Dissolved inorganic carbon (DIC) is one of the most important chemical equilibria of the oceans and is largely responsible for controlling the pH of seawater [11–13]. DIC exists in seawater in four forms: CO₃²⁻, HCO₃⁻, aqueous CO₂ and carbonic acid (H₂CO₃). At a pH of 8.2, ∼88% of DIC is in the form of HCO₃⁻ and ∼11% in the form of CO₃²⁻. Only ∼0.5% is in the form of CO₂ and H₂CO₃. When CO₂ dissolves in seawater, the net effect is to increase the concentrations of H₂CO₃, HCO₃⁻ and H⁺, and decrease the concentration of CO₃²⁻, thereby lowering pH. Atmospheric CO₂ (pCO₂) today is ∼100 ppm greater than the pre-industrial value of ∼280 ppm causing the average ocean surface pH to drop by 0.1 pH unit, which represents a H⁺ increase of ∼30% [14]. This pH decrease may seem small; however it is a change in one of the most basic environmental norms of marine life to have occurred throughout the entire genetic history of most species.

The extent to which calcifying marine life is affected by such changes largely depends on the saturation state of CaCO₃ (Ω) the product of Ca²⁺ (estimated from salinity) and CO₃²⁻ (calculated from DIC and total alkalinity) divided by a solubility constant. Ω is different for the three forms of CaCO₃ produced by marine life: high magnesium calcite (secreted by coralline algae), aragonite (secreted by corals, macro-algae and some molluscs) and calcite (secreted by most molluscs, forams, also Palaeozoic corals). Saturation states of surface waters are highest in the tropics and lowest at high latitudes mainly because CO₂ is more soluble in colder waters, thus lowering the concentration of CO₃²⁻. CaCO₃ solubility also increases with depth, the depth at which CaCO₃ becomes soluble being known as the saturation horizon. The depth of this horizon varies latitudinally and tends to shallow toward the poles, but also varies locally due to changes in ocean circulation patterns, such as in regions of upwelling. The depths of the saturation horizons for calcite, aragonite and high magnesium calcite are very different.

The aragonite saturation horizon (ASH) currently varies between ∼100–800 m depth at the equator and [15] and has been estimated to be decreasing at a rate of ∼1–2 m/yr [16]; however on the west coast of North America large areas of the continental shelf are now impacted by aragonite-undersaturated waters where upwelled water is further acidified by uptake of CO₂ from the atmosphere [17].

Calcification by marine biota is not known in the detail of ocean chemistry [18,19] and it is also much more complicated, especially where it involves algal symbionts which respond to different environmental parameters. The list of actual or potential variables affecting physiological responses is indeed long: temperature, light, environmental history, acclimation, geographic location (temperate or tropical), respiration rate, photosynthesis, net productivity, reproductive effort, symbionts (type and density and growth rate) and form (massive or branching). Some of these variables acting in synergy are already known to cause different responses to change in pCO₂ and Ω_arag. For example, coral calcification has been found to decrease [18,20,21], increase [22], or be unaffected [20,23–26] by temperature at increased pCO₂ in different taxa in different experimental regimes. Other organisms, especially those who can utilize enhanced levels of HCO₃⁻, have shown increased rates calcification created by increased pCO₂ [27–29].

In principle, OA can be accommodated by the same mechanisms that organisms use to counter metabolically produced H⁺ and CO₂. These include passive buffering of intra- and extracellular fluids, control of ion exchange across semi-permeable membranes and physiological mechanisms to accommodate short-term metabolic spikes. However, corals do not have dedicated excretory organs as do many higher invertebrates and no marine life has the option of suppressing metabolic activity to wait out periods of OA. Similarly, none of the options for compensating short-term acidity such as dissolution of tests or shells are available over generational time spans. One review of short-term responses to acidified water [12] cites 39 studies indicating a range of deleterious impacts of OA on marine life. The same limitations potentially apply to freshwater habitats where the pH is much less buffered, and is commonly <7. Freshwater habitats with low pH exclude most calcifying invertebrates except molluscs and crustaceans which have compensating shell chemistries and elaborate excretory organs. However, there are rare instances where such adaptation also occurs in the marine realm such as in mussels near hydrothermal vents [30].

The oceans permit marine life to construct carbonate exoskeletons and allow these skeletons to be consolidated into wave-resistent reef because they are supersaturated with carbonates—a physiological opportunity exploited by corals today and by other taxa in the geological past. Under natural conditions coral calcification is widely believed to be primarily dependent on Ω_arag [22–25,31–38], also Ca²⁺ [39] rather than pH. Deposition of this aragonite is an extracellular process where an organic matrix combined with minerals is precipitated in diurnal to seasonal bands by a single cell layer, the calicoblastic epithelium, adjacent to the skeleton. This layer facilitates transport of Ca²⁺ ions and DIC into a minute space filled with a chemically buffered fluid—the calicoblastic space—between the epithelium and the skeleton. The primary source of carbon is internal metabolic CO₂ which is in excess during daylight when skeletal formation is maximized. A secondary source is DIC in external seawater [40]. Either way, a cation exchanger precipitates aragonite at the skeleton surface.

