Introduction

It has been debated for a long time how common hybridization is among bryophytes and whether it is important for evolutionary processes. There is solid evidence that hybridization via allopolyploidy is fairly common in mosses and liverworts (reviewed by Wyatt et al. 1989; Stoneburner et al. 1991; Bishler and Boisselier-Dubayle 1997; Ricca and Shaw 2010; Shaw et al. 2013). On the other hand, there are few well-documented cases of hybridization at the homoploid level and most concern peat mosses Sphagnum (Natcheva and Cronberg 2007a, b; Ricca and Shaw 2010). In bryophytes, the diploid sporophyte is the primary hybrid, and spores formed in the sporophyte after meiosis are recombinants of the parental genomes, which, if viable, could germinate to form a new haploid gametophyte generation. Different kinds of evidence have been used to indicate hybridity in bryophytes, including observations of morphologically intermediate sporophytes (Khanna 1960; Pettet 1964; Anderson and Lemmon 1972), mostly in weedy bisexual species. Such an example is the cross between Pleuridium subulatum (Hedw.) Lindb. and Ditrichum pallidum (Hedw.) Hampe.—one parent having a stegocarpous spore capsule (P. subulatum) and the other a cleistocarpous capsule (D. pallidum) producing hybrids (Pleuriditrichum marylandicum) having intermediate capsule (Andrews and Harmann 1959). Artificial hybridization was early attempted (von Wettstein 1928; Burgeff 1943), and more recent evidence comes from molecular techniques (reviewed by Shaw 2009). However, recombinants of hybrid origin have rarely been documented in natural populations, and it has been suggested that homoploid hybridization is rare because hybrid sporophytes in most documented cases have failed to produce viable spores (reviewed by Natcheva and Cronberg 2004). Nevertheless, many examples of hybridization may remain undetected since molecular data suggest that recombinants inherit a skewed representation of the parental genomes (Shaw 1994, 1998; Natcheva and Cronberg 2007a, b), so that they display only occasional “misplaced” morphological character rather than complete intermediacy.

Homalothecium sericeum and H. lutescens (Brachytheciaceae) are perennial dioicous pleurocarpous bryophytes commonly found in Europe, showing a more regional distribution in North America (Nyholm 1965; Hofmann 1998; Huttunen et al. 2008; Lieske 2010; Rosengren et al. 2016; GBIF 2018a, b). The species occur in distinct habitats with a comparatively high pH (median and range of pH is 6.5 and 5.9–7.6 for H. sericeum versus 7.1 and 6.2–7.9 for H. lutescens; Tyler and Olsson 2016); H. sericeum grows epiphytically on trunks of broad-leaved trees or on calcareous stone or mortar in forests or parks, whereas H. lutescens grows on calcareous soil, usually in open habitats such as pastures. The two species display a number of distinguishing characters in both the gametophyte and sporophyte generations (Table 1). They have similar triangular and longitudinally plicate leaves. In dry condition, H. sericeum is easily recognized by its short secondary branches that form curled shoots with a typical golden lustre. In comparison, the branches of H. lutescens are rarely curled and less densely packed. It grows in loose mats with ascending shoots, whereas H. sericeum grows tightly connected to its substrate. The sporophytes differ in several traits, notably by H. sericeum having erect spore capsules and H. lutescens having horizontally oriented spore capsules. A partially intermediate taxon, H. lutescens var. fallax, has been considered to have a hybrid origin by some authors (Nyholm 1965; Hofmann 1998). Both species are facultatively nannandric, meaning that fertilization by swimming male gametes can be mediated by either normal-sized males or tiny dwarf males that grow epiphytically on females. An experimental study revealed that spores of both species germinate equally well on females of H. lutescens and form fertile dwarf males, suggesting an obvious route for hybridization (Rosengren and Cronberg 2015).

Table 1 Comparison of morphological characters of H. lutescens and H. sericeum (Hofmann 1998) and H. lutescens var. fallax (Nyholm 1965)—a variety suspected to be a hybrid between H. lutescens and H. sericeum by Hofmann (1998)
Full size table

The purpose of this study is to test the hypothesis that H. lutescens and H. sericeum hybridize in mixed (sympatric) populations. Such populations occur on the Baltic island of Öland where Lars Hedenäs and Frida Rosengren (personal communications) have observed aberrant gametophytic morphotypes, proposed to be of hybrid origin. We compare morphological expression of leaf characters (gametophyte) and capsule (sporophyte) inclination between allopatric and sympatric populations. We also use 85 specifically selected SNP markers, with alternative alleles diagnostic for the two parental species, to identify hybrid sporophytes in sympatric populations. Based on earlier studies of hybrid sporophytes in other groups of bryophytes, we expected the intermediate spore capsules in Homalothecium to indicate primary hybridization, which means that they would show heterozygotic expression of species-specific SNPs markers.

