Morphological and Molecular Characters Differentiate Common Morphotypes of Atlantic Holopelagic Sargassum

: Since 2011, massive new strandings of holopelagic Sargassum have been reported on the coasts of the Caribbean, northern Brazil, Guiana

Holopelagic Sargassum species, historically most abundant in the Sargasso Sea and open Gulf of Mexico [21,22], float and grow at the sea surface during their entire lifetime, as they have no benthic stage.Individual thalli decay and sink to depth [23] or can wash ashore, removing them from the pelagic system.At the sea surface, Sargassum thalli tend to aggregate and form mats of different structure and density [24], which host a diverse community of marine organisms [25][26][27][28].These drifting aggregations can travel long distances under the action of winds, waves, and currents [22,24,29].Since 2011, unprecedented massive strandings of holopelagic Sargassum have been reported on the coasts of the eastern and western Caribbean, northern Brazil, Guiana, and West Africa (reviewed in Fidai et al. [30]).These recurrent strandings are associated with extensive Sargassum aggregations in the tropical North Atlantic, as reported by satellite imagery [31], hydrographic modeling [32], and field campaigns [24,33].Satellite images and direct observations show that the Great Atlantic Sargassum Belt (GASB) phenomenon has been annually variable in magnitude [31].Thus, the absence of reliable growth and dispersion models for holopelagic Sargassum complicates bloom forecasting efforts [34][35][36][37] and the establishment of suitable management procedures.
Species identification adds another level of complexity to understanding the origins and composition of these recent massive standings.The growth response [38][39][40][41][42][43] and morphology-induced windage [44,45] indeed differ between Sargassum morphotypes.In addition, in the aggregations, the most abundant morphotype has shifted across annual blooms.Schell et al. [33] reported, in the North Atlantic Ocean between 10 and 30 • N, three dominant morphotypes described by Parr [21], namely, Sargassum natans VIII, Sargassum natans I, and Sargassum fluitans III, with S. natans VIII noted as the predominant form inundating eastern Caribbean coasts.Later, S. fluitans III dominated beached Sargassum in Mexico during 2016 and S. natans I contributed substantially to the community composition in the same location in 2018 [46].On a finer temporal scale, the morphotype composition of beached holopelagic Sargassum in Barbados switched from S. fluitans III dominating during spring/summer 2021 to S. natans VIII-dominated during autumn 2021/winter 2022 [27].
Integrative taxonomy [47], combining a morphological study with molecular analyses, can clarify whether the different holopelagic Sargassum morphotypes observed in recent years correspond to different species or to intra-specific variants.As Winge [20] and Parr's [21] morphotype descriptions did not follow the Botanical Code (International Code of Nomenclature for algae, fungi, and plants; [48]) for species description and the subsequent discontinuity in classification based on inconsistent character selection [49], a comprehensive morphological description of each common morphotype is warranted.Moreover, molecular analysis challenged traditional taxonomic divisions within the genus and questioned the use of some morphological or chemical characters for species delineation and taxonomy [50,51].Even the monophyly of the Sargassum genus has been disputed [52].
Early sequencing efforts revealed no divergence between holopelagic species nor between holopelagic and some benthic species [53][54][55][56].Such results may have arisen from the selection of markers that, while standard for macroalgal studies, did not capture the regions where polymorphisms among holopelagic Sargassum were located [53][54][55][56].Moreover, some included the incorrect identification of the specimens sequenced [54,56].Subsequent comparisons of full-length mitogenomes, with a documented correct morphotype identification, indicated a closer proximity between S. natans I and S. natans VIII when compared to S. fluitans III; the divergence levels were around 0.02% at the mitogenome level between S. natans I and S. natans VIII and 0.3% between S. natans I/VIII and S. fluitans III [57].In a more extended temporal and geographic analysis of holopelagic Sargassum, targeting multiple mitochondrial gene segments each with multiple polymorphisms, genetic distances among holopelagic morphotypes were in the range of those between pelagic and benthic species and on the same order as divergence among some benthic species [58].
The present study aimed to clarify the taxonomic status of holopelagic Sargassum morphotypes presently found in the Atlantic Ocean.A broad suite of morphological characters and multi-gene molecular analyses were conducted across a large sample of individuals representing all three common holopelagic morphotypes.With this deeper understanding of Sargassum taxonomy, the scientific record can be rectified and an appropriate level of differentiation can be incorporated into future modeling, forecasting, and ecological studies.

