Introduction

Ocean acidification is caused by seawater absorption of anthropogenically-derived CO21, and is a major threat to many marine ecosystems through alterations in organism physiology and changes in ecological interactions2. Ocean acidification will impact organisms via increased dissolution rates of calcareous species3,4, altered behaviour of marine invertebrates and fish2, and by increasing the supply of dissolved inorganic carbon (DIC) to some autotrophs5,6,7. These effects will be particularly strong at an ecosystem level if they affect either foundation or keystone species2,8,9. Macroalgae provide food and habitat to tens of thousands of species in shallow-water regions throughout the world10,11, and all utilise external CO2 and/or bicarbonate (HCO3) for photosynthesis, with CO2 being the ultimate substrate for the enzyme RuBisCO12,13 which fixes inorganic carbon. Calcareous macroalgae can be directly impacted by ocean acidification due to increased dissolution and reduced calcification rates7,14,15,16. Indirectly, calcareous macroalgae face increased competition from non-calcareous macroalgae (hereafter “fleshy macroalgae”) that could benefit from elevated CO217,18,19. Coralline algae are predicted to be at particular risk from ocean acidification14,20,21; they provide substrata for the settlement of marine invertebrates and also create and bind together reefs in shallow waters from the poles to the tropics22. However, while the negative response of obligate calcifying macroalgae to ocean acidification is well established3,23, there are still large gaps in our knowledge regarding general patterns of how fleshy macroalgae will respond to ocean acidification, making predictions of future ecological dynamics difficult.

Physiological mechanisms underpinning the responses of fleshy macroalgae to ocean acidification are poorly understood5, although they are likely related to methods of DIC uptake18, summarised in Fig. 1. Carbon dioxide concentrating mechanisms (CCMs) allow continued DIC uptake when CO2:O2 ratios are high (e.g. when seawater is highly modified by photosynthesis), or when CO2 concentrations are insufficient to maximise the rate of photosynthesis12. CCMs act mostly via the active uptake of HCO3, and there are multiple types of CCMs with varying energetic efficiencies12,24. The presence of CCMs in macroalgae suggests that ocean acidification may not benefit fleshy macroalgae, as DIC is not always limiting for growth or photosynthesis25. Conversely, macroalgae lacking CCMs (hereafter “non-CCM” macroalgae), or with CCMs with low affinities for DIC, could benefit from ocean acidification due to the alleviation of any CO2 limitation6,18. Additionally, DIC uptake via a CCM is marginally more energetically expensive than CO2 uptake via diffusion when the lowest costs of photorespiration (i.e. oxygenase rather than carboxylase activity with subsequent energetic waste) are included: this energetic cost of using a CCM assumes that leakage of DIC from the cell could occur because DIC concentrations within the cell, compared to seawater, increase24. Therefore, species could gain energetic savings from increasing CO2 concentrations if external seawater CO2 concentrations exceed that produced internally by the CCM, and replacement of the CCM with diffusive CO2 entry to Rubisco could occur24,26,27. This replacement occurs within one or a few generations in photosynthetic organisms, leading to the downregulation of the CCM and long-term energetic savings, which could manifest as benefits such as elevated growth rates, or competitive advantages due to ocean acidification18.

Figure 1: Responses of fleshy macroalgae to ocean acidification in relation to their CO2 concentrating mechanisms (CCMs), showing the different physiological groups defined here, mechanisms defining how increased CO2 could directly affect their physiology, and higher order flow-on effects.
figure 1

Pictured are species which represent those with no CCM (Udotea petiolata), low affinity CCM/lower future CCM activity (Caulerpa prolifera and Dictyota dichotoma) and high affinity CCM with no change in future CCM activity under ocean acidification (Halopteris scoparia and Cystoseira foeniculacea).

