Hydrothermal regeneration of ammonium as a basin-scale driver of primary productivity

Hydrothermal vents are important targets in the search for life on other planets due to their potential to generate key catalytic surfaces and organic compounds for biogenesis. Less well studied, however, is the role of hydrothermal circulation in maintaining a biosphere beyond its origin. Here we explored this question with analyses of organic carbon, nitrogen abundances, and isotopic ratios from the Paleoproterozoic Zaonega Formation (2.0 Ga), NW Russia, which is composed of interbedded sedimentary and mafic igneous rocks. Previous studies have documented mobilization of hydrocarbons, likely associated with magmatic intrusions into unconsolidated sediments. The igneous bodies are extensively hydrothermally altered. Our data reveal strong nitrogen enrichments of up to 0.6 wt.% in these altered igneous rocks, suggesting that the hydrothermal fluids carried ammonium concentrations in the millimolar range, which is consistent with some modern hydrothermal vents. Further, large isotopic offsets of approximately 10 ‰ between organic -bound and silicate-bound nitrogen are most parsimoniously explained by partial biological uptake of ammonium from the vent fluid. Our results, therefore, show that hydrothermal activity in ancient marine basins could provide a locally high flux of recycled nitrogen. Hydrothermal nutrient recycling may thus be an important mechanism for maintaining a large biosphere on anoxic worlds.


Introduction
Earth is the only inhabited planet we know of, but it is not the only body in the solar system that is (or was) volcanically active with liquid water in direct contact with fresh volcanic rock.Over the past few decades, the Mars Reconnaissance Orbiter and several rover missions have identified serpentine, phyllosilicates, and silica deposits on Mars that testify to hydrothermal fluid flow in the distant past (e.g., Ehlmann et al., 2010;Viviano et al., 2013;Yen et al., 2021).The Cassini spacecraft and Earth-bound telescopes have detected evidence of active hydrothermal plumes emanating from Enceladus and Europa, respectively (Waite et al., 2006;Roth et al., 2014;Paganini et al., 2020).These finding are tantalizing because hot springs and deep-marine vents have been associated with prebiotic chemical reactions that may be crucial for an independent origin of life.In particular, the gradients in temperature, pH, redox potential, and water composition that typify these environments, along with numerous catalytic minerals, can lead to the formation of organic compounds (Martin et al., 2008).
Hydrothermal vents thus combine two mantras in Astrobiology: "follow the water" and "follow the energy".
Up to the modern day, hydrothermal vents on Earth harbor diverse ecosystems that are adapted to these unique physicochemical conditions.These ecosystems are sometimes regarded as extremophilic oddities of the biosphere that are only remotely related to the more easily visible organisms on land and at the sea surface that play a prominent role in modern biogeochemical processes.For example, carbon, nitrogen, and oxygen cycles have been dominated by the activities of cyanobacteria, eukaryotic algae, land plants and their symbionts since oxygenic photosynthesis evolved and rose to ecological dominance in the Proterozoic and early Phanerozoic, respectively (Falkowski et al., 2004;Brocks et al., 2017;Sánchez-Baracaldo et al., 2021).However, this view does not apply to early Earth or other planetary bodies where oxygenic photosynthesis has not (yet) evolved, land masses are rare or absent, and the flux of nutrients from weathering of the continents is greatly reduced.For such alien, yet potentially habitable worlds, it is imperative to develop a better understanding of how volcanic and hydrothermal processes could supply and recycle nutrients.In other words, hydrothermal vents may well be cradles of life, but their ability to sustain and maintain a diverse biosphere also needs to be elucidated.
Previous studies have shown that vent fluids today can disperse significant amounts of bioavailable iron and copper (Sander & Koschinsky, 2011;Schine et al., 2021) into the open ocean, and they may also have played a role in elevating marine cobalt (Swanner et al., 2014), copper (Stüeken, 2020), nickel (Konhauser et al., 2015), and phosphorus levels (Rasmussen et al., 2021) in the Archean (> 2.5 billion years ago, Ga).Furthermore, N2 and lightning-derived nitrite may be converted abiotically to ammonium in hydrothermal settings (Brandes et al., 1998;Brandes et al., 2008;Smirnov et al., 2008), which has prompted hypotheses about the origin of biological N2 fixation in these environments (Preiner et al., 2018).Lastly, and most noteworthy for this study, hydrothermal activity can mobilize ammonium from organic-rich sediments (Lilley et al., 1993) and thus act as an efficient recycling mechanism of fixed nitrogen.The importance of nutrient recycling is not to be underestimated, considering that more than 99% of the biomass of primary producers in the modern surface ocean is subject to oxidative remineralization in deeper waters (Emerson & Hedges, 1988).This remineralization process maintains the deep-marine nutrient reservoirs, which are eventually re-supplied to the surface ocean by upwelling.On early Earth, when the deep ocean was largely anoxic (Lyons et al., 2014), this oxidative recycling pathway would have been severely suppressed (Kipp & Stüeken, 2017).Anaerobic biomass degradation can release ammonium, as exemplified by anoxic environments (e.g., Boudreau & Canfield, 1988).However, recent data from the Black Sea, a modern anoxic basin, suggest that biomass degradation is reduced by 50% under anoxic conditions compared to oxic settings (Jessen et al., 2017).Computational modeling also suggests high degrees of biomass burial for the Precambrian ocean (Kipp et al., 2021), which indicates that nutrient regeneration would have been less efficient.Thus, post-depositional recycling aided by hydrothermal processes may have been important for sustaining the biosphere on a basinal scale.However, despite previous indications of ammonium recycling and biological utilization from Archean and Proterozoic settings (Stüeken et al., 2021a;Stüeken et al., 2021b), the magnitude of this flux has so far not been quantified.
To test how much bioavailable ammonium can be recycled by hydrothermal fluids, we turned to the Paleoproterozoic Onega Basin in NW Russia (ca.2.0 Ga, Hannah et al., 2008a).The Zaonega Formation is famously known for unusually high organic matter contents of up to 40% (Melezhik et al., 1999) and syn-sedimentary magmatic activity in the form of lava flows and intrusions into wet sediments (Črne et al., 2013).Previous work has documented hydrocarbon seeps that probably resulted from heating of organic-rich sedimentary strata by magmatic intrusions (Qu et al., 2012;Qu et al., 2020).This package of rocks is ideal for addressing our research question because (a) it shows evidence of hydrothermal alteration of shallow buried organic matter during deposition of overlying strata and (b) the interspersed hydrothermally altered lava flows and intrusions can act as an archive of ammonium concentrations in the fluids by comparison to N-poor unaltered volcanic rocks.We stress that this basin is extreme in its organic matter content and degree of volcanic activity that goes beyond other, more quiescent, settings.However, these features make the Onega Basin a useful endmember where the effects of hydrothermal circulation on nitrogen cycling can be isolated from background processes.

