Biomineralization control related to population density under ocean acidification

Abstract

Anthropogenic CO₂ is a major driver of current environmental change in most ecosystems¹, and the related ocean acidification (OA) is threatening marine biota². With increasing pCO₂, calcification rates of several species decrease³, although cases of up-regulation are observed⁴. Here, we show that biological control over mineralization relates to species abundance along a natural pH gradient. As pCO₂ increased, the mineralogy of a scleractinian coral (Balanophyllia europaea) and a mollusc (Vermetus triqueter) did not change. In contrast, two calcifying algae (Padina pavonica and Acetabularia acetabulum) reduced and changed mineralization with increasing pCO₂, from aragonite to the less soluble calcium sulphates and whewellite, respectively. As pCO₂ increased, the coral and mollusc abundance was severely reduced, with both species disappearing at pH < 7.8. Conversely, the two calcifying and a non-calcifying algae (Lobophora variegata) showed less severe or no reductions with increasing pCO₂, and were all found at the lowest pH site. The mineralization response to decreasing pH suggests a link with the degree of control over the biomineralization process by the organism, as only species with lower control managed to thrive in the lowest pH.

Several studies on the influence of pH on crystallography and texture of calcified regions are ex situ, short-term experiments on isolated organisms⁵, providing important information, but unrepresentative of natural ecosystems and failing to assess long-term effects⁶. There is a great need of long-term analyses on OA effects on marine ecosystems acclimated to high pCO₂, as found around CO₂ vents. Vents are not perfect predictors of future oceans, owing to pH variability, proximity of unaffected populations, and co-varying environmental parameters⁷. However, vents acidify seawater on sufficiently large temporal and spatial scales to integrate ecosystem processes⁶, acting as “natural laboratories”. In Papua New Guinea vents, reductions in coral diversity, recruitment, abundance, and shifts in competitive interactions are found⁸. In Mediterranean vents, decreased diversity, biomass, trophic complexity, and abundance in many calcifying and non-calcifying organisms, and increases in macroalgae and seagrasses are observed^7,9,10.

We assessed, along a natural pH gradient, the effect of pCO₂ on the mineralization and abundances of the aragonitic scleractinian B. europaea, the aragonitic tube-forming gastropod V. triqueter, the brown alga P. pavonica, which deposits aragonite on the thalli surface, the green alga A. acetabulum, whose cups’ outer surfaces are calcified with aragonite and a small amount of whewellite (calcium oxalate), and the non-mineralized brown alga L. variegata. The mineralization is biologically controlled in V. triqueter (i.e., mineral is deposited in confined nucleation sites under complete biological control with minimal environmental effects), biologically induced in P. pavonica and A. acetabulum (i.e., it is strongly affected by the environment with minimal biological control), wheras B. europaea may represent an intermediate and still controversial situation¹¹. We aimed to assess, for the first time, changes in the mineralization and abundance of species along a pCO₂ gradient in relation to their control over biomineralization.

Seawater chemistry

Mean pH, CO₂, saturation of calcite (Ω_calc), and of aragonite (Ω_arag) differed among Sites (Kruskal-Wallis test/ANOVA, p < 0.001). The median pH values were 8.1 (Site 1), 7.9 (Site 2), 7.8 (Site 3) and 7.7 (Site 4), with increasing variability towards Site 4 (Fig. 1; Fig. S1; Table S1).

Fig. 1.

Range in pH_TS and mean percentage cover of Balanophyllia europaea, Vermetus triqueter, Padina pavonica, Acetabularia acetabulum and Lobophora variegata along the pCO₂ gradient. pH measures were 103-110 per Site. Horizontal bars indicate the median pH. Error bars are 95% confidence intervals.

Mineralogy

Aragonite was the only mineral phase in B. europaea skeletons (Fig. S2). Organic matrix (OM) content was homogeneous among Sites (Kruskal-Wallis test, p > 0.05; Fig. S3; Table 1). Skeletal texture displayed fibers evolving from a center of calcification (Fig. S4), but the morphology of these centers, and fiber thickness (600 ± 200 nm) were not related to pCO₂ (Fig. S5). Hardness was homogeneous among sample regions and study Sites (5100 ± 600 MPa). The elastic Young’s modulus decreased (i.e., the skeleton became less stiff) along the aboral to oral direction and was lower at Sites 2 and 3 than at Site 1 (Table S2). Crystal quality and fiber thickness usually increases when crystallization occurs under lower supersaturation¹². While a reduction of precipitation rate with pCO₂ could increase crystal quality, other mechanisms could involve OM molecules. Corals appear to maintain a high pH at the nucleation sites within the calicoblastic layer, possibly expending significant amount of energy¹³. This is supported by the lack of increase in aragonite fiber thickness with decreasing seawater Ω_arag, which would be expected if Ω_arag of the calcification site decreased¹². The reduction of skeletal stiffness with declining pH is likely associated with an increase in porosity, as confirmed by preliminary Nuclear Magnetic Resonance analyses. B. europaea skeletons have already been found to increase their porosity in stress conditions, such as increasing temperature¹⁴.