There are a wide array of options available to corals for regulating skeleton formation, the most apparent being the physiological isolation of the calicoblastic space from the surrounding seawater by the actions of buffers, an extracellular control mechanism requiring increased energy expenditure [41]. It can also be regulated by controlling carbon exchange between corals and their zooxanthellae in zooxanthellate taxa exhibiting light-enhanced calcification, an intracellular control mechanism [42,43]. Potentially, OA can disrupt these mechanisms by decreasing CO₃²⁻ supply, decreasing pH, or by altering the buffering system [25].

In broader terms, the subject of coral calcification has remained a black box for over thirty years [44] and to a large extent it still is. Elucidation of the internal workings of this box will certainly shed light on the ways OA might affect corals and it may well be the case that different taxa have different control mechanisms. This would be hardly surprising given that that most scleractinian families diverged in the Mesozoic Era and that the enhancement of calcification by zooxanthellae has evolved independently several times [45]. It is even possible that different colonies of the same species can adopt different controls depending on the environment in which they occur. However, the external view of the box is clear enough: coral calcification usually varies with CO₃²⁻ availability and, given their physiological dependence on the properties of seawater, it is also dependent on adequate Ω_arag. Just how high Ω_arag must be for adequate calcification will be a sliding scale from a pre-industrial high of 3.6 at the equator to a range of numbers depending on taxon, habitat, depth and location. The bottom of the scale can be reasonably estimated to be ∼2, which will occur in most reef regions this century (see below). However, given that Scleractinia characteristically have a wide range of strategies to deal with almost anything there will likely to be many exceptions to the norm; e.g., the finding that the Atlantic Ocean coral Favia fragum can still grow in seawater with Ω_arag = 0.22, although with growth abnormalities [38].

It is not yet possible to estimate the proportion of marine species occupying coral reefs; however as the diversity of most metazoan phyla peaks in reef habitats, that number is likely to be at least 25% of total marine biodiversity. Impacts on reef diversity are therefore of major global significance. Furthermore, because most reef habitats are of carbonate construction, any loss of coral diversity will have a major knock-on effect to all reef-dwelling taxa.

One natural impact of acidification on reef corals is via rainwater which commonly has a pH of 5.6 or less. Low salinity caused by rainwater can decimate reef flat corals, but those which survive are certainly tolerant of short-term acidity. Microatolls of Porites [46] are one such survivor. These have an outer rim of living colony surrounding a large central area which is usually excavated. This excavation may be due to bioerosion, but where water is regularly ponded for several hours each tidal cycle in areas of high rainfall it is likely to be enhanced by dissolution. Whether this is so or not, the living part of the colony shows a high resilience to repeated short-term acidity.

Extensive degradation of open-ocean coral communities in the vicinity of urban coastal developments is now common, especially in south-east Asia and the Indian Ocean perimeter. In most cases the primary cause (including sedimentation, salinity change, physical damage, mass bleaching and Acanthaster outbreaks) is visually evident; however clear-water impacts are seldom studied although many are potentially valuable sites for water quality assessments which incorporate local factors such as community metabolism, sediment dissolution, freshwater input and organic carbon cycling.

Natural CO₂ springs are widespread and clearly offer natural experiments of OA impacts. One such occurrence was a spring on the Italian coast where naturally occurring species with carbonate skeletons were displaced in favor of species without skeletons [47]. Similar occurrences in coral reefs have recently been observed in Indonesia (Komodo), not yet studied, and Papua New Guinea (Milne Bay) where detailed studies are currently underway [48]. The study demonstrated a wide range of tolerance among coral taxa, with Porites being the most tolerant and fast-growing and structurally complex corals such as branching Acropora the least. Down to pH 7.8, the primary impact of OA was therefore a reduction in biodiversity, structural complexity and coral juvenile densities rather than coral cover, as well as an increase in macroalgae and bioeroders. The study also showed that coral reef development ceased at a pH of less than 7.7 units.

Latitudinal sensitivity to OA will become an increasingly important issue during the 21st century as diversity changes resulting from the poleward migrations of tropical species are offset by predicted impacts of high latitude OA [8,49]. To date there no studies of tropical zooxanthellate corals demonstrating this, existing studies being confined to the temperate or azooxanthellate species Cladocora caespitosa [26] Lophelia pertusa [50] and Oculina arbuscula [51].

Sensitivity to depth is not relevant to reef corals, but it is to azooxanthellate taxa which have a similar diversity and which commonly occur to depths of >1000 m (but occasionally >6000 m), perhaps limited by the ASH, especially in the northern Pacific where the species diversity is low. The potential impact of the decreasing ASH [52] will occur over a wide latitudinal range, impacting the large bioherms of Lophelia found off the Scandinavian coast [53]. Presumably there will be a time-delay in the transmission of decreased Ω_arag from the surface to the depth where these corals occur, however acidification impacts are likely to have already commenced and, irrespective of impacts on living corals, acidification will impact deep sea bioherms by compromising the structural integrity of the framework.

