High CO 2 decreases the long‐term resilience of the free‐living coralline algae Phymatolithon lusitanicum

Abstract Mäerl/rhodolith beds are protected habitats that may be affected by ocean acidification (OA), but it is still unclear how the availability of CO 2 will affect the metabolism of these organisms. Some of the inconsistencies found among OA experimental studies may be related to experimental exposure time and synergetic effects with other stressors. Here, we investigated the long‐term (up to 20 months) effects of OA on the production and calcification of the most common mäerl species of southern Portugal, Phymatolithon lusitanicum. Both the photosynthetic and calcification rates increased with CO 2 after the first 11 months of the experiment, whereas respiration slightly decreased with CO 2. After 20 months, the pattern was reversed. Acidified algae showed lower photosynthetic and calcification rates, as well as lower accumulated growth than control algae, suggesting that a metabolic threshold was exceeded. Our results indicate that long‐term exposure to high CO 2 will decrease the resilience of Phymatolithon lusitanicum. Our results also show that shallow communities of these rhodoliths may be particularly at risk, while deeper rhodolith beds may become ocean acidification refuges for this biological community.


2015)
. Coralline algae are of particular interest to investigate as they conduct both photosynthesis and calcification simultaneously .
Beds of unattached red coralline algae (mäerl/rhodoliths) are important carbon sinks and nursery grounds for commercial species, being also major players in the global carbon cycle through the production of CaCO 3 sediment (van der Heijden &  McCoy . These habitats are found from the poles to the tropics, and can form extensive beds of 20,900 km 2 in the tropical south-west Atlantic (Amado-Filho et al., 2012). In spite of their global distribution and importance, rhodolith beds have received less attention with respect to other more geographically restricted ecosystems such as coral reefs. Mäerl beds are globally distributed habitats especially sensitive to OA and global warming because of their slow growth rates and the high solubility of their high-Mg calcite skeletons (McCoy & Kamenos, 2015;Williamson, Najorka, Perkins, Yallop, & Brodie, 2014). In a future ocean, filamentous algae are expected to be better competitors than coralline algae (Short, Kendrick, Falter, & McCulloch, 2014). The degradation and replacement of mäerl beds by fleshy algae communities is likely to affect the world's carbon cycle, with wide repercussions, including the fishery industry .
Laboratory experiments have revealed mixed and sometimes contradictory responses of the photosynthesis, calcification, and respiration of coralline algae to OA, even among closely related species. Positive, negative, and parabolic responses have been observed , hampering the withdrawal of reliable conclusions and thus complicating the prediction of future consequences of OA on coralline algae. The variability in results have been attributed to length of experimental exposure (see Ragazzola et al., 2012Ragazzola et al., , 2013, differences in algal physiology, methodologies used (gas bubbling vs. acid-base additions, e.g.,), and/or pre-experimental period (see Hurd et al., 2009).
Several authors have highlighted the importance of long-term experiments in evaluating the responses of coralline algae to OA (e.g., Martin et al., 2013;McCoy & Kamenos, 2015;Ragazzola et al., 2013). Long-term studies can reveal very different results with respect to short-term ones (McCoy et al., 2016) and provide important information on the potential for physiological acclimation (see Hurd et al., 2009;Martin et al., 2013;Ragazzola et al., 2013).
Nevertheless, the response of coralline algae to OA has mainly been investigated through short-term experiments .
The longest OA experimental studies conducted with coralline algae lasted for a year (Martin & Gattuso, 2009;Martin et al., 2013), using the species Lythophyllum cabiochae (crustose morphology). In contrast, the longest experiments with mäerl species ranged from 1 to 3 months (e.g., Burdett et al., 2012;Kamenos et al., 2013;Noisette, Duong, Six, Davoult, & Martin, 2013;Noisette, Egilsdottir, Davoult, & Martin, 2013;Ragazzola et al., 2012) to a maximum of 10 months (see Ragazzola et al., 2013). Ragazzola et al. (2013) compared the responses of the species Lithothamnion glaciale (mäerl morphology) after 3 and 10 months of high-CO 2 treatment and obtained different results with time. After 3 months, algae cultivated under high CO 2 maintained their growth rate but reduced their cell wall thickness, while, after 10 months, the cell wall thickness was maintained, but there was a reduction in the growth rates. The authors concluded that long-term experiments provide a better analog for understanding the organism's response to OA and a more accurate projection of future impact. Long-term experiments, where the synergistic effects of light, temperature and other factors are also investigated, are essential to understand how ecosystems will respond to global change.
