Early Triassic seawater sulfate drawdown

Early Triassic seawater sulfate drawdown

Available online at www.sciencedirect.com ScienceDirect Geochimica et Cosmochimica Acta 128 (2014) 95–113 www.elsevier.com/locate/gca Early Triassic...

4MB Sizes 10 Downloads 34 Views

Available online at www.sciencedirect.com

ScienceDirect Geochimica et Cosmochimica Acta 128 (2014) 95–113 www.elsevier.com/locate/gca

Early Triassic seawater sulfate drawdown Huyue Song a,⇑, Jinnan Tong a,⇑, Thomas J. Algeo a,b,⇑, Haijun Song a,c, Haiou Qiu d, Yuanyuan Zhu e, Li Tian a, Steven Bates f, Timothy W. Lyons f, Genming Luo a, Lee R. Kump g a

State Key Laboratory of Biogeology and Environmental Geology, China University of Geosciences, Wuhan 430074, China b Department of Geology, University of Cincinnati, Cincinnati, OH 45221-0013, USA c State Key Laboratory of Palaeobiology and Stratigraphy, Nanjing Institute of Geology and Palaeontology, CAS, Nanjing 210008, China d Faculty of Material Science and Chemistry, China University of Geosciences, Wuhan 430074, China e Wuhan Institute of Geology and Mineral Resources, Wuhan 430205, China f Department of Earth Sciences, University of California, Riverside, CA 92521, USA g Department of Geosciences and Earth System Science Center, Pennsylvania State University, University Park, PA 16802, USA Received 19 June 2013; accepted in revised form 4 December 2013; available online 15 December 2013

Abstract The marine sulfur cycle is intimately linked to global carbon fluxes, atmospheric composition, and climate, yet relatively little is known about how it responded to the end-Permian biocrisis, the largest mass extinction of the Phanerozoic. Here, we analyze carbonate-associated-sulfate (CAS) from three Permo–Triassic sections in South China in order to document the behavior of the C–S cycle and its relationship to marine environmental changes during the mass extinction and its aftermath. We find that d34SCAS varied from +9& to +44& at rates up to 100& Myr1 during the Griesbachian–Smithian substages of the Early Triassic. We model the marine sulfur cycle to demonstrate that such rapid variation required drawdown of seawater sulfate concentrations to 64 mM and a reduction in its residence time to 6200 kyr. This shorter residence time resulted in positive covariation with d13Ccarb due to strong coupling of the organic carbon and pyrite burial fluxes. Carbon and sulfur isotopic shifts were associated with contemporaneous changes in climate, marine productivity, and microbial sulfate reduction rates, with negative shifts in d13Ccarb and d34SCAS linked to warming, decreased productivity, and reduced sulfate reduction. Sustained cooling during the Spathian re-invigorated oceanic overturning circulation, reduced marine anoxia, and limited pyrite burial. As seawater sulfate built to higher concentrations during the Spathian, the coupling of the marine C and S cycles came to an end and a general amelioration of marine environmental conditions set the stage for a recovery of invertebrate faunas. Variation in seawater sulfate during the Early Triassic was probably controlled by climate change, possibly linked to major eruptive phases of the Siberian Traps. Ó 2013 Elsevier Ltd. All rights reserved.

1. INTRODUCTION 1.1. Oceanic perturbations during the Permian–Triassic transition

⇑ Corresponding authors. Tel.: +86 027 67883450 (H. Song, J.

Tong). Tel.: +1 513 556 4195 (T.J. Algeo). E-mail addresses: [email protected] (H. Song), [email protected] (J. Tong), [email protected] (T.J. Algeo). 0016-7037/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.gca.2013.12.009

The end-Permian mass extinction (EPME) at 252 Ma was the largest biocrisis of the Phanerozoic, with a 90% species-level extinction rate among marine invertebrates (Erwin, 1994; Alroy et al., 2008). It was followed by an interval of disturbed marine environmental conditions


H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113

characterized by widespread anoxia, large carbon (C) isotope excursions, low-diversity ecosystems, and rapid evolutionary turnover that lasted for at least 2 million years into the Early Triassic (Wignall and Twitchett, 2002; Payne et al., 2004; Bottjer et al., 2008; Brayard et al., 2009). The marine sulfur (S) cycle was also severely perturbed as indicated by large S-isotopic (d34S) fluctuations (Newton et al., 2004; Marenco et al., 2008a; Kaiho et al., 2012) and possible substantial drawdown of the seawater sulfate reservoir (Luo et al., 2010). All of these perturbations have been hypothesized to have resulted from massive eruptions of the Siberian Traps magmatic province (Wignall, 2007; Korte et al., 2010; Algeo et al., 2011). Recent work has begun to uncover the relationships among these environmental and faunal changes, which may be causally connected through extreme climatic warming (Joachimski et al., 2012; Romano et al., 2012; Sun et al., 2012), intensified water-column stratification (Horacek et al., 2007; Song et al., 2013), and expansion of oceanic oxygen-minimum zones (OMZs; Algeo et al., 2010, 2011; Winguth and Winguth, 2012). Although oxygen deficiency was probably a major biotic stressor in the thermocline and deep ocean, shallow-marine environments were generally well-oxygenated throughout the crisis except for brief anoxic episodes (Algeo et al., 2007, 2008) likely linked to upward chemocline fluctuations (Kump et al., 2005; Riccardi et al., 2006). Many shallow-marine invertebrates are sensitive to even low concentrations of hydrogen sulfide (H2S) in seawater (Wang and Chapman, 1999), and H2S toxicity thus may have been an important kill mechanism in such settings (Ward, 2006). Analysis of changes in the marine S cycle and its linkage to global C cycle perturbations has the potential to provide new insights into the nature of oceanic environmental changes during the Permian–Triassic transition and their relationship to the marine biocrisis. 1.2. Permian–Triassic seawater sulfate d34S studies Existing S isotopic studies provide only a general understanding of changes in the marine S cycle during the Permian–Triassic crisis interval. Early studies of marine evaporites yielded evidence of a large positive excursion in seawater sulfate d34S in the Early Triassic (Claypool et al., 1980), a pattern confirmed by more recent Phanerozoic sulfate d34S records (Strauss, 1997; Kampschulte and Strauss, 2004). However, the sporadic stratigraphic occurrence of marine evaporites cannot yield a continuous seawater sulfate d34S record, as shown by studies focused on the Permian–Triassic boundary (PTB) in Italy (Cortecci et al., 1981) and Saudi Arabia (Worden et al., 1997). A high-resolution record is needed, however, given the likelihood of high-frequency variation in seawater sulfate d34S during the Early Triassic. Such variation would have been a natural consequence of drawdown of the seawater sulfate reservoir following massive deposition of Late Permian marine evaporites (Anderson et al., 1972; Tucker, 1991; Hay et al., 2006). Marine carbonates are known to contain trace amounts of sulfate (Takano, 1985; Staudt and Schoonen, 1995), and the S isotopic composition of this carbonate-associated

sulfate (CAS) can serve as a proxy for seawater sulfate d34S (Kampschulte et al., 2001; Kampschulte and Strauss, 2004). Recent studies of Permo–Triassic d34 SCAS have focused mainly on a narrow stratigraphic window around the EPME and PTB (Kaiho et al., 2001, 2006; Newton et al., 2004; Gorjan et al., 2007; Luo et al., 2010). These studies have reported a wide range of isotopic values (+10& to +25&) without yielding a coherent pattern of secular d34 SCAS variation (Fig. 1A). Variation in d34 SCAS during the Early Triassic has been the examined in several studies (Riccardi et al., 2006; Marenco, 2007; Marenco et al., 2008a; Kaiho et al., 2012), but small sample numbers and/or lack of detailed biostratigraphic information limited the stratigraphic resolution achieved in these studies (Fig. 1B). In addition, none of these studies has documented any relationship between d34 SCAS and other geochemical proxies (e.g., d13 Ccarb ) that might yield insights regarding operation of the marine S cycle during the crisis interval. In the present study, we analyzed d34 SCAS in three sections (Dajiang, Lower Guandao, and Upper Guandao) from the Great Bank of Guizhou (GBG), a carbonate platform in the Nanpanjiang Basin of the South China craton in the eastern Paleotethys Ocean (Fig. 2). We obtained a nearly continuous record of d34 SCAS variation from the late Changhsingian (latest Permian) through the end of the Ladinian (Middle-Late Triassic boundary) with a temporal resolution of <0.1 Myr, improving substantially on existing Permo–Triassic d34 SCAS records (Fig. 1). We also compared our d34 SCAS record with a high-resolution d13 Ccrab profile for the same study units. The results of the present study provide new insights regarding controls on (1) the P–Tr marine S cycle, (2) the extent of seawater sulfate drawdown, and (3) its relationship to ocean stratification, redox conditions, and primary productivity rates. 2. GEOLOGICAL SETTING The study area is located in the Nanpanjiang Basin of the Permo–Triassic South China craton, which was bordered on the north and east by the carbonate Yangtze Platform and on the west and south by the Paleotethys Ocean (Fig. 2A; Tong and Yin, 2002; Yin et al., 2014). The three study sections (Dajiang, Lower Guandao, and Upper Guandao) are located on the Great Bank of Guizhou, a carbonate buildup in the Nanpanjiang Basin of south-central China (Fig. 2B; Lehrmann et al., 2003, 2006). The Great Bank of Guizhou evolved from a low-relief bank rimmed with oolite shoals and gentle basin-margin slopes in the Early Triassic, to a Tubiphytes reef-rimmed platform in the Middle Triassic (Fig. 2C), and finally to high-relief erosional escarpment before the platform was drowned and overwhelmed by siliciclastic turbidite deposition at the beginning of the Late Triassic (Enos et al., 2006). Water depths were no more than a few tens of meters on top of the platform but up to several hundred meters in adjacent basins (He et al., 2005; Payne et al., 2006). Lower and Middle Triassic strata in the study area range in thickness from 1500 m in the platform interior to 600 m on its flanks (Enos et al., 2006; Lehrmann et al., 2006).

H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113


Fig. 1. d34SCAS data from earlier studies of (A) the narrow stratigraphic interval around the EPME-PTB, and (B) the entire Early Triassic. The studies in A document pronounced variation in d34SCAS around the EPME-PTB but do not reveal a coherent secular pattern; the study in B documents large changes in d34SCAS between Early Triassic substages but lacks temporal resolution. Bed numbers, conodont zonation, and vertical scale in A are for the Meishan D global stratotype section and point (Yin et al., 2001; Zhang et al., 2007). Time scale from Algeo et al. (2013). Abbreviations: H. chx. = H. changxingensis; EPME = end-Permian mass extinction; PTB = Permian–Triassic boundary.

