Episode of intense chemical weathering during the termination of the 635 Ma Marinoan glaciation

Significance

At least two extreme glaciations (dubbed as snowball Earth events), each lasting millions of years and affecting the entire globe, occurred during the Cryogenian Period (∼720–635 Ma). These glaciations are recorded by globally distributed glacial deposits overlain by cap carbonates. According to the snowball Earth hypothesis, cap carbonate deposition was driven by intense continental weathering during deglaciation, but geochemical evidence is lacking. Using Mg isotopes as a proxy, we reconstruct the history of chemical weathering during and following the terminal Cryogenian Marinoan glaciation in South China. Our data confirm an episode of strong chemical weathering during the termination of the 635-Ma Marinoan glaciation, but its climax occurred before cap carbonate deposition.

Abstract

Cryogenian (∼720–635 Ma) global glaciations (the snowball Earth) represent the most extreme ice ages in Earth’s history. The termination of these snowball Earth glaciations is marked by the global precipitation of cap carbonates, which are interpreted to have been driven by intense chemical weathering on continents. However, direct geochemical evidence for the intense chemical weathering in the aftermath of snowball glaciations is lacking. Here, we report Mg isotopic data from the terminal Cryogenian or Marinoan-age Nantuo Formation and the overlying cap carbonate of the basal Doushantuo Formation in South China. A positive excursion of extremely high δ²⁶Mg values (+0.56 to +0.95)—indicative of an episode of intense chemical weathering—occurs in the top Nantuo Formation, whereas the siliciclastic component of the overlying Doushantuo cap carbonate has significantly lower δ²⁶Mg values (<+0.40), suggesting moderate to low intensity of chemical weathering during cap carbonate deposition. These observations suggest that cap carbonate deposition postdates the climax of chemical weathering, probably because of the suppression of carbonate precipitation in an acidified ocean when atmospheric CO₂ concentration was high. Cap carbonate deposition did not occur until chemical weathering had consumed substantial amounts of atmospheric CO₂ and accumulated high levels of oceanic alkalinity. Our finding confirms intense chemical weathering at the onset of deglaciation but indicates that the maximum weathering predated cap carbonate deposition.

The worldwide distribution of Cryogenian glacial deposits (∼720–635 Ma) implies the occurrence of global glaciations or snowball Earth events, with ice sheet extending to paleo-equatorial regions (1, 2). The termination of the end-Cryogenian Marinoan glaciation was followed by ocean oxygenation and diversification of complex eukaryotes (3–5), suggesting the possible linkage between glaciation and evolution (6). As a marker for the end of the Marinoan glaciation, globally distributed cap carbonates immediately above the glacial deposits indicate a sharp transition from an icehouse to a greenhouse climatic condition (7–9). The “snowball Earth” hypothesis posits that deglaciation was abrupt and was triggered by extremely high atmospheric CO₂ concentration (pCO₂) [∼350× present atmospheric level (PAL), about 10^–3.5 bar] accumulated during global freezing for tens of million years (8, 10). High atmospheric pCO₂ levels are expected to cause intense chemical weathering on continents, which, in turn, elevates seawater alkalinity and results in the rapid precipitation of cap carbonates all over the world (10, 11). This interpretation also explains the consistently negative carbon isotopic signatures of cap carbonates (10). Thus, cap carbonate precipitation coupled with intense chemical weathering is one of the key predictions of the snowball Earth hypothesis. However, direct geochemical evidence for intense chemical weathering during deglaciation is not available (12), and the temporal relationship between intense chemical weathering and cap carbonate precipitation remains equivocal. In this study, we use Mg isotopes as a geochemical proxy to constrain the intensity of chemical weathering during the deposition of the terminal Cryogenian Nantuo Formation and the overlying cap carbonate of the basal Ediacaran Doushantuo Formation in the Yangtze Platform of South China.

Tracer for the Intensity of Chemical Weathering

Traditional proxies for the intensity of chemical weathering include chemical index of alteration (CIA) and strontium isotopes (⁸⁷Sr/⁸⁶Sr). However, CIA is sensitive to the compositions of source rocks (13), and Sr isotopes are susceptible to diagenetic alterations (14). Magnesium (Mg) isotopes have several advantages to surmount these obstacles. First, owing to the limited fractionation during magmatic processes (15), the primary components of the upper continental crust (UCC, e.g., granitoids and basalts) display a narrow range of Mg isotopic compositions [δ²⁶Mg = –0.3 to –0.1, Fig. 1A; δ²⁶Mg is the per mil deviation of ²⁶Mg/²⁴Mg with respect to the standard Dead Sea Magnesium (DSM3) (16)], minimizing the source effect encountered in CIA study. In addition, unlike Sr isotopes, the Mg isotopic system in dolomite and siliciclastic rocks is immune to secondary alteration related to low-grade metamorphism and diagenesis (17, 18). More importantly, Mg isotopes significantly fractionate (up to 2‰) during chemical weathering, with the preferential retention of ²⁶Mg in weathering residues (Fig. 1A) (19–21), and, as such, δ²⁶Mg values of weathering residues become progressively higher with increasing intensity of chemical weathering (22). As shown in the weathered basalt profiles, δ²⁶Mg values of weathering residues increase gradually from unaltered bedrock to the most weathered saprolites, displaying a positive (although not necessarily linear) correlation with CIA (19, 22) (Fig. 1B). Given the aforementioned merits, Mg isotopes can serve as a more reliable tracer for chemical weathering.

Fig. 1.

Compilation of Mg isotope data. (A) Boxplots showing the Mg isotopic compositions of major Mg reservoirs. The box encompasses the 25 and 75 percentiles, the vertical line inside the box represents the median value, the whiskers represent the 2.5 and 97.5 percentiles, and the gray dots are outliers. The cyan and orange vertical dashed lines represent average δ²⁶Mg of seawater (SW, –0.83) (47) and UCC (–0.22) (32), respectively. See SI Text and Dataset S1 for references. (B) Cross-plot of δ²⁶Mg versus CIA for weathered residues in basalt weathering profiles. Green bar, fresh basalt (21, 22); cyan squares, saprolites developed from basaltic diabase dike, South Carolina (22); purple triangles, soils above basalts, Iceland (48); red circles, saprolites developed from tholeiitic basalts South China (21); orange diamonds, bauxites developed from Columbia River basalts, United States (19). (Inset) In detail, the relationship between CIA and δ²⁶Mg values of saprolites from South China and Columbia River.

Geological Setting and Sample Description

The Cryogenian succession in the Yangtze Platform consists of, in ascending order, the Chang’an/Tiesiao Formation, representing the early Cryogenian (or Sturtian) glacial deposits, the interglacial Fulu/Datangpo Formation, and the Nantuo Formation equivalent to the terminal Cryogenian or Marinoan glaciation (23). The Nantuo Formation is dated between 654 ± 3.8 Ma and 635.2 ± 0.6 Ma (24, 25), and is overlain by the 3- to 5-m-thick cap carbonate of the basal Doushantuo Formation (25, 26). Although cap carbonate atop Marinoan glacial deposits may be time-transgressive, the magnitude of this diachroneity is likely <10³–10⁴ y (9). In South China, the contact between the Nantuo glacial deposits and the Doushantuo cap carbonate is continuous, without a sedimentary break, particularly in the sampled sections that were deposited in deep basinal environments beyond the reach of wave activities (26).

