Saturating light and not increased carbon dioxide under ocean acidification drives photosynthesis and growth in Ulva rigida (Chlorophyta)

Carbon physiology of a genetically identified Ulva rigida was investigated under different CO2(aq) and light levels. The study was designed to answer whether (1) light or exogenous inorganic carbon (Ci) pool is driving growth; and (2) elevated CO2(aq) concentration under ocean acidification (OA) will downregulate CAext-mediated dehydration and alter the stable carbon isotope (δ13C) signatures toward more CO2 use to support higher growth rate. At pHT 9.0 where CO2(aq) is <1 μmol L−1, inhibition of the known use mechanisms, that is, direct uptake through the AE port and CAext-mediated dehydration decreased net photosynthesis (NPS) by only 56–83%, leaving the carbon uptake mechanism for the remaining 17–44% of the NPS unaccounted. An in silico search for carbon-concentrating mechanism elements in expressed sequence tag libraries of Ulva found putative light-dependent transporters to which the remaining NPS can be attributed. The shift in δ13C signatures from –22‰ toward –10‰ under saturating light but not under elevated CO2(aq) suggest preference and substantial use to support photosynthesis and growth. U. rigida is Ci saturated, and growth was primarily controlled by light. Therefore, increased levels of CO2(aq) predicted for the future will not, in isolation, stimulate Ulva blooms.


Introduction
The world's oceans are a sink for CO 2 that has been released through anthropogenic processes since the indus-trial revolution (~1850). This process buffers climate change in the terrestrial system, but perturbs the seawater carbonate system, and is reducing the pH of the surface ocean, termed ocean acidification (OA) (Takahashi et al. 2014). In the next century, atmospheric CO 2 concentrations are projected to increase from 395.93 ppm (October 2014;www.CO2now.org) to a maximum of 1142 ppm in 2100 (IPCC 2013): the 192% increase will consequently reduce global surface ocean pH by 0.4 units (Meehl et al. 2007). OA may have a profound impact on organismal carbon physiology, trophic dynamics, biogeochemical cycles, and ecosystem functions and services (Gattuso and Hansson 2011;Roleda and Hurd 2012;Koch et al. 2013).
Seaweeds are macroscopic, multicellular marine algae. They are foundation species and provide energy to higher trophic levels. They belong to three distinct evolutionary groups: green (Chlorophyta), brown (Ochrophyta), and red (Rhodophyta) seaweeds. Most seaweeds are able to take up both CO 2 and HCO À 3 from the surrounding seawater as inorganic carbon (Ci) sources for photosynthesis. CO 2 can enter seaweed cells through the plasma membranes by passive diffusion. However, this process is very slow in water and dissolved carbon dioxide (CO 2(aq) ) makes up less than 1% of the total Ci. On the other hand, HCO À 3 is the most abundant ionic form of Ci (92%) but requires specific carriers for transport across membranes (Bj€ ork et al. 1992;Sharkia et al. 1994).
To overcome the low availability of CO 2(aq) to supply the ribulose-1,5-bisphosphate carboxylase/oxygenase (Ru-BisCO) for photosynthetic carbon fixation, most seaweeds have developed strategies of utilizing HCO À 3 . These strategies include as follows: (1) extracellular carbonic anhydrase (CA ext )-mediated HCO À 3 dehydration and subsequent passive diffusion of CO 2 into the cell (Bj€ ork et al. 1992).
(2) Direct HCO À 3 uptake by the anion-exchange protein AE1, similar to that present in mammalian red blood cells (Drechsler and Beer 1991;Drechsler et al. 1993;Sharkia et al. 1994). The internal inorganic carbon (Ci) pool is stored primarily as HCO À 3 ; consequently, this requires intracellular carbonic anhydrase (CA int )-mediated HCO À 3 dehydration to supply CO 2 to RuBisCO for photosynthetic carbon fixation (Fern andez et al. 2014). For the seaweed genus Ulva, that is the focus of this study, the relative contribution of CA ext -mediated HCO À 3 utilization versus direct HCO À 3 uptake to supply CO 2 for photosynthesis varies substantially among different species where species are identified based on traditional taxonomic classification . (3) An indirect HCO À 3 use operates in conjunction with metabolically mediated proton (H + ) flux where H + extrusion creates a local acidified zone within the diffusion boundary layer at the seaweed surface, thereby stimulating a pH-dependent conversion of HCO À 3 to CO 2 (Bj€ ork et al. 1992;Beer et al. 2008). Such uncatalyzed or spontaneous dehydration of HCO À 3 can, however, be slow and may be insufficient to support high photosynthetic and growth rates (Cook et al. 1986).
Ulva is a globally ubiquitous seaweed, well known for causing massive macroalgal blooms that can have negative environmental and economic impacts (McGlathery 2001;Sun et al. 2008;Pang et al. 2010). The drivers of Ulva primary production and growth are variously attributed to light (Aldridge and Trimmer 2009), inorganic nutrients (Coutinho and Zingmark 1993), CO 2 (Xu and Gao 2012), and their interactions. Under a saturating light (200-500 lmol photons m À2 s À1 ) growth of Ulva curvata can become limited by the supply of inorganic nitrogen (Coutinho and Zingmark 1993). When grown under a subsaturating light (100 lmol photons m À2 s À1 ) (but unspecified nutrient concentration), the growth of Ulva prolifera cultivated from spores was reported to increase by~20% d À1 when CO 2 was increased by 156% (Xu and Gao 2012). Whether or not a saturating light will further enhance the growth of Ulva under elevated CO 2 is unknown.
There are many studies showing that HCO À 3 is the primary source of Ci for Ulva species (Drechsler and Beer 1991;Bj€ ork et al. 1992Bj€ ork et al. , 1993Drechsler et al. 1993;Sharkia et al. 1994). Therefore, it is unlikely that the predicted increase in CO 2(aq) due to OA will have an effect on the rates of photosynthesis and growth in Ulva (e.g. Xu and Gao 2012). Nevertheless, researchers continue to pursue this line of research (e.g. Pajusalu et al. 2013;Olischl€ ager et al. 2013), suggesting that increased levels of CO 2(aq) will cause increased growth rate in Ulva. Here, we demonstrate that light, and not Ci, is the main driver for growth of Ulva and suggest that increased levels of CO 2 (aq) predicted for the future will not, in isolation, stimulate photosynthesis and growth of Ulva.
This study is important and timely because it questions the validity of the current dogma that increasing CO 2 will promote (harmful) algal blooms and drive ecosystem "winners and losers". There is currently a very limited understanding of algal carbon physiology on an organismal level (e.g. K€ ubler et al. 1999) and on the environmental drivers of basic physiological mechanisms (e.g. Raven 1991) such as photosynthesis and growth. Moreover, the combined physiological and molecular approaches to investigate key driver for Ulva growth confers novelty in this work.
The photosynthetic carbon physiology and growth of a genetically identified Ulva species under different CO 2(aq) and light levels were investigated. We hypothesized that (1) under current concentration of CO 2(aq) , exogenous Ci is saturating for photosynthesis and growth of Ulva rigida; (2) U. rigida has several putative HCO À 3 use mechanisms other than the known inhibitor-sensitive CA extcatalyzed dehydration of HCO À 3 and direct uptake of HCO À 3 through the anion-exchange (AE) port; and (3)  future will not affect U. rigida growth. The experiment was designed to answer whether (1) light or exogenous Ci pool is limiting U. rigida growth; (2) elevated Ci (CO 2 and HCO À 3 ) under OA will support a higher growth rate; (3) CA ext -mediated HCO À 3 dehydration is downregulated when Ulva is grown under a high CO 2(aq) concentration; and (4) the hypothetical shift to more CO 2 use under OA will alter the stable carbon isotope (d 13 C) signature (cf Maberly et al. 1992;Raven et al. 2002;Giordano et al. 2005) and the higher available Ci under OA relative to constant nutrient supply will increase the molar carbon to nitrogen (C:N) ratio (cf van de Waal et al. 2010).

