Introduction

Corals comprise a dynamic and highly diverse consortium of microorganisms, including bacteria, archaea, fungi, viruses, endolithic autotrophs and symbiotic dinoflagellates, and as such represent the paradigm of a holobiont (Rohwer et al., 2002; Bourne and Munn, 2005; Rosenberg et al., 2007; Wegley et al., 2007). Interactions among coral associates influence holobiont physiology and health. The diversity and distribution of host–symbiont relationships are altered by environmental changes such as increased temperature (Banin et al., 2000; Ben-Haim et al., 2003; Bourne et al., 2007), nutrient enrichment (Szmant, 2002; Garren et al., 2009) and diseases (Sunagawa et al., 2009). Such alterations in the coral resident microbial community structure may either present as a stress response, or facilitate more rapid and versatile adaptation of the holobiont to changes in environmental conditions. The latter supposition is termed the ‘probiotic hypothesis’ (Reshef et al., 2006) and should be considered when examining the influence that environmental changes have on corals.

One environmental change extensively researched in recent years is the projected increase in atmospheric CO2. Atmospheric CO2 partial pressure (pCO2) has been increasing since the pre-industrial period, presently exceeds 380 p.p.m., and is expected to increase to 700 p.p.m. or more by the end of this century (IPCC, 2007). One of the direct outcomes of rising atmospheric pCO2 is ocean acidification. In light of these atmospheric CO2 projections, global ocean models predict surface pH reductions of 0.2–0.4 pH units by 2100 (Caldeira and Wickett, 2005).

Changes in ocean pH are expected to impact marine ecosystems and cause shifts in community structure (Ishimatsu et al., 2005; Bibby et al., 2007; Hoegh-Guldberg et al., 2007; Kuffner et al., 2007; Martin et al., 2008), especially among calcifying organisms (Gattuso et al., 1998; Orr et al., 2005). Several studies have shown that reduced pH produces a variety of physiological effects in the holobiont, such as altered dinoflagellate productivity (Crawley et al., 2010), increased susceptibility to bleaching (Anthony et al., 2008) and decreased coral calcification (Fine and Tchernov, 2007; Anthony et al., 2008). However, other studies have reported no change, or even an increase, in coral calcification in response to lower pH (Ries et al., 2009; Jury et al. 2010; Rodolfo-Metalpa et al., 2010).

Bacterial associates are also expected to respond to changes in pH. Indeed, studies examining the response of coral-associated microbial communities to decreased pH conditions (Vega-Thurber et al., 2009; Meron et al., 2011) have reported changes in bacterial species composition. Specifically, these studies have found an increase in bacteria belonging to families that include known coral pathogens, such as Vibrionaceae and Alteromonadaceae (Bourne and Munn, 2005; Ritchie, 2006; Arboleda and Reichardt, 2009; Sunagawa et al., 2009; Meron et al., 2011), as well as families previously isolated from diseased, injured or stressed marine invertebrates, such as Rhodobacteraceae (Meron et al., 2011). In addition, increased levels of secondary metabolites and stress resistance factors (for example, antibiotics and toxins) were observed at low pH (Vega-Thurber et al., 2009; Meron et al., 2011), suggesting that the coral holobiont exhibits a stress response to reduced pH.

These previous studies were all done under controlled laboratory conditions. Here, we used a natural pH gradient off the coast of Ischia (Gulf of Naples, Italy) to examine the impact of pH on the coral microbial community. This gradient results from a natural marine CO2 flux from volcanic vents (Hall-Spencer et al., 2008). We monitored the response of the microbial communities of two Mediterranean coral species, Balanophyllia europaea and Cladocora caespitosa, to this pH gradient. Our results provide insight into the potential effects of predicted ocean acidification on corals, at both the holobiont and community level.

Materials and methods

Experimental design

The experiment was carried out in Castellod’ Aragonese off Ischia Island, Italy (40°043.84′N; 013°57.08′E) (Supplementary Figure S1), where volcanic CO2 emissions from the seabed acidify a large shallow-water area. The vent gas comprises 90–95% CO2, 3–6% N2, 0.6–0.8% O2, 0.2–0.8% CH4 and 0.08–0.1% Ar (no sulfur) (Hall-Spencer et al., 2008). The experimental setup was designed to compare ambient pH (site C, pH 8.1) with three reduced pH sites: site B1 (mean pHT 7.38, min 6.83), site B2 (mean pHT 7.53, min 7.10) and site B3 (mean pHT 7.77, min 7.22). See Supplementary Table S1 for additional data and carbonate chemistry.

