There is more to maerl than meets the eye: DNA barcoding reveals a new species in Britain, Lithothamnion erinaceum sp. nov. (Hapalidiales, Rhodophyta)

ABSTRACT Due to the high plasticity of coralline algae, identification based on morphology alone can be extremely difficult, so studies increasingly use a combination of morphology and genetics in species delimitation. A DNA barcoding study was carried out on maerl-forming coralline algae using the mitochondrial cytochrome oxidase 1 gene, CO1, and the plastid gene, psbA, on field specimens from Falmouth and Oban together with herbarium specimens from the Natural History Museum, UK, and the Smithsonian Institution, Washington, USA. Results revealed the presence in the north of Britain of a new species, Lithothamnion erinaceum Melbourne & J. Brodie, sp. nov., which was previously misidentified as Lithothamnion glaciale. The results also indicated that Lithothamnion lemoineae, which had earlier been recorded from Britain, was not present. One of the biggest concerns at present is how organisms will respond to climate change and ocean acidification, and it is imperative that investigations are put on a firm taxonomic basis. Our study has highlighted the importance of using molecular techniques to aid in the elucidation of cryptic diversity.


Introduction
Non-geniculate coralline algae are a morphologically diverse group with red pigmentation and calcified cell walls (Cabioch & Giraud, 1986), currently placed in various orders of the subclass Corallinophycidae (Le Gall & Saunders, 2007). Unlike geniculate coralline algae (e.g. Corallina), they lack non-calcified articulations between calcified segments. Non-geniculate coralline algae exist as crusts or free-living thalli (rhodoliths and maerl). Free-living forms can aggregate to form structurally and functionally complex habitats, maerl beds, that are known to support a high level of biodiversity including commercial species of scallop (Hall-Spencer et al., 2003) and rare and unusual species (Steller et al., 2003;Peña et al., 2014a). Maerl beds are found from high to low latitudes and in the low intertidal zones to 150 m depth (Foster, 2001). With their sensitivity to physical disturbances and slow growth rates (average 1 mm yr -1 ) (Freiwald & Henrich, 1994;Blake & Maggs, 2003;Wilson et al., 2004), maerl beds are seen to be ecologically fragile and a non-renewable resource. The  (Maddock, 2008), and appear on the OSPAR list of threatened and/or declining species and habitats (Region III) (OSPAR, 2008). Maerl beds are also, under the JNCC interpretation of the EC Habitats Directive, a protected habitat within the Annex I category 'sandbanks which are slightly covered by seawater at all times' and Shallow inlets and Bays (JNCC, 2007). Maerl beds already have to withstand human-induced pressures, e.g. dredging (Hall-Spencer & Moore, 2000) and fish farming (Hall-Spencer et al., 2006), that inhibit their ability to provide complex habitats. With the added stressor of climate change these organisms will be under mounting pressure.
As coralline algae utilize the most soluble polymorph of calcium carbonate, high Mg-calcite, to form their skeletons (Andersson et al., 2008), they are thought to be more susceptible to ocean acidification, the decreasing carbonate saturation state in the ocean caused by increasing levels of anthropogenic CO 2 , than other calcifying organisms (Martin et al., 2008). This has led to predictions that calcified macroalgae will be impacted by future global change (Brodie et al., 2014). However, the effect of climate change is species-specific (Ries et al., 2009;Noisette et al., 2013) and some species have shown evidence of acclimation to future CO 2 conditions (Ragazzola et al., 2013). It is therefore crucial to put any analysis on a firm taxonomic basis to assess potentially subtle differences in physiology in order to be able to determine how coralline algal habitats will be affected by future climate change.