The upper part of the Nantuo Formation and the overlying Doushantuo cap carbonate were sampled from two localities: a drill core (14ZK) from the Songtao county and an outcrop section (Bahuang) in the Tongren city, both in northeastern Guizhou Province, South China. These two sections are paleogeographically located in the deep-water basinal facies in Neoproterozoic (23) (Fig. S1). The Nantuo Formation recovered from the 14ZK core is 225 m thick and consists of alternating siltstone, sandstone, pebbly sandstone, and diamictite (Fig. 2). The focus of this study is on the upper Nantuo Formation, which can be divided into six lithological units (Fig. 2). Unit I is composed of 40-m-thick massive diamictite (only the upper 20 m was sampled), followed by 15-m-thick pebbly sandstone/siltstone alternations (unit II). The rest of the Nantuo Formation can be subdivided into, in stratigraphic order, 28-m-thick sandstone and siltstone of unit III, 8-m-thick pebbly sandstone intercalated with siltstone (unit IV), 10-m-thick siltstone (unit V), and 3-m-thick diamictite associated with siltstone (unit VI). In the Bahuang section, only siltstone of the topmost 1 m of the Nantuo Formation was sampled, and the sampled interval can be correlated with unit VI in the 14ZK core (Fig. S2). The Nantuo diamictites and pebbly sandstone/siltstone are generally considered as deposition in a glacial marine environment, representing proximal glaciomarine and distal glaciomarine facies, respectively, whereas the pebble-free sandstone and siltstone may represent normal marine deposition without the influence of glaciation (27, 28). The Nantuo Formation in these two sections is capped by the basal Ediacaran Doushantuo cap carbonates, which are composed of microcrystalline dolomite and dolomicrite with widespread sheet cracks (Fig. S2).

Fig. S1.

(A) Late Neoproterozoic paleogeographic map of South China Block (modified from ref. 23). Sample localities are marked by circled numbers. (Inset) Map shows the location of the Yangtze Platform (highlighted in blue) and the Cathaysia Block. (B) Geological map of studied area, showing the location of the 14ZK core and the outcrop section at Bahuang (modified from ref. 60).

Fig. 2.

Stratigraphic profiles of δ²⁶Mg and δ¹³C of the upper Nantuo Formation and the basal Doushantuo Formation cap carbonate in the 14ZK core and the Bahuang section. Units I through VI refer to six lithostratigraphic units in the upper Nantuo Formation. Pink dashed line represents the stratigraphic correlation line. For different grain sizes, c/s is clay/silt, f is fine sand, m is medium sand, c is coarse sand, and g is granules.

Fig. S2.

Field, drill core, and petrographic photographs. (A) Photograph of drill core showing sandstone/siltstone with parallel bedding in unit II of 14ZK core. (B) Photograph of drill core indicating diamictite in unit I of 14ZK. (C) Petrographic photomicrograph showing petrography of siltstone in unit IV at 14ZK. (D) Thin section photomicrograph of sandstone in unit II at 14ZK. (E) Thin section photomicrography showing poorly sorted matrix of diamictite in unit I at 14ZK under perpendicular polarized light. (F) Field photograph showing the contact between the Doushantuo cap carbonate and the underlying pebbly siltstone in the topmost part of the Nantuo Formation at the Bahuang section. (G) Massive diamictite in the Nantuo Formation at Bahuang section. (H) Thin section photomicrograph showing calcite cements infilling sheet-crack structures in the cap carbonate at Bahuang section. (I) Thin section photomicrograph of siltstone in the topmost part of the Nantuo Formation at Bahuang section. (J) Thin section photomicrograph of pebbly sandstone in the topmost part of the Nantuo Formation at Bahuang section. Arrows highlight lonestones in the Nantuo Formation.

Results

In the 14ZK core, δ²⁶Mg remains constant at ∼+0.05 in the diamictite of unit I and pebbly sandstone and siltstone of unit II, but displays a sharp increase to +0.56 at the base of unit III (Fig. 2 and Table S1). High δ²⁶Mg values (+0.56 to +0.95) persist throughout units III and IV, followed by a sharp decline to ∼ +0.14 in unit V. The topmost part of the Nantuo Formation (units V and VI) is characterized by relatively low δ²⁶Mg, varying between –0.15 and +0.28. The siliciclastic component of the Doushantuo cap carbonate in the 14ZK core also has low δ²⁶Mg, varying between +0.17 and +0.24 (Fig. 2). In the Bahuang section, both the siltstone in the top of Nantuo Formation (equivalent to unit VI in the 14ZK core) and the siliciclastic component of the cap carbonate have relatively low δ²⁶Mg values, ranging from –0.36 to +0.40 and –0.46 to +0.42, respectively. The δ¹³C of the Doushantuo cap carbonate ranges from –11.6 to –4.0 in the 14ZK core and from –5.5 to –2.8 at the Bahuang section (Fig. 2 and Table S1).

Table S1.

Magnesium isotopic compositions of standards and fine-grained siliclastic components in the upper Nantuo Formation and basal Doushantuo Formation cap carbonate, along with carbon and oxygen isotopic compositions of Doushantuo cap carbonate

