Potential local adaptation of corals at acidified and warmed Nikko Bay, Palau

Ocean warming and acidification caused by increases of atmospheric carbon dioxide are now thought to be major threats to coral reefs on a global scale. Here we evaluated the environmental conditions and benthic community structures in semi-closed Nikko Bay at the inner reef area in Palau, which has high pCO2 and seawater temperature conditions with high zooxanthellate coral coverage. Nikko Bay is a highly sheltered system with organisms showing low connectivity with surrounding environments, making this bay a unique site for evaluating adaptation and acclimatization responses of organisms to warmed and acidified environments. Seawater pCO2/Ωarag showed strong gradation ranging from 380 to 982 µatm (Ωarag: 1.79–3.66), and benthic coverage, including soft corals and turf algae, changed along with Ωarag while hard coral coverage did not change. In contrast to previous studies, net calcification was maintained in Nikko Bay even under very low mean Ωarag (2.44). Reciprocal transplantation of the dominant coral Porites cylindrica showed that the calcification rate of corals from Nikko Bay did not change when transplanted to a reference site, while calcification of reference site corals decreased when transplanted to Nikko Bay. Corals transplanted out of their origin sites also showed the highest interactive respiration (R) and lower gross photosynthesis (Pg) to respiration (Pg:R), indicating higher energy acquirement of corals at their origin site. The results of this study give important insights about the potential local acclimatization and adaptation capacity of corals to different environmental conditions including pCO2 and temperature.

Increases in atmospheric carbon dioxide (CO 2 ) simultaneously cause both ocean warming and acidification, which are now thought to be the major threats to coral reefs on a global scale 1,2 . Coral mass bleaching events are now occurring more frequently compared to the past, and ocean warming is predicted to further increase the susceptibility of corals to bleaching in the coming decades 3,4 . Some studies have indicated that zooxanthellate corals may have the ability to adapt to high temperature environments, such as by shuffling to Symbiodiniaceae types that have higher tolerances to warmer temperature [5][6][7] . In addition to global warming, increases of seawater partial pressure of CO 2 (pCO 2 ) and decreases of aragonite calcium carbonate saturation (Ω arag ) have been reported to decrease calcification rates of most reef calcifiers including corals and crustose coralline algae (CCA) 8,9 . Meanwhile, it has been shown that the tolerance of organisms to high pCO 2 can differ among species and even within species 10 . Hence, there is now wide interest in understanding how reef organisms will respond to ocean warming and acidification at the community level, and to examine if organisms are able to acclimatize or adapt to these environmental changes.
Here we investigated a semi-closed bay (Nikko Bay) in the inner reef area in Palau ( Fig. 1) with high pCO 2 and high-temperature conditions, and counter-intuitively, high coral coverage 11,12 . The elevated pCO 2 observed in this bay has been suggested to be mainly due to natural biological activity and seawater circulation 12,13 . Nikko Bay's corals have shown little evidence of bleaching during the 1998 mass bleaching on other reefs of Palau and during the 2010 thermal stress event 11,14 . A global mass bleaching event occurred in 2016-2017 3 , however, it   Table S2) and at other coral reefs 30 . This contradicts with most conditions at CO 2 vents where the seawater pH is highly variable temporally [18][19][20] . Average Chl-a and nutrient concentration values inside Nikko Bay were slightly but significantly higher than those outside the bay (Wilcoxon-test, p < 0.01, Fig. 2f-g, Table S1). Daytime average total alkalinity (TA) and dissolved inorganic carbon (DIC) were significantly lower within the bay compared to outside the bay (Wilcoxon-test, p < 0.01, Fig. 2c, d, Table S2) and the TA-DIC diagram indicated that the low pH and high pCO 2 within Nikko Bay were mainly caused by low seawater TA due to calcification and by high DIC due to respiration (Fig. 3). By using the calculated mean water residence time within the bay (71 days 12 ), mean net calcification (Gn) and net primary production (Pn) rates within the bay were calculated to be 22.7 mmol CaCO 3 m −2 d −1 and − 6.9 mmol C m −2 d −1 . These values were lower than the net calcification rates found at most reefs 31 , however the positive net calcification at seawater Ω arag of 2.44 within the bay contradicts with previous studies suggesting that coral reef formation is restricted to seawater Ω arag higher than ca. 2.8 (pCO 2 lower than 560 μatm) 32 , and also with CO 2 seep studies indicating that reef development ceases where pH is lower than 7.7 (Ω arag 2.1) in Papua New Guinea 18 , and lower than 7.9 in the Mariana Islands 20 . When CO 2 gas exchange is considered, the heterotrophy of Nikko Bay becomes even higher. Although there are no wind speed data measured directly inside Nikko Bay, wind speed measured at PICRC station was 0.51 ± 0.80 m s −1 (N = 4320) during this period, and wind can be expected to be weaker in Nikko Bay, which is surrounded by islands. Using the average pCO 2 of seawater in the bay (670 μatm, Table S2), atmospheric pCO 2 (390 μatm, calculated from mole fraction of CO 2 in dry air data collected at Guam), and gas exchange coefficient 33 utilized; gives the largest gas flux under this condition), the pCO 2 flux was calculated to be about 5.6 mmol m −2 d −1 . Using this value, the Pn in Nikko Bay would be approximately -12.5 mmol m −2 d −1 .
