Epibenthic Harmful Marine Dinoflagellates from Fuerteventura (Canary Islands), with Special Reference to the Ciguatoxin-Producing Gambierdiscus

The relationship between the ciguatoxin-producer benthic dinoflagellate Gambierdiscus and other epibenthic dinoflagellates in the Canary Islands was examined in macrophyte samples obtained from two locations of Fuerteventura Island in September 2016. The genera examined included Coolia, Gambierdiscus, Ostreopsis, Prorocentrum, Scrippsiella, Sinophysis, and Vulcanodinium. Distinct assemblages among these benthic dinoflagellates and preferential macroalgal communities were observed. Vulcanodinium showed the highest cell concentrations (81.6 × 103 cells gr−1 wet weight macrophyte), followed by Ostreopsis (25.2 × 103 cells gr−1 wet weight macrophyte). These two species were most represented at a station (Playitas) characterized by turfy Rhodophytes. In turn, Gambierdiscus (3.8 × 103 cells gr−1 wet weight macrophyte) and Sinophysis (2.6 × 103 cells gr−1 wet weight macrophyte) were mostly found in a second station (Cotillo) dominated by Rhodophytes and Phaeophytes. The influence of macrophyte’s thallus architecture on the abundance of dinoflagellates was observed. Filamentous morphotypes followed by macroalgae arranged in entangled clumps presented more richness of epiphytic dinoflagellates. Morphometric analysis was applied to Gambierdiscus specimens. By large, G. excentricus was the most abundant species and G. australes occupied the second place. The toxigenic potential of some of the genera/species distributed in the benthic habitats of the Canary coasts, together with the already known presence of ciguatera in the region, merits future studies on possible transmission of their toxins in the marine food chain.


Introduction
Ciguatera fish poisoning (CFP), the most important food-borne illness caused by fish consumption in the world, is produced by ciguatoxins (CTX) which are suggested to be transferred from epiphytic dinoflagellates of Gambierdiscus and Fukuyoa genera into the food web [1,2]. The incidence of CFP in tropical and subtropical areas has been extensively reported since antiquity [3] but a spreading into more temperate regions of both CFP cases and Gambierdiscus and Fukuyoa populations have been reported in the last decade. In Europe, where CTXs are considered an emerging threat, the incidence of ciguatera episodes was first recorded in 2004 in the Canary Islands and Madeira [4][5][6][7]. There is awareness that global warming can cause spreading of CTX-producing dinoflagellates into higher latitudes not currently affected by CFP [8,9]. This concern has prompted Intergovernmental Panel on Climate Change to alert about the effect of global warming on the increase of CFP occurrence [10]. In fact, recent studies report Gambierdiscus and Fukuyoa species in more temperate waters of Japan, Mediterranean Sea, Canary Islands, and along the eastern coasts of North and South America [11][12][13][14][15][16].
Some authors have described that ciguatera incidence and the prevalence of Gambierdiscus and Fukuyoa cells are not always positively well correlated [17][18][19][20]. Different toxicity among Gambierdiscus species and changes in their interannual relative abundances were suggested to cause those differences between CFP outbreaks and Gambierdiscus detection [18]. Subsequent studies demonstrated high variability in the toxic potential among species. Higher toxicity has been reported, for example for G. polynesiensis in the Pacific [21][22][23], and G. excentricus in Caribbean Sea and the Canary Islands, in comparison with other species in the same regions [23,24]. This emphasizes the need for implementing adequate methodologies for the unequivocal identification of species in these genera, as well as for their quantification.
The difficult morphological differentiation among species of Gambierdiscus and, consequently, the problem for species-specific cell counts by traditional microscopy-based methodologies has been abundantly mentioned in the literature [25,26]. Therefore, the unequivocal identification of Gambierdiscus cells relies in most occasions on molecular techniques, mainly on rDNA sequences of cultures. Furthermore, semi-quantitative techniques (qPCR) have been described for most species and ribotypes of Gambierdiscus [27][28][29]. However, such methods cannot always be implemented, whereas light microscopy, despite its limitations, can still provide useful information.
In the present study, Gambierdiscus cells were morphologically characterized to determine to what extent their differences in morphology and size are useful for their specific identification. The methodology used was based on the parameters described by Bravo et al. [26] for the five species found in the Canary Islands so far, excepting the very recently reported G. belizeanus by Tudó et al. [30]. These morphological traits include cell depth measurement and the shapes of the second apical (2 ) and second antapical (2"") plates, as well as the position of the Po plate.
For some time, while the responsible agent of ciguatera was unknown, other benthic dinoflagellates apart from Gambierdiscus were associated with this syndrome, like Ostreopsis and Prorocentrum. As it was later discovered, this was due to the potential production of palytoxin and palytoxin-like compounds in Ostreopsis, and okadaic acid, dinophysistoxins-1, 2, and 4, and prorocentrolide in several benthic species of Prorocentrum like P. lima (see references in [31]). P. hoffmannianum has been isolated from benthic communities in the Canary Islands and confirmed to produce okadaic acid and three analogs [32]. Furthermore, the single species of Vulcanodinium described so far, V. rugosum, has been described to synthesize potent bioactive compounds like pinnatoxins and portimine, though it was never associated with human poisonings [33,34]. In consequence, although Gambierdiscus and Fukuyoa are the vector species for CTX in fish and CFP outbreaks, other species of those dinoflagellates cannot be discarded to cause some kind of harmful episode.
Studies in ciguatera endemic areas have described the effects of structural complexity of coral reefs on benthic harmful dinoflagellate communities. Thus, the different environmental driving factors that govern each community influence the benthic dinoflagellate assemblages [35,36]. Moreover, macrophyte host preferences as well as epiphytic dinoflagellate associations have been described in some regions [37,38]. The results, however, are sometimes contradictory due to the difficulty of understanding such complex benthic habitats [38]. The spatial distribution patterns of macrophytes depending on factors such as temperature, lighting, and wave exposure have been extensively studied on the coasts of the Canary Islands [39][40][41] and the composition and spatial distribution of marine macrophytes on the littoral of Fuerteventura exhibit a higher proportion of warm water species than on the rest of Islands of this archipelago [42][43][44]; it is not known, however, whether epiphytic harmful dinoflagellates are preferentially distributed in some of them. The specific genera examined in the present study comprised Gambierdiscus, Prorocentrum, Coolia, Sinophysis, Ostreopsis, Vulcanodinium, and Scrippsiella. All of them were surveyed from macrophytes from two locations in Fuerteventura Island with different macrophyte communities. The objectives of this study were: (1) to know if there is any preferential associations of benthic harmful dinoflagellates; (2) advance on the knowledge of the relationships of benthic dinoflagellate assemblages with different macrophyte communities; and (3) to know the most abundant species of Gambierdiscus in the benthic macrophyte communities examined, for which a morphological study was carried out.

