Ocean acidification reverses the positive effects of seawater pH fluctuations on growth and photosynthesis of the habitat-forming kelp, Ecklonia radiata

Ocean acidification (OA) is the reduction in seawater pH due to the absorption of human-released CO2 by the world’s oceans. The average surface oceanic pH is predicted to decline by 0.4 units by 2100. However, kelp metabolically modifies seawater pH via photosynthesis and respiration in some temperate coastal systems, resulting in daily pH fluctuations of up to ±0.45 units. It is unknown how these fluctuations in pH influence the growth and physiology of the kelp, or how this might change with OA. In laboratory experiments that mimicked the most extreme pH fluctuations measured within beds of the canopy-forming kelp Ecklonia radiata in Tasmania, the growth and photosynthetic rates of juvenile E. radiata were greater under fluctuating pH (8.4 in the day, 7.8 at night) than in static pH treatments (8.4, 8.1, 7.8). However, pH fluctuations had no effect on growth rates and a negative effect on photosynthesis when the mean pH of each treatment was reduced by 0.3 units. Currently, pH fluctuations have a positive effect on E. radiata but this effect could be reversed in the future under OA, which is likely to impact the future ecological dynamics and productivity of habitats dominated by E. radiata.

, because CO 2 is more depleted in 13 C (i.e. its δ 13 C is lower) compared to HCO 3 − 41 . Thus, we also hypothesized that 5) E. radiata Δ 13 C (i.e. the difference between tissue and source seawater DIC in each treatment) would increase with declining pH (i.e. increasing CO 2 ) as a result of increased use of diffusive CO 2 over active uptake of HCO 3

Results
Field Measurements. Seawater pH showed clear diel cycles at all 3 sites where pH was measured. The range in pH on the total scale (pH T ; all subsequent field measurements are referred to on the total scale) was larger (0.40 units) within the more sheltered, shallower sites (where pH was measured only during daylight hours over 3 days) than at deeper, more wave-exposed sites (0.05-0.09 units; over 21 days). Seawater pH within the sheltered, shallow (1.5 m depth) E. radiata/Phyllospora comosa bed at Darlington, Maria Island, displayed a clear increase over the course of the day, with a minimum of pH 7.97 ± 0.06 at 08:00 on day 2 and maximum of 8.37 ± 0.01 at 14:00 on day 3 (Table 1). pH was tightly correlated with oxygen concentration (r = 0.89 ± 0.06, P < 0.001). Seawater pH in Fortescue Bay in late autumn and mid-winter showed a similar diel pattern, but it had a smaller range than at Maria Island; pH fluctuated 0.08 pH units (8.01 to 8.09) at 12.5 m depth over 3 days. At the mouth of Fortescue Bay in early spring, pH varied by 0.09 pH units (8. pH within sealed bags containing adult E. radiata sporophytes in the shallow E. radiata/Phyllospora comosa bed increased significantly more (from 8.00 ± 0.02 to 8.72 ± 0.05) over the course of the day than control bags without E. radiata (from 8.00 ± 0.02 to 8.14 ± 0.03; ANOVA, F 2,17 = 142.31, P = <0.01). Oxygen concentrations also increased significantly more (84.43%) within the bags containing E. radiata compared to controls (ANOVA, F 1,17 = 36.86, P < 0.01). Oxygen concentration was positively correlated with pH, both in bags with and without E. radiata (r = 0.94 ± 0.02).
Laboratory experimental conditions. pH treatments within both of the experiments were maintained within 0.05 units of the treatment target throughout all experiments, and the standard error of pH in each tank for each treatment was smaller than the measurement error (< 0.01). The pH T of treatments (with treatment pH NBS labels in parentheses) was 7.78 (7.8), 8.09 (8.1) and 8.35 (8.4) for growth experiment 1 and 7.54 (7.5), 7.80 (7.8) and 8.06 (8.1) for growth experiment 2. Alkalinity (A T ) did not vary substantially across treatments within experiments, being 2294.57 ± 10.79 μ mol kg −1 (mean ± s.e.) in the first experiment and 2352.66 ± 16.40 μ mol kg −1 in the second. DIC measured in each treatment was inversely related to the pH of that treatment (Table S1). Salinity was 35.00 ± 0.01 in each treatment.