Materials and Methods
Algal material and stock culture conditions Sheet-forming Ulva thalli ( Fig. 1) were collected on 5 October 2011 from the subtidal at the entrance (3 m depth) of Otago Harbour, southern New Zealand (Aramoana, 45.8°S, 170.7°E) and transported in a cooled container to the laboratory. Several algal discs ( Fig. 1 inset) were excised from one individual and cultivated in 5-L glass vessels with nutrient-enriched seawater, formulated using the ESNW recipe (Harrison et al. 1980;Berges et al. 2001), which was used throughout the experiment. To avoid significant pH drop, the ESNW was prepared with one-third of the full strength concentrations of phosphate, iron-EDTA (pH <6.0), trace metals II (pH <6.0) and vitamins stock solutions and 5 lmol L À1 of NaNO 3 -N; the final seawater pH was pH T 7.97,~0.08 pH unit lower than ambient seawater. The Ulva clones were grown under photosynthetic photon fluence rate (PPFR) of 50 lmol photons m À2 s À1 over the waveband 400-700 nm (Philips TLD 18W/840 Cool White; Philips, Amsterdam, The Netherlands) and 12 h: 12 h, light:dark cycle at 13°C in a Phytotron Climate Simulator cabinet (Contherm Scientific Ltd, Lower Hutt, New Zealand). Clones were grown below the saturating irradiance of 79 AE 9 lmol photons m À2 s À1 (R. Rautenberger, unpubl. data) to avoid rapid growth, nutrient depletion, and swarmer release. After 8 months clonal propagation, algal discs (3 cm 2 , 25 mg FW) were used in the experiments below.
Two specimens grown from the clonal cultures were prepared as herbarium vouchers and deposited in the herbarium of the University of Otago (OTA; Thiers 2014) under the accession numbers OTA63969 and OTA63970. Material of the latter specimen was prepared for genetic identification by excising a disc from the freshly harvested thallus and drying it in silica gel.