Few studies have investigated the complex and tightly linked processes of photosynthesis, calcification, and respiration on coralline algae (e.g., Martin et al., 2013;Noisette, Duong, et al., 2013;Noisette, Egilsdottir, et al., 2013). These three metabolic processes are able to alter the pH of seawater, thus modifying the carbon speciation (Hurd et al., 2009). The increase in pH during photosynthesis increases the CaCO 3 saturation state and consequently fosters calcification, while respiration decreases pH, hindering calcification (see Gao et al., 1993;Martin et al., 2013). A better understanding of how these processes are inter-related under an OA scenario is required to elucidate how mäerl species will respond to global environmental changes (Hurd et al., 2009;Martin et al., 2013).
The current challenge in OA research is to understand how whole ecosystems react to a range of climate-related stressors (Dupont & Pörtner, 2013a). Recent model projections have demonstrated that regionally distinct patterns of complex oceanic change are evident, and considerations of multistressor patterns can help to guide laboratory and field studies (Boyd, Lennartz, Glover, & Doney, 2015). However, most experimental studies do not consider the interactions encompassing environmental factors such as irradiance, temperature, or nutrients, which are determinant on the algae's respiration, photosynthesis, and calcification. Thus, more realistic manipulations, in which natural fluctuations in these factors are included, will reveal more true-to-life responses (Boyd et al., 2016).
Long-term field experiments incorporating environmental variability (e.g. McCoy, 2013), as well as organismal adaptation, may provide unparalleled insights into the effects of global change (Kroeker, Kordas, & Harley, 2017) on coastal ecosystems. Moreover, the better understanding of the environmental effects in combination with the sublethal effects reported in laboratory experiments will help to forecast the long-term ecological effects of global change (Kamenos, Perna, Gambi, Micheli, & Kroeker, 2016). Still, experimental studies are not usually complemented by the monitoring of the seasonality of natural communities, leading to an important gap between experimental and community-scale studies (McCoy & Kamenos, 2015).
The synergistic effects of environmental stressors such as temperature (see Vásquez-Elizondo & Enríquez, 2016) and high irradiance (Gao & Zheng, 2010;Yildiz, Hofmann, Bischof, & Dere, 2013) are expected to aggravate and/or accelerate the detrimental effect of OA, and are determinant factors for mäerl/rhodolith beds, especially for those located at more shallow areas and intertidal zones.
Mäerl beds in Algarve (southern Portugal) are mainly composed of unattached, monospecific branches of the nongeniculate coralline red algae Phymatolithon lusitanicum (Peña et al., 2015). Recent molecular analyses have confirmed P. lusitanicum as a new species (Peña et al., 2015). The species Lithothamnion corallioides and Phymatolithon calcareum are also present on the Algarve mäerl beds, but their occurrence is rare (Carro, López, Peña, Bárbara, & Barreiro, 2014;Peña et al., 2015). These two species are especially abundant in the British Isles and Brittany, but they have been gradually replaced by Phymatolithon lusitanicum in Galicia (NW Spain) (Carro et al., 2014). P. lusitanicum is particularly abundant in subtidal mäerl beds of the Atlantic Iberian Peninsula, in Galicia (4-13 m) and Algarve (15-25 m), but it has also been detected in Ireland (intertidal zone to 6 m), in the Alborán Sea (40-48 m) and in the Balearic Islands (54-64 m) (Peña et al., 2015). The largest rhodoliths of P. lusitanicum in the Iberian Peninsula have been found in the southernmost locations and show their thickest morphology in Algarve (Carro et al., 2014; lusitanicum referred as Phymatolithon sp3 in Peña et al., 2015).
The effect of OA on the mäerl species from southern Portugal is so far unknown. The objective of this study was to investigate, for the first time, the long-term (20 months) effects of OA on the photosynthesis and calcification of Phymatolithon lusitanicum (Peña et al., 2015), the main species of the most southwestern mäerl/rhodolith beds in Europe. The effect of OA on respiration was also assessed.