In March 2008, 28 polyps of B. europaea and 12 colonies of C. caespitosa at 3–7 m depth were collected from around Castello Aragonese. These corals do not occur normally in the area of the vent plume. The corals were transported to a nearby laboratory (Stazione Zoologica A. Dohrn), where they were maintained in flow-through aquaria continuously supplied with seawater pumped from 3 m depth at ambient temperatures. The turnover rate of seawater in the 150-l aquaria was 30% h−1. Light irradiance of 100 μmol photons m−2 s−1 was provided by True-Lite (NRG Saver Supply, Henderson, TX, USA) and Sunlux (Sunlux Technologies, Bangalore, India) fluorescent tubes, and measured using an Li-Cor 4π spherical underwater quantum sensor (LI-193SA; Lambda Instruments, Lincoln, NE, USA). The photoperiod was 12:12 h light/dark. After 4 days recovery, samples were secured using epoxy glue (HoldFast, Cleveland, OH, USA) onto individual tagged plastic plates (4–6 × 4–6 cm2).

Seven B. europaea polyps and three colonies of C. caespitosa were transplanted to 3–4 m depth at each test site (sites B1-B3) on the south side of the Castello near the CO2 vents, where pHT were reported to vary from 7.2–7.9 units (Hall-Spencer et al., 2008). In addition, seven B. europaea polyps and three colonies of C. caespitosa were transplanted to a reference site 100–150 m away from the vents where the pH was normal (Site C, mean pHT 8.0–8.2). In summary, two PVC plates (30 × 60 cm) were positioned at each of the four sites, and each plate was considered an experimental block (block A and B). Samples were collected after 7 months and frozen immediately in liquid nitrogen for subsequent molecular work, or brought back to the laboratory as live animals for further analysis.

Measurements of seawater carbonate chemistry

Seawater temperature was measured at 15 min intervals in 2008 and 2009 using Hobo Onset loggers positioned near the coral sites. Total alkalinity (AT) and pH in total scale (pHT) were measured frequently in situ (Supplementary Table S1). Water samples were collected in glass bottles next to the transplants or in aquaria, and the pHT was immediately measured using a meter accurate to 0.01 pH units (Methrom pH mobile) calibrated using TRIS/HCl and 2-aminopyridine/HCl buffer solutions (Dickson et al., 2007). Seawater samples were then passed through 0.7 μm pore size filters (GF/F Whatman) and treated with 0.05 ml of 50% HgCl2 (Merck, Analar, Whitehouse, NJ, USA), and stored in the dark at 4 °C pending analysis. Three replicate 20 ml sub-samples were analyzed at 25 °C using a titration system composed of a pH meter with a Methrom pH electrode and a 1-ml automatic burette (METHROM, Herisau, Switzerland). pH was measured at 0.02 ml increments of 0.1 N HCl. AT was calculated from the Gran function applied to pH variations from 4.2 to 3.0, as mEq l−1 from the slope of the curve HCl volume versus pH. Titrations of total alkalinity standards provided by A.G. Dickson (batch 99 and 102) were within 0.7 μmol kg−1 of the nominal value. Parameters of the carbonate system (pCO2, CO32−, HCO3, CT and saturation state of calcite (Ωc) and aragonite (Ωa)) were calculated from pHT, mean AT, temperature and mean salinity (38) using the free-access CO2sys package. AT and salinity were virtually constant throughout.

Chlorophyll fluorometry, protein and chlorophyll concentration assays

To study the photosynthetic capacity of symbiotic dinoflagellates we used an imaging pulse amplitude modulated fluorometer (Walz GmbH, Effeltrich, Germany). After collection, each sample was dark-adapted for 20 min before various parameters of individual polyps were assayed: the maximum quantum yield of photosystem II (Fv/Fm), electron transport rate (ETR) and non-photochemical quenching (NPQ). Maximum quantum yield of photosystem II was calculated using the Genty equation (Genty et al., 1989): Fv/Fm=(Fm−F0)/Fm.

