of 47
Journal Pre-proofs
A volatile sulfur sink aids in reconciling the sulfur isotope mass balance of
closed basin lakes
Antoine Crémière, Christopher J. Tino, Maxwell E. Pommer, Xingqian Cui,
Matthew Roychowdhury, Roger E. Summons, Alex Sessions, J. Fredrick
Sarg, Timothy W. Lyons, Jess F. Adkins
PII:
S0016-7037(24)00020-6
DOI:
https://doi.org/10.1016/j.gca.2024.01.008
Reference:
GCA 13290
To appear in:
Geochimica et Cosmochimica Acta
Received Date:
31 May 2023
Revised Date:
4 January 2024
Accepted Date:
8 January 2024
Please cite this article as: Cr
émiè
re, A., Tino, C.J., Pommer, M.E., Cui, X., Roychowdhury, M., Summons, R.E.,
Sessions, A., Fredrick Sarg, J., Lyons, T.W., Adkins, J.F., A volatile sulfur sink aids in reconciling the sulfur
isotope mass balance of closed basin lakes,
Geochimica et Cosmochimica Acta
(2024), doi:
https://doi.org/
10.1016/j.gca.2024.01.008
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. Please note that, during the production process, errors
may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
© 2024 Elsevier Ltd. All rights reserved.
1
A volatile sulfur sink aids in reconciling the sulfur isotope mass
balance of closed basin lakes
Antoine Crémière
1,*
, Christopher J. Tino
2
, Maxwell E. Pommer
3
, Xingqian Cui
4
, Matthew
Roychowdhury
1
, Roger E. Summons
5
, Alex Sessions
1
, J. Fredrick Sarg
3
, Timothy W. Lyons
2
, Jess F.
Adkins
1
1
Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena,
California, USA
2
Department of Earth and Planetary Sciences, University of California, Riverside, CA 92521, USA
3
Department of Geology and Geological Engineering, Colorado School of Mines, Golden, CO, USA
4
School of Oceanography, Shanghai Jiao Tong University, 200030 Shanghai, China
5
Department of Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology,
Cambridge, MA 02139, USA
*corresponding author:
cremiere@caltech.edu
2
1. Abstract
Sedimentary rocks from the early Eocene Green River Formation comprise the largest known
lacustrine oil shale deposits, contain remarkably well-preserved fossils, and provide a unique record
of climate evolution across the Early Eocene Climate Optimum, a period of high atmospheric CO
2
.
The depositional environment of these intermountain lakes spanned from relatively fresh and
fluvially influenced to expanded and stratified saline closed basin lakes under the influence of the
Laramide orogeny and alternating humid and arid climatic conditions. As the surface area of the lakes
expanded, alkalinity and salinity increased, with depositional cycles that linked to evaporative cycles
and marked by an increasing abundance of organic matter-rich shales (TOC > 10 wt. %).
Simultaneously, an intriguing 20 ‰ positive shift in the sulfur isotope composition of sulfide minerals
and organic sulfur is observed. Given that this trend cannot be simply explained by a change in the
source of sulfate delivered to the basin, the evolution of biogeochemical sulfur cycle and the balance
of fluxes in response to basin evolution remain unresolved. Here, we combine the sulfur isotope
compositions of pyrite, organic sulfur, and carbonate-associated sulfate and molecular proxies of
euxinia in samples from the Uinta basin's depocenter. We find that organic matter-rich sediments
reflect deposition in a stratified water column with enhanced burial of pyrite and sulfurized organic
matter, while organic-lean facies present evidence of, at least transiently, euxinic conditions reaching
the photic zone during arid conditions presumably because of evaporation. As the lake became both
saline-stratified and euxinic, we observe that
34
S values of all measured sulfur-bearing sedimentary
proxies increase and evolve along a 1:1 line, a trend independent of facies that we interpret as
reflecting a sulfate-limited system despite saline conditions. The isotopic mass balance of sulfur
fluxes implies the existence of a sink of sulfur depleted in
34
S that is spatially decoupled from burial in
the depocenter. Modeling sulfur biogeochemical processes in a saline stratified lake system allows us
to estimate that at least 50 % of the sulfur entering the lake could have been lost from the upper part
of the euxinic water column where the fractionation factor imparted by microbial sulfate reduction is
expressed. We propose that the overall isotopic enrichment of the system was caused by H
2
S
degassing during arid climate intervals, presumably enhanced by transient water column mixing
events. Further, episodic intrusion of euxinic bottom waters into the upper part of the water column
might have triggered mass fish and plankton mortality, consequently facilitating the formation of
these exceptionally fossiliferous and organic matter-rich rocks. Our study finds that volatile
outgassing may be an underappreciated mechanism for the sulfur mass balance of stratified
lacustrine systems.
3
Keywords:
Sulfur isotopes, Green River Formation, Saline Lake, Euxinia, Pyrite, Carbonate-associated
sulfate, Organic sulfur.
4
2. Introduction
The biogeochemical cycle of sulfur is central in the regulation of Earth’s surface redox conditions
on geological timescales and is intimately linked to the cycling of other key elements such as oxygen,
carbon, and iron (Berner and Canfield, 1989; Farquhar et al., 2000; Fike et al., 2015). While the
modern ocean is well oxygenated and relatively well mixed, euxinic conditions—defined by the
presence of hydrogen sulfide (H
2
S) in the water column—occurred in abundance during multiple
intervals of Earth’s history, often concurrent with warming climate, and were responsible for some of
the largest biological crises (Canfield, 1998; Poulton et al., 2004; Gill et al., 2011). The emergence of
euxinia is typically associated with (1) oxygen deficiency, sometimes induced by water column
density stratification hindering mixing of an upper, oxygenated water layer with deeper waters and
(2) enhanced organic matter accumulation stimulating microbial sulfate reduction (MSR). MSR is
associated with the metabolic oxidation of organic carbon using sulfate as a terminal electron
acceptor, reducing it into H
2
S, which can accumulate in stagnant bottom waters. Modern meromictic
lakes (i.e., those with a permanently stratified water column due to the presence of a strong salinity
and/or temperature gradient) are generally considered useful analogs for ancient ocean chemistry
because of potential similarities in term of biogeochemical processes. For instance, stratified lake
systems with low sulfate concentration (low mM to μM levels) provide a window into the ferruginous
Precambrian ocean (Crowe et al., 2008, 2014; Busigny et al., 2014; Swanner et al., 2020; Roland et
al., 2021), while lakes with a euxinic monimolimnion have been considered analogous to certain
conditions of the Proterozoic era (Canfield et al., 2010; Gomes and Hurtgen, 2015). Importantly,
understanding the response of the sulfur cycle to major redox changes in aquatic environments
requires a quantitative assessment of how biogeochemical processes can be preserved in
sedimentary rocks.