The Dajiang section, deposited in the interior of the Great Bank of Guizhou (Fig. 2C), consists of the Upper Permian Wujiaping Formation and the Lower Triassic Daye and Anshun formations (Fig. 3A). The Wujiaping Formation comprises 60 m of skeletal packstone and grainstone, and the base of the Daye Formation consists of 16 m of thrombolitic microbialite (Lehrmann, 1999; Wang et al., 2005a, 2005b; Payne et al., 2006). The EPME is at the base of the microbialite (Lehrmann et al., 2003; Luo et al., 2010), with the PTB located 2 m higher based on the first occurrence of Hindeodus parvus, the index taxon for the PTB (Yin et al., 2001). Above the microbialite, the Daye and Anshun formations consist of thinly bedded, weakly bioturbated micritic limestone and dolomite strata containing sporadic interbeds of small gastropods and bivalves (Lehrmann et al., 2003, 2006). The two Guandao sections, deposited on the flank of the Great Bank of Guizhou (Fig. 2C), are located about 5 km northwest of the Dajiang section. The Lower Guandao section comprises the whole of the Lower Triassic, and the Upper Guandao section spans the uppermost Spathian through lowermost Carnian. At Lower Guandao, the Induan Luolou Formation consists of a lowermost Griesbachian shale overlain by thin-bedded lime mudstone, packstone, and grainstone containing few fossils but abundant peloids

and clasts (Fig. 3B). The upper part of Luolou Formation consists mainly of thin-bedded limestone, interspersed with intervals of shale and dolostone. The conspicuous increase of fossils in the upper part of the formation reflects the recovery of marine fauna during the Spathian. The Upper Guandao section is located about 200 m north of the Lower Guandao section. The base of the section contains several volcanic ash layers, overlain by 400 m of medium- to thick-bedded lime packstone and grainstone of Middle Triassic (Anisian–Ladinian) age (Fig. 3C). The choice of study sections above offered several major advantages. First, the platform carbonates in these sections are nearly free of siliciclastic influence and represent an open-marine (i.e., unrestricted) depositional environment (Lehrmann et al., 2006; Payne et al., 2006). Second, the conodont, algal, and foraminiferal fossil components of these sections have been examined in earlier studies (Payne et al., 2004; Wang et al., 2005a,b; Chen et al., 2009; Song et al., 2011, 2012). Third, high-resolution carbonate carbon isotope profiles have been generated for all three sections (Krull et al., 2004; Payne et al., 2004; Tong et al., 2007). These earlier investigations provided a detailed biostratigraphic and carbon isotope framework (Fig. 3) within which the results of the present study were evaluated.


H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113

3. METHODS 3.1. Sampling protocol A total of 203 carbonate samples were collected from the Dajiang, Lower Guandao, and Upper Guandao sections for analysis of d34 SCAS and CAS concentrations ([CAS]). Samples were collected at intervals of 1–3 m throughout each section, with care taken to avoid shale and breccia beds. Weathered surfaces and obvious diagenetic phases were trimmed off prior to powdering. Samples were crushed into small pieces in the lab, and then the freshest pieces were chosen and ground in a rock mill. 3.2. Carbonate-associated sulfate analysis CAS was extracted using the method of Burdett et al. (1989) as modified by Luo et al. (2010). For each sample, 100–150 g of powder was washed in a 10% NaCl solution for 24 h. After gravitational settling, the supernatant was aspirated and the residual powder was rinsed with 18.25 MX distilled water. The residual material was then washed for 24 h in 6% NaClO solution. The supernatant was aspirated again, and the residual powder was rinsed in DDI water. The residual powder was then dissolved in 6 M HCl for 12 h. The mixture was filtered through a 0.45-lm membrane to remove any insoluble matter. A 10% BaCl2 solution was added to the filtrate in order to precipitate BaSO4, which was then dried and weighed. The BaSO4 was mixed with V2O5 and SiO2 and combusted under vacuum in the presence of copper turnings for a quantitative conversion of barite sulfur to SO2. Isotopic determinations were performed on a Finnigan MAT 251 mass spectrometer at the State Key Laboratory of Geological Processes and Mineral Resources, China University of Geosciences (Wuhan). Results are expressed in standard delta notation (d34S) as per mille deviations versus the international VCDT standard. Analytical reproducibility was generally better than ±0.3&. 3.3. Elemental, d13Ccarb, and d34Spy analyses Major-element and trace-element concentrations were determined on whole-rock samples using a wavelength-dispersive Rigaku 3040 X-ray fluorescence (XRF) spectrometer at the University of Cincinnati. Results were calibrated using both USGS and internal laboratory standards (analyzed by XRAL Incorporated using XRF and INAA). Analytical precision based on replicate analyses was better than ±2% for major and minor elements and ±5% for trace-elements, and detection limits were 1– 5 ppm for different trace elements. Total S concentrations were measured using an Eltra 2000 C–S analyzer. Data quality was monitored via analysis of the USGS standard SDO-1, which yielded an analytical precision (2r) of ±5 % for S. Most of the carbonate C-isotope data used in this study (n = 403) were generated by Payne et al. (2004). A subset of samples from the Dajiang section (n = 59) were analyzed for d13 Ccrab as part of the present study, yielding paired

Fig. 2. Paleogeography of study sections: (A) location of Nanpanjiang Basin in South China craton (modern coordinates); inset rectangle represents panel B. (B) Locations of Guandao and Dajiang study sections on Great Bank of Guizhou; colour scheme matches facies shown in panel C. (C) Stratigraphic position of Upper and Lower Guandao and Dajiang study sections on N–S cross-section of the Great Bank of Guizhou. B and C modified from Lehrmann et al. (2006). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

d34SCAS–d13Ccarb analyses for these samples. For carbonate C-isotope analysis, about 150–400 lg of powdered sample was placed in a 10 ml Na-glass vial, sealed with a butyl rubber septum, and reacted with 100% phosphoric acid at

H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113


Fig. 3. Litho/bio/chemostratigraphy of (A) Dajiang, (B) Lower Guandao, and (C) Upper Guandao. Sources: d34SCAS data (green circles, this study), d13Ccarb data (red circles, Payne et al., 2004; blue diamonds, this study); conodont zonation (Payne et al., 2004; Wang et al., 2005a,b; Tong et al., 2007). N1–N4 and P1–P4 denote negative and positive d13Ccarb excursions per Song et al. (2013); these excursions facilitate correlations between the study sections as well as with other Lower Triassic sections globally. In A, note (1) scale change at 100 m, and (2) the age of the Wujiaping Formation in the study area is Changhsingian (late Late Permian) rather than Wujiapingian (early Late Permian; Tong et al., 2007). Abbreviations: H. p. = Hindeodus parvus, I. i. = Isarcicella isarcica, I. s. = I. staeschi, Ns. = Neospathodus, Chang. = Changhsingian, EPME = end-Permian mass extinction, L = Luolou Formation, WJ = Wujiaping Formation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

72 °C after flushing with helium. The evolved CO2 gas was analyzed for d13C and d18O using a MAT 253 mass-spectrometer coupled directly to a Finnigan Gasbench II interface (Thermo Scientific) at the China University of Geosciences in Wuhan, China. Isotopic values are reported as per mille relative to the Vienna Pee Dee belemnite (VPDB) standard. Analytical precision was better than ±0.1& for d13C based on replicate analyses of two laboratory standards (GBW 04416 and GBW 04417). A subset of samples from the Lower Guandao section (n = 20) were analyzed for d34Spy, yielding paired d34SCAS– d34Spy analyses for these samples. Pyrite sulfur was extracted using the chromium reduction method (Canfield et al., 1986). Pyrite extractions were carried out in 20 ml concentrated HCl and 40 ml 1 M chromous chloride solutions for 2 h while heated under an N2 atmosphere. The pyrite sulfur was collected as silver sulfide, which was weighed to determine pyrite sulfur concentrations. Sulfur isotope measurements were performed at the University of California at Riverside using a ThermoFinnigan Delta V continuous-flow stable-isotope ratio mass spectrometer.

3.4. Time scale The Permian–Triassic timescale has undergone major revisions in recent years based on ID-TIMS dating of zircons in volcanic ash layers, mainly from South China (Mundil et al., 2010). Current age estimates are 254.1 ± 0.1 Ma for the Wujiapingian–Changhsingian boundary, 252.28 ± 0.08 Ma for the EPME, and 252.17 ± 0.06 Ma for the Permian–Triassic boundary (Shen et al., 2010, 2011; Fig. 4). The age of the Induan–Olenekian boundary is 251.1 Ma, although durations of the substages of the Induan Stage remain under debate (Guo et al., 2008; Wu et al., 2012); we adopted estimated durations of 0.73 and 0.37 Myr for the Griesbachian and Dienerian substages, respectively, as per Guo et al. (2008). Radiometric dates constrain the ages of the Smithian–Spathian, Spathian–Anisian (Early-Middle Triassic) and Ladinian–Carnian (Middle-Late Triassic) boundaries to 250.6, 247.2, and 237.0 Ma, respectively (Lehrmann et al., 2006; Ovtcharova et al., 2006; Mundil et al., 2010). These studies have provided critical data for a revised timescale for the Late


H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113

of influence of the inverse-distance-squared weight function (i.e., window width). The standard deviation range for X at a given time t was calculated as the mean plus or minus the standard deviation (r), with the latter determined as: X ððX t  X i Þ2  ð1=ð1 þ ððt  ti Þ=kÞ2 ÞÞ rðX t Þ ¼ 0:5 X ð1=ð1 þ ððt  ti Þ=kÞ2 ÞÞ nÞ= ð2Þ

Fig. 4. Age-thickness model for the three study sections. Dajiang and Lower Guandao exhibit a pronounced increase in sedimentation rates during the Griesbachian–Smithian interval (cf. Algeo and Twitchett, 2010; Algeo et al., 2011). Differences in section-specific sedimentation rates were due to differential subsidence across the Great Bank of Guizhou. Arrows show tie-points for construction of composite section in Fig. 6. Timescale from Algeo et al. (2013). Abbreviations: G = Griesbachian, D = Dienerian, S = Smithian, Ch = Changhsingian Stage, In = Induan, LPm = Late Permian.

where n is the fractional standard deviation range of interest (i.e., 0.68 for ±1r, 0.95 for ±2r). The choice of k can significantly influence LOWESS results. Smaller k concentrate weight on local X values, producing a LOWESS curve with higher-amplitude and more rapid secular variation, whereas larger k distribute weight more evenly through the entire sample set, producing a smoother LOWESS curve. In this study, model values for each parameter were calculated at 0.1-Myr steps from 253.0 to 250.5 Ma (late Changhsingian to early Spathian) and at 0.25-Myr steps from 250.0 to 237.0 Ma (mid-Spathian through Middle Triassic) in order to take into account differences in data density and variability between these intervals (n.b., 250.5–250.0 Ma is a transition interval). For each interval, a range of k values was tested in order to determine the value that optimized the proportion of total variance accounted for by the LOWESS curve, i.e., r2L/r2T, where r2T is the total variance in a given dataset and r2T is the variance accounted for by the LOWESS curve (Fig. 5). In general, larger k produce a smoother LOWESS curve and, hence, account for a reduced proportion of total dataset variance for a given parameter. We determined empirically that k values of 0.014 and 0.051 maximized r2L/r2T for the 253.0–250.5 and 250.0– 237.0 Ma intervals, respectively (Fig. 5), and gradationally intermediate k values were used for the 250.5-to-250.0-Ma transition interval.