Growth experiment 1: ambient seawater, mean pH 8.1. Relative growth rates of blades were on average 47% greater in the fluctuating pH treatment than in all other treatments ( Table 2, Fig. 1a). Net photosynthetic rates were higher (on average 253% higher) in the fluctuating pH treatment compared to the static pH 7.8 and pH 8.1 treatments ( Table 2, Fig. 2a). RNA:DNA ratios were significantly higher in the fluctuating and static pH 8.4 treatments than at constant pH 8.1 and pH 7.8 (mean 101% higher, Tukey's Honestly Significant Difference (THSD), P < 0.01 for all significant differences) Fig. 3a) and this was caused by RNA content being 198% higher in these treatments ( Table 2, Fig. S1a). There was no significant effect of pH treatments on DNA content ( Table 2, Fig. S1c). There was no effect of treatment on rETR max (Table 2, Fig. S4a). F v /F m was highest at pH 8.4 (0.73 ± 0.01) and lowest at pH 8.1 (0.69 ± 0.01; Table 2, THSD: P = 0.015, Fig. S2a), but at all times was within the range indicating that photosystem II (PSII) was functioning normally. Δ 13 C values of individuals increased as the day time pH of treatments decreased (Fig. 4a). Δ 13 C values were significantly higher in the fluctuating treatment compared to the static pH 7.8 treatment ( Table 2). C:N ratios were significantly higher in the static pH 7.8 treatment than in all other treatments (  Fig. 1b). However, net photosynthesis was 49% lower in the fluctuating pH treatment relative to the static pH 7.8 treatment ( Table 2, THSD: P = 0.03, Fig. 2b). There were no significant differences between treatments in RNA:DNA ratios ( Table 2, Fig. 3b), total RNA, and total DNA content ( Table 2, Fig. S1). There was no effect of experimental treatment on rETR max (Table 2, Fig. S4b). Although F v /F m was significantly lower in the fluctuating pH treatment (0.69 ± 0.01, mean ± s.e.) than the pH 7.8 (0.72 ± 0.01) and 8.1 (0.73 ± 0.01) treatments ( Table 2, Fig. S2b), again these values indicate normal functionality of PSII. Similar to Experiment 1, Δ 13 C values increased with decreasing pH during the day, and were significantly higher in the pH 7.5 and 7.8 treatments compared to the fluctuating pH treatment ( Table 2, TSHD: P = 0.02 and 0.03, Fig. 4b). C:N was similar for all treatments ( Table 2, Fig. S3).

Discussion
We measured diel changes in pH within E. radiata beds of up to 0.4 units, and found that when these fluctuations are simulated in laboratory experiments, rates of blade growth and photosynthesis (measured as O 2 evolution)   of juvenile E. radiata increased compared to static pH treatments. However, this positive effect of fluctuating pH on growth and photosynthetic rates was not apparent when the experimental pH was reduced in all treatments  Bars within panels sharing a letter are not significantly different, as revealed by Tukey's Honestly Significant Difference tests (α = 0.05). by 0.3 units, simulating future OA. Moreover, pH fluctuations under simulated OA had a negative impact on net photosynthesis relative to static pH treatments with the same mean pH. These findings are important in the context of predicting the responses of kelp-based communities to OA. Past research has considered that metabolically-induced pH fluctuations in habitats dominated by photosynthetic species (macroalgae, seagrasses, corals) could act as a refuge from OA for calcifying species because higher pH during the day might facilitate calcification 22,26,43,44 . However, this refuge may be less effective than previously considered in a future, reduced pH ocean, if these pH fluctuations no longer benefit the kelp that are responsible for the pH changes. Increased variability in pH has a negative impact on coralline algae under ambient mean seawater pH, the effects of which are enhanced by OA 22,37 . However, in the case of coralline algae, reduced pH in general is known to slow growth and net calcification, possibly through elevated dissolution rates 22,43,45 . The mechanisms responsible for the negative impacts (relative to fluctuating pH under current conditions) of fluctuating pH under OA conditions on the non-calcareous E. radiata are unclear (see below). Further research is required to determine whether other wide-spread and abundant habitat-forming macroalgal species will respond to pH fluctuations predicted for the future in a similar way to E. radiata, or whether this is a species-specific response.