The stable sulfur isotopic compositions (
34
S) of sedimentary sulfur-bearing minerals and organic
compounds can constrain the rates and pathways of biological and chemical redox reactions. For
example, the difference in sulfur isotope composition between sulfate and sulfide minerals (
34
S
sulfate-
sulfide
) has been employed to infer the evolution of sulfate content in ancient ocean (Canfield and
Farquhar, 2009; Crowe et al., 2014; Algeo et al., 2015; Fakhraee et al., 2019). The
34
S of sulfate can
be recorded in carbonate-associated sulfate (CAS) or in sulfate evaporite minerals, with the former
being sensitive to diagenetic alteration and the latter requiring evaporitic conditions, making the
latter less common in the rock record (Marenco et al., 2008; Rennie and Turchyn, 2014; Fike et al.,
2015). Sulfide minerals, with the end-product being pyrite (FeS
2
), precipitate in reduced sediments
and possibly euxinic water columns when reactive iron consumes H
2
S (Berner, 1984; Raiswell and
Canfield, 1998; Suits and Wilkin, 1998). Due to the isotope effects imparted by MSR, the
34
S of
5
product sulfide is typically depleted in
34
S compared to reactant sulfate by as much as 70 ‰ (e.g.,
Wortmann et al., 2001; Canfield et al., 2010; Sim et al., 2011a). The MSR fractionation factor (
MSR
) is
thought to be controlled by the enzymatic machinery of sulfate reducers such that it is inversely
correlated with cell-specific sulfate reduction rates (Harrison and Thode, 1958; Rees, 1973; Chambers
et al., 1975; Habicht and Canfield, 1997; Brunner et al., 2005; Wing and Halevy, 2014). The efficiency
of the enzymatic machinery and thus the magnitude of
MSR
appears to differ between bacterial
strains (Bradley et al., 2016) and is affected by several environmental factors such as the quantity
and reactivity of organic substrates (Sim et al., 2011b; Leavitt et al., 2013), sulfate and sulfide
concentrations (Wing and Halevy, 2014), and nutrients (Sim et al., 2012). Thus, decoding how the
34
S value preserved in sulfide minerals can relate to environmental conditions, particularly given its
dependence on multiple features of depositional environments, presents a major challenge (Canfield
and Thamdrup, 1994; Fike et al., 2015; Pasquier et al., 2017; Lang et al., 2020). Although precipitation
of sulfide minerals is generally assumed to be the main sink of reduced sulfur in sedimentary basins,
a fraction of H
2
S and sulfur intermediate species such as polysulfides can react abiotically with
organic matter (OM) (Amrani, 2014). The relative timing of OM sulfurization versus pyrite formation,
as well as redox conditions during deposition, can be inferred by their
34
S relationship (Anderson
and Pratt, 1995; Shawar et al., 2018; Raven et al., 2019). The importance of organic sulfur (OS)
cycling in both ancient and modern oceans has recently gained attention in part due to its role in
promoting OM sequestration (Hülse et al., 2019; Fakhraee and Katsev, 2019; Raven et al., 2021;
Gomez-Saez et al., 2021). However, the importance of this process in euxinic lacustrine systems is
not well-constrained.
The Eocene Green River Formation in the Green River Basin (USA) is a classic paleo-soda lake
system. The formation includes a series of continental basins, specifically the Uinta, Piceance, and
Greater Green River basins, which emerged during the Laramide Orogeny (Dickinson et al., 1988;
Smith et al., 2008). These lakes formed during the early Eocene and span the Early Eocene Climate
Optimum, a period of high atmospheric pCO
2
and climate perturbation, which, in addition to
tectonism, had a major influence on the evolution of their depositional environment (Tänavsuu-
Milkeviciene et al., 2017; Birgenheier et al., 2020). In the aftermath of the Paleocene/Eocene
Thermal Maximum (Zachos et al., 2001), shales rich in organic matter (OM) in the Parachute Creek
Member are interpreted to have been deposited in an alkaline lake system with a redox-stratified
water column, according to the canonical depositional model (Bradley and Eugster, 1969;
Desborough, 1978; Boyer, 1982; Tuttle and Goldhaber, 1993;
Tänavsuu-Milkeviciene
and Sarg, 2012).
Despite evidence of strong saline-evaporitic conditions, as suggested by the presence of unusual,
soluble forms of Na-carbonates (Dyni, 1996; Lowenstein and Demicco, 2006; Demicco and
6
Lowenstein, 2019) and extensive deposition of carbonate minerals (e.g., Desborough, 1978), no
primary Ca-sulfate-bearing evaporite has been observed (Bradley, 1948; Tuttle and Goldhaber, 1993).
It has been hypothesized that euxinic conditions facilitated by a strong density gradient and active
MSR in the water column kept sulfate contents low and prevented the precipitation of sulfate
minerals at the Green River Formation (Tuttle and Goldhaber, 1993). However, the timing of when
these conditions emerged and whether sulfurization played a role in the deposition of related OM-
rich sediments remain unclear. Concurrent with an increase in the deposition frequency of OM-rich
shales is a remarkable ca. 30 ‰ positive
34
S shift in both pyrite and OS, which has previously been
attributed to an increase in the burial fluxes of
34
S depleted pyrite and OS (Tuttle and Goldhaber,
1993). This possibility implies, by mass balance, that the sulfate reservoir became isotopically
heavier, a hypothesis that remains to be tested. Taking advantage of modern analytical techniques
with increased sensitivity (Paris et al., 2013; Phillips et al., 2021), we revisit the sulfur isotope
composition of sulfide minerals and OS in both OM-rich and -poor facies and provide new constraints
on water column sulfate through analysis of CAS in sedimentary rocks deposited in the Uinta Lake
depocenter. We compare the information provided by these sulfur proxies with molecular
biomarkers diagnostic of photic zone euxinia and evaluate putative modern and ancient analogs
within the sedimentological context and basin evolution of the Green River Formation. Using a sulfur
mass balance model of a stratified lake, we further explore the geochemical and physicochemical
conditions under which the sulfur isotope enrichment could have occurred.
3. Geological context
The Green River Formation (Fig. 1) is one of the longest-lived (ca. 6 Myr; (Smith et al., 2008))
continuous continental climatic record of the early Eocene, a critical period marked by perturbation
of the carbon cycle and global warming because of large release of carbon into the ocean-
atmosphere system initiated at the end of the Paleocene (Zachos et al., 2008; McInerney and Wing,
2011). This evolution of early Eocene climate as well as tectonic adjustments during the Laramide
orogeny are considered the main drivers of stratigraphic change in the Green River Formation
(Johnson, 1985; Carroll and Bohacs, 1999; Carroll et al., 2006; Smith et al., 2008; Davis et al., 2008;
Birgenheier et al., 2020). The unit is also of economic importance as it contains one of the largest
known oil shale deposits (Dyni, 2006). Deposited in the eastern portion of the depocenter of the
Uinta basin (Fig. 1b), the strata from the Coyote Wash core record the long-term evolution (> 2 Myr;
Smith et al., 2008) from open to closed hydrologic conditions with increasing salinity during the latter
(Pitman, 1996; Davis et al., 2008, 2009). Below, we describe the main characteristics of the three lake
phases present in the Coyote Wash core (Fig. 2 and 3), in line with previous studies (e.g. Birgenheier
7
et al., 2020; Johnson, 1985; Tänavsuu
-
Milkeviciene and Sarg, 2012). That core does not, however,
contain the Lower Green River Formation, a freshwater lake phase containing the Carbonate Marker
Unit with an age estimated at ca. 54 Ma (Birgenheier et al., 2020).