Some Rhodophyta are extremely difficult to discriminate taxonomically as many specimens lack key identifying features and have highly variable morphologies (Carro et al., 2014). Non-geniculate coralline algae can be particularly difficult to identify using morphological characteristics alone as they acclimate to their local habitats and are known to have convergent morphology (Steneck, 1986). This is due to a wide range of environmental factors that are known to affect maerl and hence the structure of the maerl bed (Wilson et al., 2004;Carro et al., 2014;Dutertre et al., 2015). Since Saunders (2005) pioneered the use of DNA barcoding for red macroalgae, which was soon adopted by Robba et al. (2006), there has been an increase in the acceptance of molecular techniques as a tool for species identification. Since then the use of molecular identification has revealed cryptic diversity in all orders of red algae tested (Robba et al., 2006), highlighting the misidentification of species due to misleading morphologies (Hind et al., 2014), and helped to uncover evolutionary history (Hind & Saunders, 2013). Species identification studies of maerl beds in Europe (Carro et al., 2014;Pardo et al., 2014;Hernández-Kantún et al., 2015;Peña et al,. 2015b), the Caribbean (Peña et al., 2014b) and Brazil (Costa et al., 2014) have shown hidden diversity. For example, eight species of maerl-forming taxa, including five Lithothamnion species, were identified in maerl beds in Guadeloupe, Caribbean. No sequence matches were found, with only Lithothamnion cf. ruptile (Foslie) Foslie identified based on a morphological assessment. This, therefore, highlights the previous lack of understanding of the diversity of Lithothamnion (Peña et al., 2014b). A similar underestimation of diversity has been found in maerl beds elsewhere, such as the Iberian Peninsula. These beds were thought to contain L. corallioides and Phymatolithon calcareum, but a third species, P. lusitanicum V. Peña, was also identified (Peña et al., 2015a) gradually replacing both L. corallioides and P. calcareum southwards.
These studies have shown how DNA barcoding can successfully attribute coralline algae to separate species, regardless of morphology (Carro et al., 2014), and revealed the cryptic diversity within Lithothamnion. Studies such as Bittner et al. (2011) and Nelson et al. (2015) have shown the benefits of systematic and intensive molecular studies in revealing the extent of diversity within Corallinophycidae. Nelson et al. (2015) revealed, through a multi-gene phylogenetic study on the New Zealand corallines, a new order called Hapalidiales that now contains the genus Lithothamnion. Bittner et al. (2011) suggested that as Lithothamnion is such a large and important component of the Hapalidiales, delineating its taxon boundaries would be critical in helping to understand the taxonomy and regional diversity of the order.
The genus Lithothamnion currently includes 90 taxonomically accepted species (Guiry & Guiry, 2016). It is the largest genus within the subfamily Melobesioideae, with a global distribution (Guiry & Guiry, 2016). Lithothamnion species are characterized by their flared epithallial cells, lack of secondary pit connections, presence of lateral cell fusions, long subepithallial initials, uniporate gametangial conceptacles, multiporate tetra/bisporangial conceptacles and a monomerous thallus (Irvine & Chamberlain, 1994). Four species have been reported for the British Isles (Table 1; Irvine & Chamberlain, 1994;Brodie et al., 2015): L. corallioides, L. glaciale Kjellman, L. lemoineae Adey and L. sonderi Hauck. Lithothamnion lemoineae was described in Adey (1970). The type specimen was collected from Merchant Island, East Penobscot Bay, Maine in 1961 and deposited in the Smithsonian Institution Herbarium (Adey, 1970). The reported presence of L. lemoineae in Britain was questioned by Adey & Adey (1973), who considered that L. lemoineae was restricted to the northwest Atlantic. However, this view was apparently overturned when in the 1980s two Lithothamnion specimens collected from the littoral zone at Whitley Bay, Northumberland (north-east coast of England) were identified as L. lemoineae by Yvonne M. Chamberlain and L. lemoineae maerl was observed in Orkney by Peter Sansom (Irvine & Chamberlain, 1994).
