Are Large Sulfur Isotope Variations Biosignatures in an Ancient, Impact-Induced Hydrothermal Mars Analog?

Discrepancies have emerged concerning the application of sulfur stable isotope ratios as a biosignature in impact crater paleolakes. The first in-situ δ 34 S data from Mars at Gale crater display a ~75‰ range that has been attributed to an abiotic mechanism. Yet biogeochemical studies of ancient environments on Earth generally interpret δ 34 S fractionations > 21‰ as indicative of a biological origin, and studies of δ 34 S at analog impact crater lakes on Earth have followed the same approach. We performed analyses (including δ 34 S, TOC wt%, and SEM imaging) on multiple lithologies from the Nördlinger Ries impact crater, focusing on hydrothermally altered impact breccias and associated sedimentary lake-fill sequences to determine whether the δ 34 S properties define a biosignature. The differences in δ 34 S between the host lithologies may have resulted from thermochemical sulfate reduction, microbial sulfate reduction, hydrothermal equilibrium fractionation, or any combination thereof. Despite abundant samples and instrumental precision currently exclusive to Earth-bound analyses, assertions of biogenicity from δ 34 S variations > 21‰ at the Miocene Ries impact crater are tenuous. This discourages the use of δ 34 S as a biosignature in similar environments without independent checks that


Introduction
Impact crater lakes may have housed the earliest (> 3.5Ga) habitable environments on Earth and Mars.In these locations, impact-induced hydrothermal activity would have provided heat in opposition to a faint young Sun and could have sourced compounds thought necessary for prebiotic synthesis of life's building blocks as well as the bioessential elements needed to sustain life (Abramov and Kring, 2005;Cockell et al., 2020;Farmer, 2000;Kacar et al., 2020;Kring, 2000;Osinski et al., 2013).These attributes make impact craters appealing targets for origin-oflife studies and the search for extinct life on Mars.In fact, impact craters have been the sites of multiple surface exploration missions, including the landing of the Perseverance rover in Jezero crater (Goudge et al., 2015;Grotzinger et al., 2015;Osinski et al., 2020;Mangold et al., 2021).
The collective understanding of nutrient cycling in ancient crater lakes has long relied on studies of geochemical proxies at well-preserved analog environments on Earth's surface (Hays et al., 2017;Léveillé, 2010;Osinski et al., 2020).One such proxy is the ratio of the two most abundant stable isotopes of sulfur (i.e., δ 34 S = [( 34 S/ 32 S)sample/( 34 S/ 32 S)Vienna-Canyon Diablo Troilite − 1] × 1000), which has the potential to illuminate both internal and surface processes on Earth and Mars (Nielsen et al., 1991;Canfield, 2001;Franz et al., 2019;King and McLennan, 2010).Sulfur has an expanded valence octet that allows for a redox continuum, with oxidation states ranging from -2 to +6.Stable S-isotope ratios can be fractionated during chemical exchanges across this range by kinetic and equilibrium reactions linked to both abiotic and biological processes (Canfield, 2001;Baune and Böttcher, 2010;Sakai, 1957;Seal, 2006;Surkov et al., 2012;Thode et al., 1961).
Mars can be considered a S-rich planet, evidently possessing a higher relative surface abundance of S than present-day Earth (Clark et al., 1982;Franz et al., 2019).Sulfur has been examined on the Martian surface in a multitude of forms, including  and pyrite [FeS2]); Ca/ Mg/Fe-sulfates, such as anhydrite [CaSO4]; and the Fe-hydroxidesulfate mineral jarosite [KFe3(SO4)2(OH)6] (Franz et al., 2019;Squyres et al., 2004;Vaniman et al., 2014).The reference base of the stable S-isotope system, Vienna-Canyon Diablo Troilite ([VCDT]: δ 34 S = 0‰, 34 S/ 32 S = 441.626±0.039x 10 -4 ; Ding et al., 2001), presents a practical comparative scale for both Mars and Earth, because it reflects the S isotope composition of our solar system's primordial terrestrial material (Hoefs, 2021;Sharp, 2017;Thode et al., 1961).Sulfur isotope ratios in Martian crater settings therefore have great potential to provide novel insights into past environmental conditions, including the presence of life.However, before this geochemical signature can be applied as a potential proxy for biological activity, it is imperative to develop a more thorough understanding of the range of possible S-isotope-fractionating processes that can operate in impact craters.
The complexity surrounding S fractionation mechanisms has provided the basis of multiple recent studies investigating S cycling in ancient impact craters (e.g., Kring et al., 2021;Schaefer et al., 2020Schaefer et al., , 2022)).Hydrothermal mineral assemblages from borehole samples of Chicxulub crater's peak rim display anhydrite and pyrite with Δ 34 Ssulfate-sulfide values of 25‰ to 54‰ (Kring et al., 2021).There, consistent visual evidence of pyrite framboids bolsters an interpretation of microbial sulfate reduction [MSR] as the primary fractionation mechanism (Kring et al., 2021).At Haughton crater in the Canadian High Arctic, a Δ 34 Ssulfate-sulfide of 44‰ was observed between gypsum and pyrite, along with a Δ 34 Ssulfate-sulfide of 71‰ between gypsum and marcasite [orthorhombic FeS2].Due to the magnitude of these differences, these data have also been interpreted as MSR (Parnell et al., 2010).However, there is a lack of reported sedimentary framboidal pyrite, the formation of which is often attributed to high rates of dissimilatory sulfate reduction associated with organic matter degradation (Passier et al., 1999;Rickard, 2021).This form notably differs from euhedral pyrite, which can form in sediments via lower microbial process rates, overgrowths on framboids (Passier et al., 1999), recrystallization (Liaghati et al., 2005), or a combination thereof.In situ analyses of S isotopes have been made by the Mars Science Laboratory's [MSL] Sample Analysis at Mars [SAM] instrument suite at Gale crater (Franz et al., 2017).The reported values span from -47 (± 14)‰ to 28 (± 7)‰, which is noted as similar to the terrestrial range observed on Earth (Franz et al., 2017;Hoefs, 2021).A multi-step abiotic fractionation mechanism was proposed, which included atmospheric oxidation and ultraviolet photolysis of volcanogenic S, followed by hydrothermal equilibrium fractionation (Franz et al., 2017).These contradictory interpretations-specifically, thermophilic MSR at impact craters on Earth versus a set of abiotic reactions for the first in situ surface δ 34 S analyses from Mars-call for additional understanding of sulfur isotope ratios and related controls at analog sites.
To shed further light on these contrasting studies and their respective implications, we have performed a study of multi-phase sulfur speciation and isotope partitioning at the Ries crater in southern Germany.Our aim is to determine if and when large S isotope fractionation events in crater lake systems are definitively linked to life.Accordingly, we explore the three plausible mechanisms known to produce significant δ 34 S variation in crater settings: MSR, thermochemical sulfate reduction [TSR], and hydrothermal equilibrium fractionation.

