Phanerozoic variation in dolomite abundance linked to oceanic anoxia

In the Li et al. (2021) paper, we presented a geodata set that shows the dolomite abundance throughout the Phanerozoic and linked the variations to oceanic anoxia. We welcome this opportunity to further discuss the origin of Phanerozoic dolomites in response to the Comments of Rivers et al. (2021) and Hood and Wallace (2021).

In the Li et al. (2021) paper, we presented a geodata set that shows the dolomite abundance throughout the Phanerozoic and linked the variations to oceanic anoxia. We welcome this opportunity to further discuss the origin of Phanerozoic dolomites in response to the Comments of Rivers et al. (2021) and Hood and Wallace (2021).
Firstly, we agree with both sets of authors that massive dolomite horizons of the Phanerozoic are largely a replacement product of marine calcium carbonate during diagenesis, but the proportion of such type of dolomites (secondary dolomites) in the geological record remains unknown. The dolomites of the Phanerozoic can be mainly categorized as "primary" and "secondary" types. Primary dolomites refer to dolomites that formed by in situ stabilization within the water column or pore waters during early diagenesis. Whereas secondary dolomites result from a process whereby Mg 2+ ions replace Ca 2+ ions in calcite in both shallow-and deep-burial stages.
Hood and Wallace state that primary and secondary dolomites can be distinguished by petrographic analysis. However, it is a great challenge to determine the type of ancient dolomites based on petrological observation, because most dolomites undergo recrystallization that alters their primitive textures (e.g., Land et al., 1975). Two examples are given here: (1) A spherulitic inclusion is surrounded by dolomite cement in Figures 1A and 1B. The petrographic appearance suggests these crystalline dolomites are of secondary types. However, geochemical characteristics ( 13 C = −2.4‰ to −3.2‰,  18 O = −3.4‰ to −3.8‰) from within the spherulitic inclusions, as well as the depleted  34 S values (−22.4‰ to −25.5‰) of the trapped framboidal pyrite (arrow in Fig. 1A), all support a bacterial origin (Nielsen et al., 1997).
(2) Early Triassic dolomites from China (Li et al., 2018) show a concentric zonation pattern under cathodoluminescence (Fig. 1C), typical of many secondary dolomites. However, abundant fossilized bacteria ( Fig. 1D) were found in the core of these dolomites, indicating the initial core was of primary origin. Both examples show the difficulty of determining dolomite type from petrographic appearance. Therefore, distinguishing primary from secondary dolomite remains a problem that our geodata aim to address. The rationale behind our use of the data set to determine the type of dolomite is the idea that primary dolomites formed in the water column or at the very shallow burial stage in which changes in the marine environment can pose an impact on the formation of dolomites. In contrast, dolomitization (secondary dolomites) at the deep burial or later diagenetic stages would not be affected by changes in oceanic conditions. Therefore, couplings among oceanic conditions, biological events, and dolomite abundance might be good indicators for primary dolomite. Hood and Wallace argue that many dolomite occurrences in our data set are results of dolomitization during deep burial. Such dolomitization is indeed a common geological process and usually takes place on a regional-basinal scale (e.g., Machel, 2004). It is therefore, unlikely to account for the peaks in dolomite abundance that we find are associated with environmental perturbations and mass extinctions. Rivers et al. argue that the method we utilized for calculating dolomite abundance allows thin units to skew abundances and that, if the data are normalized for total carbonate thickness, there would be dolomite peaks in the Furongian, Middle Jurassic, and Miocene. However, we prefer to assess the occurrence of dolomite units because this provides a measure of global extent, whereas section thicknesses are controlled by other factors such as accommodation space availability controlled by local tectonics and sea-level variations. Rivers et al. also argue that only 5 out of 12 documented anoxic events appear to correlate with local peaks in our dolomite abundance curve (Li et al., 2021, our figure 1), making the "link" specified by our title tenuous. The non-coincidence between anoxia and dolomite peaks are confined to three intervals: the late Cambrian, the Triassic-Jurassic boundary, and within the Cretaceous. It may be that anoxia was not sufficiently extensive at this time (or was restricted to deeper oceanic waters in the case of the Cretaceous examples) to stimulate widespread dolomite formation. Coupled mass extinction and global oceanic anoxia appears to be required to achieve dolomite peaks, as suggested by our geodata set in our figure 1.

Rivers et al. discuss several Cenozoic examples from
Qatar and state that these dolomites are replacive in origin (secondary dolomites). However, the replacement process has never been achieved in laboratory experiments at room temperatures, even with high Mg/Ca (Land, 1998), and the lack of conclusive textural evidence for the replacement of aragonite by dolomite makes the replacive model problematic (Brauchli et al., 2016). Instead, the close association between microbial mats and dolomites suggests that microbes are involved in the dolomite formation (Bontognali et al., 2010). We agree with Rivers et al. that dolomite can form in oxic pore waters, and abiotic dolomites can be synthesized at low temperatures. But the experiment of Liebermann (1967) that Rivers et al. cited was performed at 43 ± 2 ℃, a temperature that is too high for most Phanerozoic seas according to isotope thermometry (Song et al., 2019). A series of experiments at room temperatures (25 °C) have shown that aerobic microbes (Sánchez-Román et al., 2009), anaerobic microbes (Vasconcelos et al., 1995), and organic and inorganic compounds (Zhang et al., 2012) all can catalyze the precipitation of primary dolomites. Among the anaerobic microbes, sulfate-reducing microbes are the most studied since the discovery of primary dolomites in a coastal lagoon (Vasconcelos et al., 1995). Moreover, abiotic compounds that can promote primary dolomite formation such as HScarboxyl groups are, to some extent, the by-product of microbial activities.
Finally, Rivers et al. show, using an example, that marine hydrocarbon source rocks (mudstone) commonly bear calcite rather than dolomite, as a supposed counterpoint to our model. However, this example is inapplicable to our study because our model is based on the data that only include marine carbonate rocks, whose depositional environment is different from that of clastic rocks. Hood and Wallace argue that the lack of consideration of the depositional settings in our data set means that many dolomite occurrences may be unrelated to marine environmental conditions and biodiversity. In fact, all sections we considered were deposited in the marine environment including the section from Fu et al. (2004), which was deposited in the deep basin and then subjected to subaerial exposure. Hood and Wallace also argue that dolomite formed in restricted near-shore environments does not necessarily reflect oceanic conditions or ecosystems. However, sulfur isotope records from the Middle and Late Triassic evaporites that were deposited in restricted environments show good agreement with those of open marine sulfate (Bernasconi et al., 2017), indicating that restricted near-shore environments can also reflect the geochemistry of open oceans.
The dolomite problem has lasted for more than 200 years and it will continue. Numerous genetic models had been proposed for dolomite formation, including the early Sabkha model, a mixing model, and the latest microbial model. The development of new techniques and the application of large geodata sets will hopefully continue to shed new light on these old problems.