Introduction

Reactive nitrogen’s (NR) vital role in the origin of life and predicting the potential for life on other planets has led to decades of work investigating clues to NR’s origins and source magnitudes on early Earth. In contrast to modern environments where a primary NR species is nitrate, early Earth’s reducing environments were dominated by ammonia/ammonium (NH3/NH4+), until the Great Oxidation Event (~ 2.3 Ga)1. Researchers have postulated several abiotic NH3 production scenarios including direct creation of NH3/NH4+ via hydrothermal and photochemical reduction or by fixation processes (e.g., volcanic, lightning) creating oxidized N species followed by reduction2,3,4,5. However, investigations into the potential magnitude of abiotic production mechanisms suggest they were too minor to sustain the Archean biosphere thus implicating an early biotic source6. Geological evidence and the presence of NH4+ enriched phyllosilicates suggest biotic nitrogen fixation originated as early as 3.8 to 3.95 Ga and isotopic and genomic evidence supports nitrogen fixation using molybdenum-based nitrogenase occurring as early as 3.2 Ga7,8,9,10,11,12,13,14,15. Whether abiotic or biotic-sourced, continuous flux of NR to early habitats would have been essential for the origin and proliferation of life. These transport mechanisms (e.g., subduction, runoff, atmospheric deposition, hydrothermal interactions, water–air flux) are as important as the source of NR itself.

The isotopic composition of nitrogen in the rock record provides evidence of paleo NR sources and processing at the time of deposition and allows reconstructions of plausible evolutionary pathways for Earth's biogeochemical N cycle. Focusing on the Archean (4.0 to 2.5 Ga), the δ15N values in the rock record vary greatly (− 10.2 to + 50.4‰) (Supplementary Fig. 1)16 and researchers have proposed various mechanisms, both biotic and abiotic, which would have produced this range. Stüeken et al., Ader et al., and references therein16,17, provide a detailed account of potential scenarios which resulted in the Archean data set. In summary, the lower δ15N values (down to − 4‰) in the Paleoarchean (4.0 to 3.2 Ga) have been described as representing NH4+ assimilation by thermophilic microbes, metasomatic alteration, or biological N2 fixation via V or Fe nitrogenases (rather than Mo) while higher δ15N values (up to 12.2‰) were interpreted as metamorphic overprinting or incomplete biological processing17,18. During the Mesoarchean (3.2 to 2.8 Ga), a tight range in δ15N values (+ 1.1 ± 1.9‰) across the majority of the data has been suggested as representing Mo N-fixation11,19, thus evidencing the nitrogenase enzyme had evolved by this period. The noted excursion from these consistent values is the very high δ15N values (up to 50‰) at the approximate transition of the Meso- to Neoarchean era. These high values could not be explained by potential post-depositional alteration and was originally interpreted as reflecting the onset of nitrification and its fractionation effects20. A subsequent investigation proposed the high δ15N values are a product of large fractionation associated with volatized NH3 from alkaline lakes21. The consistently positive δ15N values across the Neoarchean (2.8 to 2.5 Ga) data set along with ancillary evidence of oxygenation are purported to implicate the onset of competing nitrification and denitrification during this period.

These are plausible explanations for the existing Archean N isotope rock record, which is focused on marine and lacustrine sedimentary rocks. No nitrogen isotopic studies have so far been conducted on Archean terrestrial hot springs, although such sites may have been crucial for the origin and early evolution of life. Here, I present a theoretical framework for nitrogen cycling in such terrestrial hydrothermal settings. I find that NH3 outgassing could have produced a wide range of δ15N, largely driven by temperature rather than pH. I employ a large data set of the chemical and physical components of modern hydrothermal features to produce evidence that the water–air NH3 flux of equivalent Archean features could have been a significant source of bioavailable N to Archean life in neighboring sites that were NH3-undersaturated and thus able to absorb NH3 from the atmosphere. This water–air flux could be a key component of early Earth’s chemical reactor in which separate environmental settings were linked through physical exchange of reactants and products2,22.