Geological Setting and Materials
The Zaonega Formation comprises a ca.1500 m thick succession of sedimentary and volcanic rocks in the middle-upper part of the Paleoproterozoic Onega Basin succession at the south-eastern margin of the Fennoscandian Shield.It occurs above the carbonate-and evaporite-dominated Tulomozero Formation, which is characterized by anomalously positive δ 13 Ccarb values of the Lomagundi-Jatuli isotope excursion (Brasier et al., 2011;Prave et al., 2021).The Zaonega succession consists mainly of interbedded sandstones, greywackes, mudstones, and dolostones interlayered with basaltic lava flows and intersected by gabbroic sills (Črne et al., 2014).The latter have pepperitic contacts consistent with intrusion into wet, unconsolidated sediments.The upper part of the Zaonega Formation in particular houses numerous magmatic bodies and is exceptionally rich in organic matter, with total organic carbon (TOC) abundances above 10% in many intervals (Filippov, 1994;Qu et al., 2012).
The material used in the study was sampled from drill core 12AB obtained by the Fennoscandian Arctic Russia -Drilling Early Earth Project (FAR-DEEP) and coordinated by the International Continental Scientific Drilling Program (ICDP).This core intersects 504 m of the upper part of the Zaonega Formation and contains three magmatic bodies separated from each other by sedimentary rocks.The magmatic bodies at 484-414 m and 119-55 m (a package of sedimentary rocks at 113-95 m) have pepperitic contacts with extensive contact zone alteration and are interpreted as sills that intrude wet sediments.The magmatic body at 367-315 m contains a few scattered amygdales in the upper part, has a sharp unaltered upper contact to the overlying sedimentary rocks, and is interpreted to represent a lava flow.The magmatic bodies are referred hereafter as lower (484-414 m), middle (367-315 m), and upper (119-55 m; note that this consists of two separate sills).We selected samples from core 12AB with particular focus on the hydrothermally altered magmatic bodies to assess the effects of hydrothermal fluid flow on nutrient mobility.