pH fluctuations were not spatially or temporally uniform within the Ecklonia radiata beds, further indicating that the ability of habitats dominated by photosynthetic species to act as a refuge from OA is context-dependent. The largest range in pH observed over 24 hours (0.40 pH units) was at the shallowest and most wave-sheltered site (Maria Island), and the smallest range in pH (0.05 pH units) was at the outermost site at Fortescue Bay at 25 m depth which is both the most wave-exposed and deepest site. It was not our objective to determine the environmental factors responsible for different pH ranges within E. radiata beds, nor do the data allow us to do so, however the consistent trends of pH fluctuations we observed support findings that the extent of mixing and water retention within a habitat and depth 44,46,47 influence the magnitude of pH fluctuations in macrophyte systems. While we used several methods to measure pH of the seawater in situ, they each followed the recommended best practice 48 and we do not consider that differences between methods influenced the trends observed. Before effective predictions can be made regarding the extent of pH change likely to occur in various macrophyte-dominated habitats, further research is required that combines physical modelling with long-term, spatially-extensive measurements of pH, water motion, depth and biotic characteristics that are replicated through space and time. Once these data are obtained, the role that macrophyte habitats could play in modifying the effects of OA on resident organisms can be more thoroughly understood.
Contrary to our initial hypotheses, the net photosynthetic and growth rates of Ecklonia radiata were not elevated in treatments with reduced daytime pH (and more CO 2 ) relative to present ocean conditions. This supports past research indicating that increased concentrations of DIC, particularly CO 2 , associated with OA may not benefit this species 6 . The Δ 13 C of E. radiata displayed a clear trend of a greater reliance on diffusive CO 2 as a carbon source for photosynthesis in treatments with lower daytime pH, as hypothesized initially. However, we saw no evidence of increased growth rates as might be expected if CCMs are down-regulated, despite the increasing reliance on diffusive CO 2 at low pH; this is similar to the growth responses of Ulva rigida and Macrocystis pyrifera 31,36 to elevated CO 2 .
Rather than increased growth at higher DIC concentrations (i.e. lower pH), we found that the rates of E. radiata growth and photosynthesis were maximal under the experimental treatment that most closely simulated the pH fluctuations that it currently encounters in the field (daytime pH 8.4, night-time pH 7.8), suggesting that it is physiologically adapted to these pH/DIC conditions. We suggest that the specific combination of a daytime pH 8.4 and night-time pH 7.8 provides a seawater carbonate chemistry that is conducive to growth, photosynthesis, and RNA synthesis by Ecklonia. The static pH 8.4 treatment itself appeared to be slightly beneficial to Ecklonia because RNA synthesis was increased, along with evidence of increased O 2 evolution, although there was no evidence of increased growth. The explanation for this finding of stimulated metabolism when the daytime pH was 8.4 may be related to enhanced bicarbonate uptake at higher pH, which has been previously observed at very high pH (~9.0) for green seaweeds, but not brown seaweeds 49 . However, we know relatively little of the diversity of bicarbonate uptake mechanisms in kelps, although this knowledge is required to further understand how they will respond to OA (see references within 31,50 ). This is the first documentation of seaweed metabolism being enhanced by diel fluctuations in seawater carbonate chemistry. The mechanisms are unknown but are likely to involve metabolic processes that are enhanced by higher H + and/or DIC at night, along with decreased H + and/or DIC during the day. Further studies are needed including those using enzyme inhibitors and gene expression, to elucidate why this specific combination of pH/DIC during the day and night stimulated growth of E. radiata. The next step in quantifying how seawater carbonate chemistry influences growth and photosynthesis of macroalgae should investigate the activity of different CCMs in macroalgae grown for long periods of time under pH treatments similar to ours. This can be done by combining direct assessments of specific CCMs 50 with gene expression data.