Study Sites
Cotillo is located in the NW of Fuerteventura Island (Canary archipelago, Figure 1A,B). In this station (28 • 41 18.34" N/14 • 0 48.14" W), macrophytes were sampled in a "charco"-a type of coastal pool of medium size quite abundant in the Canary Islands-of approximately 4 × 10 3 m 2 located to the left of Marfolín beach, on an extensive rocky platform that extends just north of the town of El Cotillo. This pool-maximum depth of 2 m at low tide and 3 m at high tide-has a mostly rocky bottom alternating with small sand spaces. This type of pools constitutes a particular environment with great environmental variability in a limited space that displays a high biodiversity. In Cotillo the rocky platform extends in depth up to more than one kilometer offshore reaching bathymetric levels greater than 20 m. Samples were taken from the "charco" by snorkeling during low tide up to 1.5 m deep and by scuba diving until 6 m deep.
Playitas is located on the eastern side of Fuerteventura Island ( Figure 1A,B). The station (28 • 13 39.80" N/13 • 59 1.69" W) was situated just to the left of the port in the town Las Playitas. The platform extends with gentle slope in the infralitoral where small tidal ponds were sampled on foot. Then, samples were taken by snorkeling until 2-3 m and by scuba diving until 6 m deep.  (3) to know the most abundant species of Gambierdiscus in the benthic macrophyte communities examined, for which a morphological study was carried out.

Study Sites
Cotillo is located in the NW of Fuerteventura Island (Canary archipelago, Figure 1A,B). In this station (28°41′18.34″ N/14°0′48.14″ W), macrophytes were sampled in a "charco"-a type of coastal pool of medium size quite abundant in the Canary Islands-of approximately 4 × 10 3 m 2 located to the left of Marfolín beach, on an extensive rocky platform that extends just north of the town of El Cotillo. This pool-maximum depth of 2 m at low tide and 3 m at high tide-has a mostly rocky bottom alternating with small sand spaces. This type of pools constitutes a particular environment with great environmental variability in a limited space that displays a high biodiversity. In Cotillo the rocky platform extends in depth up to more than one kilometer offshore reaching bathymetric levels greater than 20 m. Samples were taken from the "charco" by snorkeling during low tide up to 1.5 m deep and by scuba diving until 6 m deep.
Playitas is located on the eastern side of Fuerteventura Island ( Figure 1A,B). The station (28°13′39.80″ N/13°59′1.69″ W) was situated just to the left of the port in the town Las Playitas. The platform extends with gentle slope in the infralitoral where small tidal ponds were sampled on foot. Then, samples were taken by snorkeling until 2-3 m and by scuba diving until 6 m deep.

Field Sampling and Cell Enumeration
Sixty-seven samples of macrophytes were collected in the two stations in September 2016 (16th-17th in Playitas and 18th-19th in Cotillo) (Table 1, Figure 1). The macrophyte samples were carefully collected with surrounding water in plastic bags, placed in a plastic bottle and shaken to detach epiphytes. Afterwards, the gross materials were removed through a 300 µm opening nylon mesh and the remaining seawater was filtered again on a 20 mm nylon mesh to concentrate the cells.

Field Sampling and Cell Enumeration
Sixty-seven samples of macrophytes were collected in the two stations in September 2016 (16th-17th in Playitas and 18th-19th in Cotillo) (Table 1, Figure 1). The macrophyte samples were carefully collected with surrounding water in plastic bags, placed in a plastic bottle and shaken to detach epiphytes. Afterwards, the gross materials were removed through a 300 µm opening nylon mesh and the remaining seawater was filtered again on a 20 mm nylon mesh to concentrate the cells. Aliquots from these samples were fixed in situ with formaldehyde for identification and enumeration in the laboratory.
Formaldehyde-fixed epiphyte samples were stained with Fluorescent Brightner 28 (Sigma, St Louis, MO, USA) [45] for dinoflagellates identification and counted under UV light using an Axiovert 125 epifluorescence inverted microscope (Carl Zeiss AG, Germany) at 400× magnification. Quantitative data were obtained for the following genera of benthic dinoflagellates: Gambierdiscus, Prorocentrum, Coolia, Sinophysis, Ostreopsis, Vulcanodinium, and Scrippsiella. Cell abundance was expressed as cells per gram wet weight of host macrophyte (abbreviated as cells g −1 in the results section). For this purpose, fresh macrophytes were weighted after being manually drained just after collection.