Methods
The outer ~0.5 cm of all core samples was cut off with a water-cooled rock saw, and the interiors were hammered into 2-5 mm-sized chips with a steel pestle on a steel plate.The chips were placed into an acid-washed and muffled (500 °C) glass beaker and cleaned by swirling them for 10-20 seconds in DI-water (18 MΩ/cm), methanol, DI-water, 1M HCl (reagent grade), and again DI-water.The rinsed chips were dried in a closed oven at 70 °C overnight and then pulverized with an agate ball mill.Muffled silica sand was run through the mill in between samples.Analyses of this sand showed total N concentrations of less than 1.5 µg/g.The powder was stored in muffled scintillation vials.
For organic carbon and nitrogen isotope analyses, roughly 0.5 g of powder was weighed into a muffled glass centrifuge tube and mixed with 10 ml of 2M HCl using a clean glass rod.The tubes were capped loosely to allow CO2 gas to escape and placed into an acid-proof oven at 70 °C overnight.The next day, the acid was decanted after centrifugation.Complete dissolution of all carbonate was ensured by adding a few drops of fresh acid at room temperature and monitoring effervescence.The acid was then washed out with 20 ml of DIwater in three iterations such that the residual washed sample was pH-neutral.The decarbonated powder was dried in the oven and then transferred into muffled scintillation vials.For the analyses, aliquots of powder were weighed into tin capsules (8 x 5 mm, Thermo Fisher) and analyzed by flash-combustion in an EA IsoLink with a ramped GC oven coupled via a Conflo IV to a MAT253 IRMS (Thermo Finnigan) (Stüeken et al., 2020).The EA was equipped with a dual-reactor setup with Cr2O3 at 1020 °C in the combustion reactor and Cu wire at 650 °C in the reduction reactor.Water vapor was trapped with a magnesium perchlorate column at room temperature.For samples with C/N ratios above c.120, the analyses were repeated with an additional CO2 trap packed with soda lime.The He flow in the EA was set to 180 ml/min for the first 110 seconds (valve V2 open) and then lowered to 50 ml/min (valve V2 closed) for the rest of the run.The GC oven was ramped from 35 °C to 240 °C after the elution of the N2 peak to improve the elution of CO2.Empty tin capsules were analyzed in each run, and their signal was subtracted from all samples and standards.The measurements were calibrated with the reference materials USGS-40 and USGS-41 and expressed in delta notation as δ 15 Nbulk = ([ 15 N/ 14 N]sample/[ 15 N/ 14 N]air -1) x 1000, and δ 13 Corg = ([ 13 C/ 12 C]sample/[ 13 C/ 12 C]VPDB -1) x 1000.A series of USGS-41 with different weights was used to calibrate peak areas for total nitrogen (TN) and total organic carbon (TOC) concentrations.USGS-62 was used for quality control and gave a long-term reproducibility of 0.46 ‰ (2 S.D.) for both δ 15 Nbulk and δ 13 Corg and better than 10% relative error for TN and TOC abundances.The measured average values of +20.21 ‰ for δ 15 Nbulk and -14.96 ‰ for δ 13 Corg are in good agreement with expected values (+20.17 ± 0.06 ‰ and -14.79 ± 0.04 ‰, respectively, Coplen, 2019).Of the 125 samples, 38 were analyzed in replicates, and their average reproducibility was 1.02 ‰ (2 S.D.) for δ 15 Nbulk, 0.92 ‰ (2 S.D.) for δ 13 Corg, and relative error for TN was 11% and 9% for TOC.For samples with total N peak areas of less than 8 Vs, δ 15 Nbulk could not be measured reliably (SD > 2 ‰) and is therefore not reported.Our results are supplemented by 47 data from a lower-resolution study of the sedimentary strata in the same drill core by Kump et al. (2011).
For metal abundance measurements, aliquots of untreated powder were sent to Australian Laboratory Services (ALS) in Dublin, Ireland.Samples were analyzed by ICP-MS and ICP-OES, following digestion in HNO3, HF, HClO4, and HCl (method ME-MS61).Seven samples were analyzed in duplicates, and the average reproducibility (RE) was better than 3 % for major elements and better than 5% for all minor elements.Enrichment factors for Zn and Cu were calculated relative to average oceanic basalt (White & Klein, 2014) and normalized to Al as follows: EFbasalt = ([X]measured/[Al]measured)/([X]basalt/[Al]basalt), where [X] is the concentration of Zn or Cu, respectively.For K, which has been significantly affected by postdepositional metasomatism (see below), the measured abundance was first normalized to the sum of Al2O3 + CaO + Na2O + K2O and then the ratio calculated of measured K2O to the initial K2O of the assumed source rock.We assumed that the basalt that had only undergone mild alteration (here the interior of the lower mafic body at 412-368 m, Fig. 2a) can be used as a reference point for the mafic bodies and average upper Archean crust (Fedo et al., 1995) for the sedimentary rocks.Lastly, we calculated the provenance index, Psi, following the methodology of Lipp et al. (2020;2021).