Predicting how the multifaceted physiological effects of ocean acidification will combine to influence macroalgal assemblages is difficult. This challenge is amplified by complex ecological and environmental processes that exert pressures on macroalgal species in ways that are difficult to capture in laboratory experiments. The ecological outcomes of combined positive and negative effects of elevated CO2 can be observed at volcanic CO2 seeps28, which provide a field setting in which to test hypotheses regarding how ocean acidification could influence multi-species assemblages. Coralline algae and calcareous green algae (Halimeda and Acetabularia spp.) decline at seeps where CO2 concentrations are similar to and below those predicted to occur in the ocean at the end of the century28. As CO2 concentrations rise closer to the seeps, non-calcareous algae increase in abundance, but not diversity29, and include assemblages comprised of diatoms, cyanobacteria, filamentous macroalgae and fucoids28,30,31. Many species (or groups of species) belonging to the morphological category of filamentous or turf macroalgae benefit from elevated CO232. However, this effect is not ubiquitous29, particularly in tropical ecosystems33, and not all natural ecosystems or seep locations contain ephemeral turf species. In addition, the term turf comprises many unrelated taxa that could have different physiologies34, which will therefore respond in different ways to elevated CO218,33. Therefore, it is presently unknown why some fleshy macroalgal species appear to benefit from elevated CO2 concentrations at these sites and others species do not32,35.

It has been suggested that macroalgae capable of utilising additional CO2 may benefit at the expense of species that are not capable of utilising increasing CO232. While this seems elementary, determining which species will utilise additional CO2 is impossible without an understanding of organismal physiology. Seaweed functional form36 – such as the presence or absence of calcification – can determine the likely-hood of an organism responding negatively to ocean acidification. However, using functional form or morphological groups alone will not determine whether a species is capable of utilising additional CO2 under ocean acidification. Functional form or taxonomic groups have been inadequate in explaining the responses of terrestrial plants to climate change, and it has been suggested that assigning plants to physiological functional groups may serves as a better predictive tool37,38.

Hepburn et al.18 suggested a template for predicting the responses of the future abundance of macroalgae to elevated CO2 based on their inorganic carbon physiology: calcareous species will decline, non-CCM species will increase, and CCM species will either be unaffected or may increase. However, to date these predictions regarding fleshy macroalgae remain untested in the field, and the groups used by Hepburn et al. may even be too broad. Here, we refine this template and combine it with the predictions of Connell et al.32, who state that species capable of utilising additional CO2 will benefit from ocean acidification at the expense of those that cannot. For fleshy macroalgae, we group species as: (1) non-CCM species; (2) CCM species capable of utilising additional CO2; and (3) CCM species not capable of utilising additional CO2. We expand group (2) into species that have either (a) CCMs with a low affinity for DIC (carbon-limited CCM species), or (b) have CCMs that can be downregulated (i.e. lower CCM activity under elevated CO2 concentrations compared to ambient conditions). Group (3) contains CCM species with high affinities for CO2, whose CCM does not downregulate due to additional CO2. For calcareous macroalgae, we split species into two groups based on prior observations at CO2 seep locations and in laboratory manipulation experiments3,23,35. These include group (4) the obligate calcifiers (calcareous reds and greens), and (5) non-obligate calcifiers (calcareous browns, Ochrophyta; Padina spp.) which would be grouped based on the characteristics of their CCM).

Macroalgal tissue δ13C can be used to determine the presence or absence of a CCM, and δ13C can indicate changes in CO2:HCO3 uptake, when corrected for other factors39,40. The δ13C of CO2 and HCO3 in seawater is between ~−10 and 0‰41. Macroalgal uptake of CO2 versus HCO3 results in different tissue δ13C, with sole CO2 uptake via diffusion over the plasmalemma theoretically resulting in values more negative than −30‰, sole HCO3 uptake resulting in values less negative than −10‰. Values between −30‰ and −10‰ can result from either diffusive CO2 influx, with higher δ13C values indicating a large resistance to diffusion with the same CO2 concentration difference, or HCO3 influx with higher values indicating a smaller fraction of the pumped HCO3 leaking out as CO2. Therefore, an organism utilising both CO2 and HCO3 would result in values in δ13C values in between −30 and −10. Here we define species mentioned in this study as having evidence for a CCM if δ13C > −30‰ and as having evidence of a lack of a CCM if δ13C < −30‰. Accumulation of compounds such as lipids could alter tissue δ13C values, as transforming photosynthate into lipids could decrease δ13C40. In past research on phytoplankton, lipid content per cell increased dramatically under elevated CO2, and lipids had much lower δ13C than the bulk organic matter42. The δ13C of source seawater DIC could also influence tissue δ13C, regardless of CO2:HCO3 uptake ratios.