Materials and methods

Sources of material

Allopatric and sympatric populations were collected from different localities in south Sweden, including Öland and the mainland region of Skåne (Table 2). Populations from Öland were collected from the Great Alvar, a steppe-like and open area with a mosaic of base-rich and weakly acidic soils overlying the limestone bedrock (Königsson 1968; Bengtsson et al. 1988; Runyeon-Lager and Prentice 2000) or from the adjacent area with deciduous forest called Mittlandsskogen. The majority of the localities of Skåne were old, abandoned gravel or chalk pits (for H. lutescens, but not for H. sericeum), or artificial habitats created by remnants from such industrial activities (Rosengren 2015).

Table 2 The study sites of parental and putative hybrid populations
Full size table

Sampling

Most of the material was sampled from Öland in November 2014, along line transects, each sample consisted of a gametophyte colony with mostly several sporophytes, separated by a minimum of two metres to avoid double sampling of the same clone. The sampling sites represented (1) sympatric populations (Hb1–7) sampled where the two species grow in close vicinity and intermediate capsules were observed along calcareous stone walls; (2) allopatric populations of H. lutescens (Hl 1–2) collected from the ground at a calcareous sand steppe area and at a border area between the Great Alvar and the Midland forest; and (3) allopatric populations of H. sericeum (Hs 1–3) collected on the top of calcareous stone walls belonging to three different churchyards. The samples from the sympatric populations were deliberately subjective in that we actively chose to include colonies that had somewhat abnormal appearance in shoot architecture and branching pattern; no such colonies occurred in the allopatric populations. Additional material was later collected from allopatric populations in Skåne, to represent a reference area in which hybridization is less likely since sympatric populations are rare. Samples for each of these sites consisted of a more limited number of colonies from separate spots (i.e. not sampled along transects). Samples of H. sericeum came from tree trunks (Hs 4) and various kinds of stone walls (Hs 5–7), whereas H. lutescens came from the ground in former limestone quarries or gravel pits (Hl 3–6). Material for SNP marker extraction and selection (see below) was also collected from allopatric populations in Skåne. Gametophyte shoots with sporophytes were brought to lab for morphological studies and DNA extraction, and the remaining material of all samples was air-dried and preserved as voucher specimens. All voucher specimens are deposited at the Botanical Museum, Lund University, and are obtainable on request.

Morphological study: Leaf measurements

From each female shoot, about 10 branch leaves (c. 2.5 cm from shoot tip) were dissected and mounted on a glass slide, from which a single complete representative leaf was selected to be photographed under a dissecting microscope (i.e., the whole leaf) and under a compound microscope (i.e., leaf cells). Images of leaves (and capsules, see below) were captured by a Nikon DS-2Mv digital camera and images of leaf cells were taken by a Nikon DS-Fi1 digital camera. A suite of 14 measured characters and 4 derived characters were scored for branch leaves (Fig. 1 and Table 3). In total, 240 samples were measured: 59 samples of H. sericeum from allopatric populations, 60 samples of H. lutescens from allopatric populations, and 121 samples from sympatric populations. All leaf image measurements were made using the Measure tool in the imaging software package NIS-Elements AR 3.0. Capsule measurements. Images of sporophytes were used to score capsule orientation in relation to the seta (inclination) by using the Angle tool in ImageJ (Schneider et al. 2012) (Fig. 2). Unopened ripe capsules characterized by dark green- to brown-coloured capsules with opercula but no calyptra were scored from the same individuals used for gametophyte morphology, but some individuals lacked mature sporophytes, and in some cases two spore capsules were measured from the same colony. In total, 213 living capsule samples—74 samples representing allopatric populations of H. sericeum, 25 samples from allopatric populations of H. lutescens and 113 samples from the seven sympatric populations—were measured and characterized. Statistical analyses. (1) The 14 morphological leaf characters were tested for normality using the Shapiro-Wilk test and equality of variance using one-way ANOVA with Student’s t test for pairwise comparisons (ggplot2 package; Wickham 2009) in RStudio version 3.3.1 (RStudio Team 2016). All samples from one region were treated as a regional population. (2) A Pearson correlation test was used to display the relationship between leaf length (mm) and width (mm) in all leaf specimens from the three origins—allopatric populations of H. lutescens, allopatric populations of H. sericeum, and sympatric populations. (3) Patterns of variation in morphological characters were analysed using principal component analysis (PCA) on the 14 leaf characters using the R packages “FactoMineR” (Lê et al. 2008) and “Factoextra” (Kassambara 2017). (4) The inclination of capsules from the three origins—allopatric populations of H. lutescens, allopatric populations of H. sericeum, and sympatric populations—were tested for equality of variance by a one-way ANOVA followed by the Tukey test for the significance of differences between the means of capsule inclination. All analyses were performed in RStudio.

