This study was carried out from April to December 2015 at two reef sites located in the northern Mexican Caribbean: Isla Contoy (IC) and Puerto Morelos (PM). Both sites are protected natural areas of the Mesoamerican Reef Barrier System (Figure 1). Ixlaché reef is located 1 km south of IC. It is a small reef 800 m long and 60 m wide. Depth varies between 0.5 and 3 m, macroalgae cover is 60%, and is influenced by coastal currents, waves, tides, and winds predominantly from the southeast and polar fronts. On the other hand, the Bonanza reef site is part of the PM reef; it is a shallow area of <2 m depth and macroalgae cover is higher than 75%.

Two macroalgae species were selected for this study, each representing a distinct functional group (Littler et al., 1983; Steneck and Deither, 1994). Amphiroa is a slow-growing rate coralline articulated macoalgae present at relatively low cover in the sites of study. Dictyota is fast-growing rate fleshy, foliose macroalgae with a high cover. Both sites selected for this study are macroalgae-dominated reefs where Dictyota, Turbinaria, Lobophora, and Laurencia are very common, although patches of the coralline algae Amphiroa still remain.

At each site, 10 specimens of each genus of macroalgae were collected. The distance between collected specimens was at least 10 m according to Lobel et al. (1988). Collected macroalgae were stored in plastic bags with seawater and placed into a cooler with water of the site to maintain the temperature. Algae were fixed with formaldehyde at 4% final concentration and stored until analysis.

In the laboratory, macroalgae samples were strongly shaken during 2 min to detach epiphytic dinoflagellates. Macroalgae were separated from the seawater and the excess of water was removed with absorbent paper. Macroalgae were weighed and fixed in formaldehyde. A total of 400 samples were processed.

Water temperature was recorded every 3 h during the study period. HOBO Pro-v2 sensors were placed at the reef sites anchored at the level of the macroalgae mats. For nutrient determinations, water samples were collected with a 50 ml syringe, previously washed with a 10% HCl solution, connected to a filter holder containing a Whatman® membrane filter of 0.22 μm size pore and 2.5 mm diameter. Filtered seawater was recovered into a 30 ml Nalgene® polypropylene bottles that were stored in a cooler with ice. Once in the laboratory the samples were stored at −80°C. The quantification of NOx- (NO3- + NO2-) and PO43- was performed by using a Skalar® segmented flux autoanalyzer following the techniques described by Strickland and Parsons (1972) and Grasshoff et al. (1983).

The abundance of epiphytic dinoflagellates was estimated with the help of a Carl Zeiss inverted microscope model Axiovert 40 CFL at 10X magnification using a Sedgewick-Rafter counting chamber. Abundance is expressed in cells·g⁻¹ of wet weight macroalgae (Reguera et al., 2016).

For dinoflagellate species identification the following specialized literature was used: Fukuyo (1981), Faust (1990, 1993, 1994, 1995, 1997, 1999), Faust et al. (1996), Litaker et al. (2009), and Almazán-Becerril et al. (2015, 2016).

Dinoflagellate observations and measurements were made with an Axio Imager A2 Zeiss microscope at 40X magnification using the differential interference contrast and epifluorescence techniques. Photographs of the specimens were taken with an AxioCam ICc1 digital camera coupled to the microscope. The image processing was carried out with the software Axiovision®.

The non-parametric Mann–Whitney U-test (α = 0.05) was used to compare the dinoflagellate abundance at each sampling date between sites, and the non-parametric Wilcoxon rank test (α = 0.05) was used to evaluate the potential differences between sites during the whole period of sampling by paired data. The Spearman rank correlation was applied to describe the community structure on basis of the associations between species in both macroalgae. The association of temperature and nutrient concentration was assessed using the same correlation test.