The Ries Crater Paleolake
A closed-basin lake formed in the Ries basin shortly after the impact and existed for 1-2 Myr (Arp et al., 2021a;Füchtbauer et al., 1977;Jankowski, 1977;Montano et al., 2021).Successive weathering of suevite followed by Bunte Breccia, in combination with diminishing hydrothermal activity, caused multiple distinct shifts in the lake's chemistry through time (Arp et al., 2013a).These shifts correlate approximately with the four stratigraphic units established by Füchtbauer et al. (1977) (Fig. 2).The first waters in the basin were transient playas that deposited sandstones and reworked suevite, which are now referred to as the basal member or unit A (Füchtbauer et al., 1977).This interval was followed by the formation of a permanent, eutrophic, high-pH alkalisaline lake that deposited the bituminous shales and laminated marls of the organic-rich laminite member (unit B) (Füchtbauer et al., 1977;Arp et al., 2013a;Rullkötter et al., 1990;Stüeken, Tino et al., 2020).Changes in the dominant weathering and alteration source from suevite to Bunte Breccia led to a shift toward saline waters with circum-neutral pH and deposition of calcareous claystones of the marl member (unit C) (Arp et al., 2013a).Then, a change to alternating hyposaline and hypersaline conditions led to deposition of the clay member (unit D) (Arp, 2006;Arp et al., 2013a;Füchtbauer et al., 1977).The youngest preserved lake sediments still show evidence of hydrologically closed conditions with no evidence for an outlet (Arp et al., 2017).Eventually, the lake basin filled with sediment, and lake waters permanently disappeared.

Impact-induced Hydrothermal Activity
Hydrothermal alteration was pervasive within the central basin, as indicated by the complete replacement of impact-generated glasses in crater suevite (Osinski, 2005;Stöffler, 1977).Meltrich layers within suevite sequences from multiple drill cores point to temperatures of ~600°C for the first post-impact materials that settled in the basin (von Engelhardt, 1990).The earliest sustained hydrothermal fluids were likely derived from surface meteoric waters and groundwaters from the surrounding country rock.These fluids ranged from 200-300°C as indicated by Kmetasomatism and chloritization in crater suevite (Osinski, 2005;Osinski et al., 2004).The second or main stage of alteration occurred via weakly alkaline fluids as temperatures decreased through the 200-100°C window and was characterized by intermediate argillic alteration and zeolitization products within crater suevite (Osinski, 2005).Muttik et al. (2010) used stable oxygen [δ 18 O] and hydrogen [δD] isotope data from crater suevite to constrain the temperatures of the final alteration stage to 43-112°C.Temperature increased monotonously with depth in association with convective circulation, and the coolest stage was likely the longest-lived (Muttik et al., 2010).Quantitative constraints on the duration and intensity of hydrothermal activity as it relates to the paleolake waters are presently limited.Stratigraphically, hydrothermal activity occurred throughout the basal member and extended into the laminite member (Arp et al. 2013b).During deposition of the laminite member, Ca-bearing fluids of both hydrothermal and non-hydrothermal origin were locally introduced to alkaline waters, leading to the formation of carbonate spring mounds and travertines (Arp, 1995;Arp et al., 2013b).The observation of streamer carbonates in the Erbisberg spring mound (coordinates: 48°49′52.34″N,10°30′42.56″E)indicate that 70°C waters were still emerging from the system several 100 kyr after impact and that hydrothermal effects extended at least to the crater's inner ring (Arp et al., 2013b).While the subsurface impact breccias faced significant exposure to hydrothermal fluid temperatures above the current known limits of life (~121°C), there is no reason to assume that the earliest lake waters, or the sediments deposited from them, ever existed at uninhabitably high temperatures.Such high temperatures would require elevated pressure to maintain water in a liquid state, which is not plausible for this lake system in the Miocene.Given the lateral and vertical extent of hydrothermal activity, drill cores and field sites are highly complementary at Ries crater.This study therefore includes samples from both.

Sample Collection
We analyzed the research drill core Nördlingen 1973 [FBN 73], which contains one of the most extensive paleolake sedimentary sections among the Ries drill cores and has been the subject of multiple geochemical studies, which provide valuable context for our δ 34 S work (Füchtbauer et al., 1977;Muttik et al., 2010Muttik et al., , 2011;;Osinski, 2005;Stüeken, Tino et al., 2020;von Engelhardt, 1997).For our examination of the paleolake sediments, 81 samples were selected across 241 m of stratigraphy (Fig. 2; 21-262 m depth in the core or 403-164 m above modern sea level [a.s.l.]).To demonstrate that the paleolacustrine δ 34 S data collected from the FBN 73 drill core are representative for the whole basin, 27 samples from four exploratory Nördlinger Ries [NR] wells were included (Fig. 3; NR-10, NR-20, NR-30, and NR-40 drilled by BEB Erdgas und Erdöl GmbH in the early 1980s; permission for use was provided by ExxonMobil; Rullkötter et al., 1990;Barakat et al., 2012;Barakat et al., 2013).All lacustrine sediment samples in this study have undergone minimal post-depositional geothermal heating.Vitrinite reflectance data from FBN 73 samples suggest a maximum temperature of 60°C (Wolf, 1977).Subsequent biomarker studies of NR cores pointed similarly to very minor thermal overprints (Barakat and Rullkötter, 1995).
Field sampling was included to complement the drill core analyses, with the hope of elucidating the S composition of both the water column of the alkaline lake and the subaerial weathering environment that fed into the basin at that time.Eight samples were collected from two carbonate mounds-Wallerstein travertine (n = 2, 48°53′20.3″N,10°28′28.9″E)and Goldberg spring mound (n = 6, 48°51′40.7″N,10°25′19.8″E)-foranalysis of carbonate-associated sulfate [CAS] content as a proxy for the SO 2 4 -composition of the water column during the alkaline lake phase (Fig. 4).The Wallerstein samples are from the base of that mound, which is composed of non-fossiliferous carbonates with intercalations of brine shrimp fecal pellets (Pache et al., 2001), signifying that they were deposited synchronously to the pellet-rich lower part of the laminite member (Jankowski, 1977) and formed before the Goldberg mound samples.The sampled portions of the Goldberg and Wallerstein mounds in this study were formed subaqueously (i.e., in a mixing zone of sub-lacustrine spring and alkaline lake water; Arp, 1995;Pache et al., 2001).Multi-isotope ( 87 Sr/ 86 Sr, δ 13 C, δ 18 O) constraints on mixing ratios at Ries spring precipitation zones suggest that the majority of SO 2 4 -available for incorporation as CAS was derived from the lake water (Pache et al., 2001).To evaluate possible S contributions to the inner basin from the surficial weathering of outer suevite, three samples of outer suevite were collected from Riedweg (n = 1, 48°54′54.4″N,10°26′39.8″E)and Otting (n = 2, 48°52′24.1″N,10°47′55.0″E).

