Comparative mitochondrial and chloroplast genomics of a genetically distinct form of Sargassum contributing to recent “Golden Tides” in the Western Atlantic

Abstract Over the past 5 years, massive accumulations of holopelagic species of the brown macroalga Sargassum in coastal areas of the Caribbean have created “golden tides” that threaten local biodiversity and trigger economic losses associated with beach deterioration and impact on fisheries and tourism. In 2015, the first report identifying the cause of these extreme events implicated a rare form of the holopelagic species Sargassum natans (form VIII). However, since the first mention of S. natans VIII in the 1930s, based solely on morphological characters, no molecular data have confirmed this identification. We generated full‐length mitogenomes and partial chloroplast genomes of all representative holopelagic Sargassum species, S. fluitans III and S. natans I alongside the putatively rare S. natans VIII, to demonstrate small but consistent differences between S. natans I and VIII (7 bp differences out of the 34,727). Our comparative analyses also revealed that both S. natans I and S. natans VIII share a very close phylogenetic relationship with S. fluitans III (94‐ and 96‐bp differences of 34,727). We designed novel primers that amplified regions of the cox2 and cox3 marker genes with consistent polymorphic sites that enabled differentiation between the two S. natans forms (I and VIII) from each other and both from S. fluitans III in over 150 Sargassum samples including those from the 2014 golden tide event. Despite remarkable gene synteny and sequence conservation, the three Sargassum forms differ in morphology, ecology, and distribution patterns, warranting more extensive interrogation of holopelagic Sargassum genomes as a whole.

most abundant in the North Atlantic Ocean subtropical gyre, also referred to as the Sargasso Sea (Figure 1), and also occur in the Gulf of Mexico and Caribbean Sea. Holopelagic Sargassum has been called the "golden floating rainforest of the Atlantic Ocean" (Laffoley et al., 2011) and functions as an ecosystem engineer that creates a unique floating pelagic biome in substrate-poor, low-nutrient open-ocean waters. The floating Sargassum supports over 100 species each of invertebrates and fishes, including ten endemic taxa. Sargassum also serves as a nursery habitat for a host of important commercial and threatened species, including large pelagic fish such as tuna and bill fish, and four species of endangered sea turtles (Coston-Clements, Settle, Hoss, & Cross, 1991).
Reports of golden tides like those in the Caribbean have also been reported in western Africa and Brazil (De Széchy, Guedes, Baeta-Neves, & Oliveira, 2012;Smetacek & Zingone, 2013) impacting tourism, food security, and the limited budgets of coastal towns trying to remove the rotting biomass from their beaches. Whether these golden tides represent changes in distribution of existing biomass or result from unusual accumulations due to higher growth rates ("blooms") has not been established, but the most popular hypothesis is that nutrients supplied by the Amazon and Congo River basins, and also equatorial and coastal upwelling regions along west Africa are allowing fast-growing Sargassum to reach very high concentrations in an area known as the North Equatorial Recirculation Region (NERR, Figure 1), with subsequent flushing toward the Caribbean (Johnson, Ko, Franks, Moreno, & Sanchez-Rubio, 2013). The impacts of potential "bloom" conditions on the structure of the Sargassum populations and the functional diversity of the community of attached and mobile fauna dependent on the Sargassum biome are unknown. We also do not know the impacts of the Sargassum on coastal ecosystems and the potential for associated fauna to become invasive once the rafts of algae wash ashore.
Only two species of holopelagic Sargassum are recognized in contemporary taxonomic guides: S. natans and S. fluitans with a range historically limited to the Sargasso Sea, Gulf of Mexico, and Caribbean (Figure 2a,b). Extensive collections and early taxonomic work on holopelagic Sargassum (Parr, 1939;Winge, 1923) revealed distinct morphological forms of S. natans (I, II, VIII, IX) and S. fluitans (III, X).
Two forms, S. natans I and S. fluitans III, proved most abundant in extensive field surveys in the 1930s (Parr, 1939), and despite annual sampling, none of the rarer forms were documented again until the "re-emergence" of S. natans VIII in 2014 (Schell, Goodwin, & Siuda, 2015;Figure 2c, d). These authors noted that during their November 2014 cruise, Sargassum concentrations were an order of magnitude higher than previously recorded in their 20-year data set. Although morphological features identified the accumulating Sargassum as the formerly rare form S. natans VIII, no genetic analyses have confirmed this observation or Parr's original morphology-based descriptions.
F I G U R E 1 Map of sampling locations where we collected Sargassum (yellow dots with black centers), including open-ocean Atlantic and Caribbean, as well as specimens collected from the shore (Caribbean and Cape Cod, USA). Stations where Sargassum was absent from net tows are shown as blue dots, and locations of mitogenome samples as red asterisks. The heatmap that overlies from approximately 38-63°W and 0-22°N depicts estimates of Sargassum accumulations from satellite data integrated over the 12-day period coincident with the timing of the sample collection in that region. The scale of the heatmap at the lower right shows percent of the ocean surface covered by Sargassum from 0% to >0.1% (satellite data courtesy of University of South Florida Oceanography Lab, http://optics.marine.usf.edu). The yellow line graph at the bottom of the figure shows Sargassum quantity from the cruise net tows, showing the peak around 50°W and good correspondence with estimated densities from the satellite data In order to understand changes in Sargassum population structure and potentially mitigate the impact of the Sargassum golden tides, marine scientists and managers have to be able to identify the algal species involved and where it is coming from. However, recent efforts to characterize members of the Sargassum subgenus Sargassum have encountered challenges due to low genetic variability among some clades (Cheang et al., 2010;Cho, Lee, Ko, Mattio, & Boo, 2012;Dixon et al., 2014;Mattio, Bolton, & Anderson, 2015;Mattio et al., 2013) and a lack of marker genes capable of distinguishing between closely related taxa in general (Camacho, Mattio, Draisma, Fredericq, & Diaz-Pulido, 2014;Mattio & Payri, 2010). Thus, new molecular ecology resources are urgently needed to aid this endeavor. The use of mitogenomes to differentiate between different species of Sargassum has recently provided a bit of traction in terms of offering new sets of genetic markers for population studies of Sargassum species from the Pacific (Bi & Zhou, 2016;Liu, Pang, & Chen, 2016;Liu, Pang, Li, & Li, 2015;Liu, Pang, & Luo, 2014). In addition to mitochondrial genomes, the first Sargassum chloroplast genome from the benthic species S. horneri recently became available providing another source of potential marker genes .
Here, we explored the potential for mitogenomes and chloroplast coding regions to distinguish between Atlantic Sargassum holopelagic species by generating and comparing complete mitochondrial genomes and partial chloroplast genomes from all of the known holopelagic Sargassum species using next-generation sequencing approaches. From our reference mitogenomes, we developed two marker gene primer sets that are capable of differentiating between all three holopelagic Sargassum forms encountered in the Atlantic: S. natans I, S. natans VIII, and S. fluitans III. Using all Sargassum mitogenomes available to date including the six new ones we generated, we also provide a phylogenetic placement for the three forms of Sargassum most commonly encountered in the tropical and subtropical seas of the western Atlantic.