Sample Collection
Holopelagic Sargassum specimens were randomly collected during two scientific cruises in the North and tropical Atlantic Ocean in 2017, with the goal of studying the composition and characteristics of Sargassum aggregations.The Caribbean expedition (http://dx.doi.org/10.17600/17004300,accessed on 28 March 2024) used the R/V Antea from 19th June to 13th July 2017, and extended from the area of extensive Sargassum bloom development off northeast Brazil up to the Sargasso Sea (25 • N), with a focus along French Caribbean islands.The Transatlantic expedition (http://dx.doi.org/10.17600/17016900,accessed on 28 March 2024) used the M/Y Yersin from 6th to 24th October 2017, and crossed the Atlantic between 8 • and 12 • N, from Cape Verde to Martinique (see [24] for details).
Holopelagic Sargassum aggregations were randomly sampled at 31 of 49 stations during the two oceanographic cruises (Figure 1).The three morphotypes were simultaneously present at all 31 sampling stations, but proportions differed by site.Sargassum fluitans III was dominant nearly everywhere (around 70%), with ~15% each for S. natans VIII and S. natans I.However, at two stations (18 and 20) in the south Sargasso Sea, S. natans VIII dominated (~70%) and S. natans I and S. fluitans III contributed ~15% each.level between S. natans I and S. natans VIII and 0.3% between S. natans I/VIII and S. fluitans III [57].In a more extended temporal and geographic analysis of holopelagic Sargassum, targeting multiple mitochondrial gene segments each with multiple polymorphisms, genetic distances among holopelagic morphotypes were in the range of those between pelagic and benthic species and on the same order as divergence among some benthic species [58].
The present study aimed to clarify the taxonomic status of holopelagic Sargassum morphotypes presently found in the Atlantic Ocean.A broad suite of morphological characters and multi-gene molecular analyses were conducted across a large sample of individuals representing all three common holopelagic morphotypes.With this deeper understanding of Sargassum taxonomy, the scientific record can be rectified and an appropriate level of differentiation can be incorporated into future modeling, forecasting, and ecological studies.

Sample Collection
Holopelagic Sargassum specimens were randomly collected during two scientific cruises in the North and tropical Atlantic Ocean in 2017, with the goal of studying the composition and characteristics of Sargassum aggregations.The Caribbean expedition (http://dx.doi.org/10.17600/17004300,accessed on 28 March 2024) used the R/V Antea from 19th June to 13th July 2017, and extended from the area of extensive Sargassum bloom development off northeast Brazil up to the Sargasso Sea (25° N), with a focus along French Caribbean islands.The Transatlantic expedition (http://dx.doi.org/10.17600/17016900,accessed on 28 March 2024) used the M/Y Yersin from 6th to 24th October 2017, and crossed the Atlantic between 8° and 12° N, from Cape Verde to Martinique (see [24] for details).
Holopelagic Sargassum aggregations were randomly sampled at 31 of 49 stations during the two oceanographic cruises (Figure 1).The three morphotypes were simultaneously present at all 31 sampling stations, but proportions differed by site.Sargassum fluitans III was dominant nearly everywhere (around 70%), with ~15% each for S. natans VIII and S. natans I.However, at two stations (18 and 20) in the south Sargasso Sea, S. natans VIII dominated (~70%) and S. natans I and S. fluitans III contributed ~15% each.At each station, collected Sargassum thalli were identified and sorted by morphotype.For each Sargassum morphotype, two to three replicate fronds were randomly selected for analysis.A total of 264 individuals of holopelagic Sargassum were collected to carry out joint morphological and genetic analyses.All specimens have been kept as vouchers with a corresponding fragment stored in silica gel at room temperature for further molecular analysis.Eight specimens of benthic Sargassum, five of S. hystrix, two of S. furcatum, and one of S. polyceratium var.ovatum were also collected by scuba-diving and snorkeling in the French Caribbean islands of Guadeloupe and Martinique and used as out-groups in phylogenetic analyses.All specimens are kept in the Herbarium of the Mediterranean Institute of Oceanography at Aix-Marseille University, Marseille, France (HCOM).Herbarium abbreviations follow Thiers [59].

Morphological Analyses
Measurements were made on dried specimens for comparison with dried specimens stored in different herbaria housed in HCOM.Fifty-four characters were identified and selected based on analysis of all available descriptions from the literature of Sargassum identification keys from the North Atlantic Ocean [49,[60][61][62][63][64][65] (Table S1).These 54 characters were observed and measured on 264 individuals; they correspond to all the specimens collected during the cruises.
Non-Multi-Dimensional Scaling (nMDS) was performed based on the resulting 54 characters (coding 0, 1) and 264 individual matrices (Table S1), using S17 Bray-Curtis similarity distance.PERMANOVA and pair-wise tests were applied to detect significant differences between the three examined morphotypes.P-values were obtained by 999 permutations of residuals under a reduced model [66][67][68] The final number of sequences obtained for each marker depended on PCR and sequencing success, and on the observed levels of variability between Sargassum morphotypes; fewer samples for psbA were sequenced, because that marker exhibited the lowest level of variability.The numbers of sequences analyzed per marker and morphotype are presented in the resulting phylogenetic trees and haplotype networks (see Results).The PCR reaction mix for all loci was prepared for a final volume of 20 µL with GoTaq ® Flexi Buffer 1X, GoTaq ® DNA Polymerase 0.02 U/µL, dNTP 0.2 mM, MgCl 2 5 mM, forward and reverse primers 0.5 µM each, BSA 0.8 mg/mL, and 2 µL of diluted DNA.All PCR programs included 3 min at 94 • C, 35 cycles with [1 min at 94 • C, 1 min or 1 min 30 s at annealing temperature (Table 1), and 1 to 2 min at 72 • C], and 5 min at 72 • C. The PCR products were sent for Sanger sequencing at Eurofins Genomics.Each unique sequence (i.e., haplotype) obtained in this study is available in GenBank under accession numbers MT422788 to MT422805, and the final sequence alignments have been submitted in Zenodo.

