We sought to determine whether the Indo-Pacific reef-building coral Seriatopora hystrix performs in a similar manner in the laboratory as it does in situ by measuring Symbiodinium density, chlorophyll a (chl-a) concentration, and the maximum quantum yield of photosystem II ( ) at the time of field sampling (in situ), as well as after three weeks of acclimation and one week of experimentation (ex situ). Symbiodinium density was similar between corals of the two study sites, Houbihu (an upwelling reef) and Houwan (a nonupwelling reef), and also remained at similar levels ex situ as in situ. On the other hand, both areal and cell-specific chl-a concentrations approximately doubled ex situ relative to in situ, an increase that may be due to having employed a light regime that differed from that experienced by these corals on the reefs of southern Taiwan from which they were collected. As this change in Symbiodinium chl-a content was documented in corals of both sites, the experiment itself was not biased by this difference. Furthermore, increased by only 1% ex situ relative to in situ, indicating that the corals maintained a similar level of photosynthetic performance as displayed in situ even after one month in captivity.

1. Introduction

Molecular biology promises to yield insight into the subcellular mechanisms underlying the stable mutualism between reef-building scleractinians and dinoflagellates of the genus Symbiodinium [1, 2], as well as their macromolecular responses to changes in their environment [35]. The latter topic is of particular interest given that global climate change (GCC)-driven temperature and pCO2 increases have been hypothesized to lead to more frequent coral bleaching events in the years to come [6]. Alongside other anthropogenic pressures, such GCC-derived threats have generated an urgent need to shift the monitoring of coral reef health from a retroactive process to a proactive one [7]. Assessment of reef health is currently conducted by visual surveys in which the number of dying or dead corals is quantified (e.g., [8]). However, such late-stage manifestations of health decline likely occurred well after the initial insult. An analysis of the expression or activity of subcellular biomarkers, such as stress genes and proteins, may allow for the determination of which corals are at risk from anthropogenic impacts on a proactive timescale. Such a monitoring approach could potentially allow for scientists and managers to work together to mitigate local-scale insults to reef stability, such as water pollution [9], prior to extensive loss of coral.

In order to validate the efficacy of certain gene and protein-level biomarkers for proactive coral health assessment, their behavior in response to environmental stress must first be determined in the laboratory. Unfortunately, little attention has been given to ensure that laboratory-borne results do not carry experimental artifacts emerging from the transplantation of corals out of the field and into aquaria. For instance, corals are routinely fragmented with chisels or pneumatic drills, removed from the ocean, transported to a laboratory, and then used in experiments within several hours or days (e.g., [1012]). Their physiological responses to the experimental treatment could therefore be masked by their recovery response to fragmentation and subsequent incubation in an environment that may differ greatly from the reef from which they were sampled.

If molecular biomarkers are to ultimately gauge the health of corals on a proactive timescale, not only must laboratory-reared corals be shown to behave in a similar manner as in situ, but the expression and/or activity of these macromolecules must also be shown to either correlate with or predict a certain phenotype. To date, there have been few attempts to document correlations between a particular physiological response and the expression of the underlying genes and proteins. Putnam et al. [13] found no correlation between ribulose-1,5-bisphosphate carboxylase/oxygenase large subunit (rbcL) gene and RBCL protein expression in larvae of the scleractinian coral Pocillopora damicornis, in contrast to what they had hypothesized. However, Putnam et al. [13] did not attempt to draw inferences from correlations between physiological parameters, such as Symbiodinium density, and subcellular response variables, such as gene expression. Such covariation analyses could ultimately allow for the development of molecular or cellular biomarkers (sensu [14]) that could be used for reef coral health assessment provided that strong correlations are documented between indices of physiological performance, such as growth rate, and expression of certain macromolecules.