Differences in the morphological characteristics of these four British species can allow identification Table 1. Species, distribution, colour and the ability to exist as free-living maerl of the four Lithothamnion species previously reported to occur in the UK (adapted from Irvine & Chamberlain, 1994 (Table 1). Lithothamnion lemoineae generally forms crusts with short, rounded branches roughly 1-2 mm in diameter, is pink-violet to brown in colour when fresh, has large pit connections within the cortical cells and strongly developed cell zonation, and lacks staining bodies (Adey, 1970). Staining bodies (phosphotungstic haematoxylin), were described in crustose corallines by Adey (1966) and thought to be potential storage proteins (Adey & McKibbin, 1970). Lithothamnion glaciale forms crusts and maerl, with longer, more complex branching that is usually greater than 2 mm in diameter. Specimens are reddish to deep pink in colour when fresh, with small pit connections, weakly developed or no cell zonation and lack staining bodies (Adey, 1970;Irvine & Chamberlain, 1994). Lithothamnion sonderi occurs mainly as flat, thin crusts with small mounds/protuberances (Chamberlain, 1992;Irvine & Chamberlain, 1994). Mature plants have a red glossy texture with white conceptacles, making them easy to identify (Chamberlain, 1992), but young sterile plants are indistinguishable from L. glaciale apart from the presence of staining bodies (Irvine & Chamberlain, 1994). Lithothamnion corallioides can occur as crusts but mainly grows in unattached forms with branch thicknesses of < 1 mm, is brownish pink when fresh, and has large primary pit connections within the cortical cells. Lithothamnion corallioides is a warm temperate species mainly found in the south of the UK, whereas L. glaciale, L. sonderi and L. lemoineae are cold temperate more northerly species (Irvine & Chamberlain, 1994). Lithothamnion tophiforme (Esper) Unger is also included here as it is a species found in the arctic and subarctic regions of the northeast Atlantic along with L. glaciale (Freiwald & Henrich, 1994;Irvine & Chamberlain, 1994;Adey et al., 2005). Lithothamnion tophiforme, like L. glaciale, forms crusts and maerl with longer complex branches > 2 mm thick, is a reddish orange in colour when fresh and contains staining bodies (Adey, 1970). Even though these characteristics provide a first step in distinguishing these species, it can still be difficult to identify them and hence the need for molecular studies on maerl beds (e.g. Carro et al., 2014;Costa et al., 2014;Pardo et al., 2014;Peña et al., 2014b).
Using the mitochondrial CO1 and plastid psbA gene, an initial DNA barcoding study of the maerl beds in Falmouth and Oban, Britain, as well as historical material stored in the algal herbarium (BM) at the Natural History Museum, London, UK, was undertaken. This study revealed cryptic diversity within the maerl beds in Scotland and the North of England. The choice of markers in this study followed other molecular identification studies on maerl beds (Carro et al., 2014;Pardo et al., 2014;Peña et al., 2014b;Hernández-Kantún et al., 2015). The CO1 marker has been used due to its fast evolution rate and therefore the ability to reveal cryptic diversity and phylogenetic relationships (Hebert et al., 2003a, b). The plastid psbA gene can be seen as a viable alternative for routine DNA barcoding (Carro et al., 2014) and phylogenetic analysis (Bittner et al., 2011). As the psbA gene evolves more slowly than the mitochondrial CO1 gene, it is used to recover deep phylogenetic relationships (Yoon et al., 2002) and to resolve intrageneric relationships (Seo et al., 2003;Yang & Boo, 2004). The results of the initial barcoding study led to a phylogenetic analysis using historical and contemporary material of Lithothamnion around Britain, as well as an assessment of the disparity between molecular and morphological diversity.

Materials and methods
Crust and maerl specimens from St Mawes Bank, Falmouth, England (mean depth 6 m; July 2014) and Loch Creran, Oban, Scotland (depth 9.6 m; October 2014) were hand collected by scuba divers. Rhodolith, crusts and maerl specimens from the algal herbarium (BM) at the Natural History Museum, London, UK and the type specimen of L. lemoineae from the US National Herbarium (USNC), Smithsonian National Museum of Natural History, Washington D.C., USA, were included in the analyses. Individual specimens were identified as Phymatolithon calcareum, Lithothamnion corallioides, L. glaciale or L. lemoineae based on morphology. In total 35 specimens from Oban, 32 specimens from Falmouth and 33 herbarium specimens from around Britain and Newfoundland, Canada were analysed (Table S1).

DNA extraction, amplification and sequencing
For DNA extraction a sample (1 cm in length), visually free from epiphytes, was taken from the tip of a branch of each specimen. DNA extraction was performed using the QIAamp DNA Micro Kit (Qiagen) following manufacturer's instructions. The type specimen of L. lemoineae was sequenced separately to avoid contamination. Two gene fragments were amplified: (1) a standard DNA barcode of 664 bp, the 5' mitochondrial end of the cytochrome oxidase 1 unit (CO1-5P), using primers M13LF3 and M13Rx (Saunders & Moore, 2013); and (2) a 950 bp barcode of the plastid Photosystem II thylakoid membrane protein (psbA), using primers psbAF1 and psbAR2 (Yoon et al., 2002).