Macrophyte Sampling
The macrophyte community of the two sampling stations presented remarkable differences in species composition. During sampling, the most representative species (or groups of species) of each station were collected. Although macrophyte composition was not the target of the study, different macrophyte communities in Cotillo and Playitas were clearly evidenced (Table 1).
In the intertidal zone of Playitas, red algae (Rhodophyceae) belonging to the orders Ceramiales, Corallinales, and Gigartinales which formed a thick turf (named as turf in Table 1) was characteristic. There, the most representative species were: Hypnea spinella, Jania adhaerens, Centroceras gasparrinii, Amphiroa fragilissima, and Palisada perforata. Added to those and frequently as an epiphyte, the filamentous cyanobacteria Blennothrix lyngbyacea was also found. Among the brown algae (Phaeophyceae), much less abundant in the intertidal zone, erect foliose species of Dictyotales as Padina pavonica and Stypopodium zonale were collected (Table 1). Other species of Ceramiales, such as the erect, filamentous, and profusely branched Lophocladia trichoclados and Cottoniella fusiformis, as well as ribbon-like Dictyotales, such as Canistrocarpus cervicornis, Dictyota dichotoma, and D. humifusa, were sampled at depths of more than two meters.
Macrophyte community was mainly formed by a very diverse assemblage of erect brown and red algae in the "charco" sampled in Cotillo station. Dictyotales as Dictyota spp, Stypopodium zonale, Padina pavonica, and the also foliose Lobophora schneideri were dominant ( Table 1). Species of Sphaceraliales forming erect arborescent tufts as Halopteris scoparia and H. filicina were also collected. Among the Rhodophyceae, species from Bonnemaisoniales (as the arborescent with duster-like appearance Asparagopsis taxiformis), Nemaliales (as the cylindrical dichotomously branched Galaxaura rugosa) and Ceramiales (as Lophocladia trichoclados) were the most common macrophytes in the "charco". At 2 m depth, Asparagopsis taxiformis was the dominant species and species of Dictyotales as Lobophora schneideri prevailed deeper.
Since epiphyte abundances are clearly related to differences in structure and wet weight to surface area ratios of macrophytes and an estimate of the surface/weight ratio has not yet been established, the macrophytes were categorized into four types based on external morphology classification, modified from definitions in Parsons and Preskitt [46]: (1) Type 1: Foliose (laminar thallus); (2) Type 2: Ribbon-like (several times forked ribbon-shaped thallus); (3) Type 3: Entangled clumps (thallus with cylindrical axes, 0.2-2.0 mm diameter, branched and entangled); (4) Type 4: Filamentous (thallus with thin cylindrical axes, ≤0.2 mm diameter, profusely branched and tree-like).

Epiphytic Dinoflagellate Assemblages
A principal component analysis (PCA) was performed to analyze the data describing the composition of epiphytic dinoflagellates. It was conducted using logarithmically transformed cell concentrations and the statistical software package SPSS. The Kaiser-Meyer-Olkin measure of sampling adequacy was 0.66 and Bartlett's test of sphericity, which tests for the presence of correlations among variables, was significant at p < 0.001. In addition, non-parametric rank-based test (Kruskal-Wallis) was performed using the statistical software package SPSS version 14.0 (SPSS Inc., Chicago IL) to compare the distribution of the abundance values of dinoflagellate species from both stations.

Morphometric Analysis and Abundances of Gambierdiscus
Morphological analyses were performed on individual cells of Gambierdiscus isolated from epiphytic samples. Measurements of the epitheca and hypotheca of the same specimen were made by placing individual cells between two coverslips, which allowed them to be observed and photographed from their apical and antapical views. The morphologies of a total of 30-40 cells from each sample were studied. Cell morphology determinations were based on measurements of two thecal plates: the second apical (2 ) plate, located on the epitheca, and the second antapical (2"") plate on the hypotheca, following the methodology described by Bravo et al. [26]. Three morphometric parameters were used as follows: (1) R1 as an assessment of the rectangular vs. the hatchet shape of the 2 plate; (2) R2 representing the position of Po in the lateral edge of the 2 plate and, therefore, the degree of eccentricity of Po in the cell; and (3) R3 as an indicator of the elongation of the 2"" plate. In addition, cell depth (D), corresponding to the dorso-ventral diameter was also used. These parameters were selected following the most relevant bibliography on Gambierdiscus morphology as mentioned by Bravo et al. [26]. These authors define the parameter values for each species based on a study performed with culture cells. All measurements needed for those morphometric calculations were made on Calcofluor-stained cells using digital imaging software (ZEN lite, ZEISS Microscopy) and an epifluorescence microscope (Leica DMLA, Wetzlar, Germany) equipped with a UV light source and an AxioCam HRc (Carl Zeiss, Jena, Germany) digital camera. Concentrations of the five species of Gambierdiscus were estimated from the percentages of cells identified for each species and the total concentration value of the genus counted in each sample as explained above (section of field sampling and cell enumeration).

Abundances of Epiphytic Dinoflagellates
Cells of Gambierdiscus, Prorocentrum, Coolia, Sinophysis, Ostreopsis, and Vulcanodinium genera were identified in the two sampled stations but appeared in different ratios. In Cotillo, Prorocentrum, Coolia, Gambierdiscus, and Vulcanodinium were present at percentages higher than 10% (26%, 25%, 17%, and 15%, respectively) whereas Synophysis and Ostreopsis represented 10% and 8%, respectively ( Figure 1B). On other hand, Vulcanodinium and Ostreopsis prevailed in Playitas (58% and 29%, respectively), while Prorocentrum, Coolia, Gambierdiscus and Sinophysis were less abundant (7%, 5%, 1%, and 0.2%, respectively; Figure 1B). The genus Scrippsiella was only detected in three samples from Cotillo (reaching up to 486 cells gr −1 ), was not included in the statistical analyses. Total dinoflagellate abundances were higher in Playitas station. The abundance mean values for all genera are plotted in Figure 1C. The differences between stations were highly significant both for genus composition and abundances. Statistical values (mean, standard deviations, maximum and minimum) of abundances of all species and stations are shown in Table 2. Significant differences (p < 0.001) were found between the distribution of the abundances of Gambierdiscus, Sinophysis, Ostreopsis, and Vulcanodinium from both stations; on the contrary, no significant differences were found for Prorocentrum and Coolia (Table 2).