Analysis of Sequence Data
Sequences were visualized and aligned with UGENE v 1.31.1 [71].For each marker, all Sargassum sequences available were retrieved from Genbank and aligned with the new sequences obtained here.Sequence alignments were performed with MUSCLE [72] in UGENE.Markers were first analyzed separately, then some were concatenated for tree or network construction.Concatenation was not always possible because of the very unbalanced distribution of Sargassum species across markers in Genbank.
For phylogenetic analyses, two datasets were used: a psbA alignment allowing a comparison with distant Phaeophyceae taxa, and a rbcL-cox3 concatenated alignment for the Sargassum genus.Other markers were not used because of the lack of corresponding sequences in Genbank or difficulty aligning evolutionary distant taxa (as in the case of mitochondrial IGS).A Maximum Likelihood (ML) phylogenetic reconstruction was performed with IQ-TREE 1.6.8[73] and a Bayesian Inference (BI) with MrBayes 3.2 [74,75].With IQ-TREE, 1000 ultrafast bootstraps were used, along with the bnni option to reduce risk of overestimating branch support with ultrafast bootstrap [76].In the case of combined datasets, the ModelFinder option of IQ-TREE was used to find the best partition scheme [77].For BI, a sample frequency of 100 over 10 6 generations (burnin 250,000) was used for rbcL-cox3 and 2•10 6 generations (burnin 10 6 ) was used for psbA.Convergence of the MCMC analysis was assumed when the average standard of split frequency was below 0.01, when the log probability of the data did not show any trend, and when the potential scale reduction factor was close to 1.0.Trees were visualized with FigTree 1.4.4 [78].For the analyses based on psbA sequences, Cutleria and Bachelotia sequences were used as outgroups [79].For rbcL-cox3, according to the present results and to Mattio and Payri [51], the tree was rooted with the clade containing two species, Sargassum ringgoldianum Harvey and Sargassum thunbergii (Mertens ex Roth) Kuntze, belonging to the subgenus Bactrophycus.Because of different representation in Genbank, the psbA and rbcL-cox3 dataset did not lead to the same species lists; previous S. fluitans and S. natans rbcL sequences were not included, as available sequences for these taxa were only partially aligned with the ones obtained in this present study.
As a complementary approach, haplotype networks were built with the TCS 1.21 software [80], in order to study relationships among the most closely related sequences, to include indels in the analysis, and to visualize the frequency of the different haplotypes.Indels longer than one base pair were transformed into a single base-pair indel before analysis.Networks were built independently for five datasets: mt spacer IGS, a concatenation of partial cox2 and cox3, nad6, and 16S rRNA mitochondrial sequences, as well as part of the plastidial rbcL.

Morphological Analyses
According to the morphological analysis of the 54 characters (Table S1), all individuals of Sargassum examined during the oceanographic cruises unambiguously belonged to one of three dominant morphotypes described by Parr [21], namely, Sargassum fluitans III, Sargassum natans I, and Sargassum natans VIII.Importantly, no intermediate morphology was observed.Analysis of all collected individuals (n = 264) based on all characters showed three distinct, clearly separated clusters, matching with the three morphotypes (Figure 2).In addition, these clusters differed significantly, as a whole and by pairs (Table 2).
one of three dominant morphotypes described by Parr [21], namely, Sargassum fluitans III, Sargassum natans I, and Sargassum natans VIII.Importantly, no intermediate morphology was observed.Analysis of all collected individuals (n = 264) based on all characters showed three distinct, clearly separated clusters, matching with the three morphotypes (Figure 2).In addition, these clusters differed significantly, as a whole and by pairs (Table 2).Among the 54 characters analyzed, several were particularly discriminating: the overall shape, the nature of the axis (thorny or smooth), the blades (size and shape), and the vesicles (air bladders; size, presence or absence of apical spine-like appendage; length of pedicel) (Figures 3 and 4).Briefly, Sargassum fluitans III was characterized by (1) a short to loose, globular or cylindrical, open, and stringy shape, (2) a thorny axis, (3) blades that are lanceolate with an asymmetrical base, and (4) short pedicellate, oblong vesicles without an apical spine-like appendage but frequently with a foliaceous margin.Sargassum natans I was characterized by (1) a mostly cylindrical, sometimes pyramidal shape, (2) a  Among the 54 characters analyzed, several were particularly discriminating: the overall shape, the nature of the axis (thorny or smooth), the blades (size and shape), and the vesicles (air bladders; size, presence or absence of apical spine-like appendage; length of pedicel) (Figures 3 and 4).Briefly, Sargassum fluitans III was characterized by (1) a short to loose, globular or cylindrical, open, and stringy shape, (2) a thorny axis, (3) blades that are lanceolate with an asymmetrical base, and (4) short pedicellate, oblong vesicles without an apical spine-like appendage but frequently with a foliaceous margin.Sargassum natans I was characterized by (1) a mostly cylindrical, sometimes pyramidal shape, (2) a smooth axis, (3) linear blades that are narrow with a symmetrical base, and (4) vesicles, frequently with a long pedicel, a foliaceous margin, and an apical spine-like or reduced foliaceous appendage.S. natans VIII was well-characterized by (1) its measurements larger than those of the two other morphotypes and a mostly pyramidal shape, (2) a smooth axis, (3) lanceolate and broad blades with a symmetrical base, and (4) vesicles that are shortly pedicellate, usually without an apical spine-like appendage (Figures 3 and 4; see key for identification).
frequently with a long pedicel, a foliaceous margin, and an apical spine-like or reduced foliaceous appendage.S. natans VIII was well-characterized by (1) its measurements larger than those of the two other morphotypes and a mostly pyramidal shape, (2) a smooth axis, (3) lanceolate and broad blades with a symmetrical base, and (4) vesicles that are shortly pedicellate, usually without an apical spine-like appendage (Figures 3 and 4; see key for identification).

