Materials and methods

Samples of Chondrophycus anabeliae and Laurencia digitata were collected in Cayo Muerto, Parque Nacional Morrocoy, Estado Falcón, Venezuela (10^o 55’ 48.08” N, 68^o 15’ 31.73” W) in 2015. For each sample, small fragments of the thallus were dried in silica gel for molecular analyses, and the remaining material was preserved in 4 % formalin-seawater or pressed as herbarium vouchers for morphological studies. For morphological examination, transverse and longitudinal hand sections were stained with 0.5 % aqueous aniline blue and acidified with 1N HCl. For each specimen studied, a minimum of 20 measurements of each morphometric character were made. Measurements are given as length × diameter. Images of whole specimens were taken with a Sony W5 digital camera (Sony, Tokyo, Japan), and details of branches and branchlets were captured with Sony W5 coupled to a Stemi SV 6-Zeiss stereomicroscope (Zeiss, Göttingen, Germany). Microscopic diagnostic features were taken with the Sony W5 coupled to a Nikon Eclipse E-200 optical microscope (Nikon, Tokyo, Japan). Voucher specimens were deposited in the herbaria of University of São Paulo (SPF) and Botanical Institute, São Paulo (SP), University of Carabobo, Venezuela (LUC), and Metropolitan Autonomous University (UAMIZ). Abbreviations follow Index Herbariorum (^{Thiers 2020}).

For molecular studies, DNA extraction followed manufacturer´s instructions of the DNeasy Plant Mini Kit (Qiagen, Valencia, USA). PCR protocols for both markers followed ^{Cassano et al. (2019)}, using for rbcL the following pairs of primers: FrbcLstart-R492, F492-R1150 and F993-RrbcS (^{Freshwater & Rueness 1994}), and for COI-5P the primer pair GAZF1-GAZR1 (^{Saunders 2005}). For PCR amplification was used the PCR Master Mix (Promega, Madison, Wisconsin USA) in a final volume of 25 μl. The reactions were performed in a Techne TC-4000 thermocycler (Bibby Scientific, Staffordshire, UK). All PCR products were analyzed by electrophoresis in 1 % agarose to check product size and were purified with MicroSpinTMS-300 HR Columns (GE Healthcare Life Sciences, Piscataway, New Jersey, USA) as per manufacturer’s instructions.

Sequencing reactions were made using the same PCR primers mentioned above, and the BigDye Terminator Cycle Sequencing Ready Reaction kit (Applied Biosystems, Foster City, California, USA) on an ABI PRISM 3730 Genetic Analyzer (Applied Biosystems). Consensus sequences and multiple sequence alignments for both rbcL and COI-5P were generated using BioEdit 7.0.4.1 software (^{Hall 1999}). Multiple sequence alignment for rbcL consisted of 93 sequences, including three newly generated sequences from Cayo Muerto, Venezuela; the remaining sequences were downloaded from GenBank (Appendix 1). Three Rhodomelacean species were used as outgroups, Chondria acrorhizophora Setchell & N.L.Gardner, Chondria collinsiana M.Howe, and Chondria dasyphylla (Woodward) C.Agardh (Appendix 1). Neighbor-joining (NJ) analysis was conducted in PAUP v4.0 beta10 (^{Swofford 2002}) with 2,000 bootstrap replicates. The most appropriate model of sequence evolution for maximum likelihood (ML) and Bayesian inference (BI) was selected using jModeltest v2.1.10 (^{Darriba et al. 2012}) under the Akaike information criterion (AIC) as implemented on the online server CIPRES Science Gateway v3.3 (^{Miller et al. 2010}). The model selected was the general-time-reversible model of nucleotide substitution with invariant sites and gamma-distributed rates for the variable sites (GTR+I+G). Maximum likelihood (ML) analysis was performed using IQ-Tree v1.4.3 (^{Nguyen et al. 2015}) with 1,000 boostrap replicates on the IQ-Tree web portal. BI analysis was performed using MrBayes v3.2.2 (^{Ronquist et al. 2012}). For BI analysis, two runs with four MCMC chains (one hot and three cold) were conducted with 4,000,000 generations and sampling every 1,000 generations, starting with a random tree. The first 100,000 generations in both runs were discarded as burn-in to build the consensus tree.