Figure 1
. a. Early Eocene paleogeographic map of Northern America (after Blakey and Ranney, 2017). b.
Location of the studied Coyote Wash core in the eastern depocenter, where profundal facies are observed, of
Uinta Basin, Green River Formation (modified from Birgenheier et al., 2020; Roehler, 1992; Smith et al., 2008).
The dashed line represents the boundary between Utah and Colorado.
Phase I (Fig. 2a), which belongs to the Douglas Creek member, is interpreted as an overfilled basin
type following Carroll and Bohacs (1999). The corresponding sediments, deposited in littoral to
sublittoral environments (Ryder et al., 1976; Remy, 1992; Tänavsuu-Milkeviciene et al., 2017;
Birgenheier et al., 2020), consist mainly of fluvial-lacustrine facies, such as deltaic siliciclastic sediments
and coated-grain, micritic, and possibly microbial carbonates (Pommer et al., 2023). Consistent with
meteoric fluvial inflow, these carbonates exhibit low
13
C and
18
O values, < 2 and < -5 ‰ VPDB,
respectively, that increase collinearly toward the upper part of the section, indicative of a gradual
transition from a relatively fresh to brackish water environment (Pitman, 1996; Pommer et al., 2023).
While arid to semi-arid climate and episodic flooding during hyperthermal events characterize phase
I, a shift toward a more subtropical humid climate and more perennial fluvial systems occurred
throughout lake phase II as the Early Eocene Climate Optimum waned (Fig. 2b; Gall et al., 2017). Cyclic
deposition of fluctuating profundal facies during the corresponding net transgressive period of phase
II is characterized by interbedded OM-lean and OM-rich carbonate mudstones (Fig. 2c and d), which
represent fluctuating profundal and balanced lake regimes (e.g., Carroll and Bohacs, 1999). These
facies form a remarkable sequence of alternating, 1 to 10 m thick layers, including both OM-rich (R)
zones, with total organic carbon contents up to 40 % (Tuttle and Goldhaber, 1993; Fouch et al., 1994;
Cumming et al., 2012), and intervals of OM-lean (L) shale, with lateral continuities that can be
8
correlated across basins (Cashion and Donnell, 1972, 1974). Starting with the Mahogany zone (i.e., the
most OM-rich zone (R-7) containing Na-rich evaporite deposits), Phase III represents a more stable,
less saline lake phase dominated by the deposition of profundal OM-rich carbonate mudstones. Due
to their expansion, the Uinta and Piceance Creek basins merged into one large saline/alkaline lake (Fig.
1b). An overall more humid period due to little to no hyperthermal events is thought to characterize
this upper stratigraphic section (Birgenheier et al., 2020).
H
2
S
(
?
)
S
a
l
i
n
e
m
o
n
i
m
o
l
i
m
n
i
o
n
O
2
L
a
m
i
n
a
t
e
d
s
e
d
i
m
e
n
t
s
O
2
B
r
a
c
k
i
s
h
w
a
t
e
r
E
p
i
l
i
m
n
i
o
n
a
.
P
h
a
s
e
I
b
.
P
h
a
s
e
s
I
I
/
I
I
I
S
i
l
i
c
i
c
l
a
s
t
i
c
s
e
d
i
m
e
n
t
s
c
.
O
M
-
l
e
a
n
d
.
O
M
-
r
i
c
h
Figure 2
. Conceptual lake evolution and simplified depositional models for the Green River Formation (Johnson,
1985; Tuttle and Goldhaber, 1993; Tänavsuu-Milkeviciene et al., 2017; Birgenheier et al., 2020; Pommer et al.,
2023). a. A homogeneous brackish oxic water column during lake phase I, an overfilled basin type associated
with deposition of siliciclastic sediments. b. A stratified water column with the presence of anoxic and putative
euxinic conditions in a dense saline monimolimnion, with a globally higher lake level that characterized most of
lake phases II and III, fluctuating profundal and balanced filled basin type that led to the deposition of
laminated OM-rich and -lean sediments. c. Deposition of OM-lean samples under arid climate conditions and
limited run-off. d. Deposition of OM-rich samples under humid climate conditions and related enhanced
riverine inflow.
4. Materials and methods
4.1.
Sampling
The Coyote Wash-1 (CW) core was drilled during the early 1980s in Utah (40.02284°N,
109.31069°W), USA, by the United-State Geological Survey (USGS) (Scott and Pantea, 1982) and is
9
located in the eastern part of the lake depocenter of the Uinta basin (Fig. 1b). The core is about 506
m long and encompasses the Douglas Creek and part of the Parachute Creek members of the Green
River Formation. A total of 31 samples were hand-drilled at the USGS Core Research Center in
Denver, Colorado, USA, in the center of the half core, discarding the top ca. 1 cm to avoid external
contamination. Due to highly laminated nature of the sediments, sampling was done, as much as
possible, on representative OM-rich and -lean lithofacies at relatively regular intervals throughout
the core. About two-thirds of the samples were collected from the latter lithofacies, which is more
suitable for CAS analysis due to higher carbonate and lower OM and sulfide contents.
4.2.
Molecular organic proxies
Between 0.5 and 2.0 g of finely powdered samples were microwave-extracted with 20 mL of a
(90/10; v/v) dichloromethane-methanol solution. The temperature was steadily increased from 20 to
100°C over 8 minutes and then maintained at 100°C for 20 minutes. The solvent extract, namely
bitumen, was separated from kerogen by decantation, and both were dried at room temperature.
After asphaltene precipitation and removal of the elemental sulfur, the bitumen was further
separated into non-polar and polar fractions using a gravity fed silica gel column and eluting with
hexane:dichloromethane (80/20; v/v) and dichloromethane:methanol (80/20; v/v), respectively. The
non-polar fraction containing saturated and aromatic hydrocarbons was analyzed using an Agilent
7895B gas chromatograph coupled to an Agilent 7010A triple-quaddrupole mass spectrometer under
dynamic multi-reaction monitoring mode conditions as described previously (French et al., 2015).
4.3.
Composition and isotopic analyses of chromium reducible sulfur and sulfur
in kerogen
Following the solvent extraction, 100 to 300 mg of finely ground sample was subjected to a
chromium-reducible sulfur (CRS) extraction. CRS was partitioned using a standard chromium reduction
method, whereby a boiling solution of chromous chloride with hydrochloric acid under a stream of N
2
gas evolves H
2
S gas from the reduced inorganic phases of sulfur in each sample (Canfield et al., 1986).