Sample	Depth, m	Description	δ26Mg, ‰	δ26Mg 2SD	δ25Mg, ‰	δ25Mg 2SD	δ13C, ‰	δ18O, ‰
14ZK core
14ZK-C10	3.00	Cap dolostone	n.d.	n.d.	n.d.	n.d.	−4.41	−6.49
14ZK-C9	2.00	Cap dolostone	0.17	0.07	0.10	0.05	−11.63	−5.33
14ZK-C8	1.50	Cap dolostone	n.d.	n.d.	n.d.	n.d.	−5.60	−11.13
14ZK-C7	1.00	Cap dolostone	n.d.	n.d.	n.d.	n.d.	−4.00	−7.68
14ZK-C6	0.20	Cap dolostone	0.24	0.07	0.13	0.05	−5.83	−9.33
14ZK-C5	0	Cap dolostone	0.17	0.07	0.11	0.05	−4.80	−7.69
14ZK-C4	−0.50	Siltstone	0.18	0.07	0.10	0.05
14ZK-C3	−1.00	Siltstone	0.08	0.07	0.05	0.05
14ZK-C2	−1.48	Siltstone	0.28	0.07	0.15	0.05
14ZK-C1	−1.78	Siltstone	0.23	0.07	0.14	0.05
14ZK-1	−2.08	Diamictite	−0.15	0.07	−0.09	0.07
14ZK-2	−2.88	Diamictite	−0.04	0.07	−0.01	0.07
14ZK-3	−3.68	Sandstone	0.19	0.07	0.06	0.07
14ZK-4	−4.48	Sandstone	0.19	0.07	0.13	0.05
14ZK-5	−5.28	Sandstone	0.15	0.07	0.09	0.05
14ZK-6	−6.08	Sandstone	0.06	0.07	0.02	0.05
14ZK-7	−6.88	Sandstone	0.23	0.07	0.12	0.05
14ZK-8	−7.68	Sandstone	0.16	0.07	0.12	0.05
14ZK-9	−8.48	Sandstone	0.06	0.07	0.06	0.05
14ZK-10	−9.28	Siltstone	0.02	0.07	0.04	0.05
14ZK-11	−10.08	Siltstone	0.14	0.07	0.08	0.07
14ZK-15	−13.68	Sandstone	0.79	0.07	0.47	0.05
14ZK-19	−18.10	Sandstone	0.86	0.07	0.5	0.05
14ZK-22	−21.50	Sandstone	0.72	0.07	0.39	0.05
14ZK-26	−25.20	Sandstone	0.76	0.08	0.38	0.07
14ZK-29	−27.60	Sandstone	0.69	0.08	0.35	0.07
14ZK-31	−29.70	Siltstone	0.95	0.08	0.51	0.07
14ZK-38	−38.00	Siltstone	0.87	0.08	0.44	0.07
14ZK-42	−41.00	Sandstone	0.89	0.08	0.49	0.07
14ZK-45	−44.10	Sandstone	0.39	0.07	0.22	0.04
14ZK-48	−47.10	Sandstone	0.56	0.08	0.3	0.07
14ZK-50	−49.00	Sandstone	−0.09	0.08	−0.01	0.07
14ZK-56	−53.50	Sandstone	0.06	0.07	0.07	0.05
14ZK-60	−57.70	Diamictite	0.01	0.07	0.02	0.05
14ZK-66	−63.20	Diamictite	0.09	0.07	0.09	0.05
14ZK-71	−66.80	Diamictite	0.05	0.07	0.09	0.05
14ZK-77	−72.70	Diamictite	0.05	0.07	0.06	0.05
14ZK-80	−74.80	Diamictite	0.06	0.07	0.09	0.05
14ZK-83	−78.50	Diamictite	0.09	0.07	0.08	0.05
Bahuang section
BH-20	1.85	Cap dolostone	−0.13	0.07	−0.02	0.06	−3.28	−6.08
BH-18	1.55	Cap dolostone	−0.11	0.07	−0.02	0.07	−2.83	−8.59
BH-17	1.25	Cap dolostone	−0.46	0.07	−0.23	0.07	−5.47	−10.75
BH-13	0.70	Cap dolostone	0.1	0.07	0.09	0.07	−5.47	−6.94
BH-10	0.40	Cap dolostone	0.01	0.07	0.02	0.07	−3.63	−8.20
BH-9	0.20	Cap dolostone	0.16	0.07	0.11	0.07	n.d.	n.d.
BH-8	0.10	Cap dolostone	0.42	0.07	0.27	0.07	−3.52	−6.81
BH-7	0	Siltstone	−0.36	0.07	−0.19	0.07
BH-5	−0.10	Pebbly siltstone	0.29	0.07	0.19	0.07
BH-4	−0.25	Pebbly siltstone	0.25	0.07	0.13	0.07
BH-3	−0.30	Pebbly siltstone	0.32	0.07	0.17	0.07
BH-2	−0.34	Pebbly siltstone	0.41	0.07	0.21	0.07
BH-1	−0.39	Pebbly siltstone	0.4	0.07	0.26	0.07
San Carlos olivine		Standard	−0.25	0.07	−0.11	0.05
Replication			−0.21	0.07	−0.12	0.06
Hawaii seawater		Standard	−0.81	0.06	−0.43	0.05
Duplicate			−0.85	0.07	−0.4	0.07
Replication			−0.78	0.08	−0.38	0.07

2SD, 2 times SD; n.d., not detected due to being below the detection limit.

Discussion

Stratigraphic fluctuation in δ²⁶Mg observed here cannot be attributed to the lithological variation alone, because δ²⁶Mg is not dependent on lithologies, as shown in Fig. 2. For example, siltstones and sandstones in units III and IV have exclusively high δ²⁶Mg values, whereas these same lithologies in units II and V have significantly lower values (Fig. 2). Nor could δ²⁶Mg variations be attributed to changes in sediment sources (e.g., fluctuations in hydrodynamic sorting of sediments and modification of the fluvial system), because consistently high δ²⁶Mg values occur in units III and IV regardless of lithological variations. This inference is further supported by the poor correlations of δ²⁶Mg with K/Ti, Mg/Ti, and Fe/Ti (Fig. S3), suggesting that variations in δ²⁶Mg are largely independent of lithological compositions. Furthermore, postdepositional alterations of δ²⁶Mg can also be ruled out on the basis of a dearth of characteristic metamorphic minerals and the poor correlation between δ²⁶Mg and K metasomatism indicator (e.g., K/Ti molar ratio) (Fig. S3A). These observations agree with previous studies that found limited Mg isotope fractionations in siliciclastic sedimentary rocks during diagenesis and low-grade metamorphism (17, 29). A possible influence of modern outcrop weathering can also be excluded because both Bahuang outcrop and the 14ZK core show consistent δ²⁶Mg profiles. Finally, the observed δ²⁶Mg patterns are not artifacts associated with incomplete carbonate leaching, because there is no significant correlation between Ca/Ti and δ²⁶Mg in leached residues (Fig. S3C).

Fig. S3.

Cross plots of the molar ratios of (A) K/Ti, (B) Mg/Ti, (C) Ca/Ti, and (D) Fe/Ti vs. δ²⁶Mg values of fine-grained siliciclastic components in Nantuo (NT) and Doushantuo (DST) formations.

Both X-ray diffraction (XRD) results and major elemental data reveal that siliciclastic components dispersed in the cap carbonate are compositionally similar to the matrix of the underlying siliciclastic rocks of the Nantuo Formation (Fig. S4 and Table S2), indicating a common and predominantly terrestrial source for the fine-grained siliciclastic components in both the Nantuo Formation and the overlying Doushantuo cap carbonate (30). As such, fine-grained siliciclastic extractions from the Nantuo Formation and the Doushantuo cap dolostone carry information about chemical weathering on continents and hence are sensitive records of chemical weathering (31). Despite their similar mineralogical compositions, siliciclastic extractions from the Nantuo and Doushantuo formations have different Mg isotopic compositions, probably because δ²⁶Mg is a more sensitive indicator of extremely intense chemical weathering when mineralogical and elemental indices such as CIA become saturated (Fig. 1B, ref. 20). Thus, we suggest that the variation in δ²⁶Mg observed here is best interpreted as fluctuation in the intensity of chemical weathering during the meltdown of Nantuo/Marinoan glaciation.