This study is the first to examine the effects of fluctuations in pH on a non-calcareous macroalgae, and our findings build on those of earlier studies 22,37,38 which highlight the importance of acknowledging pH fluctuations when assessing how ecologically important calcifying and non-calcifying primary producers might be affected by ocean acidification in dynamic coastal environments. The finding that the positive effects of fluctuating pH were not apparent under OA has implications for the abilities of kelps to act as refugia for calcifying organisms in the future (cf 27,28,43 ). This study raises many questions, but it now seems clear that it is important to ascertain how the magnitude of macrophyte-induced diel pH fluctuations in shallow environments influence the physiology and ecology of marine species now and in a future lower pH ocean. How different magnitudes of fluctuations combine with changes in mean pH to influence organisms now needs to be addressed for a range of species before we can begin to predict how long-term changes in near-shore pH due to OA could impact these ecosystems.
Scientific RepoRts | 6:26036 | DOI: 10.1038/srep26036 Methods Field measurements. To determine whether pH within Ecklonia radiata beds follows a daily cycle similar to those occurring in other ecosystems 22,24 , seawater pH (and where possible dissolved oxygen) was measured on five different sampling occasions at two general locations, viz. Darlington (Maria Island), and Fortescue Bay, Tasmania: (1) In a shallow (1.5 m), sheltered E. radiata/Phyllospora comosa bed, an environment likely to have extreme pH fluctuations (Darlington), at regular intervals between 07:30 and 17:30 on each day between April 19-21 2014. At this location pH and dissolved oxygen (DO) were measured using a pH meter (Thermo Scientific Orion Star A216 pH/RDO/DO meter), pH electrode (Thermo Scientific Orion 8107 BNUMD Ross Ultra pH/ATC Triode) and DO probe (Thermo Scientific Orion 087100MD Field RDO probe). Seawater was collected in situ using bottles (100 ml plastic sealed bottles) because surge prohibited the use of sensors in the shallow water; (2) At 7 m and 25 m depth on an exposed coast (Fortescue Bay) where pH fluctuations are likely to be minimal, between 5-26 September 2014. In situ SeaPHOX loggers with SeaFET pH sensors were used to measure pH and DO hourly at these sites; (3) In a moderately exposed site at 12.5 m (Fortescue Bay), most representative of E. radiata beds on the east coast of Tasmania and where pH change is likely to be intermediate between the situations described in 1) and 2), between 20-22 May 2014. An ENVCO pHTempion combined pH and temperature logger was used to measure pH at 5 min intervals as we did not have access to the primary sensors (the seaPHOX) at this time; (4) Within sealed bags containing either seawater only, or seawater and an adult sporophyte of E. radiata, to determine the extent to which E. radiata metabolism can change pH in the field, and to separate the changes caused from photosynthesis and respiration of phytoplankton from that caused by E. radiata, at Darlington, on each of 18, 19 and 21 April 2014 using the pH meter, probe and DO probe described in 1); and (5) Within the shallow (7 m) exposed E. radiata bed (described in 2) above) and on the adjacent soft-sediment substratum (Fortescue Bay, 20 m depth), to determine whether pH changes are localised within beds or evident over larger spatial scales, using a Niskin sampler and the pH meter, probe and DO probe described in 1) and 4), between 06:45 and 16:45 on 27 November 2014.