Genetic identification and life history of the cultured strain
The Ulva strain (SBDN 247) was genetically identified using the methods in Heesch et al. (2009), except for the following differences: DNA was extracted and PCR products were purified using commercial kits (NucleoSpin â Plant II, Macherey-Nagel, D€ uren, Germany, and PureLink â PCR Purification Kit, Invitrogen, Germany, respectively). Sequence alignments of the large subunit of the plastidencoded RuBisCO gene region (rbcL) were analyzed under the maximum-likelihood (ML) criterion using the default settings in RAxML v.7.2.2 (Stamatakis 2006). Our rbcL sequence (European Nucleotide Archive [ENA] accession no. LK022428) was found to be 100% identical to rbcL sequences from U. rigida samples collected in New Zealand (e.g. GenBank accession number EF110302; Heesch et al. 2009), Europe (EU484408; Loughnane et al. 2008) and Chile (AY422564: Hayden and Waaland 2004). The phylogenetic analysis placed our strain in a well-supported clade with the above sequences (data not shown), confirming its identification as Ulva rigida C.Agardh.
Swarmers released from clonal tissue were positively phototactic indicating that the cultivated strain was a gametophyte, that is, the haploid generation (Guiry and Guiry 2014). Cultivation of the released gametes failed to develop parthenogenetically, suggesting that swarmers were most likely male gametes, which, in contrast to female swarmers, have lower capacity to germinate without fertilization (Koeman and van den Hoek 1981). number 207233-90-7) are inhibitors with high specificity for blocking the catalyzed external dehydration of HCO À 3 and the direct HCO À 3 uptake through the anion-exchange (AE) port, respectively (Bj€ ork et al. 1992;Axelsson et al. 1995Axelsson et al. , 1999Axelsson et al. 2000;Herfort et al. 2002;Suffrian et al. 2011). Both mechanisms operate independently: the addition of these inhibitors in the absence of other HCO À 3 uptake mechanisms can result in an almost complete inhibition of net photosynthesis (Axelsson et al. 1995;Larsson and Axelsson 1999;Fern andez et al. 2014). A concentration of 300 lmol L À1 DIDS to inhibit the direct HCO 3 À uptake and 100 lmol L À1 AZ to inhibit CA ext were selected based on the dose response curves of Herfort et al. (2002), and the standard utilization of these concentrations across studies on the Ci-use mechanisms of micro-and macroalgal species (e.g. Bj€ ork et al. 1992; Axelsson et al. 1995Axelsson et al. , 19992000;Young et al. 2001;Herfort et al. 2002;Suffrian et al. 2011;van Hille et al. 2014). The 10 mmol L À1 stock solution of AZ (≥99%, Sigma-Aldrich, St. Louis, MO, USA) was prepared by dissolving the powder in a basic medium (10 mmol L À1 NaOH in MilliQ, 18.3 MO cm) while the 30 mmol L À1 stock solution of DIDS (≥80% elemental analysis, Sigma-Aldrich) was prepared by dissolving the powder in Mil-liQ. The stock solutions were freshly prepared and kept at 4°C and dark. Algal discs were acclimated for 2 days under pH T 9.0 where Ci is mainly available as HCO À 3 (700 lmol L À1 ) with minimal CO 2(aq) (<1 lmol L À1 ). Seawater was adjusted using equal amounts of 0.2 mol L À1 NaOH and 0.2 mol L À1 NaHCO 3 (Roleda et al. 2012a). After 2 days, inhibition of photosynthetic O 2 evolution (n = 3) under the same pH T = 9.0 was measured inside a 154-mL transparent acrylic glass chamber equipped with an optode, that is, FOXY-R fiber optic oxygen sensor coupled to the USB-2000 spectrophotometer (Ocean Optics, Dunedin, FL, USA) and the PC interface. The seawater was continuously stirred (650 rpm) to create a homogenous O 2 profile. To avoid photorespiration due to high O 2 concentrations and temperature, the seawater was initially adjusted to 100 AE 20 lmol O 2 L À1 using N 2 gas and kept constant at 12 AE 1°C, respectively. Net oxygen evolution of individual U. rigida discs was recorded (OOISensor 1.0 software; Ocean Optics Inc., FL, USA) at a PPFR of 250 lmol photons m À2 s À1 (400-700 nm). Oxygen concentration inside the chamber was recorded for 20 min before the inhibitors were added and then for a further 2 9 15 min following the sequential addition of the two inhibitors (e.g. DIDS-AZ and AZ-DIDS; n = 3). The oxygen concentration was measured continuously for a total of 50 min per sample. The rate of oxygen evolution was calculated using a linear regression for every stepwise incubation period, that is, the whole incubation time (20 min) before application of the inhibitors, and after every application of each inhibitor (2 9 15 min). The oxygen concentration (lmol O 2 L À1 ) was measured after Millero and Poisson (1981) and Garc ıa and Gordon (1992), and corresponding % inhibition of net photosynthetic (NPS) rate was calculated.
In silico detection of carbon-concentrating mechanisms elements in expressed sequence tag libraries of Ulva To test our hypothesis on Ulva having different Ci uptake mechanisms other than the known inhibitor-sensitive CA ext -catalyzed dehydration of HCO À 3 and direct uptake of HCO À 3 through the AE port, we searched one expressed sequence tag (EST) library of U. prolifera (Jia et al. 2011) publicly available in the dbEST database (Boguski et al. 1993) for putative carbon-concentrating mechanism (CCM) elements. A number of described CCM proteins of the unicellular green alga Chlamydomonas reinhardtii were used in a tblastn search against the U. prolifera EST data set and included the following protein sequences: alpha carbonic anhydrases ( ). Retrieved sequences were further subjected to a reciprocal blastx search against the National Center for Biotechnology Information (NCBI) Genbank nonredundant (nr) database. Furthermore, a Vec-Screen was run on retrieved sequences to check for vector contamination.