Fig. S4.

XRD chromatograms of (Upper) the representative Nantuo Formation sample (14ZK-83) and (Lower) the representative Doushantuo cap carbonate (14ZK-C5). The Nantuo samples are dominated by feldspar (albite), quartz, smectite, illite, and chlorite, and Doushantuo cap carbonate is dominated by dolomite, but also composed of quartz, chlorite, illite, and albite.

Table S2.

Major element concentrations (ppm, calculated on the basis of postleaching residues), and K/Ti, Mg/Ti, Ca/Ti, and Fe/Ti molar ratios of fine-grained siliciclastic components in the Nantuo and Doushantuo formations

Sample	Depth, m	Al	Ca	Fe	K	Mg	Mn	Ti	Na	P	K/Ti	Mg/Ti	Ca/Ti	Fe/Ti
14ZK core
14ZK-C9	2.00	29,631.3	70.2	35,680.2	59,744.0	778.5	31.1	10,394.6	n.d.	70.9	7.07	0.15	0.01	2.94
14ZK-C6	0.20	46,358.9	281.5	47,066.2	86,668.4	827.1	364.4	12,814.6	n.d.	191.8	8.32	0.13	0.03	3.15
14ZK-C5	0	21,481.6	603.4	11,699.4	74,082.8	1,436.8	83.4	11,414.4	n.d.	100.9	7.99	0.25	0.06	0.88
14ZK-C4	−0.50	59,149.7	536.7	39,423.7	53,968.9	4,839.0	192.1	7,855.9	n.d.	110.2	8.46	1.23	0.08	4.30
14ZK-C3	−1.00	135,216.2	2,526.6	27,534.1	75,330.0	16,214.0	261.8	9,950.7	n.d.	159.3	9.32	3.26	0.30	2.37
14ZK-C2	−1.48	50,605.2	364.9	33,903.7	65,726.3	2,750.5	222.6	10,442.8	n.d.	128.0	7.75	0.53	0.04	2.78
14ZK-C1	−1.78	93,097.3	601.4	40,955.3	59,531.4	8,626.3	221.2	8,296.5	n.d.	132.7	8.83	2.08	0.09	4.23
14ZK-1	−2.08	151,957.9	7,774.6	69,281.6	60,820.6	24,991.4	830.7	8,325.3	14,963.3	954.8	8.99	6.00	1.12	7.13
14ZK-2	−2.88	58,588.9	10,259.7	29,964.0	84,482.0	8,760.7	308.2	8,880.1	20,791.2	401.1	11.71	1.97	1.39	2.89
14ZK-3	−3.68	142,405.8	8,827.0	68,535.8	56,389.1	24,667.2	963.5	7,092.0	14,238.7	873.7	9.79	6.96	1.49	8.28
14ZK-4	−4.48	85,209.3	3,813.9	39,092.0	33,724.7	13,454.3	509.4	4,566.6	8,600.8	519.8	9.09	5.89	1.00	7.34
14ZK-5	−5.28	78,450.4	3,230.5	35,706.2	30,554.9	12,205.8	380.5	4,046.9	7,596.0	468.1	9.29	6.03	0.96	7.56
14ZK-6	−6.08	20,762.1	1,006.0	9,633.5	18,495.2	2,784.4	88.2	1,358.1	4,335.1	147.9	16.76	4.10	0.89	6.08
14ZK-7	−6.88	100,133.1	3,380.7	41,884.7	38,378.2	14,923.7	423.7	4,886.4	9,721.6	582.8	9.67	6.11	0.83	7.35
14ZK-8	−7.68	129,554.4	7,243.4	62,690.0	57,070.1	21,511.2	834.3	7,730.2	14,713.9	861.8	9.09	5.57	1.12	6.95
14ZK-9	−8.48	106,036.5	4,269.9	51,909.6	42,657.5	16,858.0	534.7	5,802.7	11,289.5	718.3	9.05	5.81	0.88	7.67
14ZK-10	−9.28	66,055.8	2,832.5	31,173.6	26,428.6	9,960.6	332.9	3,225.5	7,330.7	412.1	10.08	6.18	1.05	8.28
14ZK-11	−10.08	62,682.7	13,289.1	32,194.5	22,941.8	13,123.6	1,453.6	3,306.4	7,332.7	381.8	8.54	7.94	4.82	8.35
14ZK-15	−13.68	96,922.5	11,208.8	46,886.3	35,312.7	16,787.3	1,285.3	5,488.2	10,912.7	553.9	7.92	6.12	2.45	7.32
14ZK-19	−18.10	15,959.6	5,103.0	12,580.7	58,343.5	2,561.0	191.9	7,247.1	16,083.2	197.7	9.91	0.71	0.84	1.49
14ZK-22	−21.50	79,887.5	2,496.4	25,948.2	27,575.0	9,407.1	400.0	3,797.3	11,821.4	410.7	8.94	4.95	0.79	5.86
14ZK-26	−25.20	9,186.3	585.6	2,991.1	3,409.6	796.6	58.9	517.1	1,936.3	61.0	8.11	3.08	1.36	4.96
14ZK-29	−27.60	20,458.5	13,872.5	11,851.0	41,974.2	4,375.4	653.3	6,071.6	15,578.8	157.6	8.51	1.44	2.74	1.67
14ZK-31	−29.70	9,959.1	11,440.4	5,971.3	39,586.0	1,577.2	336.3	3,707.6	11,143.9	163.7	13.14	0.85	3.70	1.38
14ZK-38	−38.00	67,901.1	4,416.2	29,622.4	24,980.1	9,508.4	429.4	3,147.2	9,915.0	395.7	9.77	6.04	1.68	8.07
14ZK-42	−41.00	103,489.8	4,429.5	43,549.4	37,872.1	15,802.3	731.8	4,622.1	13,409.2	640.3	10.08	6.84	1.15	8.08
14ZK-45	−44.10	143,772.8	6,276.6	68,812.2	51,266.5	22,096.4	940.4	6,547.0	21,706.9	870.6	9.64	6.75	1.15	9.01
14ZK-48	−47.10	33,440.0	2,677.3	22,978.9	41,607.0	4,063.8	234.3	3,940.0	13,261.0	294.7	13.00	2.06	0.82	5.00
14ZK-50	−49.00	72,804.2	2,748.5	34,691.8	30,357.4	11,865.3	411.0	3,978.6	11,431.9	421.5	9.39	5.96	0.83	7.47
14ZK-56	−53.50	114,170.7	3,605.8	46,942.3	37,151.4	16,772.8	609.4	4,455.5	17,516.8	705.5	10.26	7.53	0.97	9.03
14ZK-60	−57.70	112,734.7	3,608.0	55,159.3	38,709.1	19,000.0	683.4	5,213.5	18,500.5	672.1	9.14	7.29	0.83	9.07
14ZK-66	−63.20	26,241.3	8,895.2	12,545.2	54,631.0	3,600.0	315.1	5,197.6	16,781.0	277.0	12.94	1.39	2.05	2.07
14ZK-71	−66.8	53,248.2	2,417.9	27,806.5	33,111.0	7,578.4	290.0	2,947.3	14,196.3	332.7	13.83	5.14	0.98	8.09
14ZK-77	−72.70	105,353.8	3,645.2	49,932.4	35,680.0	14,785.2	741.0	5,501.9	17,507.1	669.5	7.98	5.37	0.80	7.78
14ZK-80	−74.80	73,054.6	6,028.6	35,681.7	28,149.9	10,232.8	954.4	3,990.0	n.d.	500.3	8.68	5.13	1.81	7.67
14ZK-83	−78.50	106,835.7	10,151.9	50,237.2	43,145.0	14,665.1	1,347.3	6,179.8	22,441.9	783.7	8.59	4.75	1.97	6.97
14ZK-89	−84.50	74,002.1	4,796.5	31,724.2	23,501.0	10,463.8	709.7	3,232.4	13,985.5	455.6	8.95	6.47	1.78	8.41
Bahuang section
BH-20	1.85	97,330.8	6,704.7	23,936.4	50,640.1	14,873.9	641.2	4,233.8	n.d.	219.8	14.72	7.03	1.90	4.85
BH-19	1.65	123,175.5	8,103.9	19,026.1	60,263.2	19,609.7	157.3	7,098.1	550.1	260.3	10.45	5.53	1.37	2.30
BH-18	1.55	96,172.6	4,963.0	17,411.0	49,483.6	14,463.0	471.2	4,695.9	n.d.	247.9	12.97	6.16	1.27	3.18
BH-17	1.25	135,478.3	2,709.2	30,789.9	57,010.9	20,124.1	155.8	5,891.3	396.7	397.6	11.91	6.83	0.55	4.48
BH-16	1.15	113,184.3	9,142.9	30,332.5	54,727.5	19,220.5	187.8	4,854.5	350.1	501.8	13.88	7.92	2.26	5.36
BH-13	0.70	103,181.0	603.4	30,793.1	48,217.7	11,996.8	180.0	5,367.5	475.2	366.4	11.06	4.47	0.13	4.92
BH-10	0.40	110,522.8	2,076.9	49,172.4	67,387.5	14,416.0	1,282.1	5,923.1	883.2	710.8	14.00	4.87	0.42	7.12
BH-9	0.20	136,813.9	1,097.2	83,805.5	67,403.5	17,950.2	976.3	6,879.8	537.5	931.9	12.06	5.22	0.19	10.44
BH-8	0.10	74,787.3	496.0	107,147.9	37,329.9	7,988.1	760.0	4,208.8	395.9	344.1	10.92	3.80	0.14	21.82
BH-7	0	102,125.3	4,161.1	23,349.8	55,638.8	12,243.5	283.6	7,319.4	2,521.7	251.6	9.36	3.35	0.68	2.73
BH-6	−0.05	131,346.6	2,189.4	39,281.7	70,363.8	15,416.3	273.9	9,698.0	2,301.3	313.0	8.93	3.18	0.27	3.47
BH-5	−0.10	105,914.4	1,444.6	30,177.6	51,277.1	10,507.6	137.3	6,156.2	n.d.	584.4	10.25	3.41	0.28	4.20
BH-4	−0.25	125,635.3	680.9	48,767.5	76,472.6	10,311.6	790.3	11,007.6	9,001.5	670.2	8.55	1.87	0.07	3.80
BH-3	−0.30	128,937.8	651.5	49,482.4	66,800.8	9740.7	601.7	8,698.1	8,808.1	558.1	9.45	2.24	0.09	4.88
BH-2	−0.34	117,980.8	417.1	29,488.0	53,408.7	11,515.6	712.7	5,227.2	6,355.8	271.6	12.58	4.41	0.10	4.84
BH-1	−0.39	105,464.7	1,007.4	28,037.2	55,550.2	9,550.2	113.4	6,524.2	n.d.	414.5	10.48	2.93	0.19	3.68