Organic carbon and nitrogen
Our results (Table 1) are in good agreement with those of Kump et al. (2011), where data overlap (see Figs. 3, 4).The organic carbon isotope data also agree well with measurements by Qu et al. (2012).Within the sedimentary horizons, TN broadly increases upsection, roughly concurrently with total K enrichments (Fig. 3).Nitrogen is usually introduced into sediments through biomass burial because it is a major constituent of proteins, DNA, and RNA in cells.However, in several places in these samples the behavior of nitrogen across the core contrasts with TOC as reflected by highly variably C/N ratios: (1) Between the lower and middle mafic bodies at 414-367 m, TOC decreases, while TN remains roughly constant or perhaps slightly increases.This small increase in TN parallels an increase in δ 15 Nbulk from 1.8 ‰ to 6.6 ‰.
(2) Above the middle mafic body, TOC decreases by two orders of magnitude from high values up to 11% in mudstones (310-270 m) to 0.1% in the sandstone-rich section at 200-240 m, whereas TN increases by one order of magnitude from a few hundred µg/g in the mudstone to a few thousand µg/g in the sandstone.Both shifts are statistically significant (p < 0.005) and explain the drop in C/N ratios in this interval.δ 15 Nbulk increases from -1.68 ‰ to 8.10 ‰ across this interval and continues to increase to as high as 9.7 ‰ at 120 m.
(3) Unusually high TOC values of up to 40% occur between 136-156 m, which has previously been interpreted as a fossilized oil spill onto the seafloor (Qu et al., 2012).TN abundances are high in this section but comparable to adjacent beds.In contrast, δ 15 Nbulk values show a drop of several permil within the oil spill horizon to as low as 2.26 ‰.
(4) Within the mafic bodies, TN is most enriched relative to TOC, that is, C/N ratios are consistently low.In the lower mafic body (482-416 m), which likely represents a postdepositional intrusion (Črne et al., 2013), both elements show enrichments at the top; TN is also enriched at the base.In the middle and upper mafic bodies (368-315 m, 119-113 m, and > 94.5 m), no spatial patterns are evident, but the highest TN abundances are comparable to those of adjacent sedimentary beds, whereas TOC is consistently below the sedimentary values.δ 15 Nbulk within the lower and middle mafic bodies clusters around a mean of 5.3 ± 1.4 ‰, similar to some of the sedimentary strata in between.Also in the upper mafic bodies, δ 15 Nbulk values of up to 9.1 ‰ match those of surrounding sedimentary strata.
Within the mafic bodies, TN shows moderate to strong correlations with K/Al ratios (r 2 = 0.3 with p = < 10 -5 for the lower mafic body at 481-416 m; and r 2 = 0.8 with p = < 10 -5 for the middle mafic body at 353-323 m, Fig. 5a) and no correlations with TOC (r 2 = 0.1 and 0.0, respectively).Within the sedimentary strata only, TN is moderately correlated with K/Al (r 2 = 0.3, p = 10 -8 , Fig. 5b).Relative to organic carbon, sedimentary TN shows two populations of data where one group (mostly from below 250 m and above 155 m) shows a strong covariance with TOC, driven by samples with > 15 % TOC, and the second group (mostly from 250-155 m) is uncorrelated (Fig, 5c).

Metal abundances
All metal abundance data generated in this study are shown in Table 2.In terms of the provenance indicator Psi (Lipp et al., 2020;Lipp et al., 2021), the mafic bodies fall close to the mafic endmember, as expected (Fig. 4).This mafic signature is also consistent with the mineralogy, which is dominated by plagioclase and pyroxene in the least altered interior of the lower mafic body (Fig. 2a).The sedimentary strata up to about 250 m broadly reflect this mafic signature before shifting to more felsic compositions above 250 m, likely reflecting a change in provenance.The lower mafic body (484-414 m) shows strong enrichments in K in its top and bottom relative to the weakly altered interior.This trend parallels a mineralogical shift from plagioclase and moderately altered pyroxene in the interior (Fig. 2a) to highly altered pyroxene and replacement of all plagioclase by phyllosilicates at the top (Fig. 2b).Concurrent with these K-enrichments, Cu is slightly depleted, but Zn remains unchanged or shows subtle change only.The middle mafic body (367-315 m), which also shows highly altered residual pyroxene and complete replacement of plagioclase (Fig. 2c), is variably enriched in K as well as Cu and displays an overall slightly higher average for Zn and P.There are few data for the upper mafic bodies, but these are the most K-and N-rich samples.