Fig. 1
figure 1

Measurements of branch leaf characteristics of Homalothecium spp. Leaf characters (L1-L18) are explained in Table 3

Full size image
Table 3 Branch leaf characters and scale units in morphometric analyses of two parental species, H. sericeum and H. lutescens from allopatric and sympatric populations
Full size table
Fig. 2
figure 2

(a) Measurement of capsule orientation in relation to the seta in Homalothecium as the angle (in degrees) between the seta and spore capsule at the basis of the spore capsule. (b) Capsule inclinations of individuals from the allopatric populations fell into the black ranges of variation; in the sympatric populations individuals occurred with capsule inclinations ranging outside typical capsule inclinations of the pure species (red zone: 150°-164°), indicating hybrid origin

Full size image

Genetic study: SNP marker extraction and selection

We assessed the genetic relationships and degrees of genomic admixture of the two putative parental species in sporophytes using single nucleotide polymorphism (SNP) markers derived from transcriptomic data. Gametophyte shoots with sporophytes were collected from morphologically distinct H. lutescens and H. sericeum from allopatric populations in Skåne (see Table 2). The gametophyte shoots and sporophytes were acclimatized in the constant cold room (ca. 12 °C) for 3 days before RNA extraction. Total RNA was extracted by using the RNeasy Plant Mini Kit (Qiagen) with liquid nitrogen following the protocol used by Rosengren (2015). To get a high number of expressed genes, approximately 20 gametophyte and 20 sporophyte samples were separately pooled for construction of cDNA libraries of H. lutescens and H. sericeum. Six paired-end libraries were conducted by next-generation sequencing (NGS) on Illumina MiSeq Platform at the Sequencing Facility at the Department of Biology (Lund University). In total, we fed 34.5 megabase pairs (Mb) of raw sequences to run on a computer platform at the computational infrastructure UPPMAX, Uppsala University (http://www.uppmax.uu.se). The required programs consisted of several modules executed consecutively in an automated fashion. All scripts running in UPPMAX are provided in Supplementary materials Appendix I. We used the Trinity software (launched 2014-07-17) for de novo reconstruction of transcriptomes from RNA-seq data from gametophytes of H. sericeum to create a reference sequence (total no. of assembled bases: 51,692,526). After reference construction, we remapped short DNA sequences (reads) from both H. lutescens and H. sericeum to the reference using the aligner program Bowtie (version 1.1.2) and exported output alignments in the standard SAM format for SNP calling in SAMtools (version 0.1.19). Trinity, Bowtie and SAMtools were run on the command lines in UPPMAX. A final number of 79,314 contigs (total contigs created: 324,101) representing 42,481,248 reads (H. lutescens: 10,289,896; H. sericeum: 32,191,352) remained to be analysed from the remapping to the reference sequence. The SNP calling was performed by a custom made Perl-program (available upon request). In total putative 4631 SNPs with the read coverage of the SNP position at least 20, within 100 bp upstream or downstream of the targeted SNP positions, were extracted from the SAMtools. 133 SNPs which displayed species-specific variants (alleles) expressed in both the gametophyte and sporophyte generations were chosen for SNP genotyping. We blasted the contigs to predict their cellular compartmentalization (nuclear, mitochondrial or chloroplast) and to ensure that all three kinds were represented. The mitochondrial and chloroplast SNPs are expected to show maternal inheritance and therefore unlikely to show heterozygotic pattern in sporophytic tissue (setae). These plastid SNPs were primarily used to identify the maternal parent species in sporophytes (although they still would have a potential to signal hybridization if combined with alleles in other loci signalling origin from the other parent). DNA extraction. For allopatric populations of H. lutescens and H. sericeum (see Table 2), young gametophyte tissue (haploid) was sampled for DNA extraction using the Qiagen DNeasy Plant Minikit from the same individuals as used for the morphological study. These samples were used to check that the selected SNPs loci had alternative alleles specific for H. lutescens and H. sericeum. For samples from sympatric populations, we extracted DNA from setae from the same samples as used for capsule inclination and leaf morphology. The seta is the stalk supporting the capsule of a moss, and a part of the diploid sporophyte and therefore in principle containing a complete set of chromosomes from each parental gametophyte if hybridization occurs. The reason for not extracting also the capsule was that we planned to sow the spores to test spore germination in a separate study (Sawangproh et al. 2020; Sawangproh and Cronberg, in prep). Our initial plan was to also analyse the haplotypes of the maternal individuals, but this was not done due to a temporary problem with extraction failure (see Table 2). SNP genotyping. SNP genotyping was performed on DNA samples of gametophytes and setae at SciLifeLab Uppsala with the multiplexed primer extension (SBE) chemistry of the iPLEX assay. The amount of DNA was measured by PicoGreen method by SciLifeLab prior to the analyses. The mean (±SD) DNA concentrations of extracts from gametophyte and seta samples were 1.79 ± 1.81 ng/μl and 0.26 ± 0.38 ng/μl, respectively. The preferred total DNA amount per sample in the analysis was 20 ng (5 ng of DNA added into the analysis of each of the four multiplex SNP panels). Since there was at least 20 μl of extract in microtubes, the amounts were usually enough for the analysis, although the DNA concentration varied a lot among samples. In some cases a lower amount was successfully used, although not all loci gave detectable signal.