Here, n.d. is short for not detected.

Intensity of Chemical Weathering During and Following the Marinoan Glaciation.

Although there are second-order variations in the δ²⁶Mg profiles, the most prominent feature of the δ²⁶Mg is a strong positive excursion in units III and IV of the Nantuo Formation. Importantly, because δ²⁶Mg is a sensitive indicator of the intensity of chemical weathering (Fig. 1B), the positive excursion with extremely high δ²⁶Mg values in units III and IV of the upper Nantuo Formation suggests an episode of exceedingly strong chemical weathering during the termination of the Marinoan glaciation. Below this excursion, δ²⁶Mg values of the massive diamictite in unit I and the pebbly sandstone in unit II, which represent the proximal and distal glaciomarine deposits, respectively, are close to the average UCC value (32), suggesting low intensity of chemical weathering during the Marinoan ice age. This inference is consistent with low compositional and textural maturity of glacial deposits in units I and II, reflecting predominantly physical weathering but negligible or weak chemical weathering (33).

In comparison, high δ²⁶Mg values in units III and IV indicate the sediments went through intense chemical weathering before deposition. Such intense chemical weathering is unlikely to have persisted during the entire Marinoan glaciation that might have lasted tens of million years (34), as this would have generated diamictite sediments with uniformly high δ²⁶Mg values. Additionally, the persistence of intense chemical weathering during the entire Marinoan glaciation is inconsistent with the extremely low temperature and glacier coverage of continents during glaciations (35). Nor could the positive δ²⁶Mg excursion of +0.90 result from recycling of preexisting sediments/sedimentary rocks, because δ²⁶Mg values of siliciclastic rocks of all other ages statistically range from –0.60 to +0.60 (95% confidence interval, Fig. 1A). Therefore, the positive δ²⁶Mg excursion in units III and IV can only be interpreted as evidence for an episode of extremely strong chemical weathering on continents toward the end of the Marinoan glaciation, because glacial marine deposition is expected to occur during the waning stage of glaciation (27). The exact temporal relationship between this episode of intense chemical weathering and the final exit from the Marinoan glaciation, however, is uncertain due to the absence of high-precision radiometric age constraints.

The subsequent decline in δ²⁶Mg in the topmost part of the Nantuo Formation (units V and VI) appears to reflect a decrease in the intensity of chemical weathering. Low intensity of chemical weathering continued into the cap carbonate deposition, as evidenced by low δ²⁶Mg of the siliciclastic components in the Doushantuo cap carbonate (Fig. 2).

