全球变化知识资源中心

Ocean acidification results in co-varying inorganic carbon system variables. Of these, an explicit focus on pH and organismal acid–base regulation has failed to distinguish the mechanism of failure in highly sensitive bivalve larvae. With unique chemical manipulations of seawater we show definitively that larval shell development and growth are dependent on seawater saturation state, and not on carbon dioxide partial pressure or pH. Although other physiological processes are affected by pH, mineral saturation state thresholds will be crossed decades to centuries ahead of pH thresholds owing to nonlinear changes in the carbonate system variables as carbon dioxide is added. Our findings were repeatable for two species of bivalve larvae could resolve discrepancies in experimental results, are consistent with a previous model of ocean acidification impacts due to rapid calcification in bivalve larvae, and suggest a fundamental ocean acidification bottleneck at early life-history for some marine keystone species.

Ocean acidification (OA) is described as an imbalance between the acidic influence of rapidly accelerating anthropogenic CO₂ emissions and the slow buffering response due to weathering of continental rock and carbonate marine sediment, causing increased acidity of marine waters^{1, 2}. The release of CO₂ from fossil fuel emissions and cement production, and decreasing uptake efficiency of CO₂ by land and sea has resulted in the fastest increase in p_CO2 (partial pressure of carbon dioxide) in the past 800,000 years³. Conversely the natural mechanisms that buffer acidic perturbations from increasing p_CO2 occur over timescales of hundreds of thousands to millions of years^{1, 2}. Modern anthropogenic changes in the open ocean have tightly coupled aqueous p_CO2, pH and mineral solubility responses, but it was not always thus. Previous instances of elevated p_CO2 in the geologic record, such as the Cretaceous, seem to coincide with significantly elevated alkalinity⁴, and were fairly benign with respect to OA, with elevated p_CO2 not indicative of low pH or mineral corrosivity. Throughout the geologic record and in many coastal habitats the marine carbonate system decouples, resulting in changes in pH, p_CO2 and saturation state that do not follow the co-variance assumed for modern open-ocean average surface waters⁵.

Effects of ocean acidification on a suite of marine organisms have been the subject of significant recent work. Although many experimental results have shown equivocal impacts when taken in composite, the process of calcification has mostly exhibited negative sensitivity to OA (ref. 6). Physiological processes that may experience OA sensitivity occur across all taxa in nearly all natural waters; however, persistent calcified structures can elevate species that precipitate calcium carbonate to keystone status in marine waters. Bivalves, which provide a number of critical ecosystem services, have been noted as particularly sensitive to OA (refs 7, 8, 9, 10). Some experiments have even found OA impacts at present-day, compared with pre-industrial, p_CO2 levels¹¹. Marine bivalves seem to be sensitive to OA owing to the limited degree to which they regulate the ionic balance and pH of their haemolymph (blood)^{12, 13, 14, 15}, and acute sensitivities at specific, short-lived, life-history stages that may result in carryover effects later in life^{16, 17, 18, 19, 20}. Bivalve larvae are particularly sensitive to OA during the hours- to days-long bottleneck when initial shell (called prodissoconch I or PDI) is formed during embryogenesis¹⁷. Before PDI shell formation, larvae lack robust feeding and swimming appendages and must rely almost exclusively on maternal energy from eggs; and during calcification of PDI the calcification surfaces are in greater contact with ambient seawater than during following shell stages¹⁷. Failure of larvae to complete shell formation before exhausting maternal energy reserves leads to eventual mortality, as seen in well-documented oyster hatchery failures¹⁸. So far, the prevailing physiological mechanism identified for OA effects on organisms has been in their ability to regulate internal acid–base status; however, short-term exposure impacts and carryover effects documented in bivalve larvae^{18, 19, 20, 21} and greater exposure of PDI calcification to ambient seawater¹⁷ points to another mechanism for the early larval sensitivity not captured by regulation of internal acid–base chemistry²².