Multiple sequence alignment for COI-5P consisted of 69 sequences, including three newly generated sequences; the remaining were downloaded from GenBank (Appendix 1). One Rhodomelacean species was used as outgroup, Chondria baileyana (Montagne) Harvey (Appendix 1). The neighbor-joining (NJ) analysis was conducted in PAUP with 2,000 bootstrap replicates. Intra- and interspecific divergence values of rbcL and COI-5P were calculated using uncorrected “p” distances in PAUP.

Results

Molecular study. The rbcL final alignment included 93 sequences of 1,448 bp in length.

The Chondrophycus clade had full support (Figure 1). The two rbcL sequences of Chondrophycus anabeliae from Venezuela were identical and formed a subclade with Ch. anabeliae from Mexico, its type locality, with high to moderate supports (Figure 1), diverging by 1.34 %. This subclade was sister to Chondrophycus sp. 3 from New Caledonia plus Chondrophycus sp. from Australia from which it diverged by 5.17-5.9 %. The subclade formed by Chondrophycus cf. undulatus (Yamada) Garbary & J.T.Harper, Ch. sp.1, Ch. sp. 2 from New Caledonia, Chondrophycus dotyi (Y.Saito) K.W.Nam from Hawaii, and Chondrophycus tronoi (E.Ganzon-Fortes) K.W.Nam from Philippines diverged by 1.86 % (Ch. sp.1 vs Ch. cf. undulatus) to 6.4 % (Ch. doyti vs Ch. tronoi). The interspecific divergence within Chondrophycus genus ranging from 1.86 % to 7.8 % (Ch. anabeliae from Mexico vs Ch. sp.1 from New Caledonia).

Laurencia s.s. clade was highly supported (Figure 1). The rbcL sequence of L. digitata from Venezuela joined with L. digitata from South Africa, its type locality, with high to moderate support, diverging by 0.85 %. Laurencia digitata is sister to Laurencia cf. kuetzingii A.J.K.Millar from New Caledonia, diverging by 2.14-2.19 %. This subclade was resolved as sister to Laurencia pumila (Grunow) Papenfuss plus L. pumila var. dehoopiensis Francis, Bolton, Mattio & R.J.Anderson from South Africa plus Laurencia karachiana Bibi, Cassano & Rasheed from Pakistan with high to moderate supports. The interspecific divergence between sequences of L. digitata and L. pumila plus L. pumila var. dehoopiensis was 4.36-4.89 %, whereas L. digitata diverged from L. karachiana by 4.7-4.9 %.

The COI-5P final alignment included 69 sequences of 644 bp in length (Figure 2). We were unable to obtain COI-5P sequence for L. digitata from Venezuela due to contamination, even after several amplification attempts. The two identical COI-5P sequences of Ch. anabeliae fromVenezuela joined to the sequence of Ch. anabeliae from the type locality generated in this study (MN597440), diverging only by 0.3 % (Figure 2). This subgroup joined to Chondrophycus succisus (A.B.Cribb) K.W.Nam (as Laurencia succisa A.B.Cribb) from Molokai (Hawaii, USA) with 5.4-5.7 % of divergence. The subgroup formed by one sequence of Ch. dotyi and four of Ch. cf. undulatus, all from the Hawaiian Islands, showed low genetic divergence (0.16-0.48 %) suggesting that these samples represent the same taxonomic entity. The divergence between these two subgroups (Ch. anabeliae-Ch. succisus and Ch. dotyi-Ch. cf. undulatus) ranging from 8.13 % to 8.76 %.

Morphological study. Chondrophycus anabeliae ^{Sentíes, M.T.Fujii, Cassano & Dreckmann in Sentíes et al. 2016: 261, figures 1-18}. (Figures 3A, 4A-E, 5A-D, 6A-D)

Type locality. Mexico, Quintana Roo, Isla Mujeres, Garrafón de Castilla; holotype UAMIZ 1240!