Geochemical Analyses
The outer rims of all drill core and field samples were trimmed with either a rock saw or manually operated rock chipper, depending on sample hardness.The resulting interiors were hammered into subcentimeter-sized chips, rinsed with deionized H2O, and dehydrated at 50°C for a minimum of 48 hours.The rock chips were then pulverized in a ball mill.
Total organic carbon [TOC], total inorganic carbon [TIC], total sulfur [TS], chromiumreducible sulfide [CRS], and CAS data for all FBN 73, SUBO 18, and field samples were collected in the Lyons Biogeochemistry Laboratory at the University of California, Riverside [UCR].Total Carbon [TC], TIC, and TS were measured via combustion on an Eltra CS-500 carbon-sulfur analyzer, all with precision better than ±0.1 wt%.TOC was then calculated as the difference between measured TC and TIC.
The chromium reduction method (e.g., Canfield et al., 1986) was used to assess S present in sulfide minerals.Specifically, bulk sample powders were mixed with a 1 M CrCl2 (acidified to 0.5 M HCl) solution under a stream of N2 gas (i.e., anoxic conditions) and brought to a boil to evolve hydrogen sulfide [H2S] from present CRS phases.That H2S was trapped in a silver nitrate solution as silver sulfide [Ag2S], which was then quantified gravimetrically and converted to wt% CRS in the sample, assuming all H2S precipitated as Ag2S.For carbonate-rich samples (TIC wt% > 11), powders were decarbonated with 1N HCl prior to chromium reduction.In studies of ancient sediments, CRS is typically dominated by sulfur hosted in pyrite.Consistent with these past studies, previously published petrographic analyses of SUBO 18 confirm pyrite as the predominant S phase in the Ries impact breccia (Reimold et al., 2012;Reimold et al., 2013).Standard deviations [σ] for CRS wt%, as measured from repeat (n = 6) analyses of a randomly selected SUBO 18 sample and repeat analyses (n = 7) of a randomly selected FBN 73 sample, were σ = 0.03 wt% and σ = 0.07 wt%, respectively.Sulfur isotope ratio (δ 34 S) measurements of those same sets showed σ = 0.23‰ (FBN 73) and σ = 0.16‰ (SUBO 18).
To extract CAS, carbonate samples were rinsed in a 5% NaCl solution for three consecutive nights (the solution being refreshed each day) to remove all easily soluble sulfates.This step was followed by two consecutive deionized H2O rinses of 24 hours each.To remove organic-S and trace sulfides, the samples were then treated with a 30% H2O2 solution, the volume of which was calculated to be ~100x in excess of the organic content as inferred from TOC wt% measurements (assuming a simplified CH2O stoichiometry).This step was followed by two rinses with deionized H2O of 24 hours each.The purified carbonates were then acidified with a 4N HCl solution with 5 wt% stannous chloride [SnCl2] until effervescence ceased.Insoluble materials were removed by filtration, followed by addition of a saturated barium chloride [BaCl2] solution to quantitatively precipitate the SO 2- 4 as BaSO4.The volume of added BaCl2 was calculated to be roughly 40x in excess of the CAS as estimated by conservatively assuming that the measured wt% TS was entirely CAS.CAS concentrations (mg/kg) were then calculated via gravimetry of the filtered and dried BaSO4.Triplicate analyses of a randomly selected carbonate sample yielded a coefficient of variation [CV; ratio of σ to the mean] = 0.1% for CAS contents, and δ 34 S values for that same triplicate agreed to within 0.1‰.
Sulfur isotope ratios for the CRS and CAS were measured at UCR using a Thermo Scientific Delta V Plus continuous-flow stable isotope ratio mass spectrometer [IRMS].Sulfur from Ag2S and BaSO4 was converted to SO2 and separated via gas chromatography [GC] on a Costech ECS 4010, with gas flow to the IRMS controlled by a Thermo Scientific ConFlo III open split interface.For δ 34 SCRS, samples were bracketed by the international reference standards IAEA-S1, -S2, and -S3.For δ 34 SCAS, samples were bracketed by IAEA-SO5, -SO6, and NBS-127 (for the δ 34 S values of these standards, see Appendix 1 of Sharp, 2017).Precision for all standards in all runs was better than σ = 0.3‰.
Bulk sediment (i.e., including all S in a given sample) and kerogen enrichments of NR-10, NR-20, NR-30, and NR-40 were analyzed at ICBM Oldenburg for TC and TS using a Carlo Erba EA 1108 elemental analyzer.Carbonate was removed with dilute HCl, and TOC was determined coulometrically using a Ströhlein-type 702 Coulomat.TIC contents were obtained from the difference between TC and TOC.The kerogen fraction was separated as described in Barakat and Rullkötter (1995), where solvent-extracted residues are demineralized with a HF:HCl (1:1) treatment followed by additional Soxhlet and ultrasonic extractions.Bulk sediment and kerogen samples were analyzed for δ 34 S using combustion-isotope-ratio-monitoring mass spectrometry [C-irmMS]-specifically, a Finnigan MAT 252 mass spectrometer coupled to a Carlo Erba EA 1108 elemental analyzer via a Finnigan MAT ConFlo II open split interface, as described in detail by Böttcher et al. (1998) and Böttcher and Schnetger (2004).IAEA-S1, -S2, and -S3, as well as NBS 127, were used to calibrate the δ 34 S values to VCDT.Precision for the measurements was better than σ = 0.2‰.Isotope values reported in "‰" are equivalent to "mUr" (milli-Urey; Brand and Coplen, 2012).
Backscattered electron [BSE] imaging (Fig. 6) was performed in the Brounce Geochemistry Laboratory at UCR using a JEOL JCM-7000 benchtop scanning electron microscope [SEM].A combination of manual observations and montage imaging was used to examine the entirety of polished surfaces.
4.1.2.Data from the four exploratory wells (Fig. 3, Supplemental Table 2) allow for a wider breadth of δ 34 S examination in terms of both sedimentary host phases and spatial distributions within the lake basin.Bulk and kerogen δ 34 S data (δ 34 SBulk and δ 34 SKerogen) show trends that are similar, in both magnitude and pattern, to the δ 34 SCRS values from FBN 73.For any given sample where both δ 34 SBulk and δ 34 SKerogen were measured (n = 3), the differences were minor (≤ 0.5‰) between the two measured phases.Viewed within this context, our bulk measurements indicate that δ 34 S is uniformly elevated in the lower laminite member, regardless of which S-bearing phase dominates the isotopic mass balance (e.g., kerogen or pyrite) within a given sample.Moving up core, the combined data series rises from a minimum δ 34 SBulk of 7.4‰ to a maximum δ 34 SBulk of 44.2‰.There is a consistent up-core decrease over the remainder of the sample set, with data from the four uppermost adjusted sample depths (using NR-30 as a zero-reference line) falling below 15‰ (mean = 12.65‰, n = 4, 100.0 m -51.5 m).

