Hardie (2003) identified intervals of “aragonite” and “calcite seas” by modeling Mg/Ca ratios in Neoarchean to Neoproterozoic seawater using estimates of hydrothermal and weathering fluxes of key ions. His results agree with most published reports of aragonite pseudomorphs with the exception of the time period between 2.5 and 2.6 Ga. During this interval, his model predicts calcite seas, but aragonite pseudomorphs have been reported (Hardie, 2003). Hardie's preferred interpretation for the difference between model results and the presence of aragonite pseudomorphs is that the described pseudomorphs were not originally aragonite. I disagree with this interpretation.
Evidence for Neoarchean Aragonite Precipitation.Sumner and Grotzinger (2000) documented evidence for aragonite precipitation in four geographically diverse carbonate units spanning 2.9–2.5 Ga. The 2.6–2.5 Ga Campbellrand-Malmani carbonate platform, South Africa, contains abundant pseudomorphs that Hardie (2003) argues probably were gypsum. In this 2-km-thick platform, aragonite pseudomorphs are abundant and consist of fibrous marine cements in addition to radiating bundles of decimeter-long crystals. The fibrous cements contain most of the features typical of calcite replacing aragonite (Loucks and Folk, 1976; Assereto and Folk, 1980; Mazzullo, 1980; Sandberg, 1985; Peryt et al., 1990), including: (1) relict crystal morphology defined by inclusions and [Mn] that demonstrates the crystals were fibrous and had blunt to feathery fiber bundle terminations; (2) replacement by optically unoriented, equant to elongate calcite crystals with unit extinction; and (3) strontium concentrations of up to 3700 ppm.
The fibrous cements show no morphological similarity to any reported gypsum textures; they are entirely analogous to Phanerozoic aragonitic cements using the same identification criteria used in carbonate platforms of any age with the exception that relict aragonite inclusions have not been identified (see Sumner and Grotzinger, 2000).
Some large botryoidal and prismatic crystal pseudomorphs interpreted by Sumner and Grotzinger (2000) show similarities to gypsum pseudomorphs with respect to the size and geometrical arrangement of crystals. However, crystal morphology, petrographic characteristics, and geochemical compositions are similar to the fibrous cement pseudomorphs rather than gypsum pseudomorphs. The only similarity in crystal properties to gypsum are pseudohexagonal cross sections in some prismatic pseudomorphs, which is also consistent with an aragonite precursor. Measurements of interfacial angles are consistent with either an aragonite or gypsum precursor due to the sensitivity of results to small errors in cross section orientation, the absence of micrite drapes on pseudomorphs, and the originally fibrous character of macroscopically prismatic pseudomorphs. Errors were estimated to be 5–10°, which are too high to aid in primary mineral identification. However, the petrographic and geochemical characteristics provide very strong evidence for an aragonite precursor mineralogy for most decimeter-scale pseudomorphs (Sumner and Grotzinger, 2000).
Gypsum Pseudomorphs. Rare gypsum pseudomorphs are present in Neoarchean carbonates (e.g., Simonson et al., 1993) and have been distinguished from aragonite pseudomorphs based on crystal morphology and replacement petrographic characteristics (see Sumner and Grotzinger, 2000, for discussion). The presence of some gypsum pseudomorphs does not require reinterpreting all aragonite pseudomorphs as gypsum pseudomorphs, as suggested by Hardie (2003). The identification of both precursor minerals adds confidence that the decades of high-quality research on methods to distinguish primary mineralogy in ancient sediments is consistent and reliable when suites of properties are considered together.
Alternative Interpretation of Model Results. The presence of aragonite pseudomorphs is well documented in 2.5–2.6 Ga carbonates deposited in storm- and tidally influenced facies (Sumner and Grotzinger, 2000; Sumner, 2001). The presence of Mg-calcite cements in the same platforms (Simonson et al., 1993; Sumner and Grotzinger, 2000) also supports a sea-water chemistry typical of “aragonite seas” during late Neoarchean time. Thus, alternative interpretations of the disagreement between Hardie's (2003) model results for aragonite-calcite seas and observational data should be considered. Three possible origins for this disagreement include: (1) the ionic composition of average river water was different when the atmosphere contained little or no O2 and more CO2; (2) the flux of river water was different when continental mass, freeboard, and climate may have been significantly different; (3) plutonic rock generation in North America may be an inaccurate representation of Neoarchean seafloor spreading rates.
Overall, evidence for aragonite precipitation from 2.6 to 2.5 Ga sea-water should be used to constrain models of Neoarchean ocean chemistry rather than being discarded as unlikely on the basis of early attempts to model chemical fluxes.