We recorded species of the genera Gambierdiscus, Coolia, Prorocentrum, Ostreopsis, Amphidinium, Bysmatrum, and Sinophysis. Identification of the species of Gambierdiscus and Coolia is difficult based only on the characters observed (size and shape of the cells) during the evaluation of abundance. Therefore, hereafter the Gambierdiscus and Coolia specimens will be referred as Gambierdiscus spp. and Coolia spp. Five species of Ostreopsis were observed but only two of them accounted for more of the 95% of the abundance of this genus. These two morphotypes of Ostreopsis, were clearly differentiated during the counting and were designated as O. cf. marina and O. heptagona. Finally, six Prorocentrum species were found: P. belizeanum, P. emarginatum, P. hoffmannianum, P. lima. P. cf. levis, and P. rathymum. Gambierdiscus spp. Coolia spp., Ostreopsis cf. marina, O. heptagona, and the six Prorocentrum species were the most abundant groups found in the samples, therefore, the analysis was constrained to these taxa. Figure 2 shows the main species recorded in the zone of study.

Taking into account the entire sampling period, total abundance of epiphytic dinoflagellates was higher at PM on both algae (Tables 1, 2; Figure 3). However, when the total abundance and the abundance by group were compared at each sampling date, it became evident that the numeric differences were not always significant for all groups (Table 3). Total dinoflagellate abundances on Amphiroa were significantly higher at PM during all sampling dates (p < 0.05, Mann–Whitney test) except for Sep17 and Nov09 when no differences were detected. In Dictyota total abundances were higher in PM during the entire study period, except on Jun22 when no differences were detected.

In PM, the maximum mean abundance recorded in Dictyota was 38,471 ± 11,420 cells·g⁻¹ on Apr16. In the case of Amphiroa, the maximum abundance was 8,803 ± 964 cells·g⁻¹ and was recorded on May08. At IC the maximum density of dinoflagellates on Dictyota occurred on Dec02 (16,160 ± 3,770 cells·g⁻¹), and on Amphiroa the highest abundance was 16,160 ± 3,770 cells·g⁻¹ in May08. In PM, the maximum individual abundance was registered on Dictyota during Apr16 (112,293 cells·g⁻¹).

Dominant species during most of the period were Ostreopsis heptagona and O. cf. marina Therefore, the variability of total abundance of epiphytic dinoflagellates were due the variation in the contribution of these species at each site. In fact, the contribution of Ostreopsis species varied linearly with total abundance (Figures 4A,B).

The contribution of O. cf. marina was important at both sites, but O. heptagona was important only in PM because this species was absent at IC during almost the entire sampling period. O. cf. marina and O. heptagona reached maximum densities of 53,231 and ~46,462 cells·g⁻¹ on a Dictyota sample during Apr16.

The contribution of Gambierdiscus spp. to total abundance was marginal on both macroalgae and at both sites: from 0 to 3.26% on Amphiroa, and from 0 to 4.6% on Dictyota. The highest abundance of this group on Dictyota was found at PM on Aug10 (178 ± 60 cells·g⁻¹); and on Amphiroa the maximum abundance was observed on Dec02 (71 ± 38 cells·g⁻¹).

The most abundant species of Prorocentrum were P. hoffmannianum and P. lima on both algae, representing the largest fraction of this genus (50–94%). Prorocentrum was important (65% of total abundance) when Ostreopsis abundances were low. Maximum abundance of all Prorocentrum species was registered at PM on Dictyota (4,144 cells·g⁻¹).

The contribution of Coolia spp. to total abundance was lowest at PM. This genus did not present a contribution higher than 10% on both algae. In contrast, in IC on some dates it reached more than 30% on both macroalgae. The abundance of this group was lower than 1,000 cells·g⁻¹on, and its abundance on Amphiroa was always lower than 300 cells·g⁻¹.