Genetic Analyses
Molecular phylogenetic analyses were conducted for pbsA and rbcL-cox3 concatenated sequences.The phylogenetic tree made with psbA did not reveal any variation among the three morphotypes, and the sequences were identical to benthic species such as Sargassum vulgare (Figure S2).Thus, the tree did not clarify the phylogenetic position of

Genetic Analyses
Molecular phylogenetic analyses were conducted for pbsA and rbcL-cox3 concatenated sequences.The phylogenetic tree made with psbA did not reveal any variation among the three morphotypes, and the sequences were identical to benthic species such as Sargassum vulgare (Figure S2).Thus, the tree did not clarify the phylogenetic position of the holopelagic samples but confirmed that they belong to the Sargassum genus.Identical psbA sequences were also observed between close species within the Sargassaceae family, The molecular identification of the three holopelagic Sargassum morphotypes, among each other and in comparison with benthic species, was dependent upon the marker used for analysis.The new mt spacer IGS for the S. fluitans III morphotype was identical to previous sequences from S. fluitans III.New mt spacer IGS sequences for the S. natans VIII and S. natans I morphotypes were identical to each other and to S. natans VIII and S. natans I sequences from Amaral-Zettler et al. [57].The TCS network showed a divergence of six mutations between the S. natans complex and S. fluitans III, including two indels of two base pairs (bps) each (Figure 6).For plastidial rbcL, the fragment sequenced did not completely match with the ones previously sequenced for S. natans and S. fluitans.Therefore, no comparisons could be made beyond this dataset.The morphotypes S. natans VIII and S. natans I shared the same haplotype with the benthic Sargassum analyzed in this study.This haplotype was separated by two substitutions in rbcL and a 9 bp deletion in the spacer (not shown) from the morphotype S. fluitans III haplotype (Figure 6).One haplotype of morphotype S. fluitans III (station 20) was separated from other morphotype S. fluitans III samples by four mutations, and from the other multi-species haplotypes by one mutation and a 6 bp indel.The mitochondrial 16S rRNA, cox2-cox3 concatenated, and nad6 se-Figure 5. Phylogenetic tree based on the concatenation of rbcL and cox3 sequences.The topology shown here corresponds to the output of the Maximum Likelihood analysis.Bootstrap values higher than 80%, and posterior probabilities higher than 0.9 are indicated at the left of the corresponding group.Newly generated sequences are in red.For the sequences retrieved from GenBank, the accession number is indicated together with the species name (corrected with current taxonomic status).
The molecular identification of the three holopelagic Sargassum morphotypes, among each other and in comparison with benthic species, was dependent upon the marker used for analysis.The new mt spacer IGS for the S. fluitans III morphotype was identical to previous sequences from S. fluitans III.New mt spacer IGS sequences for the S. natans VIII and S. natans I morphotypes were identical to each other and to S. natans VIII and S. natans I sequences from Amaral-Zettler et al. [57].The TCS network showed a divergence of six mutations between the S. natans complex and S. fluitans III, including two indels of two base pairs (bps) each (Figure 6).For plastidial rbcL, the fragment sequenced did not completely match with the ones previously sequenced for S. natans and S. fluitans.Therefore, no comparisons could be made beyond this dataset.The morphotypes S. natans VIII and S. natans I shared the same haplotype with the benthic Sargassum analyzed in this study.This haplotype was separated by two substitutions in rbcL and a 9 bp deletion in the spacer (not shown) from the morphotype S. fluitans III haplotype (Figure 6).One haplotype of morphotype S. fluitans III (station 20) was separated from other morphotype S. fluitans III samples by four mutations, and from the other multi-species haplotypes by one mutation and a 6 bp indel.The mitochondrial 16S rRNA, cox2-cox3 concatenated, and nad6 sequences separated the three morphotypes and confirmed matches between the sequences obtained in this study and earlier published sequences for all three holopelagic morphotypes.The 16S rRNA sequences for S. fluitans III differed from the benthic species and from S. natans I by a single substitution and from S. natans VIII by two substitutions (Figure 6).Two S. natans VIII sequences from the current dataset matched the S. natans I sequences, and one S. hystrix sequence matched the S. fluitans III sequences.In the cox2-cox3 network, haplotypes from morphotypes S. natans VIII and S. natans I were only separated by one substitution in the cox3 partial fragment (Figure 6).Haplotypes from morphotypes S. natans VIII and S. fluitans III were separated by four substitutions, one in cox3 and three in cox2.The nad6 network was composed of multiple grouped haplotypes representing each holopelagic Sargassum morphotype, with one dominant haplotype in each group (Figure 6).Two S. natans VIII and one S. fluitans III matched the dominant S. natans I haplotype, which differed from the dominant S. fluitans III haplotype by two nucleotide substitutions and from the dominant S. natans VIII by one nucleotide substitution.