Carbonate Mounds
δ 34 SCAS values of the carbonates (Fig. 4, Supplemental Table 3) show a range from 26.4‰ to 44.5‰ (mean = 32.1‰,n = 8, σ = 5.5‰).Analyses from Wallerstein travertine (δ 34 SCAS = 36.3‰,44.5‰) display higher values than those from the Goldberg spring mound (mean = 29.3‰,n = 6, σ = 1.8‰).CAS contents range from 1075 mg/kg to 4091 mg/kg, and there is strong negative correlation between CAS content and δ 34 SCAS, where samples with lower CAS contents tend to have higher δ 34 SCAS (R 2 = 0.81).TIC for all carbonate samples is 11.6 wt% or greater (compared to 12 wt% for pure CaCO3), attesting to their very low insoluble, siliciclastic contents.CRS contents were 30 mg/kg or lower and represent ≤ 5% of the measured TS in any given sample, suggesting that the risk of isotopic contamination from CRS oxidation during CAS extraction is low.

Impact Breccias 4.2.1. SUBO 18 (Crater Suevite)
The lowest sample depths (99.35 m -86.4 m) of SUBO 18 are composed of impact melt breccia (n = 13) and have δ 34 SCRS values that generally decrease up core, ranging from 9.1‰ to -19.0‰ (Fig. 2, Supplemental Table 1).The transition into crater suevite at 82.92 m coincides with a continued decrease until the minimum of -31.4‰ is reached at 68.59 m depth.This minimum is 4.5‰ lower than the next lightest value (-26.9‰ at 78.23 m) and does not appear to be representative of crater suevite (mean: -20.2‰, n = 20, σ = 5.8‰) or the entire set of impact breccia samples (mean: -15.8‰, n = 33, σ = 9.1‰).δ 34 SCRS then increases monotonically over the remainder of the core, reaching the crater suevite maximum of -7.1‰ at 30.73 m, near the top of the analyzed interval.CRS wt% spans from 0.04% to 2.57% and differs in its relationship with δ 34 SCRS depending on rock type (Fig. 5a): crater suevites show a moderate negative correlation (R 2 = 0.54, p < 0.001), with high CRS wt% tending to correspond with low δ 34 SCRS, while impact melt breccias show no discernable relationship (R 2 = 0.03).TOC wt% varies from < 0.1% (i.e., below detection limit [b.d.l.]) to 0.8%, but the majority of samples display low values; all but four of the 34 analyses have ≤ 0.3 wt% TOC.TOC wt% is consistently low in the samples with highest CRS wt% (Fig. 5b)-specifically, all samples with CRS contents > 1.3 wt% (n = 15) also have TOC ≤ 0.1 wt%.
BSE imaging (Fig. 6) reveals that Fe-sulfides are mainly disseminated throughout the groundmass (i.e., not in clasts) of both the crater suevite and the impact melt breccia.This observation is in agreement with previous petrographic and geochemical analyses of SUBO 18 (Reimold et al., 2012;Reimold et al., 2013), which revealed that these sulfides are predominantly pyrite of post-impact origin, with minor occurrences of sphalerite [(Zn, Fe)S] in melt particles.Morphologically, the vast majority of the observed pyrite is non-framboidal, regardless of sample depth.Only one ~100 μm 2 area of one sample (68.59 m) was observed to contain framboid-like morphologies, which are ~1 μm in diameter and fracture-filling.

Field Samples (Outer Suevite)
All three field samples of outer suevite display CRS contents of 10 mg/kg or less.As a result, isotope measurements were not performed on these samples.These values are notably below 0.04 wt%, the observed CRS minimum for crater suevite.TOC was ≤ 0.1 wt%, and TS was < 0.1 wt% for all outer suevite samples.