Permian and Early Triassic (Algeo et al., 2013; Fig. 4) that differs somewhat from the most recent iteration of the standard Phanerozoic timescale (Gradstein et al., 2012). The timescale of Algeo et al. (2013) was used in age and rate calculations of the present study. 3.5. LOWESS curves LOWESS (LOcally WEighted Scatterplot Smoothing) curves were calculated from our d34SCAS, d13Ccarb, and [CAS] datasets for the Late Permian to Middle Triassic interval. The LOWESS procedure determines a best-fit trend for irregularly distributed time-series data using an inverse-distance-squared weight function (cf. Cleveland et al., 1992). The mean value of parameter X at a given time t is: Xt ¼



ð1=ð1 þ ððt  ti Þ=kÞ Þ  X i Þ= 34


X 2 ð1=ð1 þ ððt  ti Þ=kÞ ÞÞ


where Xi is the d SCAS ; d Ccrab , or [CAS] value and ti is the age of sample i, and the equation was integrated over the full dataset (i.e., i = 1 to n, where n is the total number of samples). The parameter k determines the temporal range

Fig. 5. The inverse-distance-squared weight function (k) used in LOWESS curve modeling versus the proportion of total dataset variance explained by the LOWESS curve (r2L/r2T) for the 253– 250-Ma (red) and 250–237-Ma (blue) intervals of the d34SCAS dataset. k values of 0.014 (up arrow) and 0.051 (down arrow) for these two intervals, respectively, maximize r2L/r2T, which is 0.535 of total dataset variance. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113

The LOWESS curves were used to calculate rates of variation in d34 SCAS and d13Ccarb, i.e., od34SCAS/ot and od13Ccarb/ot. Rates of variation were calculated as: @X Lððtþðtþ1ÞÞ=2Þ [email protected] ¼ ðX LðtÞ  X Lðtþ1Þ Þ=ðt  ðt þ 1ÞÞ


where XL(t) is the value of the LOWESS curve for parameter X at time t. Calculated rates are sensitive to variations in k. A choice of larger (smaller) values of k would yield lower (higher) rate estimates, but we objectivized this process through optimization of k to maximize r2L/r2T (see above). Calculated rates are also sensitive to data density and errors in isotopic measurements. Despite modest uncertainties in absolute rate estimates, relative patterns of secular variation in od34SCAS/ot and od13Ccarb/ot are robust. 4. RESULTS Chemostratigraphic data for the 3 study sections (Fig. 3) were composited within a high-resolution biostratigraphicd13C correlation framework (Fig. 6). The study sections exhibit generally strong variation in [CAS], d34 SCAS , and d13Ccarb from the latest Permian through the Smithian substage of the Early Triassic, with diminishing variability thereafter. [CAS] ranges from 37 to 552 ppm, with a mean of 135 ± 90 ppm (Fig. 6A). [CAS] is variable throughout the Early Triassic, peaking at the PTB, in the middle Griesbachian, and in the mid-Spathian, and then declining to uniformly low values (<100 ppm) in the Middle Triassic. d34 SCAS ranges from +8.7& to +44.1&, with a mean of 25.9 ± 6.9& (Fig. 6B). Below the EPME, d34 SCAS varies between +16.6& and +26.2&. d34 SCAS fluctuates strongly during the Griesbachian through Smithian, with positive excursions in the lower Griesbachian (P1s; to +35&), upper Griesbachian (P10 s; to +35&), and Dienerian–Smithian boundary (P2s; to +44&), and negative excursions at the PTB (N1s; to +9&), in the mid-Griesbachian (N10 s; to +11&), lower Dienerian (N2s; to +16&), and upper Smithian (N3s; to +16&). The mid-Spathian is characterized by a final, broad positive excursion (P3s; to +33&), after which d34 SCAS declines in the late Spathian and then fluctuates weakly in the range of +10& to +24& during the Middle Triassic (Fig. 6B). The numbering scheme corresponds to earlier-identified excursions in the marine carbonate d13C record (Fig. 6D; Song et al., 2013), with an ‘s’ appended to denote ‘sulfate’. The larger fluctuations in d34 SCAS are not simply an artifact of increased variance in the Griesbachian–Smithian portion of the dataset because (1) all of the d34 SCAS excursions discussed above are defined by large numbers of samples, (2) the d34 SCAS LOWESS curve exhibits shifts in the mean trend that are substantially larger than the coeval standard deviation range (Fig. 6B), and (3) d34 SCAS shows statistically significant covariation with d13Ccarb (Fig. 6D, F). Rates of secular d34 SCAS variation (i.e., od34SCAS/ot) are variable but generally high during the Griesbachian– Smithian, averaging 35& Myr1 and peaking at >100& Myr1 in the early Griesbachian (Fig. 6C). These rates are considered accurate because most shifts in d34 SCAS are well defined, e.g., a10& positive shift during 100 kyr in the early Griesbachian (P1s) or a 10&


negative shift during 400 kyr in the Smithian (N3s; Fig. 6B), yielding od34SCAS/ot of 100& and 25& Myr1, respectively (Fig. 6C). od34SCAS/ot declines sharply in the Spathian and Middle Triassic, an interval characterized by uniformly low rates of isotopic change (<20& Myr1). Differences in od34SCAS/ot between the Griesbachian–Smithian and the Spathian-Middle Triassic intervals are unlikely to be due to differences in sample density (which is about 10X greater in the former; Fig. 6A, B) because d34 SCAS shows both a reduced range and coherent secular variation during the Spathian to Middle Triassic despite relatively low sample density. d13Ccarb ranges from 3.0& to +7.6&, with a mean of 1.6 ± 1.6& (Fig. 6D). Below the EPME, d13Ccarb shows limited variation (+2.3& to +3.2&). The composite d13Ccarb profile for the study sections exhibits the negative (N1–N4) and positive (P1–P4) excursions that are characteristic of the Lower Triassic globally (Song et al., 2013). Although the N1 and N2 excursions are generally well-defined, the intervening P1 excursion is typically broad and of low positive amplitude (cf. Horacek et al., 2007; Tong et al., 2007). The N1 excursion should probably be viewed as two closely spaced events in time, the first at the PTB and the second in the mid-Griesbachian (e.g., Magaritz et al., 1992; Xie et al., 2007), here labeled as N1 and N10 , respectively (Fig. 6D). This bipartite character of the N1 excursion is confirmed by our d34 SCAS profile, which shows two corresponding negative excursions in the early to middle Griesbachian (N1s and N10 s; Fig. 6B). The largest d13Ccarb excursions are P2 at the Dienerian–Smithian boundary (to +7.6&), N3 in the late Smithian (to 3.0&), and P4 in the earliest Middle Triassic (to +4.8&; Fig. 6D). Following the P4 excursion, d13Ccarb stabilizes in the range of +1.3& to +2.8& during the Middle Triassic. Rates of secular d13Ccarb variation (i.e., od13Ccarb/ot) are modestly elevated in the latest Permian to Dienerian interval (to 10& Myr1) but rise sharply in the late Dienerian-early Smithian, peaking at nearly 40& Myr1 (Fig. 6E). Thereafter, od13Ccarb/ot declines first to 10–15& Myr1 in the late Smithian-early Spathian and then to <5& Myr1 from the mid-Spathian through the Middle Triassic. d34 SCAS and d13Ccarb show pronounced patterns of covariation through most of the Late Permian-to-Middle Triassic study interval. This is demonstrated by calculation of evolutive correlation coefficients (r) for the d34 SCAS and d13Ccarb profiles (Fig. 6F). Positive covariation is evident throughout most of the study interval, in which r values are statistically significant (r P 0.67; p(a)<0.05) or subsignificant, with the exception of a 1.5-Myr-long interval in the late Spathian that shows significant negative covariation (Fig. 6F). The general pattern of positive C–S isotopic covariation was further validated by paired analysis of d34 SCAS and d13Ccarb for a subset of 59 samples from the Dajiang section. Both the d34 SCAS and d13Ccarb profiles show relatively depleted values around 10– 20, 170–200, and >400 m, and enriched values around 80– 100, 220–240, and 300–340 m, although the similarities are obscured somewhat by a long-term rise in d34SCAS that is not matched by d13Ccarb (Fig. 7A). The r2 value for these paired d34SCAS–d13Ccarb analyses is a statistically robust 0.37 (Fig. 7B).


H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113

Fig. 6. Composite profiles for (A) [CAS], (B) d34SCAS, (C) od34SCAS/ot, (D) d13Ccarb, (E) od13Ccarb/ot, and (F) correlation coefficient (r) for d34SCAS-d13Ccarb (calculated using a 7- to 11- point sliding window, with width variation for the purpose of noise reduction; gray fields represent statistically significant values of r). LOWESS curves with ±1 s.d. ranges (in pink) are shown in A, B, and D. In D, N1–N4 and P1– P4 denote negative and positive C-isotope excursions per Song et al. (2013); in B, N1s–N4s and P1s–P3s denote corresponding d34SCAS excursions. The curves in C and E represent minimum rates of secular isotopic variation because of the smoothing effects of the LOWESS procedure. Warming and cooling intervals of Early Triassic are from Joachimski et al. (2012) and Sun et al. (2012). Note change in vertical scale at 250 Ma. Abbreviations: D = Dienerian, EMTB = Early-Middle Triassic boundary, EPME = end-Permian mass extinction, Pm. = Permian, PTB = Permian–Triassic boundary. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Differences in the S-isotopic compositions of co-occurring CAS and pyrite (i.e., D34SCAS-py) were determined for a subset of 20 samples from the Lower Guandao section (Fig. 8). D34SCAS-py is variable but generally large for the Griesbachian through late Smithian, ranging from 17& to 57& with a mean of 33 ± 11&. Three samples from the latest Smithian and Spathian yielded much smaller D34SCAS-py values, with a mean of 12 ± 6&. Most of the variation in D34SCAS-py is due to variation in d34Spy, which shows weak positive covariation with d34SCAS (Fig. 9A) but strong negative covariation with D34SCAS-py (Fig. 9B). All of the data are presented in electronic annex EA-1. 5. DISCUSSION 5.1. Primary character of d34SCAS measurements We infer that the d34 SCAS and d13Ccarb data of the three study sections are primary marine values, or nearly so, because geochemical relationships show no evidence of diagenetic or procedural influences. Pyrite and organic S

generally are strongly 34S-depleted relative to co-occurring sulfate (Zaback and Pratt, 1992), and oxidation of pyrite during CAS extraction can influence measured d34 SCAS values (Marenco et al., 2008b), especially in samples in which total S concentrations are low, as in the present study units (<0.3% S in all samples). [CAS] does not show significant covariation with either total S (r2 = 0.05; Fig. 10A) or total Fe (r2 = 0.10; Fig. 10B). Further, d34 SCAS exhibits no covariation with either total S or total Fe (r2 = 0.00 and 0.01, respectively; Figs. 10C, D). Collectively, these geochemical relationships indicate that [CAS] and d34 SCAS were probably not measurably influenced by the pyrite and organic sulfur fractions of the sediment. Geochemical relationships also provide no indication that diagenesis has systematically affected CAS concentrations or d34 SCAS values. Ca/Mg ratios provide a measure of the degree of dolomitization of carbonate sediment, a process that can influence the sulfur isotopic composition of CAS (Marenco et al., 2008a). However, the study samples are overwhelmingly composed of calcite rather than dolomite, as reflected in Ca/Mg ratios mostly >20

H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113


Fig. 7. Paired d34SCAS and d13Ccarb analyses of Dajiang samples (C-isotope data original to this study) shown (A) versus stratigraphic elevation, and (B) as d34SCAS-d13Ccarb crossplot. In A, N1–N3 and P1–P2 denote positive and negative C-isotope excursions per Song et al. (2013). In B, positive covariation exists between d34SCAS and d13Ccarb; the r2 value is significant at 0.37.