Interactive light 3 CO 2 experimental set-up
Six discs of U. rigida clones (each 3 cm 2 ) were contained in each 24 9 650 mL Perspex flow-through culture vessels. Specimen was pre-incubated under pH T 7.96 and nutrient replete condition (11.6 AE 0.7 lmol L À1 NO 3 -N, 9.1 AE 0.4 lmol L À1 PO 4 -P and 0.6 AE 0.1 lmol L À1 NH 4 -N; AE SD, n = 3), exposed to PPFR of 30 lmol photons m À2 s À1 (400-700 nm, 12 h: 12 h, light:dark) inside a temperature-controlled walk-in culture chamber set to 13 AE 1°C. A preliminary rapid ETR-E curve measurement calculated a saturating irradiance E k = 79 AE 9 lmol photons m À2 s À1 , and no photo-inhibition of the maximum electron transport rate (ETR max ) was observed up to a maximum PPFR of 600 lmol photons m À2 s À1 (R. Rautenberger, unpubl. data). Accordingly, two light treatments (PPFR of 400-700 nm) were set as follows: a limiting light (LL) of 31 AE 9 lmol photons m À2 s À1 and a saturating light (SL) of 274 AE 18 lmol photons m À2 s À1 . Philips HPI-T 400 W quartz metal halide lamps (Philips) provided the experimental light and the PPFR measured using a 4p quantum sensor (QSL-2101, Biospherical Instruments Inc., San Diego, CA, USA). The LL treatment was achieved using layers of neutral density black screen cover.
The response of U. rigida to a shift in pCO 2 was investigated under two treatments slightly higher than today's pCO 2 (471 latm) and that predicted for year 2100 (1224 latm). The 160% higher CO 2(aq) between the low and high pCO 2 treatments is slightly lower than the 192% increase projected in 2100 (Meehl et al. 2007;IPCC 2013). After three days pre-incubation described above, each culture vessel was randomly assigned to different pCO 2 /pH T (high CO 2 /low pH T 7.59 = HC; low CO 2 /high pH T 7.97 = LC) and light (LL; SL) treatments under higher nitrate concentration (90 lmol L À1 NO 3 -N). The two factorial experiment resulted in four treatment combinations of light and pCO 2 : HC/LL, HC/SL, LC/LL and LC/SL (n = 6).
Manipulation of the seawater carbonate system was achieved using a modified version of the automated pHcontrolled culture system described by McGraw et al. (2010). Briefly, target pH (pH T ) levels were achieved by adding HCl and NaHCO 3 (Hurd et al. 2009;Riebesell et al. 2010) to the nutrient-enriched seawater and measured spectrophotometrically to an accuracy of 0.03 pH units. Thereafter, the pH-adjusted seawater automatically supplied fresh medium to the respective randomly assigned culture vessel. Seawater in each of the 24 vessels was refreshed every 4.4 h (=49 a day change in medium). Within the 4.4-h incubation period, a maximum 0.18 units increase in pH due to photosynthesis was estimated (Cornwall et al. 2012); thereby persistently exposing the algae to a specific range of the pH treatment. Each culture vessel was provided with water movement using magnetic bars stirred at 650 rpm. Five hundred milliliter of the acid-base manipulated seawater corresponding to pH T 7.59 and 7.97 was collected and fixed with mercuric chloride. Total alkalinity (A T ) of samples was measured using the closed-cell titration method described by Dickson et al. (2007). A T , pH, salinity, and temperature were used to calculate carbonate chemistry parameters using the program SWCO2 (Hunter 2007). The seawater carbonate chemistry is presented in Table 1.
After 7 days, samples were analyzed for growth, photosynthetic efficiency, pigments, stable isotope signatures, and internal and external carbonic anhydrase (CA) activity. For biochemical analyses, samples were stored either at -20°C (pigment) or -80°C (CA) until analysis.

Growth rate
Algal discs (Fig. 1 inset) from each culture vessel were blotted dry and photographed by a 14-megapixels resolution digital camera (Lumix DMC-FT10; Panasonic, Osaka, Japan). Algal surface area (cm 2 ) was analyzed by comparing the pixel density of algal discs to a known reference area using the software ImageJ version 1.47b (National Institutes of Health, Bethesda, MD, USA, http://rsb.info. nih.gov/ij/). Relative growth rate (RGR; % d À1 ) was calculated from the surface area at the start and end of the experiment after L€ uning (1990).

Chlorophyll a fluorescence
The Chla fluorescence of each algal disc was measured submerged in their respective pH treatment using a PAM-2000 fluorometer (Walz GmbH, Effeltrich, Germany) following the protocol of Rautenberger et al. (2009) at 13 AE 1°C. After 5-min dark incubation, basic (F 0 ) and maximum fluorescence (F m ) was measured and F v /F m was calculated (Schreiber et al. 1995). Electron transport rate (ETR) vs. irradiance (E) curves (ETR-E curves) were recorded from discs exposed to incrementally increasing (every 30 s) actinic light (AL) intensities (7-500 lmol photons m À2 s À1 ; 400-700 nm). All saturation pulses were set to >9000 lmol photons m À2 s À1 and 0.8 s. The ETRs through the photosystem II (PSII) were calculated by multiplying the intensity of incident AL, the proportion of incident AL intensity that was absorbed by the measured disc, the fraction of absorbed AL which is most probably received by PSII (0.5) and the PSII operating efficiency (Baker 2008). Maximum electron transport rates (ETR max ), the initial slopes of ETR-E curves, called a, and the light saturation points of ETRs (i.e., E k ) were estimated from ETRs plotted against the incident AL intensity and calculated according to the model of Jassby and Platt (1976) Porra et al. (1989) and normalized to algal biomass (lg mg À1 FW).