After taking these measurements, all fragments were collected, frozen and kept in the dark. Subsequently, the coral tissue was removed using an air brush with filtered seawater. The slurry was homogenized using a DIAX 100 homogenizer (Heidolph Instruments GmbH & Co. KG, Schwabach, Germany) for 10 s before centrifugation at 4000 g for 5 min. The dinoflagellate pellet was re-suspended, and aliquots removed for cell counts and chlorophyll measurements. Chlorophyll was extracted in 100% acetone at 4 °C for 5 h, and chlorophyll a measured using a spectrophotometer (Ultrospec 2100 pro, GE Bioscience, Maryland Heights, MO, USA) and appropriate equations (Jeffrey and Humphrey, 1975). Cell counts were obtained from images obtained from a digital camera (Digital Sight, Nikon, Tokyo, Japan) attached to a microscope (D-Eclipse, Nikon), and image analysis software (MATROX Inspector 2.1, Matrox Imaging, Dorval, QC, Canada) was used to count cells in each of four fields per sample. Chlorophyll concentration per cell was calculated by dividing the total chlorophyll by the counted Symbiodinium cells. Total protein concentration and host protein concentrations were analyzed using the Quick Start Bradford Protein Assay Kit (Bio-Rad Laboratories, Hercules, CA, USA). Optical density at 595 nm was read using a spectrophotometer (Ultrospec 2100 pro, GE Bioscience). Owing to the morphology of the studied species (a single polyp in B. europaea and phaceloid in the case of C. caespitose) the number of symbiotic dinoflagellates and protein concentration were normalized per polyp. The differences in chlorophyll fluorometry, protein and chlorophyll concentration between pH treatments were tested by one-way analysis of variance (ANOVA) and post-hoc Tukey's HSD test.

Coral fractionation and DNA extraction

Corals were separated into tissue and skeleton fractions; the tissue was extracted from the coral using an air brush and centrifuged at 9300 g for 15 min. To minimize any cross contamination of tissue in the skeleton fraction, the skeletons were repeatedly cleaned and washed with sterile seawater, before being crushed with a mortar and pestle. For each pH treatment, the replicates of the separated fraction (that is, tissue or skeleton) were pooled together, homogenized and then DNA was extracted using the UltraClean Soil DNA kit according to the manufacturer's guidelines (MoBio, Carlsbad, CA, USA).

PCR amplification of 16S rRNA genes, clone libraries and sequencing

Bacterial primers 8F and 1492R (Lane, 1991) were used to amplify the 16S rRNA gene from the various fractions. 16S rRNA genes were amplified in a 25 μl reaction mixture comprising 2.5 μl of 10 × buffer, 2.5 μl of dNTP mixture (2.5 mM), each primer at 5 μM, 1 ng of template DNA, and 2.5 U of Dream Taq DNA polymerase (Fermentas, International Inc., Burlington, ON, Canada). Amplification conditions for the PCR included an initial denaturation step of 94 °C for 3 min, followed by 29 cycles of 94 °C for 35 s, 56 °C for 35 s, and 72 °C for 1.5 min and a final extension step of 72 °C for 7 min. Reaction products were checked for size and purity on a 1% agarose gel. PCR products from tissue and skeleton samples were ligated into the pCRII-TOPO-TA cloning vector (Invitrogen, Madison, WI, USA). Cloning of the 16S rRNA libraries and sequencing was carried out at The Genome Center (Washington University, St Louis, MO, USA) using the reverse primer 907R (5′-GTCAATTCMTTTGAGTTT-3′), as described by Lane et al. (1985). The nucleotide sequence data for clones reported in this paper were deposited in the GenBank nucleotide sequence database under accession numbers JQ235845–JQ236575.

Phylogenetic and statistical analysis

Operational taxonomic unit (OTU) analysis of the sequences was done using MOTHUR software (http://www.mothur.org) (Schloss et al., 2009), rarefaction and Shannon–Wiener index of diversity were calculated. In addition, representative sequences of the OTUs classified by MOTHUR were compared with the NCBI database (National Center for Biotechnology Information; http://www.ncbi.nlm.nih.gov) and RDP Classifier (Ribosomal database project; http://rdp.cme.msu.edu) using the BLASTn algorithm (Wang et al., 2007). Phylogenetic trees were calculated using the neighbor-joining method and constructed using the ARB phylogenetic program software (Ludwig et al., 2004). The extent of similarity between the clone libraries obtained from each fraction was assessed by Unifrac analysis using Unifrac P-value (Lozupone et al., 2006). The Correlation and regression analyses were formed using STATISTICA (version 7.1) software (Stat Soft Inc., Tulsa, OK, USA).