Discussion
Morphology strongly supports the delineation between the three holopelagic Sargassum morphotypes (S. fluitans III, S. natans I, and S. natans VIII).No individuals displayed intermediate characters, and individuals defined by their morphotype were clearly separated into groups specific to each (Figure 2).S. fluitans III, described by Parr [21], morpho-

Discussion
Morphology strongly supports the delineation between the three holopelagic Sargassum morphotypes (S. fluitans III, S. natans I, and S. natans VIII).No individuals displayed intermediate characters, and individuals defined by their morphotype were clearly separated into groups specific to each (Figure 2).S. fluitans III, described by Parr [21], morphologically corresponds to the description of S. fluitans (Børgesen) Børgesen [18,19].Consequently, S. fluitans III can be renamed as Sargassum fluitans var.fluitans.S. fluitans X, the other described S. fluitans morphotype [21], is rare; upon further analysis, it should be assigned the appropriate nomenclature.Moreover, S. natans I and S. natans VIII are syntopic, i.e., living together in the same habitat at each station, which ruled out the hypothesis that the observed morphological differences were related to their habitat differences.These two taxa, namely, S. natans I and S. natans VIII, could be in an early stage of speciation (i.e., the gray zone following de Queiroz's terminology [81]).Thus, we conservatively propose their formal separation into varieties as Sargassum natans var.natans (S. natans I) and Sargassum natans var.wingei (S. natans VIII).
Molecular results confirmed the phylogenetic positioning of Atlantic holopelagic Sargassum species, in close proximity to each other and among benthic species within a group corresponding to the section Sargassum of the subgenus Sargassum, as previously suggested (e.g., [51,54]).Nevertheless, within the section Sargassum, the low divergence of the sequences used as well as the reduced taxonomic sampling were not sufficient to unambiguously resolve the phylogenetic affinities of holopelagic species and, therefore, do not allow the discussion of the question of a single or multiple evolutionary origin of holopelagic Sargassum.The difficulty in resolving the phylogeny of this group could be due to a combination of a relatively recent evolution of the pelagic lifestyle, along with the slow molecular evolution of mitochondrial and plastidial genes considered.
In other Sargassum species, mitochondrial cox3 or plastidial rbcL showed intraspecific variation, sometimes useful for phylogeographic studies [50,70,82].Therefore, the lack of diversity inside holopelagic Sargassum species and observed with our marker set may not be a fixed rule.The discrepancy between morphological differentiation and low molecular divergence may be explained by this slow rate of molecular evolution of cytoplasmic genomes in Sargassum species.More generally, the observation of morphological differences can usually be explained by the interaction between genetic and environmental factors, but not in this case.There are also cases where species that are very similar genetically may exhibit significant morphological differences due to environmental influences or other factors (e.g., [83,84]), but holopelagic Sargassum taxa live in sympatry and are always found mixed together, even if their abundance can vary from year to year and/or place to place [27].
Regarding species/morphotype delineation using molecular markers, the examined genes exhibited only a few mutations separating the three morphotypes, as reported by Amaral-Zettler et al. [57] and Dibner et al. [58] and demonstrated in previous studies on nuclear, mitochondrial, and chloroplastic markers dedicated to the genus Sargassum [85][86][87].The present study confirmed weak but clear genetic differences (three to six mutations) for the mitochondrial IGS, cox2, cox3, and plastid rbcL genes, between morphotypes S. natans I and VIII, and S. fluitans III [57].An additional analysis of the mitochondrial 16S rRNA and nad6 genes also confirmed weak but clear genetic differences between S. natans I and S. natans VIII [58].In all cases, genetic data from these 2017 samples were consistent with previous sampling efforts between 2012 and 2016 [57] and between 2015 and 2019 [58], which seems to indicate a relative stability of composition at that scale.The fact that observed genetic differences between S. natans I and S. natans VIII were low may be the result of the aforementioned low rate of evolution for cytoplasmic genomes.In future studies, other more polymorphic markers may reveal a higher divergence between the two morphotypes.
Within the genus Sargassum, similarities in molecular marker sequences between distinct species are frequent.Yoshida et al. [86] demonstrated similarities in the nuclear ITS2 sequences of four allied Sargassum species belonging to the subgenus Bactrophycus: Sargassum boreale, S. confusum, S. microceratium, and S. pallidum.In that work, S. boreale was identified as a new species because of the presence of a large deletion of 28 bp, which was stable in all specimens of S. boreale studied [86].No nucleotide difference was observed in the ITS2 sequences of Sargassum mcclurei and S. quinhonense belonging to the subgenus Sargassum although their morphology is totally different [85].The separation of the two species was not supported by the ITS2 phylogeny; nevertheless, S. mcclurei and S. quinhonense are distinct species as no intermediate morphological form has been reported [85].On another Sargassaceae genus, Rohfritsch et al. [88] demonstrated that a divergence of almost 17 mutations on 2102 concatenated base pairs was enough to confirm that Turbinaria ornata and T. conoides were separate species.
The proposed revision is particularly critical at this time, as considerable confusion exists regarding the morphological identification of holopelagic Sargassum species in the literature.To address and resolve these inconsistencies, a thorough historic review of holopelagic Sargassum taxonomy is presented here.Early taxonomic classification of holopelagic Sargassum recognized two species, Sargassum natans (L.) and S. hystrix (Agardh) var.fluitans, based on several morphological traits that clearly distinguished one species from the other [18,19].In addition to differences in the shape of blades and vesicles, S. hystrix var.fluitans was notable for branches that are "provided with short spinules."Winge [20], based on hundreds of preserved samples collected from numerous cruises across the North Atlantic between 1911-1922, identified "8 different drifting Sargassumspecies."Parr [21], based on several cruises in the North Atlantic from 1933-1935, advanced Winge's [20] Sargassum taxonomy and provided the first dichotomous key for drifting Sargassum species, as well as detailing their unique distributional patterns.Notably, the second couplet in Parr's [21] key addresses the nature of Sargassum branches as 'thorny' for all S. fluitans morphological types (III and X) versus branches as 'smooth' for all S. natans morphotypes (I, II, VIII, and IX). S. natans morphotypes are further distinguished by the presence (S. natans I) or absence (S. natans II, VIII, and IX) of an apical spine on the vesicles, and then the size and shape of blades.
Parr [21] is recognized as the definitive source on holopelagic Sargassum taxonomy.Indeed, early reviews of Sargasso Sea science always referenced Parr's [21] taxonomic and distribution studies [89][90][91].Investigations examining the changing distribution and abundance of holopelagic Sargassum [22,92,93] compared their findings to Parr [21].Investigations of the epibionts [94][95][96][97] and motile fauna [98][99][100] that live on different types of holopelagic Sargassum looked to Parr [21] for proper species identification.Studies examining the growth and production of different holopelagic Sargassum species [101,102] also reference Parr [21].Even though these studies were familiar with the true diversity of holopelagic Sargassum morphotypes, in practice, S. natans I and S. fluitans III were consistently the most abundant, and the need to consider rare morphotypes (S. natans II, VIII, and IX, and S. fluitans X) waned over time.Eventually, Parr's [21] numeric system was dropped and S. natans I and S. fluitans III were simply referred to as S. natans and S. fluitans.