Table 1.

Summary of the main effects of OA on the mineral distribution and content in the calcified organisms with increasing proximity to the CO₂ seep (declining pH). Uncertainties are one standard deviation.

			Balanophyllia europaea		Vermetus triqueter		Padina pavonica		Acetabularia acetabulum
	pH	Ω arag	min. phase	% min. phase#	min. phase	% min. phase#	min. phases	% min. phases	min. phases*	% min. phases
Site 1	8.1	3.6	A	97.4 ± 0.4	A	98.0 ± 0.5	A CSh$	73.9 ± 8.3	A Wh	42.9 ± 3.9
Site 2	7.9	2.4	A	97.80 ± 0.04	A	98.3 ± 0.2	A CSh	71.5 ± 10.6	A Wh	24.2 ± 6.9
Site 3	7.8	2.1	A	97.8 ± 0.2	A	98.1 ± 0.1	A CSh	59.6 ± 4.9	A Wh	22.7 ± 7.6
Site 4	7.7	1.5	-	-	-	-	A$ CSh	61.6 ± 4.5	Wh	25.4 ± 8.0

A, Wh and CSh indicate aragonite, whewellite and monohydrate calcium sulphate salts. The % of mineral phase(s) is reported as mass ratio.

^*Amorphous globular particles were also detected.

^#The content of organic matrix is the complement to 100 in B. europaea and V. triqueter where no significant amounts of water were detected in the skeletons and shell-tubes. P. pavonica and A. acetabulum contained a significant amount of bounded water and their percentage of organic matrix significantly increased with pCO₂. When two phases were present, the most abundant one is indicated in bold.

^$This phase was detected in traces.

The apical regions of V. triqueter shell-tubes of similar diameter (~5 mm) were analysed. Aragonite was the only mineral phase (Fig. S2). Shell-tubes from Site 1 had four layers with different crystal texture and variable relative thickness among samples (Fig. 2). The first two layers were prismatic, with the second one made of regular prisms with a square section (500 nm side). The third was cross-lamellar with regular fibers 500 nm thick. The fourth external layer was spherulitic, with several bores and channels (Fig. 2). In one of six samples from Site 2, and in four of six from Site 3, the two internal layers were not present. Different hardnesses and elastic Young’s moduli were associated with different textures, without significant variations among Sites (Table S2). OM content was homogeneous among Sites (Kruskal-Wallis test, p > 0.05; Table 1; Fig. S3). The unaffected mineralogy of V. triqueter shell-tubes is possibly due to the exquisite control of biological macromolecules on mollusc mineralization, which occurs in confined sites resulting in complex crystalline textures (Fig. 2). During shell-tube growth, the external layers are the first to be deposited, followed by the internal ones¹¹. A pH reduction can reduce the growth rates (calcification) of mollusc shells¹⁵, which likely explains the absence of the internal layers in shell-tubes from the most acidic Sites.

Fig. 2.

SEM images of a longitudinal section of Vermetus triqueter shell-tube. a) Apical region of shell-tube. High magnification images were taken along the dotted line from inside (P1) to outside (S) the tube. b) Shell-tubes in Site 1 consisted of four layers (P1, P2, Cl, S): c) first prismatic layer (P1); d) second prismatic layer (P2), with square cross-sections; e) cross-lamellar layer (Cl); f) spherulitic layer (S) containing bores and channels. The first two layers were not present in one sample from Site 2 and in most of the samples from Site 3.

Padina pavonica thalli tips from Site 1 were mainly aragonitic with traces of hydrate calcium sulphate salts (CSh). Aragonite concentration decreased with increasing pCO₂, and almost only CSh was observed at Site 4 (Fig. S2). The presence of CSh was investigated in detail by further tests and considerations (see Supplementary Information). The overall mineral content differed among Sites (Kruskal-Wallis test, p < 0.05; Table 1, Fig. S3), and declined with increasing pCO₂ (Spearman’s rho of the correlation with pH = 0.592, p < 0.01). In all samples, many aragonite needle-like crystals were aggregated or merged forming bundles. CSh crystals appeared as prisms or tablets (Fig. 3, Fig. S6). The decalcification left the thalli free of mineral deposits (Fig. S6). In P. pavonica, CSh increased with increasing pCO₂. Some stabilization of CSh salts by algal molecules^16,17, probably polysaccharides, is expected, as in seawater CSh converts to gypsum.