Hydrothermal alteration and ammonium mobility
The Onega Basin experienced significant hydrothermal fluid flow that strongly altered the mafic bodies (Fig. 2, 4) (Črne et al., 2013;Paiste et al., 2020).The depletion of Cu from the lower intrusive mafic body (484-414 m), without concurrent depletion in Zn, is consistent with a highly saline hydrothermal fluid, such as those associated with sedimentary Cu ore deposits where Cu is mobilized from mafic source rocks by deep sedimentary brines (Hitzman et al., 2010).In the case of the Onega Basin, it is likely that such saline fluids were sourced from underlying evaporites as recovered in the Onega Parametric Core (Blättler et al., 2018).The middle mafic body (367-315 m), which is thought to represent a lava flow, is enriched in Cu and slightly enriched in Zn.The relative timing between the lava flow and the lower intrusive body is unknown.However, given the lack of Zn depletion in the intrusion (lower mafic body), a plausible source for these elements in the lava (middle mafic body) are the encasing sediments.In other words, hydrothermal fluids likely mobilized Cu, Zn, and other elements from the sedimentary strata and at least in part deposited them within the lava flow as it underwent alteration.
In the lower mafic body (484-414 m), K is enriched in the same intervals that show Cu depletion.This K-enrichment likely reflects the formation of K-feldspar and sericite, which is a typical alteration phase in hydrothermal settings (Robb, 2005).The K can be derived either directly from seawater (e.g., Brauhart et al., 2001) and/or from the breakdown of K-feldspar initially contained in now K-depleted sediments that sit above or below the mafic body (e.g., Fedo et al., 1995).The strong correlation between K and TN in the altered mafic rocks and the high TN concentrations compared to typical values of a few µg/g in unaltered basalts (Johnson & Goldblatt, 2015) indicate that this K-rich fluid was also enriched in ammonium (NH4 + ).Ammonium and K + have the same charge and a similar ionic radius, and therefore they can substitute for one another in mineral lattices (Busigny & Bebout, 2013).For the Zaonega Formation, this process of ammonium incorporation into phyllosilicates was previously proposed by Kump et al. (2011).The most likely source for the ammonium are the organicrich sedimentary strata, as exemplified by modern hydrothermal fluids that are typically ammonium-rich if sediments occur along the fluid flow path (Lilley et al., 1993).A similar transfer of material from organic-rich sediments into the altered intrusion (lower mafic body) has been documented for sulfide (Paiste et al., 2020).If the TN contents of all samples along the core are compared to their respective TOC contents and an assumed C/N ratio of 10 for initial biomass (Godfrey & Glass, 2011), it turns out that the most organic-rich samples have lost N, while the most organic-poor samples have gained it (Fig. 6c and d), which further supports the idea of TN transfer from organic-rich sedimentary strata into altered mafic rocks.
Importantly, the TN-gain is reflected not only in altered mafic rocks but also in some sedimentary strata.Major element systematics (Fig. 7b) indicate that significant parts of the sedimentary horizons have also undergone K-metasomatism, as sedimentary strata above ~250 m show K-enrichments (Fig. 4, 7).Hence, the relatively high TN concentrations above 250 m (Fig. 3), despite variable TOC levels, reveal that fluids enriched in K and ammonium (and possibly other elements such as Cu and Zn) circulated through most of the sedimentary strata and mafic bodies of the basin.

Quantifying the hydrothermal ammonium load
The TN data from the altered mafic rocks can provide an estimate of the ammonium concentrations of those fluids.When ammonium is incorporated into secondary phyllosilicates, it adsorbs onto the mineral surface and is taken up into the interlayers of the mineral lattices, as observed in modern sedimentary pore waters where ammonium is incorporated into diagenetic illitic clays (Schroeder & McLain, 1998).Boatman & Murray (1982) derived a correlation between the fluid and mineral N content as a result of this process: Nclay = aNH4+/([1/Kads] + aNH4+) ‧ Γ ‧ MN, where Nclay is the amount of adsorbed ammonium in g/g, aNH4+ is the activity of ammonium in the fluid (i.e., the molar concentration multiplied by an activity coefficient of 0.57 for seawater), MN is the molar mass of nitrogen (14.01 g/mol), Kads is an empirically determined adsorption constant (5.1 L/mol for average mud), and Γ is the cation exchange capacity for average clay (0.96 mmol/g).Their formulation successfully reproduces measured concentrations of modern marine settings (Rosenfeld, 1979;van Raaphorst & Malschaert, 1996).Applying their calibration to the TN-rich altered mafic bodies of the Zaonega Formation, we find that dissolved ammonium levels were as high as 50 mM.These high levels are comparable to modern anoxic pore waters in organic-rich sediments and black smoker fluids (Rosenfeld, 1979;Boudreau & Canfield, 1988;Lilley et al., 1993).For comparison, less extreme marine sedimentary pore waters may contain ammonium levels in the sub-mM range (Caetano et al., 1997;Engström et al., 2005).Hence, such high levels are indicative of the degradation of large amounts of biomass.However, these concentrations would have exceeded the levels one would expect to find at the bottom of restricted marine basins as a result of anaerobic biomass degradation in the absence of hydrothermal processes.This is exemplified by the modern Black Sea, where ammonium levels are approximately 0.1 mM and hydrothermal vents are absent (Brewer & Murray, 1973;Konovalov et al., 2006).Hence, the amount of ammonium liberated by maturation of organic matter passing through the oil window and by circulation of hydrothermal fluids in the Onega basin was likely well above those seen in volcanically quiescent settings of that time.
Ammonium incorporation into clays is affected by salinity and pH (Kithome et al., 1999;Hou et al., 2003), where ammonium incorporation decreases with decreasing pH and increasing salinity.Because the hydrothermal fluid in the Onega Basin was more saline and acidic than modern seawater, given the documented presence of evaporites deeper in the stratigraphy (Blättler et al., 2018), we would thus be underestimating the true amount of ammonium in the fluid.