Each PCR run contained 2.5 µl NH 4 RXN buffer, 1.5 µl MgCl 2 , 0.5 µl dNTP stock, 0.5 µl Biotaq DNA polymerase (all from Biotaq DNA Polymerase kit, Bioline Ltd, UK), 18 µl PCR quality H 2 O, 0.1 µl of forward primer, 0.1 µl of reverse primer (each primer had a 10 µM concentration) and 1-2 µl of DNA, totalling 25 µl of the PCR mix. A standard protocol of PCR (initial denaturation 94°C for 5 min, then 35 cycles of denaturation at 94°C for 30 s; annealing at 50°C for 30 s; extension at 72°C for 1 min; final extension at 72°C for 5 min and final hold at 10°C) was used for psbA. For CO1 a modified PCR protocol was used (initial denaturation 94°C for 2 min; then 5 cycles of denaturation at 94°C for 30 s; annealing at 45°C for 30 s; extension at 72°C for 1 min; a further 35 cycles of denaturation at 94°C for 30 s; annealing at 46.5°C for 30 s; extension at 72°C for 1 min; final extension at 72°C for 7 min and final hold at 10°C) (Saunders & Moore, 2013).
Amplification success was determined using gel electrophoresis. Samples that had successfully amplified were cleaned using the Illustra GFX PCR DNA purification kit following manufacturer's protocol (GE Healthcare, UK). Dideoxy cycle sequencing was performed on the samples using Big Dye v1.1 (Life Technologies, UK) containing 2 ng/100 base pairs of amplicon size and 1 µm primer concentration in 10 µl volume. Amplification was performed using the Techne Thermo cycler (Bibby Scientific) (28 cycles; 10 s at 96°C, 5 s at 50°C and 4 min at 60°C). Excess dye labelled nucleotides from the sequence reaction were removed with ethanol-sodium acetate precipitation and sequenced products were dried, resuspended and run on a 3730 XL capillary DNA analyser (Applied Biosystems). All sequences were submitted to GenBank and all codes can be found in Table S1.

Data analysis
Forward and reverse sequences were aligned and edited using the programs Seqtrace 0.9.0 (Stucky, 2012) and Bioedit 7.2.5 (Hall, 1999). Bayesian and maximum likelihood optimality criteria were used to analyse phylogenetic relationships (Guindon et al., 2010;Ronquist et al., 2012). Both datasets included sequences from GenBank, related sequences found through BLAST and one outgroup sequence each. Phymatolithon calcareum was the chosen outgroup as sequences from type material have already been published. Phymatolithon calcareum is another maerl-forming species in the Hapalidiales, which occupies the same niches as Lithothamnion around Britain. It is related yet sufficiently genetically distant to form a suitable outgroup. The CO1 dataset had 14 sequences while the psbA dataset had 10 sequences. Jmodeltest v2.1.1 (Guindon et al., 2010;Darriba et al., 2012) was used to select the best fit model by analysing the Akaike information criterion (AIC); the GTR + I model was selected for all datasets.
Bayesian analysis was implemented in MrBayes, version 3.2.3 (Ronquist et al., 2012). Two runs of four chains for 200 000 generations were employed for all analyses, with sampling occurring every 50th generation. The mean standard deviation of the split frequencies and the potential scale reduction factor (PSRF) was used to determine the stationarity of the Markov Chain Monte Carlo (MCMC); by 200 000 generations all analyses had converged. Consensus trees were constructed after 100 000 generations. Maximum likelihood analyses were implemented using PhyML version 3.0 (Guindon et al., 2010) and 100 bootstrap replicates were run to generate bootstrap support statistics (Williamson et al., 2015).

Species delimitation
For species delimitation, the whole sequence dataset was used plus GenBank sequences used in the phylogenetic study, totalling 72 sequences for the CO1 gene and 82 sequences for the psbA gene (Table S1). Two different methods were applied, the Automatic Barcode Gap Discovery (ABGD) and the General Mixed Yule Coalescent method (GMYC). The ABGD (Puillandre et al., 2012) aims to statistically find the barcode gap (the gap between intraspecific diversity and interspecific diversity). The method takes a dataset and primarily partitions the dataset based on the gap in the pairwise differenceswhere the mode of the distribution of intraspecific differences is lower than the mode(s) of interspecific divergence. This method is repeated on the primary partition until no further splitting occurs. The analysis was performed at http://wwwabi.snv.jussieu.fr/pub lic/abgd/abgdweb. All parameters were set to default (P min = 0.001, P max = 0.100, Steps = 10, Number of bins = 20) apart from x, minimum gap width, and the distance. X was set to 1.0, while the Jukes-Cantor (JC69) distance was selected.