| Biological material
The thalli were collected by SCUBA diving in Armação de Pêra, Faro

| Experimental system
The experimental system used in this study is described in detail in Sordo, Santos, Reis, Shulika, and Silva (2016). Briefly, seawater is pumped from an adjacent coastal lagoon in front of CCMAR field station and passes through a preliminary mechanical filtration and through two in-line filters of 10-20 and 5 μm before entering a preliminary 2,000 L reservoir. Here, seawater is continuously bubbled (compressed air at ambient conditions) with air stones to reach equilibrium with the atmosphere and further routed through two 16 and 8W UV filters before being distributed into three 200 L head tanks.
The CO 2 -air mix bubbling system is located in an isolated container where photoperiod and temperature are adjusted seasonally.
This system works in an open circuit with low water flow rates. CO 2 injection is controlled by a solenoid valve coupled to an Infrared Gas Analyzer (WMA-4 IRGA; PPSystems, USA), a PID controller (PID330; TEMPATRON, UK) and an air-flushing equilibrator (Frankignoulle, Borges, & Biondo, 2001). A mix of food-grade CO 2 and ambient air is injected into the three 200 L head tanks using an air pump. This is a novel OA experimental design, conceived to run long-term experiments (Sordo et al., 2016).

| Experimental design
Based on the CO 2 emission perspectives of the International Panel on Climate Change (IPCC IS92) (IPCC, 2000), two enrichment levels of pCO 2 = 550 μatm (2050 projected scenario) and pCO 2 = 750 μatm (2100 projected scenario) were simultaneously tested together with a control level without enrichment (pCO 2 ~ 400 μatm). A total of 270 thalli (90 per pCO 2 level) were labeled with numbered plastic tags attached with nylon fishing wire and randomly distributed by the eighteen 25-L aquaria and maintained at the different CO 2 levels for 20 months. The aquaria were regularly cleaned to control the growth of turf algae. The growth of the thalli was followed during the experiment using the nondestructive buoyant weight technique.
In addition, unlabeled thalli were kept in the aquaria for metabolic measurements.
Photoperiod was adjusted using a timer to the desired L:D (light and dark, hr) according to natural fluctuations. It varied from 10:14 in December to 15:9 in June. The ambient light source consisted of two 12W led strings, green and white, above the aquariums. The photosynthetic photon flux density was kept at ca. 8 μmol photons m −2 s −1 .
These values were calculated based on field-collected data from an annual sampling cycle. Air and water temperatures were controlled with an AC apparatus and water chillers (Sunsun HYH-0.5 D-C, China).

| Seawater parameters
The seawater and air temperature were monitored every 5 min using HOBO temperature loggers (Onset Corp.). Salinity (CO310 conductivity meter, VWR, USA), pH (Orion 8103SC pH meter, Thermo scientific, USA), temperature (Roth digital thermometer, Hanna, EU), and dissolved oxygen (Symphony SB90M5, VWR, USA, accuracy ± 0.2 mg/L; ±2%) were measured in each aquarium on a biweekly basis. Total alkalinity (TA) was measured at different points of the system: the source 2,000 L tank, the three head 200 L tanks, and the 25 L aquariums where the algae were kept. The pCO 2 at each treatment was recorded every 20 min by a gas analyzer (WMA-4; PPSystems, USA).
In parallel to this study, we continuously monitored the temperature and PAR levels in the natural sampling site using autologging sensors. During the experiment, temperature was gradually adjusted to the values observed under natural conditions in the field.

