The more we learn about something, the more complex it tends to become. This is nicely demonstrated by the example of calcite-aragonite seas. Secular changes in the prevalent mineralogy of abiotic calcium carbonate precipitates have long been known, since Sandberg’s (1983) discovery of Phanerozoic oscillations. Abiotic carbonates (e.g., oolites, cements) precipitate in equilibrium with ambient water, thus their composition may inform us about the chemical state of seawater. Three periods of aragonite seas alternating with two of calcite seas from the latest Precambrian to today have been recognized. Sandberg was wrong about the driver—he favored CO2 levels over Mg/Ca—but he got the pattern right. The hypothesis that the Mg/Ca ratio in seawater was a dominant control, not only of the mineralogy of abiotic precipitates (Hardie, 1996) but perhaps also of the prevalence of sediment-producing calcifiers and reef builders (Stanley and Hardie, 1998), has led to renewed interest in these trends in the late 1990s. Morse et al. (1997) showed experimentally that temperature may be important, but most authors embraced tectonically driven fluctuations in Mg/Ca as prominent, largely ignoring temperature.

Balthasar and Cusack (2015, p. 99 in this issue of Geology) use an advanced experimental approach to revisit the temperature-versus-Mg/Ca dependence in precipitation of abiotic calcium carbonate phases. Surprisingly, and in contradiction of previous work, they find that aragonite and calcite co-exist over a wide range of experimental conditions. There is a gradual change in the proportion of mineral phases with variation in Mg/Ca and temperature, rather than a threshold at which mineral phases change from one to the other. Also surprising is the weak influence of CaCO3 saturation state and pCO2. The calcite content in experimental precipitates increases at lower temperatures and Mg/Ca, but boundary conditions for pure calcite seas are so extreme that they are unlikely to have been met during the past 540 m.y. For example, aragonite proportions <1% would be expected at Mg/Ca of 1 and temperatures of <15 °C. Such low Mg/Ca may have occurred in extreme “calcite seas,” but the low temperatures were unlikely in the Phanerozoic tropics. Balthasar and Cusack emphasize that substantial geographic variation in aragonite/calcite proportions must have existed, especially in calcite seas. This agrees with common observations of aragonitic ooids in tropical calcite seas (Prasada Rao, 1990; Adabi, 2004). Pure aragonite seas are more plausible given the experimental results. Abiogenic aragonite thus would more commonly occur in calcite seas than calcite in aragonite seas.

Sandberg (1983) realized that aragonite and calcite may co-occur under aragonite-facilitating conditions, but probably underestimated their overlap. Balthasar and Cusack’s results, combined with differences between Mg/Ca models and uncertainties in paleoclimates, make the classical concept of calcite and aragonite seas fuzzy: there is little black and white, but a lot of gray.

The surprisingly strong impact of temperature on mineral phases might be developed into a paleotemperature proxy in deep time, if Mg/Ca in seawater and the proportion of aragonite in abiotic carbonates become better constrained. Stable oxygen isotopes are still our major source for paleotemperature estimates, but have problems due to diagenetic impacts (Pearson et al., 2007), and lack of knowledge of the oxygen isotopic composition of seawater, which is dependent on continental ice volume and may have changed over the Phanerozoic (Veizer et al., 1997). Where calcite-aragonite co-exist in experiments, (Mg/Ca: 0.5–2.5; Balthasar and Cusack, their figure 1), we can derive a multiple linear regression model to estimate temperature (T in °C) at a known percentage of abiogenic aragonite (A) and seawater Mg/Ca:

The adjusted R2 for this regression is 0.73 (p < 0.001), sufficient to justify more experimental work. Given the restrictions on Mg/Ca in the model, the proxy would mostly be applicable to calcite seas; that is, from Cambrian Stage 2 to the Mississippian, and from the Late Jurassic to the Paleogene. Although Mg/Ca varies among models, fluid inclusions and other geochemical proxies provide independent evidence (Lowenstein et al., 2001; Siemann, 2003), so that ambient Mg/Ca may soon be better constrained than traditional “global” temperature. Quantifying the proportions of original aragonite and calcite in ancient samples is challenging, but criteria appear robust (Sandberg, 1983; Wilkinson et al., 1985).