To evaluate the correlation among seawater carbonate chemistry and benthic community structure, six sites (N2-N7) along with the Ω arag gradient (1.28-3.51) inside the bay and one site outside of Nikko Bay (N1) were selected for benthic community observation (Fig. 1, Table S3). Even though the seawater inside the bay was warmer and more acidified than the seawater outside the bay, hard coral coverage inside the bay (N2-N7) ranged  Table S5). On the other hand, there was a significant increase in soft coral (= octocoral) coverage with an observed decrease of Ω arag (GLM, p = 0.01), which follows previous results at a CO 2 vent at Iwotorishima in southern Japan showing high coverage of soft coral at a high pCO 2 site 19 . Turf algae coverage increased with Ω arag (GLM, p = 0.04), contradicting previous observations at a CO 2 vent in the Mariana Islands 20 (Table S5, Fig. S1). In Nikko Bay, the coral community was found to differentiate along with the Ω arag gradient observed from the outer reef to the inner reef area 23 . Here we found that although coral coverage was not affected, the hard coral community structure showed differentiation among sites within the inner reef bay area, and this structure was mainly predicted by seawater Ω arag , dissolved oxygen (DO), Chla, nitrate plus nitrite (NO 3 − + NO 2 − ) concentration, pCO 2 and temperature ( Fig. 4, Table S6). Site N1 (outside of the bay) was characterized by high Ω arag (3.51), low pCO 2 (395 μatm), low temperature (29. 3 °C), low Chl-a (0.55 μg/L), high DO (6.09 mg/L), and was dominated by Acropora spp. (coverage 16.5 ± 4.1%), while site N5 was characterized by low Ω arag (1.28), high pCO 2 (1,305 μatm), high temperature (30.5 °C), high Chl-a (1.68 μg/L), low DO (4.52 mg/L), and was dominated by Merulinidae spp. (15.6 ± 5.2%, Fig. 4, Tables S3-S5). Both Acropora spp. and massive Porites showed a slight but positive correlation with seawater Ω arag (GLM, p = 0.04, Fig. S1, Table S5), suggesting that species belonging to these genera are sensitive to OA, though other environmental factors may also have interactively affected the coverage of those species. Branching Porites (mainly consisting of Porites cylindrica) showed the highest coverage, accounting for 22 to 79% of hard coral cover inside the bay (Fig. 4g, Table S4), and there was no significant correlation with branching Porites spp. and Ω arag (GLM, p = 0.16, Table S5. These results suggest that the high coral cover observed within Nikko Bay is related to the potential acclimatization or adaptation capacity of corals such as P. cylindrica to high pCO 2 (low Ω arag ) seawater.