Results

Leaf morphology

Results indicated that statistical techniques based on normality assumptions could be used for most morphological characters in our study of leaf morphology. Most of the leaf characters examined for the allopatric populations of H. lutescens and H. sericeum are similar within species; some leaf characters slightly differ between the two regional populations, i.e. Öland and Skåne, for example, width of leaf at base and width of median cell; and number of teeth at base is significantly larger for allopatric populations of H. sericeum from Öland than Skåne. Gametophytes from sympatric populations showed more or less clear intermediacy in several leaf characters such as width at base of leaf lamina, leaf surface area, tooth length at base of leaf lamina, and tooth number at base of leaf lamina (Fig. 3a, f, g, and h, respectively), whereas other characters fell out of range of both species, exemplified by the ratio between leaf length and width (Fig. 3b and c; Fig. 4). As for the general leaf dimensions, specimens from allopatric populations of H. lutescens had the longest and widest leaves, followed in size by specimens from allopatric populations of H. sericeum, whereas the sympatric population had the shortest and narrowest leaves (Fig. 4). All fourteen quantitative leaf characters for each allopatric and sympatric populations were showed in Table S1.

Fig. 3
figure 3

Box plot of width at base of leaf lamina (mm) (a), box plot of length of leaf lamina (mm) (b), box plot of width of leaf lamina (mm) (c), box plot of width of basal leaf cell (μm) (d), box plot of width of median leaf cell (µm) (e), box plot of leaf surface area (mm2) (f), box plot of tooth length at base of leaf lamina (µm) (g) and box plot of tooth number at base of leaf lamina (h) in allopatric regional populations of Homalothecium sericeum and H. lutescens and sympatric regional population. Red outline box plot = allopatric regional population of H. sericeum; blue outline box plot = sympatric regional populations; green outline box plot = allopatric regional population of H. lutescens; white-filled box plot = populations from Öland and grey-filled box plots = population from Skåne. Student’s t test for pairwise comparisons (ns = p > 0.05, * = p ≤ 0.05, ** = p ≤ 0.01, *** = p ≤ 0.001, **** = p ≤ 0.0001)

Full size image
Fig. 4
figure 4

Relationship between width and length of leaf lamina of Homalothecium leaf specimens (N = 240) collected from the allopatric populations of H. lutescens (N = 60) and H. sericeum (N = 59) and the sympatric population (N = 121). The regression line of data points from each population is also displayed

Full size image

The PCA based on the leaf morphology data set reveals that 48.54% of the observed variation is explained by the first three factors (PC1 = 19.56%, mostly summarizing variation in general leaf size and traits related to leaf tooth; PC2 = 16.45% mostly reflecting width of leaf cells in different parts of the leaf; PC3 = 12.43% mostly determined by leaf size and length of apical leaf cells; Table 4). A plot of individual component scores along the first principal component (x-axis) and the second principal component (y-axis) (Fig. 5) showed almost complete separation of allopatric populations of H. lutescens and H. sericeum with individuals from the putative hybrid populations somewhat overlapping with the two clusters from allopatric populations, mostly filling the intermediate space between the allopatric populations.

Table 4 Factor loadings for the first three principal axes of a PCA based on leaf morphology of Homalothecium lutescens and H. sericeum from allopatric and sympatric populations
Full size table
Fig. 5
figure 5

Principal component analysis of 14 leaf characters from 240 specimens representing allopatric and sympatric populations of Homalothecium lutescens and H. sericeum. The first two axes (PC1 and PC2) representing together 36% of variation are shown. The colours and shapes of data points correspond to the population of the specimens. Leaf shapes are exemplified for selected leaves (with sample ID no). The encircled surfaces represent 95% confidence level of the sample means with the centre of each cluster marked by a larger symbol. Factor loadings are presented in Table 4

Full size image

Capsule inclination

In the allopatric populations, the capsules of H. sericeum were almost erect (152–180 degrees) whereas the capsules of H. lutescens were markedly inclined (125–154 degrees), in accordance with the generally accepted species characteristics. The capsule specimens from the sympatric populations, however, showed a wider range of variation in capsule inclination from strongly inclined to intermediate to erect (122–180°). The mean of capsule inclinations from the three different populations were significantly different (Table 5). The capsule inclinations falling between 150 and 160° were out of the capsule inclination range of the allopatric populations of both species and thus indicative of a hybrid origin (Table 5). Out of 101 capsule samples with leaf data from the sympatric populations, 44 capsules were H. sericeum-like (43%), 31 capsules were H. lutescens-like (31%), and 26 capsules were intermediate (26%). In the sympatric populations of H. lutescens and H. sericeum 12 out of 26 putative hybrid capsules fell into the morphospace of H. lutescens (46%), 4 into the morphospace of H. sericeum (15%), 9 into the overlapping morphospace of H. lutescens and H. sericeum (35%), and 1 outside the morphospace of both H. lutescens and H. sericeum (4%) (Fig. 6).