The evolved H
2
S was trapped quantitatively by precipitation as silver sulfide (Ag
2
S), which was
quantified gravimetrically. This extracted sulfur is generally expected to primarily originate from pyrite,
although other sulfide minerals are sometimes present in these sediments (Tuttle and Goldhaber,
1993). Hereafter, we use the term 'pyrite' in a broader context to encompass sulfide minerals in
sediments. The residue of this extraction, which contained a mixture of kerogen and silicate minerals,
was washed twice with 18.2
Millipore Milli-Q water and dried at 50°C. Sulfur isotope
measurements were made targeting 2-4
μg
of sulfur from either Ag
2
S or the residue of the CRS and
solvent extractions (kerogen) weighed into tin capsules. The organic carbon and sulfur contents of
10
kerogen (TOC and OS) and sulfur isotope measurements of kerogen and CRS were performed using a
Thermo Scientific elemental analyzer IsoLink combustion system coupled to a Delta V Plus Isotope
Ratio Mass Spectrometer (EA-IRMS). Sulfur isotope compositions are reported in conventional
units
relative to the Vienna Canyon Diablo Troilite (VCDT) reference, corrected for sulfur content in tin
capsules (Phillips et al., 2021), with an estimated
34
S value of ca. 5 ‰. The data are calibrated against
international reference materials (IAEA-S1;
34
S = -0.35 ± 0.11 ‰, IAEA-S2;
34
S = 21.47 ± 0.11 ‰, IAEA-
S3;
34
S = -31.35 ± 0.07 ‰). All samples were run at least in duplicate. The analytical reproducibility for
34
S values is better than ± 0.26‰ for CRS and ± 0.39 ‰ for OS. Total carbonate content (CaCO
3
in wt.
%) was determined by measuring the weight loss during decarbonation. This measurement was
combined with carbon and sulfur contents as measured by IRMS to calculate TOC, OS, and CRS
contents as wt. % of dry sediments. Total sulfur content (TS) was calculated as the sum of CRS, OS, and
CAS contents (see below).
4.4.
Carbonate-associated sulfate
Due to the possible presence of elemental sulfur in our samples (Sassen and Chinn, 1989), and to
reduce the amount of bitumen that might interfere with the column chemistry employed for
isolating CAS, we used an aliquot of the powders extracted for bitumen analysis (see section 4.2) as a
starting material for CAS. To remove water-soluble sulfate, which is distinct from structurally bound
sulfate in carbonate minerals, powders as well as procedural blanks were immersed in a 10 % (w/w)
NaCl solution, sonicated overnight, and subsequently resuspended three times with 18.2
water
in microcentrifuge tubes. This entire step was repeated twice, which typically avoids the potential
contamination of water-soluble sulfate (Tostevin et al., 2017; Edwards et al., 2019). Weighed dried
powders were then transferred into cleaned microcentrifuge tubes and dissolved in 0.25 N Seastar
(C*) HCl within 1 to 2 hours, dried down, and taken up in 0.01 % (v/v) C* HCl. The dissolution was
conducted with an excess of carbonate relative to the amount of acid to target, as much as possible,
the dissolution of primary to early diagenetic carbonate phases such as calcite and aragonite while
minimizing the dissolution of later diagenetic carbonates, such as dolomite and siderite (Pommer et
al., 2023). Sulfate from each sample was then isolated on an anion exchange column containing one
mL of Bio-Rad AG1-X8 resin for 5-10 mg of carbonate. If the presence of iron carbonates was
suspected based on the visual observation of yellowish-reddish residue, the samples and procedural
blanks were filtered through 0.22
μm
precleaned nylon filters prior to column loading. Following the
procedures in Paris et al. (2014), the resin was preconditioned with 2 × resin volume 10 % (v/v)
reagent grade HNO
3
, 2 × resin volume 33 % (v/v) reagent grade HCl, and 2 × resin volume 0.5 % C*
HCl. Following 3 × resin volume rinses of the column with 18.2
water, retained sulfate was eluted
with 3 × 2 resin volume 0.45 N C* HNO
3
. Samples were then dried down on a hot plate in a PicoTrace
11
hood. Quality control was maintained by running seawater and deep-sea coral consistency standards,
as well as 5 to 9 blanks with each batch of columns (20-30 in each batch).
To quantify CAS abundances, purified samples were re-dissolved in 18.2
water, and a 10 %
(v/v) aliquot was removed for concentration measurements. Sulfate concentrations were measured
on a Dionex ICS-3000 ion chromatography system equipped with a 2 mm AS4A-SC column using a 1.8
mM Na
2
CO
3
/ 1.7 mM NaHCO
3
eluent. Based on the measured sulfate concentrations, samples were
dried down, re-dissolved a final time in 5 % (v/v) C* HNO
3
, and diluted in 2 mL autosampler vials to
attain a sulfate concentration matching that of a Na
2
SO
4
bracketing standard (either 15
μM
or 20
μM).
Sulfur isotope ratios were measured on a Neptune Plus multi-collector inductively coupled
plasma mass spectrometer (MC-ICP-MS) equipped with a CETAC Aridus II desolvating nebulizer
system at Caltech using the method of Paris et al. (2013). Samples were processed in two batches.
First, all samples were dissolved (3-21 mg of dissolved carbonates), but about half the samples did
not yield enough sulfate for reliable
34
S/
32
S measurements. The second batch targeting low CAS
content samples was processed via larger amounts of dissolved carbonate (40 to 100 mg).
Accuracy was monitored by including seawater and deep-sea coral consistency standards in each
Neptune run. The average
34
S value was 21.10 ± 0.05 ‰ (n = 3, ranging from 20 to 40 nmol) and
21.4 ± 0.5 ‰ (n = 4, ranging from 4 to 14 nmol) for seawater standard analyzed in batches 1 and 2,
respectively, and 22.4 ± 0.15 ‰ (n = 3) for the deep-sea coral standard (batch 1) and 1.0 ± 1.5 ‰ (n =
14) for all procedural blanks. The average procedural blank size was 0.56 ± 0.03 (n = 5) nmol of sulfur
in batch 1. For batch 2, larger blank content was observed due to the larger volume of resin and
reagents used. Specifically, column blank size was 2.4 ± 0.06 (n = 5) nmol, whereas total procedural
blanks increased with the increasing amount of HCl C* used to dissolved carbonate. The blank almost
doubled for the maximum amount of HCl C* used (i.e., 4.6 ± 0.3 for 10 mL). As a result, each sample
was corrected for the volume of HCl C* used in addition to the column blanks. All listed uncertainties
are
standard errors representing a combination of internal and intermediate errors during the
same analytical session, including drift correction via intermediate samples
(John and Adkins, 2010)
and propagated uncertainties resulting from subtracting the procedural blank isotopic composition
and amount.
