Grimmia shevockii and G. insolita, two new species endemic to California

ABSTRACT Introduction California is a biodiversity hotspot for bryophytes, and its northern counties are among the most endemic-rich areas for bryophytes in North America. Having 29 species in California, Grimmia Hedw. is the third richest moss genus in the state, but recent fieldwork suggests that this number is an underestimate. The identity of some specimens endemic to the Klamath Mountains of California and previously identified as G. brevirostris R.S.Williams remains unclear, and they potentially represent a new species. Two further specimens of Grimmia from Siskiyou County, California, do not correspond to any species described so far. Here we present morphological and phylogenetic arguments for recognising these plants as two new moss species in the genus Grimmia subg. Orthogrimmia Schimp. Methods Morphological characters across potential new species were reviewed and compared with those of related species. Phylogenetic analyses based on the plastid trnS–trnF and nuclear ITS regions were conducted. Key results and conclusions The molecular and morphological analyses revealed two new species of Grimmia for California, which are described and illustrated: G. shevockii J.Muñoz, I.Solano & D.Quandt and G. insolita J.Muñoz, I.Solano & D.Quandt. The former is resolved in a monophyletic clade with high support, whereas the latter has incongruent positions in the plastid and nuclear phylogenies, thus it is hypothesised that it could have a hybrid origin. Both of the two new species and G. hamulosa undoubtly pertain to Grimmia. The richness of the bryophyte flora of California provides a clear warrant for future research.


Introduction
California contains a great diversity of habitats, which makes it one of the most species-rich territories in the USA (Burge et al. 2016).Its moss flora is indeed rich, hitherto including ∼640 species (580 in Norris and Shevock 2004; updated by Shevock, personal communication 2022), and new species from the area are still being described.Additionally, its northern counties are among the most endemic-rich areas for bryophytes in North America (Carter et al. 2016).From 2011 to 2016, 13 new species of bryophytes were described, which suggests that the bryophyte flora of California deserves further research.
The family Grimmiaceae is the second most speciesrich in California, having 61 species.Within this family, Grimmia Hedw.has been considered to include 71 (Muñoz and Pando 2000), 95 (Greven 2003) or 51 (Maier 2010) species in total, 17 of which were described after 2000 (https://www.tropicos.org).Grimmia is the third largest moss genus in California, having currently 29 accepted species (Norris and Shevock 2004).
Over the past few decades, James R. Shevock and collaborators have collected bryophytes intensively throughout California, including many Grimmia specimens.Their efforts have resulted in new additions and changes to the Californian bryoflora.
In the summer of 2013, R. I. Hastings, D. R. Toren, J. R. Shevock and C. P. Dillingham, while conducting fieldwork at the area where the type of Grimmia brevirostris R.S.Williams (Williams 1920) was collected in Plumas National Forest, found G. hamulosa Lesq.(Lesquereux 1868) growing with plants that they considered to be another taxon and that they interpreted as G. brevirostris (Dillingham 2015;Shevock et al. 2021).Consequently, they elevated G. brevirostris from synonymy with G. hamulosa as "a species worthy of recognition" (Shevock et al. 2021).Those two taxa had been synonymised by Muñoz (2000) after examining their types, and this treatment was followed in the catalogue of Californian mosses by Norris and Shevock (2004).
Later, Maier (2010) excluded both species from Grimmia based on morphological traits.According to her, both species had the peristome teeth ornamented with fine papillae arranged in longitudinal rows, a morphological character not found in other species of Grimmia, and thus she excluded both from Grimmia but without formally proposing any new generic placement.In the case of G. hamulosa, Maier moreover suggested a connection to the genus Schistidium, as she considered the guide cells to be arranged in a medial position, contrary to what is typical in Grimmia, where guide cells are commonly the ventral layer.However, the generic position of G. hamulosa and G. brevirostris seems to be within Grimmia, as the guide cells in the types of both these species, although not greatly differentiated, are clearly situated in a ventral position, and furthermore the setae are longer than the exserted capsules and have relatively large calyptrae, whereas in Schistidium the setae are shorter than the capsules and the calyptrae are very small, barely covering the operculum.
On several occasions, Shevock and colleagues collected plants with blackish, crisped leaves in the northern counties of California, and for these they also used the name Grimmia brevirostris (Shevock et al. 2021).The specimens were sent to the senior author, who realised that they differ from the type of G. brevirostris and indeed they pertain to a species not yet described.
Besides these controversial previous specimens, in 2020 Shevock collected in Siskiyou County, California, another two samples with flexuous and yellowish-or orange-green, mostly muticous vegetative leaves but having long hairpoints in the perichaetial leaves.These were considered to be an undescribed species of Grimmia, as reported by Shevock et al. (2021).
All the specimens included in the present study belong morphologically to Grimmia subg.Orthogrimmia Schimp., which includes the species of Grimmia with straight setae and leaves which are V-shaped in transverse section and with flat margins (Muñoz 1998).The subgenus comprises two sections: sect.Donnianae (Loeske) J.Muñoz, in whose species the proximal marginal cells of the leaves have thin-walled cells which are all alike, mitrate calyptrae, and annulus compound and revoluble; and sect.Montanae I.Hagen, in which the proximal marginal cells of the leaves have transverse walls thicker than the longitudinal walls, cucullate calyptrae, and the annulus simple and persistent.
Against this background, and given the possibility that the diversity of Grimmia in California remains underestimated, the aims of the present study were: (i) to resolve, using morphological and molecular characters, the discrepancies in the taxonomic status of G. brevirostris and G. hamulosa, and thus clarify how many species are involved in this puzzle and confirm whether the inclusion of these two species in Grimmia is appropriate; and (ii) to investigate whether the two specimens from Siskiyou County indeed represent a new species of Grimmia, and if so, to explore its systematic relationships.