Table 5 Capsule inclination of specimens from allopatric and sympatric populations of H. lutescens and H. sericeum collected in Sweden
Full size table
Fig. 6
figure 6

The positions of putatively hybrid sporophytes from the sympatric populations of H. lutescens and H. sericeum superimposed in the PCA from Fig. 5, based on leaf morphology of the maternal gametophytes. The morphospace of leaves from allopatric populations of H. lutescens and H. sericeum are shown as encircled surfaces (blue circle = H. lutescens and yellow circle = H. sericeum). The red circle represents the morphospace of individuals from the sympatric populations. Each square represents a hybrid sporophyte specimen, collected on its corresponding gametophyte (sample ID number). The colours of the square corresponds to different hybrid indications, i.e. hybrid capsules inferred by intermediate capsule inclination (), hybrid capsules inferred by SNP markers but SNPs mainly typical for either H. lutescens () or H. sericeum (), hybrid capsules inferred by both intermediate capsule inclination and SNP markers but SNPs mainly typical for H. lutescens ()

Full size image

SNP variation

In total, 111 SNPs (out of 133 SNPs) distributed across 90 contigs were successfully genotyped. Out of 98 nuclear SNPs, 85 turned out to possess species-specific alleles for either H. lutescens or H. sericeum, and the remaining 16 were polymorphic in sympatric populations of one or the other species, or in one case, both. The 85 species-specific SNPs gametophytes were selected in our analysis to indicate hybrid influence (“admixture”) of sporophytes in sympatric populations (Table 6). Admixture was indicated in sporophytes by (1) a presence of species-specific homozygous SNP expression representing both parental species in the same seta but in different loci (only seen in locus 31, 51, 70, and 80) and (2) heterozygous nuclear SNP expression in one or more loci (i.e. having species-specific alleles for both parental species; only occurring in locus 31, 51, 70 and 80). Analysis of 85 SNP markers in 100 seta samples (13 DNA samples failed to produce SNP genotypes) indicated 53 sporophytes as pure H. lutescens, 14 sporophytes as pure H. sericeum, and 33 as admixed sporophytes (only putative hybrid sporophytes with intermediate capsule inclination and/or admixed sporophytes inferred by SNP markers are shown in Table 7). Thus, no sporophytes appeared to be primary hybrids because heterozygotic expression was rare. The results, rather, suggest secondary back-crossing or other kinds of introgression (admixing). Overall, admixed sporophytes showed biased SNP markers towards either H. lutescens (22 sporophytes = 67%) or H. sericeum (11 sporophytes = 33%). The positions of SNP-inferred admixed sporophytes superimposed on leaf sample of the associated female gametophyte ordination are shown in Fig. 6.

Table 6 Frequency of polymorphism of different forms within and between SNPs of gametophyte individuals of H. lutescens and H. sericeum
Full size table
Table 7 Distribution of species-specific alleles in 85 nuclear SNP loci for 33 out of 100 sporophyte (seta) samples collected from 18 sampling points (Patch No. I-XVIII) from sympatric populations (Hb1-Hb6) that showed signs of hybrid origin by SNP markers or spore intermediate capsule inclination in sympatric Homalothecium populations
Full size table

Discussion

The results from our analyses show a clear signature of hybridization and introgression in sympatric populations of the two Homalothecium species in traits of gametophytes (branch leaves) and sporophytes (capsule inclination) as well as genetic markers (SNP genotypes). These three indicators of hybridization show a non-overlapping pattern, with individuals displaying hybrid character states in one or two indicators, rather than all three. We believe that the non-overlapping pattern of the three indicators of hybridization is explained by extensive introgression and in the following we will discuss this conclusion, what assumptions and limitations are found in the data sets and how the results fit into a wider context.

Morphological evidence of hybridization

Homalothecium lutescens and H. sericeum are sometimes considered difficult to separate (Lieske 2010) since they are closely related and share some common morphological characters (Ignatov and Huttunen 2002). Our results show that allopatric populations of H. lutescens and H. sericeum are morphologically well differentiated, as demonstrated by the PCA (Fig. 5), which confirms the characters that have traditionally been used for species identification. For example, H. sericeum is conspicuously toothed at leaf basal margin and produces more or less straight capsules, whereas H. lutescens is denticulate at leaf basal margin and produces curved capsules (Nyholm 1965; Hofmann 1998). In contrast, gametophytes from putatively hybridizing sympatric populations showed more or less intermediate character states for several leaf characters such as width at base of leaf lamina, width of basal leaf cell, width of median leaf cell, leaf surface area, tooth length at base of leaf lamina, and tooth number at base of leaf lamina (Fig. 3). Notably, the general leaf shape was relatively shorter and narrower in the sympatric populations as compared with any of the allopatric populations (Fig. 4). It could be argued that the morphological expression in our leaf samples represents the wild phenotypes (without prior common garden culturing) and consequently that the phenotypes from the sympatric populations result from some specific environmental conditions, rather than hybridization. However, we want to point out that the morphology of each species, collected in allopatric populations of environmentally different regions such as in Öland and Skåne, does not differ substantially, suggesting that the studied branch-leaf traits are largely species-specific rather than environmentally controlled. Thus, we have reason to believe that the morphological analysis of branch leaves provides evidence for hybridization between H. lutescens and H. sericeum in sympatric populations.