Extracellular and intracellular carbonic anhydrase activity
Carbonic anhydrase activity was measured according to the pH drift method of Wilbur and Anderson (1948) and Haglund et al. (1992a) using 50 mmol L À1 Tris HCl buffer (adjusted to pH 8.5, 4°C), 2 mmol L À1 dithiothreitol (DTT), 15 mmol L À1 ascorbic acid and 5 mmol L À1 EDTA (disodium salt). The external carbonic anhydrase (CA ext ) activity was analyzed from frozen algal discs (60 AE 20 mg FW). Each disc was placed in a 20-mL scintillation vial containing 10 ml of the extraction buffer and equipped with a micro stirrer bar (10 9 3 mm) for the enzymatic reaction. The glass vial was placed inside an icecontaining 100-mL plastic container to maintain the temperature at 0-2°C, sitting on top of a magnetic stirrer to stir the solution. Temperature and pH were simultaneously measured using a ROSS electrode (Orion 8107BNUMD) coupled to Orion 3-Stars Plus pH Benchtop meter (Orion, Thermo Fisher Scientific, Waltham, MA, USA). When pH stabilized at 8.3, 5 mL of ice-cold CO 2 -saturated water was added. The time (sec) taken for the pH to drop by 0.4 units, in the pH interval of 8.3-8.1, was recorded. The internal carbonic anhydrase (CA int ) activity was subsequently measured from the same algal disc (Fern andez et al. 2014). The algal disc was ground to fine powder in liquid N 2 -frozen mortar and pestle. The ground tissue (60 AE 20 mg) was analyzed following the protocol described above. Relative enzyme activity (REA) was computed after the formula of Haglund et al. (1992b).

Carbon to nitrogen ratios and stable carbon isotope signatures
Molar C:N ratios and d 13 C signatures (&) were analyzed after algal discs were dried at 60 AE 1°C for 24 h. Discs were ground in a mortar and pestle to a fine powder and combusted in a CE NA1500 Elemental Analyzer (Carlo-Erba Instruments Ltd, Hindley Green, Wigan, UK) interfaced to an IRMS 20-20 continuous flow mass spectrometer (Europa Scientific Ltd, Crewe, UK). Corrections for drift were made automatically every five samples from an EDTA standard with a known isotope ratio.

Data analyses
Means AE standard deviations (SDs) were calculated from five (n = 5) or six (n = 6) measurements per treatment. Statistical analyses were performed using JMP Pro 10.0 (SAS Institute Inc., Cary, NC, USA) and R versions 2.7 and 2.15 (The R Foundation for Statistical Computing, http://www.R-project.org). Two-way ANOVA and t-test were used to identify statistical differences of the means within and between treatments. Tukey's honestly significant difference (HSD, P < 0.05) test was used as a post hoc test.

Results
Carbon physiology and in silico detection of putative CCM from EST libraries Under pH T 9.0, initial blocking of the direct HCO À 3 uptake through the AE port by DIDS resulted in 44% inhibition of the net photosynthetic rate (NPS) in U. rigida. Subsequent inhibition of the CA ext by AZ to arrest the catalyzed dehydration of HCO À 3 to CO 2 resulted in an additional 39% inhibition of photosynthesis. When the order of inhibition for HCO À 3 use was reversed, AZ and DIDS caused 29% and 26% of the NPS, respectively. Total inhibition of the NPS was higher when direct HCO À 3 uptake through the AE port was blocked first (Ttest, P = 0.0195; Fig. 2). Depending on the order of application of the inhibitors, 17-44% of NPS remained unaccounted.
In the absence of sequence information from U. rigida, the search for putative CCM elements in the EST libraries of U. prolifera (Jia et al. 2011) identified two a-CA and three c-CAs (Table S1). However, the putative targeting signal peptides could not be resolved because the sequences retrieved from the EST data were not full length. Moreover, the exact localization of these proteins (e.g. chloroplast or periplasm) could not be defined. No b-CA was found in the EST libraries of U. prolifera. Furthermore, four putative HLA3 ABC transporters and three chloroplast carrier proteins (CCP1) and five mitochondrial transporter protein were found (Table S1). The transcriptome of a second species Ulva linza also identified putative CCM genes (Zhang et al. 2012). Among those found are at least one a-CA localized in the chloroplast lumen, several low CO 2 -induced proteins, a chloroplast Ci transporter (LCIB), several ABC transporters as well as the nuclear transcriptional regulators of CCM elements (CIA5, LCR1).

Growth rate
Relative growth rates (RGR; Fig. 3) of U. rigida discs grown under limiting light (LL) were similar under low pCO 2 (LC; 8.5 AE 2.7% d À1 ) and high pCO 2 (HC; 9.1 AE 3.6% d À1 ). When discs were exposed to a saturating light (SL), RGRs were 2.279 and 2.359 higher under LC and HC, respectively (Fig. 3). Statistical analysis showed that RGRs were strongly influenced by the experimental irradiance (ANOVA, P < 0.001) but not by pCO 2 (ANOVA, P = 0.2473). The interaction between irradiance and pCO 2 did not significantly affect the growth rate (ANOVA, P = 0.5245).

Chlorophyll a fluorescence
The maximum quantum yield of PSII (F v /F m ) of U. rigida ranged from 0.807 to 0.813 (Table 2). Neither light (ANOVA, P = 0.152) nor the seawater carbonate chemistry (ANOVA, P = 0.391) or their interaction (ANOVA, P = 0.593) had an effect on F v /F m .
All electron transport rate-irradiance (ETR-E) curve parameters (Table 2), that is, ETR max , E k and a, were significantly different between light (ANOVA, P < 0.0001) but not significantly different between seawater carbonate chemistry (ANOVA, P > 0.05). No significant interactive effect of the independent variables was observed in any of the ETR-E curve parameter (ANOVA, P > 0.05). The ETR max and E k were significantly higher at saturating light (LL < SL; Tukey's HSD test, P < 0.05), while the photosynthetic efficiency, a, was significantly higher at low light (LL > SL; Tukey's HSD test, P < 0.05).