| Photosynthesis, respiration, and calcification
Net photosynthesis, respiration, and calcification rates of algae exposed to the different pCO 2 treatments were determined three times during the experiment (at 0, 11, and 20 months of treatment) through short time incubations. After 11 months (March 2013), incubations were conducted at different irradiance levels (0, 8, 45, 195, and 450 μmol photons m −2 s −1 ), and the effect of high CO 2 on the respiration and photosynthetic rates of algae was determined. Light intensities were selected based on the average levels of PAR determined in the natural beds and on parameters from previously determined light-response curves for this species (Sordo et al., 2016). At Preweighted algae individuals were placed in 500 ml Erlenmeyer flasks (n = 4) filled to the top and sealed to prevent gas loss by leakage. For the photosynthetic and calcification measurements, 20 g of mäerl was incubated for 1 hr per light intensity and CO 2 level. Dark respiration (R d ) and calcification (G d ) were measured using 40 g of mäerl incubated for two and a half hours per CO 2 level. To guarantee robust respiration results, the algae were left in the dark for 30 min prior to the incubations.
Samples for dissolved oxygen and total alkalinity (TA), to calculate the calcification rates of algae (Steller, Hernandez-Ayon, Riosmena-Rodriguez, & Cabello-Pasini, 2007), were collected at the beginning and at the end of the incubations. Temperature, pH, and salinity were measured. Dissolved oxygen was analyzed by direct spectrophotometry as described in Labasque, Chaumery, Aminot, and Kergoat (2004). Briefly, a calibration curve was performed with six different concentrations of KIO 3 (0, 40, 80, 160, 240 and 320 μmol/L). Cl 2 Mn 3 mol/L, NaOH 8 mol/L, and NaI 4 mol/L were added to the calibration curve as well as to the water samples right after collection. After acidification with H 2 SO 4 10 mol/L, the io- The alkalinity (TA) anomaly technique (see Smith & Key, 1975) was used to determine the calcification rates (Steller et al., 2007).
When 1 mole of CaCO 3 precipitates, TA decreases in the ratio of 1:2 (see Wolf-Gladrow, Zeebe, Klaas, Koertzinger, & Dickson, 2007). Total alkalinity was determined using the Gran titration method as in Lewis and Wallace (1998). Seawater was sampled at the beginning and end of the incubations and at different points Oceanography, USA). The carbonate chemistry of seawater samples was determined from measured pH, TA, temperature, and salinity using the software CO 2 SYS (Lewis & Wallace, 1998) with the constants of Mehrbach, Culberson, Hawley, and Pytkowicz (1973) (refitted by Dickson & Millero, 1987), that is, dissolved inorganic carbon (DIC) and saturation state of aragonite (the saturation state of high Mg-calcite is closer to aragonite than calcite). Light and dark calcification rates (G) were calculated from the increment in total alkalinity, normalized by the incubation time, the volume of the chamber, and the fresh weight of the incubated thalli according to the formula described in Steller et al. (2007): where ΔTA is the difference between initial and final total alkalinity (μmol CaCO 3 /L), V is the incubation volume (L), Δt is the incubation time (hours), and FW is the fresh weight of the sample (g).

| Calcium carbonate and organic matter content (HCl acidification)
After 20 months of high CO 2 exposure, organic matter and CaCO 3 composition in the algae were determined by eliminating CaCO 3 through acidification (see Steller et al., 2007). Twenty-four individual rhodoliths per CO 2 level (n = 24) were weighed and then dried to constant weight. After this step, each rhodolith was placed in 25 ml of 5% HCl which was changed every 24 hr until no bubbles were observed and all CaCO 3 was dissolved. Samples were rinsed with distilled water and dried to constant weight in a drying oven at 60°C.
Organic matter and CaCO 3 were determined by subtraction from the preacidified dry weight.