(Fig. 11A, B) as well as the limited stratigraphic occurrence of dolomite beds (Fig. 3). Neither [CAS] nor d34 SCAS covaries significantly with Ca/Mg ratios (all r2 < 0.10; Fig. 11A, B). We therefore infer that the d34 SCAS results of our study are free of possible effects related to dolomitization. Mn/Sr ratios provide a measure of the degree of burial diagenesis of carbonate sediments, reflecting uptake of Mn2+ under reducing porewater conditions and loss of Sr during aragonite-to-calcite recrystallization reactions. [Mn] <300 ppm and Mn/Sr ratios <2 are considered indicative of relatively unaltered carbonate sediments (Brand and Veizer, 1980; Brand, 2004). Most of the present study samples (136 out of 143) exhibit low Mn/Sr values (<1.0; Fig. 11C, D), suggesting very limited burial diagenesis. Further, neither [CAS] nor d34SCAS covaries significantly with Mn/Sr ratios (all r2 < 0.15; Fig. 11C, D), supporting the interpretation of limited burial diagenesis. Additional considerations that support this inference are (1) d34 SCAS covaries strongly with d13Ccarb (Figs. 6F and 7), and (2) LOWESS trends account for the majority of total variance in the d34 SCAS and d13Ccarb, with r2L/r2T equal to 56% and 89%, respectively (Fig. 6B, D). These observations are

consistent with the presence of a recognizable secular signal of primary marine origin in each geochemical profile. CAS concentrations are known to decrease with progressive burial diagenesis (Staudt and Schoonen, 1995; Hurtgen et al., 2006; Gill et al., 2008), rendering the significance of absolute CAS concentrations uncertain. The isotopic composition of CAS potentially can be altered by diagenetic processes, e.g., bacterial sulfate reduction and sulfide oxidation (Kampschulte and Strauss, 2004). However, these earlier studies have concluded that much of the diagenetically reworked sulfate is reincorporated into secondary carbonate phases, and that diagenetic losses commonly does not measurably affect the d34S of CAS in bulk samples. 5.2. Seawater sulfate drawdown during the Early Triassic Several methods have been used to estimate sulfate concentrations in ancient seawater ([SO42]SW). Low [CAS] has been cited as evidence of reduced [SO42]SW (Kah et al., 2004; Gellatly and Lyons, 2005). However, care must be exercised in interpreting CAS concentration data since


H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113

Fig. 8. (A) d34SCAS and d34Spy, and (B) D34SCAS-py for Lower Guandao. For d34SCAS, red diamonds represent samples with paired d34Spy analyses (red triangles), and blue diamonds represent other samples. Note the abrupt decrease in D34SCAS-py in the late Smithian. The gray field represents our best estimate of D34SCAS-py (32& to 40&) for sulfate-unlimited pyrite formation in Early Triassic seas; the two samples yielding D34SCAS-py >40& are regarded as anomalous. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

>80% of CAS can be lost during diagenesis (Lyons et al., 2004). In this study, the concurrent increase in [CAS] and reduction in od34SCAS/ot during the Spathian (Fig. 6A, C) may have resulted from increasing seawater sulfate concentrations following an interval of strong drawdown during the Griesbachian–Smithian. However, [CAS] does not show a clear relationship to d34SCAS during the pre-Spathian, and [CAS] declines to uniformly low values during the Middle Triassic (Fig. 6A), an interval for which there is no independent evidence for seawater sulfate drawdown. Thus, [CAS] is probably an unreliable proxy for [SO42]SW. The sulfate concentration of seawater is related to rates of variation in its S isotopic composition (od34S/ot), which can be proxied by od34SCAS/ot (Kah et al., 2004). Low rates of od34SCAS/ot are associated with a large seawater sulfate reservoir, as for the modern ocean, in which [SO42]SW is 28 mM (Millero, 2005) and od34SCAS/ot has not exceeded 0.5& Myr1 since 65 Ma (Paytan et al., 1998). On the other hand, seawater sulfate d34S has the potential to vary rapidly at low [SO42]SW, although such variation may be limited if the fluxes and isotopic compositions of the sources and sinks remain relatively uniform. For this reason, maximum rather than mean rates of od34SCAS/ot are more useful for assessment of [SO42]SW. In the study units, od34SCAS/ot averages 35& Myr1 and ranges up to 100& Myr1 during the Griesbachian–Smithian (Fig. 6C). Indeed, these values must be viewed as minimum estimates of the true maximum od34SCAS/ot values because the LOWESS

calculation procedure results in smoothing of high-frequency variation within the d34SCAS sample set (Fig. 6B). The high values of od34SCAS/ot in the study units support the inference of strong sulfate drawdown in Early Triassic seawater (cf. Luo et al., 2010). Another approach to evaluating [SO42]SW is based on sulfur isotopic fractionation between co-occurring sedimentary sulfate and pyrite (D34SCAS-py) (Habicht and Canfield, 1996, 1997, 2001). Modern euxinic lacustrine and marine environments in which sulfate concentrations are low (61 mM) typically yield D34SCAS-py of <10&. For example, the Black Kichier and Great Kichier lakes in Russia have sulfate concentrations of 1 and 0.55 mM and are characterized by D34SCAS-py of 10& and 1.9&, respectively (Canfield et al., 2010; Sim et al., 2011). On the other hand, modern environments in which sulfate concentrations are >1 mM typically yield D34SCAS-py of >10&, and many fully marine systems have D34SCAS-py of 45 ± 15& (Canfield and Thamdrup, 1994; Bru¨chert et al., 2001; Werne et al., 2003). However, D34SCAS-py can be influenced by factors other than [SO42], including sulfate reduction rates, organic types, and biochemical differences among sulfate-reducing microbes (Detmers et al., 2001; Bru¨chert, 2004; Kleikemper et al., 2004; Sim et al., 2011), and values as high as 72& have been reported from environments with [SO42] significantly lower than that of seawater (Canfield et al., 2010). All of these D34SCAS-py values relate to bacterial sulfate reduction in either the water column or an open

H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113

Fig. 9. (A) d34Spy versus d34SCAS, and (B) d34Spy versus D34SCAS-py for Lower Guandao. In B, negative covariation exists between d34Spy and D34SCAS-py; the r2 value is significant at 0.72.

diagenetic system (in which sediment at or just below the sediment–water interface remains in contact with the water column). In marine systems containing large quantities of dissolved sulfate, the concentration and isotopic composition of aqueous sulfate does not vary significantly as a function of progressive sulfate reduction, and D34SCAS-py will be maximized. In contrast, closed or semi-closed diagenetic systems (such as sediment porewaters) are characterized by Rayleigh distillation, which results in a progressive increase in the d34S of both sulfide and residual sulfate as [SO42] diminishes and, thus, in smaller D34SCAS-py. Under such conditions, pyrite commonly forms at a range of degrees of system openness, resulting in a broad pyrite d34S distribution with D34SCAS-py values that are reduced (sometimes greatly so) relative to open-system values. In the study units, samples of latest Smithian to Middle Triassic age yield relatively small D34SCAS-py (12 ± 6&; Fig. 8B). Given lack of evidence for water-column sulfide in the South China region at that time, we infer that these are closed-system values from diagenetic pyrite. In contrast, Griesbachian–Smithian samples yield large D34SCAS-py (36 ± 4&; Fig. 8B) that are more likely to represent open-system pyrite formation. The transition from high to low D34SCAS-py values in the latest Smithian thus signifies a shift from predominantly open- to closed-system conditions. Furthermore, the degree of openness of pyrite formation is broadly related to ambient benthic redox conditions, because a high flux of syngenetic pyrite framboids from an euxinic water column to the sediment generally skews


observed D34SCAS-py distributions toward high values. Although not identified in the present study samples, framboidal pyrite has been found in Lower Triassic carbonates of the nearby Dawen section (Algeo et al., 2008) as well as elsewhere in South China (Shen et al., 2007). The transition from high to low D34SCAS-py values in the latest Smithian thus may signify a shift from euxinic (or intermittently euxinic) to persistently well-oxygenated watermass conditions at the study site. We note, however, that the relationship of system openness to benthic redox conditions is not invariant, and that exceptions to the pattern above exist both in the modern ocean (e.g., Aller et al., 2010) and in low-sulfate lacustrine systems (e.g., Gomes and Hurtgen, 2013). In the study units, the largest D34SCAS-py values for the Griesbachian–Smithian (36 ± 4&, except for a few outliers; Fig. 8B) suggest strong fractionation of sulfur isotopes between co-occurring sulfate and pyrite in Early Triassic seas and, hence, relatively high [SO42]SW. These values are considerably larger than D34SCAS-py for Archean and Mesoproterozoic marine formations, which are typically <10& and from which [SO42]SW estimates of <200 lM have been derived (Shen et al., 2001; Habicht et al., 2002; Luo et al., in review). However, they are at the lower end of the range for modern marine D34SCAS-py values (45 ± 15&) and, hence, presumably indicative of seawater sulfate concentrations below the modern level of 28 mM. Thus, while [SO42]SW can be constrained by both od34SCAS/ot and D34SCAS-py, it is not immediately evident that these two parameters of the present study units are yielding similar estimates for Early Triassic [SO42]SW. Further, existing estimates of Early Triassic [SO42]SW based on the composition of fluid inclusions in halite (20 mM; Horita et al., 2002; Lowenstein et al., 2005) and modeling of od34SCAS/ot (4 mM; Luo et al., 2010) are not in particularly close agreement. The marine S cycle, which has a limited number of fluxes with well-defined S-isotopic ranges (Bottrell and Newton, 2006), is amenable to analysis through modeling (Kah et al., 2004). Subaerial weathering yields a riverine sulfate source flux (FQ) of 10  1013 g yr1 with an average d34S of +6&. Seawater sulfate (modern d34S of +20&) is removed to the sediment either as isotopically heavy CAS or evaporite deposits (FEV = 6  1013 g yr1, D34SSW-ev of 0 to +4&) or as isotopically light pyrite (FPY = 4  1013 g yr1, D34SSW-py of 30& to 60&; Habicht and Canfield, 1997). To more narrowly constrain the [SO42]SW of Early Triassic seas, we adapted the model of Kah et al. (2004) to calculate [SO42]SW based on model inputs for D34SCAS-py and od34SCAS/ot. They calculated maximum rates of isotopic change for seawater sulfate based on estimates of D34SCAS-py and the mass of sulfate in seawater (MSW): @d34 S CAS [email protected]ðmaxÞ ¼ F PY xD34 S CASpy =M SW


We reorganized this equation to calculate seawater sulfate concentration based on estimates of D34SCAS-py and od34SCAS/ot(max): MSW ¼ k 1  FPY  D34 SCASpy [email protected] SCAS [email protected]ðmaxÞ



H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113

Fig. 10. [CAS] versus (A) total S and (B) total Fe, and d34SCAS versus (C) total S and (D) total Fe. None of the relationships are statistically non-significant at p(a)<0.01. In B–D and in Fig. 11, elemental data are available only for the Lower and Upper Guandao sections.