Pigments
All LL-grown discs had a significantly higher biomassbased content of both Chla and Chlb (Table 2) compared with SL-grown discs (ANOVA, P < 0.0001). The content of both chlorophylls was not affected by pCO 2 (ANOVA, P > 0 .05). Light and pCO 2 had an interactive effect on both Chla (ANOVA, P = 0.027) and Chlb (ANOVA, P = 0.026). When grown under SL, the Chla content was $ 35% lower than discs grown under LL. Similarly, the Chlb content in LL-grown discs was $ 50% higher than those grown in SL. The highest Chla/b ratio of 2.2 (LC and HC) were calculated for SL-grown discs, whereas LLgrown discs had lower ratios of 1.64 (LC) and 1.76 (HC).

Extracellular and intracellular carbonic anhydrase activity
The CA ext activities (Fig. 4A) ranged between 3.54 and 4.28 REA g À1 FW and were similar in all treatments (ANOVA, P > 0.05), whereas CA int activities (Fig. 4B) were 20% higher under SL compared with LL (ANOVA, P = 0.002) irrespective of pCO 2 . An increase in pCO 2 (ANOVA, P = 0.943) and the interaction between light and pCO 2 (ANOVA, P = 0.387) did not significantly change the CA int . Regardless of pCO 2 , the CA int was 1.5-29 higher compared with the CA ext under LL and SL, respectively.