5. Results
The geochemical data, which include the concentration and isotopic composition of sulfide
minerals, OS, and CAS along with molecular proxies, are summarized in Table 1. The TOC and TS
values from our samples span a wide range, from 0.3 to 41.6 wt. % and from 8 ppm to 1.77 wt. %,
respectively, and present coherent relationships when classified by lithofacies. For example, OM-rich
12
facies have elevated TOC and TS values, typically above ca. 9 and 0.7 wt. %, respectively. Similarly,
both pyrite and OS range widely from extremely low content in OM-lean samples (i.e., close to or
below detection limit for pyrite) to maxima of 1.5 and 0.96 wt. % in OM-rich samples, respectively.
The content of OS is more abundant than pyrite in some of the OM-rich facies. Most of the paired
34
S values of pyrite and OS are close (i.e., < 6 ‰ different) and increase with decreasing depth in the
core from ca. 10 to 40 ‰, reaching a maximum at the Mahogany bed (R7) and decreasing slightly in
the overlying section. This trend is remarkably independent of the lithofacies and consistent with a
previous study from the Green River Formation (Tuttle and Goldhaber, 1993). The
34
S values for CAS
show a similar stratigraphic trend. An exception is the OM-lean samples in lake phase I, which
present significantly higher
34
S values when compared to coexisting pyrite and OS. Observed CAS
concentrations also display a wide range, from 8 to 11,294 ppm, with higher content found in
association with high TS in OM-rich samples.
Key biomarker proxies are presented as ratios for hydrocarbons that are diagnostic for biota
and/or environmental conditions. The proportion of C
29
sterane isomers (C
29
/
C
27-29
steranes), which
is a proxy for the proportion of green algae versus other types, ranges between 0.26 and 0.56, with
significant differences (
p
< 0.001) between organic-lean and organic matter-rich layers. The ratio of
chlorobactan
e/β-caro
tane, indicative of the photic zone euxinia intensity, spans a large range
between < 0.01 and 0.22. The ratio is significantly higher in phase 2 compared to phase 3 (
p
= 0.05)
and is also often higher in organic lean relative to the organic matter-rich samples (
p
= 0.03).
Additionally, the ratio of (chlorobactane+okenane)/isorenieratane, which is a proxy for depth of the
euxinic interface within the photic zone, is significantly higher in organic-lean than organic matter-
rich layers (
p
= 0.001).
13
Figure 3.
Lithostratigraphy and geochemistry of the Coyote Wash core subdivided into 3 lake phases (
Birgenheier et al., 2020; Tänavsuu
-
Milkeviciene and Sarg, 2012). a. Fisher assay (Scott and Pantea, 1982)
overlaid with total organic carbon (TOC) of lithofacies classified samples by relative organic matter (OM)
content. b. Total sulfur (S) content. c. Sulfur isotope composition of organic sulfur (OS) and pyrite. d. Sulfur
isotope composition of carbonate-associated sulfate (CAS). e. Organic proxy (Chloro: chlorobactane,
normalized to
-carotane) for the anoxygenic phototrophic green sulfur bacteria
Chlorobiaceae
. Horizontal
grey bands show organic matter-rich zones, with the richest Mahogany zone (R7) standing at ca. 680 m.
14
Depth
CaCO
3
TOC
TS
OS/TOC
OS
34
S-OS
S-pyrite
34
S-
pyrite
CAS
34
S-CAS
Lake
phase
Sample
SN#
(m)
OM-
(wt. %)
(wt. %)
(wt. %)
(mol %)
(wt.
%)
(‰ VCDT)
(wt. %)
(‰ VCDT)
(ppm)
(‰ VCDT)
C
29
/
C
27-29
chloro.
/β-caro.
(chloro.+oke.)
/isoren.
CW-1
554.07
lean
59.6
5.1
0.04
0.24
0.033
26.0
0.006
-
15
21.6
0.34
0.000
1.7
CW-2
591.62
lean
57.9
2.6
0.11
0.32
0.022
34.2
0.090
29.7
30
25.3
0.37
0.004
10.7
CW-3
610.00
rich
28.9
13.6
1.25
0.77
0.278
34.6
0.969
38.1
156
35.3
0.56
0.010
2.4
CW-4
645.72
lean
47.7
2.2
0.02
0.17
0.010
29.4
0.008
-
22
-
0.40
0.001
0
CW-5
659.47
rich
29.7
21.5
1.38
0.71
0.408
36.0
0.954
35.9
2244
33.5
0.53
0.001
0.4
CW-6
667.45
lean
29.9
1.7
0.28
0.28
0.012
38.8
0.269
40.6
61
40.8
0.40
0.001
2.8
CW-7
671.50
rich
31.4
17.8
1.16
0.78
0.370
40.8
0.783
39.9
1038
38.4
0.51
0.001
1.3
CW-8
674.25
poor
46.5
7.0
0.22
0.34
0.064
34.5
0.159
32.2
78
29.8
0.54
0.001
0.6
CW-9
678.15
rich
7.9
40.7
1.65
0.89
0.965
39.2
0.681
-
1790
-
0.54
0.002
0.6
CW-10
681.29
rich
75.1
10.1
0.17
0.57
0.153
31.8
0.018
34.6
27
33.9
0.56
0.012
1.5
CW-11
681.56
rich
12.9
41.6
2.34
0.7
0.777
32.0
1.516
29.8
11294
29.2
0.54
0.017
1.0
CW-12
691.96
lean
47.2
4.9
0.15
0.42
0.055
42.9
0.090
42.3
136
37.3
0.53
0.079
11.0
CW-13
703.91
rich
15.3
27.3
1.18
0.85
0.618
37.1
0.556
37.8
1403
36.7
0.54
0.001
0.4
III
CW-14
705.06
lean
70.0
1.4
0.01
0.28
0.010
38.8
0.002
-
42
29.0
0.36
0.002
0.6
CW-15
724.66
lean
52.0
1.4
0.03
0.2
0.007
37.1
0.018
41.5
48
-
0.45
0.010
8
CW-16
739.29
lean
66.9
1.7
0.19
0.36
0.017
23.0
0.171
29.6
260
29.5
0.54
0.098
21.6
CW-17
762.40
lean
54.8
1.2
0.01
0.12
0.004
35.1
0.003
-
42
-
0.38
0.007
12
CW-18
773.52
lean
56.3
1.7
0.02
0.14
0.006
29.5
0.013
26.5
28
-
0.51
0.221
10
CW-19
783.79
lean
58.8
2.2
0.01
0.13
0.008
28.9
0.004
-
42
-
0.43
0.003
9
CW-20
795.22
lean
48.5
8.3
0.11
0.46
0.102
16.0
0.005
-
41
14.6
0.48
0.087
27.6
CW-21
809.52
lean
70.1
1.1
0.03
0.19
0.006
7.6
0.022
20.5
186
18.8
0.45
0.001
4.1
CW-22
825.40
rich
29.8
15.4
0.87
0.89
0.365
30.7
0.418
30.6
9103
29.6
0.51
0.011
1.5
CW-23
857.98
lean
39.1
4.1
0.09
0.24
0.026
19.6
0.065
22.9
17
21.0
0.31
0.084
5.4
II
CW-24
865.11
rich
30.8
20.3
1.06
0.47
0.255
25.7
0.792
20.5
1320
20.8
0.53
0.001
0.7
15
CW-25
895.96
lean
80.0
2.0
0.02
0.19
0.010
18.8
0.007
-
37
13.3
0.46
0.144
13.2
CW-26
910.89
rich
22.8
9.1
0.15
0.58
0.141
20.3
0.007
21.1
26
19.9
0.42
0.138
3.7
CW-27
929.67
lean
88.5
0.3
0.00
0.12
0.001
14.5
-
-
8
26.2
0.51
0.033
-
CW-28
969.29
lean
17.2
0.8
0.03
0.04
0.001
-
0.030
13.1
15.1
29.6
0.33
0.021
3.6
CW-29
1026.57
rich
17.5
11.6
1.24
0.48
0.148
14.2
1.043
9.1
8732
9.2
0.45
0.014
0.3
CW-30
1043.94
lean
95.7
1.1
0.04
0.57
0.017
25.1
0.017
19.1
224
41.0
0.35
0.013
0.5
I
CW-31
985.42
lean
87.1
0.6
0.06
0.23
0.004
17.5
0.045
17.9
201
26.3
0.26
0.023
0.4
Table 1.