| Calcification/growth (buoyant weight technique)
The calcification/growth of algae was determined using the buoyant weight (BW) technique, firstly proposed for coral growth by Jokiel, Maragos, and Franzisket (1978) and also used in coralline algae (e.g., Short et al., 2014 andSteller et al., 2007). A total of 270 thalli (90 per CO 2 level) were tagged at the beginning of the experiment. The individual weight of each thallus was determined after 7, 12, 16, and 20 months, by suspending the sample in a beaker filled with filtered seawater by a nylon string attached to an electronic balance (Sartorius 0.1 mg).
The weight of CaCO 3 at each sampling time (Wcc) was calculated as (Steller et al., 2007); where Wcc is the dry weight of the CaCO 3 , Wb is the buoyant weight of the algae, Dw is the density of the seawater (1.03 g/cm 3 ), and Dcc is the density of CaCO 3 (2.71 g/cm 3 ). Replacing the densities of seawater and CaCO 3 , the following equation is obtained; The total growth was calculated by dividing the difference between the initial and the final buoyant weight by the initial weight of each replicate. The calcification rate was calculated by dividing the total growth by the number of days between measurements.

| Determination of photosynthetic pigments
Samples for pigments were collected at the end of the experiment.
Algae were frozen in liquid nitrogen and stored at −80°C until analysis. For chlorophylls and carotenoids, approximately 1.5 g FW was ground in liquid nitrogen, immediately extracted in 5 ml of 100% acetone and centrifuged for 5 min at 2,000 ×g and 4°C.
Chlorophylls a and d concentrations were calculated according to Ritchie (2008); where:  The Pearson correlation coefficient was used to assess the linear dependence between photosynthesis and calcification after 11 months , and between growth rate and total growth with CO 2 and time.

| Statistical analyses
The effect of CO 2 on the pigment concentrations was analyzed using a one-way ANOVA. When significant differences were found, a post hoc test (Student-Newman-Keuls, SNK) was applied to explore differences between treatments. As carotenoid data presented equal variance but did not meet the assumptions of normal distribution (Shapiro-Wilk test), data went through a square root transformation prior to analysis.

| RE SULTS
Salinity, temperature, and dissolved oxygen did not differ among CO 2 treatments. During the 20 months of the experiment, salinity ranged from 34 to 38 psu and temperature fluctuated in a similar way as the natural beds (14-20°C) (Figure 1). Dissolved oxygen at the mesocosms remained constant (~6.7 mgO 2 /L). Total alkalinity (TA) was identical at different points of the experimental system both in CO 2 and control treatments and ranged between 2,449 and 2,536 μmol/kgSW (Table 1).

The mean CO 2 values under control conditions ranged from 397
to 470 μatm of CO 2 , while the mean values for the intermediate CO 2 treatment ranged from 539 to 577 μatm of CO 2, and the mean values for the high CO 2 treatment ranged from 740 to 804 μatm of CO 2 ( Figure 1 and Table 1).

| Effects of CO 2 and time of exposure on photosynthesis and of CO 2 on respiration. Photosynthesis and calcification relationship
Photosynthetic rates increased significantly with increasing light

| Effects of CO 2 and time of exposure on calcification
After 11 months of high CO 2 exposure, the calcification rates of The calcification rates of Phymatolithon lusitanicum maintained at low light levels of 8 μmol photos m −2 s −1 decreased significantly TA B L E 1 Carbonate chemistry for each CO 2 concentrations (400, 550 and 750 μatm) after 11 and 20 months of high CO 2 . Total alkalinity (TA), salinity (Sal.), temperature (T) and pH were measured while dissolved inorganic carbon (DIC) and aragonite saturation state (Ω arag ) were calculated using CO 2 SYS software. Values expressed as means ± standard error (n = 11)  These results suggest that both CO 2 and time of exposure have a negative effect on the growth/calcification of these algae.