To determine this possibility, colonies of P. cylindrica were reciprocally transplanted between two inner reef bays; a reference site at Malakal Bay (site M1) and a site in Nikko Bay (site N5, Fig. 1) that had different seawater temperatures and pCO 2 conditions (Table S7). As a result, it was found that while the calcification rate of P. cylindrica originating from M1 significantly decreased when transplanted to N5, the calcification rate of corals from N5 did not show significant differences when transplanted to either M1 and N5 (Fig. 5a, Table S8). Most previous tank experiments have reported a decrease of calcification rates of P. cylindrica at high pCO 2 34,35 or under high pCO 2 and high temperature conditions 36 . Additionally, in contrast to massive Porites, P. cylindrica was found to have less capacity of up-regulating calicoblastic calcifying fluid pH, suggesting a high sensitivity to increases of seawater pCO 2 37 . Additionally, the skeleton density of P. cylindrica did not show significant differences among sites (Fig. S2), again contradicting previous studies that showed lower skeleton densities of corals www.nature.com/scientificreports/ at a CO 2 vent 38 and naturally acidified sites 39 . The net photosynthesis (Pn) rate of P. cylindrica transplanted to site N5 had a significantly higher value (p = 0.04) regardless of their origin (Fig. 5b, Table S8), which may be related to the slightly higher nutrient concentrations at N5 site (Table S7). Respiration (R) rates showed interactive effects among transplanted site and origin site, and the R rate of both M1 and N5 corals was significantly lower when transplanted to their origin site (Fig. 5c, Table S8). As a result, there were also interactive effects among the transplanted site and origin site with regards to gross photosynthesis (Pg):R, with higher values when transplanted to their original site (Fig. 5d, Table S8), indicating higher energy acquirement of corals at their own origin site. Interestingly, the corals Acropora pulchra, Porites lutea and Coelastrea aspera in a semi-enclosed lagoon of New Caledonia with low pH, high temperature, low oxygen conditions but high coral coverage, were found to exhibit lower calcification, higher respiration (R) and lower Pg:R compared to corals outside of the lagoon 25 . Acclimatization of corals at the New Caledonia lagoon was suggested to be caused by high respiration through potentially high heterotrophy of corals within the lagoon, which has high organic carbon sedimentation 25 . A comparatively high heterotrophy of the corals in Nikko Bay is also suggested as zooplankton abundance (particularly copepod abundance) was observed to be higher at site N5 compared to reference site M1 (Fig. 6), and this may partially alleviate the effects of high temperature and high CO 2 by enhancing their energy availability 40,41 . However, taking into account that only the calcification rate of P. cylindrica at site M1 decreased when transplanted to site N5, potential epigenetic or genetic adaptation to the environmental conditions found within the bay appears to have occurred for Nikko Bay corals. This is also indicated by other findings that showed Pocillopora acuta within Nikko Bay had higher calcification rates when transplanted to their original site than out of the bay, while P. acuta from out of the bay were not able to survive when transplanted within the bay 42 . P. cylindrica from N5 was also found to host two types of Cladocopium subclade C1 43 (former Symbiodinium 'Clade C'), as well as Durusdinium 43 (former Symbiodinium 'Clade D'), which are known to be tolerant to high temperatures 44 , while P. cylindrica from the other sites only hosted Cladocopium subclade C1 (Fig. S3). These differences in Symbiodiniaceae, particularly at the most sheltered Nikko Bay site, may be another adaptation mechanism of corals to the environment found within Nikko Bay. However, molecular studies evaluating the potential genetic differentiation of those host corals within the bay are first needed before implying the occurrence of local www.nature.com/scientificreports/ adaptation. Thus, for further understanding, molecular studies evaluating the potential genetic differentiation of these host corals within the bay are essential in evaluating of the possibility of local adaptation. From the present study, coral community structure was found to change according to the seawater environmental conditions within the bay, and corals living within the bay such as P. cylindrica could maintain their fitness in the warmed and acidified conditions found within the bay. However, interpretation of these results as related to future climate change should be taken carefully, as several other environmental factors including Chl-a, DO,  www.nature.com/scientificreports/ inorganic nutrient concentrations, and light intensities also varied among sites. Additionally, corals within this bay have been suggested to have been continuously exposed to the unique environment within Nikko Bay for at least the past 150-500 years 12 , while climate change is predicted to continue for the next few decades to centuries. Nevertheless, these results give important insights about the potential acclimatization and adaptation capacities of corals to different environmental conditions, even at small spatial scales, on coral reefs.