| Sample collection and DNA extraction
We collected samples for our six reference mitochondrial and partial chloroplast genomes and 155 marker gene validations from 71 independent clumps of Sargassum from across the NW Atlantic Ocean and in the Caribbean Sea (see Figure 1 and Table S1 for details). Samples of holopelagic Sargassum were collected in towed neuston nets, as well as in dip nets during four separate cruises in 2012, 2014, 2015, and 2016 in the North Atlantic Ocean. Additional samples of clumps that had stranded on shore were collected by hand in 2014 and 2016 (see Table S1).
Clumps of macroalgae were assigned species identifications using morphological characters defined by Parr (1939) including frond characteristics, the presence or absence of thorns on the axes, and spinelike appendages on the air bladders (vesicles), as well as by blade (leaf) size. According to Parr's definitions, S. fluitans III can be distinguished from both S. natans I and S. natans VIII based on the presence of thorns on its axes. S. fluitans also does not have spine-like appendages on its vesicles that are usually present on vesicles of S. natans I and occasionally found on vesicles of S. natans VIII. The two species of S. natans can also be distinguished based on their blade width. S. natans I has blades that are much narrower than S. natans VIII.

| Library construction and sequencing
We constructed genomic libraries for representative samples of each Sargassum form (number of individuals sequenced in parentheses): S. natans VIII (three specimens); S. natans I (one specimen); and S. fluitans III (two specimens). Table S1 and Figure Table S1 contains specific details on library preparation and sequencing performed for each sample.

| Bioinformatics and phylogenetics
Resulting Illumina raw reads were merged and quality-checked using a series of Python scripts: "iu-merge-pairs" with the "enforce-Q30-check" option, to remove low-quality reads and merge pairs, followed by "iu-filter-merged-reads" to retain only merged pairs with no mismatches in the overlapping region (Minoche, Dohm, & Himmelbauer, 2011). We used CLC genomic workbench to assemble and map reads back to single mitochondrial genomic contigs (http://www.clcbio.com) and Geneious v. 8.0.5 (Biomatters, Ltd, Aukland, New Zealand) for gene annotation and visualization of the mitogenomes by importing annotation tracks from existing sequenced mitogenomes of other Sargassum species in GenBank.
We used a similar strategy for mapping chloroplast gene reads back to the S. horneri chloroplast genome in order to access variation of polymorphic sites in the chloroplast coding regions of our representative Sargassum species.
For our phylogenetic analyses of mitogenomes, we removed any positions of uncertain homology in the mitochondrial alignments and inferred phylogenetic trees using 34,290 homologous nucleotide positions in RAxML version 7.2.8 (Stamatakis, 2006) and a general time reversible substitution model with gamma and invariant sites implemented through Geneious. One thousand bootstrap replicates determined the confidence of the branch support in the resulting phylogeny.