Chemical weathering is partially controlled by surface environmental factors, such as temperature and hydrologic cycle, which are ultimately related to atmospheric pCO₂ levels (36). The episode of intense chemical weathering, as evidenced by high δ²⁶Mg in units III and IV, requires a meltdown of continental ice sheets and exposure of land surface with a high level of atmospheric pCO₂ (8, 10). The subsequent decline in chemical weathering in units V and VI and the cap carbonate may result from a drawdown in atmospheric pCO₂ levels after this episode of intense chemical weathering. As a possible outcome of rapid CO₂ drawdown, renewed glacier formation may have ensued, leading to the deposition of distal glacial marine sediments in unit VI in the topmost part of the Nantuo Formation and the occasional presence of dropstones in the cap carbonate. Therefore, the Mg isotopic data demonstrate that the climate during deglaciation was more dynamic than previously thought and that cap carbonate deposition followed an episode of strong chemical weathering and the major phase of deglaciation, both of which were driven by a high level of atmospheric pCO₂ toward the end of the Marinoan glaciation.

Implications for Cap Carbonate Deposition.

The snowball Earth hypothesis predicates that cap carbonate precipitation is associated with intense chemical weathering during the termination of global glaciation (8). However, our Mg isotopic data indicate that cap carbonate deposition postdates the climax of chemical weathering and deglaciation. Instead, chemical weathering intensity returned to background levels during cap carbonate deposition.

The delayed response of the sedimentary system to deglaciation and chemical weathering may be attributed to the low seawater pH when atmospheric pCO₂ was high—perhaps as high as 210 PAL to 550 PAL (37, 38)—at the beginning of the deglaciation. Seawater pH at equilibrium with an atmospheric pCO₂ level of 550 PAL would be ≤6.5 (39), which is significantly lower than the seawater pH of 8.3 to 8.7 during cap carbonate deposition as estimated from boron isotopes (41) (Fig. 3A). Thus, cap carbonate precipitation seems to require a substantial rise in seawater pH as well as a sharp decline in atmospheric pCO₂ levels, so as to raise the carbonate saturation state (Ω) in seawaters (Fig. 3). Assuming cap carbonate deposition at pH of 8.3 (40, 41) and seawater dissolved inorganic carbon (DIC) of 10 mM (SI Text), the equilibrated atmospheric pCO₂ level would be ∼5.2 PAL (Fig. 3A) and Ω would be 4.5 (Fig. 3B), implying that cap carbonate deposition did not occur until more than 97 to 99% of atmospheric CO₂ (210 PAL to 550 PAL) had been consumed by chemical weathering. Similarly, if we accept an atmospheric pCO₂ level of 33 PAL on the basis of triple-oxygen isotopes of cap carbonates (42), cap carbonate deposition began after 85 to 95% CO₂ consumption, at seawater pH of 7.5, and carbonate Ω of 1.54 (Fig. 3). The gap between deglaciation and cap carbonate precipitation might be on the order of 10⁵ y, inferred from a previous modeling study that estimated the length of time needed for chemical weathering to reduce atmospheric pCO₂ levels from 270 PAL (assumed pCO₂ levels at end of glaciation) to 1 PAL (background pCO₂ level) at 40 °C (43).

Fig. 3.

Modeling results showing the relationships between the partial pressure of atmospheric CO₂ (pCO₂), (A) seawater pH, and (B) carbonate saturation state (Ω) in an equilibrated atmosphere−ocean system, assuming a sea surface temperature of 10 °C and salinity of 35‰. Contour lines indicate different concentrations of total DIC in seawater. Orange shade refers to the estimated seawater pH and Ω at the onset of Snowball deglaciation (8, 37, 38). Blue line and green shade represent the range of seawater pH and Ω during cap carbonate deposition based on triple-oxygen isotopes (42) and boron isotopes (40, 41), respectively.

A drop in atmospheric pCO₂ levels is consistent with persistently low δ²⁶Mg values and thus low intensity of chemical weathering in the topmost part of the Nantuo Formation and the Doushantuo cap carbonate (Fig. 2). We suggest that, as chemical weathering consumes atmospheric CO₂, seawater alkalinity would be elevated, leading to the progressive increase of carbonate saturation state in seawater (Fig. 3B) and eventually cap carbonate precipitation. In this scenario, the cap carbonate precipitated from supersaturated seawater with DIC mainly derived from atmospheric CO₂, explaining the negative carbon isotope signals (Fig. 2) and widespread isopachous cements and aragonite fans in the cap carbonate (7, 44, 45).

In summary, Mg isotopic data indicate that the deposition of the Doushantuo cap carbonate was preceded by an episode of intense chemical weathering during the termination of the 635 Ma Marinoan glaciation, and cap carbonate precipitation occurred when atmospheric pCO₂ and chemical weathering intensity returned to lower levels. In this scenario, cap carbonate deposition was a delayed response to an episode of intense chemical weathering, which ultimately supplied Ca²⁺ and bicarbonate to support cap carbonate precipitation. This episode of intense chemical weathering not only led to the global precipitation of cap carbonate after the Marinoan glaciation but may have also brought abundant nutrients to the ocean, which in turn might have triggered primary productivity, ocean oxygenation (3), and, eventually, the diversification of complex eukaryotes in the early Ediacaran Period (46).

Methods

Sample Preparation.

To minimize the effect of modern weathering, drill core samples and least-weathered outcrop samples were selected for analysis. Thin sections were prepared to guide the sampling of fine-grained siliciclastic components, so as to keep away from pebbles or large-sized clasts in diamictites. Powders (300 mg to 500 mg) were obtained from different parts of the matrix using a handheld drill and were mixed to produce a homogenous sample. To remove calcareous contents, sample powder (∼30 mg) was first leached in 10 mL of 0.5 N acetic acid in an ultrasonic incubator for 3 h. After centrifugation, the residue was further leached in 5 mL of 1 N hydrochloric acid for 1 h to ensure complete removal of carbonate minerals. After centrifugation, leaching residues were washed in deionized water at least three times, and then dried down overnight at 60 °C.

Elemental and Mineralogical Analyses.

Postleaching residues were completely dissolved in a mixture of concentrated HF, HNO₃, and HCl. Elemental compositions were determined on a Spectro Blue Sop Inductively Coupled Plasma Optical Emission Spectrometer at Peking University. Elemental concentrations were calculated on the basis of residue mass. The external reproducibility for major and minor elements (Na, Mg, Al, K, Ca, Fe, Mn, Ti, and Sr) is better than 5%. Mineral compositions of Doushantuo cap carbonate (whole-rock powders) and Nantuo Formation (postleaching residues) were analyzed using a PANalytical X'Pert Pro MPD X-ray Diffractometer. Mineral phases were identified by peak positions, and mineralogical abundances were quantitatively determined by peak areas.

Carbon Isotope Analysis.

About 0.2 mg of carbonate sample powders was weighted to react with 1 mL of 100% phosphoric acid at 72 °C to liberate CO₂ from carbonate. CO₂ gas was introduced to MAT253 mass spectrometry at Louisiana State University for δ¹³C and δ¹⁸O analysis. Carbon and oxygen isotope values are reported in the delta notation as permil (‰) deviation relative to the Vienna-Pee Dee Formation belemnite standard. The analytical precision is ±0.1‰ (1σ).

Magnesium Isotope Analysis.