Epiphytic Dinoflagellate Assemblages and Macrophyte Associations
Different dinoflagellate assemblages among the six dinoflagellate genera were revealed through Principal Component Analysis (PCA) of their abundances. Component 1 (PC1) grouped four species: Ostreopsis, Prorocentrum, Coolia, and Vulcanodinium, whereas component 2 (PC2) was more associated to Gambierdiscus and Sinophysis (Figure 2A). On PC2, these last two species were negatively correlated with Ostreopsis. The two components explained 62% of the variance (31.6% for component 1 and 30.6% for component 2). Factor loadings of the genera projected on the PCA plot show a clear separation of Ostreopsis and Vulcanodinium from Gambierdiscus and Sinophysis (Figure 2A), whereas the relationship of Prorocentrum and Coolia was not so evident. The components were differently associated to the two stations. As shown in Figure 2A, while Ostreopsis and Vulcanodinium were more associated to Playitas, Gambierdiscus and Sinophysis were more to Cotillo station. The different macroalgae composition in the two stations and its differential association to the different genera of dinoflagellates is showed in Figure 2B.
Abundances of the different dinoflagellate genera in the main macrophytes of the two stations are plotted in Figure 3. The means of the abundances of Gambierdiscus and Sinophysis presented the highest values in the macrophytes of Cotillo compared to those of Ostreopsis and Vulcanodinium, which are higher in the macrophytes of Playitas.
The study of the abundances of the dinoflagellate genera in the different types of macrophytes showed that those with a filamentous structure clearly presented the highest abundances in all the genera studied ( Figure 4). However, within that type of macrophytes remarkable differences were found depending on the dinoflagellates. Thus, Gambierdiscus and Sinophysis presented the highest abundance values in the filamentous Halopteris and Asparagopsis, whereas Spyridia and Lophocladia showed the highest abundances of Ostreopsis and Vulcanodinium (Figures 3 and 4). Macrophytes with an entangled clump structure as turf species characteristic of Playitas station ranked second for Prorocentrum, Coolia, and Ostreopsis ( Figure 4). On the contrary, the foliose macrophyte Lobophora occupied second position for Gambierdiscus and Sinophysis and the ribbon-like Dyctiota and Canistrocarpus in the case of Vulcanodinium. Concentrations of Prorocentrum and Coolia presented a more homogenous distribution among all macrophyte species (Figure 3).   Table 1). Crosses represent the factor loadings of the different dinoflagellate genera.
Abundances of the different dinoflagellate genera in the main macrophytes of the two stations are plotted in Figure 3. The means of the abundances of Gambierdiscus and Sinophysis presented the highest values in the macrophytes of Cotillo compared to those of Ostreopsis and Vulcanodinium, which are higher in the macrophytes of Playitas.  Table 1). Crosses represent the factor loadings of the different dinoflagellate genera.  Table 1.
The study of the abundances of the dinoflagellate genera in the different types of macrophytes showed that those with a filamentous structure clearly presented the highest abundances in all the genera studied ( Figure 4). However, within that type of macrophytes remarkable differences were found depending on the dinoflagellates. Thus, Gambierdiscus and Sinophysis presented the highest abundance values in the filamentous Halopteris and Asparagopsis, whereas Spyridia and Lophocladia showed the highest abundances of Ostreopsis and Vulcanodinium (Figures 3 and 4). Macrophytes with an entangled clump structure as turf species characteristic of Playitas station ranked second for Prorocentrum, Coolia, and Ostreopsis (Figure 4). On the contrary, the foliose macrophyte Lobophora occupied second position for Gambierdiscus and Sinophysis and the ribbon-like Dyctiota and Canistrocarpus in the case of Vulcanodinium. Concentrations of Prorocentrum and Coolia presented a more homogenous distribution among all macrophyte species (Figure 3).   Table 1.

Morphological Characterization of Gambierdiscus Species
Based on cell sizes (cell depth denoted as D) and the parameters R1, R2, and R3 (related to the plate's morphology, see Material and Methods) 91% of the specimens were classified within the five Gambierdiscus species detected previously in the Canary Islands: G. australes, G. caribaeus, G. carolinianus, G. excentricus, and G. silvae. The values of the parameters (D, R1, R2, and R3) and the corresponding classification are scattered in Figure 5. G. excentricus was separated from all other species by the excentricity of Po (represented by parameter R2) ( Figure 5A) excepting the overlap of some specimens with G. australes. In those cases, R1 and R3 relationship was useful for identification ( Figure 5B). Figure 5A shows as G. excentricus and G. silvae were the most easily discriminated species basing in size and R2. In addition, the scattered plotting of R1 (denoting the shape of 2′ plate) and R3 (shape of 2'''' plate) efficiently separated G. silvae and G. caribaeus from the rest of the species ( Figure 5B). For classification of those species, size was also useful following the description by the same authors previously mentioned. Low overlap percentages were observed between the groups of G. australes and G. excentricus (1.9% of the total cells) regarding excentricity of Po (represented by parameter R2), the most differentiating trait between those species ( Figure 5A). Notwithstanding, the general appearance of the cell as well as the general shape of 2′ and 2'''' plates helped to classify them. G. australes and G. caribaeus were the most similar species. Both coincide in the three following traits: rectangular shape of 2′ plate (R1), cell size (D) and position of Po (R2) ( Figure 5). The shape of 2'''', more elongated in G. australes than in G. caribaeus, was the most useful trait to discriminate them ( Figure 5B). However, the overlapping in that parameter was also remarkable. Due to this, it was not possible to separate 9% of the total cells which were comprised in the G. australes/caribaeus group.  Table 1.