In most natural waters the dissolved inorganic carbon (DIC) system controls both pH and the thermodynamic mineral solubility (saturation state), but in different ways. pH is determined by the ratio of dissolved concentrations of CO₂ to carbonate ions, whereas saturation state is predominantly controlled by absolute carbonate ion concentration. The potential that organisms will respond differently to pH (ratio) or mineral saturation state (abundance), highlights how the decoupling of carbonate system variables in coastal zones⁵ or geologic time^{1, 2} provides a formidable challenge in interpreting and predicting organismal responses to OA. The seemingly simple experimental perturbation of CO₂ bubbling results in the equilibrium redistribution of the acid–base species with pH, saturation state, p_CO2 and dissolved inorganic carbon (DIC) all changing simultaneously. The co-variance of carbonate parameters leaves interpretation of experimental responses unclear if organismal sensitivity to each parameter is physiologically distinct, particularly if the importance of each process varies across ontology (for example, respiration, shell formation, feeding rate). The underlying mechanisms of organismal sensitivity to OA may therefore not be constrainable without special experimental techniques.

We conducted series of experiments in which we applied a unique chemical manipulation approach to decouple the carbonate system parameter-covariance and evaluated larval growth and development of two bivalve species: the Pacific oyster, Crassostrea gigas, and the Mediterranean mussel, Mytilus galloprovincialis. Through simultaneous manipulation of DIC and alkalinity, we generated a 4 × 4 factorial design with aragonite saturation state (Ω_ar) and p_CO2. Our experimental design separated Ω_ar and p_CO2 effects on larval responses; and responses to pH were evaluated by examining responses to pH within a p_CO2 and Ω_ar treatment level. Using this approach, we assessed which carbonate system parameter is most important to early larval shell development and growth: pH, p_CO2, or Ω_ar.

Chemistry manipulations.

We simultaneously altered abundance and ratio of DIC and alkalinity to provide three orthogonal experimental axes in pH, p_CO2 and saturation state of the calcium carbonate mineral aragonite (Ω_ar) (Fig. 1 and Supplementary Table 1). The sensitivity of these parameters to DIC:alkalinity means that there is some variability within treatment suites, but that variability was far less than the differences among treatments. We were able to replicate treatment conditions via DIC and alkalinity, as evidenced by the concordance between expected versus measured values (Supplementary Fig. 1). At the termination of the 48 h incubation period we found p_CO2 generally increased by approximately 10–30% relative to initial conditions. The greatest p_CO2 increases were in treatments with the poorest larval development, probably due to elevated microbial respiration associated with larval mortality in these treatments.

a, Values for the two experiments on Mytilus galloprovincialis. b, Values for the two experiments on Crassostrea gigas. Circle and square symbols represent chemistry for the first and second experiments, respectively.

Our results initially seems contradictory to the physiological basis for understanding ocean acidification impacts on organisms; particularly the overarching role of seawater pH, acid–base regulation, and extracellular acidosis in marine organisms^{12, 13, 22, 23, 24}. Specifically, we found that seawater pH seems to have little to no measurable effect on early larval shell development and growth, except for the case where pH and Ω_ar are both very low (Figs 2 and 3). At these low levels, seawater pH probably becomes very important (particularly to bivalves, which show limited ability to regulate extracellular pH), as eukaryote intracellular pH typically ranges from 7.0 to 7.4 (ref. 25) and additional energy is needed to maintain physiochemical gradients crucial for passive and active cross-membrane ion transport²⁶ if extracellular pH approaches these values. Some species seem to be able to mitigate acidosis via bicarbonate accumulation; however, ability to do so is variable across taxa, and bicarbonate accumulation often requires several days to months^{14, 15, 22}. During the transient (days) early larval stage it is unlikely that bivalve larvae have the time or physiological capacity to compensate for acidosis²², with their limited energy budget and the embryological development taking place during this time period. Therefore, although seawater pH effects on organismal acidosis may also be at work during this early larval stage, we have experimentally shown that any pH effect is overwhelmed by the impact of saturation state during initial shell formation. The likelihood of organisms experiencing such low pH conditions without coinciding low-Ω conditions is also very unlikely (Fig. 1 and Supplementary Table 1). Therefore, the conclusions from this study do not contradict the importance of pH on marine bivalve larvae, but rather highlight the overwhelming significance of saturation state at this critical bottleneck for bivalve larvae.