Fig. 3.

SEM images of Padina pavonica thalli and Acetabularia acetabulum cups. In ‘a’ and ‘b’, left column indicates low-magnification images and right column indicates high-magnification images. P. pavonica (a) became less mineralized with increasing pCO₂. The aggregation and shape of aragonite crystals was lightly affected by pCO₂. In A. acetabulum (b), needle-like crystals of aragonite (An) were above and among the globular aggregates (Ga). The amount of biomineralized material was reduced with increasing pCO₂, wheras the center of the cup became more populated by spherical aggregates. Aragonite needles are shown in the inset for Site 2 and 3, and whewellite is shown in the inset for Site 4.

Acetabularia acetabulum was mainly aragonitic, with minor presence of whewellite (calcium oxalate; Fig. S2). All samples contained about 10% (w/w) of water. The content of mineral phases, estimated after pyrolysis and release of water, differed among Sites (Kruskal-Wallis test, p < 0.01; Table 1 and Fig. S3) and declined with increasing pCO₂ (Spearman’s rho of the correlation with pH = 0.555, p < 0.01). The aragonite/whewellite ratio decreased with increasing pCO₂. Globular granules of amorphous material rich in Ca and S were observed on the surface of needle-like aragonite aggregates at Sites 1-4 (Fig. 3 and Fig. S6). Whewellite was the only phase detected at Site 4 (Fig. S2; Fig. 3), where it appeared as prisms. After decalcification, the cup surface was free of mineral deposits (Fig. S6). In line with a mainly chemical mineralization control¹⁷, increasing pCO₂ did not affect deposition of whewellite in A. acetabulum. The persistence of amorphous globular particles rich of calcium and sulphur was unexpectedly observed when aragonite disappeared. This phase, soluble in acetic acid solution, may be the result of interactions between Ca ions and the sulphonated groups of polysaccharides¹⁷, overexpressed by the algae as a response to acidification.

Within each sample, P. pavonica and A. acetabulum showed a marked reduction of mineralized areas with increasing pCO₂ (Fig. 4). Their aragonite content decreased with decreasing Ω_arag. At a Ω_arag of 1.5, aragonite was not observed, probably because this saturation level is too low to sustain its nucleation process. The observed changes in morphology of aragonite crystals (Fig. 3) are associated with seawater chemistry¹⁸.

Fig. 4.

In situ pictures of Padina pavonica and Acetabularia acetabulum at Sites 1 and 4, showing the reduction of calcified material (white areas) with increasing pCO₂.

Abundance

The percentage of cover for all species, except A. acetabulum, differed among Sites (Kruskal-Wallis test, p < 0.001), and decreased with increasing pCO₂ (Table S3). B. europaea and V. triqueter were not found at Site 4 (pH = 7.7, Ω_arag = 1.5; Fig. 1). Densities of A. acetabulum were homogeneous among Sites (Kruskal-Wallis test, p > 0.05). This suggests that P. pavonica and A. acetabulum are not obligate calcifiers, persisting in high pCO₂ where also the non-calcifying species L. variegata survives (Fig. 1).

A changing benthic community was associated with the pCO₂ gradient (Supplementary Video). The coverage of a coral, a mollusc, and two macroalgae, one of which is a calcifier, declined with increasing pCO₂ (Fig. 1). This is in agreement with previous investigations at CO₂ vents, documenting dramatic reductions in calcifying macroalgal abundance^7,8.Calcifying organisms seem the most sensitive to elevated pCO₂¹⁹, even if their response is not consistent^8,15,20. A. acetabulum, B. europaea and V. triqueter were previously found only outside vent areas, at a pH_TS of 8.14⁷. Instead, in our survey, B. europaea and V. triqueter survived up to pH_TS 7.8 and algal species up to pH_TS 7.7 (Fig. 1). The response of macroalgae to pCO₂ is expected to vary among species. As an example, increased cover with decreasing pH was observed for Padina spp at CO₂ seeps in Vulcano and Papua New Guinea²¹, and for L. variegata at CO₂ seeps in Ischia¹⁰, which contrasts with our data. The responses of non-calcifying macroalgae to elevated pCO₂ is variable²², depending on energy availability²³. The nature of CO₂-induced shifts in macroalgal community structure is likely to vary depending on other environmental factors, such as nutrient availability, temperature and solar radiation²⁴. For instance, L. variegata were sampled in Ischia at 1 m depth and in our study at 8-12 m. Growth of the brown alga Dictyota ciliolata decreases with decreasing light intensity²⁵. In the brown alga Fucus vesiculosus, photosynthesis and growth decrease with increasing depth²⁶ from 1 to 6 m. At the high light of the shallow Ischia site, L. variegata may compensate for the negative effect of lowered pH through enhanced growth.