In order to address these two deficiencies in the coral biology field, specimens of the Indo-Pacific reef-building coral Seriatopora hystrix were collected from two reefs nearby Taiwan’s National Museum of Marine Biology and Aquarium (NMMBA), and their physiology was assessed both in situ and after a previously published experiment [4] to determine whether this coral demonstrates a similar physiology in the laboratory (i.e., ex situ) as it does in the field (i.e., in situ). Then, a previously published dataset was explored again in order to document the degree of correlation between four physiological parameters: growth, Symbiodinium density, chlorophyll a (chl-a) concentration (both areal and cell specific), and the maximum quantum yield of photosystem II ( ) [4], and the expression of four Symbiodinium genes (Table 1). It was hypothesized that expression of Symbiodinium photosynthesis genes, such as rbcL, photosystem I (psI, subunit III), and phosphoglycolate phosphatase (pgpase), would be associated with increases in , which is often used as a proxy for photosynthetic performance in Symbiodinium populations within other reef-building corals [15]. The ultimate goal of this work was to begin to develop a standard operating procedure for manipulative experiments with molecular endpoints with S. hystrix, which has recently emerged as a model for understanding the impacts of GCC on coral physiology [3, 14, 16].

2. Materials and Methods

2.1. Coral Collection, Manipulative Experiment, and Physiological Parameter Analysis

A reciprocal transplant was previously conducted in the laboratory whereby S. hystrix specimens from Houbihu, a reef characterized by extensive upwelling [17], were exposed to either a fluctuating temperature treatment (23–29°C over a 5 hr period) or a stable one (26°C) for 7 d [4]. Conspecifics from Houwan (Figure 1(a)), a nonupwelling reef, were simultaneously exposed to the same two treatments, and sampling was conducted at the end of the experiment only. Coral nubbins ( 2 g) from both sites of origin were acclimated at constant seawater conditions (described in [4]) for three weeks prior to the experiment to allow for recovery from the fragmentation, transplantation, and husbandry processes. However, from analysis of the published results alone [4], it was unclear whether the experimental nubbins performed in a similar manner as in situ after four weeks of husbandry.

On the evening prior to collecting the six S. hystrix colonies from their respective field sites, a diving pulse amplitude modulating (PAM) fluorometer (Walz, Germany) was used to calculate the maximum dark-adapted of each sampled colony at 18 : 30 as described in [5]. It was hypothesized that corals of the two sites would demonstrate similar values in situ. Furthermore, although the ex situ values of nontransplanted corals (see [4] for details) were hypothesized to be similar to those documented in situ, the ex situ values of transplanted corals were hypothesized to be lower than those documented in situ.

The following day, six colonies (Figure 1(b)) from each site were removed from the ocean, and three 50 mg biopsies were fragmented with pliers from each ( biopsies per site) prior to transportation of the colonies to the laboratory (a 2 km distance away). One biopsy was submerged in 500 μL of TRIzol (Life Technologies, Grand Island, NY, USA), while the other two were immersed in 1-2 mL of RNALater (Ambion, Austin, TX, USA). Coral biopsies within one of the two tubes containing RNALater were stored at 4°C for one month. One hour after returning to the laboratory, both the coral samples and RNALater in the other six tubes were decanted into a mortar and homogenized with a pestle for several minutes. The RNALater-coral tissue slurry was then frozen at −20°C for one month.

We have previously found that RNALater does not preserve the integrity of coral RNA to an extent that is suitable for downstream analyses, such as real-time PCR. It was hypothesized that this poor capacity for fixation could be due to the inability of the salts within the RNALater solution to penetrate the coral tissue-skeleton interface. Therefore, homogenization in RNALater could potentially lead to better protection of RNA against ribonucleases, as this would allow for the fixative to come into contact with and permeate a greater proportion of both the coral and Symbiodinium cells. If this method led to RNA of similar quantity and quality to that of samples stored in TRIzol, this would represent a preferable means for preserving corals sampled in situ since it would not require the transport of liquid nitrogen or toxic chemicals (i.e., TRIzol) into the field. Only the 18 biopsies generated from the six coral colonies collected from Houwan were processed for this fixation strategy comparison.

Upon arriving at NMMBA, branches ( 2 g) were removed from each of the 12 colonies (Figure 1(c)), wrapped in aluminum foil, and frozen at −20°C. The tissue was removed from the branches as in [4] the next day, and Symbiodinium density and chl-a concentration were determined as in [18]. Surface area (SA) was calculated with the wax dipping method developed by [19]; 12 wooden dowels of a range of known dimensions and SAs were created and dipped twice in vats containing molten paraffin wax in order to generate a standard curve of SA versus the difference in wax mass accumulated between the 1st and 2nd wax dips. The wax mass differences of the branches were then compared against the standard curve to calculate SA.

