Molecular phylogenies support taxonomic revision of three species of Laurencia (Rhodomelaceae, Rhodophyta), with the description of a new genus

. The systematics of the Laurencia complex was investigated using a taxon-rich data set including the chloroplast ribulose-1,5-bisphosphate carboxylase/oxygenase large subunit ( rbc L) gene only and a character-rich data set combining mitochondrial cytochrome oxidase 1 (COI-5P), the rbc L marker, and the nuclear large subunit of the ribosomal operon (LSU). Bayesian and ML analyses of these data sets showed that three species hitherto placed in the genus Laurencia J.V.Lamour. were not closely related to Laurencia s. str. Laurencia caspica Zinova & Zaberzhinskaya was the sister group of the remaining Osmundea Stackh. species, L. crustiformans McDermid joined Palisada and L. ﬂ exilis Setch. consisted of an independent lineage. In light of these results a new genus, Ohelopapa F.Rousseau, Martin-Lescanne, Payri & L.Le Gall gen. nov., is proposed to accommodate L. ﬂ exilis . This new genus is morphologically characterized by four pericentral cells in each vegetative axial segment; however, it lacks ‘corps en cerise’ in cortical cells and secondary pit connections between cortical cells, which are characteristic of Laurencia . Three novel combinations are proposed to render the classi ﬁ cation Molecular phylogenies support taxonomic revision of three species of Laurencia (Rhodomelaceae, Rhodophyta), with the description of a new genus.


Introduction
The Laurencia complex consists of species which historically belonged to the genus Laurencia J.V.Lamour. (Rhodomelaceae, Ceramiales). Although it is easy to assign specimens to the Laurencia complex, it is far more diffi cult to delimit species and genera within the complex due to extensive morphological plasticity and the signifi cance of inconspicuous anatomical characters.
Over the last two decades, the diversity within the genus Laurencia has been revealed by thorough morphological and molecular studies, leading to the recognition of distinctive anatomical features, as well as genetic divergences usually diagnostic at the generic level. The Laurencia complex has been split, leading to the resurrection or proposal of new genus names (Table 1). Currently, the Laurencia complex consists of seven genera: Chondrophycus (Tokida & Y.Saito) Garbary & J.T.Harper (Garbary & Harper 1998), Coronaphycus Metti (Metti et al. 2015), Laurencia (Lamouroux 1813), Laurenciella V.Cassano, Gil-Rodríguez, Sentíes, Díaz-Larrea, M.C.Oliveira & M.T.Fujii (Cassano et al. 2012), Osmundea Stackh. (Nam et al. 1994), Palisada K.W.Nam (Nam 2007) and Yuzurua (K.W.Nam) Martin-Lescanne (Martin-Lescanne et al. 2010). Despite the recognition of genera within the Laurencia complex, a limited number of species have been assigned to the four resurrected genera and the three recently described genera only include one (Laurenciella and Yuzurua) or two species (Coronaphycus) each (Table 1). Therefore, Laurencia s. str. is still by far the most speciose genus within the complex, and according to AlgaeBase (Guiry & Guiry 2016) it currently accommodates 145 recognized species and 27 infraspecifi c taxa among no fewer than 421 taxa (Guiry & Guiry 2016), strongly suggesting that steps toward taxonomic revision are necessary. Presently, most species of the genus Laurencia s. str. share the characters of four pericentral cells, trichoblastic spermatangia, tetrasporangial development from pericentral cells (Nam et al. 1994;Nam 1999), 'corps en cerise' (globular, hyaline bodies) in cortical cells and secondary pit connections between cortical cells. Reproductive characters are not always available, while the other anatomical features are diffi cult to observe, particularly in herbarium specimens. Consequently, the systematic revision of the Laurencia complex relies mostly on molecular databases to complete comprehensive phylogenetic analyses.
In the course of recent fl oristic surveys we have collected many specimens of the Laurencia complex from their type localities, including Laurencia fl exilis Setch., L. caspica Zinova &Zaberzhinskaya andL. crustiformans McDermid (e.g., Sherwood et al. 2010a, 2010b). These three species have character combinations that do not exactly match the description of Laurencia s. str. Laurencia fl exilis has four pericentral cells, but unlike the type species of Laurencia, L. obtusa (Huds.) J.V.Lamour., it lacks secondary pit connections and 'corps en cerise', (Saito 1967, Masuda et al. 1999. Abe et al. (2006) resolved this taxon with moderate support as the earliest diverging lineage within the Laurencia complex in an rbcL phylogeny. Regrettably, Abe's data are not publicly available and in the present study we generate novel sequences to address the question of the affi nity of this taxon.  (McDermid 1989, Abbott 1999. It shares with Laurencia the presence of secondary pit connections between cortical cells and 'corps en cerise', but it differs by bearing tetrasporangia in a right-angle arrangement, as observed in species of Palisada and Yuzurua. The aim of this manuscript is to assess the generic affi nities of L. fl exilis, L. caspica and L. crustiformans within the Laurencia complex by generating molecular data for specimens collected at the type localities of these three taxa, and inferring phylogenetic relationships from a character-rich combined dataset (LSU + rbcL + CO1) and a taxon-rich rbcL only dataset.

