populated south has lower haplotype diversity and different haplotype composition. The two other species investigated also had different haplotype composition, and further investigations might determine whether this represents a general pattern in Scandinavia.


Introduction
Among bryophytes, wetland species may have been the subject of the largest number of studies of genetic variation and diversity.The results of these studies have led to a better understanding of several processes, such as cryptic speciation, hybridisation, incomplete lineage sorting, and horizontal plastid gene transfer (Hedenäs and Eldenäs 2007;Natcheva and Cronberg 2007;Bączkiewicz et al. 2017;Hedenäs et al. 2021;Meleshko et al. 2021;Hedenäs 2022).They have also improved our understanding of the phylogeography of both peat mosses (Sphagnum) and so-called 'brown mosses' (Stenøien et al. 2010;Mikulášková et al. 2017;Hedenäs 2019;Manukjanová et al. 2020;Hedenäs et al. 2022).
It seems unlikely that the habitat of Scorpidium revolvens should be naturally much rarer than that of Sc. cossonii in the south, because non-calcareous regions dominate most of southern Scandinavia according to the online geological map provided by the Swedish Geological Survey (https://apps.sgu.se/kartvisare/kartvisare-berg-50-250-tusen.html;accessed 30 May 2023).Reasons for the discrepancy include the possibility that the more intensive agriculture and forestry, and the earlier atmospheric deposition of S and N compounds, especially in the south, affected poorly buffered mires more strongly than calcareous fens, or that the former type of fen was drained more often (e.g.Naturvårdsverket 2007Naturvårdsverket , 2022)).The poorly buffered and sensitive fen habitat where Sc. revolvens grows could therefore have to a high extent been lost in the south (Kooijman 2012;Kolari et al. 2021).
Whatever the reason for the relative rarity of Sc. revolvens in southern Scandinavia, this could potentially result in a lower genetic diversity in the southern compared with the northern portion of its distribution range (Frankham 1996;Leimu et al. 2006).
The present study was carried out to answer the following questions concerning Scorpidium revolvens in Scandinavia: (i) What are the phylogeographical patterns within this species?, and (ii) Is its genetic diversity in the more sparsely populated southern region lower than that farther north?Additionally, because preliminary results suggested that the genetic diversity of Sc. revolvens is indeed lower in the south, the regional diversity pattern of this species was compared with that of two other wetland mosses, focusing on three geographical regions for which sufficient data exist for all three species.

Studied material
Scorpidium revolvens is a medium-sized to robust pleurocarpous fen moss that often becomes purplish to blackish red when it grows exposed to sunshine.Scorpidium cossonii was not distinguished from Sc. revolvens in much of the earlier literature, and Sc.revolvens was frequently, but erroneously, considered to grow in mineral-rich fens.However, in contrast to Sc. cossonii, Sc. revolvens grows in fens of intermediate mineral richness (Kooijman and Hedenäs 1991;Graham et al. 2019).It also differs from Sc. cossonii in having a double chromosome number, being autoicous rather than dioicous, and having longer median stem leaf cells with more longly fusiformly narrowed ends.Additional distinguishing features of Sc. revolvens have been discussed by Hedenäs (1989).It frequently reproduces sexually, producing easily dispersed small spores (13.5-21.0μm; Hedenäs 1989).Scorpidium revolvens is widespread and often common in (northern) temperate to subpolar areas of the northern hemisphere, whereas it is infrequent or mostly rare in the southern hemisphere (Hedenäs 2003a(Hedenäs , 2003b)).
The intraspecific patterns within Scorpidium revolvens were explored based on 63 specimens.Four specimens represent Svalbard (SVA); 16, the northern Scandinavian lowland (NLO); 15, the southern Scandinavian mountains (SMO); 15, the middle Scandinavian lowland (MLO); and 13, the southern Scandinavian lowland (SLO).For the delimitations of these five regions, see Figure 1.Except in the cases of specimens from outside Sweden, these regions correspond approximately with the lowlands of northern Sweden, the mountains of middle Sweden, the lowlands of middle Sweden, and southern Sweden in Hedenäs (2019).Additionally, a single specimen from the middle of the mountain range was included.Scorpidium cossonii was chosen as an outgroup, based on the findings of Hedenäs and Eldenäs (2008).The studied specimens are listed in Appendix 1.
Sequence data for comparison with the rich-fen species Scorpidium cossonii and a species that grows in a habitat similar to that of Sc. revolvens but is more frequent throughout its range, namely Sarmentypnum exannulatum (Schimp.)Hedenäs, were extracted from Hedenäs (2019), with enough specimens being available from the Sc.revolvens regions NLO (23 and 37 specimens, respectively), SMO (22 and 14,respectively) and SLO (12 and 20, respectively) (Appendix 2).