Morphological Characterization of Gambierdiscus Species
Based on cell sizes (cell depth denoted as D) and the parameters R1, R2, and R3 (related to the plate's morphology, see Material and Methods) 91% of the specimens were classified within the five Gambierdiscus species detected previously in the Canary Islands: G. australes, G. caribaeus, G. carolinianus, G. excentricus, and G. silvae. The values of the parameters (D, R1, R2, and R3) and the corresponding classification are scattered in Figure 5. G. excentricus was separated from all other species by the excentricity of Po (represented by parameter R2) ( Figure 5A) excepting the overlap of some specimens with G. australes. In those cases, R1 and R3 relationship was useful for identification ( Figure 5B). Figure 5A shows as G. excentricus and G. silvae were the most easily discriminated species basing in size and R2. In addition, the scattered plotting of R1 (denoting the shape of 2 plate) and R3 (shape of 2"" plate) efficiently separated G. silvae and G. caribaeus from the rest of the species ( Figure 5B). For classification of those species, size was also useful following the description by the same authors previously mentioned. Low overlap percentages were observed between the groups of G. australes and G. excentricus (1.9% of the total cells) regarding excentricity of Po (represented by parameter R2), the most differentiating trait between those species ( Figure 5A). Notwithstanding, the general appearance of the cell as well as the general shape of 2 and 2"" plates helped to classify them. G. australes and G. caribaeus were the most similar species. Both coincide in the three following traits: rectangular shape of 2 plate (R1), cell size (D) and position of Po (R2) ( Figure 5). The shape of 2"", more elongated in G. australes than in G. caribaeus, was the most useful trait to discriminate them ( Figure 5B). However, the overlapping in that parameter was also remarkable. Due to this, it was not possible to separate 9% of the total cells which were comprised in the G. australes/caribaeus group.

Diversity and abundances of Gambierdiscus species
Total abundances of genus Gambierdiscus reached up to 3.8 × 10 3 cells gr −1 in Cotillo station and 8•10 2 cells gr −1 in Playitas. Means and standard deviations are showed in Table 2. The abundance distributions between the two stations showed significant differences (p < 0.001) ( Table 2). Individual morphometric analyses as mentioned in the previous section revealed, at least five species of Gambierdiscus in the two stations: G. australes, G. caribaeus, G. carolinianus, G. excentricus, and G. silvae. Figure 6 shows the percent cell concentrations of Gambierdiscus species. Significant differences were only detected in the distribution of percentages of G. australes and G. excentricus between the two stations (p < 0.01). G. excentricus was the most abundant of the five species, representing as average 56% and 75% in Cotillo and Playitas, respectively, followed by G. australes (mean of 24% and 18% in each station, respectively). The number of specimens which could not be identified unequivocally, denominated as G. australes/caribaeus, was quite abundant in Cotillo but rare in Playitas station (mean of 12% and 3%, respectively). G. caribaeus and G. silvae presented mean abundances of 4% and 5% in Cotillo and 1% and 3% in Playitas, respectively. Finally, G. carolinianus was even less represented in the two stations ( Figure 6).

Diversity and abundances of Gambierdiscus species
Total abundances of genus Gambierdiscus reached up to 3.8 × 10 3 cells gr −1 in Cotillo station and 8 × 10 2 cells gr −1 in Playitas. Means and standard deviations are showed in Table 2. The abundance distributions between the two stations showed significant differences (p < 0.001) ( Table 2). Individual morphometric analyses as mentioned in the previous section revealed, at least five species of Gambierdiscus in the two stations: G. australes, G. caribaeus, G. carolinianus, G. excentricus, and G. silvae. Figure 6 shows the percent cell concentrations of Gambierdiscus species. Significant differences were only detected in the distribution of percentages of G. australes and G. excentricus between the two stations (p < 0.01). G. excentricus was the most abundant of the five species, representing as average 56% and 75% in Cotillo and Playitas, respectively, followed by G. australes (mean of 24% and 18% in each station, respectively). The number of specimens which could not be identified unequivocally, denominated as G. australes/caribaeus, was quite abundant in Cotillo but rare in Playitas station (mean of 12% and 3%, respectively). G. caribaeus and G. silvae presented mean abundances of 4% and 5% in Cotillo and 1% and 3% in Playitas, respectively. Finally, G. carolinianus was even less represented in the two stations ( Figure 6).