Inorganic and organic geochemical results of the Coyote Wash core.
Abbreviations: C
29
= C
29
steranes,
C
27-29
=sum of C
27
, C
28
and C
29
steranes, chloro. = chlorobactane,
-carot.=
-carotane,
oke. = okenane, and iso.= isorenieratane.
16
6. Discussion
Reconstructing the redox state of a water column, whether in marine or lacustrine systems, relies
on our ability to interpret geochemical proxies preserved in sediments. In the first part of this
discussion, we focus on the remarkable isotopic relationships among the different measured sulfur
proxies, all of which become progressively heavier along a 1:1 line during lake phase II. By comparing
these results with potential analogs and combining them with organic biomarkers indicative of photic
zone euxinia, we infer the geochemical structure and evolution of Uinta Lake’s water column. In the
second part, we build a sulfur mass balance model to appraise the sulfur budget of the Uinta Lake
and discuss the mechanism responsible for this secular isotopic enrichment observed in both sulfide
and sulfate proxies. Over the course of this discussion, we refer to an idealized stratified water
column with two layers, as depicted in Fig. 2b, where the epilimnion represents an uppermost mixed
oxic water layer overlaying the monimolimnion, an anoxic, possibly euxinic saline layer.
6.1.
Paleo-environmental reconstruction of the Uinta Lake
6.1.1. Evaluation of CAS data and implications for change in redox state of the water
column
The mineralogical composition of carbonates from the Green River Formation is remarkably
diverse and encompasses variable mixtures of Ca-Mg-Fe-Na-bearing carbonate minerals of both
autochthonous and diagenetic origins (Milton and Fahey, 1960; Smith and Robb, 1966; Smith and
Milton, 1966; Milton, 1971; Desborough and Pitman, 1974; Desborough, 1978; Cole and Picard, 1978;
Mason and Surdam, 1992; Boak and Poole, 2015; Pommer et al., 2023). Indeed, a detailed
petrographic characterization of the samples studied here shows that they include a wide range of
carbonate minerals interpreted to have precipitated in the upper (e.g., low-Mg calcite and aragonite)
to deeper water column and diagenetic environments (e.g., dolomite and ankerite) where sulfate is
likely to be exhausted (Pommer et al., 2023). However, the microcrystalline nature of these different
carbonate components precludes their physical separation by micro/hand-drilling, a sampling
approach that could otherwise match up with the high sensitivity of CAS analysis by MC-ICP-MS (Paris
et al., 2014; Present et al., 2015; Johnson et al., 2020). An additional challenge for constraining the
origin of our CAS signal is associated with the oxidation of sulfur-bearing minerals, and possibly OS,
which can lead to the incorporation of sulfate in carbonates with an isotopic composition like that of
the oxidized sulfur (Marenco et al., 2008; Mazumdar et al., 2008; Wotte et al., 2012; Rennie and
Turchyn, 2014). Previous studies have discussed the potential for oxidation of these sulfur-bearing
17
compounds syn- and post-depositionally, as well as during laboratory sample processing (Marenco et
al., 2008; Mazumdar et al., 2008; Present et al., 2015; Johnson et al., 2021). Given that the Coyote
Wash core has been stored for a few decades, making them vulnerable to oxidation, and that we
used a protocol with rapid (< 2 h) carbonate dissolution and carbonate in excess to minimize pyrite
oxidation, we assume that most of the oxidation signal discussed below was generated prior to
sample processing. We show that independent of the degree of pyrite oxidation, the CAS signal in
lake phases II and III always closely follows the isotopic enrichment measured in pyrite and OS. We
interpret this key feature as indicative of the isotopic evolution of lake sulfate, wherein significant
sulfate consumption resulted in the formation of sulfide minerals with a similar isotopic composition
to that of sulfate.
+
2
0
o
f
f
s
e
t
1
:
1
l
i
n
e
-
5
o
f
f
s
e
t
L
a
k
e
p
h
a
s
e
I
I
I
I
I
I
δ
3
4
S
-
C
A
S
(
,
V
C
D
T
)
1
0
2
0
3
0
4
0
5
0
δ
3
4
S
-
p
y
r
i
t
e
(
,
V
C
D
T
)
1
0
2
0
3
0
4
0
O
M
-
l
e
a
n
O
M
-
r
i
c
h
<
1
0
-
3
1
0
-
2
1
0
-
1
>
1
S
-
p
y
r
i
t
e
(
w
t
.
%
)
δ
3
4
S
-
C
A
S
(
,
V
C
D
T
)
1
0
2
0
3
0
4
0
1
/
C
A
S
(
p
p
m
-
1
,
l
o
g
1
0
s
c
a
l
e
)
0
.
0
0
0
1
0
.
0
0
1
0
.
0
1
0
.
1
P
a
n
e
l
d
.
T
S
(
w
t
.
%
)
1
0
3
1
0
2
1
0
1
1
C
A
S
(
p
p
m
)
1
0
1
1
0
2
1
0
3
1
0
4
T
S
(
w
t
.
%
)
0
0
.
1
0
.
2
0
.
3
C
A
S
(
p
p
m
)
0
1
0
0
2
0
0
3
0
0
a
.
c
.
d
.
b
.
18
Figure 4.
a.
