Material and methods

Fifty samples, consisting of compressed, flattened, and cylindrical specimens, identified as morphospecies of Gracilariaceae (Table S1) according to ^{Dreckmann (2012)}, were collected in 12 localities along the Yucatan Peninsula, at a depth of 1.5-2.0 m, in three different years (Table S1). Apical sections for molecular analysis were preserved in silica gel until DNA extraction was done. Samples collected for morphological observations were preserved in 3 % formaldehyde in seawater. Fresh specimens of each species were mounted on herbarium sheets and incorporated into the algal collections under a single voucher number, if they were collected in a same site and date (Table S1), at Metropolitan Herbarium UAMIZ (Herbarium abbreviations follow the online Index Herbariorum sciweb.nybg.org/science2/IndexHerbariorum.asp).

Total DNA for molecular analysis was extracted from 5-10 mg of dried tissue using a Qiagen DNeasy Plant Mini Kit (Qiagen, Valencia, California USA) according to the manufacturer’s protocols. The mitochondrial COI-5P region was amplified using the primers GazF1 and GazR1 (^{Saunders 2005}). The chloroplast rbcL region was amplified using the primers F8 (^{Shimada et al. 1999}), R753 (^{Freshwater & Rueness 1994}), and F765 (^{Wang et al. 2000}) R1381 (Freshwater & Rueness 1994). The PCR procedure followed ^{Núñez-Resendiz et al. (2015)}. PCR products were purified with QIAquick Gel Extraction Kit (Qiagen, Valencia, California USA) and sequenced commercially (Macrogen Inc., Seoul, Korea). The same set of primers was used for sequencing. The sequences generated were assembled and edited using the program Sequencher^® version 5.4.5. The final alignment with sequences from GenBank of other species of Gracilariaceae (Table S2) was performed using Bioedit (^{Hall 1999}). Cryptonemia seminervis (C. Agardh) J. Agardh and Laurencia mutueae Sentíes, Cassano & Dreckmann were included as outgroup in both data sets. We analysed the COI-5P and rbcL data sets separately. Phylogenetic analyses using maximum likelihood (ML) and Bayesian inference (BI) were performed separately with partitioned codons. The evolutionary model selected was GTR+I+G (general time reversible + invariable sites + gamma distribution) determined based on the Maximum Likelihood ratio test implemented by TOPALi version 2 software (^{Milne et al. 2009}). ML analysis was performed using RAxML software (^{Stamatakis 2006}) with the GTR+I+G model. Support for each branch was obtained from 1000 bootstrap replications. BI analysis was performed using MrBayes 3.2.2 (^{Ronquist et al. 2012}). Four chains of Markov chain Monte Carlo were used, starting with a random tree, and sampling the data every 500 generations for 5 × 10⁶ generations. 25 % of trees were discarded as burn-in. Pairwise distances values (p distance) were calculated using Mega X (^{Kumar et al. 2018}).

To delimit the Gracilariaceae species in the trees, we ran three DNA - based species delimitation methods for both data sets (COI-5P and rbcL): Automatic Barcoding Gap Discovery (ABGD) (^{Puillandre et al. 2012}), Bayesian variant of Poisson Trees Processes model (bPTP) (^{Pons et al. 2006}), and the General-Mixed-Yule-Coalescent (GMYC) (^{Zhang et al. 2013}). The ABGD method was done in the web interface (bioinfo.mnhn.fr/abi/public/abgd/abgdweb.html), with the following criteria: intraspecific variability (P) between 0.001 (Pmin) and 0.1 (Pmax), minimum gap width (X) of 0.1, Kimura-2-parameters, 50 screening steps and 20 Nb bins. The bPTP model was done via interface web (species.h-its.org/ptp), using the ML topology (see above). The analysis consisted of 100,000 generations, with a thinning every 100 generations and a burn-in of 25 %. For GYMC analyses, we generated an ultrametric tree for GYMC analyses in BEAST 1.8.2 software (^{Drummond et al. 2012}), from rbcL and COI-5P after removing identical sequences in the alignments. A coalescent constant size tree prior was set under an uncorrelated lognormal relaxed clock and GTR+G+I site model. The analysis was set up for 10 million generations and a sampling frequency of 5000. Before performing the GMYC analyses, we checked the estimated samples size with Tracer 1.6 (^{Rambaut et al. 2014}). The maximum clade credibility tree was computed using TreeAnnotator v1.8.3 (^{Drummond et al. 2012}). The resulting ultrametric tree was imported into the GMYC web server (species.h-its.org/gmyc), running the single threshold.