Here, we assess whether a mechanistic understanding of macroalgal DIC physiology can be used to predict changes in macroalgal communities along a seawater pCO2 gradient at Vulcano, Italy. We hypothesized that (1) at Vulcano the δ13C values of all macroalgal species in group 2 (possessing CCMs, with low affinities for CO2, or having a downregulated CCM) would increase (i.e. they would be depleted in δ13C; evidence for increased reliance on CO2) as pH declines; (2) the percentage cover of non-CCM species (group 1) and those CCM species capable of utilising additional CO2 (group 2) would increase at all sites as CO2 increases; (3) the percentage cover of all green and red calcareous species (group 4) and fleshy macroalgae from group (3) would decrease as CO2 increases; and (4) lipid content (a potential influence on δ13C) of all species would not differ significantly across locations with differing CO2 at Vulcano.

Results

The abundance of 17 macroalgal species varied significantly along a gradient of CO2 (PERMANOVA, F = 8.58, P < 0.01; Figs 2 and 3). Udotea petiolata was not present in the surveys at the pH 8.04 and 7.89 sites, while Caulerpa prolifera and Dilophus fasciola were not present at the pH 8.04 location. Dictyota dichotoma was the species whose abundance increased the most between the pH 8.04 and 7.69 locations (17-fold increase). This was followed by Caulerpa racemosa (6 fold increase), Padina pavonica (5-fold increase), Sargassum muticum (3-fold increase), and Dictyopteris polypodioides (2-fold increase). Conversely, crustose coralline algae (CCA) and 8 individual species of macroalgae decreased in abundance between the pH 8.04 and 7.69 locations. CCA had <1% cover at the pH 7.69 and 7.89 locations, while Halopteris scoparia, Cystoseira crinita, C. barbarta and C. brachycarpa had <1% cover at the pH 7.69 location. Four other species decreased in abundance between the pH 8.04 and 7.69 locations. In order from the largest to the smallest decrease in percent cover these were Cystoseira foeniculacea, Codium bursa, Cystoseira compressa, and Acetabularia acetabulum. Most species made minor contributions to the percent cover overall, except Cystoseira foeniculacea and Dictyota dichotoma which showed opposing shifts in abundance between locations (Fig. 2). C. foeniculacea had the highest percent cover at the pH 8.04 (44.78% cover) and pH 7.89 (24.98%) locations, but was the second most abundant species at the pH 7.69 location (8.46%). D. dichotoma had the highest abundance at the pH 7.69 location (40.14% cover), the 2nd highest at pH 7.89 (14.04%) and the 9th highest at pH 8.04 (2.33%).

Figure 2: Percentage cover of macroalgal species from locations with three mean seawater pH values at Vulcano, Italy at three locations differing in mean seawater pH.
figure 2

(a) Species with physiologies predicted to benefit from elevated CO2 concentrations. (b) Species with physiologies predicted to not benefit from elevated CO2 concentrations. Each bar represents mean ±1 s.e. n = 3 transects per site and 10 quadrats per transect.

Figure 3
figure 3

Proportional cover of species predicted to benefit from elevated CO2, those with unknown physiologies, and those predicted not to benefit from elevated CO2 at three different CO2 seep sites: (a) Vulcano, (b) Ischia, and (c) Methana. Proportional cover of the physiological groups: non CCM (CO2 concentrating mechanism), unknown (possible non-CCM), high affinity CCM, low affinity CCM, unknown CCM, calcifying reds, greens and browns at CO2 seep locations with different mean seawater pH. (d) Vulcano. (e) Ischia. (f) Methana. See methods for more details.