Specimen collection
Newly sequenced specimens were collected in different localities around the world and are listed in Table 2, along with their valid names and GenBank accession numbers (NCBI GenBank) as well as the herbarium where the specimen was deposited. Samples were dried in silica gel or prepared as herbarium specimens and some of them were stored in 5% buffered formalin in seawater. Specimens from the Hawaiian Islands were collected as part of the Hawaiian Rhodophyta Biodiversity Survey (Sherwood et al. 2010a(Sherwood et al. , 2010b. During a monitoring project by Environment & Resource Technology Limited (ERT Caspian) for BP Global in Sangachal Bay, Azerbaijan, Caspian Sea, in connection with existing and future operations at the Sangachal oil pipeline terminal (ERT Caspian 2002), the distribution of red algae was surveyed in June 2001 (by grab sampling) and 2002 (snorkeling). Laurencia caspica was collected at depths of 4-11 m, mostly more than 1 km from shore, on calcareous hard substrata and barnacle shells. The salinity in this area is 12-13 PSU, typical of the Caspian Sea (Lüning 1990: 123), and macroalgal diversity is very low, with only six red algal species recorded (ERT Caspian 2002) from the 30 species known from the Caspian (Lüning 1990 (2017) 6

Sample preparation for anatomical studies
Morphology was observed under dissecting and compound microscopes and permanent slides were made from sections stained with acidifi ed aniline blue stain (4% in seawater) and mounted in 50% dilute Karo corn syrup.

Extraction, amplifi cation and sequencing
Total cellular DNA was extracted from tissues preserved in silica gel and dry herbarium specimens using the DNeasy Plant Mini Kit (QIAGEN, Valencia, California, USA or Qiagen GmbH, Hilden, Germany). From 0.5 × 10 -3 to 10 -3 mg of proteinase K was added to the lysis buffer to improve the DNA yield. Except for L. caspica, the rbcL coding region (1467 bp) was amplifi ed using the following combinations of primers: F-rbcL start × R-753 (Freshwater & Rueness 1994) for the 5' end, rbcLFC × 1011R (Nam & Choi 2000) or F-577 × R1381 (Freshwater & Rueness 1994) for the middle fragment and F-993 × R-rbcS start (Freshwater & Rueness, 1994) for the 3' end. The protocol used for PCR amplifi cations was presented in Martin-Lescanne et al. (2010).
For L. caspica, PCR amplifi cation used a PTC-200 DNA Engine (MJ Research Inc.). All PCR amplifi cations were carried out using the published primers rbcLFC as the forward primer and rbcLRD as the reverse primer , and all reactions contained 200 μM each of dATP, dCTP, dGTP and dTTP, 0.3 μM of each primer, 2.5 mM MgCl 2 and 1.6 units of Taq polymerase (Bioline). The PCR cycle used was as previously indicated . The mitochondrial marker cytochrome oxidase subunit 1 (CO1) was amplifi ed using the primer pair GAZF1 and GAZR5 with the PCR conditions of Saunders (2005) and Clayden & Saunders (2014). LSU was amplifi ed as three overlapping fragments using primers T01N/T20, T04/T08 and T05/T15, and using the PCR primers and the internal primers T10, T16N, T19N, T22, T24, T25, T30 and T33, following the protocols of Harper & Saunders (2001) and Le Gall & Saunders (2010).
The resulting PCR products were purifi ed and used as templates for cycle sequencing reactions with the same primers as those used for the initial amplifi cations. These steps were performed by Genoscope (www.genoscope.fr, Evry, France) or Eurofi ns (France). For L. caspica, PCR fragments were purifi ed using the High Pure PCR Product Purifi cation Kit (Roche Diagnostics Ltd., Lewes, UK), according to the manufacturer's instructions. The PCR products were then directly sequenced commercially by MWG-Biotech, Ebersberg, Germany.