Photographs of thalli were taken with a Nikon D7000 digital camera. Microscopic cross-sections were made by hand using a razor blade and mounted in 80 % Karo^®/distilled water solution. Photomicrographs were taken using an Olympus DP12 digital camera adapted to an Olympus BX51 microscope (DIC and bright-field). Morphological measurements were obtained from micrographs using SigmaScan^© Pro automated image analysis software (Jandel Scientific, Sausalito, California).

Results

From our results with COI-5P and rbcL sequences, supported by DNA species delimitation methods and genetic distances, we were able to confirm the phylogenetic identity of ten species of Gracilariaceae (Figures 1, 2) distributed in the Yucatan Peninsula. Of the nine Gracilariaceae morphospecies, only Crassiphycus corneus, C. usneoides and Gracilariopsis tenuifrons confirmed their phylogenetic identity, while the remaining of morphospecies correspond to another identity, and are mentioned below: the samples identified as Gracilaria blodgettii and G. tikvahiae correspond to G. microcarpa; the samples identified as G. damicornis correspond to G. flabelliformis; the samples identified as G. mammillaris correspond to G. flabelliformis, G. occidentalis, G. hayi or G. suzanneae; the samples identified to Crassiphycus caudatus and G. cf. isabellana could be correspond to new phylogenetic identities. We have recognized eight species previously recorded in the area: Crassiphycus caudatus, C. corneus, C. usneoides, Gracilaria flabelliformis, G. cf. isabellana, G. microcarpa, G. occidentalis, and Gracilariopsis tenuifrons; and two species were new records for the Mexican Atlantic: G. hayi and G. suzanneae.

Figure 1 A. G. hayi from Bahía Tortuga. UAMIZ-1425. Scale bar = 4cm. B. Cross section of G. hayi showing cortical and medullary cells. UAMIZ-1425. Scale bar = 60 µm. C. G. suzanneae from Punta Xen. Scale bar = 4 cm. UAMIZ-1412. D. Cross section of G. suzanneae showing cortex and medullary cells. UAMIZ-1412. Scale bar = 60 µm. 

Figure 2 A. Gracilariopsis tenuifrons from Cancún. UAMIZ-1415. Scale bar = 10 cm. B. Crassiphycus caudatus from Bahía Tortuga. UAMIZ-1416. Scale bar = 2.5 cm. C. corneus from Puerto Progreso. UAMIZ-1417. Scale bar = 4 cm. D. C. usneoides from Puerto Progreso. UAMIZ-1420. Scale bar = 4 cm. E. Gracilaria microcarpa from Punta Xen. Scale bar = 4 cm. UAMIZ-1409. F-G. G. cf isabellana from Km 33 Carretera Champotón. UAMIZ-1431. Scale bar = 1 cm. H. G. occidentalis from Punta Xen. UAMIZ-1426. Scale bar = 4cm. I. G. flabelliformis from Bahía Tortuga. UAMIZ-1411. Scale bar = 4cm. I. G. flabelliformis from Xcalacoco. UAMIZ-1429. Scale bar = 3cm. 

Gracilaria hayi Gurgel, Fredericq & J.N.Norris in ^{Gurgel et al. 2004; 178, figure 7 A-F}(Figures 1A-B). Thallus flattened, 5 cm long, dark red (Figure 1A). Dichotomous branching in one plane. Branches flattened up to 1 cm wide, non-constricted at the base, smooth margins, rounded apices (Figure 1A). Thalli not adhering to paper. Gradual transition between medulla and cortex. Cortex with 1-2 cell layers, cells oval to round, 5-10 µm in diameter (Figure 1B). Subcortex composed of 2 cell layers, cells oval of 26-44 µm in diameter (Figure 1B). Medulla of 2-3 cell layers, cells oval to round, 120-170 µm in diameter (Figure 1B). Reproductive characters were not observed in sequenced specimens.