Description. Plants forming erect tufts up to 5 cm high, reddish-brown to yellowish-brown, terete to partially compressed axes (Figures 3A, 4A), cartilaginous in texture, not adhering to herbarium paper when dried. Thalli attached to the substratum by a discoid holdfast. Erect branches irregularly alternate and spirally arranged, usually with 2-3 (4) orders of branches. The main axes are terete and first-order branches partially compressed mainly in the upper third portion of the thalli, up to 2,000 μm in diameter, and slightly narrowing towards the terete apices. Ultimate branchlets cylindrical to clavate and truncate at the apices, 500-2,800 μm long and 525-1,000 μm in diameter.

In surface view, the outermost cortical cells are translucent, isodiametric-polygonal in the middle portions, 17.5-32.5 × 17.2-37.5 μm without secondary pit connections (Figure 4B). Subcortical cells are pigmented, larger and connected to each other by secondary pit connections (Figure 4C). Outermost cortical cells translucent and subcortical cells present 1 (-2) crystals per cell (Figure 4C, D). In transverse section, thalli formed by two cortical cell layers, and four or five layers of medullary cells (Figure 4E). The cortical cells of translucent outer layer are quadratic, cuneiform to rectangular, smaller than the inner layer cells, measuring 17.5-30 × 22.5-30 μm in the ultimate branchlets (Figure 5A), and elongated, 47.5-55 × 27.5-35 μm in the middle portions of main axes. The inner layer of cortical cells is composed of pigmented and elongated cells, measuring 40-65 × 35-50 μm (Figure 5B) in the middle portions of main axes. Medullary cells are rounded or slightly radially elongated, measuring 65-145 × 47.5-107.5 μm in the middle portions of the main axes. Medullary cell walls uniformly thickened, but lenticular wall thickenings are absent. Each vegetative axial segment cuts off two pericentral cells (Figure 5C) that are slightly smaller than the medullary cells of the surrounding layer. In median longitudinal sections through a branchlet, the outer cortical cell walls near the apices are markedly projecting beyond the surface (Figure 5D).

Tetrasporangial branchlets are cylindrical or slightly compressed, simple or compound, 500-1,500 × 575-825 μm (Figure 6A). At the apex of fertile branches, each axial segment produces one fertile additional pericentral cell situated oppositely to the pre-existing two pericentral cells which remain vegetative (Figure 6B). The additional cell cuts off two pre-sporangial cover cells distally abaxially positioned in relation to the tetrasporangial initial (Figure 6C). Subsequently, one post-sporangial cover cell is produced and continues to divide, contributing to cortication around the tetrasporangia. Tetrasporangia are arranged in a right-angled pattern in relation to fertile branchlets (Figure 6D). Mature tetrasporangia are tetrahedrally divided, 50-100 μm in diameter. Gametangia were not observed.

Examined material. Venezuela. Estado Falcón: Parque Nacional Morrocoy, Cayo Muerto, 19 May 2015, tetrasporophyte, S. Ardito, M.T. Fujii, A. Sentíes, V. Cassano (SPF58487, SP470468, LUC7611, UAMIZ 1405). GenBank accession number for rbcL (MN597441, MN597442) and for COI-5P (MN597438, MN597439).

Distribution and habitat. This species is currently recorded only for Mexico (^{Sentíes et al. 2016}), and Venezuela (this study). Epilithic specimens were collected growing in shallow waters on rocky coastline, northeast of Cayo Muerto. The environment is considered as an intertidal zone with medium-sized rocks and moderately strong waves.

Laurencia digitata ^{Francis, Bolton, Mattio & R.J.Anderson in Francis et al. 2017: 812}, Figure 5. (Figures 3B, 7A-F, 8A-D)

Type locality. South Africa, KwaZulu-Natal, Cape Vidal; holotype BOL150572.

Description. Plants forming small, very intricate cushion-like tufts, up to 5 cm high, yellowish-brown, terete, cartilaginous in texture, adhering to herbarium paper when dried (Figure 3B). Thallus attached to the substratum by a discoid holdfast, and basal descending branches. Erect branches irregularly alternate and spirally arranged, with up to 3 orders of branches (Figure 7A). Main axes 275-525 μm in diameter in middle portion of the thallus. Ultimate branchlets are cylindrical to clavate with truncated tips, 452-950 × 225-325 μm in diameter.