Molecular methods
Nuclear ITS and plastid rpl16 sequences were generated for the Scorpidium revolvens specimens, as described by Hedenäs (2009), except as follows.The amplified PCR products were purified from excess primers and nucleotides by using ExoSap-IT (Applied Biosystems, Waltham, MA, USA).For all specimens, 5 μL of ExoSap-IT was added to 20 μL of PCR product and incubated at 37°C for 30 min; this was 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 (Amsterdam, The Netherlands) for single-stranded sequencing using the Applied Biosystems 3730XL sequencer.

Sequence editing and analysis
The nucleotide sequence fragments for each DNA region were edited and assembled using PhyDE 0.9971 (http://www.phyde.de/index.html;accessed 25 May 2023).The assembled sequences were aligned manually in PhyDE.Regions of partially incomplete data at the beginning and end of the sequences were identified and 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, which was included in the analyses.The sequence alignments used in the analyses are available as Supplemental material 1. GenBank accession numbers for the specimens are listed in Appendix 1.
ITS paralogues are occasionally encountered in bryophytes (Košnar et al. 2012;Hedenäs et al. 2019); however, the ITS chromatograms included in the present study did not show 'messy' patterns or noise that could suggest paralogy, and the 5.8S gene was invariable among specimens (Shaw et al. 2002;Feliner and Rosselló 2007).The revealed ITS variation was thus interpreted as being among homologous haplotypes.
A preliminary analysis did not reveal well-supported incongruences between ITS and rpl16 for the included specimens, and all molecular data were thus primarily analysed in combination to explore relationships.Initial analyses by TCS (Clement et al. 2000) and Neighbor-Net split networks in SplitsTree 4.12.6 (Huson and Bryant 2006) revealed reticulation, and the relationships were therefore analysed in a network context.Potential support for lineages in a tree context was evaluated by jackknife analyses (1000 replications) carried out using the program TNT (Goloboff et al. 2003).
The program TCS was used to identify haplotypes, based on ITS and rpl16 data separately.To investigate patterns of haplotype variation among the regions (see Figure 1), an analysis of molecular variance (AMOVA) was performed using GENALEX 6.501 (Peakall andSmouse 2006, 2012), with nuclear and plastid markers as separate loci.Pairwise Φ PT (an analogue of F ST , i.e. genetic divergence among populations) was estimated using GENALEX, and the same program was used to calculate the effective number of haplotypes (Ne), Shannon's information index (I; sometimes called Shannon's diversity index), and the haplotype diversity (H) for each region.Differences in I between the regions were calculated using Hutcheson's t-test (Hutcheson 1970; see also https://www.dataanalytics.org.uk/comparing-diversity/,accessed 21 February 2023).
Arlequin version 3.5.1.3(Excoffier and Lischer 2010) was used to calculate nucleotide diversity (π), the average number of pairwise differences among regions, and Tajima's D. Tajima's D test of selective neutrality was used to decide whether the regional populations are probably stable in size, expanding or decreasing (Tajima 1989).Tajima's D test was preferred over Fu's F S test (Fu 1997), because recombination levels are unknown (Ramírez-Soriano et al. 2008).The null hypothesis in these analyses is that no differences exist in haplotype and nucleotide diversity or composition among the studied regions.
For both Scorpidium cossonii and Sarmentypnum exannulatum, an AMOVA was run to explore patterns of haplotype variation among the three regions and pairwise Φ PT , and Ne, I and H were calculated for the three regions.Differences in I between the regions were evaluated by Hutcheson's t-test.These calculations were done as described above.