These five sampling procedures were used to capture a range of examples of pH change that could occur in E. radiata beds. Electrodes were calibrated initially using pH 7 and pH 9 NBS buffers on site, then pH on the total scale (pHT) was calculated afterwards using Tris and amp buffers at 14 °C. All pH measurements are given on the total scale, unless otherwise noted. The DO probe was calibrated by measuring the oxygen concentration of seawater that had been bubbled with air for 10 min (100% saturation) and by measuring the DO content of seawater that had been bubbled with N 2 gas for 5 min (0%) at 14 °C. These methods of calibration were used for all measurements in the study, with the exception of the seaPHOX's which were calibrated by taking DIC and A T samples of seawater at deployment, a mid-point and collection. See supplementary methods (SI 1) for a full description and rationale of the methods of water sampling and pH measurement. Intact sporelings (n = 24) were assigned randomly to one of 4 pH treatments (n = 6 for each treatment). We labelled treatments with names referring to the NBS scale to 1 decimal place, but measured pH on the total scale using Tris and amp buffers 48 (see results for values). The treatments were: "pH 8.1", "pH 7.8", "pH 8.4", and a "fluctuating pH" treatment which consisted of "pH 8.4" during the day and "pH 7.8" at night. All individuals were kept at 14 °C on a 12:12 light:dark cycle, with lights turned off at 20:00 and on at 08:00 in a temperature controlled room. Light levels during the day were 10 μ mol photons m −2 s −1 from day 1 to day 5 of the experiment to reduce stress associated with high light after collection (C.E. Cornwall, C.D. Hepburn and C.L. Hurd, unpublished data), and then raised to 28 μ mol photons m −2 s −1 from day 6 onwards, with the experiment conducted for a total of 21 days. One individual (replicate 5, constant pH 8.4) became necrotic early in the experiment and was discarded and excluded from analysis.
The pH of the experimental seawater collected from Bruny Island, Tasmania was modified either by bubbling with CO 2 gas to reduce pH (chemically simulating biological release of CO 2 ) or by bubbling with N 2 gas to raise pH (simulating the biological uptake of CO 2 ). These methods altered the concentration of DIC without altering A T , as is recommended 32 . Bubbling of N 2 gas also removed dissolved O 2 so subsequent re-equilibrium of O 2 to 100% saturation was undertaken for the high pH treatments by bubbling O 2 gas through the seawater. pH manipulations for each experimental treatment took place in the same container in a random order each day, and the same tubing was used for all gas delivery for every treatment to avoid pseudoreplication, following Fig. 3c from Cornwall and Hurd 51 . Modified water was placed in header tanks, which consisted of 5l low density polyethylene bags lined with a metallic film to keep pH constant for short term storage. Seawater then drained from the bags into the experimental tanks over time. The fluctuating pH treatment was achieved by changing bags over at 08:00, 11:00, 14:00, 17:00, and 20:00, with bags containing the appropriate seawater for the treatment, in other words using a step-wise approach 22,37,52 and where pH was 8.4 in the day and 7.8 at night. Each individual was grown in a separate 650 ml culture tank with constant water motion (using magnetic stirrers) to break down the diffusion boundary layer. See 22,43 for a description of the chambers. Trials were conducted to test the extent that juvenile E.
Scientific RepoRts | 6:26036 | DOI: 10.1038/srep26036 radiata could alter pH in the culture tanks under the experimental light conditions (i.e. at 0 and 28 μ mol photons m −2 s −1 ), and revealed that tanks required exchange of seawater every 3 hours during the day and every 12 hours at night to ensure that pH remained within 0.05 units of the target. Growth experiment 2: OA conditions. Approximately 50 E. radiata sporelings were collected from the site at Fortescue Bay (as above) on August 20 th 2014. All experimental conditions and methods were the same as the growth experiment #1 described above, except that the target pH of each treatment was reduced by 0.3 pH NBS units across all treatments to simulate conditions expected to occur in the future as a result of OA. Three individual sporophytes (two replicates subject to constant pH 7.5 and one replicate under constant pH 7.8) became necrotic and were discarded during the experiment.