Symbiodinium identification

Dinoflagellate Symbiodinium were identified using denaturing gradient gel electrophoresis (DGGE) of the internal transcribed spacer-2 region of ribosomal DNA (rDNA). Briefly, the ITS-2 region was amplified using the primers ‘ITSintfor2’ and ‘ITS2clamp’ (LaJeunesse and Trench, 2000) with the following thermal cycling conditions: an initial denaturation step of 94 °C for 3 min followed by 35 cycles of 1 min at 94 °C, 1 min at 55 °C and 1 min at 74 °C. Amplification success was confirmed by gel electrophoresis on a 1.5% agarose gel. Subsequently, products were separated by DGGE (35–75% gradient) using a CBS scientific system and visualized, after SYBR gold staining, using a UV transilluminator. Dominant bands were excised from the gel and reamplified using the same primer set without the GC clamp. The resulting amplicons were purified and sequenced using BigDye Terminator v3.1 cycle sequencing kit and an Applied Biosystems 3130xl Genetic Analyzer (Foster City, CA, USA). Sequences were identified using the NCBI database and the BLASTn algorithm.

Results

Photophysiology of B. europaea and C. caespitosa along the pH gradient

All coral fragments from both species survived collection, recovery, transplantation and experimentation without exhibiting any visual signs of disease. At the end of the experiment, both species at the reference site exhibited similar Fv/Fm values. The only variation observed was that of C. caespitosa in the low pH site (pHT 7.3) that displayed lower Fv/Fm values (one-way ANOVA with post-hoc Tukey's HSD test P<0.001, Figure 1), no significant difference was observed for B. eruopeae (one-way ANOVA P>0.05). Similarly, no significant differences in ETR and NPQ were detected in either specie between the different pH sites (one-way ANOVA P>0.05, data not shown).

Figure 1
figure 1

Maximum dark-adapted Fv/Fm values for C. caespitosa (gray) and B. europaea (white) exposed to a natural pH gradient. The reference site (C) is 300 m from B1. Means were compared by one-way ANOVA, with post-hoc Tukey's HSD test. Different letters represent significant differences (P<0.05). Note that only in C. caespitosa a significant difference was observed.

In C. caespitosa, chlorophyll concentration per symbiodinium cell was higher in the lower pH sites (pHT 7.3 and 7.5) as compared with the higher pH site (pHT 7.8) and the control site (pHT 8.1). Statistically significant differences were detected between lower pH (pHT 7.3) and pHT 7.8 (one-way ANOVA, with post-hoc Tukey's HSD test (P<0.05), Figure 2a). B. europeae did not show such a change (one-way ANOVA, P>0.05), although the overall levels (mean 5.212E-06 ± 1.087E-06) were higher than those obtained for C. caespitosa (1.577E-06 ±2.153E-07).

Figure 2
figure 2

Chlorophyll levels (a) and protein concentrations (b) in C. caespitosa exposed to a natural pH gradient. Protein concentration was calculated per polyp, whereas the chlorophyll concentration was calculated per symbiodinium cell as described in Materials and methods. Data are expressed as means±s.e.m. Means were compared by one-way ANOVA, with post-hoc Tukey's HSD test. Different letters represent significant differences (P<0.05).

Symbiodinium identity and density in B. europaea and C. caespitosa across the CO2 vent

Symbiodiunium analysis revealed that C. caespitosa was found to be dominated by Symbiodinium B2 (that is, there were no other Symbiodinium taxa detectable at the resolution of DGGE, which typically detects symbionts present at 5–10% of the total symbiont community (Thornhill et al., 2006)). In contrast, B. europaea contained Symbiodinium B2 mixed with a novel Symbiodinium that was 95% similar to ‘Mediterranean A’ (Visram et al., 2006; Hunter et al., 2007).

Symbiodinium densities (cells per polyp) did not vary between the sites in both corals (one-way ANOVA, P>0.05). For C. caespitosa the average density was 1.81E+06 ± 2.6E+05 and for B. europaea average was 1.23E+07 ± 6.33E+05. It should be noted that a significant increase in protein concentration (mg per polyp) was observed for C. caespitosa taken from the site at the lowest pH (pHT 7.3) compared with pHT 7.8 (one-way ANOVA, with post-hoc Tukey's HSD test (P<0.05); Figure 2b).