Diversity and abundances of Gambierdiscus species
Total abundances of genus Gambierdiscus reached up to 3.8 × 10 3 cells gr −1 in Cotillo station and 8•10 2 cells gr −1 in Playitas. Means and standard deviations are showed in Table 2. The abundance distributions between the two stations showed significant differences (p < 0.001) ( Table 2). Individual morphometric analyses as mentioned in the previous section revealed, at least five species of Gambierdiscus in the two stations: G. australes, G. caribaeus, G. carolinianus, G. excentricus, and G. silvae. Figure 6 shows the percent cell concentrations of Gambierdiscus species. Significant differences were only detected in the distribution of percentages of G. australes and G. excentricus between the two stations (p < 0.01). G. excentricus was the most abundant of the five species, representing as average 56% and 75% in Cotillo and Playitas, respectively, followed by G. australes (mean of 24% and 18% in each station, respectively). The number of specimens which could not be identified unequivocally, denominated as G. australes/caribaeus, was quite abundant in Cotillo but rare in Playitas station (mean of 12% and 3%, respectively). G. caribaeus and G. silvae presented mean abundances of 4% and 5% in Cotillo and 1% and 3% in Playitas, respectively. Finally, G. carolinianus was even less represented in the two stations ( Figure 6).   (Figure 7). Regarding depth in the water column, Canistrocarpus cervicornis and Dictyota implexa were the macrophytes collected at greater depths (3.5 m. and 6 m. respectively) where concentrations of G. excentricus were estimated to be higher than 10 3 cells gr −1 . That species accounted for 96% and 76% of Gambierdiscus spp. in those samples, respectively.

Discussion
A great deal of research and communication efforts have been carried out during the last decade on the study of tropical and subtropical benthic HABs mainly those associated with ciguatera outbreaks and Gambierdiscus. However, with the exception of a few areas and dinoflagellate genera, the knowledge on benthic harmful microalgae abundance and distribution is still very scarce [47]. That knowledge has become even more essential considering the current expansion of some harmful benthic dinoflagellate species to temperate regions. Ciguatera is an emerging human poisoning in Europe since the first outbreak occurred in the Canary Islands archipelago and in Madeira in 2004 [3,48]. Since then, populations of the CTXs-producer dinoflagellates, Gambierdiscus and Fukuyoa, have been documented both in those regions of Macaronesia and in Mediterranean Sea, though no CFP episodes have been confirmed in the latter region [3]. Yet, there are few data in the literature on harmful benthic dinoflagellates in the Canary Islands other than Gambierdiscus. Studies on this topic are increasing since the emergence of ciguatera on the Islands. It is remarkable that two species of Gambierdiscus, G. excentricus and G. silvae, and two of Coolia, C. canariensis and C. guanchica, have been described in the last decade from  Table 1).

Discussion
A great deal of research and communication efforts have been carried out during the last decade on the study of tropical and subtropical benthic HABs mainly those associated with ciguatera outbreaks and Gambierdiscus. However, with the exception of a few areas and dinoflagellate genera, the knowledge on benthic harmful microalgae abundance and distribution is still very scarce [47]. That knowledge has become even more essential considering the current expansion of some harmful benthic dinoflagellate species to temperate regions. Ciguatera is an emerging human poisoning in Europe since the first outbreak occurred in the Canary Islands archipelago and in Madeira in 2004 [3,48]. Since then, populations of the CTXs-producer dinoflagellates, Gambierdiscus and Fukuyoa, have been documented both in those regions of Macaronesia and in Mediterranean Sea, though no CFP episodes have been confirmed in the latter region [3]. Yet, there are few data in the literature on harmful benthic dinoflagellates in the Canary Islands other than Gambierdiscus. Studies on this topic are increasing since the emergence of ciguatera on the Islands. It is remarkable that two species of Gambierdiscus, G. excentricus and G. silvae, and two of Coolia, C. canariensis and C. guanchica, have been described in the last decade from samples from the Canary archipelago [49][50][51][52]. In addition, genera as Gambierdiscus, Ostreopsis, Prorocentrum, Coolia, and Vulcanodinium had already been reported in the same region [26,53].