Gracilaria suzanneae L.P.Soares, C.F.D.Gurgel & M.T.Fujii in Soares et al. 2018; 346, 347, figures 1-22 (as 'suzannae') (Figures 1C-D). Thallus flattened, 8 cm long, brown with reddish apices, coriaceous texture (Figure 1C). Trichotomous to irregular branching. Branches flattened, 2-3 cm long, 1-7 mm wide, non-constricted at the base (Figure 1C). Apical branches with a lacerate appearance. Thalli not adhering to paper. Cortex with 2-4 cell layers, cells oval to rectangular, 5-27 µm in diameter (Figure 1D). Subcortex composed of 1-2 cell layers, cells oval of 19-49 µm in diameter (Figure 1D). Medulla of 2-3 cell layers, cells oval to round, 206-250 µm in diameter (Figure 1D). Reproductive characters were not observed in sequenced specimens.

The COI-5P alignment was 614 base pairs (bp) in length and included 55 sequences, 29 newly determined; rbcL alignment was 1,204 bp in length and included 69 sequences, 30 newly determined. With both markers, ML and BI analyses had identical topologies; only ML tree, with BI values integrated and ML bootstrap values are shown (Figures 3, 4). The topologies resulting with both markers showed that our specimens were included in three monophyletic groups (I, II, III). Group I, with the maximum phylogenetic support in both analyses, was integrated with the species of the genus Gracilariopsis (Figures 3, 4). Of all the specimens collected, identified morphologically in the field as Gracilariopsis, only the presence of Gp. tenuifrons was confirmed with a distribution from Campeche to Quintana Roo. Group II, with high support (COI-P, BI = 1.0, ML = 99 %; rbcL, BI = 1.0, ML = 100 %), included species of the genus Crassiphycus, and the presence of three species was confirmed (Figures 3, 4): C. caudatus with distribution in Campeche; C. corneus and C. usneoides distributed along the three States of the Yucatan Peninsula. Group III, also well-supported (COI-P, BI = 1.0, ML = 97 %; rbcL, BI = 0.99, ML = 99 %), included the species of the genus Gracilaria, and six species were confirmed (Figures 3, 4): G. flabelliformis distributed from Campeche to Quintana Roo; G. hayi distributed in Campeche; G. cf isabellana distributed in Quintana Roo; G. microcarpa distributed from Campeche to Quintana Roo; G. occidentalis distributed in Campeche; and G. suzanneae distributed in Campeche.

Figure 3 Maximum likelihood topology based on COI-5P sequences data. BI values (left) followed by ML bootstrap (right) on branches. Asterisks indicate full support (BI = 1.0, ML = 100 %), dashes indicate values below 70 %. Bars at the right of the tree indicate the results of the three species delimitations methods: Automatic Barcoding Gap Detection (ABGD), the General-Mixed-Yule-Coalescent (GMYC) and the Bayesian variant of Poisson Trees Processes model (bPTP). Sequences generated in this study are in boldface. S.P.S., substitutions per site. 

Figure 4 Maximum likelihood topology based on rbcL sequences data. BI values (left) followed by ML bootstrap (right) on branches. Asterisks indicate full support (BI = 1.0, ML = 100 %), dashes indicate values below 70 %. Bars at the right of the tree indicate the results of the three species delimitations methods: Automatic Barcoding Gap Detection (ABGD), the General-Mixed-Yule-Coalescent (GMYC) and the Bayesian variant of Poisson Trees Processes model (bPTP). Sequences generated in this study are in boldface. S.P.S., substitutions per site. 