Secondly, the GMYC method, developed by Pons et al. (2006), aims to delimit species by finding the transition between species level and population level evolutionary processes by analysing differences in branching patterns on a gene tree. A Bayesian phylogenetic analysis was conducted in BEAST v1.5.3. (Drummond et al., 2012) using a GTR + I + G model with a strict clock and using a constant Yule speciation tree prior. Markov chain Monte Carlo (MCMC) analyses were run for 20 million generations for both CO1 and psbA. For each run, trees were sampled every 1000th generation. Tracer v.1.5 was used to diagnose the output for convergence, which was when all model parameters for the combined log files exceeded an estimated sample size (ESS) of 200. The GMYC analysis was performed on maximum clade credibility trees calculated with TreeAnnotator v.1.4.8 and using the SPLITS package for R (http://r-forge.r-project.org/projects/splits/). Both the single threshold and the multiple threshold, where the analysis incorporates a variable transition from coalescent to speciation among lineages (Monaghan et al., 2009), were fitted on a fully dichotomous ultrametric tree.

Morphology
Sporangial conceptacles were observed in histologically prepared specimens that had been decalcified in 1% hydrochloric acid. Longitudinal sections of the decalcified specimens were stained with aniline blue, according to the methods of Brodie & Guiry (1988). Conceptacle measurements were made using ImageJ analysis software (Rasband, 1997(Rasband, -2011. For cell morphology images, longitudinal sections of the protuberances of specimens were mounted on aluminium stubs and coated with gold. Samples were then imaged using the Hitachi S-3500N scanning electron microscope (SEM). For cell measurements, specimens were scanned using synchrotron-based X-ray tomographic microscopy at the TOMCAT beamline at the Swiss Light Source, Paul Scherrer Institute, Villigen, Switzerland (Stampanoni et al., 2006). For each tomographic scan, 1501 projections over 180 degrees were acquired at 20 keV, 10× objective, field of view of 1.6 × 1.2 mm 2 , pixel size 0.65 × 0.65 µm 2 . The exposure time ranged between 100 and 200 ms. Projections were rearranged into corrected sonograms, and reconstructed using optimized fast Fourier transformations and gridding procedures (Marone et al., 2010). The final data were exported as TIFF images. Cell measurements were acquired in the Avizo lite V9.0 (VSG) software package. Cell measurements were based on calcified inorganic material and therefore length and diameter measurements are of the cellular space rather than the whole cell.

Molecular phylogenetic analysis
Analysis of CO1 and psbA sequence data of contemporary and historical specimens from Oban and Berwick revealed the presence of both Lithothamnion glaciale and an unidentified species of Lithothamnion, while specimens from Falmouth were identified as either L. corallioides or Phymatolithon calcareum. The psbA gene of the type material of L. lemoineae, collected from Maine, USA, was successfully sequenced and matched with other L. lemoineae sequences from Canada. The material initially identified as L. lemoineae from Whitley Bay, North Northumberland, Britain, however, did not match the genetic sequences of the type material or Canadian specimens. One specimen (HLE02) matched GenBank sequences of L. glaciale, while the other specimen (HLE01) corresponded to Lithophyllum orbiculatum and therefore the sequence was not used in the analysis.
In the CO1 phylogeny there was high bootstrap and Bayesian support for the clades representing species (the relation between L. lemoineae and Lithothamnion sp. was not supported; Fig. 1), whereas in the psbA phylogeny high bootstrap and Bayesian support was seen in all clades except for between L. glaciale and L. lemoineae, where there was only Bayesian support (Fig. 2). The interspecies CO1 divergence within the cold temperate species was 34-45 bp between L. glaciale and Lithothamnion sp., 50-57 bp between L. glaciale and L. lemoineae and 50-53 bp between Lithothamnion sp. and L. lemoineae (Table 2). For psbA, the interspecies variation was 12 bp between L. glaciale and Lithothamnion sp., 12-20 bp between L. glaciale and L. lemoineae and 13 bp between Lithothamnion sp. and L. lemoineae (Table 3).