Biotic responses. Linear extension of the blade length (tip of blade to blade-stipe junction) of each sporophyte was calculated from photographs of individuals placed on a grid on days 1 and 21, using the software program ImageJ 53 . Relative growth rate was calculated relative to the initial size 54 .
Photosynthetic rates were determined on day 21 in each culture tank at experimental light levels at 11:00 after 3 hours in the light, using an Orion RDO probe (ORI087100MDW). Oxygen production was standardised to surface area of the thallus (in mm 2 ) over an hour.
The performance of PSII (F v /F m ) and maximum relative electron transport rates (rETR max ) were measured to assess any changes in photo-physiology during the experiments. F v /F m and rETR max were measured using a Pulse Amplitude Modulation (PAM) chlorophyll fluorescence meter (Diving PAM, Walz, Germany) on day 21. F v /F m measurements were made on E. radiata individuals that had been dark adapted for 15 minutes 55 . The diving PAM had a blue-light-emitting diode, and dampening and gain were set to 1 and 2 respectively. F o was above 120 on all occasions.
Tissue samples from the meristem were taken at the end of the experiment on day 21 from each individual for determination of δ 13 C, C:N ratios, and RNA:DNA ratios. C:N was measured to indicate whether nutrient limitation occurred, and along with rETRmax and F v /F m , was used only as an indication of the physiological state of the algae. δ 13 C and C:N ratios were determined using the methods outlined in 39 using a NA1500 elemental analyser coupled to a Thermo Scientific Delta V Plus via a Conflo IV. Combustion and reduction were achieved at 1020 °C and 650 °C respectively. Values were normalised to the VPDB scale via a 3 point calibration using certified reference material. Both precision and accuracy were ± 0.1‰ (1 SD). δ 13 C values were corrected to an absolute value (Δ 13 C) by determining the δ 13 C of the seawater for each treatment and subsequently using the formula: Δ 13 C = (δ 13 C seawater − δ 13 C tissue )/(1 + δ 13 C seawater /1000) 43 . For analysis, one individual was removed from consideration of C:N ratios (treatment = constant pH 8.1) as this individual returned an extreme and unrealistic value substantially lower than all other replicates from the same treatment. See Supplementary Methods (SI 1) for details of RNA and DNA measurements.
Seawater samples were collected from each replicate at the beginning of a change in seawater pH and again after 3 hours. DIC concentrations of the water samples were measured using a DIC analyser (Apollo SciTech DIC analyser model AS-C3) with an inbuilt CO 2 analyser (LI-COR LI-7000 CO 2 /H 2 O analyser). The CO 2 analyser was calibrated with a certified reference material provided by Andrew Dickson, Scripps Institute for Oceanography, San Diego, USA 48 . A T was calculated using the constants of Mehrbach 56 and the refit by Dickson and Millero 57 .
Statistical analysis. All statistical analyses were conducted using the statistical software R v. 3.1.1 58 .
Correlations between pH and oxygen concentrations during field measurements were calculated using Pearson's correlation coefficient. Two-way ANOVAs were used to assess whether there were differences between pH and oxygen concentration changes in bags placed in the field, with a fixed factor of 'Treatment' (2 levels: chambers containing Ecklonia radiata and control bags with no kelp) and a random factor 'Day' (3 levels: April 18 th , 19 th and 21 st ). In the laboratory-based experiments, one-way ANOVAs were used to estimate whether there were significant differences between pH treatments for RGR, oxygen evolution, F v /F m , rETR max , C:N ratios, Δ 13 C, RNA:DNA ratios, and RNA and DNA content. When main effects in one-way ANOVAs were significant (at α = 0.05), Tukey's Honest Significant Difference (THSD) post-hoc tests were used to determine the nature of differences between treatments. Data were checked for violations of ANOVA assumptions (normality and homoscedasticity), and passed all tests, except for Δ 13 C in the ambient pH experiment; 2 outliers were removed from the pH 8.1 treatment as they returned values that were substantially different to other replicates in the treatment.