There were more correlations between species on Dictyota than on Amphiroa in both sites (Table 4). Prorocentrum lima, P. belizeanum, and P. hoffmannianum were correlated with each other on Amphiroa at both sites but not on Dictyota. Also P. rathymum was correlated with O. heptagona on Amphiroa at PM. Given that this species was almost absent at IC, the correlation did not exist at this site. At Isla Contoy, P. hoffmannianum, and P. cf. levis were correlated with Coolia spp. The association between P. lima and P. hoffmannianum was the only association detected in both algae at both sites. On Dictyota, the community structure was more complex because it included the correlations of Coolia with Gambierdiscus, and each of these genera with the Prorocentrum species. The negative correlations of O. marina (with Gambierdiscus, P. emarginatum, and P. hoffmannianum) were particularly important, but this only occurred at IC.

The temperature varied from 24.04 to 30.20°C at IC and from 26.68 to 29.7°C at PM. Warmer temperatures were registered between August and September, and April was the coldest month in both sites (Figure 5). Over the entire sampling period, mean water temperature was slightly lower at IC (28.25°C) than at PM (28.74°C). Although small, this difference was significant (p < 0.05, Wilcoxon test). On Apr17, the temperature was 1.46°C higher at PM (27.90°C) than IC (26.44°C). Over the rest of the sampling period, the temperature differences between sites were lower than 1°C. After May29, the temperature decreased almost 1°C at both sites, followed by a gradual increase until the maximum value in August–September. This behavior is common in the zone and is related to the influence of the polar fronts coming from the north from December to February, although during the year of study, polar fronts occurred until July. For example, the event which occurred during June 12–19 was intense enough to decrease the temperature recorded on Jun22.

High NOx- (NO3- +NO2-) concentrations were measured on both sites with mean values of 8.0 and 7.8 μmol·l⁻¹ for PM and IC, respectively, but no significant differences were detected along the entire period of study. Conversely, PO4-3 concentration was different between sites: 2.3 μmol·l⁻¹ at IC and 2.4 μmol·l⁻¹ in PM (p < 0.05, Wilcoxon test). NO3- and PO4-3 showed a high correlation at PM (r = 0.95) but not for IC. Temperature did not correlate with nutrients at PM nor IC (p > 0.05, Spearman rank correlation).

The correlation between temperature and the abundance of O. cf. marina is shown in Figure 6. At both sites and on both algae this species showed a negative correlation with temperature (p < 0.05, Spearman rank correlation). Maximum abundances were detected at temperatures between 27 and 28°C, independent of the macroalgae substrate. In contrast, Prorocentrum lima and P. emarginatum were positively correlated with temperature at IC and only on Dictyota: r = 0.66 and r = 0.79, respectively. A negative correlation between O. heptagona and PO4-3 concentration was found at PM on both algae (r = −0.67 and r = 0.7 on Amphiroa and Dictyota, respectively). Correlations with NO3- only were observed at IC with the abundances of P. emarginatum, P. rathymun, and Coolia spp. on Dictyota (Table 5).

Describing the taxonomic composition of epiphytic dinoflagellates is more complicated now than in the past because at present there are more described species for each of the benthic genera. Pioneer studies on benthic dinoflagellates considered all specimens with the typical shape of Gambierdiscus to be G. toxicus, or all those individuals similar to Coolia as C. monotis. However, Faust (1995) and Litaker et al. (2009) increased to four the richness of the genus Gambierdiscus in the Caribbean: G. belizeanus, G. caribaeus, G. carolinianus, and G. carpenteri. Also the number of Coolia species increased after reports of the presence of C. tropicalis, C. santacroce, C. malayensis, and C. areolata (Faust, 1995; Karafas et al., 2015; Almazán-Becerril et al., 2016). Since the morphological identification of each of these species involves the inspection and analysis of the thecal plates, which is not a practical procedure given the high number of samples to analyse, it was necessary to join the species of each genus and consider them as taxonomic groups, although the ecological behavior of individual species of the same genus remain unclear.