By far the most damaging ecological effects of OA are not on individual taxa but on the group of corals responsible for habitat construction. Coral reefs can only exist if macroalgae are kept in abeyance by herbivores, especially fish [54]. Most herbivores, especially herbivorous fish, need the protective three dimensional habitat provided by branching corals, especially Acropora, in larval as well as adult and phases of their life cycle. Should these corals cease to grow other corals and a high proportion of other life will follow as macroalgae become dominant.

Although many geologists are at the forefront of climate change science, some have expressed skepticism of predicted impacts of OA just as others are, or were, skeptical of the existence of anthropogenic CO₂-driven climate change. In cases where biological and geological sciences meet, there is great potential for misinterpreting concepts of time [55]. Biological changes on reefs in geological time exhibit a boom-to-bust existence where the same carbonate platform is repeatedly re-populated by the same or different suits of species over intervals as short as millennia. However, reefs as geological structures are more enduring than any other biologically-derived structures on Earth. For this reason, the geological history of a reef usually says little or nothing about biologically-significant impacts over the same time interval. In this context, palaeontological and chemostratigraphic records are scattered points which have scant relevance to the fragility of living reefs or the stability of geological platforms. This type of discrepancy is analogous to the reconstructions of pCO₂ from ice cores: the zig-zag nature of ice core records is not seen in other geological sources such as reef cores and ocean sediments, yet it is the ice core record that revealed the history of environment change that profoundly affected all ecosystems.

Geological indicators of pCO₂ (pCO_2paleo) have been referred to by several authors (in non peer-reviewed publications) as evidence that reefs will not be affected by today's increases in pCO₂. In the ice-core analogy above, these records are scatter points (derived from densities of leaf stomata; carbon isotope composition of fossil soils, boron isotope composition of marine carbonates and modeling) within geological history, points which may have nothing to do with the biological ‘boom-to-bust’ cycles of reef growth with which they are being correlated. Reefs accretion may in fact cease during intervals of high pCO_2paleo without any record of this cessation being preserved. Although high pCO_2paleo does not necessarily cause OA (because it is countered by the weathering of carbonate rock in geological time), some points of high pCO_2paleo seem sufficiently well linked to intervals of reef development to indicate a tolerance for OA that exceeds any predicted to occur this century.

In broad perspective, the geological record appears to show that reefs have contracted or ceased development during geological intervals when pCO_2paleo is both high and low, although the former is clearly dominant [56,57]. At least 27 extinction events involving reefs have been identified [58,59], the best known being the five great mass extinctions that obliterated a high proportion of the earth's marine and terrestrial life. The causes of mass extinctions have attracted a wealth of study culminating in many hypotheses [60]. However, when compared with each-other, and in the light of biological knowledge, the most credible common cause is some sort of disruption of the carbon cycle, sometimes related to a slow geological process such as volcanism from sea-floor spreading, sometimes to an abrupt event such as a bolide impact. This strongly implicates OA and ocean anoxia as major drivers of mass extinctions [61] although direct geological evidence exists only for the end-Permian [62,63] and end-Triassic mass extinctions [64,65]. However, pCO_2paleo is clearly implicated in four of the five greatest reef crises (not necessarily mass extinctions) of all time [66]. The best studied among the latter is the Paleocene-Eocene Thermal Maximum (PETM), where a peak of pCO_2paleo, largely from methane [67], impacted the reefs of the Tethys [68], at that time the centre of coral diversity, causing a significant diversity reduction.

One long-term change in ocean carbonate chemistry with unclear relevance to OA is the alternation of ‘calcite seas’ (of low magnesium calcite) during greenhouse intervals with ‘aragonite seas’ (of high magnesium calcite and aragonite) during icehouse intervals. This change in the Mg²⁺/Ca²⁺ ratio of seawater is correlated with high levels of pCO_2paleo as both are driven by changes in ocean crust production [69]. These secular changes in ocean carbonate chemistry correlate well with mass extinctions. During the end-Permian extinction, low magnesium calcite-secreting taxa (including all Palaeozoic corals) were replaced by aragonite-secreting taxa [57] among which were the first Scleractinia. During much of the Jurassic and on into the Cretaceous, seas again became calcitic and so did a significant proportion of the bivalve fauna of the time [70] including rudist bivalves (a major reef-builder of the mid-Cretaceous) and, curiously, at least one coral, Coelosmilia [71]. Indeed, two modern Mediterranean corals, Oculina patagonica and Madracis pharensis lose their skeletons when placed in highly acidified aquaria, then re-grow them when returned to normal seawater [72], a result in keeping with the intriguing ‘naked coral hypothesis’ where corals are proposed to live on, without skeletons, through geological intervals of OA [73].