Composition and diversity of bacterial communities associated with B. europaea and C. caespitosa along the pH gradient

To test whether the bacterial community associated with each coral species was influenced by pH, microbial DNA was isolated from the tissues and skeletons of B. europaea and C. caespitosa from each site, and 16S rRNA clone libraries were constructed and sequenced (n=2652). Rarefaction analyses were performed on the clone libraries to determine the number of unique bacterial clones as a proportion of the estimated total diversity within each library. The rarefaction curves (cutoff 97%) did not reach an asymptote (Supplementary Figure S2a, b), indicating the actual total diversity could be much higher. Nevertheless, pattern comparisons revealed significant differences in bacterial community composition between the two corals species (Unifrac P-value, P<0.01 (Figure 3). In general, there was a subdivision according to fraction, tissue versus skeleton, for each species, rather than pH, although these distributions were not found to be statistically significant (UniFrac P-value, P>0.05).

Figure 3
figure 3

Cluster analysis of microbial community 16S rRNA clone libraries generated from the tissues and skeletons of B. europaea and C. caespitosa exposed to four pH conditions. Cluster analysis was performed by cluster environment using unifrac P-value (Unifrac Metric analysis) (Lozupone and Knight, 2005) n=2652. Squares represent the coral B. europaea (B), circles represent C. caespitosa (C), and the fractions are divided into tissue (black) and skeleton (white). *Note that 8.1 (microbes generated from B. europaea tissue exposed to ambient pH) clusters with B. europaea skeleton fraction communities. **Significant differences between the tree clusters. The bar represents a weighted UniFrac distance of 0.1.

The composition and the diversity of the bacterial communities associated with the tissues and skeletons of the two corals at each site were examined in more detail by inspecting the 16S rRNA sequences in the clone libraries (Figures 4 and 5 and Supplementary Table S2–4). The first parameter examined was the Shannon–Wiener index, which measures species diversity (richness and evenness). The correlation between the pH sites and the Shannon–Wiener index values of each fraction was significant only in C. caespitosa skeleton (r2=0.968, P<0.05) for which the Shannon–Wiener values decreased from 4.8 to 3.8 (ambient and lowest pH, respectively) (Supplementary Table S4).

Figure 4
figure 4

The impact of pH on the bacterial groups associated with tissue and skeleton composition and relative abundance of C. aespitosa and B. europaea, as well as the surrounding water. The chart was constructed based on the 16S rRNA gene sequences in clone libraries generated from the corals (top) C. caespitosa (n=1354), (bottom) B. europaea (n=1298) and sea water (n=2029). *Classes under 3%.

Figure 5
figure 5

Changes in bacterial composition and relative abundance of the Bacteroidetes (left) and Gammaproteobacteria (right) groups associated with C. caespitosa (a, b) and B. europaea (c, d) exposed to four pH conditions. The chart was constructed using BLASTn algorithm and RDP Classifier. *Marks significant correlation between the group and pH.

When examining the composition of the bacterial community, 26 class groups were found associated with C. caespitosa. In all pH conditions, the three dominant groups were: Alphaproteobacteria, Bacteroidetes and Gammaproteobacteria (Figure 4 and Supplementary Table S2). Interestingly, a negative correlation was observed between the Alphaproteobacteria and Deltaproteobacteria (r2=0.917, P<0.05) and between Gammaproteobacteria and Actinobacteria (r2=0.924, P<0.05). In Alphaproteobacteria, (most of which belong to Rhodobacterales), the most abundant clones were most similar to coral-associated bacteria FJ952800 and sponge-associated bacteria AM259856 (Supplementary Table S3 and Supplementary Figure S3). Flavobacteriales and Sphingobacteriales were the main groups in the Bacteroidetes, and the abundant clones most closely match marine bacterium GU061250, sponge symbionts EF159908 and plankton bacterium (Supplementary Table S3 and Supplementary Figure S4). Within the Bacteroidetes classes, a negative correlation was observed between Flavobacteria and Sphingobacteria (r2=0.935, P<0.05). The abundance of Flavobacteria decreased (from 54.0% to 37.0%), whereas that of Sphingobacteria increased (from 36% to 52%) (Figure 5a).

Finally in the Gammaproteobacteria, the abundant clone was most similar to Mediterranean sponge bacterium AJ581351 (Supplementary Table S3 and Supplementary Figure S6). Although Gammaproteobacteria was one of the three dominant groups associated with C. caespitosa at each site, the abundance of this group was greater at the sites exposed to the CO2 vent (13.4% of sequences at ambient pH versus 27.2%, 24% and 32.5% at pHT 7.8, 7.5 and 7.3, respectively). The increased numbers of Gammaproteobacteria were mainly in the tissue fraction, although there was also a greater abundance in the skeleton fraction at pHT 7.3 (Supplementary Table S2). Notably, there were changes in composition within the dominant groups. For example, within the Gammaproteobacteria group, the abundance of Thiotrichales was greater at the lowest pH (63.4%), such that it became the major group constituent, whereas Alteromonadales disappeared at the lowest pH. It is important to note that a significant correlation was observed between pH and Chromatiales whose abundance was reduced with decreasing pH (r2=0.954, P<0.05) (Figure 5b).