Prorocentrum has the highest richness among the benthic-epiphytic thecate dinoflagellates, and the Caribbean harbors a great fraction of these species (Faust, 1990, 1993, 1994, 1997), however, just some of them are common in the macroalgae samples which allow an easier identification of the specimens. Prorocentrum lima, P. hofmmaniannum, P. belizeanum. P. rathymum, and P. emarginatum are well-recognized under light microscope as different species.

Ostreopsis taxonomy represents a challenge due to the lack of specific morphological characters for species discrimination (Penna et al., 2005; Parsons et al., 2012), however, in this study two Ostreopsis species were identified as O. heptagona and O. cf. marina. In the first case, the contact between the plates 4′ and 4″ (according to the interpretation of Besada et al., 1982) was an adequate characteristic to identify this species. The other species was referred as O. cf. marina because the original description of this species made by Faust (1999) matches with the characteristics of the specimens observed in this study. A closely related morphological species is O. lenticularis, also reported for the Caribbean (Ballantine et al., 1988; Moreira et al., 2012, 2017), which has nearly the same shape (Fukuyo, 1981; Faust, 1999). However, the size reported for this species in the literature is consistently lower than the size of the specimens observed in this study. Therefore, the discrimination of the specimens of these two species under microscope during counting was based on the size and shape of the specimens and was ratified by plate analysis of some cells in several samples. As consequence, the taxonomic composition of epiphytic dinoflagellates cannot be unequivocally determined by using the standard methodology (rough morphometric discrimination), and for purposes of practicality and comparability it is necessary to work with this “schematic” community.

The taxonomic composition of the epiphytic dinoflagellate community on the surveyed algae of this study is quite similar to those reported in similar studies in the Caribbean (Morton and Faust, 1997; Delgado et al., 2006; Boisnoir et al., 2018) and other regions of the North Atlantic like the south of the Gulf of Mexico (Okolodkov et al., 2007, 2014) or Florida (Bomber et al., 1989) and even in the Pacific (Parsons and Preskitt, 2007; Richlen and Lobel, 2011) although each species (or genus) did not contribute in the same proportion to the total abundance. For example, Delgado et al. (2006) found that P. lima and G. toxicus were more abundant than O. lenticularis and C. monotis, whereas Boisnoir et al. (2018) reported that Ostreopsis spp. and Prorocentrum spp. were numerically more abundant than Coolia and Gambierdiscus. In the Gulf of Mexico, Okolodkov et al. (2007) found that depending of the site and the date P. lima or C. monotis could dominate the community, and in the Yucatan coast, Prorocentrum species, particularly P. rathymum was the dominant species on seagrass meadows present in Yucatan coast (Okolodkov et al., 2014).

The results of this study demonstrate that all the taxonomic groups were present on both algae at both localities, excepting O. heptagona which was almost absent at IC with only a few cells observed between September and December. As the distribution of this species comprises the Gulf of Mexico, Florida and the Caribbean (Bomber et al., 1989; Faust et al., 1996), its low abundance at IC during the period of study could be the effect of a local characteristic of the site, because at the same time a high number of specimens were detected at PM.

The short period of sampling of this study was not enough to determinate seasonal patterns for any species of dinoflagellate (although positive significant correlations were found between temperature and P. lima and P. emarginatum) and it appears that more time is necessary to characterize seasonal variability. For example, Ballantine et al. (1988) found evidence of seasonality in Gambierdiscus and O. lenticularis in Puerto Rico in a 3-year study. Also, Chinain et al. (1999) used a 5-year time series to describe the seasonality of Gambierdiscus spp. at Tahiti, and Chateau-Degat et al. (2005) used an 8-year study in French Polynesia to describe the seasonal behavior of Gambierdiscus. Higher abundances were observed during April and May 08 on both algae in PM, and in May 08 for both algae at IC, additionally the higher abundance in Dictyota at IC was detected on Dec02.