Sequence alignments and phylogenetic analyses
Sequences were obtained for both DNA strands, assembled and corrected using Sequencher TM v. 4.1 (Gene Codes Corporation, Ann Arbor, Michigan) or Codoncode (Dedham, Massachusetts). Alignments of LSU, rbcL and CO1 were performed with MEGA v. 6 (Tamura et al. 2013). The LSU alignment was adjusted by eye with the objective of minimizing gaps, and four ambiguous regions corresponding to 109 nt were excluded before phylogenetic analyses.
Two datasets were analysed. The fi rst one included 32 taxa for which COI-5P, rbcL and LSU (3 genes) were generated (giving 4801 nt alignment length) and a second one was based on only rbcL sequences (1395 nt) for 117 taxa. For the rbcL dataset, a global alignment (198 taxa) was generated with all GenBank sequences belonging to the Laurencia complex and with our newly produced sequences. We then screened the 198 sequences, of which we fi nally selected 117. We excluded sequences that were too short (less than 900 nt) and conserved only sequences that diverged by more than 2 nt (collapsetypes.pl available at http://sourceforge.net/projects/collapsetypes/). Finally, rapid NJ analyses were performed separately on sub-alignments corresponding to typical PCR primer amplifi cations (0-527, 528-701, 702-1099 and 1100-1395) to reveal a putative confl ict signal. The selection of sequences is explained in the Supplementary fi le.

Phylogenetic analyses
Bayesian and ML analyses were performed using MrBayes v. 3.2.6 (Ronquist et al. 2012) and RAxML v. 8.2.0 (Stamatakis 2014) on the cipres portal. For both BI and ML analyses, the rbcL data set was partitioned by codon position (1+2 vs 3) while the three-gene data set was analysed in fi ve partitions by gene and codon positions (codon position 1 and 2 merged into one partition). BI analyses were run with GTR + I + G model parameters estimated independently for each partition, with four heated Monte Carlo Markov Chains for 10 7 generations, with sampling intervals of 1 000 generations, to produce 10 000 trees. Nodal support was assessed by calculating posterior probability (PP) values at each node of the resulting consensus tree after a burn-in of 10% of the trees. The ML analyses were conducted with a GTRGAMMA model with ML estimates of base frequencies. The best-scoring ML tree and 1000 bootstrap trees were obtained using the rapid hill-climbing algorithm (-f a option).

Results
The rbcL only dataset (Fig. 1) and the combined dataset including CO1, rbcL, and LSU (Fig. 2) were analysed using ML and Bayesian approaches to delimit genera within the Laurencia complex and to assess phylogenetic relationships among genera. The two trees (Figs 1-2) were congruent in delimiting strongly supported lineages which overall correspond to the genera currently recognized in the Laurencia complex, with a few exceptions. Laurencia caspica was resolved as a sister taxon to all the species of Osmundea included in these analyses. Laurencia crustiformans allied with species of Palisada with strong ( Fig. 1) to full support (Figs 1-2). Laurencia fl exilis did not join the remaining species of Laurencia and its position varied in the tree depending on the analyses. It was resolved as sister to either Yuzurua (Fig. 2) or Palisada (Fig. 1). Moreover, our analyses revealed that a specimen collected in the Mediterranean Sea grouped with the recently described genus Laurenciella with a divergence from the generitype from the Canary Islands of only 0.5%. In addition, the monotypic genus Yuzurua encompassed two lineages with a divergence of 4.7%: one corresponded to a group of specimens from the Caribbean region and the second to a specimen from Guadeloupe. Intergeneric relationships were not supported by either reconstruction method in our taxon-rich dataset (Fig. 1) or in our character-rich dataset (Fig. 2).