B. europaea was found to maintain 21 associated bacterial class groups. The four dominant groups were: Alphaproteobacteria, Bacteroidetes, Deltaproteobacteria and Gammaproteobacteria (Figure 4 and Supplementary Table S2). In Alphaproteobacteria, many clones belonged to the Rhizobiales subgroup, and the most abundant clones were most similar to the Oculina patagonica mucus bacterium AY654755 (Supplementary Table S3 and Supplementary Figure S3). A significant correlation between the Alphaproteobacteria group and decreasing pH was observed within the tissue fraction in which the abundance of the Alphaproteobacteria reduced with decreasing pH (r2=0.907, P<0.05) (Supplementary Table S2 and Figure 4). In the Bacteroidetes subgroup, the abundant clones displayed the most similarity to the Paralvinella palmiformis mucus bacterium AJ441239 and planktonbacterium AY167327 (Supplementary Table S3 and Supplementary Figure S4). The Desulfovibrionales and Desulfobacterales were the main groups in the Deltaproteobacteria, and the abundant clones most closely match Desulfovibrio sp. sediment bacteria DQ522111, GQ850564 and Montipora sp. bacterium AB470955 (Supplementary Table S3 and Supplementary Figure S5). The abundance of the Gammaproteobacteria group was lower at the sites exposed to the CO2 vent; at ambient pH it was 32.6%, whereas at pHT 7.8, 7.5 and 7.3 it was 30.4%, 27.2% and 19.9%, respectively (Figure 4 and Supplementary Table S2). At lower pH the abundance of Pseudomonadales and Enterobacteriales subgroups was lower or undetectable, whereas the numbers of Vibrionales increased from 5% to 30%, at pHT 8.1 versus 7.3, respectively (Figure 5d). In the Gammaproteobacteria group, the Alteromonadales and Chromatiales were the most dominant groups (Figure 5d). The abundant clones are most similar to marine bacteria GU061987 and GU061172 and uncultured vibrio sp. AY785255 (Supplementary Table S3 and Supplementary Figure S6). Unlike what was observed for C. caespitosa, only select groups increased in abundance at the sites exposed to the CO2 vent. The best example of this was the Bacteroidetes group, the abundance of which increased from 12.8% (at ambient pH) to 19.2% and 26.3% (at pHT 7.5 and 7.3, respectively). This increased abundance was most evident in the tissue fraction, increasing from 13.5% to 18.2% and 47.8%, at pHT 8.1, 7.5 and 7.3, respectively (Table S2). Similar to what was seen for the Bacteroidetes group associated with C. caespitosa, there were changes in the composition of the Bacteroidetes group associated with B. europaea, specifically, in Flavobacteriales where a pH dependent correlation was found (r2=0.9737, P<0.05). The Flavobacteriales decreased in abundance from 89.2% at pHT 8.1 to 70%, 49% and 22.8% at pHT, 7.8, 7.5 and 7.3, respectively, whereas the Bacteroidia subgroup increased in abundance from 5% at ambient pH to 51.9% at pHT 7.3 (Figure 5c).

Discussion

A predicted global change during the twenty-first century is an increase in CO2 concentration in the atmosphere and oceans, leading to ocean acidification. The consequences of ocean acidification remain debated and more research is needed to better understand the impact of reduced pH on the ocean environment. A volcanic vent near Ischia Island (Italy) provides a unique natural environment with a pH gradient, enabling research into the influence of ocean acidification on marine organisms, including corals. In an earlier study, it was found that the abundance of organisms that deposit calcium carbonate skeletons declined significantly within vent areas (Hall-Spencer et al., 2008). Corals did not occur naturally at the vent site. However, other organisms thrived, such as seagrasses and other macroalgae, which appear to have outcompeted corals in this environment. On the basis of these observations, it has been proposed that ocean acidification can reduce biodiversity and radically alter marine ecosystems (Hall-Spencer et al., 2008).