Species delimitation
Results for the ABGD and GMYC (single and multiple) analyses (Table S2, Figs S1-S6) show that for both CO1 and psbA, the GMYC model was favoured over the null model for single and multiple thresholds (CO1 (single), P <0.001; CO1 (multiple), P <0.001; psbA (single), P <0.001; psbA (multiple), P <0.001). For both genes and both thresholds there was significant evidence for the predicted shift in branching rates from interspecific variation to intraspecific variation (Figs S1-S4, Table S2). The multiple threshold for both genes had larger confidence intervals than the single threshold and also over-split the results especially in the psbA analysis (Table S2). Using the single threshold, smaller confidence intervals and results comparable to the ABGD analysis and tree topologies were seen within the CO1 data. Therefore, we focused our interpretation only on the single threshold results (Table S2). The CO1 gene under the single threshold predicted five    clusters, whereas the psbA analysis predicted between 11 and 12 clusters with a larger confidence interval. Under the ABGD analysis five groups were delimited for each gene (Table S2, Figs S5-S6). These five groups all included individuals assigned to one of the five species Phymatolithon calcareum, Lithothamnion corallioides, L. glaciale, L. lemoineae and Lithothamnion sp. Based on these results coupled with morphological evidence we have therefore described a new species of Lithothamnion.

Comparison of Lithothamnion erinaceum sp. nov. with other cold temperate North Atlantic species
Lithothamnion erinaceum sp. nov. exhibits many features of the genus Lithothamnion such as flared epithallial cells, primary pit connections, multiporate tetra/bisporangial conceptacles and lateral cell fusions (Figs 7, 8), similar to L. glaciale (Fig. 6) (Irvine & Chamberlain, 1994). Apart from molecular data, vegetative characters can also be used to distinguish between different species of Lithothamnion (Table 4). When dried, both L. erinaceum sp. nov. and L. glaciale are a similar deep pink, whereas L. lemoineae is characterized by a much lighter pink (Figs 3-5). Lithothamnion sonderi is (mauvy) pink, and covered in white speckles (Chamberlain, 1992). Lithothamnion erinaceum sp. nov. has narrow long branches (typically < 3 mm in diameter), whereas L. lemoineae has even narrower yet shorter branches (max 2 mm in diameter). Lithothamnion glaciale branches can be variable in diameter (> 2 mm) (Adey, 1970;Irvine & Chamberlain, 1994) and are typically bulbous (Fig. 4) compared with both L. erinaceum sp. nov. (Fig. 3) and L. lemoineae (Fig. 5). Lithothamnion sonderi does not form extensive branching like the other cold temperate North Atlantic species of Lithothamnion. As branching morphology is related to environmental factors such as depth, light and wave energy (Steller & Foster, 1995), differences seen between environments will be, therefore, manifested within the branching morphology as seen in the Carro et al. (2014) study. This means that the branching morphology stated here may be different in other environments. Hence, caution must be applied when identifying species based on branching morphology in other environments. On the outer surface, the pores of L. glaciale are surrounded by 6 cells, whereas L. sonderi has a range of between 6 and 8 cells surrounding the pores. Lithothamnion erinaceum sp. nov. also has a range but the number of cells surrounding the pores are between 5 and 7 (Fig. 9). For the reproductive features, L. glaciale has buried conceptacles (Irvine & Chamberlain, 1994) (Fig. 10), whereas there is no evidence of buried conceptacles in L. erinaceum sp. nov (Fig. 11). The range of asexual conceptacle size is between 80-180 µm in height and 140-290 µm in diameter for L. erinaceum sp. nov., which is smaller than the conceptacle sizes for both L. glaciale (110-180 µm height, 150-360 µm diameter), and L. lemoineae (122-162 µm height, 178-315 µm diameter) (Adey, 1966;Irvine & Chamberlain, 1994). Lithothamnion erinaceum sp. nov. has conceptacles larger in height but smaller in diameter than L. sonderi (90-140 µm height, 143-422 µm diameter) (Chamberlain, 1992).