Genetic distance within the species of the Gracilariaceae included in our analyses varied between the recognized genera. Species of Gracilaria ranged from 4.2 % (G. occidentalis from Mexico) to 17.6 % (G. dentata from Senegal) with COI-5P and from 0.8 % (G. suzanneae from Brazil) to 6.5 % (G. isabellana from Mexico) with rbcL. Species of Crassiphycus ranged from 5.2 % (C. usneoides from Mexico) to 10.3 % (C. caudatus from Brazil) with COI-5P and from 2.1 % (C. corneus from Venezuela) to 3.1 % (C. usneoides from Mexico) with rbcL. Species of Gracilariopsis ranged from 5.2 % (Gp. andersonii from Canada) to 6.5 % (Gp. tenuifrons from Brazil) with COI-5P and from 3.3 % (Gp. lemaneiformis from Mexico) to 3.6 % (Gp. costaricensis from Costa Rica) with rbcL.

In addition, for COI-5P data set, ABGD analysis defined 26 hypothetical species (nine of these from Yucatan Peninsula), and bPTP and GYMC analyses defined 24 (seven of these are from Yucatan Peninsula). For rbcL data set, ABGD analysis defined 22 hypothetical species (eight of these are from Yucatan Peninsula), and bPTP and GYMC analyses defined 24 (seven of these from Yucatan Peninsula).

Discussion

Based on his review with morphological data and geographic distribution of the species of Gracilariaceae occurring on the Mexican Atlantic coast, ^{Dreckmann (2012)} concluded that the diversity of the family in the area was overestimated. Consequently, he proposed that from the 30 species that had currently been recorded in the Yucatan Peninsula, only 12 species were distributed in the area: five species of Gracilaria (G. blodgettii, G. cervicornis, G. damicornis, G. mammillaris, and G. tikvahiae), four species of Crassiphycus (as Hydropuntia Montagne) (C. caudatus, C. corneus, C. usneoides, and C. crassus), and three species of Gracilariopsis (Gp. cata-luziana, Gp. lemaneiformis and Gp. tenuifrons). These data were later confirmed in a new review of family diversity on the Mexican coasts (^{Dreckmann & Sentíes 2014}).

Recently, ^{Hernández et al. (2020)} demonstrated, with molecular evidence, that specimens morphologically identified as Gp. lemaneiformis from Yucatan Peninsula correspond to Gp. tenuifrons and it is distributed along Atlantic Mexican coast, and probably this species is the only distributed in the area.

Our results suggest a dramatic reduction of the floristic diversity known for the family in the Yucatan Peninsula. Of the 30 species currently recorded, our investigation can only support the phylogenetic identity of ten species, including two new records, despite our having sequenced specimens with the morphological characteristics corresponding to most of the reported species, especially those described by ^{Dreckmann (2012)}. Of the confirmed species in our study, only the species of Crassiphycus (except C. crassissimus) and Gp. tenuifrons were consistent with those proposed by ^{Dreckmann & Sentíes (2014)} and ^{Hernández et al. (2020)}, respectively. Of the species of Gracilaria, none of those proposed by ^{Dreckmann & Sentíes (2014)} was confirmed by our results. In fact, all these morphospecies turned out to be morphotypes of G. flabelliformis or G. microcarpa (recently described by ^{Dreckmann et al. 2018}). Both species are very abundant in the area, the first of flattened thalli and the second with cylindrical thalli. Except for G. suzanneae and G. hayi, originally described from Brazil (^{Soares et al. 2018}) and Panama (^{Gurgel et al. 2004}), respectively, which constitute new records for the Mexican Atlantic coasts; the rest of these Gracilaria species had already been previously recorded in the area. Gracilaria isabellana has been previously recorded from the coast of Veracruz, but although we found at the Yucatan Peninsula specimens morphologically very similar to what we had called G. cf isabellana, phylogenetic analysis and the results with DNA species delimitation methods suggested that these specimens may consist of an undescribed species. Additional morphological observations and sequence data are required to confirm its true identity.