Both Symbiodinium density and chl-a content were normalized to SA and reported as cells and μg  , respectively. The areal chl-a concentration was divided by the Symbiodinium density of the same sample to calculate the cell-specific chl-a concentration, which was reported as pg . The in situ values for these four physiological response variables (including ) were compared to those obtained after a three-week acclimation period and a one-week experimental period [4], and it was hypothesized that there might be increases in chl-a concentration (both areal and per cell) due to the use of stable, rather than fluctuating (as occurs in situ), photosynthetically active radiation (PAR) levels during the experiment itself.

During the three-week acclimation period, the experimentally fragmented coral nubbins were exposed to shaded, natural light ( 90 μmol photons ). Artificial lights were utilized and set to a 12 hr light ( 90 μmol photons )-12 hr dark cycle during the one-week experiment that followed [4], meaning that the light profile did not vary over the daytime portion of the diel cycle during the experiment, as it would in situ (see Figure 1 of [20]). Despite this utilization of a stable light regime characterized by the same average hourly PAR level that these corals experience in situ [4], corals from each of the two sites were expected to be affected similarly by this transplant from a fluctuating to a stable light regime.

2.2. RNA, DNA, and Protein Extractions

RNA, DNA, and protein were extracted in sequential fashion from the 18 Houwan samples stored in either TRIzol or RNALater (either uncrushed and stored at 4°C or homogenized and stored at −20°C). For the six samples already stored in 500 μL of TRIzol, an additional 1 mL of TRIzol was added to the samples, and the tissues were homogenized in the 1.5 mL of TRIzol with a mortar and pestle. The TRIzol-tissue mixture (1 mL) was then transferred to a new 1.5-mL microcentrifuge tube. For the 12 samples stored in RNALater, the tubes were pulse spun for 1 s to bring the coral skeletons to the bottom of the tubes, and the RNALater was decanted. The six samples that had not been previously homogenized were ground into a powder in 1.5 mL of TRIzol with a mortar and pestle, and 1 mL of TRIzol-tissue slurry homogenate was transferred to a new 1.5-mL microcentrifuge tube. The previously homogenized samples ( ) were vigorously vortexed with 1 mL of TRIzol for several seconds after having decanted the RNALater. For all 18 samples, tissues in TRIzol were pulse spun for 1 s to sediment the residual skeleton, and approximately 1 mL of TRIzol-tissue mixture was transferred to a new 1.5-mL microcentrifuge tube.

RNA and DNA were then extracted as in [13] from all the 18 samples. RNAs were precipitated with a high salt solution (250 μL) and 250 μL of isopropanol at room temperature for 10 min as in [21], and precipitated pellets were further purified with a commercial spin column kit (Plant RNA Miniprep kit, Hopegen Biotechnology, Inc., Taipei, Taiwan) as in [13]. DNAs were separated from the protein phase as in [22], and the precipitated DNAs were further purified with the AxyPrep PCR clean-up kit (Axygen Biosciences, Union City, CA, USA) as directed by the manufacturer. RNA and DNA quantity and quality were assessed as described in [23] after eluting into DEPC-treated water (30 μL) and manufacturer’s eluent (30 μL), respectively. Proteins were extracted from the organic phase of the samples fixed in TRIzol and RNALater (4°C only) as described in [3], and their quantity and quality were assessed as described therein.

2.3. Statistical Analyses

All statistics were calculated with JMP (ver. 5.0, SAS Institute Inc., Cary, NC, USA). For in situ physiological parameter comparisons between the two sites, Student’s -tests were used when the data were normally distributed and of homogenous variance (determined by Shapiro-Wilk tests and Levine’s tests, resp.). When log or square root transformations did not generate datasets suitable for parametric analyses, either the Mann-Whitney median test or the Wilcoxon rank-sum test was used instead. To compare the effects of sampling time (in situ versus ex situ) across the two temperature regimes (stable versus variable) and two sites (Houbihu versus Houwan), a two-way, repeated-measures ANOVA was used to test for the effects of site, temperature, their interaction, tank nested within temperature, site time, temperature time, and time. When the tank term was not statistically significant ( ), it was dropped from the model. One-way ANOVAs were used to assess the impact of fixation/homogenization strategy on [RNA], [DNA], and [protein], as well as the respective 260/280 and 260/230 ratios for both RNA and DNA. Tukey’s honestly significant difference (HSD) tests were used to determine individual mean differences when the model detected an overall treatment effect ( ). In all instances, error terms presented later and in figures represent standard error of the mean (SEM).