Species possessing CCMs dominated percent cover at Vulcano at all locations. However, there were noticeable differences between the percent cover of physiological groups between locations (PERMANOVA: F = 12.29, p < 0.01, Figs 3 and 4). Non-CCM (Udotea petiolata) cover was highest at the pH 7.69 location, calcareous brown (Padina pavonica) highest at pH 7.89, and the cover of calcareous reds (corallines and Peyssoniellia spp.) were lowest at pH 7.89 and 7.69 locations (Fig. 3a). We grouped the species with δ13C values that significantly decreased as pH declines as low affinity CCM species (group 2 – see the introduction), and those with δ13C values that did not vary significantly by pH location as high affinity species (group 3). We then test whether there are differences in the composition of species that theoretically should benefit from elevated CO2 (groups 1 and 2) or not (groups 3 and 4). Results indicate that there are significant differences in their distribution at the three locations at Vulcano (PERMANOVA: F = 17.22, p < 0.01, Fig. 3a). The abundance of species predicted to increase between pH 8.04 and pH 7.69 did so, and those predicted to decline in abundance between pH 8.04 and 7.69 also did (Fig. 3a).

Figure 4
figure 4

PCA plot displaying the association of different species at Vulcano (a), physiological groups at Vulcano (b) and Methana (d), and winners and losers from Vulcano (c) and Methana (e) across the different transects at both sites at locations with different mean seawater pH.

Shifts in species percent cover were related to changes in tissue δ13C, whereby species whose abundance increased by the largest amount as pH declined also tended to be the same species whose tissue δ13C was significantly depleted. We observed significant declines in δ13C at elevated CO2 locations in 5 species (Table 1). Only one species (Udotea petiolata) had δ13C ~−30‰ (Table 1), indicating the possible lack a CCM. Five of six species that were found in greater cover at elevated CO2 locations had δ13C values that were more depleted at elevated CO2 locations (Caulerpa prolifera, Dilophus fasciola, Dictyota dichotoma, Caulerpa racemosa, and Padina pavonica). All five species that declined in abundance at elevated CO2 locations had no significant differences in δ13C values. For D. polypoides, there was insufficient material collected at the pH 8.04 location to analyse δ13C values. Our seawater DIC δ13C values were relatively similar across all three locations, so we use the raw δ13C values of the macroalgae, rather than correcting for source seawater DIC δ13C.

Table 1 Mean ± 1 s.e. of δ13C of macroalgal tissue at the three locations at Vulcano, Italy, with varying mean seawater pH values.

Lipid concentrations did not vary among species examined, and did not vary by mean CO2 at the three locations at Vulcano (SI 1). The presence of a CCM was confirmed for Sargassum muticum, Halopteris scoparia, Padina pavonica and Cystoseira foeniculacea by pH drift experiments that showed pH compensation points above 9 at all locations for all species. Only for Cystoseira foeniculacea was there a significant difference between final pH compensation points between sites (SI 2).

To test the generality of our findings, we examined changes in the abundance of physiological groups of macroalgae at CO2 seeps off Ischia, Italy, and Methana, Greece. We combined published species abundances30,43 and δ13C values (SI 3) to assess changes in the five physiological groups. In Ischia, non-CCM abundance also increased at the elevated CO2 location, and calcareous red species cover declined. At Methana there were significant shifts in the physiological groups (PERMANOVA, F = 1.63, p = 0.04), with calcareous red macroalgae declining as CO2 increased (Fig. 3c). As for Vulcano, we also grouped species abundances based on whether they were predicted to benefit from elevated CO2 or not. At Methana, we see significant differences in the abundance of species predicted to benefit from elevated CO2 or not between the three different pH locations (PERMANOVA, F = 5.06, p = 0.02). The abundance of species not predicted to benefit from elevated CO2 were lowest at the pH 7.67 location, and the abundance of species predicted to benefit were highest at the pH 7.67 location. However, these changes were not as large as at Vulcano (Fig. 3), and there was some seasonality to these effects, with groups predicted to benefit from elevated CO2 being associated more with pH 7.67 locations in spring than in autumn (Fig. 4c). Species from combined groups 1 and 2 tended to increase in abundance at pH 7.80 at Ischia, whereas species from the combined groups 3 and 4 decreased in abundance relative to the pH 8.10 location. However, the data from Ischia could not be analysed statistically, as we did not have the raw data, only the mean of each location.

