An assessment of the taxonomic status of the Mediterranean endemic genus Acrodiscus Zanardini (Halymeniales, Rhodophyta)

1,2,3 Institut de Systématique, Évolution, Biodiversité, ISYEB – UMR 7205 – CNRS, MNHN, UPMC, EPHE, Muséum national d’Histoire naturelle, Sorbonne Universités, 57 rue Cuvier, CP 39 75005, Paris, France. 1,4,6 Dept. of Chemical, Biological, Pharmaceutical and Environmental Sciences – Botany, University of Messina, Salita Sperone, 31, 98166 Messina, Italy. 1,5 Laboratori de Botànica, Facultat de Farmàcia, Universitat de Barcelona, Av. Joan XXIII s/n, 08028 Barcelona, Spain.


Material and Methods
Specimens from several localities (Table 1) were used for anatomical and DNA studies.
Fresh material was pressed as herbarium specimens, with fragments preserved in 4% formalin for permanent wet-preservation or dried in silica gel for molecular studies. Vouchers are housed in the Herbarium of the University of Barcelona, Spain (BCN-Phyc), the Muséum national d'Histoire naturelle in Paris, France (PC), or in the Herbarium Messanaensis of the University of Messina, Italy (MS). Additional material (dried or formalin preserved) was examined in the herbaria of Spanish universities and institutions involved in the project "Flora Phycologica Iberica" (Manghisi et al. 2010a). For the lectotypifi cation, the Herbarium Centrale Italicum of Florence (FI), the herbarium of the Civic Natural History Museum of Venice (MCVE), the Herbarium Patavinum of the University of Padua (PAD), the Herbarium Horti Botanici Pisani of the University of Pisa (PI), the herbarium of the University of Leiden (L), the herbarium of the Natural History Museum in London (BM), and the cryptogamic herbarium of the Muséum national d'Histoire naturelle in Paris (PC) were searched for type material (Table 2; Appendix 1). Herbarium designations are given as directed by the Index Herbariorum (Thiers continuously updated).
For morpho-anatomical observations, thalli were hand-sectioned with single-edged razor blades, stained with 1% aniline blue, and observed using a Nikon Optiphot-2 equipped with a Nikon Coolpix 4500 camera (Leica Microsystems, Italy).
Sequence data generated for COI-5P, rbcL and LSU genes were submitted to BOLD (http://www. barcodinglife.org) and to GenBank (Clark et al. 2016). Accession numbers, together with collection information, are given in Table 1. DNA extraction was performed as outlined in Manghisi et al. (2010b). The barcode region COI-5P was PCR amplifi ed as detailed in Saunders & McDevitt (2012), the nuclear LSU rRNA gene was PCR amplifi ed as detailed in Harper & Saunders (2001), and the plastid rbcL gene was amplifi ed with various primer combinations as specifi ed by Freshwater & Rueness (1994) and Wang et al. (2000).
Sequences were generated using the BigDye Terminator v. 3.1 Cycle Sequencing Kit (PE Applied Biosystems [ABI], Foster City, CA, USA) and analyzed using an ABI Prism 3130XL genetic analyzer at   Taxonomy 267: 1-24 (2017)   6 the CEMAR, University of New Brunswick, Canada, or an ABI Prism 3730XL at the Genoscope (www. genoscope.fr) in Evry, France. Forward and reverse sequence reads were assembled into contigs with the software ChromasPro (v. 1.7.6.1, Technelysium Pty Ltd) and edited. Multiple sequence alignments were constructed in SeaView v. 4.3.3 (Gouy et al. 2010) and included both data from GenBank and sequences generated for the present study. Newly generated COI-5P sequences were subjected to distance analysis in PAUP* (Swofford 2002).
For rbcL analyses, an initial alignment of 728 sequences and 1258 nucleotide positions was subjected to neighbor-joining (NJ) distance analysis under a K2 nucleotide substitution model in PAUP* to identify species groups. The resulting tree was used to prepare a second alignment for subsequent phylogenetic analyses with 36 sequences representative of most genera of the order Halymeniales, including the generitypes, by the exclusion of duplicate or similar sequences (poor quality sequences, i.e., those missing more than 30% of data, were also removed), and 1211 nucleotide positions.
Similarly, an initial alignment was built with 122 LSU sequences and 2921 nucleotide positions, from which, after NJ analyses performed as above, a fourth alignment was built with 33 LSU sequences representative of most genera of Halymeniales, including the generitypes, and 2640 nucleotide positions, excluding those ambiguously aligned.
A fi fth alignment consisted of the concatenation of LSU and rbcL sequences for the same species and, when possible, for the same specimen, these including 21 sequences and 3820 nucleotide positions. GenBank accession numbers of sequences used for the fi nal alignments are listed in Table 3.
Maximum Likelihood analyses were performed with bootstrap resampling to estimate robustness of the internal nodes (Felsenstein 1985), based on 1000 replicates in PhyML, with a GTR+G+I substitution model (with all parameters estimated during the search), starting from ten random BIONJ trees (Gascuel 1997) with subtree pruning and regrafting (SPR) as branch-swapping algorithm. Furthermore, Maximum Likelihood (ML) analyses were also performed with bootstrap resampling based on 100 replicates in RAxML, with a GTR+G+I substitution model with three partitions for rbcL, corresponding to each codon position, and with four partitions for the concatenate analyses corresponding to LSU and each codon position of rbcL.
For Bayesian inference both the rbcL and the concatenate data sets were also partitioned as above. The covarion-like model (Huelsenbeck 2002) was combined with the GTR+G model of sequence evolution linking or unlinking parameters among partitions (shape, statefreq, revmat, switchrates, Tratio) and setting the prior for the site specifi c rates as "variable". Different analytical strategies were tested in order to reach convergence. Each analysis consisted of two parallel runs, each run using four chains, one cold and three incrementally heated.
A single run consisted of 5 million generations that were sampled every 1000 th tree. After completion of the two runs, likelihood values were plotted against the number of generations to evaluate when MCMC chains reached stability, in order to set an appropriate burn-in value for each analysis. Only trees saved during the stationary phase were used to reconstruct a majority-rule consensus tree and calculate the distribution of posterior probabilities.
In all phylogenetic analyses, unrooted trees were constructed, the root being subsequently designated based on previous knowledge (Withall & Saunders 2006

