Introduction

The genus Symbiodinium is a diverse group of unicellular dinoflagellates best known for their endosymbiotic interactions with scleractinian corals and other marine invertebrates and protists. Genetic studies have revealed that the genus Symbiodinium is comprised of nine major groups or clades named clades A–I that each contains multiple sub-clade types (Pochon et al. 2006; Pochon and Gates 2010). This taxonomic diversity is reflected in differences in the functional attributes of Symbiodinium. The nature and composition of endosymbiotic unions, therefore, influence host characteristics such as growth rates, reproductive output, and thermal tolerance (Kinzie and Chee 1979; Fitt 1985; Rowan et al. 1997; Rowan 2004; Little et al. 2004; Stat et al. 2008). Coral–Symbiodinium associations are essential to the survival of the host and underpin the productivity and calcification that creates habitat for the immense biodiversity that coral reefs support (e.g., Muscatine and Porter 1977). The perpetuation of these endosymbioses through time is therefore central to the maintenance of functional integrity in the coral reef ecosystems.

The majority of scleractinian corals produce asymbiotic gametes and larvae that must acquire Symbiodinium from the environment anew each generation (Harrison and Wallace 1990). This mode of endosymbiont acquisition, also called horizontal transmission, requires that Symbiodinium be present and available for acquisition in the free-living compartments of coral reef environments (sediments and surface water). These free-living Symbiodinium have also been proposed to be a source for replenishing endosymbiotic communities in hosts recovering from bleaching (Buddemeier and Fautin 1993; Baker et al. 2008). Although pivotally important to coral reef ecology, our understanding of free-living Symbiodinium diversity and ecology is still limited.

Nano-sized dinoflagellates (2–20 μm) are common in both planktonic and benthic communities (Li 2002; Werner et al. 2008). Free-living Symbiodinium in these communities have been examined using asymbiotic invertebrate hosts as collectors (Kinzie et al. 2001; Coffroth et al. 2006) and by isolating, culturing, and genotyping Symbiodinium-like cells from environmental samples (Carlos et al. 1999; Gou et al. 2003; Coffroth et al. 2006; Hirose et al. 2008b; Porto et al. 2008). Asymbiotic hosts typically acquire one or a few specific Symbiodinium genotypes (Baker 2003; LaJeunesse et al. 2004a; Stat et al. 2009) rather than visualizing the full diversity present in the environment. Similarly, diversity estimates using culturing approaches are confounded by the fact that only a subset of Symbiodinium types are easily culturable (Santos et al. 2001; Coffroth et al. 2006; Hirose et al. 2008b). Interestingly, the majority of free-living Symbiodinium types that have been successfully cultured belong to clades A and B, and some of these have not been identified as endosymbiotic types in these clades. This suggests that Symbiodinium types within the same clade may have very different habitat preferences and may be restricted to either free-living or endosymbiotic mode of living (Carlos et al. 1999; Coffroth et al. 2006; Hirose et al. 2008b; Porto et al. 2008). Microscopy, flow cytometry, and genetics are all common methods used to describe the diversity and abundance of specific taxa in the nanoplankton and microphytobenthic communities (Díez et al. 2001; Moreira and López-García 2002; Unrein et al. 2005; Werner et al. 2008). Recent studies applying these approaches to free-living Symbiodinium have revealed that they are present in high densities in reef sediments and water column (Littman et al. 2008) and that free-living Symbiodinium communities are diverse and contain representatives in clades A, B, and C (Pochon et al. 2010).

The distributions of free-living Symbiodinium types are of interest at all spatial scales, especially in relation to the biogeographic patterns that are relatively well characterized for endosymbiotic Symbiodinium (LaJeunesse 2002; Baker 2003; LaJeunesse et al. 2004b; Pochon et al. 2004; Goulet et al. 2008). Based on the latter, habitat partitioning and large-scale biogeographic distribution patterns are expected among free-living Symbiodinium communities. Habitat partitioning among free-living Symbiodinium types has been suggested by previous studies (Coffroth et al. 2006; Adams et al. 2009; Pochon et al. 2010) and likely reflects variability in environmental ranges and habitat preferences of Symbiodinium. Large-scale biogeographic patterns of free-living Symbiodinium are also expected and may correlate with those of endosymbiotic Symbiodinium depending on the degree to which symbiotic hosts rely on horizontal acquisition. For example, locations across the Hawaiian Archipelago may show less overlap between endosymbiotic and free-living Symbiodinium communities due to the predominance of symbiotic hosts that perpetuate endosymbiosis via vertical transmission (Krupp 1983; Richmond and Hunter 1990). In contrast, an increased overlap may be more common in the Caribbean, where most hosts acquire symbionts from the environment (Baker 2003; Coffroth et al. 2006). Here, we investigate this hypothesis by exploring the diversity and distribution of Symbiodinium in the surface water and sediments of coral reefs from sites on two islands in Hawai‘i and the Florida Keys. The presentation of our data in the context of previous research also provides a summary of all free-living Symbiodinium chloroplast 23S domain V hypervariable region sequences (cp23S-HVR) available to date.