Intermediate spore capsules have been used to infer hybridization in bryophytes and primary hybrids are expected to display characters of both parental species. In our study, we focussed on the inclination of capsules, as this is a major distinguishing character between the species, H. lutescens having inclined capsules and H. sericeum having erect capsules (Hofmann 1998). H. sericeum grows mainly on walls, rocks and tree trunks, whereas H. lutescens grows mainly on soil and sand but rarely epiphytically (Hofmann 1998). The contrasting capsule orientation in the two species is therefore likely explained as a consequence of selection pressures imposed by their particular habitat preferences. Hedenäs (2001) showed that erect and straight capsules types are overrepresented among epiphytes, whereas curved and inclined capsule types are more common among species growing on the ground. A probable explanation is that spore dispersal is most efficient at a horizontal capsule orientation—for epiphytic species with a straight capsule growing on a vertical substrate, the capsule orientation will effectively be horizontal. We confirmed the specific characteristics in allopatric populations, whereas sporophytes displayed a high variability of capsule inclination when the two species grew intermixed in sympatric populations. We see no reason other than hybridization that capsules with intermediate inclination should occur spontaneously in six separate sympatric populations.

Genetic evidence for hybridization

Before discussing the interpretation of the genetic data, it is worth noting that there is a fairly high and variable failure rate for SNP markers in sporophyte samples (see Supplementary materials Appendix II). We must ask if we can trust the markers and why there is such a high failure rate. Primarily, the failure rate for SNP markers can be explained by the low DNA concentration in an individual seta. As far as we know, nobody has tried to extract SNP haplotypes from only setae before, which contains much less tissue than complete sporophytes. We knew beforehand that we were close to the detection limit, although we used a method for SNP genotyping that required low amounts of DNA. The reason for extracting setae was that we wanted to sow out the spores in the spore capsules for germination tests (Sawangproh et al. 2020; Sawangproh and Cronberg, in prep).

It is possible that heterozygotic loci are overrepresented among the failing loci, since each SNP allele only represents half the amount of DNA as compared with the single allele in a homozygotic locus. A low signal level for individual alleles could therefore be a partial explanation for the low incidence of loci with heterozygotic expression. On the other hand, the genotype distribution among diploid sporophytes agrees well with their maternal origin, in cases where several sporophytes have been analysed from the same mother. Typically, rare allelic combinations (and even instances of detection failure) are similar or identical in such sibling sporophytes, supporting the credibility to of detected SNP variants.

Contrary to our initial expectation, the analysis of SNP genotypes in sporophytes with intermediate capsule inclination using a combination of 85 species-specific nuclear markers did not reveal primary hybridization in sporophytes from sympatric populations. Instead, we found only four nuclear SNPs (i.e. locus 31, 51, 70 and 80 from four separate nuclear sequences) out of 85 SNPs that appear to signal secondary hybridization (introgression). Taken alone, a low frequency of heterozygotes is a somewhat weak evidence for hybridization, since it could be argued that these four SNP loci may also be heterozygotic in one or the other of the parental species, although not in our allopatric reference populations. We also observed that individuals with mixed SNP markers generally fell into the leaf morphospace of the parent that had the strongest genomic representation in the SNP genotype, whereas those with both intermediate capsule inclination and mixed SNP markers more clearly fell into the intermediate morphospace (Fig. 6). Finally, sporophytes with intermediate capsule inclination were more commonly borne on gametophytes with a maternal origin from H. lutescens, as indicated by the SNP markers. In a subsequent larger study, comparing hybridization across different stages in the life cycle (Sawangproh et al. 2020), we did find a small fraction of individuals (3.8%) with strongly admixed genomes suggesting primary hybridization.

Evidence from spore capsule inclination

Our study is the first, to our knowledge, to report intermediate sporophytes (capsule inclination) in bryophytes supported by genetic data analysis (SNP markers). Closely related (congeneric) species of most pleurocarps, many acrocarps, peat mosses, and almost all liverworts have similar sporophytes, and there is little chance of recognizing F1 hybrids (hybrid sporophytes) by morphological examination of sporophytes (Shaw 1994; Hofmann 1997; Natcheva and Cronberg 2004). One exception is the closely related species pair Orthotrichum gymnostomum and O. obtusifolium in which Hedderson (1986) inferred hybridization based on the basis of intermediate characters of sporophytes, such as capsule shape, ribbing, seta length, peristome, exothecial cells and stomata. Other studies show that hybrid sporophytes do not necessarily have intermediate shapes but rather tend to be closer or identical to one or the other parent. For example, when crossing Physcomitrium pyriforme (Hedw.) Hampe and Funaria hygrometrica Hedw., the sporophyte morphology, including peristomal structure, appeared to be more similar to the maternal parent (von Wettstein 1923). Naturally occurring hybrid sporophytes among several species of Weissia also tended to show morphologies that were more similar to the maternal parent (Nicholson 1905, 1906), suggesting complex inheritance involving maternal effects, non-nuclear genetic factors and/or nuclear-cytoplasmic interaction (Shaw 1998). In most cases, the spores in such sporophytes have proved to be more or less completely inviable, probably because the parental species were too distantly related (Natcheva and Cronberg 2004) and therefore genomically incompatible.