Expression of four Symbiodinium genes (Table 1) was measured previously in the same samples from which the physiological data were obtained after seven days of exposure to either the stable or variable temperature regime [4]. In order to determine whether expression of any of these genes correlated significantly with any of the four physiological response variables or amongst each other, analysis of covariance (ANCOVA) was conducted after identifying the four most significant correlations with JMP’s multivariate correlation analysis program. After plotting the global trend lines across all data points, those of the individual temperature regimes within each of the two sites were plotted and assessed against each other, and slopes were considered to differ significantly at . It was hypothesized that Symbiodinium density and chl-a content might demonstrate a negative relationship due to the fact that corals with higher densities of Symbiodinium would potentially maintain lower chl-a content in response to self-shading by other Symbiodinium. Regarding the comparisons between physiological and molecular parameters, it was hypothesized that expression of the Symbiodinium photosynthesis genes may display strong degrees of correlation with chl-a content and . Furthermore, high expression of photosynthesis genes might be associated with an increase in oxygen production in vivo, which would potentially necessitate the translation of higher concentrations of reactive oxygen species (ROS)-detoxifying proteins, such as APX1. Expression of the photosynthesis genes—psI, rbcL, and pgpase—was therefore hypothesized to correlate positively with apx1 gene expression.

3. Results

3.1. Physiological Parameters

Symbiodinium density in situ (Figure 2(a)) did not differ significantly between the two sites (Wilcoxon rank-sum test, , ) and was approximately and cells for the corals from Houbihu and Houwan, respectively ( ). There was no temporal change in Symbiodinium densities between the values measured in situ (Figures 2(b) and 2(c)) and those documented ex situ (Table 2).

Areal chl-a concentration (Figure 2(d)) increased from 3 to 6 μg  between the in situ and ex situ sampling times (Table 2), and this increase was statistically significant for corals of both Houbihu (Figure 2(e)) and Houwan (Figure 2(f)). When normalized per cell (Figure 2(g)), chl-a concentration also increased significantly ( 2-fold; Table 2) from the in situ sampling time (Figure 2(g)) to the ex situ sampling time in Symbiodinium within corals of both Houbihu (Figure 2(h)) and Houwan (Figure 2(i)). A significant 1% increase in (Figure 2(j)) was observed between the in situ and ex situ sampling times (Table 2), and this effect of time was similar for Symbiodinium within corals from Houbihu (Figure 2(k)) and Houwan (Figure 2(l)). There were no significant interaction effects of site and sampling time (Table 2) on areal chl-a concentration, cell-specific chl-a concentration, or , indicating that the fragmentation, transplantation, and husbandry processes did not affect corals from one site more than the other.

3.2. Fixation Strategy Comparisons

There was a statistically significant effect of tissue storage/homogenization strategy on [RNA] (Figure 3(a); one-way ANOVA, , ). Although there were no post hoc differences, it does appear that the difference detected by the model is due to the higher [RNA] generated by samples fixed in TRIzol (mean =  ng ). When looking at the quality of the RNA with respect to protein contamination (i.e., the 260/280 ratio; Figure 3(a)), there was also a statistically significant effect of fixation strategy (Wilcoxon rank-sum test, , ), and this appears to be driven by a significantly lower ratio in samples homogenized in RNALater (mean ); however, there were no post hoc differences. The 260/230 ratio (Figure 3(a)), which reflects the degree of contamination of the purified RNAs with phenol or alcohol, did not vary across the three fixation strategies (one-way ANOVA, , ).

There was also a statistically significant effect of fixation strategy on [DNA] (Figure 3(b); Wilcoxon rank-sum test, , ) due to the approximately 5-fold higher [DNA] (mean =  ng ) emerging from samples homogenized in RNALater relative to those fixed in TRIzol and RNALater (without homogenization). The 260/280 ratios (Figure 3(b)) were similar between the three fixation strategies (one-way ANOVA, , ) and averaged across all 18 samples. The 260/230 ratio was significantly different across the three fixation techniques (Wilcoxon rank-sum test, , ) due to this parameter being significantly higher in samples homogenized in RNALater (mean ). This signifies that the DNAs from samples homogenized in RNALater had relatively less organic solvent contamination. Whether this increase in purity can be attributed more to the homogenization step itself, or merely the difference in storage temperatures (4 versus 20°C), remains to be determined.