The contemporary definitive source for benthic algae taxonomy in the western North Atlantic and Caribbean has been Taylor [49], which naturally includes a thorough account of benthic Sargassum species.However, Taylor's [49] approach to the taxonomy of holopelagic Sargassum was overly simplified and only recognized two species: S. natans and S. fluitans, making no mention of the other morphotypes.Critically, Taylor [49] did not distinguish the two Sargassum species in his dichotomous key based on the presence or absence of thorns on the branches, the well-established historical precedent, but instead used the blade and vesicle shape.In fact, Taylor [49] describes S. fluitans as having a ". ..stem smooth or very sparingly spinulose," which contradicts the historical descriptions of S. fluitans III [18][19][20][21], and the accompanying illustration of S. fluitans (see Plate 39, Figure 2) shows smooth branches and spherical vesicles that more closely resembles Winge's [20] and Parr's [21] descriptions and illustrations of S. natans VIII.A more recent authority on benthic algae in the Caribbean, Wynne's [103] checklist, also adopted Taylor's [49] simplified view of holopelagic Sargassum diversity and did not reference Winge [20] or Parr [21].Schneider and Searles [104] and Littler and Littler [63] also described two species, making no mention of additional holopelagic Sargassum morphotypes, and simplified their dichotomous keys by removing any mention of thorns along the branches.Notably, Littler and Littler [63] provide an accurate photograph of S. fluitans III with thorns evident on the branches.
When Wynne [103] and subsequent revisions are consulted as the authoritative catalog of currently accepted nomenclature for species in the region and Taylor [49], Schneider and Searles [104], or Littler and Littler [63] are the only source material used for holopelagic Sargassum identification, specimens of S. natans VIII, with smooth branches and large, spherical vesicles lacking an apical spine-like appendage, can be mistakenly identified as S. fluitans III. de Szechy et al. [105] collected drift Sargassum off the coast of Brazil during the first GASB bloom and misidentified S. fluitans as S. natans (see Table 1, Figures 4 and 5).The extensive morphological and molecular assessment of benthic Sargassum in the Caribbean by Camacho et al. [54] appears to have used a specimen of S. natans VIII for their analysis of S. fluitans (see Figures 16 and 17 in [54]).Other studies at the time provided photographs or illustrations of S. natans VIII specimens but identified them as S. fluitans [106,107].This is not surprising, considering sample collection occurred during the 2014-15 GASB event, when the composition of the bloom was dominated by S. natans VIII [33].Sissini et al. [55] in their analysis of Sargassum washing up on Brazilian beaches in 2014-15 used Taylor [49], de Szechy et al. [105], and Camacho et al. [54] to make their morphological determinations (S. natans and S. fluitans), thus calling into question their morphological identification of Sargassum specimens and the validity of their molecular results.Similarly, Gonzalez-Nieto et al. [56], in their molecular assessment of Sargassum in the Gulf of Mexico and Caribbean region, included a misidentification of S. natans VIII, labeled as S. fluitans (see Figure 4, Panel D; open habitus, spherical vesicles, and low vesicle-to-blade ratio indicate S. natans VIII).Given the likelihood that their holopelagic Sargassum samples have been improperly categorized, the merits of their molecular analyses cannot be critically assessed.
These errors in identification have persisted without much cause for concern due to the rarity of several of Parr's [21] Sargassum morphotypes and, thus, had little meaningful consequence for misidentification.However, the landscape has changed since 2011, with the development of the GASB and re-emergence of S. natans VIII [33].The improper identification of holopelagic Sargassum samples now has significant consequences and limits our ability to understand the driving mechanisms of the GASB, as recent studies have observed annual and seasonal fluctuations in the composition of holopelagic Sargassum morphotypes [27,33,46].For example, the proposal that ten species with their varieties should be synonymized under S. cymosum [56], a recognized benthic species, was accepted by Wynne [108], resulting in the removal of S. fluitans from the checklist.
We strongly reject the conclusions of Gonzalez-Nieto et al. [56] and Wynne [108] that holopelagic Sargassum are in a polytomy with benthic Sargassum species.First, all holopelagic Sargassum morphotypes are considered sterile, as no receptacles have ever been observed; thus, grouping them with sexually reproductive benthic species with well-documented receptacles is nonsensical.Second, the three most common holopelagic Sargassum morphotypes (S. fluitans III, and S. natans I and VIII) co-occur across broad geographic and environmental ranges and, yet, their morphological and molecular differences are conserved as demonstrated in the present study and Dibner et al. [58].Given the new research presented in this study, we maintain the continued species delineation between S. fluitans III and S. natans I and now formally recognize S. natans VIII as a separate variety of S. natans.
Recognition of a greater, not less, amount of diversity in holopelagic Sargassum is further supported by the research across ecological, biochemical, and physiological fields of study.Govindarajan et al. [109] observed population-level differences in the hydroid Aglaophenia latecarinata among S. fluitans III and S. natans VIII samples, despite the significant geographic overlap in their distribution.Martin et al. [110] observed differences in motile fauna abundance, diversity, and species composition across the three morpho-types.Garcia-Sanchez et al. [46] observed shifts in the abundance of holopelagic Sargassum morphotypes reaching beaches in Mexico over several years.Similarly, seasonal shifts in Sargassum morphotype abundance were observed in Barbados [27].McGillicuddy et al. [111] observed differences in the nutrient and arsenic biogeochemistry among S. fluitans III and S. natans I morphotypes.Finally, several studies have documented the different growth and environmental tolerance among holopelagic Sargassum morphotypes to varying nutrient, temperature, and salinity conditions [39][40][41][42][43].
Taxonomy is not an end in itself, but a tool allowing researchers to propose species hypotheses [47,112].Most often, accepted species are distinguished by a set of morphological characters and/or specific characteristics.In addition to the need for name agreement among scientists to help in research communications, environmental managers also need definitive and consistent names to designate the populations they administer, even if the biodiversity is generally more complex than nomenclatural partitioning, notably because speciation processes are gradual.Terrestrial botany, especially studies carried out on Magnoliophytes, offers thousands of examples of taxa that are distinguished even though intermediate forms, in particular hybrids, unite them.In the case of S. natans VIII and S. natans I, we showed that no intermediate forms exist.In addition, hybridization is, a priori, irrelevant insofar as sexual reproduction has never been observed in these holopelagic macroalgae, despite the major sampling effort that has occurred.Nevertheless, the discord between the genetic differentiation (i.e., very low) and the morphological differences (i.e., high) for S. natans I and S. natans VIII makes it challenging to propose that the two morphotypes correspond to different species.
Current morphotype naming, associated with the lack of a formal description, leads to uncertainties in the description of the holopelagic Sargassum.We, therefore, propose a taxonomic framework, with the description of a new variety and an identification key.The type specimen of Sargassum natans belongs to the morphotype Sargassum natans I (Figure S1).The Sargassum natans VIII morphotype was well-identified by Winge [20] and Parr [21], but neither with a formal diagnosis nor a creation of different taxon.Although this name was taken up later (e.g., [33,57,58] and in most of the recent papers on Sargassum inundation since 2015), it remains undescribed according to the rules of botanical taxonomy (International Code of Nomenclature of algae, fungi, and plants [48]).It is, therefore, described below, under the name of Sargassum natans var.wingei, in tribute to Professor Ö. Winge who first recognized the originality of this form.