Etymology
Ohelopapa means "strawberry" in Tahitian, an allusion to the Tahitian origin, the red color, and the many stolons developed by Ohelopapa fl exilis.

Type locality
Reef at Tahara Mountain, Tahiti.

Distribution
Tropical regions in the Pacifi c (Setchell 1926;Cribb 1983;Masuda et al. 1999), in the Indian Ocean (Silva et al. 1996) and on coasts of Japan ).

Fig. 2.
Tree inferred from COI-5P + rbcL + LSU using Bayesian analysis (BI) and including 30 specimens of the Laurencia complex and two outgroup taxa. The numbers above branches indicate Bayesian posterior probabilities (pp) and below branches indicate bootstrap values (bp) inferred from 1 000 ML bootstrap replicates (ML); pp < 0.95 and bp < 75% are not shown.

Remarks
Ohelopapa fl exilis (Fig. 3A) specimens included in this study matched previous descriptions where anatomical features were thoroughly illustrated (Setchell 1926;Masuda et al. 1999Masuda et al. , 2006. Notably, we observed, in the outermost cortical layer, translucent cells without secondary pit connections, whereas the inner cortical layer (just below this) consisted of pigmented cells with secondary pit connections (Fig. 3B). This anatomical character was highlighted by Fujii & Cordeiro-Marino (1996) Fujii et al. (2006) showed that C. translucidus pertains to genus Laurencia rather than to Chondrophycus. Interestingly, Masuda et al. (2006) used the absence of secondary pit connections between superfi cial cortical cells as a new argument to support the distinction between L. fl exilis and another morphologically similar species, L. tropica Yamada. It would be interesting in the future to test with molecular characters whether L. tropica belongs to the genus Ohelopapa and also to analyse the taxonomic signifi cance of the presence of an outermost cortical layer formed of translucent cells without secondary pit connections.

Distribution
Recorded from Hawaii and Tahiti.

Distribution
Recorded from the Black Sea (Bulgaria and Romania) and Caspian Sea (Guiry & Guiry 2016), both of which are low salinity bodies of water.

Description
The description of L. caspica is published in Russian in a book that is not widely available, so here we provide the following description. Thalli were 5-11 cm high, growing from a solid discoid basal holdfast; erect axes terete, about 1 mm in diameter, irregularly branched to three orders, with blunt apices, similar in general habit to Laurencia obtusa with the exception of the holdfast (Fig. 5A). In surface view of the cortex of live thalli, there were no 'corps en cerise'; in surface view of preserved and stained cortical preparations, secondary pit connections were absent (Fig. 5B). In transverse section of axes, pericentral cells were not distinguishable, and the cortical cells were comparatively large and slightly radially elongated (Fig. 5C). Lenticular thickenings were absent in medullary cells.
Mature non-reproductive thalli, tetrasporophytes and males were collected but females are unknown. Tetrasporangia 80-110 μm in diameter occurred in bands below the apices of lateral branches. They were produced adaxially from random epidermal cells, cut off laterally from the mother cells. Spermatangial receptacles were terminal and open cup-shaped. Spermatangial structures were of the fi lament type ; spermatangial fi laments were unbranched, bearing numerous elongate spermatangia, and terminating in single large round to ovoid cells up to 40 μm in diameter (Fig. 5D-E).