| Effect of CO 2 on the photosynthetic pigment concentration
Even if some bleaching was observed on the algae exposed to high CO 2 (personal observations, not quantified), the photosynthetic F I G U R E 2 Phymatolithon lusitanicum. Photosynthesis, respiration, and calcification after 11 months of exposure at three different CO 2 levels (400, 550 and 750 μatm) (a) Dark respiration and net photosynthesis as a function of light (8, 45, 195, and 450 μmol photons m −2 s −1 ), (b) net calcification as a function of light (0, 8, 45, 195, and 450 μmol photons m −2 s −1 ) results calculated using the total alkalinity anomaly technique, and (c) correlation of net photosynthesis and light calcification at 200 μmol photons m −2 s −1 under control and high CO 2 conditions. Different letters indicate significant differences with CO 2 and irradiance. Values are expressed as mean ± standard error (n = 4) F I G U R E 3 Net photosynthesis (μmolO 2 gFW −1 hr −1 ) of Phymatolithon lusitanicum at ~200 μmol photons m −2 s −1 (PAR) incubated for 0, 11, and 20 months under three different CO 2 levels (400, 550, and 750 μatm). Different letters indicate significant differences with CO 2 and time. Values expressed as mean ± standard error (n = 4) pigments were not affected by CO 2 concentration (Table 2).
Phymatolithon lusitanicum presented higher concentrations of phycoerythrin than phycocyanin. Chlorophyll d, a pigment present in some red algae adapted to low light conditions, was not detected.