Overview of sulfur geochemistry at the hydrothermal Ries crater lake
The SUBO 18 interval under investigation is composed of impact breccias that had hydrothermal fluids migrate through them en route to the water column of the Miocene Ries crater lake.The formation of secondary pyrite in the subsurface impact breccias likely had a direct influence on the S chemistry of the fluids that reached and accumulated in the overlying waters, and the chemistry of these lake waters is preserved in the sedimentary deposits of FBN 73.The strongest hydrothermal influence on lake water chemistry likely occurred during the basin's early evolution from transient playas to a permanent lake (i.e., an interval of several 100 kyr that was likely coincident with the basal member extending into the early laminite member).The distinct geochemical differences between SUBO 18 and FBN 73 (Fig. 2) support this overarching interpretation.The lowest δ 34 SCRS values are sequestered in impact breccia, with the highest δ 34 SCRS values residing in the laminite member of the paleolake sediments.In a general sense, this relationship would be expected in order to satisfy isotopic mass balance.Past research suggests that meteoric waters (i.e., supplied by a combination of groundwater and precipitation) and the materials that they weathered were likely the only significant sources of dissolved SO 2 4 -to the lake for much of its history (Arp et al., 2013a).During periods of hydrothermal activity, much of that SO 2 4 -was apparently scavenged in the subsurface as sulfide, resulting in average CRS contents that are higher in the underlying impact breccias compared to sediments of the laminite member.Sulfate that was not scavenged in the subsurface would have become available to the lacustrine system via ascending hydrothermal fluids.
Within the crater lake ecosystem, as archived by FBN 73 and the carbonate mound samples, δ 34 S was uniformly high.CAS data (Fig. 4) indicate that the SO 2 4 -isotope composition of the water column (δ 34 SCAS) during deposition of the laminite member ranged from 26.4‰ to 44.5‰, a spread similar to that of the δ 34 SCRS values of the FBN 73 laminite (17.5‰ to 45.1‰).Absolute concentrations of dissolved SO 2 4 -in the lake water column cannot be directly inferred from measured CAS concentrations or δ 34 SCAS (Fichtner et al., 2017;Gill et al., 2008).However, given the depositional similarities of the carbonate mounds sampled in this study, we assert that comparisons of relative CAS concentrations, at least in a first-order sense, are valid.The strong anticorrelation between CAS concentrations and δ 34 SCAS values (R 2 = 0.81) suggests that high extents of S sequestration in the subsurface impact breccias led to more positive δ 34 S values of the water-column SO 2 4 -.Furthermore, because the Wallerstein samples are older than those of the Goldberg mound, the δ 34 SCAS data of the carbonates fall on the same stratigraphically descending δ 34 S trend defined by the laminite member.The slight differences between δ 34 SCAS and laminite δ 34 SCRS values therefore indicate small or insignificant fractionation events during the transformation of dissolved SO 2 4 -(as captured by CAS) to mineral-phase sulfide (represented by CRS), or vice versa.
The data across the NR cores (Fig. 3) provide additional evidence that the general observed lake δ 34 S trend represents a homogenous signal that existed throughout the basin and that δ 34 S values are similar in magnitude across all phases measured (i.e., CRS, CAS, bulk materials, and kerogen).The NR core data also extend the argument that, at least during deposition of the lower laminite member, sulfate with high δ 34 S was distributed throughout the water column and in the sediments, with that S pool being preserved in the sediments with minimal net fractionation.This assertion is consistent with experimental studies that have demonstrated only small (< 4‰) δ 34 S fractionation during the formation of both CAS from dissolved sulfate and pyrite from dissolved sulfide (Barkan et al., 2020;Butler et al., 2004), although some 34 S enrichment has been observed in sedimentary organic matter when compared to co-existing Fe-sulfides (e.g., Kotuzov et al., 2020).
In a model approach, Arp et al. (2013a) used the SO 2 4 -concentrations found in local modern groundwaters (~0.85 mM; Winkler, 1972) and a dynamic impact breccia weathering scenario to show that water column SO 2 4 -values were likely higher than 200 μM shortly after lake formation and remained that way for the remainder of the lake's lifetime (Arp et al., 2013a;Füchtbauer et al., 1977).Therefore, a concentration-dependent metabolic isotope effect, which can lead to small fractionations from MSR in a low SO 2 4 -[below 200 μM] environment, appears unlikely (Crowe et al., 2014;Habicht et al., 2002;Harrison and Thode, 1958).However, near-quantitative MSR (i.e., an extent that is indistinguishable in the rock record from a quantitative reaction)-leading to reservoir-effect-induced sulfide isotopic compositions that closely resemble those of the parent dissolved SO 2 4 --is plausible under the high organic productivity regime of the laminite member.Microbial sulfate reduction likely occurred in the lake sediments during early diagenesis and possibly in the water column during certain intervals (i.e., euxinia), specifically in a saline hypolimnion with a sulfide surplus relative to Fe 2+ , as indicated by the presence of thiophene biomarkers in the early laminite sediments (Arp et al., 2013a;Barakat et al., 1994;Barakat and Rullkötter, 1997;Rullkötter et al., 1990).
The coincident, gradual decline in δ 34 SCRS throughout the remainder of FBN 73 could then be interpreted as reflecting a highly restricted, evaporitic lake with progressively decreasing contributions of hydrothermally derived SO 2 4 -and progressively increasing contributions of SO 2 4 from karstic groundwater and surface weathering of Bunte Breccia (Arp et al., 2013a).In any case, it is apparent that there was little-to-no observable S-isotope fractionation within the earliest subaerial lake itself.Thus, we interpret from the above data that the primary locus of fractionation (and therefore the source of both the maximum and minimum observed δ 34 S values) was the subsurface hydrothermal system.The exact nature of the S isotope fractionation mechanism in the hydrothermal system, and its relevance as a biosignature, is more complex.

Mechanisms for the observed sulfur isotope variation
There are multiple lines of reasoning that can be used to assess the observed δ 34 S variation at Ries Crater.MSR is often evoked as a means of explaining large differences in δ 34 S values between source sulfate and product sulfide [Δ 34 Ssulfate-sulfide], with offsets of up to 75‰ (Brunner et al., 2005;Canfield et al., 2010;Lyons and Gill, 2010;Sim et al., 2011;Wortmann et al., 2001).However, meteorite impacts can induce target rock temperatures well in excess of 130°C, exceeding the limits of life and possibly generating significant thermochemical sulfate reduction while using the reducing power of organic matter (Abramov and Kring, 2005;Ivanov, 2004;Machel, 2001;Osinski et al., 2013).Thermochemical sulfate reduction has been shown to yield Δ 34 Ssulfate-sulfide values of up to 21‰; this can confound interpretations that assign S-isotope fractionations to MSR, specifically in cases where δ 34 S variations are smaller than 21‰ in rocks exposed to temperatures above 100°C (Machel et al., 1995;Machel, 2001;Ohmoto and Goldhaber, 1997;Watanabe et al., 2009).Additionally, theoretical work indicates that hydrothermal activity can lead to large equilibrium fractionations between aqueous SO 2 4 -and H2S-e.g., Δ 34 SSO 2- 4 -H2S = 43.4‰at 100°C (Eldridge et al., 2016;Marini et al., 2011;Ohmoto and Goldhaber, 1997;Sakai, 1968).Equilibrium S fractionation is typically considered for magmatic hydrothermal systems, but impact events also provide a plausible setting for this effect, depending on crater size, target rock composition, and fluid chemistry (Abramov and Kring, 2007;Franz et al., 2017;Ivanov, 2004;Kirsimäe and Osinski, 2012;Marini et al., 2011;Osinski et al., 2013).Each of these scenarios (i.e., MSR, TSR, and equilibrium fractionation) warrants individual evaluation, as well as consideration for the potential simultaneous or sequential occurrence of two or more fractionation mechanisms.