½SO2 4 SW ¼ k2  MSW


where FPY is the sink flux of reduced sulfur (pyrite) from seawater, and k1 and k2 are relational dimensional constants (106 and 2.15  1020 mM g1, respectively). Kah et al. (2004) used FPY = 10  1013 g yr1, which is the total sink flux for modern seawater sulfate, in order to model maximum rates of change in od34SCAS/ot(max). While this may be appropriate for intervals of widespread euxinia in the global ocean, FPY = 4  1013 g yr1 (i.e., the modern value) may better represent interval with well-oxygenated oceans (Fig. 12A). For values of D34SCAS-py and od34SCAS/ ot(max) that are potentially representative of the modern ocean (e.g., 35& and 1.1& Myr1; see discussion below), Eq. (5) yields the modern seawater sulfate mass of MSW = 1.3  1021 g and Eq. (6) the modern seawater sulfate concentration of 28 mM. The significance of our reformulation of these relationships is that [SO42]SW now can be calculated as a function of two parameters that are measurable in paleoceanographic systems, i.e., D34SCAS-py and od34SCAS/ot(max). To explore relationships among the model parameters, we varied D34SCAS-py from 1 to 60& for five discrete values of od34SCAS/ot(max) from 1 to 100& Myr1 (Fig. 12). For any given value of od34SCAS/ot(max), [SO42]SW increases linearly with increasing D34SCAS-py. Increasing od34SCAS/

ot(max) results in a linear decrease in [SO42]SW for a given value of D34SCAS-py. Because observed values of od34SCAS/ ot are generally smaller than the maximum possible rate of change in d34SCAS, the resulting estimates of [SO42]SW are generally larger than true seawater sulfate concentrations; thus, Eq. (6) yields the maximum possible [SO42]SW. This is illustrated by a calculation for the modern ocean, using D34SCAS-py of 30–60& (e.g., Canfield and Thamdrup, 1994) and od34SCAS/ot of 60.5& Myr1 (n.b., rates derived from the Cenozoic seawater sulfate d34S record; Paytan et al., 1998). These inputs yield a range of [SO42]SW values from 40 to 110 mM, somewhat greater than the actual modern seawater sulfate concentration of 28 mM (Fig. 12A). This overestimate of modern [SO42]SW results from observed od34SCAS/ot values for the Cenozoic (i.e.,60.5& Myr1) not approaching the theoretical maximum for modern seawater (1.0–1.2& Myr1), which would be achieved only if the source flux of sulfur to the global ocean were reduced to zero (Kah et al., 2004). The relationships expressed in Eqs. (5), (6) allow calculation of maximum likely seawater sulfate concentrations for Early Triassic seas. Using values of 32–40& for D34SCAS-py (Fig. 8B) and 20–60& Myr1 for od34SCAS/ot (Fig. 6C), we calculated that the [SO42]SW of Early Triassic seas probably did not exceed 0.5–1.5 mM for FPY = 4  1013 g yr1 (Fig. 12A) or 1.2–4.2 mM for

H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113


Fig. 11. (A) [CAS] and (B) d34SCAS versus Ca/Mg, and (C) [CAS] and (D) d34SCAS versus Mn/Sr. All relationships yield r2 < 0.10 and are statistically non-significant at p(a)<0.01.

FPY = 10  1013 g yr1 (Fig. 12B). The latter values are effectively identical to the 4 mM estimate of Luo et al. (2010) and validate this earlier estimate. These relationships impose significant constraints on the range of viable [SO42]SW values for Early Triassic seas. Low seawater sulfate concentrations during the Early Triassic may have resulted from two factors: (1) drawdown prior to the Early Triassic as a consequence of massive marine evaporite deposition during the Late Permian (Anderson et al., 1972; Tucker, 1991; Hay et al., 2006); and (2) rapid expansion of oceanic anoxia during the end-Permian crisis (Wignall and Twitchett, 2002; Algeo et al., 2011), increasing the burial flux of pyrite (Luo et al., 2010). However, increased subaerial erosion rates during the Early Triassic (Algeo and Twitchett, 2010; Algeo et al., 2011) would have increased the source flux of sulfur to the global ocean at the same time and thus served as a counterbalancing influence. Current estimates are that the area of anoxic oceanic sedimentation expanded by a factor of 7X during the end-Permian crisis (Brennecka et al., 2011), and that Early Triassic weathering fluxes were 2–3X higher than prior to this event (Algeo et al., 2013). Based on these findings, we modeled the effects on [SO42 ]SW and sSO4 (i.e., the residence time of seawater sulfate) of varying the riverine source and marine-sediment sink fluxes under three scenarios (Fig. 13). For a seawater sulfate mass equal to that of the modern ocean (MSW-0Ma = 1.0), [SO42 ]SW estimates are between 5.5 and 11 mM (‘b’ in Fig. 13A) for elevated source and sink fluxes (i.e., Scenarios 2 and 3).

These values are above the previously calculated range of viable values for Early Triassic seawater (Fig. 12) and imply that the mass of seawater sulfate in the Early Triassic could not have been as large as that of the modern ocean. If seawater sulfate mass were first reduced to 0.3X that of the modern ocean, then Scenarios 2 and 3 yield [SO42]SW between 1.5 and 3.2 mM (‘c’ in Fig. 13A), consistent with estimated values for an anoxic ocean (Fig. 12B). A further reduction in seawater sulfate mass to 0.1X that of the modern ocean yields [SO42]SW between 0.5 and 1.1 mM (‘d’ in Fig. 13A), consistent with estimated values for a well-oxygenated ocean (Fig. 12A). In summary, we infer that the mass of Early Triassic seawater sulfate probably had been reduced by Late Permian evaporite deposition to between 10 and 30% of that of the modern ocean (MSW-0Ma = 0.1– 0.3), and that a contemporaneous increase in the sink fluxes relative to the source fluxes of seawater sulfate (Scenarios 2 and 3) resulted in a further decline in [SO42]SW to a range (0.5–3.2 mM) that is consistent with [SO42]SW estimates modeled on the basis of observed D34SCAS-py and od34SCAS/ ot(max) values (Fig. 12). 5.3. Carbon–sulfur cycle linkage and global climate change The marine S cycle is important, in part, because of its intimate linkage with carbon fluxes and, hence, atmospheric composition and global climate. For example, reduced (pyritic) sulfur is co-buried with reduced (organic) carbon and, thus, linked to atmospheric O2 levels (Berner, 2005),


H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113

Fig. 12. Seawater sulfate concentrations ([SO42]) versus S-isotope fractionation (D34SCAS-py) for pyrite burial fluxes of (A) 4X and (B) 10X, where X = 1013 g yr1. Diagonal lines represent constant maximum rates of change in CAS-d34S (i.e., od34SCAS/ot). Given constraints on D34SCAS-py and od34SCAS/ot, [SO42]SW can be estimated by projection to the abscissa (dashed lines). The study units yield D34SCAS-py of 32–40& (Fig. 8B) and od34SCAS/ot of 20–60& Myr1 (Fig. 6C), as shown by the red fields. Projection to the x-axis implies Early Triassic seawater sulfate concentrations of 500–1500 lM (A) or 1200–4200 lM (B). Panel A is more relevant to well-oxygenated oceans, and panel B to widely anoxic oceans. Because the LOWESS calculation procedure smoothes out high-frequency variation, observed od34SCAS/ot values (e.g., Fig. 6C) may underestimate actual maximum rates of d34SCAS variation (gray fields); consequently, the [SO42]SW estimates yielded by this procedure are maxima and actual seawater sulfate concentrations may be lower. Parameters for the modern ocean are shown in panel A: 30–60& for D34SCAS-py and 60.5& Myr1 for od34SCAS/ot (blue field). Projection to the abscissa implies maximum [SO42]SW of 40–110  103 lM, which is consistent with the observed present-day seawater sulfate concentration of 28  103 lM (Millero, 2005). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

and depletion of marine sulfate can result in a shift toward mainly methanogenic remineralization of organic matter and, thus, to an increased atmospheric flux of a potent greenhouse gas (Habicht et al., 2002). Although a staple of modeling studies (e.g., Berner, 2005), surprisingly few empirical studies have documented direct connections between the marine carbon and sulfur cycles during the Permian–Triassic transition. Algeo et al. (2007, 2008) docu-

mented positive covariation between pyrite d34S and carbonate d13C in a shallow carbonate platform succession, which they attributed to incursions of sulfidic deepwaters containing 13C-depleted dissolved inorganic carbon (DIC), and Luo et al. (2010) documented positive covariation between CAS-d34S and carbonate d13C in the lowermost Triassic microbialite facies, which they attributed to coburial of 34S-depleted pyrite and 13C-depleted organic matter. Both of these examples are from short (<10 m) sections representing the lowermost Triassic (H. parvus Zone) in South China. The degree of linkage of the marine carbon and sulfur cycles through larger intervals of the Early Triassic remains unknown. In the present study, d34SCAS shows distinct patterns of covariation with d13Ccarb. The pre-EPME record is too short to evaluate, but significant to subsignificant positive covariation exists through most of the Early Triassic, from the Griesbachian to the mid-Spathian (Fig. 6F). This covariation is reflected in nearly parallel excursions in the CAS-d34S (Fig. 6B) and marine carbonate d13C (Fig. 6D) profiles, with only minor apparent differences in timing. Following the P3 excursion of early Spathian age, the pattern of positive covariation breaks down, and the N4 and P4 excursions of the d13Ccarb profile appear to have no counterparts in the d34SCAS profile (Fig. 6B, D). It is possible, however, that a mid-Anisian negative excursion in d34S, N4s(?), is a distant echo of the N4 C-isotope excursion, and that the delayed response was due to a large increase in the mass of seawater sulfate between the mid-Spathian and mid-Anisian. Strongly reduced variation in both d34SCAS and d13Ccarb during the Middle Triassic makes the question of covariation largely irrelevant. The nearly simultaneous fluctuations in the marine carbonate d13C and sulfate d34S records during most of the Early Triassic imply similar response times for these two isotopic systems, which would be possibly only if DIC and SO42 had similar residence times in seawater. The residence time of DIC in the modern ocean is 100–200 kyr, far shorter than the 13-Myr residence time of sulfate (Zeebe and Wolf-Gladrow, 2001), making covariation of carbonate d13C and sulfate d34S impossible in modern marine sediments. Major changes in the concentration and residence time of DIC in seawater through the Phanerozoic are unlikely in view of the persistent formation of equatorial inorganic carbonates (Wilkinson et al., 1985) as well as geochemical modeling constraints (Wilkinson and Algeo, 1989). If so, then the residence time of sulfate in seawater must have been much shorter during the Early Triassic than at present. Modeling of the marine S cycle (Fig. 13) indicates the conditions under which such short residence times could have been achieved. Seawater sulfate residence times of 100–200 kyr appear possible only if (1) the mass of seawater sulfate had been drawn down to 10% of that of the modern ocean prior to the Early Triassic, and (2) an Early Triassic increase of the sink flux of sulfate substantially exceeded that of the source flux (‘d’ in Fig. 13B). The pattern of close covariation between d13Ccarb and d34SCAS thus provides another line of evidence supporting extreme drawdown of Early Triassic seawater sulfate concentrations.

H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113


Fig. 13. Variation in (A) [SO42]SW and (B) sSO4 (i.e., the residence time of sulfate in seawater) as a function of the mass of seawater sulfate (shown as a fraction of the modern value, MSW-0Ma). Scenario 1 uses the present-day values of the riverine source (1Q) and marine-sediment sink (1S) fluxes (both 10  1013 g yr1); Scenario 2 increases the source and sink fluxes by factors of 2X and 5X, respectively, and Scenario 3 increases these fluxes to 2X and 10X, respectively, in order to model inferred changes in continental weathering rates and areas of anoxic marine sedimentation during the Early Triassic (e.g., Algeo and Twitchett, 2010; Algeo et al., 2011; Brennecka et al., 2011). Point ‘a’ represents modern ocean sulfate parameters, i.e., [SO42]SW = 28 mM and sSO4 = 1.3  107 yr; see text for discussion of points ‘b’ to ‘d’.