Methods
Water quality measurements. The carbonate chemistry including total alkalinity (TA), dissolved inorganic carbon (DIC), aragonite saturation (Ω arag ), and water quality parameters including chlorophyll-a (Chl-a), turbidity (FTU), dissolved oxygen (DO) and inorganic nutrients (DIN, DIP) were measured once each during daytime and nighttime from 40 sites around Nikko Bay ( Supplementary Fig. S1, Tables S1, S2). Water quality measurements in Nikko Bay were conducted during daytime before sunset (15:00-18:00) and during nighttime around sunrise (5:00-8:00) between 17 to 19 November, 2014 at 40 sites (Fig. 1). Water samples were taken before sunset and after sunrise because of the following assumptions. Net primary production and calcification continue until compensation light intensity reaches near sunset, and therefore DIC and TA reach their minimum values around sunset in a closed system. On the other hand, respiration and dissolution continue until sunrise, so DIC and TA are at their maximum around sunrise in a closed system. Because of the highly closed nature of Nikko Bay, it is safe to assume that the range of diel changes in DIC and TA can be captured by collecting water at two time points, around sunset and around sunrise. As well, the middle value of these two time points can give the average value of the system. Seawater temperature, salinity, depth, Chl-a (fluorescent chloride sensor), dissolved oxygen (DO, phosphoresce based oxygen sensor) and turbidity (FTU, backscattering sensor, formazin reference) were measured by vertical casting (1 s logging interval) using a multi-parameter sensor (AAQ-Rinko, JFE Advantech). Salinity data was calibrated by a salinometer (PORTASAL 8410A, Guildline Instruments), and Chl-a (3 replicates) by a fluorometer (Trilogy, Turner Designs). Surface water samples for TA, DIC (2 replicates) and nutrient (DIN, DIP) measurements (4 replicates) were collected from the same sites. TA and DIC were measured within 24 h after seawater collection using an auto burette titrator (ATT-05, Kimoto Electronic Co. Ltd.), which has an accuracy of ± 2 µmol Kg −1 for TA and ± 3 µmol Kg −1 for DIC and standardized by certified reference materials obtained from A. Dickson Laboratory (Scripps Institution of Oceanography, Batch 139). When samples were not measurable within 24 h, they were poisoned with saturated mercury chloride and kept within air-tight glass vial bottles, and these samples were measured within a few days after collection. Nutrient samples were stored in a freezer (-20 °C) until measurement by nutrient autoanalyzer (AACSIII, BRAN + LUEBEE). Carbonate chemistry was calculated from measured TA, DIC and surface salinity and temperature measured at 30 cm below the seawater surface by a casting sensor using CO2SYS 45 with the constants of Mehrbach et al. 46 as refit by Dickson and Millero 47 , and aragonite solubility of Mucci 48 . The accuracy of calculated pCO 2 is estimated as approximately 2-3% of the measurement range of Nikko Bay (see Anderson et al. 49 for calculation of propagation error).
Mean net calcification and primary production in Nikko Bay were calculated from the differences between the average of inside Nikko Bay TA and DIC (n = 74 data each, Table S2) and offshore end-members, the average depth of the bay (d = 18 m), and the mean water residence time within the bay (τ = 71 days 12 ), by simply assuming these TA and DIC differences were accumulated in the bay during this residence time; mean net calcification rate (Gn) was calculated from the TA differences, τ, d, and seawater density (ρ), and primary productivity (Pn) rate was calculated from DIC and TA differences, τ, d, and ρ (see Kayanne et al. 30 and Watanabe et al. 50 ). For the calculation TA and DIC were normalized at S = 33.02 which was the mean salinity of Nikko Bay.
In addition to seawater multi-station measurements across the entire bay, several measurements were conducted during the subsequent benthic survey and coral transplantation survey. Seawater qualities (3 m depth) for the seven sites where benthic surveys were conducted were measured using the AAQ-Rinko in the same manner as above. Additionally, seawater samples were taken at 3 m depths using Van Dorn samplers to measure TA, DIC, and nutrients. During coral transplantation, bottom seawater was sampled using Van Dorn samplers on October 17, 2015 for measurement of Chl-a (3 replicates), nutrients (4 replicates), and suspended solids (3 replicates). Seawater carbonate chemistry (TA and DIC) was measured by collecting two replicate water samples three times (17 Oct, 28 Oct and 5 Nov, 2015) during transplantation. pH and temperature loggers (SPS-14, Kimoto Electrode, Japan) were also deployed at the three transplantation sites for ca. 5 days. The pH data for the loggers were calibrated using TRIS-AMP buffers and are reported in the total scale. Benthic community survey. Benthic coverage including coral community cover (at the genus level) was determined by taking 0.5 × 0.5 m photo-quadrats every meter along five 50 m transects at 3 m depth. A total of 50 photographs were taken per transect and per depth at each site; representing an area of 12.5 m 2 of benthos per depth and site. We measured coral community composition and water quality of one site outside of Nikko Bay (N1) and six sites within Nikko Bay (N2-N7) at similar depths (~ 5 m) showing different Ω arag conditions (1.39-3.45) for the analyses.