Discussion
Our study showed that HCO À 3 is the primary exogenous Ci source used by Ulva rigida to support growth and photosynthesis under saturating light, supporting the findings of previous work on other Ulva species (e.g. Bj€ ork et al. 1992;Drechsler et al. 1993;Larsson et al. 1997). However, when the known HCO À 3 use mechanisms, that is, direct HCO À 3 uptake through the AE port and external catalyzed dehydration of HCO À 3 , in U. rigida were inhibited, net photosynthesis (NPS) decreased by only 56-83% leaving the carbon uptake mechanism(s) for the remaining 17-44% of NPS unaccounted for. The remaining production cannot be attributed CO 2(aq) as the concentrations under pH T 9.0 are much too low but rather to a possible light-dependent active HCO À 3 transport system. An in silico search of EST libraries of U. prolifera (Jia et al. 2011) found putative light-dependent HCO À 3 transporters, that is, the ABC transporters of the ABCC subfamily also found in U. linza (Zhang et al. 2012), which resemble the HLA3 transporter found in Chlamydomonas (Meyer and Griffiths 2013). The ABCC transporters are present in all eukaryotes and have different transport functions, including vacuolar sequestration of toxic metabolites and transport of chlorophyll catabolites during senescence, among others (Meyer and Griffiths 2013). The HLA3 analog has been implicated in the active bicarbonate uptake in Chlamydomonas: their role in Maximum quantum yield of PSII (F v /F m ), electron transport rate-irradiance (ETR-E) curve parameters, that is, ETR max , E k , and a estimated using the hyperbolic tangent model of Jassby and Platt (1976), pigments, and molar C:N of U. rigida under different light (LL and SL; limiting and saturating light, respectively) and pCO 2 concentrations (LC and HC; low and high pCO 2 , respectively). Data are means AE standard deviations (n = 6). FWfresh weight.
a CCM is supported by genetic and physiological evidence . Several CAs are involved in the utilization of HCO À 3 . For example, the a-CAs, pCA1 and pCA2, catalyze the external dehydration of HCO À 3 in Chlamydomonas (Spalding 1998); however, these isozymes are not found in Ulva. An unclassified CA ext isozyme may be present in Ulva, but their contribution to the external dehydration of HCO À 3 to CO 2 and the active and/or passive transport of CO 2 into the cell may not be sufficient to supply and fill the internal Ci pool to support the high growth rate of the species. Instead, two putative intracellular a-CA isozymes were found in the EST libraries of U. prolifera (Jia et al. 2011). These a-type CA isozymes also identified in U. linza (Zhang et al. 2012) most likely have a subcellular localization, for example, the chloroplast lumen. Both studies, that is, Zhang et al. (2012) and Ye et al. (2014), suggest that this a-CA isozyme is responsible for the internal HCO À 3 dehydration to provide the chloroplast with sufficient CO 2 for carbon fixation. The same was recently reported in Saccharina japonica, where cDNA encoding a-CA was associated with the chloroplast envelopes and thylakoid membranes (Ye et al. 2014). Although the b-type CA was not found in the EST of Ulva, it is reported to be localized in the cytosol of another green alga Coccomyxa (Huang et al. 2011) where it facilitates Ci diffusion from the inner surface of the plasmalemma to the chloroplast envelope.
On the other hand, the exact roles of c-type CA identified in the ESTs of U. prolifera are still unknown. They are usually associated with the mitochondria of green algae and plants (Parisi et al. 2004). Mitochondrial carbonic anhydrase (mtCA) is involved in enzymatic hydration of CO 2 , produced during respiration and photorespiration, to HCO À 3 to stimulate anaplerotic b-carboxylation, where sufficient supply of HCO À 3 is required to support nonphotosynthetic biosynthetic pathways (Giordano et al. 2003). This may explain the nonphotosynthetic growth enhancement reported in U. rigida (Gordillo et al. 2001). This mechanism that recovers respiratory CO 2 presents a very effective way to ensure efficient use of Ci for photosynthesis and nonphotosynthetic Ci-use pathways specially among calcifying algae and invertebrates (Roleda et al. 2012b).
The role of CA ext in the Ci use of U. rigida was insignificant and was not regulated by either pCO 2 or light. Despite the 160% increase in CO 2 under the HC treatment (1224 latm), downregulation in CA ext activity was not observed, a finding opposite to those found in the microalga Emiliania huxleyi (Richier et al. 2011). The constant low CA ext activities (3.54-4.28 REA g À1 FW) across all treatments suggest that CO 2 uptake after catalyzed external dehydration of HCO À 3 is not the main source of Ci in U. rigida. Considering that diffusive entry of aqueous CO 2 into the cell is very slow in water, this mechanism is deemed to be insufficient for the accumulation of an internal Ci pool large enough to support the observed high growth rates (max 20% d À1 ) in this species. Moreover, CO 2 uptake requires a conversion of CO 2 to HCO À 3 to maintain internal pH and avoid CO 2 leakage from the cell (Price et al. 2008).
Conversely, CA int was 1.5-29 higher under saturating light compared with limiting light. The higher enzymatic activity facilitates the conversion of HCO À 3 to CO 2 to support the higher CO 2 requirement of RuBisCO for photosynthetic fixation driving higher growth rate under saturating light. However, the source of this internal Ci pool (under high pH 9.0) cannot be wholly attributed to the known HCO À 3 uptake mechanism where only up to 44% of net photosynthesis is supported by the direct HCO À 3 uptake through the AE port and 39% by the CA ext -mediated HCO À 3 dehydration. The light-dependent HLA3 ABC transporters described above are hypothesized to most likely contribute to direct HCO À 3 transport in order to saturate the internal Ci pool. HCO À 3 is the preferred Ci form for cellular accumulation because it is about 1000-fold less permeable to lipid membranes than CO 2 (Price et al. 2008).
When active HCO À 3 transport through the AE port was blocked first, inhibition of NPS during the 15-min period was 34% higher than when CA ext -mediated HCO À 3 use was blocked first. Subsequent application of the second inhibitor contributed to additional 39% and 26% for the DIDS-AZ and AZ-DIDS treatment, respectively. In the AZ-DIDS treatment, where the CA ext -mediated HCO À 3 use was initially inhibited, algal discs were still actively transporting HCO À 3 through the AE port that enables the cells to accumulate higher internal Ci pool contributing to lesser total inhibition of NPS. Conversely, when active HCO À 3 transport through the AE port was blocked first (DIDS-AZ), significantly higher total inhibition of NPS was observed. This suggests that different HCO À 3 use mechanisms operate simultaneously and that the active transport through the AE port contributes more Ci to the internal pool.
Earlier studies of the carbon physiology of different Ulva species that focused on the AE port and external dehydration of HCO À 3 reported contradictory results. Larsson and Axelsson (1999) reported the net photosynthetic rates of different Ulva species are primarily supported by external dehydration of HCO À 3 (34-68%) rather than direct HCO À 3 uptake (9-40%) through the AE port. Another study by Axelsson et al. (1999) reported >90% of Ulva lactuca photosynthesis is dependent on AZ-sensitive HCO À 3 dehydration mechanism. Drechsler and Beer (1991) and Drechsler et al. (1993) reported a significantly higher contribution of direct HCO À 3 uptake compared with CA ext -catalyzed HCO À 3 dehydration in U. lactuca. The above-mentioned studies on HCO À 3 use mechanisms was measured between pH 8.2 and pH 8.7. At pH >9.4, only HCO À 3 uptake via the putative DIDSsensitive AE-transporter is operational in Ulva intestinalis (formerly Enteromorpha intestinalis) (Larsson et al. 1997). Conversely, only at extremely low pH 5.6 was a higher affinity for CO 2 observed, which had no adverse effect on U. lactuca's photosynthetic performance (Drechsler and Beer 1991). It should be noted that those Ulva species reported above may have some degree of taxonomical uncertainty. Previous physiological studies on Ulva did not identify that a light-dependent HCO À 3 transporter is most likely operational in Ulva's CCM as found in this study.