Diversity and Abundance of Harmful Benthic Dinoflagellates
For the six epibenthic genera herein studied, both the mean and the maximum cell concentrations showed the following descending order: Vulcanodinium, Ostreopsis, Prorocentrum, Coolia, Gambierdiscus, and Sinophysis (Table 2). These genera are comparable to those reported in other studies in the Canary Islands [26,53], though Fernandez-Zabala et al. [53] limited the study to Gambierdiscus, Ostreopsis, Prorocentrum, and Coolia. As far as we know, there are very few reports on benthic dinoflagellates other than Gambierdiscus or Fukuyoa for other Islands from Macaronesia region. The genera Ostreopsis, Prorocentrum, and Coolia have been also reported for Cabo Verde Islands [53]. Moreover, a list of phytoplankton taxa including Ostreopsis (O. cf. ovata), Prorocentrum (P. lima and P. hoffmaniannum), Coolia sp., and Gambierdiscus excentricus are reported in Madeira [54,55]. Cell abundance comparisons from literature are controversial due to the methodological differences among studies. The main methodological problem is related to the differences on macrophyte surfaces and morphologies which make difficult the standardizations. Methodologies based on quantifying benthic dinoflagellates on artificial substrates have been developed in the last decade in order to normalize cell abundance to a standardized surface [56]. This methodology has been tested in the Canary Islands by Fernandez-Zabala et al. [53] showing that, in most cases, cell abundances of epiphytic dinoflagellates showed lower variability on artificial substrates than on macroalgae. However, a well-defined methodology to quantify epiphytic cells in macrophytes is still needed. In order to make the pertinent comparisons between macrophytes and artificial substrates, there should be a consensus on the methodologies of both procedures. This issue is particularly relevant to quantitate the potential associations between epiphytic dinoflagellates and certain macrophyte taxa.
Maximum concentrations of Gambierdiscus of 4.9 × 10 3 cells gr −1 blot dry weight of host macrophyte (n = 128, from samples collected from five Canary Islands) were already reported in Fuerteventura by Rodriguez et al. [15]. No mention of macrophyte species was given by those authors. Blot dry procedure consists in draining algae overnight over soft laboratory paper. A loss of 62% of weight on average has been reported when dry-blot macrophyte weight is used compared with the manually drained wet weight of the macrophyte used in the present paper; obviously with the corresponding increase in the concentrations of cells when blot-dried weight expression is used [26]. Taking this into consideration, estimated maximum values for Gambierdiscus from those authors and our results (3.1 × 10 3 and 3.8 × 10 3 cells gr −1 wet weight respectively) are of the same order of magnitude. On other hand, blooms of Gambierdiscus with concentrations higher than 10 4 cells gr −1 wet weight were reported in the port of La Restinga [53,57]. Further investigations carried out with standardized methodologies should be addressed to link dinoflagellate populations and their associated environmental conditions with CFP risk areas in the Canary Islands. Furthermore, the high heterogeneity in Gambierdiscus cell numbers in the region makes essential to investigate the relationships between some habitats and detected hotspot areas.
The maximum abundances of Ostreopsis found in the present study were lower than previously reported values in the region since concentrations up to 2.2 × 10 5 cells gr −1 wet weight algae had been documented [53]. Even if these numbers are lower than those for Ostreopsis blooms reported in NW Mediterranean Sea and New Zealand where they have been associated with human health problems by coastal aerosols [58], the risk of Ostreopsis proliferations in Canary Islands should be investigated. The genus Prorocentrum includes benthic species, such as P. lima and P. hoffmannianum, that produce okadaic acid and dinophysistoxins or derivatives which have been associated to Diarrhetic Shellfish Poisoning [32,[59][60][61]. Although in the present study no taxonomic studies were carried out that allowed identification at species level, the different morphologies observed in cell size and shape reveals a high specific diversity which includes both P. lima-like cells and P. hoffmannianum-like specimens. Hence the great interest to carry out taxonomic studies from this potentially toxic genus in the region. Regarding the genus Sinophysis (often observed in Cotillo station), to our knowledge it has not been associated with toxin production. The only species reported so far in the Canary Islands, S. canaliculata, harbors cyanobionts of uncertain taxonomic position [62,63].
Vulcanodinium is not a genus usually included in studies of benthic dinoflagellates, although it has been documented in benthic communities of the Canary Islands [15,26]. Its high abundance in the present study is a remarkable new interesting finding given the high concentrations observed in Playitas station. Vulcanodinium rugosum, the only one species described so far from the genus, was described in 2011 from a French Mediterranean Lagoon and is responsible for producing neurotoxic pinnatoxins (PnTXs) which have been recurrently detected in the shellfish from that region [64,65]. The morphology of Vulcanodinium cells in the samples coincide with those described by Rhodes et al. and Zeng et al. [64,66] as motile cells, however their benthic/planktonic character should be studied.
In the life strategy of this species, the phase in which vegetative division occurs is the benthic non-mobile spherical cells which are considered as cysts [66]. This type of cysts has been called division cysts which have been described in species considered planktonic but with an intense relationship with the benthos [67,68]. No human poisonings by PnTXs are known, however because of their high toxic potential, European Food Safety Authority (EFSA) have pointed out the need for more information on the oral toxicity of these compounds for risk assessment as seafood contaminant [65]. Therefore, future taxonomic, life cycle, and toxin studies are required from the organism found in the Canary Islands.