Understanding the nature of coupling between the marine carbonate d13C and sulfate d34S records can provide insights into contemporaneous oceanic processes. Reconstruction of variation in Early Triassic sea-surface temperatures (SSTs) (Joachimski et al., 2012; Sun et al., 2012) provides a framework for interpreting the d13Ccarb and d34SCAS records. Major warming events occurred during the latest Changhsingian, mid-Griesbachian, and late Smithian, contemporaneously with the N1, N10 , and N3 excursions in d13Ccarb (and their d34SCAS counterparts; Fig. 6). On the other hand, strong cooling took place in the late Dienerian, earliest Spathian, and latest Spathian, contemporaneously with the P2, (early) P3, and P4 excursions in d13Ccarb (and the P2s and P3s excursions in d34SCAS; Fig. 6). The correspondence between the SST and d13Ccarb and d34SCAS records is rather good: warming events were associated with negative C- and S-isotope shifts and cooling events with positive isotope shifts. Thus, seawater temperature may have been an important control on the Early Triassic marine carbon and sulfur cycles (although it is also possible that all three were responding to another forcing). Warming episodes during the Early Triassic are generally inferred to have been triggered by massive eruptions of the Siberian Traps magmatic system (Wignall, 2007; Algeo et al., 2011), although this relationship has not been definitively demonstrated to date. There are a number of ways in which SSTs may have influenced d13Ccarb and d34SCAS. Warm intervals may have triggered expansion of oceanic oxygen-minimum zones and, thus, chemocline upward excursions that introduced 34S-depleted sulfide and 13C-depleted DIC into the surface mixed layer (Kump et al., 2005; Algeo et al., 2007, 2011). Subsequent oxidation of hydrogen sulfide would have generated isotopically light sulfate (Riccardi et al., 2006), resulting in lower d34SCAS during warm intervals (Fig. 6). However, bacterial sulfate reduction in deep waters would have enriched the seawater sulfate pool in 34S, so it is not apparent whether this process could have altered bulk seawater

sulfate d34S. Warm intervals were probably also associated with enhanced volcanic emissions of sulfur, which, if isotopically light, would have contributed to negative shifts in seawater sulfate d34S (Newton et al., 2004). Rapid rates of change in d34SCAS during the Early Triassic (e.g., in the mid-Griesbachian and late Smithian; Fig. 6C) are associated with warming episodes, as might be expected for a volcanic control. Both oceanic mixing and volcanic eruptions are geologically rapid processes, however, that may not have been able to sustain shifts in d13Ccarb and d34SCAS that continued for hundreds of thousands to a few million years (Fig. 6). The most likely long-term control on Early Triassic seawater DIC-d13C and sulfate d34S is secular variation in the burial fluxes of organic matter and pyrite (Bottrell and Newton, 2006). Other factors being equal, increases in marine productivity and organic carbon burial fluxes result in higher d13Ccarb (through burial of 13C-depleted organic carbon) and higher d34SCAS (through coburial of 34S-depleted pyrite); decreases in productivity and organic burial have the opposite effect. With regard to the Early Triassic, increasing (decreasing) d13Ccarb and d34SCAS values are associated with cool (warm) intervals (Fig. 6). This pattern implies an association of cooling (warming) with higher (lower) marine productivity, a relationship that could have been controlled by rates of oceanic overturning circulation and mixing of nutrients into the ocean surface layer. Stratification of the oceanic water column is thought to have intensified considerably during the Early Triassic (Horacek et al., 2007; Song et al., 2013), which would have sharply reduced the level of nutrients in the ocean-surface layer. Under these conditions, relatively small changes in overturning circulation rates, which weakened (strengthened) during warm (cool) intervals, would have had a large effect on marine productivity and, hence, on d13Ccarb and d34SCAS. These inferences are consistent with the conclusions of Berner (2005) regarding a general decline in the global rate of organic matter burial and an increase in


H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113

marine-sediment S/Corg ratios during the latest Permian to Early Triassic. The highly perturbed oceanic conditions that had existed since the EPME began to ameliorate during the Spathian and effectively terminated around the Early-Middle Triassic boundary. Declining rates of variation in d34SCAS and d13Ccarb during the Spathian imply a progressive stabilization of the oceanic environment as well as increases in the mass and residence time of seawater sulfate (Fig. 6C, E). This inference is supported by the decoupling of the marine C and S cycles in the late Spathian (Fig. 6F). These developments coincided with a two-step cooling during the Spathian, reducing SSTs from a blistering >40 °C to a still warm 32 °C by the early Middle Triassic (Romano et al., 2012; Sun et al., 2012). Climatic cooling in the early Spathian was linked to a return to less intense oceanic water-column stratification (Song et al., 2013) and to a strong recovery among many clades of marine invertebrates (Brayard et al., 2006, 2009; Bottjer et al., 2008). Recovery of terrestrial ecosystems, as reflected in regrowth of coniferous forests, took place approximately at the same time (Looy et al., 1999; Galfetti et al., 2007; Hermann et al., 2010). We infer that these changes were due to a rapid tapering off of Siberian Traps magmatic activity. The timing of major oceanic changes, as documented herein, ultimately may prove to be a useful tool in refining the eruption chronology of the Siberian Traps. 6. CONCLUSIONS The d34S of carbonate-associated-sulfate (CAS), a proxy for seawater sulfate, in three Lower Triassic sections in South China varies from +8.7& to +44.1& at rates of up to 100& Myr1 during the Griesbachian–Smithian substages of the Early Triassic. Such rapid variation requires that seawater sulfate was drawn down to low concentrations (64 mM) and had a short residence time (6200 kyr). Positive covariation with d13Ccarb during this interval reflects strong coupling of the organic carbon and pyrite burial fluxes, with negative (positive) shifts in d13Ccarb and d34SCAS due to climatic warming (cooling) and decreased (increased) marine productivity and microbial sulfate reduction. Major cooling during the Spathian substage reinvigorated oceanic overturning circulation, resulting in a sustained increase in marine productivity, contraction of marine anoxia, and reduction in pyrite burial. As seawater sulfate built to higher concentrations, the close coupling of the marine C and S cycles came to an end during the late Spathian, and a general amelioration of marine environmental conditions set the stage for a strong recovery of invertebrate faunas. Variations in seawater sulfate and general oceanic conditions during the Early Triassic were probably controlled by climate change associated with major eruptive phases of the Siberian Traps. ACKNOWLEDGMENTS We thank Junhua Huang, Dabo Wang, Yong Du, and Yanling Xiong for analytical assistance, Jonathan Payne for providing the carbon isotopic data from his 2004 Science paper, and David John-

ston, Linda Kah and Maya Gomes for constructive reviews of the manuscript. This project was supported by the 973 program (2011CB808800), the Chinese National Natural Science Foundation (40830212, 41172312, 41172036, 41272372, 41240016, 41302271 and 41302010), the “111 Project” (B08030), the Open Research Program of BGEG (1016), the State Key Laboratory of Palaeobiology and Stratigraphy (133111), the Fundamental Research Funds for the Central Universities of CUG, and the US National Science Foundation (Grants EAR-0745574 and EAR-1053449) and the NASA Exobiology program for support to T.J.A. This paper is a contribution to IGCP Project 572.

APPENDIX A. SUPPLEMENTARY DATA Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.gca.2013.12.009. REFERENCES Algeo T. J. and Twitchett R. J. (2010) Anomalous Early Triassic sediment fluxes due to due to elevated weathering rates and their biological consequences. Geology 38, 1023–1026. Algeo T. J., Chen Z. Q., Fraiser M. L. and Twitchett R. J. (2011) Terrestrial–marine teleconnections in the collapse and rebuilding of Early Triassic marine ecosystems. Palaeogeogr. Palaeoclimatol. Palaeoecol. 308, 1–11. Algeo T. J., Ellwood B. B., Nguyen T. K. T., Rowe H. and Maynard J. B. (2007) The Permian–Triassic boundary at Nhi Tao, Vietnam: Evidence for recurrent influx of sulfidic watermasses to a shallow-marine carbonate platform. Palaeogeogr. Palaeoclimatol. Palaeoecol. 252, 304–327. Algeo T. J., Henderson C. M., Tong J. N., Feng Q. L., Yin H. F. and Tyson R. V. (2013) Plankton and productivity during the Permian–Triassic boundary crisis: an analysis of organic carbon fluxes. Glob. Planet. Change 105, 52–67. Algeo T. J., Hinnov L., Moser J., Maynard J. B., Elswick E., Kuwahara K. and Sano H. (2010) Changes in productivity and redox conditions in the Panthalassic Ocean during the latest Permian. Geology 38, 187–190. Algeo T. J., Shen Y. A., Zhang T. G., Lyons T., Bates S., Rowe H. and Nguyen T. K. T. (2008) Association of 34 S-depleted pyrite layers with negative carbonate d13 C excursions at the Permian– Triassic boundary: evidence for upwelling of sulfidic deep-ocean water masses. Geochem. Geophys. Geosyst., 9, Q04025, 10. Aller R. C., Madrid V., Chistoserdov A., Aller J. Y. and Heilbrun C. (2010) Unsteady diagenetic processes and sulfur biogeochemistry in tropical deltaic muds: implications for oceanic isotope cycles and the sedimentary record. Geochim. Cosmochim. Acta 74, 4671–4692. Alroy J.34 coauthors (2008) Phanerozoic trends in the global diversity of marine invertebrates. Science 321, 97–100. Anderson R. Y., Dean, Jr., W. E., Kirkland D. W. and Snider H. I. (1972) Permian Castile varved evaporite sequence, West Texas and New Mexico. Geol. Soc. Am. Bull. 83, 59–86. Berner R. A. (2005) The carbon and sulfur cycles and atmospheric oxygen from middle Permian to middle Triassic. Geochim. Cosmochim. Acta 69, 3211–3217. Bottjer D. J., Clapham M. E., Frasier M. L. and Powers C. M. (2008) Understanding mechanisms for the end-Permian mass extinction and the protracted Early Triassic aftermath and recovery. GSA Today 18, 4–10.