Microbial sulfate reduction
If the δ 34 S data are viewed in the absence of the full system context, MSR is the most straightforward way to reconcile the δ 34 SCRS difference between the highest paleolake sedimentary samples and average impact breccia values (Δ = 61‰; Fig. 2, Supplemental Table 1).Hydrothermal activity does not preclude MSR: optimal growth rates for hyperthermophilic MSR Bacteria can be at temperatures as high as 83°C, with an observed upper limit of 90°C (Thermodesulfobacterium geofontis; Hamilton-Brehm et al., 2013), while MSR Archaea also show optimal growth at temperatures as high as 83°C (Archaeoglobus fulgidus; Stetter 1988;Stetter et al. 1987) but can apparently function at temperatures ~100°C (Elsgaard et al., 1994;Jørgensen et al., 1992;Weber and Jørgensen, 2002).Canfield et al. (2000) measured MSR δ 34 S fractionation from Guaymas Basin sediments sampled at 85°C and found Δ 34 Ssulfate-sulfide of 13‰ to 28‰, comparable to results from pure thermophilic cultures (Böttcher et al., 1999).This fractionation is notably smaller than that observed in some isolates from mesophilic environments (i.e., > 60‰ in those measured by Canfield et al., 2010, andSim et al., 2011).Taken together, these studies point to TSR or hydrothermal equilibrium as being the dominant S-fractionation mechanisms in geologic settings with temperatures > 100°C (Machel, 2001) and imply that, from a δ 34 S perspective, it becomes increasingly difficult to distinguish between MSR and TSR as this temperature range is approached.
Appraisals of hyperthermophilic MSR viability aside, the lack of observed framboidal pyrite (associated with organics or otherwise) in favor of fracture-and/or void-filling forms makes possible biological origins ambiguous.In the single observed area where framboid-like pyrite is found (Fig. 6b), it is present in fracture fills.Indeed, these forms observed in SUBO 18 may not be true framboids in the sense that they do not show organized or disorganized clustering.They can instead be described as linked, near-equidimensional microcrystals.Even if we assume that all observed framboid-like pyrite was of bacterial origin, despite the variety of abiotic framboid formation mechanisms (Ohfuji and Rickard, 2005), these pyrites would be insignificant in terms of their overall contribution to δ 34 S mass balance because they compose only a small portion of the total pyrite.The potential for morphological overprinting of pyrite framboids due to hydrothermal activity (Nozaki et al., 2020) furthers ambiguity and highlights a broader issue for sulfide biogenicity, as hydrothermalism is generally considered a relevant feature for origin of life hypotheses on Earth as well as the search for life elsewhere (Deamer et al., 2019;Hendrix et al., 2019;Schwenzer and Kring, 2009).Hydrothermal activity is known to have been a feature of many impact craters on Earth and Mars (Osinski et al., 2013).
Contextual geochemical trends also contribute to MSR uncertainty.TOC wt% in the SUBO 18 samples is uniformly low, and a TOC wt% vs CRS wt% crossplot (Fig. 5b) shows low TOC wt% corresponding to high CRS wt%.This trend is the reverse of that typical of MSR, where higher TOC contents facilitate greater pyrite precipitation (e.g., Westrich and Berner, 1984).This reasoning also applies when scrutinizing the trend seen in suevite samples specifically (Fig. 5a), where higher CRS wt% corresponds to lower δ 34 SCRS; greater rates of MSR and thus pyrite formation should impart smaller fractionations (Canfield, 2001;Kaplan and Rittenberg, 1964)but the opposite is observed.Rather than invoking biology, these data may be better interpreted within the framework of post-impact TSR at temperatures exceeding 100°C.
Rayleigh fractionation in a closed-system with respect to reactant SO 2
The Rayleigh distillation scenarios above assume an ideal closed system and therefore can only provide a limited reconstruction of what was a highly dynamic environment (Fig. 8).Much of the earliest hydrothermal fluid was likely derived from groundwaters migrating into the hydrostatic void generated by the impact event (Osinski, 2005).Surface water would have percolated back downward into the breccia (Osinski, 2005) and tenuously balanced the upward movement of groundwater, particularly after reaching a hydrostatic equilibrium (there is no evidence of outlet formation at Ries crater).Thus, while Ries is not a theoretically ideal closed system, SO 2 4 -anions would have had prolonged exposure to subsurface hydrothermal circulation.
In any case, extensive Rayleigh distillation appears to have occurred: even the highest observed hyperthermophilic MSR fractionation (ε = -28) would require significant closed-system SO 2 4 removal (> 75%) to mimic the observed data.
It should be noted that an all-encompassing numerical solution that assumes the Ries crater to be a completely closed system with regards to δ 34 S is not necessary to explain our observations.This is because temporal constraints on the intensity of hydrothermal activity during the paleolake interval that hosts δ 34 S values of > 40‰ are limited.It is possible, for instance, that the δ 34 SSO 2- composition of the lake at that time reflects the latest possible stages of hydrothermal activity, when only the uppermost suevite was interacting with the lake surface water.The change in δ 34 SΣH 2S through time may be reflected in the suevite δ 34 SCRS trend, which increases moving up core.This argument allows for a scenario discussed above, where TSR generates δ 34 SSO 2- 4 of 45.1‰ and δ 34 SΣH 2S of -6.2‰ from initial δ 34 SSO 2- 4 of 0‰ and could explain the relatively elevated δ 34 SCRS of the two uppermost suevite samples (which show values of -7.1‰ and -10.0‰).For this to occur, S would need to permanently escape the restricted basin, which can happen via H2S volatilization.If the early laminite member waters were euxinic (Arp et al., 2013a;Barakat et al., 1994;Barakat and Rullkötter, 1997;Rullkötter et al., 1990), H2S derived from the hydrothermal system could have been transported to the lake's water column.Water-column sulfide is susceptible to volatilization (Fry et al., 1986;Reese et al., 2008).An example is the modern Salton Sea in California, USA, where H2S escapes to the atmosphere during seasonal mixing in a process that has been projected to remove ~24% of the basin's dissolved H2S annually (Reese et al., 2008).While this effect imparts only minor fractionation (εgas-aqueous = -1.6‰ to 3.0‰; see Baune and Böttcher, 2010;Fry et al., 1986;Sim et al., 2019), it nonetheless allows for the isotopic evolution of a restricted basin's sulfur reservoir on geologic timescales.This is because a proportion of the S derived from hydrothermal activity and isotopic distillation could have escaped via volatilization relatively early in the lake's history.If this was followed by increased relative input and mixing of non-hydrothermal, meteoric S (with its lower δ 34 S composition), then the elevated early lacustrine δ 34 S signal would be effectively diluted over time.The implication is that during formation of the lightest pyrite, δ 34 SSO 2- 4 of < 45.1‰ may have been generated but is only briefly recorded in the rock record (i.e., during the transition from the basal member into the laminite member, where paleolake δ 34 SCRS values shift from -5.2‰ to 41.9‰ over the core depths of 260.5m-248.5m).As the basin evolved, the restricted recycling of high δ 34 SSO 2- 4 could have resulted in the simultaneous occurrence of the highest suevite δ 34 SCRS and the highest paleolake δ 34 S values (Δ = 52.8‰,where δ 34 SSO 2- 4 = 45.1‰ and δ 34 SΣH 2S = -7.7‰).This spread can occur from TSR removing ~89% of a limited pool of SO 2 4 -with an initial δ 34 SSO 2- 4 value of -1.7-a plausible value for groundwater SO 2 4 -, even without the likely mixing of an isotopically lighter SO 2 4 -source such as oxidized pyrite of the Posidonia shale and/or other pyrite-bearing Jurassic and Triassic formations of the target rocks (Arp et al., 2021b).In light of the above discussion, we assert that TSR was as likely as MSR to have been the primary δ 34 S fractionation mechanism in Ries crater.