For the rest of the morphospecies (^{García-García et al. 2020}) of Gracilariaceae recorded for the Yucatan Peninsula and not confirmed in our study, eleven of them have type localities outside the Western Atlantic. G. armata, G. bursa-pastoris, G. gracilis, G. multipartite, and Gp. longissima have type localities in the European Atlantic and Mediterranean Sea; G. pacifica I.A. Abbott, Gp. andersonii, Gp. costaricensis and Gp. sjoestedtii in the American Pacific; and G. debilis and G. follifera in the Indian Ocean; which present a disjunct distribution, that is, an area of distribution interrupted by a geographical barrier (oceanic or continental). This type of distribution has been of great interest because geographic barriers can be the cause of evolutionary divergence processes between the populations involved, causing speciation. For example, regarding the Gracilariopsis species, ^{Hernández et al. (2020)} showed that the above-mentioned species are not present in the Mexican Atlantic, and that these records probably correspond molecularly to Gp. tenuifrons. So, the disjunct distribution of these species was ruled out. The morphological similarity between Gracilariopsis species is the main cause of misidentifications, which in turn leads to the wrong definition of their alleged disjunct distribution ranges and the overestimation of specific richness. This is a problem that may be occurring with Gracilaria species whose records in the American Atlantic do not correspond to their areas of distribution. However, another problem that could be the explanation of disjunct distribution for some species of Gracilariaceae family is the introduction of non-native species. Thus, the phylogenetic identified of these records not only help us clarify their taxonomic status, also their biogeographic processes.

Another important factor to consider in the molecular identification of the species found in this study is that the values of interspecific distance between them. In some cases, such as Gracilaria hayi, G. suzanneae, and G. curtissiae, the values of the interspecific distances ranged between 0.8 and 1.5 %, which is less than the range of interspecific values proposed by ^{Gurgel et al. (2004)} to delimit species of Gracilariaceae (from 2 to 3.4 %). From our results, a morphometic study of these species could provide additional evidence to support their phylogenetic independence, as it was in the case of Crassiphycus species, previously considered as cryptic species and currently discriminated by morphometric analysis (^{Vilchis et al. 2019}, ²⁰²⁰).

Considering the values of the interspecific distances, the morphological similarity between several species in the Gracilariaceae, and the results using DNA species delimitation methods, it seems that some species in the family have been wrongly described as new diversity; consequently, they must be critically reviewed. For example, in a very recent study (^{Lyra et al. 2021}), the floristic diversity for the Gracilariaceae species from Brazil was discussed under morphological and molecular evidence. Consequently, genera previously described as independent from Gracilaria due to their paraphyletic position in the phylogeny (^{Gurgel et al. 2018}, ²⁰²⁰), have been merged back into a single genus (^{Lyra et al. 2021}). Crassiphycus was one of the genera included in a broadly circumscribed Gracilaria sensu lato proposed by ^{Lyra et al. (2021)}. However, from our phylogenetic analysis, we recognize Crassiphycus as an independent genus of Gracilaria. Likewise, in the study by ^{Lyra et al. (2021)}, Crassiphycus usneoides was not included within the diversity of Gracilariaceae in Brazil, because they did not molecularly confirm its presence and discussed, in agreement with ^{Núñez-Resendiz et al. (2015)}, the difficulty of recognizing them morphologically. Previously, C. usneoides was included as part of a complex of cryptic species along with C. corneus by ^{Núñez-Resendiz et al. (2015)}, both species with type localities in Brazil. In their study they described that within the morphological spectrum of the two species included in the complex, there were two genetically independent entities, although morphologically indistinguishable. Recently, ^{Vilchis et al. (2019}, ²⁰²⁰⁾ was able to discriminate both genetic entities, by applying morphometric techniques, both by the diameter in the middle and apical portion of the thallus (significantly larger in C. usneoides than in C. corneus) and by the shape of the cells (narrower in C. usneoides). Given that C. corneus and C. usneoides were circumscribed only with morphological evidence, the characters that were used for discriminated them overlap between the two species, so it may be that in Brazil only C. corneus is distributed. However, our morphological and molecular results do support the presence of both species in the Yucatan Peninsula.

In conclusion, based on our current results, we could detect the phylogenetic identity to 10 species from the molecular analysis of nine morphospecies. These identities correspond to species previously recorded and two new records for the Yucatan Peninsula. The study demonstrated that the molecular assessment has proved to be very useful for the diversity evaluation, thus, the future phylogenetic identifies for the rest of morphospecies recorded in the area will allow a better approximation of Gracilariaceae diversity species.

Supplementary material

Supplemental data for this article can be accessed here: https://doi.org/10.17129/botsci.2915

Supplemental material, Table S1, Table S2.