These fluctuations were dominated by the contribution of Ostreopsis species which was largely variable and increased linearly with total abundance. The high abundance of Ostreopsis species was also reported at the Caribbean in Cuba (Moreira et al., 2012, 2017) and Puerto Rico (Ballantine et al., 1988) where O. lenticularis reached higher abundances than the reported in this study (Table 6). When the abundance of Ostreopsis species was removed from the data, the contribution of Coolia spp., P. lima, and P. hoffmannianum were more important, while the abundance of Gambierdiscus spp. and the rest of Prorocentrum species was almost marginal. Prorocentrum lima is a conspicuous member of benthic dinoflagellate communities in the Caribbean and Gulf of Mexico where it attaches to a variety of algae and seagrasses and at some sites could be the dominant species with maximum abundances above 10,000 cells·g⁻¹ on phaeophyte algae (Delgado et al., 2006), and up to 29,756 cells·g⁻¹ of Thalassia testudinum (Okolodkov et al., 2007). Also, at Florida Keys, Bomber et al. (1989) found that P. lima was more abundant on green algae Penicillus capitatus and Avrainvillea nigricans as compared to other brown or red algae species.

The case of Gambierdiscus abundance is remarkable because the species of this genus and those of Fukuyoa are thought to be primarily responsible for the toxin production involved in ciguatera intoxications. Although there is an intrinsic difficulty in comparing abundances between different studies based on number of cells normalized by wet weight of macroalgae due to the high variability of weight to surface area ratios between the different groups of host macroalgae (Lobel et al., 1988), there are some reports of high abundances of Gambierdiscus spp. on Dictyota in the Caribbean. For example, Ballantine et al. (1988) reported a maximum abundance slightly higher than 2,000 cells·g⁻¹ in Puerto Rico and Morton and Faust (1997) found 400 cells·g⁻¹ in Belize. Also, Bomber et al. (1989) reported a maximum mean value of 8,191 ± 300 cells·g⁻¹ in Florida. In contrast, the Gambierdiscus abundances found on Dictyota in this study were similar to the values reported by Lobel et al. (1988) at the Caribbean Island St. Barthélemy (5–56 cells·g⁻¹). The factors that control the abundance of Gambierdiscus populations in the zone remain to be solved.

In the Caribbean high abundances of Ostreopsis lenticularis has been reported for Puerto Rico (Ballantine et al., 1988) and Cuba (Moreira et al., 2012, 2017), but no reports of high densities exist for O. marina and O. heptagona in this region. Table 5 shows that maximum densities recorded for Ostreopsis species in the Caribbean vary from 10² to 10⁵ cells·g⁻¹ in Dictyota. Indeed, the highest abundance was recorded in Cuba (531,000 cel·g⁻¹) that is two times the maximum value reported by Ballantine et al. (1988) for the same species at Laurel Reef site, in Puerto Rico (235,803 cells·g⁻¹). High abundances have been reported for O. siamensis (Shears and Ross, 2009) at New Zealand coasts (1.4 × 10⁶ cells·g⁻¹ of Carpophyllum plumosum) and for O. cf. ovata at the Mediterranean (Cohu et al., 2013) where the maximum abundance recorded was 8.54 × 10⁶ cells·g⁻¹ of Dictyota spp. These results demonstrate that at temperate and tropical latitudes Ostreopsis species can reach high abundances, emphasizing the potential of Ostreopsis species to produce large benthic blooms globally (Rhodes, 2011).