Sulfur isotope composition of carbonate-associated sulfate (CAS) against the logarithm of inverse
CAS concentration (ppm
-1
reported in g of dissolved carbonate per μg of extracted sulfate). Symbol shapes
indicate lithofacies classification and are colored by sulfur content as pyrite (S-pyrite reported in weight
percent of sulfur extracted per bulk sediment), with one data point below the detection limit in grey. b. Sulfur
isotope composition of carbonate-associated sulfate (CAS) against sulfur isotope composition of pyrite, colored
by lake phase. c. TS content against CAS concentration on a log. scale and (d.) zoomed in on OM-lean samples,
with CAS concentration plotted on a linear scale. TS is calculated as the sum of S-pyrite, OS and CAS. For panels
b, c, and d, symbols are colored according to lake phase (see legend in panel b).
Sulfide minerals produced as a result of MSR are typically depleted in
34
S as compared to
contemporaneous CAS, and thus their oxidation leads to a recognizable trend of increasing CAS
content with decreasing
34
S values (Marenco et al., 2008; Loyd et al., 2012; Johnson et al., 2021).
However, in the case of sulfide minerals enriched in
34
S, such as those measured here, disentangling
the effects of oxidation over other processes that can affect CAS content and isotopic composition is
difficult. The
34
S-CAS vs. 1/CAS concentration cross plot (Fig. 4a) has a roughly inverted V-shape with
a remarkably large range (over three orders of magnitude) for CAS concentrations. High (> 1000
ppm) CAS concentrations are only found in OM-rich samples (Fig. 4a), where pyrite contents are
typically above 0.4 wt. % S. The corresponding CAS
34
S values are close to those of coexisting sulfide
minerals (Fig. 4b), a trend that is consistent with a significant amount of sulfate being inherited from
the oxidation of sulfide minerals and/or OS overprinting the primary CAS signal. Conversely, most of
the OM-lean samples, with pyrite concentrations below 0.4 wt. % S and lower CAS concentrations
ranging from 8 to 260 ppm, appear to be less prone to the effects of sulfide mineral oxidation. For
this subset of samples, there is not a clear trend of increasing CAS concentration with increasing TS
content (Fig. 4d), which should be expected for pyrite oxidation. In both OM-rich and OM-lean
samples, there is a remarkable consistency between
34
S-CAS and
34
S-pyrite (Fig. 4b), including at
both high and low TS contents, implying that pyrite oxidation may not be the solely governing
mechanism of the signals for lake phases II and III.
The OM-lean samples from the earlier lake phase I present a marked change from the linear CAS-
sulfide isotope relationship in lake phases II and III (Fig. 4b). Those data show higher
34
S CAS values,
up to 22 ‰ different, with respect to the coexisting sulfide minerals, whereas CAS contents from the
deeper lake at that time are close to or slightly
34
S-depleted relative to pyrite. In an oxic water
column like lake phase I, MSR produces sulfide depleted in
34
S, in turn leaving isotopically enriched
residual porewater sulfate that can be incorporated as CAS during the precipitation of diagenetic
carbonates (Loyd et al., 2012; Rennie and Turchyn, 2014; Present et al., 2019; Crémière et al., 2020).
19
This mechanism explains the isotopic pattern from phase I and is further supported by the isotopic
enrichment of CAS relative to riverine inputs that had an estimated maximum
34
S value of ca. 20 ‰
(Tuttle and Goldhaber, 1993). In contrast, the apparent absence of a MSR imprint on CAS in the
overlying lake phases suggests a change in diagenetic regime. We propose that the emergence of
salinity-driven water column stratification at the end of lake phase I might have been responsible for
a collapse in MSR-induced diagenetic carbonate precipitation, possibly due to a reduction in the
diffusive flux of sulfate to the sediments. The emergence of stratification and anoxia is supported by
the deposition of laminated OM-rich mudstones in lake phases II and III (e.g., Tänavsuu-Milkeviciene
et al., 2017; Pommer et al., 2023). Low sulfate content is indirectly suggested by high abundances of
3
-methylhopane depleted in
13
C, a biomarker signature that is characteristic of aerobic
methanotrophic communities thriving under high methane flux (Collister et al., 1992; Ruble et al.,
1994; Farrimond et al., 2004; French et al., 2020), given that methanogenesis is classically thought to
be outcompeted by MSR in the presence of sulfate (Lovley and Klug, 1986). The development of
euxinic conditions in a stagnant monimolimnion would result in an expansion of the locus of MSR to
include both the sediments and the water column. Under strong water column stratification, sulfate-
limiting conditions may emerge in the monimolimnion. Since the incorporation of sulfate into
carbonate as CAS necessitates the presence of sulfate, such conditions would enable the
preservation of a CAS signal, presumably acquired through pelagic carbonate precipitation in the
epilimnion where sulfate is present.
Such a line of reasoning has major implications for the stratigraphic coevolution of the
34
S
values of CAS and pyrite in phases II and III. It implies that the isotopic enrichment of both oxidized
and reduced sulfur proxies most likely reflects the long-term isotopic evolution of sulfate and sulfide
in a stratified lake. This isotopic enrichment cannot be simply explained by a change in catchment
weathering, given that the aforementioned 20 ‰
34
S value of Jurassic evaporites represents the
heaviest known source of sulfate that could have been delivered to the lake (Tuttle and Goldhaber,
1993). Moreover,
34
S values of ca. 40 ‰ are rare in the evaporite record of the Phanerozoic,
observed only in Cambrian deposits (Present et al., 2020), which are not situated in reasonable
geographical proximity to the Green River Formation. If so, the isotopic enrichment of the Uinta Lake
reflecting the long-term isotopic evolution requires a sink of sulfur depleted in
34
S to fulfill isotopic
mass balance. A well-constrained example of this concept can be found in the sulfur isotope mass
balance of the modern ocean. The main source of sulfate to the modern ocean is riverine inputs,
which have a global average
34
S of 4.8 ± 4.9 ‰ (Burke et al., 2018). The ocean is comparatively
enriched in
34
S, showing a value of 21.2 ‰. The burial flux of pyrite sourced from
34
S-depleted H
2
S
20
produced by MSR in sediments is the main sink responsible for this isotopic enrichment of marine
sulfate. Similarly, burial of sulfide minerals and OS has been proposed as a mechanism by which
residual sulfate became isotopically enriched in the Green River Formation (Tuttle and Goldhaber,
1993). However, our results indicate that sulfate is consistently carrying the same or even slightly less
enriched (< 5 ‰, Fig. 4b) isotopic signature as co-occurring sulfide. The burial flux of pyrite and OS
predominates during OM-rich facies deposition, as evidenced by their high TS content (Fig. 3),
implying that this sink of
34
S-depleted sulfur cannot be the lake depocenter burial flux (i.e.,
associated with profundal facies) that leads to system-wide
34
S enrichment. A detailed evaluation of
sulfur cycling in other lacustrine systems may provide context, support our interpretation based on
proxies and help us resolve this sulfur mass balance question.
6.1.2. Searching for potential analogs
To better contextualize the inferred, unusual geochemical characteristics of Lake Uinta’s water
column, we have determined whether other modern lake systems present similar
34
S enrichment by
compiling sulfur isotope compositions of sulfide minerals from the literature (Fig. 5a and Table S1).