Results
The total number of aligned sites in the studied 63 specimens of Scorpidium revolvens, after deletion of regions at the beginnings and ends that were incomplete for some specimens, was 648 for ITS, including 7 base substitutions, of which 6 were parsimony-informative, and 1 indel, which was informative, and 619 for rpl16 (0 base substitutions; 3 indels, all parsimonyinformative).Sequence lengths were 647-648 bp for ITS and 619-622 bp for rpl16.
Because no well-supported contradictions were found between ITS and rpl16, differences between the results based on these two markers individually and the concatenated dataset are mentioned in the text only when relevant.Based on the results for ITS and rpl16 together, both the Neighbor-Net split network and the TCS network grouped the Scorpidium revolvens specimens in a grade A and the somewhat indistinctly delimited lineages B-D (Figures 2, 3).Only lineage C received moderate jackknife support (86%).This support comes from the ITS data; rpl16 data alone did not provide jackknife support for any of the lineages (results not shown).
The geographical distributions of the four groups overlap in Scandinavia (Figure 4), where lineage B seems to be relatively more common in the northern lowland region (NLO; see Figure 1) and southern mountainous Scandinavia (SMO), including one specimen from Svalbard (SVA).Lineages C and D are relatively more common in southern and middle Scandinavia (SLO and MLO, respectively), including SMO.Grade A shows a more even distribution in the sampled area, and includes three SVA specimens.The most basal specimen in the networks comes from  Svalbard (specimen M1383; see Figures 3, 4).Overall, the specimens closest to the outgroup Scorpidium cossonii in the network were mainly collected in SVA, NLO or SMO.Specimens M282 and M1604, whose positions were suggested to be close to M1383 and M1389, respectively (see Figure 3), based on the data for rpl16 but not ITS (results not shown), were collected in northernmost Scandinavia (see Figures 3, 4).
The AMOVA results revealed that 17% of the haplotype variation was among the regions, whereas 83% was within the regions (n = 63; ITS and rpl16, 5 + 4 haplotypes; Φ = 0.165, p = 0.006, 9999 permutations).A pairwise comparison of Φ PT values and average numbers of pairwise nucleotide differences showed that the composition of SVA differed from that of all other regions (haplotype data only); the composition of SLO differed from that of SVA, NLO and SMO; and the composition of NLO and MLO differed from each other (Table 1, Figure 5A).Diversity was lowest in SVA (ITS data only) and SLO, significantly so in SLO compared with MLO for ITS and compared with the other three regions for rpl16, according to Shannon's information index (Table 2).Tajima's D test of selective neutrality did not suggest that the regional populations were increasing or decreasing (results not shown).

Discussion
Within Scorpidium revolvens, a basal grade gave rise to three different lineages with partly different geographical distributions in northwestern Europe.The basalmost haplotypes were collected in Svalbard, and numerous specimens of the basal grade were northern.In Scandinavia, the lowest genetic diversity was found in the southernmost region.Lineage B is, relatively, most frequent in the north and in the mountains, whereas the other two lineages (C and D) are more frequent in the southern portions of the sampled area.These differences could be a result of colonisation from different glacial refugia, as was suggested for Drepanocladus turgescens (T.Jensen) Broth., Rhytidium rugosum (Hedw.)Kindb., and to some degree, Dryas octopetala L. (Skrede et al. 2006;Hedenäs 2014Hedenäs , 2015;;Hedenäs and Bisang 2019).It could also be a result of differences in climate adaptations, as has been suggested for different genetic entities of Pinus sylvestris L., Scorpidium cossonii and Timmia austriaca Hedw.(Eriksson et al. 1980;Hedenäs 2009Hedenäs , 2019Hedenäs , 2021)).

Sc. revolvens
A comparison with other studies of bryophytes shows that a relatively large proportion of the haplotype variation in Scorpidium revolvens is between regions (Hedenäs 2015(Hedenäs , 2019)).The most salient pattern is that the haplotype composition of the southern region, where the species occurs relatively sparsely, differs from most other regions, and that the diversity is clearly lowest in the south.For Sc. cossonii and Sarmentypnum exannulatum, a relatively large and small proportion of the haplotype variation, respectively, was between regions.No differences in diversity were found between the regions, but in both species, SLO differed significantly from NLO and SMO in terms of haplotype composition.
In moss species that are relatively frequent in northern Scandinavia, southern disjunct regional populations seem to differ markedly in genetic composition from those in regions farther north, a finding that suggests different postglacial origins, as in Drepanocladus turgescens and Rhytidium rugosum (Hedenäs 2014(Hedenäs , 2015)).The patterns seen in Scorpidium cossonii, Sc. revolvens and Sarmentypnum exannulatum suggest that in species with more continuous distributions, the southern regional populations may also have partly different origins than the regional populations of the north.Also, in flowering plants with wide distributions, we find genetic traces of different postglacial immigration routes, for example in Melica nutans L. (Tyler 2002) and Parnassia palustris L. (Borgen and Hultgård 2003).Fen habitats in southern Scandinavia are strongly affected by human management and atmospheric pollution (Naturvårdsverket 2007(Naturvårdsverket , 2022)).The different genetic compositions found in southern regional populations of widespread fen mosses therefore suggest that southern populations may require conservation attention, even if such populations do presently not show signs of population decrease.

Figure 2 .
Figure 2. Neighbor-Net split network for Scorpidium revolvens, based on data for the nuclear ITS and plastid rpl16, using one specimen of Sc. cossonii as an outgroup.Jackknife support of at least 80% is indicated by transverse grey lines.LyL, Lycksele lappmark; MLO, middle Scandinavian lowland; NLO, northern Scandinavian lowland; SLO, southern Scandinavian lowland; SMO, southern Scandinavian mountains; SVA, Svalbard.

Figure 3 .
Figure3.TCS haplotype network for Scorpidium revolvens, based on data for the nuclear ITS and plastid rpl16 combined, using one specimen of Sc. cossonii as an outgroup.Specimens are indicated according to their numbers in Appendix 1. Black dots indicate 'missing haplotypes', that is, potential 'haplotypes' that were not recovered in the sampling.