About 5 mg of postleaching residue was dissolved in a mixture of Optima-grade HF, HNO₃, and HCl. Mg was purified by cation exchange chromatography (loaded with Bio-Rad AGW50-X8 resin, 200 to 400 mesh). Two standards (Hawaii seawater and San Carlos olivine) were processed with samples for column chemistry. The purified Mg solutions with Na/Mg, Ca/Mg, Al/Mg, K/Mg, Fe/Mg < 0.05 were analyzed by the standard-sample bracketing method on a Nu plasma Multi-Collector Inductively Coupled Plasma Mass Spectrometer at University of Washington, Seattle. Three Mg isotopes (24–26) were measured simultaneously in separate Faraday cups (H5, Ax, and L4) at low-resolution mode. The procedure blank for ²⁴Mg (<10⁻⁴ V) was negligible relative to the sample signals (2 V to 4 V). Mg isotope values are reported in δ-notation as permil deviation relative to the standard DSM3: δ²⁶Mg = [(δ^xMg/²⁴Mg)_sample/(δ^xMg/²⁴Mg)_DSM3 – 1] × 1,000, where x refers to mass 25 or 26. Multiple analyses of the San Carlos olivine and Hawaii seawater standards during the course of this study yielded δ²⁶Mg values ranging from –0.25 to –0.21 and from –0.87 to –0.78, respectively, which are consistent with published values. All samples and standards analyzed in this study fall in a single mass-dependent fractionation line with a slope of 0.519 (Fig. S5).

Fig. S5.

Three-isotope plot illustrating Mg isotopic composition of samples from the Nantuo (NT) Formation and Doushantuo (DST) cap carbonate, and in-house standards (San Carlos olivine, Hawaii seawater). All data fall along the terrestrial equilibrium mass-dependent fractionation curve (solid line) with a slope of 0.5193.

SI Text

Geological Setting and Stratigraphic Context.

South China Block was an isolated craton at low latitudes during the late Neoproterozoic (49). It consists of the Yangtze Platform in the northwest (in the present orientation) and the Cathaysia Block in the southeast (Fig. S1A). Convergence between the Yangtze Platform and Cathaysia occurred in mid-Neoproterozoic (50). In the Yangtze Platform, Neoproterozoic succession represents three tecto phases (23, 50): (i) the pre-Cryogenian (>720 Ma) early rifting phase that is characterized by siliciclastic deposition and active volcanic activities (the Liantuo Formation/Banxi Group/Xiajiang Group), (ii) the late rifting phase that is represented by the Cryogenian deposition with declined magmatic activities (the Chang’an/Dongshanfeng/Tiesiao formations, Fulu/Datangpo/Xiangmeng formations, and Nantuo Formation), and (iii) the thermo-subsidence phase that is identified by passive margin deposition of the Ediacaran Doushantuo and Dengying/Liuchapo formations. The Nantuo Formation is dated between 654 ± 3.8 Ma and 635.2 ± 0.6 Ma (24, 25), and is suggested to correlate with the terminal Cryogenian Marinoan glaciation (34). The Nantuo Formation thickens from the shelf (1 m to 150 m) to the basin environments (up to 2,000 m) (23, 50), whereas the overlying Doushantuo Formation thins from 80 m to 200 m in the shelf to <10 m to 50 m in the slope and basinal settings (23, 51). The cap carbonate at the basal Doushantuo Formation ranges from 3 m to 6 m thick, and mainly consists of dolomicrite or microcrystalline dolomite (also called cap dolostone) (23), with the development of sheet cracks in the basal 1-m unit.

Section Description.

In this study, samples of the upper Nantuo Formation and basal Doushantuo cap carbonate were collected from a drill core (14ZK core) at Daotuo village in Songtao county (28°7’54.30”N; 108°53’32.80”E), and from an outcrop section in Bahuang town in Tongren City (27°44’10.19”N;109°0’52.90”E), northeastern Guizhou Province (Fig. S1A).

The Nantuo Formation in the recovered 14ZK core is about 225 m thick, and conformably overlies black shales of the Datangpo Formation (Fig. 2). In this study, we focus on the upper part of the Nantuo Formation [812 m to 899 m below the surface (mbs)] (Fig. 2). The sampled interval can be subdivided into six lithostratigraphic members (Fig. 2) in ascending order: 80-m-thick (but only the top 35 m was sampled) massive diamictite (unit I, 879 mbs to 899 mbs), 15-m-thick pebbly sandstone/siltstone with the development of lonestones (unit II, 864 mbs to 879 mbs), 28-m-thick intercalated laminated sandstone and siltstone (unit III, 836 mbs to 864 mbs), 10-m-thick pebbly sandstone/siltstone (unit IV, 828 mbs to 836 mbs), 10-m-thick laminated siltstone (unit V, 817 mbs to 828 mbs), and 3-m-thick diamictite/siltstone unit (unit VI, 815 mbs to 818 mbs). Unit VI is conformably overlain by the basal Ediacaran cap dolostone of the Doushantuo Formation (Fig. 2), which is mainly composed of microcrystalline dolomite and dolomicrite with the development of sheet cracks (Fig. S2H).

In the Bahuang section, only the upper most 20 m of the Nantuo Formation is exposed, and only the topmost 1-m interval can be sampled. The exposed Nantuo Formation is dominated by massive diamictite (Fig. S2G), and the uppermost 0.7-m interval is composed of greenish gray siltstone with sporadic occurrences of small pebbles in the base. Such a siltstone layer can be correlated with the upper unit VI in the 14ZK core. The Nantuo Formation is conformably overlain by a ca. 3-m-thick cap carbonate of the basal Doushantuo Formation (Fig. S2F). The lower part of the Doushantuo cap carbonate consists of a 1-m-thick massive dolostone with the development of sheet cracks, followed by the deposition of thin- to medium-bedded dolostone (Fig. S2F).

Several lines of evidence, based on the petrographic, mineralogical, and geochemical studies, indicate that the studied sections have suffered limited metamorphism and diagenetic alteration. First, the sedimentary textures and rock fabrics are well preserved in both 14ZK core and the Bahuang section. All carbonate samples are composed of dolomicrite and microcrystalline dolostone without significant recrystallization (Fig. S2). Second, no metamorphic minerals have been identified in thin section observation or in XRD analysis. The siliciclastic sediments in the Nantuo Formation are dominated by feldspar, quartz, smectite, and illite (Fig. S4). Third, carbon and oxygen isotopic signatures of the Doushantuo cap carbonates (Fig. 2 and Table S1) are congruent with other well-preserved basal Ediacaran cap carbonates in South China and elsewhere in the world, arguing against significant diagenetic and metamorphic modifications.

Depositional Environments.