| D ISCUSS I ON
Our study shows the importance of establishing long-term experiments to detect thresholds of resistance to OA, and reveals that time of exposure must be considered to predict how calcified macroalgae will respond to elevated CO 2 . To our knowledge, this is the longest OA experiment with mäerl coralline algae (20 months). Previous OA short time experiments with coralline algae have shown a variety of contrasting responses such as that growth rates decreased with high CO 2 , after 20 hr (Gao et al., 1993) or 1 month (Büdenbender, Riebesell, & Form, 2011;Hofmann, Yildiz, Hanelt, & Bischof, 2012;Johnson & Carpenter, 2012), and that this decrease is intensified under high light irradiance events (Gao & Zheng, 2010;Yildiz et al., 2013) or exacerbated with warning (Büdenbender et al., 2011;Johnson & Carpenter, 2012; see Martin & Hall-Spencer, 2017 for review). On the other hand, other short time experiments have found no effect on the gross production, calcification or respiration after 3 weeks at different CO 2 concentrations (Egilsdottir, Noisette, Noël, Olafsson, & Martin, 2013), or an increase in the coralline algae's growth rate (Ragazzola et al., 2013). High CO 2 also seems to decrease the Mg concentrations on the carbonate that may reduce elasticity, thus changing the structural properties of algae (Ragazzola et al., 2016), and this decrease is exacerbated with temperature (Nash, Martin, & Gattuso, 2016).
Phymatolithon lusitanicum showed positive responses to high CO 2 during the first year, but after 2 years, control algae presented the highest rates and accumulated growth with respect to acidified algae. During the first 11 months, both photosynthetic and calcification rates increased with CO 2 . However, after 20 months of treatment, photosynthesis and calcification decreased at both concentrations of 550 and 750 μatm. Similar CO 2 /time of exposure limits have been found with other coralline algae species, for example, after 1 year (Martin & Gattuso, 2009;Martin et al., 2013) three months (Noisette, Duong, et al., 2013) or between 3 and 10 months (Ragazzola et al., 2013). Previous research suggests that in a future F I G U R E 4 (a) Calcification/growth rates (μmolCaCO 3 gFW −1 day −1 ) and (b)  high CO 2 world, coralline algae will be able to sustain high production rates for some time, but losses on their structural integrity are likely to happen (Burdett et al., 2012;Kamenos et al., 2013;Martin et al., 2013;Ragazzola et al., 2012). The calcification under lowering saturation state of calcium carbonate represents additional energetic costs that may not be sustained in the long term, in spite of increasing photosynthetic rates, resulting in a decrease in the general resistance of the organism, facilitating disease development and bleaching and reducing the algae ability to resist boring by predators (Ragazzola et al., 2012) or to compete with fleshy algae (Short et al., 2014).
After 11 months of OA, the observed increase in both calcification and photosynthetic rates may be considered a mechanism to compensate low pH and carbonate saturation state levels. Kamenos et al. (2013) demonstrated that the structure of coralline algal is more sensitive to rate rather than magnitude of acidification. At low pH, coralline algae survived by increasing their calcification rates, but the authors detected weakness in the calcite skeletons. Even if acidified algae calcified more to compensate OA, they were under a constant dissolution/calcification process (see Dupont & Pörtner, 2013b) and grew less in the long term (20 months). We found that after the first 11 months, calcification and photosynthesis grew together; however, this correlation was stronger under control conditions, and in the long term, the net production, growth rates, and total growth of algae decrease with CO 2 .
Previous research under natural conditions has shown that in the northeast Pacific, crustose coralline algae (CCA) competitive interaction and community structure have changed over 30 years due to ocean acidification (McCoy, 2013). By examining morphological change and comparing individuals of four CCA species that coexist in this area over time, McCoy and Ragazzola (2014) found that all of them maintain less skeletal material than in the past but the specific response depends on the algae's morphology. Because results change not only with CO 2 , but also with time of exposure, long-term experiments are mandatory to detect and predict the real effect that OA will have on coralline algae. So, even if algae are able to temporarily compensate the low pH by increasing their rates, they also dissolve more, and in the long term, this counterpoint has a negative effect on their overall growth.
Contrary to our observation that Phymatolithon lusitanicum respiration decreased with higher CO 2 , most authors have found that coralline algae's respiration rates do not change with high CO 2 Noisette, Duong, et al., 2013;Noisette, Egilsdottir, et al., 2013). In a previous study with Phymatolithon lusitanicum, we also found that respiration was unaffected by CO 2 but positively affected by temperature (Sordo et al., 2016). Even if we included the natural fluctuations in temperature, these fluctuations were small and the effect of temperature was not investigated during this study. It is, however, known that the effect of ocean acidification is exacerbated with warming (Martin & Hall-Spencer, 2017), and according to Vásquez-Elizondo and Enríquez (2016), coralline algae physiology can more adversely affected by temperature than by CO 2 . However, few studies have investigated the simultaneous effect of temperature and high CO 2 on the photosynthesis, respiration, and calcification of coralline algae (e.g., Martin et al., 2013;Noisette, Duong, et al., 2013;Noisette, Egilsdottir, et al., 2013).
This study emphasizes the need for more long-term OA studies with coralline algae, where other variables can be tested simultaneously. We showed that results changed with time of exposure to high CO 2 . In the first 11 months, acidified algae increased their photosynthetic and calcification rates to compensate the low saturation state of the seawater. But after a 20-month exposure, acidified algae could not compensate the corrosive condition of seawater and decreased both their photosynthetic and calcification rates. At the end of the experiment, acidified algae presented the same production rates than during the initial measurements but lower calcification rates, while control algae kept constant rates during the whole experiment and presented the highest accumulated growth.
The future outcome of mäerl beds from southern Portugal to OA will depend on other factors involved that are not easy to predict.
Phymatolithon lusitanicum may survive with minor production losses until intermediate CO 2 concentrations are reached (550 μatm, IPCC projections for 2050), but in the next century (750 μatm, IPCC projection for 2100), the species resilience to OA may decline seriously, and shallow population may be particularly susceptible to increasing high CO 2 . Deeper areas where rhodolith beds are exposed to low light may become ocean acidification refuges for this biological community.

ACK N OWLED G M ENTS
We would like to thank Silvia Albano, Bruno Fragoso, and Amrit Kumar Mishra for their help during the incubations. This paper is a contribution to the FCT project PTDC/MAR/115789/2009 funded by Fundação para a Ciência e a Tecnologia (FCT) (jmsilva@ualg.pt).
The first author, LS, was supported by the FCT doctoral grant SFRH/ BD/76762/2011. All of the undersigned authors participated in the study, and none has any potential conflict of interest.

CO N FLI C T O F I NTE R E S T
None declared.