In surface view, cortical cells have 1 (-2) corps en cerise per cell (Figure 7B). Cortical cells are arranged regularly in longitudinal rows and connected to each other by longitudinally oriented secondary pit-connections (Figure 7C). Cortical cells are rounded to polygonal and slightly longitudinally elongated in middle portions of main axes, 37.5-57.5 × 27.5-50 μm. In transverse section, the thallus has 1-2 layers of pigmented cortical cells and 3-4 layers of hyaline medullary cells (Figure 7D). Cortical cells are quadrate, cuneiform to rectangular, not arranged as a palisade, and 25-35 × 22.5-37.5 μm in the middle portions of thalli. Medullary cells are rounded to slightly radially elongated, and 55-90 × 42.5-52.5 μm, gradually increasing in size toward the center of the thallus. Each vegetative axial segment cut off four pericentral cells slightly larger than the other surrounding cells (Figure 7E). In median longitudinal sections through a branchlet, the outer cortical cell walls near the apices projecting beyond the surface (Figure 7F). Lenticular thickening absent.

Tetrasporangial branchlets are cylindrical, simple or compound, 575-2,125 × 225-400 μm (Figure 8A). At the apex of fertile branches, each axial segment produces one fertile pericentral cell, the fourth ones (Figure 8B), the other pericentral cells remain sterile. Fertile pericentral cell cuts off two pre-sporangial cover cells distally abaxially positioned in relation to the tetrasporangial initial (Figure 8D). Subsequently, one post-sporangial cover cell is produced and continues to divide, contributing to cortication around the tetrasporangia. Tetrasporangia are arranged in a parallel pattern in relation to fertile branchlets (Figure 8A, C). Mature tetrasporangia are tetrahedrally divided, 47.5-75 μm in diameter. Gametangia were not observed.

Material examined. Venezuela. Estado Falcón: Parque Nacional Morrocoy, Cayo Muerto, 19 May 2015, tetrasporophyte, S. Ardito, M.T. Fujii, A. Sentíes, V. Cassano (SPF58488, SP470469, UAMIZ 1406). GenBank accession number for rbcL (MN597443).

Distribution and habitat. This species is currently recorded only for South Africa (^{Francis et al. 2017}) and Venezuela (this study). Laurencia digitata was collected as drift specimens in shallow waters on rocky coastline, northeast of Cayo Muerto. The environment is considered as an intertidal zone with medium-sized rocks and moderately strong waves.

Discussion

Comparison of rbcL and COI-5P sequences of Ch. anabeliae from the type locality (Mexico) with our material confirmed the occurrence of this species on the Venezuelan coast, whereas L. digitata was confirmed only by comparison with rbcL sequence from the type locality (South Africa), since we were unable to generate COI-5P sequences and there are also no sequences available of this marker in databases.

Considering our results for rbcL, intraspecific value between the Venezuelan and South African L. digitata (0.85 %) is within the range observed in previous works for Laurencia s.s. (0-1.35 %) reported by ^{Cassano et al. (2012a}, ^b), ^{Metti et al. (2013)}, and ^{Collado-Vides et al. (2018)}. Interspecific divergences for the rbcL gene between L. digitata and Laurencia species closest molecularly (i.e., L. cf. kuetzingii, L. pumila, L. pumila var. dehoopiensis, and L. karachiana, 2.14-4.9 %) were within the variation observed for Laurencia s.s. by ^{Cassano et al. (2012b, 1.0-6.8 %)} and ^{Cassano et al. (2019, 2.4-3.7 %)}.

There are no data available in the literature for the rbcL gene to compare the intraspecific divergence for Chondrophycus. However, the intraspecific value obtained for Ch. anabeliae (1.34 %) is below the interspecific variation range described for the genus by ^{Cassano et al. (2012b, 1.8-6.9 %)}, by ^{Sentíes et al. (2016, 3.4-7.8 %)}, and that found in this study (1.86-7.8 %).