H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113 Bottrell S. H. and Newton R. J. (2006) Reconstruction of changes in global sulfur cycling from marine sulfate isotopes. Earth-Sci. Rev. 75, 59–83. Brand U. (2004) Carbon, oxygen and strontium isotopes in Paleozoic carbonate components: an evaluation of original seawater-chemistry proxies. Chem. Geol. 204, 23–44. Brand U. and Veizer J. (1980) Chemical diagenesis of a multicomponent carbonate system – 1: trace elements. J. Sed. Pet. 50, 1219–1236. Brayard A., Bucher H., Escarguel G., Fluteau F., Bourquin S. and Galfetti T. (2006) The Early Triassic ammonoid recovery: paleoclimatic significance of diversity gradients. Palaeogeogr. Palaeoclimatol. Palaeoecol. 239, 374–395. Brayard A., Escarguel G., Bucher H., Monnet C., Bru¨hwiler T., Goudemand N., Galfetti T. and Guex J. (2009) Good genes and good luck: ammonoid diversity and the end-Permian mass extinction. Science 325, 1118–1121. Brennecka G. A., Herrmann A. D., Algeo T. J. and Anbar A. D. (2011) Rapid expansion of oceanic anoxia immediately before the end-Permian mass extinction. Proc. Nat. Acad. Sci. USA 108, 17631–17634. Bru¨chert V. (2004) Physiological and ecological aspects of sulfur isotope fractionation during bacterial sulfate reduction. In Sulfur Biogeochemistry–Past and Present (eds. J. P. Amend, K. J. Edwards and T. W. Lyons). Geol. Soc. Am. Spec. Pap. 379, pp. 1–16. Bru¨chert V., Knoblauch C. and Jørgensen B. B. (2001) Microbial controls on the stable sulfur isotope fractionation during bacterial sulfate reduction in Arctic sediments. Geochim. Cosmochim. Acta 65, 753–766. Burdett J. W., Arthur M. A. and Richardson M. A. (1989) Neogene seawater sulfur isotope age curve from calcareous pelagic microfossils. Earth Planet. Sci. Lett. 94, 189–198. Canfield D. E. and Thamdrup B. T. (1994) The production of 34Sdepleted sulfide during disproportionation of elemental sulfur. Science 266, 1973–1975. Canfield D. E., Farquhar J. and Zerkle A. L. (2010) High isotope fractionations during sulfate reduction in a low-sulfate euxinic ocean analog. Geology 38, 415–418. Canfield D. E., Raiswell R., Westrich J. T., Reaves C. M. and Berner R. A. (1986) The use of chromium reduction in the analysis of reduced sulfur in sediments and shales. Chem. Geol. 54, 149–155. Chen J., Beatty T. W., Henderson C. M. and Rowe H. (2009) Conodont biostratigraphy across the Permian–Triassic boundary at the Dawen section, Great Bank of Guizhou, Guizhou Province, South China: implications for the Late Permian extinction and correlation with Meishan. J. Asian Earth Sci. 36, 442–458. Claypool G. E., Holser W. T., Kaplan I. R., Sakai H. and Sak I. (1980) The age curves of sulfur and oxygen isotopes in marine sulfate and their mutual interpretation. Chem. Geol. 28, 199– 260. Cleveland W. S., Grosse E. and Shyu W. M. (1992) Local regression models. In Statistical Models in S (eds. J. M. Chambers and T. J. Hastie). Wadsworth and Brooks/Cole, Pacific, Grove, pp. 309–376. Cortecci G., Reyes E., Berti G. and Casati P. (1981) Sulfur and oxygen isotopes in Italian marine sulfates of Permian and Triassic ages. Chem. Geol. 34, 65–79. Detmers J., Bru¨chert V., Habicht K. S. and Kuever J. (2001) Diversity of sulfur isotope fractionations by sulfate-reducing prokaryotes. Appl. Environ. Microbiol. 67, 888–894. Enos P., Lehrmann D. J., Wei J. Y., Yu Y. Y., Xiao J. F., Chaikin D. H., Minzoni M., Berry A. K. and Montgomery P. (2006) Triassic evolution of the Yangtze Platform in Guizhou


Province, People’s Republic of China. Geol. Soc. Am. Spec. Pap. 417, 105. Erwin D. H. (1994) The Permo–Triassic extinction. Nature 367, 231–236. Galfetti T., Hochuli P. A., Brayard A., Bucher H., Weissert H. and Vigran J. O. (2007) Smithian–Spathian boundary event: evidence for global climatic change in the wake of the endPermian biotic crisis. Geology 35, 291–294. Gellatly A. M. and Lyons T. W. (2005) Trace sulfate in midProterozoic carbonates and the sulfur isotope record of biospheric evolution. Geochim. Cosmochim. Acta 69, 3813– 3829. Gill B. C., Lyons T. W. and Frank T. D. (2008) Behavior of carbonate-associated sulfate during meteoric diagenesis and implications for the sulfur isotope paleoproxy. Geochim. Cosmochim. Acta 72, 4699–4711. Gomes M. L. and Hurtgen M. T. (2013) Sulfur isotope systematics of a euxinic, low-sulfate lake: evaluating the importance of the reservoir effect in modern and ancient oceans. Geology 41, 663– 666. Gorjan P., Kaiho K., Kakegawa T., Niitsuma S., Chen Z. Q., Kajiwara Y. and Nicora A. (2007) Paleoredox, biotic and sulfur-isotopic changes associated with the end-Permian mass extinction in the western Tethys. Chem. Geol. 244, 483–492. Gradstein, F. M., Ogg, J. G., Schmitz, M. and Ogg, G. (eds.) (2012) The Geologic Time Scale 2012. Elsevier, Amsterdam. Guo G., Tong J., Zhang S., Zhang J. and Bai L. (2008) Cyclostratigraphy of the Induan (Early Triassic) in West Pingdingshan Section, Chaohu, Anhui Province. Sci. China Ser. D 51, 21–29. Habicht K. S. and Canfield D. E. (1996) Sulphur isotope fractionation in modern microbial mats and the evolution of the sulphur cycle. Nature 382, 342–343. Habicht K. S. and Canfield D. E. (1997) Sulfur isotope fractionation during bacterial sulfate reduction in organic-rich sediments. Geochim. Cosmochim. Acta 61, 5351–5361. Habicht K. S. and Canfield D. E. (2001) Isotope fractionation by sulfate-reducing natural populations and the isotopic composition of sulfide in marine sediments. Geology 29, 555–558. Habicht K. S., Gade M., Thamdrup B., Berg P. and Canfield D. E. (2002) Calibration of sulfate levels in the Archean ocean. Science 298, 2372–2374. Hay W. W., Migdisov A., Balukhovsky A. N., Wold C. N., Flo¨gel S. and So¨ding E. (2006) Evaporites and the salinity of the ocean during the Phanerozoic: implications for climate, ocean circulation and life. Palaeogeor. Palaeoclimatol. Palaeoecol. 240, 3– 46. He W. H., Shen S. Z., Feng Q. L. and Gu S. Z. (2005) A late Changhsingian (Late Permian) deepwater brachiopod fauna from the Talung Formation at the Dongpan section, southern Guangxi, South China. J. Paleontol. 79, 927–938. Hermann E., Hochuli P. A., Bucher H., Vigran J. O., Weissert H. and Bernasconi S. M. (2010) A close-up view of the Permian– Triassic boundary based on expanded organic carbon isotope records from Norway (Trøndelag and Finnmark Platform). Glob. Planet. Change 74, 156–167. Horacek M., Brandner R. and Abart R. (2007) Carbon isotope record of the P/T boundary and the Lower Triassic in the Southern Alps: evidence for rapid changes in storage of organic carbon. Palaeogeogr. Palaeoclimatol. Palaeoecol. 252, 347–354. Horita J., Zimmermann H. and Holland H. D. (2002) Chemical evolution of seawater during the Phanerozoic: implications from the record of marine evaporites. Geochim. Cosmochim. Acta 66, 3733–3756. Hurtgen M. T., Halverson G. P., Arthur M. A. and Hoffman P. F. (2006) Sulfur cycling in the aftermath of a 635-Ma snowball


H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113

glaciation: evidence for a syn-glacial sulfidic deep ocean. Earth Planet. Sci. Lett. 245, 551–570. Joachimski M. M., Lai X. L., Shen S. Z., Jiang H. S., Luo G. M., Chen B., Chen J. and Sun Y. D. (2012) Climate warming in the latest Permian and the Permian–Triassic mass extinction. Geology 40, 195–198. Kah L. C., Lyons T. W. and Frank T. D. (2004) Low marine sulphate and protracted oxygenation of the Proterozoic biosphere. Nature 431, 834–838. Kaiho K., Kajiwara Y., Chen Z. Q. and Gorjan P. (2006) A sulfur isotope event at the end of the Permian. Chem. Geol. 235, 33– 47. Kaiho K., Kajiwara Y., Nakano T., Miura Y., Kawahata H., Tazaki K., Ueshima M., Chen Z. Q. and Shi G. R. (2001) EndPermian catastrophe by a bolide impact: evidence of a gigantic release of sulfur from the mantle. Geology 29, 815–818. Kaiho K., Oba M., Fukuda Y., Ito K., Ariyoshi S., Gorjan P., Riu Y., Takahashi S., Chen Z. Q., Tong J. N. and Yamakita S. (2012) Changes in depth-transect redox conditions spanning the end-Permian mass extinction and their impact on the marine extinction: evidence from biomarkers and sulfur isotopes. Glob. Planet. Change 94–95, 20–32. Kampschulte A., Bruckschen P. and Strauss H. (2001) The sulphur isotopic composition of trace sulphates in Carboniferous brachiopods: implications for coeval seawater, correlation with other geochemical cycles and isotope stratigraphy. Chem. Geol. 175, 149–173. Kampschulte A. and Strauss H. (2004) The sulfur isotopic evolution of Phanerozoic seawater based on the analysis of structurally substituted sulfate in carbonates. Chem. Geol. 204, 255–286. Kleikemper J., Schroth M. H., Bernasconi S. M., Brunner B. and Zeyer J. (2004) Sulfur isotope fractionation during growth of sulfate-reducing bacteria on various carbon sources. Geochim. Cosmochim. Acta 68, 4891–4904. Korte C., Pande P., Kalia P., Kozur H. W., Joachimski M. M. and Oberha¨nsli H. (2010) Massive volcanism at the Permian– Triassic boundary and its impact on the isotopic composition of the ocean and atmosphere. J. Asian Earth Sci. 37, 293–311. Krull E. S., Lehrmann D. J., Druke D., Kessel B., Yu Y. Y. and Li R. (2004) Stable carbon isotope stratigraphy across the Permian–Triassic boundary in shallow marine carbonate platforms, Nanpanjiang Basin, south China. Palaeogeogr. Palaeoclimatol. Palaeoecol. 204, 297–315. Kump L. R., Pavlov A. and Arthur M. A. (2005) Massive release of hydrogen sulfide to the surface ocean and atmosphere during intervals of oceanic anoxia. Geology 33, 397–400. Lehrmann D. J. (1999) Early Triassic calcimicrobial mounds and biostromes of the Nanpanjiang basin, South China. Geology 27, 359–362. Lehrmann D. J., Payne J. L., Felix S. V., Dillett P. M., Wang H. M., Yu Y. Y. and Wei J. Y. (2003) Permian–Triassic boundary sections from shallow-marine carbonate platforms of the Nanpanjiang Basin, South China: implications for oceanic conditions associated with the end-Permian extinction and its aftermath. Palaios 18, 138–152. Lehrmann D. J., Ramezani J., Bowring S. A., Martin M. W., Montgomery P., Enos P., Payne J. L., Orchard M. J., Wang H. M. and Wei J. Y. (2006) Timing of recovery from the endPermian extinction: geochronologic and biostratigraphic constraints from south China. Geology 34, 1053–1056. Looy C. V., Brugman W. A., Dilcher D. L. and Visscher H. (1999) The delayed resurgence of equatorial forests after the Permian– Triassic ecological crisis. Proc. Nat. Acad. Sci. USA 96, 13857– 13862.