Hydrothermal equilibrium isotope fractionation
The near-absence of organics associated with the most sulfide-enriched impact breccia samples may imply a role for hydrothermal equilibrium fractionation.After the exhaustion of reactive organic matter via TSR, equilibrium fractionation between aqueous S-bearing phases has been shown to occur (Meshoulam et al., 2016).The timescales needed for such an equilibration are highly dependent on temperature, pH, and the total concentration of S species [ΣS] (Ohmoto and Goldhaber, 1997;Sakai, 1968).According to the equations presented in Ohmoto and Goldhaber (1997), a hydrothermal solution (T = 100°C, pH = 7, ΣS = 0.01 molal) would take ~407 kyr to reach 90% of isotope exchange equilibrium (i.e., indistinguishable from an equilibrium state).At that temperature, the equilibrium fractionation for between SO 2 4 -and H2S has a 1000lnα value of about 43‰, with SO 2 4 -being substantially enriched in 34 S (Eldridge et al., 2016;Sakai, 1968).The subsequent equilibration of H2S and HS -, which is an expected occurrence at circumneutral-to-basic pH, results in a 1000lnα value of 2.5, where H2S has a higher δ 34 S than HS - (Eldridge et al., 2016;Sakai, 1968).While the hydrothermal fluid composition described above is broadly consistent with the results from previous studies of the Ries crater system (e.g., Arp et al., 2013b;Muttik et al., 2010;Osinksi, 2005;Sapers et al., 2017), generating such a large fractionation would require fluid temperatures at ~100°C for the entirety of the protracted equilibration time.Further, changes to single parameters of the fluid composition profoundly decrease the likelihood of this fractionation.Shifting the pH to ~9, which is compatible with the pH range suggested for the Ries system based on boron isotope measurements (Muttik et al., 2011), would increase the time necessary for an equilibration from ~407 kyr to an impossible ~869 Gyr.Decreases in ΣS by a factor of 10 also increase the time to equilibrium by a factor of 10 (Ohmoto and Goldhaber et al., 1997).Due to these rigid theoretical constraints, if hydrothermal equilibrium were ever reached at Ries crater, it would have been brief and occurred at temperatures where the equilibrium fractionation is much smaller (e.g., at 300°C, 1000lnα = 19.8;Eldridge et al., 2016).Such effects would be transient and likely overprinted beyond interpretation by larger, nonequilibrium fractionations caused by TSR or MSR.However, this remains a viable abiotic mechanism in acidic and/or high ΣS fluid environments where reductant scarcity prohibits nonequilibrium fractionation mechanisms.