Species description
We propose the following new name for the morphotype Sargassum natans VIII.Sargassum natans var.wingei Thibaut, Blanfuné, Boudouresque, Siuda & Verlaque var.nov.Figures: Figures 3 and 4. Holotype: Specimen HCOM S10-S1-3 (Holotype deposited in HCOM Herbarium) Isotypes: Specimens S10-S1-1 and S10-S1-2, deposited in HCOM Herbarium.Type locality: Station 10, tropical Atlantic Ocean, geographic co-ordinates: 14 • 58.440 ′ N 59 • 07.230 ′ W. Etymology: Named for Professor Öjvind Winge, who was the first scientist to distinguish this taxon (as Sargassum sp.VIII) among the pelagic Sargassum specimens.Representative material: All samples of Sargassum natans var.wingei collected during the two expeditions (Caribbean and transatlantic expeditions).All samples are deposited in HCOM Herbarium (Table S2).Specimens sequenced and Genbank references: New haplotypes produced here are available in GenBank under accession numbers MT422788 to MT422805, and final sequence alignments have been submitted in Zenodo.Other descriptions and illustrations: Winge [20] (1923, Figure 13, as Sargassum sp.VIII); Parr [21] (1939, figs 9-10, 12-17, as S. natans (VIII) Winge); Schell et al. [33] (2015, Figure 1 in ref [33], as S. natans VIII Parr); Amaral-Zettler et al. [57]  Diagnosis: Plant holopelagic, brown to yellow-brown, without a holdfast or a distinct main axis; branches smooth without cryptostomata, 1-1.5 mm diam., ramified several times; blades simple, shortly pedicellate, flat, lanceolate to broadly lanceolate, without cryptostomata (13) 30-55 (-65) mm long and 3-10 mm wide (L/W ratio: 3-10, typically: 7-8), with coarsely irregularly serrated margins, symmetrical base, and midrib; vesicles (air bladders), more, less, or as much as the blades, smooth, without cryptostomata, spherical, rarely oblong, 2-5 (-7) mm diam., usually without foliaceous margin or apical spine-like appendage; pedicels short, 0.5-2.5 mm long, cylindrical, sometimes foliaceous with toothed margins; receptacles never observed.Distribution: Tropical Atlantic Ocean, between 7 • and 26 • N. Habitat: It is one of the three holopelagic taxa of the tropical Atlantic Ocean, with S. fluitans (Børgesen) Børgesen and S. natans (Linnaeus) Gaillon.The decision to place S. natans VIII at the level of variety, and not to elevate the morphotype to the status of a new species, is somewhat conservative.Beyond the question of renaming these taxa, the clear morphological differences between the three holopelagic Sargassum and the weak genetic differences raise significant doubts about the speciation stage of these taxa; their assignment to a species rank is even more complex as they most likely rely on asexual reproduction as their primary method of propagation, so that the biological species concept can hardly be used for delineation.Our genetic results on S. natans suggest that the two varieties could correspond to different fixed clones from a single species, where asexual lineages fixed morphological differences.However, the use of mitochondrial and plastidial DNA together is not sufficient to investigate this hypothesis.Fucales are oogamous [113], and, in oogamous brown algae, chloroplasts and mitochondria are thought to be maternally inherited [114].Therefore, these two cytoplasmic genomes are not independently inherited, and, alone, they do not correspond to independent markers.Further investigation of the species' evolutionary history could be accomplished by using nuclear markers, polymorphic within each of the two varieties.This approach would allow for a more in-depth analysis of the extent of clonality, as well as species delineation and speciation scenarios.For example, it would be interesting to test if the pelagic way of life originated once or several times in the Sargassum genus.Finally, the phylogeny of the genus Sargassum must be profoundly clarified considering that the holotype specimen defining the genus Sargassum was Sargassum bacciferum Turner (C.Agardh) currently named for the type species as S. natans (Agardh, 1820), a clonal holopelagic taxa lacking sexual reproductive apparatus.