Plant material
For the present study, we added 28 additional Grimmia specimens to the trnS-rps4-trnF and 20 to the ITS matrices of Hernández-Maqueda et al. (2007).Data on the specimens newly sequenced for this study can be found in Appendix 1.
A complete matrix for the plastid data, including 87 taxa, was first used to assess the generic position of the studied specimens.Because previous studies showed that the nuclear ITS and the chloroplast rps4-trnF regions were incongruent in big datasets for Grimmia (Hernández-Maqueda et al. 2008a), we later used a reduced matrix with 35 taxa of Grimmia subg.Orthogrimmia for both regions.Thus, we had two independent reduced matrices (henceforth Orthogrimmia matrices): the plastid Orthogrimmia matrix and the ITS Orthogrimmia matrix.GenBank accessions used for the Orthogrimmia phylogeny can be found in Appendix 2.

DNA extraction, PCR amplification and sequencing
For each specimen, DNA was extracted from some dried stem apices of the gametophyte, using the columns method (Dellaporta et al. 1983).The DNA was purifed with the Thermo Scientific GeneJET Gel Extraction and DNA Cleanup Micro Kit (with the reference K0832; Thermo Fisher Scientific, Waltham, MA, USA).The final extracted DNA of the samples was eluted with 20 μL of the elution buffer prepared by the manufacturer.
The amplification cycles were performed using an Eppendorf Mastercycler.For the plastid region, the cycles consisted of 5 min at 94°C followed by 35 cycles of 94°C for 2 min, 1 min at 55°C, and 1 min at 72°C , with a final 8 min extension at 72°C.For the ITS, the cycling conditions were 5 min at 94°C followed by 35 cycles of 94°C for 30 s, 45 s at 58°C, and 1 min at 72°C , with a final 8 min extension at 72°C.All the PCR products were gel cleaned and sequenced at Macrogen Spain (Madrid, Spain).

Phylogenetic analyses
First, the chromatograms of each sequence (forward and reverse) were checked in Geneious Prime 2022.1 (https://www.geneious.com)to eliminate chromatograms with baseline noise before creating the consensus sequence.Next, primer sequences were trimmed from the consensus sequences.Then, our consensus sequences and the datasets (plastid and nuclear matrices) were automatically aligned in MAFFT version 7 (Katoh and Standley 2013), selecting the 'Align full length sequences to an MSA' option.After the automatic alignment, the sequences were manually adjusted using PhyDE 0.9971 (Müller et al. 2006).For the phylogenetic analyses, mutational hotspot areas (poly-A/T microsatellites) were eliminated and inversions were included as reverse complements.
Furthermore, the datasets were also analysed, employing a simple indel-coding approach (Simmons and Ochoterena 2000) using the option SeqState (Müller 2004) in PhyDE 0.9971.Thus, we had two matrices per dataset: the original matrix, where the indels are considered missing information; and the combined matrix, including the original matrix with the indel matrix appended as binary data.
We estimated the phylogenetic relationships using Bayesian inference (BI) in the software MrBayes version 3.2.7a(Ronquist et al. 2012) through the CIPRES Science Gateway (Miller et al. 2010).We performed two independent analyses for each matrix, one for the original alignment and another for the indel matrix.As in the study by Hernández-Maqueda et al. (2007), the substitution model used for the original alignments without indels coded was the Generalised Time Reversible model (GTR + I + G), and for the indel partition, the binary model (F81).To analyse the datasets, we used the Markov chain Monte Carlo approach, and we performed two independent runs with four chains each for 10 7 generations, with a sample frequency of 1000 generations and a burn-in of 25%.
Additionally, to compare the BI results, we performed maximum likelihood (ML) analyses via RAxML-NG v 1.0.1 (Kozlov et al. 2019) with a bootstrap analysis (1000 replicates) to obtain the nodal support, also through the CIPRES Science Gateway.The alignments had the same evolutionary models as implemented in the BI analysis.
The trees obtained from the BI and ML analyses were rooted and compiled in TreeGraph2 (Stöver and Müller 2010).We used MEGA 11 (Tamura et al. 2021) to obtain variable and parsimony-informative positions in both datasets (with and without indels).
A first analysis was run using the complete plastid matrix and Racomitrium Brid.as outgroup to check the systematic position of the studied specimens, particularly if Grimmia hamulosa pertains to Grimmia or if it should be reaccommodated in another genus, as proposed by Maier (2010).
Finally, to check for congruence between the plastid and the nuclear regions, the Orthogrimmia matrices from both regions were concatenated and visually and statistically tested for incongruences.We used the incongruence length difference (ILD) test (Farris et al. 1994) implemented in PAUP version 4.0a168 as 'Partition homogeneity (ILD) test' (Swofford 2003).