Distribution and habitat
Acrodiscus is uncommon but widely distributed across the Mediterranean Sea ( Fig. 4; see also Guiry & Guiry 2016;Manghisi et al. 2010a); it is a sciaphilous species, found throughout the year on rocky substrata from depths of 0-50 m. Tetrasporophytes were collected in spring and autumn (Table 1).

Taxonomic history
In the year following Meneghini's (Savi 1841)     In 1949, Ercegović proposed Acrodiscus vidovichii f. cochlearis, arguing that the specimens from Dalmatia (Croatia) had "spoon-shaped", rather than the fl attened fronds described by several other workers (Ardissone 1883;Hauck 1885;Preda 1908;Zanardini 1868). Ercegović overlooked, however, the fact that populations from throughout the Mediterranean had been described as both compressed-fl at or plane (Ardissone 1883;Hauck 1885;Kylin 1956;Savi 1841;Preda 1908;Zanardini 1868) and with bent/curved sub-grooved margins (Agardh 1842;Aleem 1993;De Toni 1905;Feldmann 1939). In our experience, freshly collected specimens normally have bent/curved margins but fl atten once pressed on herbarium sheets; younger and thinner specimens, on the other hand, can be planar throughout and usually adhere to paper, whereas older, more coriaceous specimens may remain canaliculate and nonadherent. We therefore fi nd little reason to recognize a separate forma cochlearis.