Overall, there is large variability in pyrite
34
S values, ranging from -38 to 38 ‰. This relationship
likely reflects a combination of the origin of sulfate in different basins as well as the varying nature of
biogeochemical sulfur cycling and depositional conditions. Importantly, only one modern system, the
Dziani Lake on an island off the coast of southeast Africa, has pyrite
34
S values over 22 ‰. This
hypersaline-alkaline and strongly stratified crater lake (Jovovic et al., 2020; Cadeau et al., 2022) is
typified by a distinct cluster of high pyrite
34
S values, between 30 and 38 ‰, similar to those of the
Green River Formation. A second example with similar characteristics is the Ace Lake in Antarctica, a
meromictic lake with reported
34
S of H
2
S from the monimolimnion as high as 41 ‰ (Burton and
Barker, 1979), reaching the initial isotopic composition of the water column sulfate (Fig. 5b). Ancient
saline lake systems such as the early Jurassic Towaco Formation (Stüeken et al., 2019) or the
Paleoproterozoic Barney Creek Formation (Shen et al., 2002; Mukherjee et al., 2019), the latter of
which was recently reinterpreted as a saline lacustrine system based on biomarker assemblage
similarities with the Green River Formation (French et al., 2020), also present isotopically enriched
pyrite with
34
S values as high as 54 ‰ and 42 ‰, respectively. All these putative modern and
ancient analogs represent bodies of water with high concentrations of salts that resulted in water
column stratification and the subsequent development of euxinic conditions in the monimolimnion.
21
G
r
e
e
n
R
i
v
e
r
f
m
M
o
d
e
r
n
l
a
k
e
s
A
v
e
r
a
g
e
±
s
t
d
e
v
M
e
r
o
m
i
c
t
i
c
l
a
k
e
s
D
z
a
n
a
l
a
k
e
F
r
e
q
u
e
n
c
y
δ
3
4
S
p
y
r
i
t
e
(
,
V
C
D
T
)
4
0
2
0
0
2
0
4
0
6
0
S
O
2
-
4
H
2
S
D
e
p
t
h
(
m
)
0
1
0
2
0
C
o
n
c
.
(
m
M
)
0
2
4
6
δ
3
4
S
(
,
V
C
D
T
)
0
5
0
a
.
b
.
Figure 5
. a. Comparison of compiled sulfur isotopic compositions of pyrite from various modern lake systems
(196 individual measurements from 14 different lakes, see Table S1 for details) with the early Eocene Green
River Formation (Tuttle and Goldhaber, 1993 and this study). The histogram is overlapped by the average pyrite
34
S value (
1 s.d.) of each lake, shown as circles (identified meromictic systems in gray), and ordered from low
to high
34
S values. b. Water-column concentration (conc.) and isotopic profiles of lake Ace (data from Burton
and Barker, 1979). Note the dotted blue line depicting a
34
S value of 41 ‰.
While euxinic conditions and water column stratification appear to be a prerequisite for isotopic
enrichment of sulfide minerals such as that observed in the Green River Formation, it is important to
note that many examples from modern saline meromictic lakes argue otherwise (Fig. 5a). For
instance, both Mahoney Lake and Owens Lake have sulfide minerals that are significantly depleted in
34
S with respect to sulfate, with
34
S values lower than -15 ‰ (Ryu et al., 2006; Gilhooly et al., 2016).
The disparities in the isotopic characteristics of modern saline lakes likely result from variations in the
extent to which the sulfate reservoir in the monimolimnion is reduced by MSR, specifically through a
Rayleigh distillation process. Indeed, in modern euxinic lacustrine systems, the concentration of
sulfate prior to microbial consumption appears to be a key parameter controlling the difference in
the isotopic composition of sulfide minerals relative to sulfate (Gomes and Hurtgen, 2015). Under
non-limiting sulfate conditions observed generally in lakes with high (> few mM) sulfate
concentration, the kinetic isotope fractionation imparted by microbial sulfate reduction (
MSR
) results
in an apparent isotopic difference between sulfate and sulfide higher than 20 ‰. In contrast, the
isotopic composition of sulfide tends to reach the original isotopic composition of sulfate under
limiting sulfate conditions (Gomes and Hurtgen, 2013), in a way comparable to what is shown in Fig.
5b for Ace Lake in Antarctica. Therefore, to produce sulfide minerals with isotopic compositions close
22
to initial sulfate, we posit that the monimolimnion and/or the sediments of the Uinta Lake
experienced extended intervals of sulfate limitation.
6.1.3. Evidence for strong euxinic conditions
There are lines of direct and indirect evidence suggesting the presence of euxinic conditions in
the Uinta Lake. Besides the general absence of sulfate evaporite minerals, another inference of
euxinic conditions relies on the correlation between paired
34
S of pyrite and OS, which has a slope
close to 1 (Fig. 6a). This trend is independent of facies and remarkably distinct from typical marine
sediments, even those deposited during Oceanic Anoxic Events (OAEs, Raven et al., 2019), where OS
is typically enriched in
34
S by 5 to 30 ‰ relative to coexisting pyrite (Anderson and Pratt, 1995). Two
non-mutually exclusive mechanisms have been invoked for the existence of this linear relationship in
the Green River Formation (Tuttle and Goldhaber, 1993). The first involves kinetic inhibition of sulfide
mineral formation due to elevated pH, and the second is based on the production of an isotopically
homogenous sulfide reservoir from which both OS and pyrite would have formed because of sulfate-
limiting conditions, as discussed above.
O
M
-
r
i
c
h
O
M
-
l
e
a
n
+
1
0
o
f
f
s
e
t
+
5
o
f
f
s
e
t
1
:
1
l
i
n
e
L
a
k
e
p
h
a
s
e
I
I
I
I
I
I
δ
3
4
S
-
O
S
(
,
V
C
D
T
)
1
0
2
0
3
0
4
0
δ
3
4
S
-
p
y
r
i
t
e
(
,
V
C
D
T
)
1
0
2
0
3
0
4
0
O
S
/
T
S
(
w
t
.
)
0
0
.
5
1
.
0
Δ
3
4
S
O
S
-
p
y
r
i
t
e
(
,
V
C
D
T
)
1
0
0
1
0
a
.
b
.
Figure 6
. a. Paired
34
S values of organic sulfur (OS) and pyrite colored by lake phase. Symbol shapes indicate
lithofacies classification colored by lake phase. Grey data are from Tuttle et al. (1993) from multiple basins. The
solid line represents the 1:1 line. The dotted lines represents a slope of 1 with an isotopic offset of +10 ‰ in
OS, which is typical for marine sediments (Anderson and Pratt, 1995) and of +5 ‰ predicted by OS synthesis
experiments (Amrani and Aizenshtat, 2004; Amrani et al., 2008). b. Weight fraction of OS (out of TS) as a
function of the isotopic offset between OS and pyrite (
34
S
OS-pyrite
).