Figure 4 .
Figure 4. Geographical distributions of the specimens belonging to Scorpidium revolvens grade A (A) and lineages B-D (B, C, D) in Figures 2 and 3. Small grey dots indicate all sampled sites, whereas larger coloured dots indicate sites from which the respective groups were sampled.Different colours indicate different elevation spans.The Scandinavian specimens closest to the base of the network in (A) (cf.Figures 2, 3) and two specimens in (B) and (D), which are discussed in the text, are indicated with lines and specimen numbers.

Figure 5 .
Figure 5. (A) Haplotype compositions in five regions for Scorpidium revolvens, plus a single specimen from Lycksele lappmark, based on data for either the nuclear ITS or the plastid rpl16 (total, n = 63).(B) Haplotype compositions in three regions for Sc.cossonii, based on combined data for the nuclear ITS or the plastid rpl16 (total, n = 57).(C) Haplotype compositions in three regions for Sarmentypnum exannulatum, based on data for the nuclear ITS or the plastid markers (rpl16, trnG, trnL-trnF) (total, n = 71).Circle sizes are proportional to the number of specimens in a region.The plastid marker haplotype compositions are indicated by green text and green circles around the pie charts.MLO, middle Scandinavian lowland; NLO, northern Scandinavian lowland; SLO, southern Scandinavian lowland; SMO, southern Scandinavian mountains; SVA, Svalbard.

Table 1 .
Matrix of regions pairwise Φ PT values (A) and corrected average numbers of pairwise nucleotide differences (B) for ITS and rpl16 for the five regional populations of Scorpidium revolvens.a MLO, middle Scandinavian lowland; NLO, northern Scandinavian lowland; SLO, southern Scandinavian lowland; SMO, southern Scandinavian mountains; SVA, Svalbard.a Values below the diagonal are the Φ PT values (A) or number of pairwise differences (B); their corresponding probability values, based on 9999 and 1000 permutations, respectively, are shown above each diagonal.Values in bold indicate significant differences between two regions (p < 0.05).

Table 2 .
Haplotype and nucleotide diversity for the five studied regions for Scorpidium revolvens, based on data for ITS and rpl16.a Shannon's information index; n, number of specimens; Na, number of haplotypes; Ne, effective number of haplotypes; SD, standard deviation; π, nucleotide diversity.Regions: MLO, middle Scandinavian lowland; NLO, northern Scandinavian lowland; SLO, southern Scandinavian lowland; SMO, southern Scandinavian mountains; SVA, Svalbard.a Five ITS and four rpl16 haplotypes occur in these regions.Indel information was included to define haplotypes but not to determine nucleotide diversity.b Different single superscript letters (a vs b) for separate regions denote a significant pairwise difference in diversity, based on Hutcheson's t-test for Shannon's information index.

Table 3 .
Matrix of regions pairwise Φ PT values for three regions, with at least 10 specimens for Scorpidium cossonii, based on data for ITS and rpl16 (A), and for Sarmentypnum exannulatum, based on data for ITS and the combined data for plastid rpl16, trnG and trnL-trnF (B). a

Table 4 .
(Hedenäs 2009(Hedenäs , 2019))regions with at least 10 specimens for Scorpidium cossonii (13 ITS and 9 rpl16 haplotypes) and Sarmentypnum exannulatum(21 ITS and 12 plastid [rpl16,haplotypes).aShannon'sinformationindex; n, number of specimens; Na, number of haplotypes; Ne, effective number of haplotypes.Regions: NLO, northern Scandinavian lowland; SLO, southern Scandinavian lowland; SMO, southern Scandinavian mountains.aIndelinformationwas included to define the haplotypes.Regions were not significantly different from each other based on Hutcheson's t-test for Shannon's information index.As has been found for its sister species Scorpidium cossonii(Hedenäs 2009(Hedenäs , 2019)), which was used as outgroup to determine the polarity within Sc. revolvens, the present results based on data for Scandinavia suggest that the most basal specimens of Sc. revolvens within grade A (see Figures 2, 3) come from areas with cold climates.These two species are also similar in that members of the main lineages that originated from the basal grade occur in more temperate habitats.I tentatively suggest that ancestral haplotypes of Sc. revolvens are northern, and that the origin of the entire Scorpidium lineage could thus be northern.A wider geographical sampling of Sc. revolvens should be attempted to confirm or reject this hypothesis.Interestingly, early Pleistocene finds of Hamatocaulis vernicosus (Mitt.)Hedenäs, a member of the probable sister genus of Scorpidium (Hedenäs and Eldenäs 2008; Schlesak et al. 2018; Kučera et al. 2019), from northeastern and northwestern Greenland show that this species also occurred in the far north during earlier times (Bennike et al. 2022; Hedenäs and Bennike 2008).