Lectotypifi cation
Chondrus? vidovichii Menegh. was validly published in what is commonly reported as Menghini's Algologia Dalmatica (Guiry & Guiry 2016) in Atti della terza Riunione degli Scienziati italiani tenuta in Firenze nel Settembre 1841 (Savi 1841). Indeed, the latter is a congress acta in the form of a book with various sections. During the meeting of the Botany and Plant Physiology group, Meneghini showed his manuscript to the assembly. The secretary of the group, Pietro Savi, recorded the meeting events in the acta, and transcribed part of Meneghini's manuscript, reportedly titled Algologia Dalmatica, including its novelties. With specifi c reference to Chondrus? vidovichii, Savi copied the Latin diagnosis and noted the lack of reproductive structures, and made reference to an illustration that does not appear in the acta. No holotype is designated, nor is there an iconotype that might serve as one.
In the Library of Natural and Environmental Sciences of the University of Pisa, Italy, we found numerous manuscript documents belonging to Meneghini, among them the original complete manuscript of the so-called Algologia Dalmatica, actually Alghe Dalmate, enumerate ed illustrate dal professor Giuseppe Meneghini (Fig. 5 A-B), along with plates including his illustration of Chondrus? vidovichii (Fig 6). Unfortunately, the manuscript was never published.
Interestingly, in other documents there is evidence that: a) Meneghini received material from Dalmatia (Croatia) collected either by Vidovich in Sebenico (Šibenik), by Sandri in Zara (Zadar), or by Stalio in Spalato (Split); b) Menegnini received material from Vidovich in July 1841 (including a Chondrus?); and c) Meneghini dedicated to Vidovich all the new species collected by him. Consequently, it can be inferred that the type material should have been collected by Vidovich in Sebenico in July 1841.
Finally, we found that the Herbarium Horti Botanici Pisani (PI) holds a number of Meneghini specimens. In a folder labelled "742. 4. Euhymenia dichotoma Kg." is an envelope and three sheets numbered "11". The envelope contains ten specimens of A. vidovichii, three of them on numbered sheets (224,778,783). Two of the three sheets (upper and lower left) each have a fragment of the specimen drawn in the abovementioned plate (Fig. 7). Consequently, we designate as type material of Chondrus? vidovichii   Meneghini the two fragments on two of the "11" sheets that were portions of the single specimen he illustrated in the unpublished fi gure accompanying his manuscript.

Phylogenetic analyses
The DNA barcode region was generated for 15 samples from different Mediterranean localities, including the type area; the sequences are now lodged in BOLD and Genbank (Table 1). Divergence among generated sequences ranged from 0-3 bp (0-0.53%), which is a typical level of within-species variation.
Phylogenetic analyses inferred from both rbcL and LSU markers ( Fig. 8 and