Morphological examinations
Morphological examinations were carried out using Wild M8 and Leica S9i stereomicroscopes (the latter equipped with an internal camera of 10 megapixels; both Leica, Wetzlar, Germany), and a Nikon Labophot-2 light microscope (Nikon, Tokyo, Japan) equipped with a Euromex CMEX 12 camera (Euromex, Antwerpen, Belgium) of 12 megapixels.
Important morphological characters with which to identify Grimmia species rest on gametophytic traits such as leaf shape, presence of hairpoints, leaf margin curvature (plane, recurved or incurved), walls of the basal marginal cells (all alike or the transverse walls thicker than the longitudinal walls), and shape of the leaves in cross-section (concave or keeled), and on sporophytic traits such as seta orientation (straight or curved) and capsule form (ovoid or ventricose).

Results
Analysis of the type specimens of Grimmia brevirostris and G. hamulosa We studied the types of Grimmia hamulosa (The New York Botanical Garden Virtual Herbarium, http:// sweetgum.nybg.org/science/vh/specimen-details/?irn= 589243) and G. brevirostris (The New York Botanical Garden Virtual Herbarium, http://sweetgum.nybg.org/science/vh/specimen-details/?irn=280962), and our morphological results support the view of Muñoz (2000) that they are synonyms.Particularly noteworthy in the types of both taxa are the homomallous disposition of apical leaves, the bistratose limb, and the elliptical-crescent shaped costa with a distinct ventral ridge.
The second group of specimens identified as Grimmia brevirostris (Dillingham 2015;Shevock et al. 2021) and characterised by having crisped, blackish leaves, is indeed different from any other described Grimmia and so may be considered to represent a new species, and this is also confirmed by the results of phylogenetic analyses presented here.