Discussion

Predicting changes in the structure and dynamics of future macroalgae assemblages due to the direct effects of ocean acidification requires an understanding of the physiological mechanisms of DIC acquisition. The majority of macroalgal species at Mediterranean seep sites had CCMs, and the direction of the change in their abundance at elevated pCO2 at Vulcano was correlated with whether or not their δ13C values decreased: species with decreased δ13C were more likely to increase in abundance and vice versa. This decrease of δ13C with increasing CO2 most likely resulted from an increase in the reliance on CO2 during photosynthesis; we discuss other possibilities below. Species utilising additional CO2 would either receive benefits through the alleviation of DIC limitation, or through the down-regulation of energetically expensive CCMs (Fig. 1). This supports predictions that macroalgal species capable of utilising additional CO2 (i.e. non-CCM species and CCM species with low affinities for DIC) will benefit from ocean acidification, while species that cannot (high affinity CCMs with no down-regulation)18,32, and obligate calcifying species, will decrease in abundance. Although only one non-CCM species was present in our surveys at Vulcano, surveys at other Mediterranean CO2 seep sites support our suggestion that fleshy macroalgae with a predictable suite of physiological traits are able to benefit from elevated CO2 over species that have physiologies poorly suited to higher CO2 concentrations.

The dominance of fleshy macroalgae and absence of calcifying macroalgae at vent sites with high CO2 is well known28,30,43,44, and this study reveals for the first time that these patterns in macroalgal assemblages, which are consistent across several vent sites, are correlated with the DIC physiology of specific species. Macroalgae that thrived at locations with high CO2 at multiple sites were Dictyota spp. (4/4 sites)28,30,43,45, Caulerpa spp. (3/3 sites)28,45, Udotea petiolata (2/3 sites)43, and Sargassum spp. (4/4 sites)28,30,43. This includes two CCM genera containing species whose δ13C values indicate higher reliance on CO2 at Vulcano (Dictyota and Caulerpa), one non-CCM species (U. petiolata), and one CCM species whose δ13C values did not change at Vulcano (Sargassum spp.). Macroalgae that had reduced percent cover at high CO2 locations included coralline algae28,30,43,45 (4/4 sites), calcifying green algae28 (2/2 sites), some Cystoseira spp. (3/4 sites)30,43,45 and Halopteris scoparia (2/2 sites)43. This includes two calcifying groups (coralline algae and calcifying green algae) and two CCM genera with species whose δ13C values did not indicate higher reliance on CO2 at Vulcano. These trends in changing species abundances further support the hypothesis that species responses to ocean acidification will be influenced by their DIC physiology18. The direct effect of changing CO2 concentrations on macroalgal physiology is important, but it is likely that changes in species abundances due to ocean acidification will also be modified by shifting species interactions8,9 and by changes in other environmental factors or their variability46,47,48, such as light18,49.

Natural CO2 seeps are useful in determining the structure of marine assemblages at elevated CO2, but they are not perfect analogues for the effects of ocean acidification. As mean CO2 concentrations increase, variability about the mean also increases near the seeps which may exacerbate any negative effects. Coralline algae, for example, are particularly vulnerable to rapid declines in carbonate saturation/increases in proton concentrations20,21,50. Therefore, the effects of mean increases in CO2 concentrations due to ocean acidification on macroalgal assemblages could differ from the effects of volcanic acidification examined here. Other physicochemical gradients, such as water motion, may affect organisms near volcanic seeps. Although our locations at Vulcano were wave-sheltered, there was a small increase in exposure moving away from the seeps. This gradient of water motion would be expected to act on macroalgal δ13C in the reverse direction to CO2 because thinner diffusion boundary layers result in greater uptake of CO2 at more wave exposed locations39. Therefore, our results are conservative, as the decrease in macroalgal tissue δ13C would be greater if this gradient in wave exposure did not exist.