Bioavailability of fluid-mobile ammonium
To evaluate the significance of these ammonium-rich fluids for the Paleoproterozoic biosphere, a key question concerns the timing of fluid flow relative to sedimentation and biological activity.Two lines of evidence suggest that at least some of the ammonium-rich fluids emanated into the water column of the basin and probably created locally nutrient-replete conditions as follows: (1) The presence of the syn-sedimentary oil spill, which is most prominent at 156-136 m and begins to show sedimentary expressions in the form of tar balls above 170 m (Qu et al., 2012) (Fig. 2d), and (2) the isotopic composition in δ 15 N seen in the upper part of the core (Fig. 3, 5).
With regard to point (1), the oil spill interval and bracketing strata contain tar balls that strongly suggest emplacement of the oil on the seafloor, rather than post-depositional emplacement within the sediment pile, and organic carbon isotope data suggest that this oil was derived from organic matter from deeper in the succession (Qu et al., 2012).Oil forms during the maturation of buried biomass, typically as organic-rich sediments are buried to a depth of the "oil window" at the depth >1500 m, where temperatures of > 70 °C lead to the release of hydrocarbon molecules.In organic-rich sediments of the Zaonega Formation, the oil formation and migration was initiated possibly at much shallower burial depths by intrusion of mafic magmatic bodies (Qu et al., 2012).Importantly, the oil window also marks the point where most ammonium is released from degrading organic matter, as indicated by several studies of younger oil source rocks (Williams et al., 1989;Williams & Ferrell, 1991;Williams et al., 1992).It is, therefore, highly likely that the generation of this oil spill also marked the time when large quantities of ammonium were released into warm aqueous fluids that migrated through the sediment package of the Zaonega Formation and seeped into the water column, along with the methane seep (Qu et al., 2012).Evidence of microbial methane uptake (methanotrophy) in the form of low δ 13 Corg values is most conspicuous from the top of the core, but it is also apparent immediately above and below the oil spill interval, where δ 13 Corg values are lighter than the oil spill itself, which is consistent with methanotrophy concurrent with ammonium enrichment.The very low δ 13 Corg values at the top of the core may indicate higher oxidant availability (O2 or sulfate) that further boosted methanotrophy at that time (see also below).
With regard to point (2): Ammonium release from organic matter by hot fluids and uptake of ammonium into phyllosilicates are not associated with any significant isotopic fractionation (< 1 ‰) (Ader et al., 2006;Boudou et al., 2008;Koehler et al., 2019), yet the measured δ 15 N values across the core spread over a range of more than 10 ‰.In a previous study, the least altered organic matter was identified at 400-370 m, beneath the middle igneous body (Qu et al., 2012).Here, δ 15 N values fall between 4-6 ‰, which is similar to the overlying middle mafic body.Since hydrothermal mobilization and uptake of ammonium into clays does not significantly fractionate δ 15 N (Ader et al., 2006;Boudou et al., 2008;Koehler et al., 2019), it seems plausible that ammonium with an isotopic composition of 4-6 ‰ was mobilized from biomass in the 400-370 m interval and transferred into the overlying middle mafic body by hydrothermal fluids without any modification of the δ 15 N value.Within the middle mafic body at 367-315 m, δ 15 N is fairly homogeneous, which also suggests the absence of any significant fractionation during the transfer of ammonium from the underlying sediments into this altered lava flow.
In contrast, the highly altered organic matter above the middle mafic body at 315-270 m is isotopically light in δ 15 N, as low as -1.7 ‰.The high C/N ratios of these rocks suggest that they are dominated by organic-bound N with relatively little secondary N-enrichment in clays.A scatter plot of δ 15 N versus C/N ratios indicates a tendency that the lowest δ 15 N values near 0 ± 2 ‰ occur in the samples with the highest C/N ratios (Fig. 5d), which are the most Ndepleted, organic-rich sedimentary rocks, particularly at 315-270 m depth.The highest δ 15 N values (8-10 ‰) occur within sedimentary rocks immediately above and below the oil spill (155-135 m), which itself falls between 2-5 ‰.Hence, it appears that, to first order, the altered sedimentary rocks from 315-120 m and above are composed of two nitrogen endmembers: an isotopically light endmember with high C/N ratios (i.e., organic-bound N) and an isotopically heavy endmember with low C/N ratios (i.e., TN-enriched phyllosilicates).Using the kerogen isolate data from Kump et al. (2011) for this part of the core as a proxy for the organic endmember, one can calculate the isotopic offset between the organic-bound and silicate-bound fraction to be between 3-10 ‰ with some outliers up to almost 20 ‰ (Fig. 6).
It might be argued that this divergence is explicable due to regional metamorphism.Previous work on metasedimentary rocks of various grades revealed an increasing divergence between kerogen and associated silicates from prehnite-pumpellyite to amphibolite facies (Stüeken et al., 2017;Kipp et al., 2018).For the rocks of the Zaonega Formation, which have undergone regional metamorphism up to lower greenschist facies, an isotopic difference of around 2-4 ‰ between the two nitrogen phases can be expected.The reason for this metamorphism-driven isotope offset is likely related to differing bond strengths between the organic and inorganic nitrogen reservoirs in metasedimentary rocks.However, in the case of the Zaonega Formation, the isotopic difference between kerogen and silicates is, on average, larger than expected, and the two nitrogen phases are spatially well separated by many meters, as suggested by the large variability of C/N ratios across the core (Fig 3).A good illustration of this spatial separation is the interval around the syn-sedimentary oil spill, where the Ndepleted oil spill itself is isotopically lighter than the N-enriched bracketing sediments, which indicates that the isotopic divergence pre-dates regional metamorphism.It thus appears that, in this part of the section, the hydrothermally mobilized ammonium was isotopically fractionated after being released from organic matter and prior to incorporation into phyllosilicates.
In modern low-temperature (< 50 °C) hydrothermal fluids from the Juan de Fuca Ridge, Bourbonnais et al. (2012) described δ 15 N values of up to 12 ‰ in dissolved ammonium.Those values were up to 8 ‰ higher compared to background sediments and high-temperature fluids.They concluded that this isotopic enrichment in 15 N in the dissolved ammonium had been imparted by partial biological ammonium uptake or by partial biological oxidation of ammonium to nitrite and nitrate (nitrification).In the former case, the complementary isotopically light phase would have been homogenized with other biomass.In the case of nitrification, the isotopically light nitrogen would be the resulting nitrate vented into the open ocean.From the hydrothermally active Guaymas Basin, Campbell et al. (2013) documented a high abundance of anaerobic ammonium oxidizers (ANAMMOX) near diffuse vents, which could also contribute to this isotopic effect.We propose that the large δ 15 N range observed in the upper part of our core through the Zaonega Formation reflects similar biological processes.The calculated δ 15 N values for the silicate phase of 3-10 ‰ and occasionally up to 20 ‰ (Fig. 6) is likely reflective of the dissolved ammonium in the hydrothermal fluid.Such 15 N-enriched values compared to associated organic-bound N are not consistent with known abiotic processes such as hydrothermal ammonium mobilization (Ader et al., 2006;Boudou et al., 2008), clay intake (Koehler et al., 2019), or regional metamorphism (Stüeken et al., 2017;Kipp et al., 2018), but are consistent with biological metabolisms acting on the dissolved ammonium pool.For example, partial biological uptake of ammonium is associated with a fractionation of up to 27 ‰ (Hoch et al., 1992), which could explain both the 15 N-enrichment in residual ammonium and the 15 N depletion in organic matter.Alternatively, nitrification (fractionation of 14-38 ‰ from NH4 + to NO2 -) (Casciotti, 2009) could have generated 15 N-enriched ammonium and a 15 N-depleted nitrite (and nitrate) pool that was assimilated elsewhere.The latter interpretation would be in line with that of Kump et al. (2011), who also interpreted the progressive increase in δ 15 N up section as evidence of increasingly oxic conditions, and it would be consistent with anomalously low δ 13 Corg at the top of the core that has been interpreted as evidence of methanotrophy (Qu et al., 2012).Although the nitrogen data alone do not allow for distinction between ammonium assimilation and nitrification, the latter is perhaps favorable in light of the carbon isotope data.In fact, the inferred bioavailability of ammonium in the upper half of the core, coinciding with a progressive depletion in δ 13 Corg values (Fig. 1), indicates that ammonium was biologically metabolized in parallel with methane in the Onega basin, and both substrates were likely provided by volcanic and hydrothermal activity.In other words, local hydrothermal activity created not only methane-rich but also ammonium-rich conditions by recycling this nutrient from older buried biomass.