Discussion
Despite the long history of the study of maerl around Britain (Irvine & Chamberlain, 1994) our molecular results have uncovered a previously misidentified species of Lithothamnion in the maerl beds in the north of Britain. Lithothamnion erinaceum sp. nov. occurs with L. glaciale in the maerl beds in Northern Ireland, Scotland, Iceland and Norway, but appears to make a smaller contribution. It was suggested by Pardo et al. (2014) that this species, known at the time as Lithothamnion sp. 2, along with L. glaciale replaces L. corallioides and P. calcareum as the dominant maerl-forming species at higher latitudes in the North Atlantic. However, they found sequences of this species matching a Pacific Canadian coralline, suggesting that this species may also be found in the North Pacific. Even with the same habitat and similar morphology as L. glaciale the significant genetic divergence distinguished L. erinaceum sp. nov. as a separate species (Tables 2, 3). This demonstrates the difficulty of relying on morphology alone in the identification of rhodolith and maerl-forming coralline algae and shows that a morphological taxonomy may underestimate the local diversity. The results highlight the hidden diversity within the genus Lithothamnion around Britain and the importance of the use of molecular techniques to aid conventional morphological techniques in revealing true diversity, which corroborates earlier findings within the Rhodophyta (Robba et al., 2006;Walker et al., 2009;Peña et al., 2014b).
The results also confirmed the presence of the warm temperate species, L. corallioides and P. calcareum in the south of Britain and the presence of the cold temperate species L. glaciale in the north of Britain. Lithothamnion erinaceum sp. nov., the previously misidentified species, was clearly separated from the other Lithothamnion species in the phylogenetic analysis, although the relationship between L. erinaceum sp. nov. and the other cold temperate species (L. lemoineae and L. glaciale) could not be resolved (Figs 1, 2). This may indicate recent divergence. CO1 is a mitochondrial gene which is expected to evolve faster than the plastid gene psbA (Tables 2,  3). The difference in evolution rates of the two genes and our interpretation of a recent divergence would explain why the GMYC model multiple threshold and the GMYC model analysis for the psbA gene broke down in delimiting the species and over-split the results (Figs S3, S4). This difference would also explain the ABGD analysis for the psbA gene where at a larger prior limit the cold temperate species were collated into one group creating three individual groups instead of five (Fig. S6). These results therefore suggest that more data are required to fully resolve the relationship of these cold temperate species.
Another Lithothamnion species requiring further taxonomic study is L. tophiforme, an arctic species living in the deeper and colder parts of the subarctic Atlantic (Adey et al., 2005). Lithothamnion tophiforme was first described by Unger in 1858 where he illustrated a specimen from Greenland and stated it was held in the 'hof Naturalien Cabinete'. Unfortunately this specimen has never been found (Adey et al., 2005). The neotype, initially identified as L. soriferum Kjellman (Adey et al., 2005), collected from Julianehaab, Greenland, was described in Adey (1970). It is deposited in the Foslie Herbarium, Section of Natural History, Museum of Natural History and Archaeology, Norwegian University of Science and Technology, Trondheim (Woelkerling et al., 2005). There was only one finding of L. tophiforme in Britain, in the west of Ireland (Adams, 1908) and the material was unavailable for confirmation (Irvine & Chamberlain, 1994). A small number of molecular analyses have included L. tophiforme from Canada, providing SSU (Bailey & Chapman, 1998;Adey et al., 2015), cox 2-3 spacer , rbcL and psbA (Adey et al., 2015) genes sequences for this species. However, as the sequences of L. tophiforme have not been compared with sequence data of the type material, absolute identification of the specimen must be applied with caution. Ideally type material is sequenced for a confident species assignment. However, as in the case for many crustose corallines, the refusal of destructive sampling, degradation of DNA and lost material type Table 4. Characteristic differences in morphology between cold temperate species of Lithothamnion found in Britain, L. lemoineae, L. glaciale, L. sonderi and L. erinaceum sp. nov., adapted from Irvine & Chamberlain (1994) and using data from Adey (1966 Hence the second best option is to sequence material that best resembles the original specimen, preferably collected from the type locality. This has also been done for L. glaciale and L. corallioides, which is recommended here for species assignment. Even without molecular data, we are confident L. erinaceum sp. nov. is not L. tophiforme based on morphological differences. Lithothamnion tophiforme is an orange/red colour whereas L. erinaceum sp. nov. is more pinkish. Lithothamnion tophiforme has much larger asexual conceptacles, ranging between 222-355 µm in diameter (Adey et al., 2005), than those for L. erinaceum sp. nov., which range between 140-290 µm. In addition, there are differences in the geographic distribution between the two species; the waters in Scotland and the north of England can reach temperatures above 12°C (Inall et al., 2009), which is above the thermal limit of L. tophiforme (Adey et al., 2005). As the rest of British coastal waters can reach even higher temperatures, it is unlikely that L. tophiforme can survive there and is therefore found further north. However, a study of the rhodolith beds in the Svalbard Archipelago at the Arctic Circle by Teichert et al. (2012) found no evidence of L. tophiforme. This stresses the need for extensive sampling to further our understanding of the distribution of rhodolith and maerl-forming species in the arctic waters. Specifically, the northern limits of L. erinaceum sp. nov., the distribution of L. tophiforme and whether L. tophiforme replaces L. erinaceum sp. nov. at higher latitudes all need to be resolved.