Materials and methods

Collection

Seawater and sediment samples were collected from four sites in Hawai‘i and three sites in the Florida Keys during June–July 2008 (Fig. 1, Table 1). The Hawaiian sites and collection times included Lele‘iwi (N19°44′01.11″ W155°01′04.59″; 15:30–17:00 h), Puakō (N19°58′15.67″ W155°50′52.94″; 09:00–10:00 h), and Honokōhau (N19°40′10.47″ W156°01′37.46″; 11:00–12:00 h) on Hawai‘i Island, and Kāne‘ohe Bay (N 21°26′50.44″ W157°47′49.62″; 10:30–11:45 h) on O‘ahu, Hawai‘i. The Florida Keys sites and collection times were at The Cable (N24°28′10.74″ W81°43′25.02″; 09:00–10:00 h), Western Sambo (N24°28′44.88″ W81°42′59.16; 10:30–11:30 h), and Cottrell Key (N24°36′49.14″ W81°55′16.92″; 09:00–10:00 h). At each site, triplicate 10 l samples of surface seawater were collected directly over the reef at a depth of <1 m. Similarly, three to ten replicates of 2 ml surface sediment samples were collected at depths between 5 and 10 m using a microcentrifuge tube at each site. Only the surface layers of sediments determined as oxic by coloration were collected (i.e., the top 1–5 mm of sediment). Sediment layers of a gray or black coloration indicating anoxic conditions were strictly avoided. Water and sediment samples were immediately stored on ice in the dark and directly transferred to the laboratory for preservation and storage.

Fig. 1
figure 1

Map of collection sites in the Hawaiian Islands and the Florida Keys

Full size image
Table 1 Summary of all cp23S-HVR sequence types included in the phylogenetic networks (Fig. 2)
Full size table

Sample preparation

All seawater samples were first filtered by gravity through a 20-μm nylon mesh to exclude endosymbiotic Symbiodinium from invertebrate larvae (>100 μm; Edmunds et al. 2005, Nozawa and Harrison 2005, Hirose et al. 2008a), soritid foraminifera (100–15,000 μm in diameter; Pochon et al. 2007), and those trapped in coral mucus. The resulting filtrate was then subjected to further filtration through a 5-μm filter using low constant vacuum filtration at a constant 0.1 bar to capture particles of the size range 5–20 μm, including free-living Symbiodinium (5–15 μm; Stat et al. 2006). Each sediment sample was suspended in approximately 200 ml of filtered (0.22 μm) seawater and then sequentially filtered through 20- and 5-μm filters as described above.