| Confirmation of polymorphic sites via PCR
To assess whether the variant sites within mitochondrial genomes that we discovered were fixed, we designed primers that targeted the relevant cytochrome oxidase subunit 2 (cox2)

| RESULTS
Our phylogenetic analyses based on mitochondrial sequence data revealed three genetically distinct forms of holopelagic Sargassum.
The holopelagic Sargassum responsible for the golden tides in the Caribbean fell into the S. natans species complex with high bootstrap support (100%) and was not S. fluitans as often reported. Figure 3 depicts a maximum-likelihood phylogenetic tree based on complete that is, the differences between forms and species appear to be fixed.
Both nuclear 18S and 5.8S rRNA genes were identical across the forms, while the intergenic spacer between rbcL and rbcS and three sites in the rbcL gene distinguished S. fluitans from S. natans, but could not differentiate between the two S. natans forms. ITS-2 genes for S. natans I and S. fluitans III were identical, while S. natans VIII showed a single bp difference. Across the gold-standard barcoding gene, cox1, S. fluitans differed from the S. natans forms at only two sites, and the two S. natans forms were identical.
In addition to comparisons between mitogenomes and nuclear marker genes used in other studies targeting Sargassum, we also compared the chloroplast coding regions that we were able to recover from mapping reads from our six representative samples against the full-length chloroplast genome of S. horneri as a reference . Near-complete chloroplast genomes were recovered from one of the S. natans VIII samples (C6) and the S. natans I sample (C2) and included all of the coding regions reported in S. horneri.  (Table S2).
For chloroplast genomes that were most complete S. natans VIII (C6) and the S. natans I (C2), gene synteny and content were identical to S. horneri with 173 genes (Table S2), including six coding for rRNA genes (two copies each) and 28 coding for tRNA genes (with multiple copies of some, designated with a "1" or a "2" in Table S2). As described extensively in , there is a high degree of conservation in chloroplast DNA among the brown algae and our results were consistent with this finding.  with available cox2 genes for comparison, so are unlikely to be useful for population studies beyond the holopelagic forms.

F I G U R E 3 A maximum-likelihood-inferred phylogeny of mitogenomes from all available Sargassum species alongside the holopelagic species sequenced in this study (in bold). Scale bar represents evolutionary distance, and numbers at the nodes represent bootstrap confidence values. GenBank numbers follow the names of published mitogenomes
Due to limitations in using single marker genes for comparative brown algal phylogenetic studies, investigators are increasingly turning to comparative genomics of organelles as tools to differentiate between closely related taxa. Our study is the first to demonstrate genetic variability between the different morphotypes of Sargassum originally described in the first part of the 20th century. The fact that we were able to confirm that the species involved in the "golden tides" are a genetically distinct and formerly rare form of holopelagic Sargassum is important because it suggests that the absolute and relative abun- The ability to accurately identify the type of Sargassum is important as we start to address these questions, but recent publications disagree on morphological characters that differentiate between  Figure 4). Sargassum fluitans III has a range that overlaps S. natans VIII and the two can be readily confused; even with considerable expertise of the scientists identifying our samples using morphological characters alone, our molecular results showed that some specimens had been misclassified, highlighting the importance of genetic analyses to confirm the identity of these closely related species. The primers and methods described in this study address this need by providing accurate and nonsubjective identification of the holopelagic species.
Despite consistent differences at the mitogenome and marker gene levels, our molecular results demonstrate that overall, the holopelagic Sargassum species exhibit remarkable genetic similarity.
Likewise, differences in chloroplast coding regions are modest at best.  (Schell et al., 2015). This suggests that and suggest that they are relatively new species. Current theories argue that human-mediated environmental changes such as increases in temperature and nutrient loads are leading to blooms of previously "rare" genotypes already existing in a population. The increase in abundance along with asexual reproduction and gas vesicles in the holopelagic forms enhance their ability to be transported long distances and may increase their invasion potential (Paula & Eston, 1987). With the new repertoire of genetic information now available through our efforts, we are poised to expand our understanding of population-level variation in holopelagic Sargassum and its associated taxa.

CONFLICT OF INTERESTS
We have no competing interests.