Phylogenetic analyses
As shown in Table 2, when indels were coded as informative data, the number of parsimony-informative positions increased in all datasets.As expected, the number of parsimony-informative positions was nearly twice as high in the ITS Orthogrimmia compared with the plastid Orthogrimmia matrix (without indels coded as characters), despite the lower number of alignment positions.
The ML tree inferred from the complete matrix for the plastid region is shown in Figure 1, with bootstrap support (ML) above the branches and posterior probabilities (BI) below them.The values at the right correspond to the analyses with indels coded as characters.Both analyses resolved the samples of Grimmia hamulosa, G. insolita (newly described here) and G. shevockii (initially identified as G. brevirostris, and formally described here as new), as reciprocally monophyletic and all pertain to Grimmia sect.Montanae (subgenus Orthogrimmia) with strong support (posterior probability [PP] 1/1, bootstrap support [BS] 88/94).The specimens (G5 and G6 in Figure 1), which correspond to the newly described G. insolita, form a monophyletic group with G. hamulosa and G. serrana J. Muñoz, Shevock & D.R.Toren (Muñoz et al. 2002), also with strong support (PP 1/1, BS 85/78).The new species G. shevockii is strongly resolved in a monophyletic clade with high support (PP 1/1, BS 100/100), and it is sister to the clade including G. alpestris (F.Weber & D.Mohr) Schleich., G. caespiticia (Brid.)Jur. and G. reflexidens Müll.Hal., supported only by PP values (PP 0.99/0.97).Grimmia shevockii is clearly separated from G. hamulosa, which forms a monophyletic clade supported in both analyses (PP 1/1, BS 84/85).
The ILD test results showed a significant incongruence between the nuclear and the plastid regions in the combined Orthogrimmia dataset (p = 0.006), hence this concatenated dataset was not used for phylogenetic analyses, and each Orthogrimmia matrix was analysed independently.The slightly different topologies of both trees (plastid and ITS Orthogrimmia trees) confirm this heterogeneity for Grimmia sect.Montanae (Figure 2): G. insolita (sample G6 did not work for ITS) was resolved in different groups in both phylogenies, and thus we performed another ILD test without this species.This analysis showed no significant incongruences between the plastid and nuclear regions (p = 0.327), and the combined matrix without this potential hybrid species was analysed with both BI and ML methods (Figure 3).
As for Grimmia insolita, Figure 2 shows a clear incongruence between the chloroplast and the ITS phylogenies for Grimmia sect.Montanae.Whereas in the chloroplast tree, G. insolita is in the same clade as G. hamulosa and G. serrana with high support (PP 1/1, BS 85/84), in the ITS phylogeny it is closely related to G. shevockii, with strong branch support (PP 1/1, BS 99/96).Grimmia shevockii is clearly resolved in a monophyletic clade in both phylogenies with maximum branch support (PP 1/1, BS 100/100).As for G. hamulosa, the monophyletic clade is only supported in the chloroplast phylogeny (PP 1/1, BS 85/86) (see Figure 2).Furthermore, G. orbicularis also has an incongruent position in both trees, as it  is sister to G. crinita Brid.with moderate support in the ITS phylogeny, whereas is sister to G. crinita and G. capillata De Not. in the chloroplast tree (see Figure 2).Finally, the phylogeny reconstructed when both datasets were combined (Figure 3) for Grimmia sect.Montanae resolves the position of G. shevockii in a monophyletic clade with maximum support (PP 1/1, BS 100/100), sister to some specimens of G. alpestris from California (USA).These 'G.alpestris' from California cluster separately from Italian G. alpestris with high support in both analyses (PP 1/1, BS 84/88) and clearly pertain to an undescribed taxon.However, lack of mature, complete sporophytes caused us to refrain from formal description, as gametophytically they are inseparable from European G. alpestris.Lastly, G. hamulosa is resolved with high support values (PP 1/1, BS 95/95) in a monophyletic clade.
(Figure 4) Diagnosis.Leaves subulate, flexuous, usually muticous except the perichaetial ones, where the leaf apex truncates into a long, terete hairpoint; basal marginal cells with transverse cell walls thicker than the longitudinal walls, and basal juxtacostal cells long and narrow, with straight and thick walls; capsules without stomata at the urn base.
Type.USA, California: Siskiyou County, Russian Wilderness, Klamath National Forest, Scott River Drainage.
Etymology.The name refers to its strange nature: the species seems to be of hybrid origin, and at first look it is reminiscent of species in sect.Donniana, to which, however, it is not related.Habitat and distribution (Figure 5).Grimmia insolita grows on partially shaded granitic rocks in subalpine conifer and mixed conifer forests.It is known only from two close localities in northernmost California.
(Figures 6, 7) Diagnosis.Differs from any other Grimmia in its strongly curled, blackish, muticous leaves of wiry aspect, with a prominent costa bulging on the dorsal side, which in cross-section shows uniform cells with strongly thickened walls.
Habitat and distribution (Figure 5).Grimmia shevockii grows on sunny exposed rockslabs and boulders of both granitic and metamorphic rocks in subalpine conifer and mixed conifer forests dominated by Abies magnifica A.Murray, Pinus albicaulis Engelm., Pinus monticola Douglas ex D.Don, Pinus balfouriana Balf., Picea breweriana S.Watson or Tsuga mertensiana (Bong.)Carrière, primarily where water from snowmelt flows over the plants for an extended period.Currently it is known from around two dozen localities in northernmost California.Populations are on public lands administered by the US Forest Service, and the majority of populations are within National Wilderness Areas, the most restrictive and protected landscapes within the USA, and therefore this species is not of conservation concern, even though it is endemic to the Klamath Mountains of California (J.Shevock, personal communication).Description.Plants dark olive-green.Stems erect, to 1.40 cm long, central strand lacking.Leaves erect and appressed, homomallous towards the stem apex when dry, homomallous when moist, 2.3-3.0 × 0.25-0.40mm, lanceolate, acuminate, tubulose, concave, not plicate; margins plane to incurved; costa semielliptical with a ventral furrow, narrower proximally, differentiated from lamina, ventral layer 8-10 cells wide in cross-section; lamina 2(3)-stratose in the distal 2/3, not pseudopapillose; distal cells 5.0-12.5 μm long, oblate to rectangular, not bulging, not papillose, walls sinuous; proximal juxtacostal cells 7.5-12.5 × 2.5-12.5 μm, rectangular (2-4:1), walls thick and straight; proximal marginal cells 15.0-37.5 × 5-12.5 μm, rectangular (1.5-2.0:1), the transverse walls thicker than the longitudinal walls; hyaline hairpoints terete, straight, erect, to 0.4 mm long, entire to weakly denticulate.Propagula lacking.