Lowenstein T. K., Timofeeff M. N., Kovalevych V. M. and Horita J. (2005) The major-ion composition of Permian seawater. Geochim. Cosmochim. Acta 69, 1701–1719. Luo G. M., Kump L. R., Wang Y. B., Tong J. N., Arthur M. A., Yang H., Huang J. H., Yin H. F. and Xie S. C. (2010) Isotopic evidence for an anomalously low oceanic sulfate concentration following end-Permian mass extinction. Earth Planet. Sci. Lett. 300, 101–111. Luo G. M., Ono S., Kump L. R., Huang J. H., Li C., Zhou L., Liu J. and Xie S. C. (in review) Return of Archean low oceanic sulfate levels during the earliest Mesoproterozoic. Earth Planet. Sci. Lett. Lyons T. W., Walter L. M., Gellatly A. M., Martini A. M. and Blake R. E. (2004) Sites of anomalous organic remineralization in the carbonate sediments of South Florida, USA: the sulfur cycle and carbonate-associated sulfate. In Sulfur Biogeochemistry–Past and Present (eds. J. P. Amend, K. J. Edwards, and T. W. Lyons). Geol. Soc. Am. Spec. Pap. 379, pp. 161–176. Magaritz M., Krishnamurthy R. V. and Holser W. T. (1992) Parallel trends in organic and inorganic carbon isotopes across the Permian/Triassic boundary. Am. J. Sci. 292, 727–739. Marenco P. J. (2007) Sulfur isotope geochemistry and the end Permian mass extinction. Ph. D. thesis, University of Southern California. p. 189. Marenco P. J., Corsetti F. A., Hammond D. E., Kaufman A. J. and Bottjer D. J. (2008a) Oxidation of pyrite during extraction of carbonate associated sulfate. Chem. Geol. 247, 124–132. Marenco P. J., Corsetti F. A., Kaufman A. J. and Bottjer D. J. (2008b) Environmental and diagenetic variations in carbonate associated sulfate: an investigation of CAS in the Lower Triassic of the western USA. Geochim. Cosmochim. Acta 72, 1570–1582. Millero F. J. (2005) Chemical Oceanography, third ed. CRC Press, Boca Raton, Florida, p. 536. Mundil R., Palfy J., Renne P. R. and Brack P. (2010) The Triassic time scale: New constraints and a review of geochronological data. In The Triassic Timescale (ed. S. G. Lucas). Geol. Soc. London Spec. Publ. 334, pp. 41–60. Newton R. J., Pevitt E. L., Wignall P. B. and Bottrell S. H. (2004) Large shifts in the isotopic composition of seawater sulphate across the Permo–Triassic boundary in northern Italy. Earth Planet. Sci. Lett. 218, 331–345. Ovtcharova M., Bucher H., Schaltegger U., Galfetti T., Brayard A. and Guex J. (2006) New Early to Middle Triassic U–Pb ages from South China: calibration with ammonoid biochronozones and implications for the timing of the Triassic biotic recovery. Earth Planet. Sci. Lett. 243, 463–475. Payne J. L., Lehrmann D. J., Wei J. Y., Orchard M. J., Schrag D. P. and Knoll A. H. (2004) Large perturbations of the carbon cycle during recovery from the end-Permian extinction. Science 305, 506–509. Payne J. L., Lehrmann D. J., Wei J. Y. and Knoll A. H. (2006) The pattern and timing of biotic recovery from the end-Permian extinction on the Great Bank of Guizhou, Guizhou Province, China. Palaios 21, 63–85. Paytan A., Kastner M., Campbell D. and Thiemens M. H. (1998) Sulfur isotopic composition of Cenozoic seawater sulfate. Science 282, 1459–1462. Riccardi A. L., Arthur M. A. and Kump L. R. (2006) Sulfur isotopic evidence for chemocline upward excursions during the end-Permian mass extinction. Geochim. Cosmochim. Acta 70, 5740–5752. Romano C., Goudemand N., Vennemann T. W., Ware D., Schneebeli-Hermann E., Hochuli P. A., Bru¨hwiler T., Brinkmann W. and Bucher H. (2012) Climatic and biotic upheavals

H. Song et al. / Geochimica et Cosmochimica Acta 128 (2014) 95–113 following the end-Permian mass extinction. Nat. Geosci. 6, 57– 60. Shen S. Z., Henderson C. M., Bowring S. A., Cao C. Q., Wang Y., Wang W., Zhang H., Zhang Y. C. and Mu L. (2010) Highresolution Lopingian (Late Permian) timescale of South China. Geol. J. 45, 122–134. Shen S. Z.21 coauthors (2011) Calibrating the end-Permian mass extinction. Science 334, 1367–1372. Shen W. J., Lin Y. T., Xu L., Li J. F., Wu Y. S. and Sun Y. G. (2007) Pyrite framboids in the Permian–Triassic boundary section at Meishan, China: evidence for dysoxic deposition. Palaeogeor. Palaeoclimatol. Palaeoecol. 253, 323–331. Shen Y. A., Buick R. and Canfield D. E. (2001) Isotopic evidence for microbial sulphate reduction in the early Archaean era. Nature 410, 77–81. Sim M. S., Bosak T. and Ono S. H. (2011) Large sulfur isotope fractionation does not require disproportionation. Science 333, 74–78. Song H. J., Wignall P. B., Chen Z. Q., Tong J. N., Bond D. P. G., Lai X. L., Zhao X. M., Jiang H. S., Yan C. B., Niu Z. J., Chen J., Yang H. and Wang Y. B. (2011) Recovery tempo and pattern of marine ecosystems after the end-Permian mass extinction. Geology 39, 739–742. Song H. J., Wignall P. B., Tong J. N., Bond D. P. G., Song H. Y., Lai X. L., Zhang K. X., Wang H. M. and Chen Y. L. (2012) Geochemical evidence from bio-apatite for multiple oceanic anoxic events during Permian–Triassic transition and the link with end-Permian extinction and recovery. Earth Planet. Sci. Lett. 353, 12–21. Song H. Y., Tong J. N., Algeo T. J., Horacek M., Qiu H. O., Song H. J., Tian L. and Chen Z. Q. (2013) Large vertical d13CDIC gradients in Early Triassic seas of the South China craton: implications for oceanographic changes related to Siberian Traps volcanism. Glob. Planet. Change 105, 7–20. Staudt W. J. and Schoonen M. A. A. (1995) Sulfate incorporation into sedimentary carbonates. In Geochemical Transformations of Sedimentary Sulfur (eds. M. Vairavamurthy and M. Schoonen). American Chemical Society, Washington, D.C. pp. 332– 345. Strauss H. (1997) The isotopic composition of sedimentary sulfur through time. Palaeogeor. Palaeoclimatol. Palaeoecol. 132, 97– 118. Sun Y. D., Joachimski M. M., Wignall P. B., Yan C. B., Chen Y. L., Jiang H. S., Wang L. N. and Lai X. L. (2012) Lethally hot temperatures during the early Triassic greenhouse. Science 388, 366–370. Takano B. (1985) Geochemical implications of sulphate in sedimentary carbonates. Chem. Geol. 49, 393–403. Tong J. N. and Yin H. F. (2002) The Lower Triassic of South China. J. Asian Earth Sci. 20, 803–815. Tong J. N., Zuo J. X. and Chen Z. Q. (2007) Early Triassic carbon isotope excursions from South China: proxies for devastation and restoration of marine ecosystems following the endPermian mass extinction. Geol. J. 42, 371–389. Tucker M. E. (1991) Sequence stratigraphy of carbonate-evaporite basins: models and application to the Upper Permian (Zechstein) of northeast England and adjoining North Sea. J. Geol. Soc. London 148, 1019–1036. Wang F. Y. and Chapman P. M. (1999) Biological implications of sulfide in sediment–a review focusing on sediment toxicity. Environ. Toxicol. Chem. 18, 2526–2532. Wang H. M., Wang X. L., Li R. X. and Wei J. Y. (2005a) Triassic conodont succession and stage subdivision of the Guandao


section, Bianyang, Luodian, Guizhou. Acta Palaeontol. Sinica 44, 611–626 (in Chinese with English abstract). Wang Y. B., Tong J. N., Wang J. S. and Zhou X. G. (2005b) Calcimicrobialite after end-Permian mass extinction in South China and its palaeoenvironmental significance. Chin. Sci. Bull. 50, 665–671. Ward P. D. (2006) Impact from the deep. Sci. Am. 295, 64–71. Werne J. P., Lyons T. W., Hollander D. J., Formolo M. J. and Sinninghe Damste´ J. S. (2003) Reduced sulfur in euxinic sediments of the Cariaco Basin: sulfur isotope constraints on organic sulfur formation. Chem. Geol. 195, 159–179. Wignall P. B. (2007) The end-Permian mass extinction—how bad did it get? Geobiology 5, 303–309. Wignall P. B. and Twitchett R. J. (2002) Extent, duration, and nature of the Permian–Triassic superanoxic event. In Catastrophic Events and Mass Extinctions: Impacts and Beyond (eds. C. Koeberl and K. G. MacLeod). Geol. Soc. Am. Spec. Pap. 356, pp. 395–413. Wilkinson B. H. and Algeo T. J. (1989) Sedimentary carbonate record of calcium–magnesium cycling. Am. J. Sci. 289, 1158– 1194. Wilkinson B. H., Owen R. M. and Carroll A. R. (1985) Submarine hydrothermal weathering, global eustasy, and carbonate polymorphism in Phanerozoic marine oolites. J. Sed. Pet. 55, 171– 183. Winguth C. and Winguth A. M. E. (2012) Simulating Permian– Triassic oceanic anoxia distribution: implications for species extinction and recovery. Geology 40, 127–130. Worden R. H., Smalley P. C. and Fallick A. E. (1997) Sulfur cycle in buried evaporites. Geology 25, 643–646. Wu H. C., Zhang S. H., Feng Q. L., Jiang G. Q., Li H. Y. and Yang T. S. (2012) Milankovitch and sub-Milankovitch cycles of the early Triassic Daye Formation, South China and their geochronological and paleoclimatic implications. Gondwana Res. 22, 748–759. Xie S. C., Pancost R. D., Huang J. H., Wignall P. B., Yu J. X., Tang X. Y., Chen L., Huang X. Y. and Lai X. L. (2007) Changes in the global carbon cycle occurred as two episodes during the Permian–Triassic crisis. Geology 35, 1083–1086. Yin H. F., Jiang H. S., Xia W. C., Feng Q. L., Zhang N. and Shen J. (2014) The end-Permian regression in South China and its implication on mass extinction. Earth-Sci. Rev.. http:// dx.doi.org/10.1016/j.earscirev.2013.06.003. Yin H. F., Zhang K. X., Tong J. N., Yang Z. Y. and Wu S. B. (2001) The Global Stratotype Section and Point (GSSP) of the Permian–Triassic boundary. Episodes 24, 102–114. Zaback D. A. and Pratt L. M. (1992) Isotopic composition and speciation of sulfur in the Miocene Monterey Formation: reevaluation of sulfur reactions during early diagenesis in marine environments. Geochim. Cosmochim. Acta 56, 763– 774. Zeebe R. E. and Wolf-Gladrow D. A. (2001) CO2 in Seawater: Equilibrium, Kinetics, Isotopes. Elsevier Oceanography Series, p. 346. Zhang K. X., Tong J. N., Shi G. R., Lai X. L., Yu J. X., He W. H., Peng Y. Q. and Jin Y. L. (2007) Early Triassic conodontpalynological biostratigraphy of the Meishan D Section in Changxing, Zhejiang Province, South China. Palaeogeogr. Palaeoclimatol. Palaeoecol. 252, 4–23. Associate editor: David Johnston