We consider that the decreased δ13C values of specific macroalgae from elevated CO2 locations is most likely caused by changes in the ratio of CO2:HCO3 uptake. However, there are other mechanisms unrelated to increased reliance on CO2 that would result in decreased tissue δ13C. The most likely scenarios are: (1) Increases in the phosphoenolpyruvate carboxykinase (PEPCK)-based C4 pathways could result in decreased δ13C51,52, and for Udotea petiolata this could explain the <−30‰ values; a phosphoenolpyruvate carboxylase (PEPC)-based C4 pathway would result in less negative δ13C values51,53,54. (2) Lower photosynthetic rates (resulting in lower uptake of 13C enriched HCO3) by macroalgae would also result in decreased δ13C at elevated CO2. As these species tended to increase in abundance at Vulcano, their photosynthetic rates would not be expected to decline. Indeed, past research has found elevated photosystem II relative electron transport rates (used an indicator of photosynthesis in situ in that study) of Padina pavonica at elevated CO2 locations in Vulcano35. (3) Leakage of CO2 from the CCM decreasing as concentrations of CO2 increases externally, with CCM pumping of DIC into the cell remaining equal55. Apart from decreased leakage, none of these scenarios would explain why the species of macroalgae with depleted δ13C also tended to have higher cover at elevated CO2 locations throughout the Mediterranean, as most of these alternate explanations would not provide benefits. Similar decreases in tissue δ13C as CO2 concentrations increase has also occurred for other macroalgal20,56,57, seagrass58 and phytoplankton species59. However, these changes did not always correspond to changes in rates of photosynthesis, DIC acquisition, or growth. For freshwater macroalgae, decreased δ13C and a higher abundance of non-CCM species were also found in locations with elevated CO260, further supporting our conclusions. While further research is still required to examine in situ photosynthetic rates of macroalgae at seep sites and to determine the relationship between δ13C and CCM activity, we conclude based on the evidence available that increased uptake of CO2 during DIC acquisition as mean CO2 concentrations increase is the most likely and logical cause of decreased tissue δ13C in our study.

The effects of ocean acidification on macroalgal assemblages will likely depend on the natural assemblage composition in a particular region. For example in the Mediterranean and in coral reef ecosystems, the abundance of non-CCM species is low61,62 compared to temperate subtidal ecosystems18,63, meaning that greater changes in species compositions due to ocean acidification could occur in ecosystems possessing more non-CCM species, compared to those trends observed here. CCM species are relatively abundant in most shallow-water marine ecosystems18,63, particularly in warmer climates (such as in this study and on the Great Barrier Reef ref. 61), and the majority of macroalgal species in rocky and coral reefs possess CCMs18,61,62,63. Therefore, the direct effects of ocean acidification on macroalgae could manifest via increases in the abundance of low-affinity CCM species in most ecosystems dominated by macroalgae, particularly warmer locations.

Here we demonstrate a relationship between the DIC physiology of macroalgae and their changes in abundance at natural CO2 seeps in the Mediterranean. Predicting changes in macroalgal communities due to ocean acidification is complex. This is in part because we do not understand the relative roles of direct and indirect effects of ocean acidification in shaping macroalgal assemblages. In ecosystems where strong top-down control exists, responses of benthic communities could be more related to changes in higher trophic levels8,9, but when top-down control is relatively weak, responses could be dictated primarily by macroalgal DIC physiology. Additionally, the direct effects of ocean acidification could be modified by the presence of competitors or facilitators17,64,65. The challenge now is to determine whether the physiological groups defined here respond similarly elsewhere. It is clear is that most calcareous macroalgae will be negatively impacted by elevated mean CO2, but that species which are not obligate calcifiers (such as Padina pavonica) are tolerant3,14,20,35,48,66. Fleshy macroalgae respond positively if they are reliant only on CO2, or if their reliance increases as CO2 concentrations increase. The capacity to utilise additional seawater CO2 means that non-CCM species, or species with CCMs with lower affinities for DIC, will likely benefit from ocean acidification.