Relationships with the control over the biomineralization

We related, for the first time, the biological control over biomineralization with changes in the abundance of organisms along natural pCO₂ gradients. The content of biomineralized products decreased with increasing pCO₂ only in the two calcifying algal species (with a weak control over their biomineralization), but the two species with stronger biomineralization control (B. europaea and V. triqueter) tolerated less severe pCO₂ increases (pH 7.8) than the algae (pH ≤7.7). Moreover, the more tolerant algae continued to grow despite their biomineralization products being profoundly altered by pCO₂. Even if biomineralization in algae is only induced¹⁷, we can’t exclude that the switch from aragonite to other biominerals may represent a phenomenon of phenotypic plasticity²⁷, which is increasingly being found to strongly contribute to persistence in the face of climate change²⁸. The control over biomineralization may not be the only cause of the observed differences, because the coral and the mollusc have a biology completely dependent on calcification, whereas the algae do not. Moreover, the algae may benefit from pCO₂ increase in terms of photosynthesis, while the coral may be less dependent on the photosynthetic process and the mollusc does not photosynthesize. This study adds new evidence to field studies on OA effects^4,9,29,30, all indicating major ecological shifts as CO₂ rises. It documented that (i) the mineralization response to OA seems connected with the organisms’ control of biomineralization, (ii) increasing pCO₂ profoundly affects the abundance of many benthic organisms, and (iii) only the species with weaker control were observed at the lowest pH.

Methods

Study site

Fieldwork was conducted at Panarea, Italy (Supplementary Video; Fig. S1), where hydrothermally stable CO₂ emissions acidify seawater, generating a pH gradient (see Supplementary Information for details).

Carbonate chemistry

Four sampling Sites were selected (Figs. 1 and S1): a control site (Site 1), two intermediate pCO₂ (Site 2 and Site 3), and high pCO₂ (Site 4). pH (NBS scale), temperature, and salinity were measured at each Site during several surveys between July 2010 and May 2013 with a multi-parametric probe (600R, YSI Incorporated, USA) powered from a small boat and operated by SCUBA divers. Bottom water samples for determination of total alkalinity (TA) were collected and analysed using standard methods (see Supplementary Information for details). Additional temperature data were recorded every three hours by sensors (Thermochron iButton, DS1921G, Maxim Integrated Products, USA) attached in each Site from July 2010 to May 2013. Measured pH was converted to the total scale using CO2SYS software. Median pH (back-transformed hydrogen ion concentrations) were calculated for each Site. The pH, TA, salinity and temperature were used to calculate other carbonate system parameters using the software CO2SYS (Supplementary Information).

Benthic survey

Photographs of benthos (5 to 10 per Site, 50 × 50 cm for the animals, 21.0 × 29.7 cm for the algae) were used to measure the percent cover of B. europaea, V. triqueter, P. pavonica, A. acetabulum and L. variegata at each Site. See Supplementary Information for details.

Vent gas

Gas was sampled during five surveys (June 2011-May 2013) and analysed at the Laboratory of Fluid and Rock Geochemistry of the University of Florence using standard methods (Supplementary Information). Water samples were collected and tested for dissolved H₂S (Supplementary Information).

Statistical analyses

Analysis of variance (ANOVA) and the post-hoc Fisher LSD test were used to test for differences among Sites using arcsine and log-transformation for percent cover and environmental data, respectively, when necessary. Else, the non-parametric Kruskal-Wallis and Spearman’s rank correlation coefficients were used. All analyses were performed using SPSS v.20.

Biomineralization

Samples were randomly collected by SCUBA divers at all Sites and were prepared for analyses with standard methods (Supplementary Information). Microscopic observations, mechanical, and spectroscopic measurements required the preparation of cross sections of the samples (Supplementary Information). X-ray powder diffraction (XRD) and Fourier transform infrared spectroscopy (FTIR) patterns on small amounts of powdered samples were collected using standard methods (Supplementary Information). Attenuated Total Reflection FTIR (FTIR ATR) spectra of sample cross sections were acquired with standard methods (Supplementary Information). The organic matter content in the sample was determined by thermo-gravimetric analysis (TGA; Supplementary Information). Microstructures were observed using optical and scanning electronic microscopes (SEM; Supplementary Information). The mechanical properties of shell-tubes and skeletons were measured with standard nanoindentation techniques (Supplementary Information).