Associations of Benthic Harmful Dinoflagellates and Macrophyte Communities
Our data showed preferential associations of benthic dinoflagellates in benthic communities of the Canary archipelago. The population distributions of Gambierdiscus and Synophysis were significantly opposite to that of Ostreopsis and Vulcanodinium. Moreover, the two principal components from PCA were preferentially associated with two different algae communities, those of Cotillo and Playitas, respectively. Our results agree with the distinct distributions of Gambierdiscus and Ostreopsis reported by other studies (as for example [69]). These authors reported Ostreopsis spp. in greater concentration in reef areas with high wave energy, coinciding with that mentioned in the Mediterranean by Vila et al. [70]. This is also supported by the results of Grzebyk et al. [71] which reported highest abundances of Ostreopsis in turbulent coral reef habitats. However, blooms of this genus have also been registered in protected areas [31,72]. To better understand these patterns, proper identification of Ostreopsis assemblages in each case, and more information about their ecology and the environmental factors associated with their proliferations are needed. On other hand, distribution of Gambierdiscus has been more associated with sheltered zones protected from the wind and adversely affected by terrestrial inputs [71]. These authors also cite Ostreopsis and Prorocentrum to be more tolerant to terrestrial loads and exploiting different ecological niches than Gambierdiscus. These opposite niches can be determined by the spatial distribution of environmental factors, such as hydrodynamics and terrestrial contributions.
Macrophytes as important elements of benthic niches are interrelated with environmental factors. Among them, wave exposure integrates a wide variety of environmental factors being critical for the biodiversity of coastal ecosystems. It is known that hydrodynamic conditions influence the distribution of intertidal and subtidal organisms [73,74]. In this way, direct and indirect effects of waves have been reported an important driver of the distribution and biodiversity of marine macrophytes in coastal ecosystems [75]. Comparing the characteristics of the habitats studied here, Playitas station is more exposed to wave impact and with macrophyte communities mainly composed by mix red turf algae. This is very different to Cotillo, a more protected habitat with a very different macrophyte composition. Our data suggests that the "charco" in Cotillo station would provide a better niche for the development of Gambierdiscus and Sinophysis. Instead, the rocky platform exposed to waves in Playitas would configure a habitat more suitable for Ostreopsis and Vulcanodinium. Our results suggest that dinoflagellate-macrophyte associations are determined by the characteristics of the studied habitats. The environmental conditions and the microhabitats found in each location would determine the dominant organisms. On the other hand, their populations and the resulting associations could change over time. It must be taken into account that this study represents a fixed image at a certain time of the year. In that sense, further studies integrating spatial and temporal scales are needed, as these dimensions are highly relevant for management purposes and sampling strategy [76].
The fact is that macrophytes serve as habitat and function as complex ecological systems depending on their size, structure and longevity. They exhibit a great variety of epiphytic algae, as well as other microorganisms and associated mobile animals (including meiofauna, macrofauna, and fish). Therefore, given such complexity, discerning the relationship between macrophytes and epiphytic dinoflagellates still remains a difficult task. Notwithstanding, some trends appear in the literature about substrate preferences of the main benthic dinoflagellate genera (Ostreopsis, Prorocentrum, Coolia and Gambierdiscus), linked with macrophyte morphology and taxonomy (see Boisnoir et al. [38] and references therein). Regarding Gambierdiscus, this genus seems to be associated with a wide diversity of macrophyte taxa, although the epiphytic behaviour (growth and attachment) varies by species and host algae [77]. Recent authors have emphasized the importance of microhabitats in benthic communities of ciguatera endemic areas and the complexity of habitats as a determinant factor for the heterogeneity in Gambierdiscus and other epiphytic dinoflagellate distributions [36]. Environmental factors such as light and wave impact have a heterogeneous distribution and, therefore, generate a great deal of heterogeneity in the macrophyte communities and the associated dinoflagellates.
Since the beginning of studies on communities where ciguatera-producing organisms thrive, many authors have remarked that the type of substrate plays an important role in their distributions. Yet, the role of some macrophytes as potentially preferential substrates is controversial. Some of the first ciguatera studies mentioned that rhodophytes were most prone to harbor epiphytic harmful dinoflagellates [78,79]. However, other authors described opportunistic patterns regarding substrate interactions with occurrences on rhodophytes, phaeophytes, chlorophytes, and vascular plants [20,80]. The examination of substrate preferences is controversial due to the difficulty of standardizing cell abundances calculated per weight of the host macrophyte. As far as we know, no estimates of surface/weight ratio have been established which prevent accurate comparisons among the different species of macrophytes. Trying to avoid this handicap we conducted comparisons between types of macrophytes depending on their thallus architecture. The different thallus architecture determines the total surface available for epiphytic dinoflagellates and defines a range of microhabitats which offer shelter and facilitate survival. The available surface and the microhabitats number increase progressively from the two-dimensional foliose to the three-dimensional, flexible filamentous thallus with a high surface:volume ratio (types 1-4 respectively, see material and methods). Our results revealed filamentous macrophytes as preferred substrates for all dinoflagellate genera, suggesting that it shapes a very heterogeneous habitat which increases the diversity and richness of the epiphytic communities. Macrophytes that formed entangled groups also showed high concentrations of dinoflagellates, especially of the genus Ostreopsis. Nevertheless, this classification aimed to define general trends of host preferences has some limitations. For example, the delimitation between the two macrophyte types mentioned is not strict. In our study, the "entangled clumps" type coincided mainly with turf algae that occasionally included some specimens of filamentous algae. Despite these considerations, differences in macrophyte preferences between dinoflagellate genera were observed (i.e., the association between the "entangled clumps" type formed by turfs of rhodophytes and Ostreopsis vs. the preference of ribbon-like macrophytes by Vulcanodinium).

Gambierdiscus Results
In the Canary archipelago, ciguatera outbreaks could be related with local Gambierdiscus spp. including those identified until date: G. australes, G. caribaeus, G. carolinianus, G. excentricus G. silvae and G. belizeanum [15,30]. The morphometric study of these first five species performed by Bravo et al. [26] was applied in the present study with the aim to identify them in samples of Fuerteventura -take into account that the publication of the detection of G. belizeanum in the region was almost coincident with that of the present manuscript. Despite of their morphological similarity, 91% of the specimens were successfully classified at species level. G. excentricus and G. australes were the most abundant species in that order representing 83% (61% and 22%, respectively) of total Gambierdiscus spp. Taking into account that 9% of analyzed specimens were classified within the group G. australes/G. caribaeus, G. australes is almost certainly underrated. The dominance of G. excentricus and G. australes matches previous molecular results based on LSUrDNA and SSUrDNA sequences of cultures and single cells isolated from Eastern Canary Islands [15,26]. Quantification based on morphology is very time consuming and not totally effective, but species-specific quantitative PCR assays have not been yet undertaken in that region as it has been the case in other areas such as the Gulf of Mexico and the Pacific Ocean [28,81,82].
The species of the genus Gambierdiscus produce ciguatoxins (CTXs) and maitotoxins (MTXs) but only the transfer of CTXs up the food chain results in their metabolism and accumulation in fish tissues, thus potentially causing CFP in humans. Although highly toxic, MTXs do not induce CFP because of their low oral potency and inability to accumulate in the muscle tissue of fish [83,84]. It has very recently been reported that different species of Gambierdiscus contain different proportions of the two types of toxins and therefore a very different toxic potential [23,24,85]. For this reason, in order to assess the potential risk of CFP occurrence it is necessary to know the specific diversity and distribution of Gambierdiscus in the region as well as the CTXs and MTXs contained by each one. G. excentricus displays the highest content of CTXs so far [23,24,50] and its CTX-like toxicity has been comparable to that of G. polynesiensis, the predominant CTX producer in the South Pacific, a ciguatera endemic region. In contrast to the consistent toxicity characteristics of G. excentricus, analyses of G. australes have yielded variable results depending on the strains and their origins [24,30,85,86]. The toxicity of the rest of Gambierdiscus species from the Canary Islands has been very scarcely studied; neuroblastoma cell-based assay (neuro-2a CBA) revealed lower CTX-like toxicity than the former ones (or even none for G. caribaeus), although high intraspecific variability has been also reported [24,30].