Currently, there is a limited understanding of how macroalgal physiology could be used to predict general shifts in abundance or fitness under ocean acidification. This is due to the complex nature of macroalgal inorganic carbon physiology and calcification, and in part, the difficulties in thoroughly assessing the DIC physiology of numerous taxa from different geographic regions. For example, the traditional approach of examining changing photosynthetic rates over constant additions of DIC is problematic, because there is the possibility that future seawater could favour the retention of a CCM in macroalgae that are not DIC-saturated today67. However, this would only occur until the external seawater CO2 concentrations exceeded that produced internally by the CCM, so that replacement of the CCM with diffusive CO2 entry to Rubisco could occur and it is unknown whether the additional CO2 or HCO3 relieves this DIC limitation. Greater progress has been made determining mechanistically why the calcification of some macroalgal species may be impacted by declining pH and increasing CO2 concentrations15,68,69,70,71, although more work is required72. Our study provides an essential stepping stone to improving knowledge regarding how fleshy macroalgae might respond in a future elevated CO2 ocean, as the first general test of how macroalgal DIC physiology is linked to responses to natural acidification in situ. Our framework explains why elevated CO2 does not ubiquitously alter habitat types at CO2 seep sites between canopy-forming and turf species29: species compositions will change depending on resident species’ physiology, which is only sometimes related to their morphology or functional form group. From knowledge of the DIC physiology of macroalgae within a community, an improved framework can be created by which we can project the responses of coastal marine assemblages to ocean acidification.

Materials and Methods

In late May to early June 2014 we surveyed pHT (pH on the total scale, hereafter “pH”), total alkalinity (hereafter “AT”), DIC, and δ13C of DIC at three locations with a mean pH of 8.04, 7.89 and 7.69 at 2 m depth (“pH 7.69” 38°25.176′N, 14°57.658′E; “pH 7.89” 38°25.193′N, 14°57.763′E; “pH8.04” 38°25.248′N, 14°57.853′E; see SI 6) where pH, AT and some associated biogeochemical parameters have been monitored for six years35,73,74. For long term pH data and other environmental variables (e.g. H2S, total alkalinity and heavy metals) at this site see Boatta et al.74. Surveys of macroalgal abundance were conducted on the 26 May 2014, and involved three 25 m transects at each of the 3 locations. Each transect involved ten 50 by 50 cm photoquadrats, from which the percent cover of individual species were determined. Cystoseira spp. were lumped into morphological categories and later identified to species were possible. Macroalgae were grouped based on their use of DIC into non-CCM, CCM, and calcareous species based on δ13C values (after ref. 18). Calcareous species were further split into reds, greens and browns because these taxonomic groups have different calcification mechanisms75, which may influence their response to ocean acidification35,66.

Samples of each species were collected from the three locations for δ13C analysis. Due to the scarcity of some species at specific locations, not all species used here were collected at all three locations. A minimum of three replicates were collected of each species at each location where possible, with each replicate being a different individual and a total of 19 species being analysed. All samples were air dried at 60 °C for 24 hours. Then they were ground into a fine powder using a porcelain mortar and pestle and weighed in to pressed tin capsules (0.2 mg; Sercon, UK). Isotope values were determined using a Fisons NA1500 elemental analyzer coupled to a Thermo Scientific Delta V Plus via a Conflo IV. Combustion and reduction were achieved at 1020 °C and 650 °C respectively. Percent C and N composition was calculated by comparison of mass spectrometer peak areas to those of standards with known concentrations. Isotopic values for carbon are reported as δ-values (‰) relative to Vienna Pee Dee Belemnite (VPDB) and were corrected via a 4-point calibration using certified standards. Reproducibility was also monitored by the use of long term internal reference materials. Both precision and accuracy were ±0.1‰ (1 SD). All samples were run in duplicate and averaged when values differed.

The four most abundant and easily identifiable species – all of which were present at all sites – were collected on May 27th from each station for simultaneous pH drift assays. pH drift assays determine whether a species is capable of utilizing HCO3 by assessing its capacity to elevate pH above 9 – the point at which CO2 concentrations are functionally zero. For more details see refs 49,76,77. Specimens were placed into 50 ml transparent plastic containers that were maintained in a water bath for 24 hours. Water in the water bath was manually replaced every 30 minutes during daylight hours to maintain temperatures within 1.5 °C of the ambient seawater temperature (24 °C). Seawater pH was measured using a combined pH and RDO meter (Orion Star A216 pH/RDO/DO) and pH electrode (Orion 8107BNUMD – Ross Ultra pH/AIC triode), calibrated with pH 7.0 and 9.0 NBS buffers, cross referenced with Tris and Amp seawater buffers to convert to the total scale78. Oxygen was measured using an Orion RDO probe (ORI087100MDW) attached to the same meter, and was calibrated using 0% and 100% air saturation standards made by bubbling seawater with N2 or air respectively for 10 minutes.