The present study takes advantage of the same volcanic vents near Ischia Island to further understand the impact of changes in oceanic pH on various aspects of coral biology, including physiology, the type and density of symbiotic dinoflagellates, and the composition of associated bacterial communities. Notably, we observed that reef coral symbioses from the lower pH sites exhibited only minor changes in photophysiological parameters. Although Fv/Fm was consistently lower for C. caespitosa at lower pH compared with control sites (in agreement with the data of Hall-Spencer et al. 2008), this reduction was only significant at the lowest pH site (pH 7.3) (Figure 1). pH had even less impact on the Fv/Fm values for B. europaea, where none of the sites were significantly different from the controls. In addition, there was no detectable impact of pH on the ETR in either species, indicating that the thylakoid membranes were energetically coupled irrespective of pH. Similarly, NPQ was comparable at the four sites. It is possible that as the corals were collected in a low-light-intensity environment (turbulent and turbid waters) photosystem II was not stressed by high light intensities, and therefore no chronic photo-damage can be detected in any of the pH environments. Therefore, this study found that overall changes in ocean pH did not interfere greatly with the photosynthetic activity of these reef corals. These findings corroborate other studies reporting no changes in photosynthesis under ocean acidification conditions (Reynaued et al., 2003; Schneider and Erez, 2006) or only slightly altered cytoplasmic pH values in response to external pH shifts of similar magnitude (Raven and Smith, 1980; Gimmler, 2000).

A higher protein concentration per polyp was noted for C. caespitosa at the lowest pH site (Figure 2b). Previous studies have shown an increase in host protein at reduced pH (Fine and Tchernov, 2007; Anthony et al., 2008; Krief et al., 2010). This may indicate a thicker tissue, which can serve as a barrier between external and internal environments, enabling calcification despite more acidic water conditions and protecting the symbiotic dinoflagellates in the coral gastrodermis. Several recent studies have also reported thicker tissue under lower pH conditions (Anthony et al., 2008; Krief et al., 2010; Fabricius et al., 2011). Furthermore, recent boron measurements on the coral C. caespitosa held in aquaria at pHT 7.8 and from CO2 vents (Trotter et al., 2011) showed that this species can adjust its internal pH allowing calcification in undersaturated seawater.

Examining the zooxanthellae revealed that both coral species contained Symbiodinium B2, but only B. europaea contained Symbiodinium clade A. No change in Symbiodinium diversity was observed following the 7-month exposure to reduced pH. It has been proposed that tolerance to various environmental stresses, such as light or temperature, is influenced by Symbiodinium type (Baker, 2003, 2004; Knowlton and Rohwer, 2003; Rodolfo-Metalpa et al., 2006; Jones et al., 2008). Indeed, clade A has been shown to impart resistance to short-term increases in temperature under experimental conditions to its host (Rodolfo-Metalpa et al., 2006).

Although the two coral species were exposed to similar conditions, the bacterial communities associated with each species were significantly different (Figure 3), demonstrating specific coral-bacterial associations, as reported previously (Frias-Lopez et al., 2002; Rohwer et al., 2002; Ritchie and Smith, 2004; Rohwer and Kelly, 2004; Bourne and Munn, 2005). When examining the microbial community within each coral species some changes in bacterial group composition were observed at the various pH treatments. For example, a subdivision of bacterial groups for each coral can be seen according to fraction, tissue versus skeleton (Figure 3). However, we did not find significant difference between the bacterial cluster and the pH treatments. In addition, in most cases, no correlation was observed between the specific bacterial classes and the pH change. This may indicate that under these tested conditions pH is not a major driving force in determining the bacterial composition and the diversity. Our data, generated from observing corals in a natural pH gradient, contrasts with results from previous studies performed in the laboratory, which indicated a dominant pH-dependent change in the bacterial community that clustered according to pH treatment and not according to coral fraction (Meron et al., 2011). It should be emphasized that the corals examined were different in these studies and this may also determine the difference in response.