DNA analysis

Genomic DNA was extracted from the organisms on the 5-μm filter using 1 ml of a guanidinium protocol described in Adams et al. (2009). DNA was isolated and purified from 200 μl of the total DNA extraction solution. In order to achieve successful PCR amplification, some samples were diluted 1/10 or 1/50. A nested PCR strategy was used to amplify the cp23S-HVR. First, the entire cp23S-DomainV region was amplified from the genomic DNA extraction using the primers “23S1M13” and “23S2M13” (Santos et al. 2002) under the following thermal cycler conditions: 2 min at 94°C, 36 cycles of 94°C for 30 s, 50°C for 1 min, 72°C for 1 min 15 s, followed by 7 min at 72°C. This primer set is not Symbiodinium-specific and amplified the cp23S-Domain V region of other dinoflagellates, as well as Symbiodinium (Santos et al. 2002). One (1) microliter of the resulting PCR products was used as template in a second PCR to amplify the cp23S-HVR region using the Symbiodinium-specific primers “23SHYPERUP” (Santos et al. 2003) and “23SHYPERDN” (same as “23SHYPERDNM13” in Santos et al. 2003 without the M13 sequence, and first applied in Manning and Gates 2008) under the same conditions given above except with an annealing temperature of 55°C. The product of a no-template control from the first PCR was used as template in the second PCR as a negative control for the nested PCR protocol employed. In cases where the second round of amplification yielded no products after several attempts, successful amplification of the first round of PCR with the less specific primer set served as a positive control. This positive control determined that the lack of PCR products with Symbiodinium-specific primers was because of the absence of Symbiodinium in the samples rather than poor-quality DNA templates. Multiple (8–16) extraction dilutions were amplified, and final amplified products were isolated by gel extraction and purified using the QIAquick Gel Extraction Kit (Qiagen), according to the manufacturer’s protocol. Purified PCR products were pooled for each sample and cloned into the pGEM®-T Easy Vector System II (Promega). A total of 18–25 clones per sample were sequenced using an ABI 3730XL capillary-based DNA sequencer (Applied Biosciences) at the Advanced Studies in Genomics, Proteomics and Bioinformatics Sequencing Facility at University of Hawai‘i at Mānoa.

Sequence analyses

Sequences were manually edited and aligned, using Sequencher 4.5 (Gene Codes Corporation). Only sequences that were amplified from two or more independent samples or those identical to sequences deposited in GenBank were included in the analysis (Apprill and Gates 2007; Manning and Gates 2008; Pochon et al. 2010; but see Thornhill et al. 2007). Sequence types that were represented by only one clone in any given library were verified by sequencing in the reverse direction. The clade of each sequence type was identified using nucleotide BLAST (Altschul et al. 1990). Sequences belonging to each clade were aligned with closely related published sequences from GenBank, using a Clustal W alignment in BioEdit v5.0.9 (Hall 1999), and the alignment refined manually. A statistical parsimony network was generated for each cladal alignment using TCS v1.21 (Clement et al. 2000) with a maximum of 50-bp connection steps and treating gaps as a fifth state. Each sequence was assigned a unique name reflecting the locus, clade, and sequence number. For example, “chvA1” corresponds to the chloroplast hypervariable, clade A, sequence number 1 (Table 1, Electronic Supplemental Material, ESM Table S1).

Results

Symbiodinium sequence diversity

The first-round of amplification with the less specific primer set 23S1M13/23S2M13 in the nested PCR was successful in 80 samples. This verified that the quality of extracted DNA was suitable as a PCR template in these samples. Of these, the cp23S-HVR region of Symbiodinium was then successfully amplified and sequenced from 62 samples (i.e., 77.5%; Table 1). Symbiodinium was detected in 20 out of 21 surface water samples (95.2%) and 42 out of 59 of the sediment samples (71.2%). A total of 1,113 Symbiodinium sequences were recovered in our analyses, representing 29 different sequence types (Fig. 2), 13 of which are known and 16 are novel. Each Symbiodinium sequence type was found in 1-19 independent samples in this study, with an average of 3.6 samples per sequence type, though many of the sequence types (N = 18) were found in only 1 or 2 samples. The number of sequences recovered for each clade varied; 9 belonged to clade A (GQ370580–GQ370582, GQ370585–GQ370587, GQ370590, GQ370592, and GQ370594), 13 to clade B (GQ370595–GQ3705601, GQ370603–GQ370607, and JF327758), 3 to clade C (GQ370609, GQ370611, and GQ370613), 1 to clade D (GQ370614), and 3 to clade G (GQ370618–GQ370620) (Table 1, ESM Table S1). No Symbiodinium sequence types belonging to clades E, F, H, or I were found. The majority of novel sequences belonged to clades A and B, while only one novel sequence was recovered in clade C and 2 novel sequences in clade G. Each individual sample yielded 1–5 sequence types, averaging 1.5 Symbiodinium sequence types per sample.

Fig. 2
figure 2

Statistical parsimony networks of clades A, B, C, D, and G of Symbiodinium comprised of sequence types identified in this study and that of closely related published Symbiodinium types, showing a diversity of Symbiodinium in Hawai‘i (yellow) versus Florida (green) and b diversity of Symbiodinium in the sediment (brown) versus surface water (blue). The number of independent samples that contained each sequence type is indicated by the size of the circle; indices are given in each figure. Sequence types identified in this study are indicated by circles and published sequence types by squares. Names of novel sequence types are in red

Full size image

The sequence networks for clades A, B, C, and G show multiple sequences radiating out from core sequences, defined here as sequences having 3 or more sequence types radiating from them (Fig. 2). These core sequences are assumed to be ancestral (Correa and Baker 2009; Stat et al. 2009). Only 1 sequence was recovered from clade D and was distantly related (58-bp substitutions and insertions/deletions) to the only other clade D sequence known for this locus. Core sequences included chvA1, chvA2, chvA15, chvB1, chvB2, chvB17, chvC1, chvG1, and chvG10 (Fig. 2).