Conclusions
In summary, our data from the Paleoproterozoic Onega Basin in NW Russia show strong evidence of hydrothermal ammonium remobilization from organic-rich sedimentary strata concurrent with previously inferred mobilization of hydrocarbons, including methane.Mafic intrusions provided a heat source that took organic matter through the oil window and initiated ammonium release into fluids.Further evidence for such ammonium-rich fluids comes from pervasive K-metasomatism in the hydrothermally altered mafic rocks and in the upper sediment package of the core.The ammonium that was liberated by this process likely reached concentrations of several millimole per liter, as indicated by the high N-enrichments in the altered mafic rocks.Such high concentrations are consistent with modern pore waters, but they exceed the concentrations found in volcanically quiescent anoxic basins like the Black Sea.This finding highlights the ability of hydrothermal activity to drive ammonium mobilization at a large scale in the Paleoproterozoic Onega basin.Furthermore, the isotopic divergence between silicate-bound and kerogen-bound nitrogen in these rocks is most parsimoniously explained by biological fractionation imparted on the dissolved ammonium reservoir of the hydrothermal fluid.In other words, we see evidence that the hydrothermal ammonium was bioavailable and created locally N-replete conditions.It is conceivable that this nutrient enrichment enhanced biological productivity and biomass production, which may have resulted in positive feedback on the ammonium recycling flux.
In the modern ocean, where most nutrients are recycled by oxidative degradation of biomass in the oxic water column, hydrothermal nutrient sources are minor on a global scale.However, in the Precambrian, when the deep ocean was anoxic, this hydrothermal recycling mechanism could have locally offset the absence of water column recycling and created nutrient-rich refugia for microorganisms.In fact, we speculate that isotopic offsets between kerogen and silicates that have previously been documented from other succession may at least in part also reflect biological activity (Godfrey et al., 2013;Stüeken et al., 2017;Kipp et al., 2018).Evaluating this possibility will require more detailed investigation of each setting.In any case, it is likely that similar processes would occur on other planetary bodies where hot fluids interact with buried biomass.Thus, hydrothermally active basins are not only ideal sites for an independent genesis but perhaps also for the sustenance of a diverse biosphere.