In our prior laboratory experiments (carried out with a Red Sea coral species) we found that decreased pH resulted in an increase in bacteria linked to coral disease and stress, but not in any external signs of coral disease (Meron et al., 2011). Similarly, in the present field study, the corals did not show any signs of disease, but in contrast to the laboratory study, there was also no significant increase in specific pathogens or bacterial communities associated with diseased or stressed corals. For example, Rhodobacteraceae (Alphaproteobacteria), a bacterial family isolated from diseased, injured or stressed marine invertebrates (Sekar et al., 2006; Sunagawa et al., 2009) that increased in abundance in the laboratory experiments (Meron et al., 2011), did not increase in abundance at lower pH in the present study, and in fact decreased in B. europaea (Figure 4 and Supplementary Table S2). Similarly, Alteromonadaceae and Vibrionaceae (of Gammaproteobacteria), which are often associated with diseased and stressed corals, increased in abundance at lower pH in the laboratory experiments (Meron et al., 2011), but were absent or did not change in abundance in the present study (Figure 5b and d). In the case of Vibrionaceae associated with B. europaea, an increase in abundance was observed at lower pH, but these bacteria represented only 6% of the total microbial community at pHT 7.3 and no homologs of known pathogens were detected (Figure 5d). Another bacterial group detected in healthy corals (Sekar et al., 2008) but also linked with stress conditions, including pH (Vega-Thurber et al., 2009) and with black band disease are the Bacteroidetes (Cooney et al., 2002; Frias-Lopez et al., 2002, 2004; Barneah et al., 2007). As with the Vibrionaceae, the present study found no significant correlation in the abundance of Bacteroidetes associated with C. caespitosa with the pH values, although the proportion of Bacteroidetes in the microbial population associated with B. europaea increased from 12.8% at pHT 8.1% to 26.3% at pHT 7.3; this increase again was mainly evident in the tissue fraction (Figure 4 and Supplementary Table S2). It is interesting to note that a significant correlation was observed in B. europaea between Flavobacteriales group (belonging to Bacteroidetes) and reduction in pH (Figure 5c).

The diverse microbial communities associated with the corals may have an important role in the minor pH impact we observed. Kitano and Oda (2006) described a model showing how a host capable of accommodating highly diverse bacterial flora is more likely to survive fluctuating environmental conditions, such as changes in foods and pathogens. A corollary of this model is that the observed dynamic changes in the bacterial communities of the coral holobiont do not necessarily reflect stress or sensitivity but rather enable tolerance, acclimatization and survival, despite variable biotic and abiotic parameters. Further research is required to corroborate and better understand how, and under which circumstances, different coral symbionts can increase tolerance to specific disturbances.

Although in the present study the corals were exposed to a wide range of pH values some of which are much lower than what is expected in the next century, our findings suggests that pH most likely has relatively little impact on coral microbial ecology and coral photophysiology, compared with other environmental stresses, such as increased temperature (Kuhl et al., 1995; Furla et al., 1998, 2000; Hoegh-Guldberg, 1999; Al-Horani et al., 2003). This finding could be explained by the fact that corals are naturally exposed to variable pH conditions (Kuhl et al., 1995; Furla et al., 1998, 2000; Al-Horani et al., 2003). Monitoring of reef waters over time shows that several parameters, including pCO2, pH, total alkalinity, total inorganic carbon and dissolved oxygen, change diurnally due to photosynthesis and calcification during the day and respiration at night (Kayanne et al., 2000). In fact, the differences in Ωaragonite and pH values that occur within the coelenteron in light versus dark conditions are much greater than those predicted to occur due to ocean acidification within this century (Furla et al., 1998, 2000; Al-Horani et al., 2003; Venn et al., 2011). Indeed Furla et al. (1998) and Kuhl et al. (1995) showed that the coelenteric pH fluctuates from 7.4 at night to 8.6–8.9 during the day.

A key advantage of studying the impact of pH on corals in natural habitats is the large pool of microbial species in the ocean. Laboratory environments cannot mimic the dynamism and microbial diversity present in nature. Moreover, it is possible that aquarium conditions themselves contribute to stress or disturbance in the microbial community. Kooperman et al. (2007) demonstrated that the same coral species has different associated microbial communities in the laboratory compared with field conditions . Despite this, laboratory experiments are critical tools in providing a basic understanding of how pH impacts marine organisms. Naturally occurring CO2 vent sites provide important complementary systems that can serve as good models for examining the impact of changes in ocean pH in the natural environment. The current study, which examined corals maintained in a natural pH gradient, suggests that, at least for these two coral species, reduced pH does not seem to significantly reduce coral health. Fabricius et al. (2011) recently examined coral diversity in a natural pH gradient off the coast of Papua New Guinea. The authors report a decrease in coral diversity, although the existing coral species were found in good health suggesting acclimation to the pH environment. No doubt future studies will continue to take advantage of these unique natural environments to explore further the influence of reduced pH on marine ecosystems.