Sequence diversity among regions

All 5 clades, A, B, C, D, and G, and 21 out of 29 sequence types, recovered in this study were identified in Hawaiian samples, of which 14 sequence types were unique to Hawai‘i (Figs. 2a, 3). In comparison, a total of 15 sequence types were recovered from Florida samples. Of these, 8 types were found only in Florida, 7 of which belonging to clade B. Seven sequence types were common to Hawai‘i and Florida (chvA3, chvA15, chvB1, chvB2, chvB17, chvB18, and chvG1) (Figs. 2a, 3).

Fig. 3
figure 3

Proportions of free-living Symbiodinium sequence types recovered from sediment (“sed”) and surface water (“water”) samples in Hawai‘i (HI) and Florida (FL)

Full size image

Sequence diversity among habitats

Thirteen sequence types representing clades A, B, C, D, and G were identified in surface water samples. Five sequence types were only found in surface water samples (Fig. 2b), and each water sample contained 1–4 sequence types (average = 2 types/sample). In Hawai‘i, 8 sequence types from clades A–D were identified in surface water (Fig. 3), though clade B (chvB1) was only identified in one water sample from Kāne‘ohe Bay and clade D was only found at Puakō. Sequence types chvA2, chvC1, and chvC3 were the most common in Hawai‘i, and 2 of these 3 sequences were always recovered at a site. Surface water samples in Florida were dominated by clade B, which was found in every sample at every site. In Florida, 5 clade B types were identified, and type chvB2 was the most commonly encountered, being present in every sample but one. Clade G, specifically type chvG1, was found in only 2 water samples at 2 different sites, Cottrell Key and The Cable, in Florida.

In sediment samples, 21 out of the 24 sequence types identified were from clades A–C and 16 of these were found solely in sediment samples (Figs. 2b, 3). Each sediment sample yielded 1–5 sequence types, averaging 1.8 types per sample. In Hawai‘i, 18 sequence types from clades A, B, C, and G were found, though clade C was identified in only 4 (out of 21) sediment samples and clade G was only identified in samples from Puakō. In addition, clade A was the most ubiquitous, being present at every site and in most samples (13); however, no one type was particularly common, as was found for the surface water samples. In Florida, sediment samples contained 14 sequence types from clades A, B, and C; however, clade A was only present in 4 samples (out of 21) and clade C in 1 sample (at The Cable). Clade B was the most frequently found clade in Florida; however, as with Hawai‘i, no one sequence type was more prevalent than another. Of the 29 sequence types identified in this study, 8 (chvA2, chvA14, chvB1, chvB2, chvB17, chvC3, chvC13, and chvG1) were found in both surface water and sediment samples (Figs. 2b, 3); however, no sequence type was found in all habitats and regions (Fig. 3).

Discussion

Free-living Symbiodinium are necessary to establish or re-establish endosymbiosis with many marine organisms and are also likely to have important ecological roles in benthic and pelagic ecosystems (GuoFu et al. 2008; Werner et al. 2008). A number of previous studies have identified free-living Symbiodinium from coral reefs globally, but detailed characterization of their diversity is only now beginning to emerge (Carlos et al. 1999; Gou et al. 2003; Coffroth et al. 2006; Koike et al. 2007, GuoFu et al. 2008; Hirose et al. 2008b; Littman et al. 2008; Manning and Gates 2008; Porto et al. 2008; Pochon et al. 2010; Venera-Ponton et al. 2010). The overall irregularity in the detection of Symbiodinium in our environmental samples and the fact that new studies continue to find previously uncharacterized sequence types suggest that the distribution patterns of free-living Symbiodinium are spatially heterogeneous. This is not surprising given that the distribution patterns of many dinoflagellates are similarly spatially heterogeneous (Mouritsen and Richardson 2003). Part of spatiotemporal heterogeneity in free-living Symbiodinium diversity at any given location may also depend on diel vertical migration observed in cultured Symbiodinium (Yacobovitch et al. 2004) and commonly seen in other dinoflagellates in the field (e.g., Blasco 1978). Sample collection in our study generally occurred in late morning to minimize temporal variability. Expanded sampling in the future is needed to gain a more comprehensive understanding of the ecology of free-living Symbiodinium over dynamic spatiotemporal scales.