Figure 2 :
Figure 2: Photomicrographs of selected hand samples.(a) Weakly altered interior of the lower igneous body.Plagioclase (elongated grains with low birefringence) is preserved, pyroxene (medium birefringence poikilitic grains around plagioclase) is partly altered to phyllosilicates, including sericite and chlorite (fine-grained material in the bottom-left corner of the image).(b) Strongly altered upper part of the lower igneous body.Plagioclase is completely replaced and pyroxene is mostly replaced by secondary phyllosilicates.(c) Highly altered middle igneous body, similar to (b).In all three images (a-c), the blocky, nearly opaque grains are sphene, reflecting retention of Ti during alteration.(d) Tar balls (black rounded particles with low relief) within silty, calcareous shale.The shale matrix is itself fragmented (darker and lighter patches).The scale bar (bottom right corner in each image) is 250 µm in (a) and 500 µm in (b), (c) and (d).Xpl = crossed polars, ppl = plane-polarized light.

Figure 3 :
Figure 3: Stratigraphic patterns of total nitrogen (TN), total organic carbon (TOC), organic carbon isotopes, bulk nitrogen isotopes and molar C/N ratios.Orange circles = igneous bodies from this study, yellow circles = organic-rich veins within mafic rock, dark blue circles = sedimentary rocks from this study, light blue squares = sedimentary rocks from Kump et al. (2011), purple triangles = sedimentary rocks from Qu et al. (2012).Light green bands highlight data from the lower, middle and upper mafic bodies.

Figure 4 :
Figure 4: Stratigraphic patterns of total nitrogen (TN, same as in Fig. 2), potassium enrichment factors, accounting for weathering effects (K [EFwx]), zinc (Zn) and copper (Cu) enrichment factors relative to basalt (EFbasalt), the provenance index Psi (see text for details).Some additional K data are taken from Kump et al. (2011).The vertical dashed lines in the Psi plot indicate the average value for upper continental crust UCC).Black lines in the K and Cu plots highlight enrichments and depletions, respectively, in the mafic intrusive body due to fluid alteration.The small black arrows in the Zn and Cu plots indicate positions of outliers with higher enrichments that are not shown on the graph.Zn and Cu abundances are normalized to average oceanic crust (White & Klein, 2014).Light green bands highlight data from the lower, middle and upper mafic bodies.

Figure 5 :
Figure5: Scatter plots of total nitrogen versus potassium/aluminium ratios for igneous rocks (a), total nitrogen versus potassium/aluminium ratios for sedimentary rocks (b), total nitrogen versus total organic carbon (c), and bulk nitrogen isotopes versus molar C/N ratios (d).Panels (c) and (d) include all data from both igneous and sedimentary strata, following the same colour code as in Fig.2.In panels (a) and (b), data are separated by stratigraphic depth.

Figure 6 :
Figure 6: (a) Total nitrogen along the core.(b) Nitrogen isotopes for differing phases(Kump et al., 2011).The silicate data were calculated by mass balance from the measured data for kerogen extracts and bulk rock.Relative abundances of organic-and silicate-bound N are highly variable along the core without particular stratigraphic trends and are therefore not plotted.(c) Nitrogen loss and gain (both expressed as positive numbers) for all samples, calculated from measured TOC and TN abundances and an assumed C/N starting value of 10.(d) Same as (c) but plotted in percent relative to the initial abundance of nitrogen in the given sample.Light green bands highlight data from the lower, middle and upper mafic bodies.

Figure 7 :
Figure 7: Ternary diagrams for igneous bodies (a) and for sedimentary rocks (b).Average Archean crust and expected alteration trends are taken from Fedo et al. (1995).The sedimentary rocks are corrected for Ca-contributions by calcite and phosphate following the procedure of Medaris et al. (2015).

Table 1 :
Organic carbon and total nitrogen data generated in this study.

Table 2 :
Metal abundances generated in this study.