Discussion
Our molecular analyses have highlighted the well-known fact that some genera within the Halymeniaceae are not monophyletic assemblages, with various tree topologies for the family as a whole depending both on taxon sampling and phylogenetic signals of the markers (Fig 8). None of our analyses conclusively resolved the family position of Acrodiscus Zanardini, as its alliances changed depending on the reconstruction method.
The three supergeneric lineages within the Halymeniaceae emerging from our phylogenies have already been found in previous works (Manghisi et al. 2014;Mineur et al. 2010;Nelson et al. 2014), but morphological/anatomical characteristics unifying the members of the three lineages can still not be precisely specifi ed. The relationships among the several remaining genera represented in our analyses were poorly or not at all resolved, the polyphyly of some of them, such as Halymenia, Thamnoclonium and Cryptonemia, being clearly indicated. Studies such as those recently initiated for the complex that includes Grateloupia, Pachymeniopsis, Yonagunia and related genera (Calderon et al. 2014;Gargiulo et al. 2013) are needed before the generic relationships for the whole of the family Halymeniaceae can be satisfactorily resolved.
The anatomy of Acrodiscus is typically halymeniaceous, although in the absence of molecular data this evidence would not be defi nitive for family placement. Gametangial and gonimoblast structures remain unknown despite extensive herbarium collections and our examination of some 44 specimens, although the discovery of nemathecial tetrasporangia is a strong indication that gametophytes exist in some form yet to be discovered. Should they prove to be heteromorphic or cryptic stages, this would be a fi rst for a large family in which the members otherwise uniform in displaying isomorphic alternations of generations.
Although Schmitz (1889) provisionally included Acrodiscus in the genus Polyopes, the molecular data show that the two genera are not closely related. The grouping of Acrodiscus with the tropical genus Corynomorpha is not robust in our phylogenies, nor do habit and morphologies suggest a natural alliance. Nevertheless, this unlikely relationship should be further tested, as should a possible association with Felicinia and sistership with the Cryptonemia/Halymenia clade. For the moment we conclude on morphological and anatomical, as well and particularly on molecular, grounds that Acrodiscus is unquestionably an independent genus of the Halymeniaceae. The two poorly known species Cryptonemia crenulata and C. denticulata, both described by J. Agardh (1851) as having subapical sori of tetrasporangia similar to those of Acrodiscus, should be investigated as possible additional members of this presently monotypic genus.
Description of the historical herbarium material studied for the lectotypifi cation of Chrondrus ? vidovichii Menegh. Koster (1969) suggested that the Meneghini collection might be in Florence (FI) or in Kützing's collections, which are in Leiden (L) or London (BM). Therefore, the Herbarium Universitatis Florentinae was searched for the type of Meneghini, and we found that sheet FI 4796-1 holds fi ve sub-sheets, each with a specimens of Acrodiscus vidovichii, without locality, date or collector. Four of them are from Herbarium Meneghini; the fi fth one is mounted on a slide and accompanied by a manuscript annotation "Chondrus vidovichii col frutto, un pezzo del quale si staccò dalla cima e lo comprendo. -Sentirò con piacere che a quest'alga ha levato per sempre il punto interrogativo".
["Chondrus vidovichii with fruit, a fragment of which detached from the tip and I include it. -I will hear with pleasure that you have removed the question mark from this alga for ever".] The manuscript annotation has been compared to Meneghini manuscript documents, and the writing is different. The annotation is from someone sending a fertile specimen to Meneghini and hoping he will eliminate the question mark from Chondrus ? vidovichii. The fact that the specimen is fertile is actually an argument to posit that it cannot belong to Meneghini's original material at the time he described the species. For the remaining specimens there is no clear evidence of their being paratypes.
From the Nationaal Herbarium Nederland (L) we received on loan three herbarium sheets of A. vidovichii. L 0833934 belonged to Herbarium Suringar and has a single specimen with the manuscript label "Cryptopleura vidovichii Dalmazia!", but without mention of either a date or a collector. L 0833935 belonged to Herbarium Kützing and has two specimens, with the manuscript label "Cryptonemia vidovichii, * Chondrus vidovichii, Menegh. 1841 Cryptonemia dichotoma, J. Ag. 1842", the collector/ sender is Meneghini and the locality is Dalmatia; nevertheless, there is no mention of a date. The two specimens were drawn by Kützing in his Tabulae Phycologicae vol. 17 (Kützing 1867: tav. 72, fi g. d-e). Kützing in his Species Algarum (1849: 742) wrote "specim. dedit amic. Menghini", maybe in reference to these specimens. L 0833936 consists of two sheets. The fi rst one holds a specimen and fragment in a small envelope, with the manuscript label "Acrodiscus vidovichii Dalmazia lg. Vidovich", but without a date. The second sheet holds a single specimen, with the manuscript label "Acrodiscus vidovichii lg. Vidovich, com. Zanardini Dalmazien", and without date, but has another annotation in the upper part referring to Zanardini's Iconographia Phycologica Adriatica tv. LXIX, fi g. 3. In fact, the specimen corresponds to the one drawn in that fi gure.
In the Herbarium of the Natural History Museum of London (BM) we found the herbarium sheet BM000619430, with the manuscript label "Cryptonemia dichotoma J. Agardh, 1841 Nizza J. Agardh" (ex Herbarium Griffi ths), which is original material of Cryptonemia dichotoma J. Agardh 1842. Also BM000563717 and BM000563718 are both from Nizza and ex Herbarium J.G. Agardh, but without a date. Another interesting sheet is BM000569934, with the manuscript label "Acrodiscus vidovichii, Dalmazien, lg. Vidovich" (ex Herb. Weber van Bosse), because the specimen was collected at the type locality (topotype) by Vidovich, the same collector that sent material to Meneghini.
Unable to fi nd suitable material for the lectotypifi cation in the herbaria suggested by Koster (1969), we extended our search to other herbaria. In the Muséum national d'Histoire naturelle de Paris (PC) there are three interesting specimens/sheets. The fi rst one, PC0047928 (MA2052), is labelled "Cryptonemia vidovichii Zan. (Chondrus ? vidovichii Menegh., Cryptonemia dichotoma J. Ag.) Meneghini -Dalmazia"; thus, it belonged to Meneghini and was collected in Dalmatia, but there is no reference to the date of collection. The second one, PC0523534 (AR24258), is labelled "Cryptonemia dichotoma J Ag. , Nizza -