Figure 2 .
Figure 2. Two-dimensional nMDS ordination plot using S17 Bray-Curtis similarity based on 54 discriminant morphological characters of 264 holopelagic specimens of Sargassum collected during the two 2017 campaigns.

Figure 2 .
Figure 2. Two-dimensional nMDS ordination plot using S17 Bray-Curtis similarity based on 54 discriminant morphological characters of 264 holopelagic specimens of Sargassum collected during the two 2017 campaigns.

Figure 5 .
Figure5.Phylogenetic tree based on the concatenation of rbcL and cox3 sequences.The topology shown here corresponds to the output of the Maximum Likelihood analysis.Bootstrap values higher than 80%, and posterior probabilities higher than 0.9 are indicated at the left of the corresponding group.Newly generated sequences are in red.For the sequences retrieved from GenBank, the accession number is indicated together with the species name (corrected with current taxonomic status).

Phycology 2024, 4 ,
FOR PEER REVIEW 10 (Figure 6).Two S. natans VIII and one S. fluitans III matched the dominant S. natans I haplotype, which differed from the dominant S. fluitans III haplotype by two nucleotide substitutions and from the dominant S. natans VIII by one nucleotide substitution.

Figure 6 .
Figure 6.Sequence networks for mt spacer IGS, rbcL, 16S rRNA gene, concatenation of cox2 and cox3, and nad6.Sequences of S. polyceratium, S. vachellianum, S. spinuligerum, and Sargassum sp., as well as dot-filled wedges correspond to sequences retrieved from GenBank.Number of sequences per haplotype is noted.

Figure 6 .
Figure 6.Sequence networks for mt spacer IGS, rbcL, 16S rRNA gene, concatenation of cox2 and cox3, and nad6.Sequences of S. polyceratium, S. vachellianum, S. spinuligerum, and Sargassum sp., as well as dot-filled wedges correspond to sequences retrieved from GenBank.Number of sequences per haplotype is noted.

Table 1 .
. and nad6) and two plastid markers (psbA and rbcL) were amplified.Primer names, sequences, and PCR programs for all loci are presented in Table1.Primer names and sequences, annealing temperature and time, elongation time, expected size of the PCR product, and references for all loci studied in the present study.

Table 2 .
Permutation multivariate analysis of variance (PERMANOVA) and pair-wise test, based on the morphological characters of holopelagic specimens of Sargassum collected during the two 2017 campaigns (after a square root transformation of the data, and using S17 Bray-Curtis similarity).

Table 2 .
Permutation multivariate analysis of variance (PERMANOVA) and pair-wise test, based on the morphological characters of holopelagic specimens of Sargassum collected during the two 2017 campaigns (after a square root transformation of the data, and using S17 Bray-Curtis similarity).