Sites N1 to N5 were surveyed in November 2015 while sites N6 and N7 were surveyed in August 2014. No major disturbances within Nikko Bay (e.g. bleaching events) were observed between the two benthic survey dates. Benthic photographs were analyzed using CPCe software 51 . The benthic substrate directly below five random points per photograph was classified into benthic categories as described in several previous studies (e.g. Houk and van Woesik 52 , Golbuu et al. 14 , Gouezo et al. 17 , Houk et al. 15 ). The benthic categories included live corals (to the genus level), fleshy macroalgae (identified to the genus level), turf algae, other invertebrates (e.g. sponges, ascidians, gorgonians), crustose coralline algae (CCA), and non-living substrates (e.g. bare rock, rubble, sand).  Fig. 1b, c). After bringing all samples to the center, coral nubbins were each glued individually onto the top of plastic bolts using epoxy glue, and transplanted back to the same site from which they were collected. After one month, all nubbins were recovered and buoyant wet weights were measured with an electronic balance (0.1 mg precision HR-200, A&D, Japan) twice for each nubbin. Thereafter, coral nubbins were reciprocally transplanted to the two sites, with each one nubbin of 12 colonies collected from the 2 sites transplanted to each 2 sites (n = 24 per site). During transplantation, both seawater temperatures and light intensities were logged at 10 min intervals using temperature (CO-UA-002, HOBO, Onset Corp.) and light sensor loggers (DEFI-L, JFE Advantech, cleaned every weak), respectively. Eighteen days after transplantation, all nubbins were recovered and the buoyant wet weights of all nubbins were measured. Skeletal dry weights were calculated using aragonite density (2.94 g cm −3 ) according to Davis 53 , and calcification rates were calculated by the change of skeletal dry weights during transplantation and normalized by the surface areas measured using the aluminum foil technique 54 . Additionally, photosynthesis and respiration rates were measured for 9 out of 12 nubbins in each condition. Coral nubbins (36 nubbins in total) were first cleaned to remove any attached algae on the coral surface, and each nubbin was then placed individually in an airtight glass container (volume = 400 ml) filled with seawater collected from the same site where the corals were transplanted, placed under LED light (250 μmol photon m −2 s −1 ) and photosynthesis was measured during daytime (11:00-14:00). One extra incubation container without a nubbin was also added as a control. Seawater within the container was continuously stirred with a magnetic stirrer (450 rpm) during incubation and oxygen concentrations were measured at 0, 20 and 40 min using an oxygen sensor (Fibox3, PreSens) calibrated with 100% saturated and 0% oxygen water prepared by Na 2 SO 3 . The same procedure was also conducted at night (20:00-22:00) under dark conditions to measure the respiration rate. Net photosynthesis (Pn) and respiration (R) rates were calculated by the following equation, where V is volume (ml), T is time in hours, and SA is the surface area of the coral (cm −2 ). Gross photosynthesis (Pg) was calculated by Pn minus R.
Seawater volume versus coral nubbin size was care to minimize the change in carbonate chemistry of the seawater within the container during the incubation due to the photosynthesis and respiration by the corals, and the seawater pH change was kept to be smaller than 0.08, which gives in average pCO 2 change of 200 μmol.
Zooplankton. Zooplankton were sampled at night after sunset for three days (11, 12 and 13 March 2016) from three sites (N5, N7, and M1) using a 100 m Nansen plankton net (30 cm diameter). Horizontal tows were conducted 5 times at 1-3 m depth per site, and the filtered volume was recorded with a flowmeter. Zooplankton samples were split and one-half of the sample was preserved in 5% borated-buffered formalin. Individual numbers of all zooplankton and copepods of the formalin fixed samples were counted under a microscope.
Statistical methods. Differences in seawater qualities inside and outside Nikko Bay were analyzed using Wilcoxon signed-rank test. Redundancy analysis (RDA) was conducted as a constrained ordination technique to relate the coral communities to seawater environmental variables. Input for the RDA consisted of coral coverage data that were first transformed using the decostand function in the R package vegan 55 . In the present case, the Hellinger distance was used. Generalized linear models (GLMs, family = quasipoisson) were used to evaluate the relation between Ω arag and coral and other benthic coverage. The calcification rates of the transplantation experiment were evaluated with a Generalized Linear Mixed Effects model (family = Gamma) with origin site and transplanted site and its interaction as fixed effects and colony as random effect. Net photosynthesis, respiration (log transformed), and Pg:R (log transformed) of the transplantation experiment were evaluated with a linear mixed-effect model (REML) with origin site and transplanted site and its interaction as fixed effects and colony as random effect. Tukey's HSD test was conducted when there was a significant interaction. Differences of zooplankton and copepod abundances among sites were tested with student t-test. All statistical analyses were conducted using R (version 3.6.3) 56 .