In terms of abundance, both algae are used as substrate by all species. However, in terms of community structure, correlation between species can be used to know if the substrate offers common advantages for different species, or if the increase of one species implies the decrease of another. The results of this study suggest that the structure of the epiphytic dinoflagellate community is different on each of the host algae Amphiroa and Dictyota, and there are even differences in the same alga between sites. These findings suggest that Prorocentrum species and Coolia constitute the basic structure on Amphiroa, and on Dictyota the structure is given by Prorocentrum, Gambierdiscus and Coolia that share the habitat. In addition, O. marina presents habitat separation with Prorocentrum and Gambierdiscus. A negative correlation between Ostreopsis and Prorocentrum was reported by Richlen and Lobel (2011) in the Pacific, but Boisnoir et al. (2018) found the contrary in the Eastern Caribbean. Also, the habitat separation between Gambierdiscus and Ostreopsis has been reported by Bomber et al. (1989) in Florida, but in the Caribbean, Ballantine et al. (1988) did not find evidence of correlation between these genera. The dominance of Ostreopsis over the other genera of dinoflagellates could be related to its preference to sites with little water motion, as these are shallow sites (Richlen and Lobel, 2011; Cohu et al., 2013; Boisnoir et al., 2018)

The distance between study sites is nearly 50 km; both sites are shallow and located in the northern region of the Mesoamerican Barrier Reef System. Therefore it is reasonable to suppose that these sites also undergo the same climatological and oceanographic conditions, generating similar abundances and taxonomic compositions of dinoflagellate populations. However, the results showed some important differences. For example, the differences in temperature are low but significant and this could be explained by the climate patterns. The region presents three seasons: (1) the season influenced by polar fronts (locally called “nortes”) between December and February, (2) the dry season between March to May, and (3) the rainy season between June and November. The presence of a polar front can occur year-round (Henry, 1979), but its intensity is higher during winter. During 2015 the cold fronts were intense even during July (2015). Figure 5 shows the temperature series in the study sites. Both series have the same behavior but the magnitude of the variations is higher at IC than PM. The main effect of the polar fronts on the shallow aquatic ecosystems is the abrupt decrease of water temperature, but also the generation of strong movement of the surface layer of water. The energy of polar fronts decreases as they flow southward, therefore their effect on the coral reef systems should be higher at IC than PM. Since the energy of polar fronts removes host macroalgae, the turnover of dinoflagellate populations could be higher at IC, preventing their accumulation and resulting in higher abundances at PM. As the nutrient concentration remains high year-round, Dictyota can recover its biomass and abundance in short periods, providing surface for the attachment of Ostreopsis, which also grows rapidly (Ballantine et al., 1988). This implies that Dictyota and Ostreopsis are opportunistic genera exploiting an expanding habitat produced by the phase shift in the coral reefs.

The presence of Ostreopsis blooms on Dictyota along the Caribbean coasts has some ecological implications. First, toxicity has been reported in populations of O. lenticularis from Cuba and Puerto Rico (Ballantine et al., 1988; Moreira et al., 2012), as well as O. heptagona (Norris et al., 1985); in addition, some samples of O. cf. marina taken after the sampling period of this study showed toxicity on mouse bioassays and hemolytic activity (unpublished data). In fact, Ostreopsis species produce palytoxin-like compounds which are involved in human intoxications and possibly mass mortality of vertebrates and invertebrates (Ramos and Vasconcelos, 2010; Faimali et al., 2012; Patocka et al., 2017). Secondly, Dictyota is a macroalgae favored by the changes in the coral reef systems. In the sites studied these algae can form mats of several tens of square meters, at times reaching nearly 80% of cover (Delgado-Pech, 2016). However, the low palatability of Dictyota to larger benthic herbivores reduces grazing (Cruz-Rivera and Villareal, 2006) and prevents the flux of toxins to higher trophic levels. Dictyota mats are persistent, growing on other calcareous and coralline algae, corals and any other hard surface. Still, their attachment to the substrate is not firm enough to avoid detachment by the strong waves formed under the influence of polar fronts. During these events, the cover of Dictyota and the abundance of Ostreopsis decrease in the benthic environment, although, in the case of Ostreopsis, the cells (and their toxins) could remain in the water column and possibly enter the planktonic food web. If the coral-algal phase shift in the Caribbean results in dominance of Dictyota and other fleshy macroalgae, toxic Ostreopsis species could increase their abundance and distribution and cases of ciguatera and other syndrome of benthic origin could become a more serious public health problem in the region.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.