Seawater pH was measured at all three locations during the study duration on three days, and involved triplicate discrete measurements using the electrode noted above. Samples for AT and DIC were taken on June 1st from all three locations. Seawater was stored in 250 ml glass containers and poisoned immediately with HgCl2. AT was measured using an open-cell titration using a Metrohm 809 Titrando and Metrohm 800 Dosino. AT measurements were made at 25 °C using a Circu temperature bath, while DIC was measured using a LiCOR DIC analyser. Both were measured against a CRM supplied by Andrew Dickson (see ref. 78). Measured accuracy and precision of the CRM was within ±5 μmol kg−1 for both DIC and AT, within the standard error of the differences between measurements from any one location. SI 6 lists the mean seawater carbonate chemistry at the three locations. δ13C of seawater DIC was determined using a Gasbench II coupled to a Delta V Plus via a Conflo IV (Thermo Scientific; after ref. 79). Samples (0.5 mL) were injected into helium flushed 12 mL septum capped exetainers (Labco, High Wycombe, UK) containing phosphoric acid (85%). Samples were equilibrated overnight after which time liberated CO2 is passed to the IRMS in a stream of helium passing through a nafion water separator and a Poroplot Q GC column (25 m × 0.32 mm ID, 45 °C, 2.5 mL/min). Isotopic values for carbon are reported as δ-values (‰) relative to VPDB and were corrected via a 2-point calibration using certified standards. Reproducibility of replicates was 0.05‰. Four replicates samples were collected from each site, but half of these were broken by Australian Customs for the pH 7.89 and 7.69 sites. All replicate samples were then measured in duplicate.

Lipid content of macroalgae was measured by extracting samples quantitatively overnight using a single-phase dichloromethane/methanol/water mixture (after ref. 80). Following phase separation, lipids were recovered in the lower dichloromethane layer and concentrated by rotary evaporation to provide the total lipid extract. This was transferred to pre-weighed glass vials using solvent which was subsequently removed under a stream of nitrogen allowing lipid weight to be determined.

Most CO2 seeps have elevated H2S or toxic metals so detailed geochemical monitoring is needed to rule out these potentially confounding factors (e.g. at Ischia) or minimize their likely effects (e.g. at Vulcano)74. The effects of elevated CO2 at individual seep sites could be altered by other local factors such as wave exposure, nutrient concentrations or turbidity. Additionally, nature is inherently variable, and observations of changing patterns in species’ abundances at one location may only represent snapshots of variability in species’ abundances in time in space81. For these reasons we compared the patterns we observed along the Vulcano gradient with published studies on macroalgal abundances at CO2 seep sites off Ischia (Italy) and Methana (Greece). To calculate the percent cover of the five different physiological groups of macroalgae, percent cover data from Porzio et al.43 (Ischia) and Baggini et al.30 (Methana) were used and macroalgae were grouped based on their use of DIC. For fleshy species, δ13C from this study or from the literature were used to determine CCM presence or absence. SI 5 lists the details of this analysis beyond those presented in Fig. 2.

ANOVA was used to examine differences in δ13C and final O2 concentrations and pH in pH drift experiments between locations within species. All data were checked for normality and heterogeneity first, and passed all checks. When statistical differences were found in ANOVAs, Tukey’s post-hoc tests further determined differences between particular locations along the CO2 gradient. These analyses were conducted in R v. 3.20. Differences in percent cover of species (Vulcano only), the physiological groups, and the species we predict to benefit or not from elevated CO2 between the locations with differing CO2 concentrations where analysed using separate PERMANOVAs in PAST for each site (Vulcano and Methana, i.e. our study site and that published previously by Baggini et al.30). Methana was the only additional site analysed using PERMANOVA and PCA because it was the only site where published raw data was available that contained species abundances, and not morphologically grouped species.

Additional Information

How to cite this article: Cornwall, C. E. et al. Inorganic carbon physiology underpins macroalgal responses to elevated CO2. Sci. Rep. 7, 46297; doi: 10.1038/srep46297 (2017).

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.