For COI-5P gene, the intraspecific divergence values reported for the Laurencia complex are low, not exceeding 1 %. The lowest range of intraspecific divergence was observed in this study (0-0.3 %) for samples of Ch. anabeliae from the Caribbean Sea (Venezuela and Mexico), whereas divergences up to 0.52 % were reported by ^{Machín-Sánchez et al. (2014)} for Laurencia, up to 0.67 % by ^{Machín-Sánchez et al. (2016)} for Osmundea, and up to 0.7 % by ^{Cassano et al. (2012b)}, ^{Machín-Sánchez et al. (2014)}, and ^{Collado-Vides et al. (2018)} for Laurenciella. The interspecific divergence obtained in this study for COI-5P (5.4-8.76 %) is within the range found for other genera of the complex, i.e. for Laurenciella species (7.4-9.2 %, ^{Collado-Vides et al. 2018}), and for Laurencia species (2.6-10.2 %, ^{Cassano et al. 2019}).

Morphologically, Venezuelan Ch. anabeliae shares all diagnostic characters described for the species by ^{Sentíes et al. (2016)}, such as: (i) thallus slightly compressed; (ii) two cortical cell layers, the outmost layer formed by translucent cells with conspicuous cell wall projections near the apices and absence of secondary pit connections, and the inner layer formed by pigmented and pit connected cells; and (iii) tetrasporangia produced from one additional fertile pericentral cell. The presence of 1 (-2) crystals per cortical and subcortical cells is a characteristic unique for Ch. anabeliae from Venezuela; they were not found in Mexican Ch. anabeliae and were not described for other Chondrophycus species.

Venezuelan and South African Laurencia digitata share the cushion-like tufted habit and absence of lenticular thickenings. However, the South African material differed from ours by color (reddish-brown), branching pattern (alternate but subopposite and/or subverticillate at the tips of some branches), and absence of cortical cell walls projections near the apices (^{Francis et al. 2017}). The tetrasporophyte has typical features of the genus Laurencia and is described for the first time for this species.

Chondrophycus anabeliae seems to be restricted to the Caribbean Sea so far. The range of distribution of Ch. anabeliae represents the limits from the western to the southeast Caribbean, whereas L. digitata, previously considered endemic to South Africa by ^{Francis et al. (2017)}, presents a disjunct distribution in the Indian and Atlantic Oceans. The disjunct distribution of L. digitata is similar to that of L. natalensis whose occurrence was recorded only for the Indian Ocean, nevertheless with wider distribution [South Africa, Mozambique, Kenya, Mauritius, Sri Lanka and Vietnam (^{Guiry & Guiry 2020})], and it was cited for Venezuela by ^{García-Soto & Lopéz-Bautista (2019)}. However, more studies of biogeography and phylogeography are needed in the area, which will allow us to explain this further.

Although macroalgal floristic studies have been carried out for the Venezuelan coast (e.g., ^{Gómez et al. 2017}), the diversity for this region is still underestimated. Similar underestimations have already been reported for other groups of red algae (e.g., ^{Adey et al. 2015}, ^{Basso et al. 2015}, ^{Hind et al. 2015}, ^{Ardito et al. 2017}, ^{Núñez-Resendiz et al. 2018}). For this reason, it is necessary to continue performing floristic surveys that incorporate molecular-assisted alpha taxonomy to accurately identify all and potential new species from this region, especially for ecologically and economically important taxonomic groups such as those included in the Laurencia complex.

The use of rbcL gene for phylogenetic inference, and the COI-5P barcode marker for species delimitation allied to morphological study revealed the presence of two species of the Laurencia complex for Venezuela, Chondrophycus anabeliae and Laurencia digitata. Both constitute new records for the region and their first occurrence outside the type localities. Our findings expand the geographic distribution of Ch. anabeliae to the southeast of the Caribbean Sea, where the species seems to be restricted so far, whereas the first report of L. digitata for the Atlantic Ocean established a disjunct distribution of this species in the Atlantic and Indian Oceans.