Conclusion and Astrobiological Implications
In the present study, we tested the viability of stable sulfur isotopes as a biosignature in impact crater lakes through a detailed investigation of the Miocene Ries crater, which serves as an analog to early habitable locations on Mars and Earth.At Ries, elevated temperatures sustained by impactinduced hydrothermal activity allowed for the possibility of S isotope fractionation mechanisms that are rare in mesophilic environments (where δ 34 S variation > 21‰ is often representative of microbial S cycling).After examining the observed δ 34 S spread of ~77‰ at the Ries crater, we contend that TSR was at least as likely as MSR to have been the primary δ 34 S fractionation mechanism (Fig. 8).This conclusion is further supported by multiple lines of evidence, including impact breccia with TOC wt% versus CRS wt% and δ 34 SCRS versus CRS wt% relationships that are atypical of biological controls, as well as ambiguous pyrite textures.The hydrothermal system apparently dominated S isotope fractionation, as there is evidence (in the form of near-equivalent δ 34 S in CRS, CAS, bulk-S, and kerogen-S samples) that lacustrine S cycling imparted little-to-no δ 34 S fractionation during deposition of the most 34 S-enriched lacustrine pyrite (45‰).We have shown that a potential source of isotopically light SO 2 4 -(e.g., oxidized pyrite of the Posidonia shale, or from other Jurassic and/or Triassic formations) from the catchment area of the lake would have allowed, via Rayleigh distillation of residual SO 2 4 -, for TSR fractionation events of 21‰ to result in Δ 34 SSO 2- 4 -H2S distributions that match the observed Ries δ 34 S spread.Additionally, spatiotemporal uncertainties inherent to impact-induced hydrothermal activity imply that δ 34 S turnover via mechanisms such as H2S volatilization is possible at the Ries crater given extensive timescales, regardless of its closed-basin nature.
Our overarching assessment that wide-ranging δ 34 S in impact crater lakes need not be biological is in general agreement with in-situ Curiosity rover findings (Franz et al., 2017), although the specified mechanisms differ.Hydrothermal equilibration of aqueous S species at temperatures ~100°C is unlikely at Ries given constraints involving elevated fluid pH and hydrothermal activity duration (several 100 ky) (Arp et al., 2013b;Ohmoto and Goldhaber, 1997).However, this remains a viable fractionation mechanism at Gale crater, where fluid pH appears to have been acidic-to-circumneutral, and the crater's larger diameter (154 km) implies longer-lasting hydrothermal activity (Abramov and Kring, 2005;Berger et al., 2017;McLennan et al., 2013;Rampe et al., 2020).Our study, however, suggests that impact-induced TSR of abiotically derived organic matter should also be a considered S-fractionation mechanism at Gale, Jezero, and other ancient craters on Mars and Earth.Exogenous delivery of abiotic organics may have been relatively widespread early in our solar system's lifetime (Sandford et al., 2020), and it is plausible that there was/is a subsurface abiotic organic reservoir on Mars (Eigenbrode et al., 2018;Yang et al., 2020) available to be incorporated into breccia during impact events and subsequently used as a reductant for TSR.TSR could have also proceeded on ancient Mars via alternative reducing species such as dissolved H2 (Cai et al., 2022;Wordsworth et al., 2021) and has previously been proposed to explain the multiple sulfur isotope distribution of pyrite in Martian meteorite NWA 7533 (Lorand et al., 2020).Therefore, even with our multitude of well-preserved samples, wet chemical techniques, and Earth-bound instrumentation, our wide-ranging (> 21‰) δ 34 S data from an impact crater paleolake suggest that biological processes need not be involved.In other words, wide δ 34 S ranges can be abiological, and overarching geologic context is essential to any interpretation of such data.Thus, the potential ambiguities of sulfur isotope composition as a biosignature should be recognized in future studies of crater systems on Mars and Earth.Ultimately, any convincing arguments for past life on Mars or elsewhere within or beyond our solar system would likely have to be built on multiple parameters viewed within a comprehensive context.After a Rayleigh distillation process in which at least 93% of SO 2 4 -is converted to H2S (assuming a closed system), isotopically enriched SO 2 4 -of δ 34 S ≥ 45‰ reaches the water column of the lake.
Sulfide produced in this hydrothermal process either precipitates as pyrite within suevite or is volatilized out of the system.Water column SO 2 4 -is then incorporated into CAS (white arrow), as well as reduced and incorporated into lacustrine CRS and kerogen (green arrow), with little to no additional fractionation (indicated by the data in Figures 2, 3, and 4).Approximate basin structure, lithology, and horizontal scale were adapted from Stöffler et al. (2013).Thin black lines indicating lacustrine sedimentary layering are aesthetic and do not represent geologic data.Paleolacustrine sediment thickness from this lake stage (i.e., when δ 34 SCRS = 45.1‰) is magnified three-fold (see left margin) with respect to the underlying suevite layer; thickness data are from the FBN 73 core only.For a detailed analysis of changes in the vertical extent of Ries crater sediments through time, see Arp et al. (2021).

Figure 1
Figure 1 Geographical context for Ries crater, including the locations of all drill core and field sites sampled or referenced in this study.Adapted from Arp et al. (2013a) and Stöffler et al. (2013).

Figure 2
Figure 2 Stratigraphic trends of the FBN 73 and SUBO 18 drill cores, connected to a simplified diagram of their relative positions within Ries crater's concave structure.The depths used for bounding the four sedimentary units (members) of FBN 73 are from Füchtbauer et al. (1977), and the dashed stratigraphic correlation markers are adapted from Arp et al. (2021).Lithostratigraphy of SUBO 18 is adapted from Reimold et al. (2012).Angular lines near the base of the visual representation for FBN 73 represent breaks in stratigraphy and thus a deviation from relative scale at depths below 314.3 meters.

Figure 3 δ
Figure 3 δ 34 S data from bulk and kerogen analyses across the sedimentary paleolake intervals of four drill cores (NR-10, NR-20, NR-30, and NR-40) from within the inner ring of Ries crater.Sample depths were adjusted based on vertical trends, using the unadjusted NR-30 sample depths as a fixed reference point.

Figure 4
Figure 4 CAS (mg/kg) vs. δ 34 SCAS (‰) scatterplot composed of data generated from the Goldberg and Wallerstein carbonate mounds.The negative relationship with a strong coefficient of determination indicates that the highest δ 34 SCAS values were generated during intervals where the lake water had relatively low SO 2

Figure 5
Figure5Scatterplots of SUBO 18 data.(a) There is a moderate relationship for CRS wt% versus δ 34 SCRS in crater suevite samples, but no relationship between these variables in impact melt breccia samples.(b) An apparent low TOC threshold is observed for the highest CRS contents, where all breccia samples with > 1.3 wt% CRS also have TOC ≤ 0.1 wt%.

Figure 6
Figure 6 BSE Imaging data of polished SUBO 18 billets.(a) Images from four different depths show pyrite morphologies that are non-framboidal in nature.Depths: (i) 96.11 m, (ii) 90.93 m, (iii) 68.59 m, (iv) 30.75 m.(b) The only observation of framboid-like morphologies among the samples (shown at two different magnifications), where they (i) are filling fractures and (ii) are approximately 1 μm in diameter.They could alternatively be described as linked, nearequidimensional microcrystals.Both (i) and (ii) are at sample depth 68.59 m.