Can such fuzzy seas exert control on biomineralization and the evolutionary success of organisms (Stanley and Hardie, 1998; Hautmann, 2006; Porter, 2007, 2010)? Mg/Ca influences the shell composition of marine invertebrates (Ries, 2005, 2010; Ries et al., 2006), but these experiments kept temperature constant, so that we do not know the effects of changing temperature on skeletal mineralogy. If the results for abiotic carbonates are applicable to skeletal organisms, aragonitic organisms could thrive in calcite seas, at least in the tropics and subtropics, whereas calcitic organisms in aragonite seas might have a harder time, for which there is some evidence. Clades originating or acquiring skeletons in aragonite seas all had an original aragonitic mineralogy, but some aragonitic clades first appeared in calcite seas (Porter, 2010). This asymmetry may be stronger than Porter acknowledged, as more aragonitic taxa than shown emerged in the early Paleozoic calcite sea (Balthasar et al., 2011). We would also expect a latitudinal gradient in skeletal mineralogy, with more calcite at high latitudes. Indeed, modern and ancient cool-water carbonates are characterized by predominantly calcitic skeletons (Nelson, 1988), whereas todays’ tropics are dominated by aragonitic organisms.

The overall match between skeletal mineralogy and inferred oceanic state, however, is not as good as often claimed, even when climate change is taken into account. More substantial changes in mineralogical proportions are observed across mass extinction episodes than over the transition from one oceanic state to another (Kiessling et al., 2008). Aragonitic taxa increased in abundance after the Permian-Triassic mass extinction in an aragonite sea during substantial warming (Sun et al., 2012), in line with Balthasar and Cusack’s findings. However, aragonitic taxa decreased in the warming aragonite seas across the end-Triassic mass extinction, and increased in the calcite sea across the Cretaceous-Paleogene (K/Pg) boundary. According to Equation 1, and assuming Mg/Ca of 1.5, it would require a temperature increase of 15 °C to explain the 9% increase of epifaunal aragonitic taxa across the K/Pg (Kiessling et al., 2008), a time without dramatic long-term climate change (Kemp et al., 2014). Finally, factors independent of Mg/Ca may govern the fate of skeletal clades. For example, the Cretaceous coral-to-rudist transition in reefal habitats was emphasized by Stanley and Hardie (1998) as forced by decreasing Mg/Ca (see also Ries et al., 2006). Stanley and Hardie argued that the bimineralic, dominantly calcitic rudist bivalves did not directly benefit from the lowered Mg/Ca, but the aragonitic corals could not compete successfully in the calcite sea and thus declined relative to rudists. However, the rising middle Cretaceous temperatures (Wilson and Norris, 2001; Forster et al., 2007) should have facilitated growth of the corals, but they obviously did not. Warming may have imposed physiological stress on corals, whereas rudists were perhaps better adapted to warming, thus gaining competitive advantage. In summary, the experimental results for abiogenic calcium carbonate phases may not be directly transferable to organisms because physiology plays in.

Resolving fuzzy-sea scenarios can be achieved in many ways. Most importantly, calcite, aragonite, and gray states must be better constrained in time and space. Little progress has been made quantifying the proportional mineral abundance in oolites and cements since the mid-1980s. New observations need to be quantified more rigorously using point counting and, ideally, compiled in an online database, recording the stratigraphic as well as the (paleo-)geographic setting. We also require more solid data on seawater Mg/Ca to fine-tune models. Experiments on skeletal organisms should manipulate both Mg/Ca and temperature, and geochemical models should incorporate temperature more explicitly. Clearly, the last word on calcite-aragonite seas has not been spoken. Stay tuned for the next discoveries!

I thank Martin Aberhan and Axel Munnecke for commenting on a draft of this paper and Ellen Thomas for editing.