Regional mapping and paleogeographic studies indicate that the Nantuo and Doushantuo formations at the 14ZK core and the Bahuang section were deposited in deep-water lower-slope to basinal environments (23) (Fig. S1). Massive diamictite is interpreted as the rapid deposition during the meltdown of sediment-loaded ice shelf, representing the proximal glaciomarine facies with sediments sourced from continents and transported by glaciers (28). The pebbly sandstones and siltstone alternations with the presence of lonestone are interpreted as deposition with the influence of debris-laden icebergs in a distal glacial marine environment (52). The pebble-free sandstone and siltstone with thickness of >10 m represents normal marine deposition with sediment transportation by fluvial system without the direct influence of glaciers (27, 53). Three facies associations, the proximal glaciomarine facies, distal glaciomarine facies, and nonglacial marine facies, are thus identified in the Nantuo Formation. The upper Nantuo Formation (in the case of 14ZK core) records, in ascending order, the proximal glaciomarine deposition of massive diamictite (unit I), the distal glaciomarine deposition of lonestone bearing pebbly sandstone and siltstone alternations (unit II), normal marine deposition of intercalated sandstone and siltstone (unit III), pebbly sandstone with siltstone of distal glaciomarine facies (unit IV), normal marine deposition of siltstone with parallel bedding (unit V), and thin-layer of diamictite overlain by pebbly siltstone or siltstone, representing distal glaciomarine deposition (unit VI). The overlying cap carbonate may represent deposition in relatively deep water below storm wave base (54).

CO₂ in the Atmosphere−Ocean System.

In the atmosphere−ocean system, surface seawater is in equilibrium with atmospheric CO₂. The equilibrium of carbonate in seawater can be expressed by the following equations:

$${\text{CO}}_2\left(g\right)\stackrel{K_H}{\leftrightarrow }\hspace{0.17em}{\text{CO}}_2\left(aq\right),$$ [S1]

$${\text{CO}}_2\left(\text{aq}\right)\hspace{0.17em}\mathrm{+}{\hspace{0.17em}\text{H}}_2\text{O}\hspace{0.17em}\stackrel{K_1}{\leftrightarrow }\hspace{0.17em}{\text{H}}^{+}\hspace{0.17em}+\hspace{0.17em}{\text{HCO}}_3^{-},$$ [S2]

$${\text{HCO}}_3^{-}\hspace{0.17em}\stackrel{K_2}{\leftrightarrow }\hspace{0.17em}{\text{H}}^{+}+{\hspace{0.17em}\text{CO}}_3^{2-},$$ [S3]

where K_i is the equilibrium constant, and can be expressed as follows:

$$K_H=\left[{\text{CO}}_2\left(aq\right)\right]/p{\text{CO}}_2,$$ [S4]

$$K_1=\left[{\text{HCO}}_3^{-}\right]\left[{\text{H}}^{+}\right]/\left[{\text{CO}}_2\left(aq\right)\right],$$ [S5]

$$\hspace{0.17em}{K}_2=\left[{\text{CO}}_3^{2-}\right]\left[{\text{H}}^{+}\right]/\left[{\text{HCO}}_3^{-}\right].$$ [S6]

$K_H$ is the Henry’s constant for CO₂, and $K_1$ and $K_2$ are the first and second dissociation constants for carbonic acid. The values of $K_H$, $K_1$, and $K_2$ are temperature- and salinity-dependent, and are set to 4.39 $\times$ 10⁻² mol⋅kg⁻¹⋅atm⁻¹, 1.01 $\times$ 10⁻⁶ mol⋅kg⁻¹⋅atm⁻¹, 6.53 $\times$ 10⁻¹⁰ mol⋅kg⁻¹⋅atm⁻¹ at sea surface temperature of 10 °C and salinity of 35‰ under 1 atm surface pressure (55, 56). Rearranging Eqs. S4–S6, we arrive at

$$\left[{\text{CO}}_3^{2-}\right]=\frac{K_H{K}_1{K}_2}{{\left[H^{+}\right]}^2}\hspace{0.17em}p{\text{CO}}_2,$$ [S7]

$$\text{DIC}= p{\text{CO}}_2\left(\frac{1}{K_H}+\frac{K_1}{K_H\left[{\text{H}}^{+}\right]}+\frac{K_1{K}_2}{K_H{\left[{\text{H}}^{+}\right]}^2}\right),$$ [S8]

where DIC is the DIC in seawater, including all inorganic carbon components. Eqs. S7 and S8 indicates that carbonate concentration ($\left[{\text{CO}}_3^{2-}\right]$) and DIC content of seawater are related to atmospheric pCO₂ levels and seawater pH (pH = –log[H⁺]). In modern oceans, seawater DIC content is 2.3 mmol/kg (57). We consider that seawater DIC might be higher in postglacial oceans (58), because postglacial chemical weathering scavenges a mass of atmospheric CO₂, leading to an elevated DIC content in seawater (11). Assuming all atmospheric CO₂ is sequestered in the ocean as DIC, the maximum estimates of seawater DIC concentrations would be 10 mM, 7 mM, and 5 mM for atmospheric pCO₂ of 550 PAL (present atmospheric CO₂ level, i.e., 10^–3.5 bar) (38), 350 PAL (8), and 210 PAL (37), respectively. Calculation results indicate that the equilibrium seawater pH is less than 6.24 at atmospheric pCO₂ greater than 210 PAL (Fig. 3). The seawater pH during cap carbonate deposition is estimated to be 8.3 to 8.7 based on boron isotope (41), or 7.5 based on an atmospheric pCO₂ of 33 PAL according to triple-oxygen isotope data (37) (Fig. 3).

CaCO₃ precipitation is determined by the saturation state (Ω), which is related to the activity of calcium cation (Ca²⁺) and carbonate (${\text{CO}}_3^{2-}$). The saturation state Ω of a solution is defined by

$$\mathrm{\Omega }=\frac{{\gamma }_{{\text{Ca}}^{2+}}{\gamma }_{{\text{CO}}_3^{2-}}\left[{\text{Ca}}^{2+}\right]\left[{\text{CO}}_3^{2-}\right]}{K_{\text{sp}}},$$ [S9]

where ${\gamma }_{{\text{Ca}}^{2+}}{\gamma }_{{\text{CO}}_3^{2-}}$ is the product of the total ion activity coefficients of calcium and carbonate, which is about 0.0066 at salinity of 35‰ (59), and K_sp is a solubility constant of calcite. When Eq. S9 is combined with Eq. S7, we arrive at

$$\mathrm{\Omega }=\frac{{\gamma }_{{\text{Ca}}^{2+}}{\gamma }_{{\text{CO}}_3^{2-}}\left[{\text{Ca}}^{2+}\right]K_H{K}_1{K}_2}{K_{\text{sp}}{\left[{\text{H}}^{+}\right]}^2}p{\text{CO}}_2.$$ [S10]

Eq. S10 indicates that Ω is positively correlated with atmospheric pCO₂ level and pH, which ultimately link to DIC. The modeling results reveal that seawater is undersaturated (i.e., Ω < 1) when atmospheric pCO₂ is higher than 60 PAL, even if we assume that seawater Ca²⁺ concentration after global glaciation was 100 mM, which is 10-fold the modern seawater (10 mM) (Fig. 3B).