We are also confident that L. erinaceum sp. nov. is not L. sonderi due to clear morphological differences. Lithothamnion sonderi exists mainly as flat reddish crusts with white speckles, sometimes with low mounds/protuberances (Chamberlain, 1992;Irvine & Chamberlain, 1994), whereas L. erinaceum sp. nov. exists as more pinkish crusts with prominent branches. The geographic distribution of L. sonderi also differs; it is unknown in Iceland and northern Norway as it is unable to withstand the winter temperatures which are below 6-7°C (Adey, 1971). Conversely L. erinaceum sp. nov. has been found there and therefore can presumably tolerate these temperatures (Pardo et al., 2014). According to Irvine & Chamberlain (1994) L. sonderi does not exist unattached, but can form a crust on dead maerl, whereas L. erinaceum sp. nov. grows unattached (Pardo et al., 2014). Attempts to obtain sequence data from L. sonderi from around Britain gave variable results: four sequences were obtained that did not match each other extremely well but matched species from the Hapalidiales. However, the species was overgrowing a variety of different crusts. The closest match found in GenBank was for L. corallioides at 94% similarity. This suggests that we may be dealing with some new species, in which one may be L. sonderi, or even species belonging to different genera. Even though all care was taken in sequencing L. sonderi, because the specimens used were crusts overgrowing maerl or growing with a variety of crusts, the sequence variation seen could be potentially due to contamination. Therefore, for absolute confirmation of L. sonderi fresh material would need to be sequenced and if possible compared with sequence data from the type specimen.
Lithothamnion lemoineae is a rhodolith/maerlforming species predominantly found in the Northwest Atlantic, from Newfoundland south to the Gulf of Maine. Sequences of historical specimens initially identified as L. lemoineae, collected from Britain, did not match those from the type material and were incorrectly identified. Hence, we have disproved the current evidence that L. lemoineae is found in Britain (Irvine & Chamberlain, 1994), which is in agreement with Adey (1970) who did not find any evidence of L. lemoineae in Europe. This study, therefore, supports the interpretation of L. lemoineae being an endemic species in the North-west Atlantic. However, a more extensive study is required to fully rule out the presence of L. lemoineae in Britain.
This study has highlighted the need for the use of molecular techniques alongside conventional morphological taxonomy in order to be confident in estimating local species diversity of calcified red algae. Maerl/ rhodolith-forming coralline algae are highly plastic organisms known to acclimate to their environments. Large variations in morphology are not only seen within populations but also within individuals. Therefore sole morphological assessment can be extremely difficult. This study shows that molecular taxonomy is an easy method to quickly identify taxa confidently. A full re-evaluation of Lithothamnion is needed to understand the distribution and relative abundance of Lithothamnion species around Britain.

Funding
We acknowledge a NERC studentship award [NE/L501554/ 1] and the Natural History Museum, London for Leanne A. Melbourne and a Royal Society URF for Daniela N. Schmidt for providing funding.

Supplementary Information
The following supplementary material is accessible via the Supplementary Content tab on the article's online page at http://10.1080/09670262.2016.1269953 Supplementary Table S1. List of all specimens collected and sequenced, as well as GenBank sequences used in the analysis. Supplementary Table S2. The GMYC (single and multiple threshold) and ABGD analysis of the CO1 and psbA genes. Supplementary Figs S1-S4. GMYC analysis, No. of lineages against time. Fig S1; CO1 genesingle threshold. Fig S2; CO1 genemultiple threshold. Fig S3; psbA genesingle threshold and Fig S4; psbA genemultiple threshold. Red line is the transition from interspecies variation to intraspecies variation. Supplementary Figs S5-S6. ABGD analysis, number of groups against prior intraspecific divergence. Fig. S5; CO1 gene and Fig S6; psbA gene.