Elevated Ci (both CO 2 and HCO À 3 ) under OA did not cause a higher growth in U. rigida but growth rate was limited by light. This suggests that the present-day Ci concentration is already saturating for Ulva. Cyanobacterial ancestors of the green algae evolved effective HCO À 3 use mechanisms during the geologic low-CO 2 environment (Giordano et al. 2005); we suggest that this trait is most likely genetically fixed and the modern Ulva can modulate their CCM under different pCO 2 conditions (Giordano et al. 2005). Our results do not support the observation of Gordillo et al. (2001) on the same species and Xu and Gao (2012) on U. prolifera that higher pCO 2 causes an increase in rates of both photosynthesis and growth. Although the response may be species specific, the mechanistic study of Gordillo et al. (2001) used extremely high CO 2 in air (10 000 lL L À1 ) and the identity of their U. rigida is uncertain as this was not molecularly identified (F. J. L. Gordillo, pers. comm.). The results of our study concur with those of Drechsler and Beer (1991) on U. lactuca where maximum photosynthetic O 2 evolution was comparable under low (5.6) and high (8.2) pH. Furthermore, the study of Mercado et al. (2001) on the carbon physiology of three red seaweeds of the order Gelidiales inhabiting the intertidal zone also reported that photosynthesis is limited by light and not by Ci availability. The seawater Ci species, that is, CO 2 and HCO À 3 , used as substrate for carbon uptake and fixation consist of stable carbon isotopes 13 C and 12 C within each species. The present seawater consist of 1% CO 2 and 91% HCO À 3 ; of all natural carbon, only 1.1% is 13 C while 98.89% is 12 C. Relative to 13 d VPDB standard, CO 2(aq) has d 13 C = -10& while dissolved HCO À 3 has d 13 C = +1& to +1.5& (Mook et al. 1974;Mook 2005). The 13 C/ 12 C ratios (= d 13 C) of the organic cellular material has been used as a proxy of Ci use relative to bulk seawater Ci source (Giordano et al. 2005). For example, organisms with d 13 C higher than -10& (a value more positive than d 13 C of CO 2 in seawater) must involve HCO À 3 use (Raven et al. 2002). Among mixed CO 2 /HCO À 3 using algae, a significant use of CO 2 under high pCO 2 will shift d 13 C signature corresponding to more CO 2 use (i.e. toward -30&) (Maberly et al. 1992). Here, we found that regardless of pCO 2 , d 13 C signatures of U. rigida shifted upwards toward values of -10& and higher, under saturating light, which provides further support of the presence of a light-dependent active HCO À 3 transport and intracellular accumulation, and the use of available HCO À 3 to support photosynthetic carbon fixation as previously suggested by Raven et al. (2002).
The C-isotope fractionation in aquatic plants is more complex than in terrestrial plants (Hoefs 2009). Factors that control the d 13 C signature in algae include not only the availability of CO 2(aq) , but also temperature, light intensity, nutrient availability, pH, and physiological factors such as cell size and growth rate (Hoefs 2009 and references therein). In U. rigida, the increase in productivity observed under saturating light regardless of increased availability of CO 2(aq) causes a corresponding rise in d 13 C values, a response associated with more 12 C locked up in the tissue as organic matter is generally depleted in 13 C (Zeebe and Wolf-Gladrow 2001).
Moreover, the carbon fixation pathway can also influence the isotopic composition of organic matter. d 13 C signatures between -32% and -22% are characteristics of C 3 plants while d 13 C between -16& and -10& are typical for C 4 plants (Zeebe and Wolf-Gladrow 2001;Hoefs 2009 and references therein). The natural variations in the d 13 C signature of U. rigida between -22& and -10& point to the possible occurrence of a C 4 photosynthetic carbon fixation pathway, as observed in U. linza (Xu et al. 2013).
The rising atmospheric CO 2 does not only trigger OA but also contributes to global warming that strengthens the vertical stratification of aquatic ecosystems: this suppresses the nutrient supply from deep water into the surface layer. The enhanced CO 2 but reduced nutrient supply can therefore increase the C:N ratio of primary producers (e.g., phytoplankton) which are of low nutritional value to consumers (e.g., zooplankton), cascading throughout the entire aquatic food web (van de Waal et al. 2010). In our experiment, we increased CO 2 bỹ 160% and HCO À 3 by~9% while the nutrient level remained constant. In this scenario, the enhanced exogenous Ci concentration and constant nutrient supply will theoretically still increase the C:N ratio. However, the insignificant fractional increase was observed relative to light and not to CO 2 (Table 2). This suggests that changes in the tissue/cellular stoichiometry in macroalgae may not be sensitive to a change in Ci alone. The small increase in C:N under saturating light suggests that exogenous Ci concentration is already saturating for Ulva regardless of pCO 2 and saturating light increase carbon fixation.
The maximum quantum yield of PSII (F v /F m ) and other photosynthetic parameters (ETR max , E k , a) are reliable proxies to assess seaweed photosynthetic performance under environmental stress such as high PAR, UVR, and temperature (e.g., Roleda et al. 2005;Rautenberger and Bischof 2006;Hanelt and Roleda 2009;Rautenberger et al. 2009;Roleda 2009). However, we suggest that these parameters are unlikely to be sensitive to changes in seawater carbonate chemistry, that is, pCO 2 and H + , within the range likely to occur due to OA. In this study, rigorous PAM-based photophysiological measurements on U. rigida showed that ETR max , E k , and a are regulated by light, and a reduced seawater pH, simulating OA, had no effect, a finding contrary to that of Olischl€ ager et al. (2013). Moreover, and contrary to the reported decrease in chlorophyll pigments under OA in U. prolifera (Xu and Gao 2012), OA did not affect the photosynthetic pigments of U. rigida: the amounts of Chla and Chlb were regulated by light.
Algae regulate internal pH maintaining cytoplasmic pH at 7.3 AE 0.2 (Ritchie 1985;Lundberg et al. 1989;Smith and Bidwell 1989) which is 0.7 units lower than that of the current average surface seawater pH. Moreover, PSII is located in the thylakoid membrane where it is exposed to the acidified lumen (pH 5.0) and neutral to slightly basic stroma (pH 7.2-8.0) (Falkowski and Raven 2007). Therefore, the photosynthetic apparatus is already acclimated to a wide range of pH. The mechanism of how the bulk water pH may affect F v /F m is unknown; studies reporting positive or negative effects of OA on this physiological proxy should therefore be interpreted with caution as they are possible artifacts.
In conclusion, U. rigida is insensitive to OA. The present-day seawater Ci pool is saturating for photosynthesis and growth, and these parameters were primarily controlled by light rather than elevated CO 2(aq) . For photosynthetic carbon fixation, HCO À 3 is the primary Ci species assimilated by U. rigida. Aside from the known catalyzed external HCO À 3 dehydration and direct HCO À 3 uptake through the AE port, another inhibitor-insensitive HCO À 3 transport mechanism is most likely present. An in silico search of CCM elements in EST libraries of Ulva found putative light-dependent HCO À 3 transporters, that is, the ABC transporters of the ABCC subfamily, found in both U. prolifera and U. linza. Neither a downregulation in extracellular CA-mediated HCO À 3 dehydration nor a shift to CO 2 use was observed under high CO 2(aq) . The shift in d 13 C signatures in U. rigida toward -10& under saturating light under low and high CO 2(aq) but not toward -30& under elevated CO 2(aq) suggests preference and substantial internal HCO À 3 accumulation to support photosynthesis and growth regardless of CO 2 concentrations. Despite the limited effect of OA and PPFR, the interaction of OA with other climate change stressors, for example, eutrophication and warming, may elicit different effects and warrants further investigation.