The hypervariable region of the chloroplast 23S domain V (cp23S-HVR) was selected as the molecular marker in this study because it is the only marker specific enough to distinguish Symbiodinium from other closely related dinoflagellates in environmental samples (Manning and Gates 2008; Pochon et al. 2010). Although cp23S-HVR provides less taxonomic resolution than either the nuclear ITS, or cp23S-Domain V within which it resides, the phylogenetic patterns resolved by all of these markers are largely consistent with one another (Stat et al. 2009; Pochon et al. 2010). An additional advantage of this locus is that it is coding and may be directly linked to physiological performance. The cp23S gene transcribes for the chloroplastic ribosomal RNA (rRNA), which is strongly conserved through evolution (Harris et al. 1994). Chloroplastic 23S-rRNAs are most closely related to eubacterial 23S-rRNAs (Gray 1988), which forms the peptidyl transferase center that helps protein folding and could ultimately affect physiology (Samanata et al. 2008). Interestingly, Symbiodinium types that were distinguished by cp23S-DomainV sequences have different growth rates under different temperatures in culture (Kinzie et al. 2001, Santos et al. 2002), suggesting that this marker links to phenotypic traits that are important.

Free-living Symbiodinium diversity as compared to symbiotic counterpart

The general structure of the statistical parsimony networks of free-living Symbiodinium based on cp23S-HVR is similar to networks generated for Symbiodinium in other studies for the ITS-2 and cp23S-DomainV in that the derived types radiate out from few ancestral types in each clade (LaJeunesse 2005; Correa and Baker 2009; Stat et al. 2009) (Fig. 2). The diversity of free-living sequence types in clade A recovered in this study is high and much greater than, and in part distinct from, the endosymbiotic Symbiodinium diversity known for this clade (Stat et al. 2006; Correa and Baker 2009). Although our understanding of both endosymbiotic and free-living Symbiodinium diversity is far from complete, this difference supports the suggestion of Coffroth et al. (2006) and Hirose et al. (2008b) that at least some clade A Symbiodinium do not engage in endosymbiosis and are exclusively free-living. Similarly, the high number of free-living clade B sequence types, mostly previously uncharacterized by this marker, is in agreement with a recent study (Manning and Gates 2008) and suggests that some clade B Symbiodinium may also be exclusively free-living. The contrastingly low diversity of free-living clade C Symbiodinium is surprising. Many corals, particularly those in the Pacific, associate with clade C Symbiodinium (e.g., Pochon et al. 2006), and thousands of Symbiodinium cells are expelled from hosts on a daily basis (Stimson and Kinzie 1991; Jones and Yellowlees 1997; Baghdasarian and Muscatine 2000). Consequently, the diversity of clade C Symbiodinium in the environment is expected to be as high as found in endosymbioses. Our results indicate otherwise and suggest that some clade C Symbiodinium may be confined to endosymbiotic habitats. Occurrence of free-living clade G Symbiodinium in both Hawai‘i and Florida is intriguing. Although clade G Symbiodinium has been found in association with a wide range of invertebrates and protists, their distribution range was previously considered to be restricted to the Indo-Pacific (van Oppen et al. 2005; Pochon et al. 2006; Goulet et al. 2008; Granados et al. 2008). Recent phylogenetic analyses of the genus Symbiodinium using the nuclear large subunit ribosomal DNA have revealed that clade G contains two genetically divergent sub-clades. One of these has so far only been found in soritid foraminifera from Guam, Micronesia (Pochon et al. 2001, 2007), and the other sub-clade is most commonly associated with excavating sponges of the genus Cliona (Schönberg and Loh 2005; Schönberg et al. 2008). Recently, the latter sub-clade has also been found in symbiosis with excavating sponges in the Caribbean (Granados et al. 2008; Hill et al. 2011). Interestingly, clade G had not been recovered from the free-living Symbiodinium community in Hawai‘i and the Caribbean (Manning and Gates 2008; Porto et al. 2008), but our study retrieved three clade G sequence types, one of which (chvG1) was found in both Hawai‘i and Florida, and is identical to numerous cp23S-HVR Symbiodinium sequences previously reported from Caribbean excavating sponges (see Table 1, ESM Table S1).