We have previously argued¹⁷ that during PDI shell formation in bivalve larvae the rapid rate of calcification (as shown in Fig. 4) and increased exposure of crystal nucleation sites to seawater puts an important kinetic-energetic constraint on the larvae; thereby mandating an Ω_ar sensitivity (as in equation (1)). The classical representation of the calcification rate (r) following the standard empirical formulation is:

Representative scanning electron micrographs of Pacific oyster larvae at 10 h (a), 14 h (b) and 16 h (c) post fertilization. Over the course of development from a to c the formation of the periostracum (wrinkled) is seen, followed by increasing amounts of hardening by calcium carbonate until, by 16 h, the prodissoconch I shell is formed and fully calcified and the periostracum is taut over the shell surface. Larvae were reared at 23 °C and salinity = 34, under atmospheric CO₂.

In marine waters, the increase of p_CO2 decreases saturation state and pH, but their declines approach potential thresholds differentially. We have plotted the change in Ω_ar and pH as p_CO2 increases for typical upwelling conditions in Oregon’s coastal waters (Fig. 5). We found acute responses of bivalve larvae begin to manifest at saturation states (Ω_ar) of ∼1.2–1.5 (Figs 2 and 3). Other studies have documented sub-lethal chronic exposure effects in Pacific oyster larvae (∼2.0: refs 18, 35), Olympia oyster larvae (∼1.4, ref. 19), Eastern oyster larvae (∼1.9, ref. 11), and California mussel larvae (∼1.8, ref. 37). Although far from an exhaustive list of experimental studies, placing these Ω_ar values in context of the present conditions in the California Current ecosystem illustrate two key points. First, there is limited remaining capacity for Oregon’s coastal waters to absorb more CO₂ before sub-lethal Ω_ar thresholds are crossed for bivalve larvae. Increasing atmospheric CO₂ pushes saturation state across these thresholds more frequently and with greater magnitude in the California Current^{38, 39}. Second, these saturation state thresholds will be crossed long before recently documented pH changes found to be physiologically important in molluscs⁸ (often >0.3 pH units). If transient conditions during spawning are unfavourable for bivalve larvae, in hatcheries or in the wild, then these impacts would result in diminished larval supply and recruitment to adult populations.

Conditions calculated for total alkalinity = 2,300 μmol kg⁻¹, temperature = 13 °C, and salinity = 33. Symbols are values of p_CO2. Chronic and acute effects due to saturation state decreases from experiments have been noted for bivalve larvae. The Δ 0.3 pH was previously noted as significant to many physiological processes in molluscs⁸.

Water collection and stripping dissolved inorganic carbon.

For each experiment, 1 μm filtered seawater was collected from Yaquina Bay. The alkalinity was reduced by the addition of trace metal grade HCl in near-alkalinity equivalence, followed by bubbling with ambient air for 48 h to strip (DIC) as CO₂. The acidified, stripped seawater was then 0.22 μm-filtered, pasteurized, and stored at 2–5 °C. Before treatment manipulation, the seawater was bubbled with 0.2 μm-filtered outside air until atmospheric conditions were achieved, then carbonate DIC and alkalinity values were determined for manipulations.

Experimental manipulation.

A 4 × 4 factorial experimental design was developed to target 16 total treatment combinations of p_CO2 and Ω_ar (saturation state with respect to aragonite; Fig. 1 and Supplementary Table 1), with triplicate 500 ml biological oxygen demand (BOD) bottles per treatment. Two separate experiments were conducted with each species. DIC and alkalinity concentrations were calculated for each of the 16 target treatment combinations (p_CO2 and Ω_ar). Experimental treatments were created by gravimetric addition of mineral acids and bases to the decarbonated seawater in gas-impermeable bags customized with Luer lock fittings. Aliquots of a concentrated, ambient-p_CO2, solution of Na₂CO₃ and and NaHCO₃ were added to adjust DIC to target treatment levels followed by 0.1N HCl to adjust alkalinity. Immediately following chemical manipulation, the bags with treatment water were stored without head-space at 2–5 °C for up to several weeks before spawning broodstock. Antibiotics were added to BOD bottles (2 ppm chloramphenicol and 10 ppm ampicillin),