Discussion
Grimmia shevockii, specimens of which were originally named G.brevirostris in Dillingham (2015) and Shevock et al. (2021), is a very distinct species characterised by its strongly curled, blackish and muticous leaves of wiry aspect, and a prominent costa consisting of uniform cells with strongly thickened walls (see Figures 6, 7).All our phylogenetic analyses resolved G. shevockii within Grimmia subg.Orthogrimmia (see Figures 1-3).Most of the collections of Dan Norris in UC were originally identified as G. incurva, a moss not previously reported as occurring in California but reported from western North America in the Rocky Mountains of Colorado.Clearly Norris was familiar with G. hamulosa from his many collections of this very distinctive Grimmia in the Sierra Nevada, and realising that this was a different taxon, applied the name G. incurva to accommodate them.However, in G. incurva the proximal marginal cells of the leaves have thin walls all alike, and the costa is delicate, with a ventral layer 2 cells wide, whereas in G. shevockii the transverse walls of the proximal basal cells are thicker than the longitudinal ones, and the strong, wiry costa is 6-8 cells wide in its ventral layer (see Figure 6).
Grimmia hamulosa is a Pacific Coast endemic morphologically close to G. serrana, with which it is resolved in a well-supported clade in the combined chloroplast and ITS analysis (see Figure 3).Both have a semi-elliptical costa with a narrow furrow in the ventral side, a costa morphology very infrequent in Grimmia, but G. serrana has multistratose intramarginal bands, unique to this taxon.
Grimmia insolita is a rather puzzling species that macroscopically resembles species in Grimmia sect.Donniana by having leaves with margins plane to the base and slightly incurved apically.However, its marginal basal leaf cells have transverse cell walls thicker than the longitudinal walls, which points to sect.Montana instead.Its leaves are subulate, flexuous when dry, and usually muticous except the perichaetial leaves, where the leaf apex truncates into a long, terete hairpoint as in Coscinodon horridus (Muñoz & H.Hespanhol) Hugonnot, R.D.Porley & Ignatov (Muñoz et al. 2009).The incongruence between the chloroplast and nuclear phylogenies (see Figure 2) suggests a possible hybrid origin of G. insolita.Previous studies have shown cases of reticulate evolution in other species of Grimmia, such as G. tergestina Bruch & Schimp.and G. beringiensis Ignatov & Ignatova, based on different tree topologies using nuclear and plastid markers (Hernández-Maqueda et al. 2008a;Ignatova et al. 2016).These potential hybrid species could be further evidence of the process of interspecific hybridisation in bryophytes, which has long been considered by bryologists as a rare evolutionary process for bryophytes (Natcheva and Cronberg 2004;Sawangproh and Cronberg 2021), although many cases of hybridisation in mosses are known today (Ignatov et al. 2019).
Finally, the samples of Grimmia alpestris from Italy and from California clustered in different groups with high support (see Figure 3).Both molecular and morphological differences indicate that they are probably different species.The lack of material of enough quality in the Californian samples, particularly mature sporophytes, prevent us from formally describing them as a new taxon.

Figure 1 .
Figure 1.The maximum likelihood (ML) tree inferred from the complete trnS-trnF region with ML bootstrap support shown above the branches (with indels coded = right, without indels = left) and posterior probabilities below (with indels coded = right, without indels coded = left).Red boxes show the main targets of this study.

Figure 2 .
Figure 2. The maximum likelihood (ML) trees of the chloroplast and ITS regions for Grimmia sect.Montanae with ML bootstrap support shown above the branches (with indels coded = right, without indels = left) and posterior probabilities below (with indels coded = right, without indels coded = left).Sample G6 did not work for ITS.

Figure 3 .
Figure 3.The maximum likelihood (ML) tree of the combined chloroplast and ITS regions for Grimmia sect.Montanae.Bayesian inference and ML support values are indicated above and below the branches respectively (with indels coded = right, without indels = left).Red boxes show the main targets of this study.Note that the potential hybrid Grimmia insolita was excluded from the analysis.

Table 1 .
Sequences of the primers used in the present study.