Of the 29 free-living Symbiodinium sequence types recovered in this study, 11 types (chvA1, chvA2, chvA3, chvB1, chvB2, chvB3, chvB10, chvC1, chvC3, chvD1, and chvG1) have previously been identified as endosymbiotic. In Hawai‘i, 5 sequence types (chvA1, chvA2, chvB1, chvC1, and chvC3) previously known as endosymbiotic (Santos et al. 2001; Pochon et al. 2010) were identified as free-living in this study. The overlap between free-living and symbiotic communities in Hawai‘i was greater than was found in Pochon et al. (2010), which is likely due to the additional sampling in Hawai‘i in this study. Nonetheless, the overlap between symbiotic and free-living Symbiodinium communities in Hawai‘i is relatively small, perhaps suggesting that some Symbiodinium are restricted and optimized to the endosymbiotic environment and have a limited ability to survive outside their hosts once expelled (Hill and Ralph 2007). Interestingly, most coral hosts in Hawai‘i transmit their dinoflagellate symbionts vertically without relying on the environmental Symbiodinium pools (Krupp 1983, Richmond and Hunter 1990); therefore, the limited overlap between symbiotic and free-living communities is expected. A larger overlap would then be predicted in Florida where much more symbiotic hosts acquire Symbiodinium from the environment than those in Hawai‘i (Baker 2003; Pochon et al. 2010). The endosymbiotic Symbiodinium communities in Florida, however, have not been thoroughly characterized using the cp23S-HVR marker (Hill et al. 2011), so it is impossible to test this prediction at this time.

Free-living Symbiodinium diversity patterns in different regions

Comparison of free-living Symbiodinium sequence diversity from Hawai‘i and Florida reveals some interesting biogeographic patterns. The free-living Symbiodinium sequence types belonging to clades A, B, C, D, and G were recovered in Hawai‘i, and those of clades A, B and G were found in Florida (Figs. 2a, 3). Furthermore, 85.1% of the free-living diversity in Florida was found in clade B. In contrast, 82.3% of the free-living diversity in Hawai‘i was found in clades A and C (sequences in circles in Figs. 2, 3). The dominance of clade B in free-living Symbiodinium community in Florida interestingly mirrors the endosymbiotic Symbiodinium community for the Caribbean (LaJeunesse 2002; Baker 2003; Goulet et al. 2008). Additionally, higher genetic diversity of free-living Symbiodinium was detected in Hawai‘i (21 sequence types) as compared to Florida (15 sequence types) (Fig. 3). However, it is unclear if this pattern reflects the ecology or is correlated with the uneven sample sizes. Lastly, the sequence types that were found in both Hawai‘i and Florida were all ancestral sequence types, except for two, chvA3 and chvB18 (Fig. 2a). These ‘generalists’ may have been widespread before the Indo-Pacific and Atlantic oceans were separated and remained undifferentiated because of limited sexual reproduction (LaJeunesse 2005). However, because dinoflagellate endemism is generally considered rare (Taylor et al. 2008, but see Thornhill et al. 2009), more derived types may be found elsewhere with further sampling.

Diversity patterns in different habitats

Some interesting patterns also emerge when genetic diversity of free-living Symbiodinium in surface water and sediment are compared with one another. The most abundant and ancestral sequence types (except chvA14) were found both habitats, while the less abundant and more derived sequence types were generally found in sediments (Fig. 2b). Further, a relatively smaller number of sequence types dominated surface waters, chvA2, chvC1, and chvC3 in Hawai‘i and chvB2 in Florida (Fig. 3). In contrast, sequence types were almost equally represented in sediments of each region (Fig. 3). Given the differences in sample size and collection protocols to accommodate the solid and liquid nature of environmental samples, the Symbiodinium diversity in the two habitats cannot be directly compared. However, the observed diversity of Symbiodinium sequence patterns characteristic for each environment is noteworthy. It is possible that such diversity patterns are indicative of ecological differences in free-living Symbiodinium communities in these habitats. The specificity of certain Symbiodinium sequence types to a particular habitat, and their abundances and capacity to move between these habitats are largely unknown at this stage but certainly warrant further investigation.