In our study, intermediate sporophytes tended to occur in individuals with genomes dominated by one of the parents as indicated by SNP markers, and they are clearly not indicative of primary hybridization. In primary hybridization, we expect genes that are responsible for parental differences in sporophyte morphology to be represented by species-specific alleles in the F1 hybrid sporophytes. In such cases, intermediacy may be visible if the traits (e.g., capsule inclination) are controlled by few genes and show co-dominance. In secondary backcrossing (introgression), the alleles that are responsible for the sporophyte morphology may be recombined in a different way. If inheritance of genes controlling sporophyte morphology is simple (involving few genes) and these genes not strongly linked to other traits, their distribution could be more or less random in a population of introgressed individuals and, as we experience in our study, decoupled from other morphological indicators of hybridization. We therefore urge against uncritical use of intermediate sporophytes as evidence for primary hybridization.

Introgression—the ghost of hybridization

It is sometimes assumed that individuals of hybrid origin should also display intermediacy in vegetative morphological characters, but this assumption may be misleading and result in failure to recognize hybridization and introgression, especially if supporting molecular data is missing (Rieseberg et al. 1993). Rieseberg (1995) suggested that non-intermediate morphology is common where past hybridization and intensive introgression occur in sympatric populations of vascular plants. Bryological studies have shown that morphology of hybrid gametophytes is more similar to either the female (more common) or the male parent (Nicholson 1910; Allen 1935, 1945; Pettet 1964; Holmen and Scotter 1971; Delgadillo 1989; Shaw et al. 2012, 2013; Werner et al. 2014) or displays to some degree a mixed combination of parental characters (Burgeff 1943; Boisselier-Dubayle et al. 1998; Såstad et al. 1999, 2001; Flatberg 2005; Orzechowska et al. 2006) so that only occasional traits seem to be misplaced (Cronberg and Natcheva 2002; Natcheva and Cronberg 2007a, b).

Our interpretation is that introgression is extensive in the sympatric populations of Homalothecium, most of the admixed individuals display a majority of SNPs from one parent and occasional SNPs that appear to be misplaced. Pereira et al. (2019) recently found evidence for extensive introgression in the sibling moss species Syrrhopodon annotinus and S. simmondsii that occur in sympatry but in different habitats in lowland Amazonian rainforests. They observed that the two species show a clear morphological differentiation and a low but significant genetic differentiation using 183 SNPs, but interspecific pairs of individuals were more closely related when they were geographically closer, suggesting that isolation-by-distance is stronger than the interspecific reproductive barrier and pointing to extensive introgression between the two species. Our observation of occasional misplaced SNPs resembles earlier studies of peat mosses, which display limited mixing of genetic markers in hybridizing populations. Without individuals that show a strong mixing of the parental genomes, it is difficult to demonstrate hybridization in the absence of additional evidence. In studies of the species pair Sphagnum capillifolium and S. quinquefarium, Natcheva and Cronberg (2007b) were able to show that hybrid capsules could produce viable spores, which turned out to inherit the chloroplast genes from the maternal parent, whereas most of the nuclear markers came from the father. Since a very small fraction of spores from the hybrid capsules germinated, it was suggested that only those spores that for stochastic reasons happened to recombine a majority of the genome from one parent were viable, i.e. most of the spores were aborted due to incompatibilities between the genomes. In other words, individuals with a more balanced genomic composition of the parental species may be absent when homoploid hybridization takes place. A similar situation may occur in the Homalothecium species, but we could imagine other, mutually non-exclusive, explanations for limited genomic mixing in this study. Firstly, both species are nannandric, which means that dwarf males can be generated by spores that germinate on normal-sized females. Spore dispersal by wind in bryophytes is strongly leptokurtic (Miles and Longton 1992), which means that many spores may actually land directly on the sporulating maternal gametophyte. Rosengren et al. (2016) inferred that repeated back-crossing by son-mother fertilization occurs from time to time in H. lutescens. For statistical reasons, each back-crossing event will mean that half of the paternal genetic variation is lost. Since dwarf males of Homalothecium are annual or possibly biannual and therefore have a high turnover rate, this kind of genome homogenisation is bound to be faster than in species with normal-sized males or females. On the other hand, dwarf males are most likely to be exposed to different selection pressures than normal-sized individuals, which is important since both represent the haploid generation—no variation is hidden as recessive alleles. Secondly, our method of selecting species-specific SNP markers may also play a role in our potential to detect hybrids. If hybridization has occurred episodically in the past, we may have excluded markers that occur in parts of the genome that are interchangeable between the species and therefore potentially polymorphic in one or both species. Recent studies of various organisms at the genomic level have stressed the semipermeable nature of species boundaries (Harrison and Larson 2014), meaning that some parts of the genomes are mixed between sister lineages in hybrid zones and others not. Finally, our approach to include only individuals with mature sporophytes from the sympatric populations may mean that we have excluded individuals that are sterile (no sex organs) or failing to produce sporophytes after cross-fertilization (aborted sporophytes). Such individuals may possess more strongly mixed genomes. In additional data sets from the same sympatric populations we include non-sporulating gametophytes, a few of which appear to be more close to primary hybrids, and we also perform germination tests to estimate spore germination frequencies (Sawangproh et al. 2020; Sawangproh and Cronberg, in prep).