Despite notably higher protein concentrations (Figure 3(c)) from samples fixed in TRIzol (mean  ng ) relative to those fixed in RNALater (nonhomogenized, mean  ng ), this difference was not statistically significant (Mann-Whitney median test, , ). Given that the total protein values (Figure 3(c)) were derived directly from the protein concentrations, the statistical test results were identical. This signifies that total protein yields were similar between the two fixation strategies.

3.3. Correlations

The four most statistically significant correlations within the ex situ dataset were found to be between cell-specific chl-a concentration and Symbiodinium density (Figure 4(a)), Symbiodinium psI mRNA expression and (Figure 4(b)), psI and pgpase mRNA expression (Figure 4(c)), and apx1 and pgpase mRNA expression (Figure 4(d)). Regarding the former, there was a statistically significant (linear regression -test, , ), negative correlation between these two parameters. There was also a statistically significant ( , linear regression -test, , ), negative association between psI mRNA expression and (Figure 4(b)). ANCOVA revealed a significant difference between the two field sites within the stable temperature treatment ( , ) for the latter association. There was a slight, positive relationship between these two parameters in Houwan samples exposed to a stable temperature regime, whereas there was a slight negative correlation between them in Houbihu samples exposed to a stable temperature regime.

A statistically significant ( , linear regression -test, , ), positive correlation was observed between Symbiodinium pgpase and psI mRNA expression (Figure 4(c)). For the samples from Houwan exposed to a stable temperature regime, the slope was significantly lower than that of all other interaction groups, and, specifically, was significantly lower than the slope of the regression line corresponding to data from coral samples from Houwan exposed to a variable temperature regime (ANCOVA, , ). Finally, there was a statistically significant ( , linear regression -test, , ), positive association between Symbiodinium pgpase and apx1 gene expression (Figure 4(d)).

4. Discussion

Previous researchers have rarely ensured that their experimental corals behaved in a similar fashion ex situ as in situ. In fact, it is likely that both the transportation and fragmentation processes are stressful to corals, and, for that reason, a three-week recovery period was utilized herein in order to provide specimens with sufficient time to heal and acclimate to the laboratory environment. Tissues of all nubbins overgrew the fishing lines used for suspension prior to the initiation of the experiment and some parameters, such as the Symbiodinium density, did not change between the time of collection (i.e., in situ) and the termination of the experiment (i.e., ex situ). However, chl-a cell−1 approximately doubled over the four weeks of husbandry, and the light regime utilized in the experimental aquaria may have accounted for this difference.

The use of subsaturating PAR levels (e.g., 80–100 μmol  ) has become a hallmark of manipulative experiments in the coral biology field (e.g., [24]), possibly because older studies (e.g., [25]) found that corals bleached in captivity when in situ PAR levels were employed. The 90 μmol  PAR level employed herein was chosen because this is the approximate level this coral experiences in situ in the reefs of southern Taiwan from which the experimental colonies were collected [4]. That being said, S. hystrix populations in southern Taiwan are routinely exposed to saturating PAR levels (e.g., 400 μmol ) for at least several hours each day (Mayfield, unpublished findings), and these higher PAR levels may partially determine the standing chl-a concentrations within the Symbiodinium cells. The reason for this is because Symbiodinium, as well as all other photosynthetic organisms, must establish chl-a levels that are associated with a degree of electron harvesting which results in a balance between high levels of carbon fixation and an acceptable amount of ROS production [26]. While the same average hourly PAR was experienced by the corals ex situ as in situ, the lower diel maximum PAR experienced by the corals ex situ may have contributed to the ability for their Symbiodinium populations to accumulate high cellular chl-a levels without a contingent increase in risk for ROS production. Although low expression levels of the Symbiodinium oxidative stress marker apx1 were documented in these corals, direct measurements of ROS levels are needed to confirm this relationship between chl-a levels and ROS production in Symbiodinium.