Discussion
In the present study we focused on assessing the generic affi nities of some species of Laurencia which displayed atypical characters for the genus Laurencia s. str.
Contrary to other Laurencia species, L. fl exilis has neither secondary pit connections between cortical cells nor 'corps en cerise' but axial cells of its sterile branches have four pericentral cells like species of Laurencia s. str. (Masuda et al. 1999, observed both in specimens from Malaysia and Setchell's original material from Tahiti preserved in SAP; Masuda et al. 2006, based on both Setchell's original material and Japanese specimens). In the molecular analyses of Abe et al. (2006), L. fl exilis was resolved as an independent divergent lineage of Laurencia s. str. with no supported relationships with any other genera. In our analyses, we confi rm that L. fl exilis does not have strong affi nities with any of the currently described genera of the Laurencia complex and we therefore propose here to accommodate L. fl exilis in Ohelopapa gen. nov.
When Nam (2007) elevated Palisada to generic rank, he included a cladistic analysis based on the anatomical characters of Laurencia crustiformans and resolved it within a species group for which he proposed novel combinations to be accommodated in Palisada. Surprisingly, he did not propose the transfer of Laurencia crustiformans to Palisada. Given that our molecular phylogeny is congruent with his cladistic analysis, we here propose the new combination Palisada crustiformans. Furnari from the Mediterranean in the anatomical features considered important in this genus, sharing a solid discoid holdfast, terete axes, lacking 'corps en cerise' and secondary pit connections in the cortex, lacking lenticular thickenings in the medulla, and having cup-shaped spermatangial receptacles with individual spermatangial fi laments terminating in 1-2 large cells (Fig. 5;Serio et al. 1999;Furnari 2001). The main difference is that whereas O. caspica is up to 11 cm high, and the axes are less than 1 mm in diameter, O. maggsiana is smaller (up to 5.5 cm) but more robust, up to 3 mm in diameter. Also, in O. maggsiana cortical cells are markedly radially elongated in longitudinal section ("palisadelike", Serio et al. 1999;Fig. 5), whereas in O. caspica cortical cells are isodiametric in longitudinal section (Fig. 5C). Osmundea maggsiana is apparently known only from a single collection from Pantelleria Island, Straits of Sicily (Serio et al. 1999), consisting of males and tetrasporophytes but no cystocarpic thalli. Osmundea caspica also resembles O. hybrida from the Atlantic, but the spermatangial fi laments terminate in single cells rather than in groups of up to 20 large cells as in O. hybrida (Maggs & Hommersand 1993). The spermatangial development on fi laments terminated by large sterile cells and the origin of tetrasporangia from outer cortical cells observed in O. caspica are distinctive characters of the genus Osmundea.
Despite its evident morphological similarities with Osmundea, L. caspica has never been transferred to Osmundea. It is widespread and common in the Caspian Sea (Zinova 1967), and it has also been reported from the Black Sea (Gómez Garreta et al. 2001). It is the only species of Osmundea that occurs in low-salinity conditions, occurring in the Caspian at 12-13 PSU.
In addition to clarifying the generic affi nities of some species so far known as Laurencia, this paper provides new records for the two recently described genera Laurenciella and Yuzurua. Laurenciella Here, we further extend the distribution of the genus to the Mediterranean Sea; however, we question whether this new record is conspecifi c with L. marilzae as it falls in the grey zone between intra-and interspecifi c divergence (0.5% of divergence in rbcL). Unfortunately, our specimens were pressed as herbarium vouchers, a technique which does not preserve tissues of members of the Laurencia complex well enough for a thorough anatomical study. The genus Yuzurua was proposed to accommodate L. poiteaui (J.V.Lamour) M.Howe, the type locality of which is Santo Domingo, Dominican Republic, based on molecular data for specimens from Mexico and USA (Martin-Lescanne et al. 2010). In the present study we report one specimen of Yuzurua from Guadeloupe which displays a divergence that clearly indicates that the two lineages are not conspecifi c (4.7%). Further anatomical studies are needed to assess whether we need to describe a new species for this Guadeloupian record.

Conclusion and perspectives
This study, which is the fi rst examination of the Laurencia complex based on both a taxon-rich dataset (rbcL) as well as a character-rich data set (combining COI-5P + rbcL + LSU), has once again indicated that the Laurencia complex is more diverse than currently recognized at the generic level and has confi rmed that molecular-assisted systematics constitutes a helpful tool to assign species of the notoriously diffi cult Laurencia complex to a genus. Despite the fact that the genera of the Laurencia complex are currently well delimited using molecular characters, relationships among them are still poorly resolved. More molecular data, as well as broader taxon sampling, are still necessary to further improve our understanding of this taxonomic complex in an evolutionary framework. Advances in next generation sequencing create novel opportunities to develop phylogenomic analyses. The recent publication of a plastid genome of a species of Laurencia (Verbruggen & Costa 2015) paves the way toward plastome phylogenetic analyses of the Laurencia complex which will likely break the stranglehold of unresolved phylogenetic relationships.