Despite limited genomic mixing and strong genomic representation of one or the other parent in admixed SNP genotypes, individuals from the sympatric populations had characteristics that were out of range of any of the parental species from the allopatric populations, in leaf shape, having smaller leaves. These individuals also tended to look habitually intermediate (in terms of branching and general growth pattern) although this was not recorded. Similarly, hybrids of Sphagnum capillifolium and S. quinquefarium differed systematically from the parental species in a controlled growth experiment with different watering and shading regimes. The hybrids performed worse under drought stress but tended to regenerate faster than either of the parental species (Natcheva and Cronberg in Natcheva 2006) and as a possible explanation, it was proposed that adaptive gene combinations may have been disrupted during recombination.

Because species of the genus Homalothecium sometimes grow sympatrically, it has been speculated that hybridization plays an evolutionary role within the genus (Hofmann 1998; Hedenäs et al. 2009). Variable gametophyte characters could suggest multiple origins and interspecific crossing; for example, Homolothecium aeneum shares characters with H. nevadense and H. aureum as well as H. pinnatifidum, without having any diagnostic characters of its own. To investigate this, Hofmann (1997) made a biometrical analysis, which revealed that branch leaves of H. aeneum are intermediate between H. aureum and H. nevadense, in accordance with the theory of a hybrid origin and also similar to our results from the sympatric populations of H. lutescens and H. sericeum. Although H. lutescens and H. sericeum have been suspected to hybridize, and specimens with intermediate gametophyte morphology have been recognized as H. lutescens var. fallax (Hofmann 1998, see Table 1), hybrid sporophytes have not been documented. We cannot verify that var. fallax is a hybrid taxon because the critical characters do not completely agree with the morphospace of individuals in our sympatric populations. The original description (by Philibert in Schimper 1876, as H. fallax) states that the vegetative characters resemble H. philippeanum but gives no specific information about the traits of normal leaves. Having inspected the type specimen, Hofmann (1998) notes that the only difference between H. fallax and H. lutescens is the erect, straight capsule, which led her to suggest that H. fallax is a hybrid between H. lutescens and H. sericeum, or possibly that the straight capsule is caused by a rare mutation. She does not mention anything regarding intermediate leaf characters, although she had observed occasional specimen of H. lutescens with variable capsule shapes from a number of countries in Europe. Nyholm (Nyholm 1965; Table 1) having not seen the type of H. fallax states that Scandinavian plants referred to this taxon has been found from Öland and some other regions, not including Skåne. She mentions that they have more or less creeping stems with few rhizoids and crowded erect branches, like a robust H. sericeum. In agreement with our observations from sympatric populations, she points out that the angular cells have less denticulate and shorter teeth than H. sericeum (i.e. intermediate between the species). She also mentions that the capsule is almost erect and with shorter operculum than H. lutescens. To summarize, it is difficult to directly link var. fallax to hybrid origin(s), but it is probable that the taxon sometimes has been applied for putative hybrids.

The predominance of SNP markers typical of one or the other of the parental species in admixed sporophytes suggests that introgression is bidirectional—meaning that hybrid gametophytes can backcross repeatedly with either H. lutescens or H. sericeum. Bidirectional hybridization is a possible explanation why some leaf samples of gametophytes from sympatric populations showed similar morphology typical to one of the parents although the molecular data revealed signs of hybridization. In sympatric populations on Öland, H. sericeum generally grows above H. lutescens. Therefore, H. sericeum is more likely to act as paternal plants, since the sperm cells are more easily transported downwards with flowing or splashing water. Also spores of H. sericeum could be expected to more easily disperse downwards at the local scale, and therefore deposit on female shoots of H. lutescens and dominate the population of dwarf males. A tendency for spores to be deposited close to the sporulating female shoot could be even more pronounced if spore dispersal is negatively affected by intermediate capsule inclination or peristomal dysfunction as a consequence of hybridization.

Conclusions

Our study shows that H. lutescens and H. sericeum are incompletely reproductively isolated and that interspecific hybridization and introgression occur in sympatric populations. Branch leaves of Homalothecium plants in sympatric populations show intermediate morphology in a number of characters but are generally smaller in size compared with those of either species in allopatric populations. Contrary to our expectations, intermediate capsule inclination is not a morphological marker for primary hybridization, but rather an indication of diffuse introgression. Most of the sporophyte genotypes that contained SNP markers from both species displayed asymmetrical genomic contribution from the two parents, indicating that hybrid sporophytes with low recombination of parental genomes can develop normally on maternal gametophytes and perhaps also produce viable hybrid spores. Overall, sometimes stronger morphological affinity of branch leaves to one or the other of the parents of admixed individuals and asymmetrical SNP markers in sporophytes suggest bidirectional hybridization and introgression. It appears that ecological isolation is a stronger barrier against hybridization than post-zygotic genomic incompatibilities—in the vast majority of populations of H. lutescens and H. sericeum in south Sweden, the species do not occur together at all or at least not in immediate vicinity.