Despite the documented ex situ increase in both areal chl-a, as well as chl-a , relative to in situ values, corals from Houbihu and Houwan responded similarly in terms of this increase in Symbiodinium chl-a content. Furthermore, dark-adapted values measured after three weeks of acclimation and one week of experimentation were similar to those measured on the evening before field sampling. This suggests that, although chl-a cell−1 doubled, the efficiency of the photosystem II complexes of the Symbiodinium populations was similar to field levels. As a final comment on chl-a effects, Symbiodinium populations tended to have higher chl-a concentrations when they were present at lower densities (Figure 4(a)). This tendency may have allowed corals with lower Symbiodinium densities to photosynthesize at a similar level as those with higher densities and could be related to a decrease in self-shading of Symbiodinium in these coral tissues.

Our small-scale correlation analysis identified other sets of parameters whose degree of correlation did not conform to our expectations. This is evident in the negative association between Symbiodinium psI gene expression and (Figure 4(b)). Whereas Symbiodinium with more photosystem mRNA transcripts present in their chloroplasts would appear to have a higher capacity for photosystem protein expression, and thereby electron capture, such does not appear to be the case in the Symbiodinium populations within the corals sampled herein. This could be because the respective protein is regulated posttranscriptionally. Putnam et al. [13] found a negative relationship between rbcL gene and RBCL protein expression, and so it is possible that a negative relationship between psI gene and PSI protein expression also exists. The extent to which Symbiodinium proteins involved in photosynthesis are regulated posttranscriptionally would therefore represent a promising question for future study.

In corals exposed to the variable temperature treatment only, a positive relationship was observed between expression of two Symbiodinium photosynthesis genes, pgpase and psI. Corals exposed to a variable temperature regime could be hypothesized to necessitate a more concerted regulation of photosynthesis gene expression in order to quickly modify expression levels of the respective proteins in response to rapid and dramatic changes in temperature. This hypothesis is based on the idea that, all else being equal, the rate of translation of a protein will be higher if the intracellular concentration of its respective mRNA is also higher. If Symbiodinium populations express high levels of both genes, then they may be able to more quickly adjust expression of the respective proteins during the acclimation response to abrupt temperature changes, such as those associated with upwelling events. This may explain why Symbiodinium populations in coral samples of the variable temperature treatment that expressed high levels of pgpase were also likely to express high levels of psI. This hypothesis could also account for the significant correlation between expression of Symbiodinium pgpase and apx1 in samples from Houbihu exposed to a variable temperature regime, though this correlation might simply be due to the increase in ROS generated at times at which photosynthesis is occurring at high levels [4, 14]. Future efforts should seek to verify whether changes in expression of these photosynthesis genes, as well as their respective proteins, actually drive increases in carbon fixation in order to validate their capacity to serve as molecular biomarkers of the photosynthetic performance of Symbiodinium.

5. Conclusions

As a potential acclimation response to a light regime that differed from the in situ condition, the Symbiodinium populations within the S. hystrix specimens studied herein were found to undergo a doubling of their chl-a concentration after four weeks of husbandry. Although the results of a previously published work [4] are not discredited by these results given that these concentration changes were similar in corals of both study sites, we nevertheless recommend that researchers track the recovery of corals transported to the laboratory on a more finely tuned timescale, such as within several hours of their fragmentation and consequent incubation in laboratory aquaria, in order to demonstrate that the experimental samples have recovered and acclimated to their ex situ environment. A conservative approach was taken herein by utilizing a lengthy acclimation time. However, too extensive an acclimation period could begin to influence the physiology of corals in a manner that causes them to perform differently ex situ compared to in situ. Although corals herein appeared to be physiologically competent after four weeks of husbandry, the fact that their chl-a content doubled demonstrates the sensitivity of S. hystrix to changes in its environment. Indeed, this coral is amongst the most environmentally sensitive species studied to date [27] and may require a longer acclimation time than more robust species, such as massive poritids.


The authors would like to thank Drs. Peter Edmunds and Hollie Putnam for fruitful discussions on experimental design and interpretation of the data. Thanks are also given to Pei-Hsun Chan for assistance with the laboratory work. ABM was funded by an international postdoctoral research fellowship (OISE-0852960) from the National Science Foundation of the United States of America, and the laboratory analyses were funded by the PADI Foundation, PADI Project Aware, The Journal of Experimental Biology, and the Khaled bin Sultan Living Oceans Foundation.