Results and discussion

Water–air ammonia flux of modern hydrothermal features

The water–air NH3 flux of 1022 hydrothermal features (terrestrial hot springs) across the globe were modeled according to previous approaches23,24,25,26,27 using available literature values of NH4+(aq), salinity, pH, temperature, wind speed and NH3 atmospheric concentrations (SI Methods; Supplementary Table 1, Fig. 1). Hydrothermal settings can lead to aqueous systems with wide ranging chemical and physical characteristics which are well represented in the data set. Temperature, pH, salinity and NH4+ concentration ranges were 298 to 443 K (332 ± 20), 1.2 to 9.7 (4.9 ± 2.3), 0 to 101 ppt (1.4 ± 4.3) and 0 to 46,400 (555 ± 2,764 µM), respectively. Modeled NH3 fluxes ranged from − 0.003 to 626 kg N m−2 year−1 with negative values indicating a NH3 flux into the feature and positive values indicating a flux to the atmosphere. The average flux was 2.1 kg N m−2 year−1, with an outlier removed average flux of 1.2 kg N m−2 year−1 (outliers were determined using inter-quartile range) and the lower 95% of data average of 0.2 kg N m−2. The ability of hot springs to produce these high NH3 fluxes have been previously evidenced from the NH3 concentration of gases collected directly from two springs in Yellowstone Park, USA (0.1 and 0.7 mol% NH3(g))28,29. NH3 fluxes are often associated with alkaline waters as a higher pH increases the amount of NH4+ partitioning to NH3. However, with widely varying T the portion of NH3 available to flux can be significant even at low pH (Fig. 2A). For instance, at 25 °C and 100 °C the pH at which the NH4+ to NH3 ratio is 50:50 is 9.2 and 7.4, respectively. If this 75 °C increase is applied across the data set, the average flux increases by a factor of 4.2. This has been recounted in a previous sensitivity study which reports T as the primary parameter driving NH3 water–air flux24. For comparison, if a similar simple sensitivity test is performed on this data set where pH for all sites is increased by 1 unit (i.e., 1 order of magnitude for [H+]), the average flux only increases by a factor of 1.3. While previous studies have pointed to a need for alkaline conditions to increase available NH3, temperature can greatly increase available NH3 and alkaline conditions are not a prerequisite for water–air flux.

Figure 1
figure 1

Locations of the 1022 hydrothermal features used as modern analogues to estimate Archean water–air NH3 flux. Data and references are included in the Supplemental Information.

Full size image
Figure 2
figure 2

(A) Fraction of NH3 and NH4+ in aqueous solution with respect to pH and temperature. (B) The δ15N values of NH4+(aq) residual, NH3(aq) instantaneous, and NH3(aq) accumulated and are modeled with respect to Rayleigh fractionation and the fractionation effect (35.6‰) associated with the average temperature of the observed hydrothermal features (333 Kelvin). Additional less significant fractionation occurs during outgassing (NH3(aq) to NH3(g)) and is not included in the model here but the effect is calculated as 6.5‰ at 333 Kelvin (SI)30. A starting NH4+ pool of 0‰ was chosen as the δ15N in sedimentary rocks may be representative of the N2 fixing organisms present in the early Archean and the δ15N of the sedimentary record between 3.2 and 2.5 Ga falls around values of 0‰31.

Full size image

To portray the significance of the modeled flux from these hydrothermal features, I determined the portion of Archean Earth surface required to be covered by terrestrial hydrothermal features in order for the mean and outlier-removed mean water–air flux to equal the estimated (1) Phanerozoic land to ocean N flux, (2) modern rock weathering N flux, (3) “modern pre-industrial” atmospheric deposition flux and (4) modern biological fixation flux (Table 1). With respect to modeled outlier removed average flux (1.2 kg N m−2 year−1), the Archean Earth’s surface would need to be 0.017% hydrothermal features to match the amount of biogenic N fixation on modern Earth. For reference, the combined area of these surficial features would be equivalent to the size of Serbia or the state of Minnesota, USA; a feasible scale relative to the volcanic and hydrothermal activity which would have been widespread during periods of the Archean32. While this flux could be seen as a first order loss of bioavailable N from the hot spring environment, the transport and deposition to other environments would significantly complement previously hypothesized abiotic and biotic source mechanisms to provide bioavailable N on early Earth. The transport of this fluxed NH3 would be most impactful in nearby terrestrial or aquatic environments due to the direct relationship between concentration and deposition flux (i.e., F = νd × [NH3]). Terrestrial hydrothermal features with large NH3 water–air flux would create an atmosphere higher in NH3 and adjacent undersaturated hot springs or other aquatic settings would readily act as a sink for NH3. For instance, the modern ocean can act as a source or sink of NH324,25,33 and an Archean ocean adjacent to hot springs may have acted as a sink. Long distance transport of NH3 flux may have been hampered due to UV photolysis but this is dependent on the potential presence of a protective organic haze which may have shielded NH3 from photolysis34,35,36. This is a scenario which is still debated in the literature2,7,32.

Table 1 Water–air flux NH3 estimates and the corresponding percentage of Archean Earth coverage needed for the flux to equal Phanerozoic N land-to-ocean export flux and modern rock weathering, biological fixation and pre-Industrial Revolution atmospheric deposition N flux.
Full size table

Archean hydrothermal waters would have been NH4+ poor and unable to produce a large NH3 flux unless they had circulated through sedimentary formations featuring older organic matter40,41. For the water–air NH3 flux to be as pronounced as those from modern analogs, previous biotic and/or abiotic processes would have needed to supply substantial NH4+ to Paleoarchean sediments. As abiotic sources were estimated to be too low to produce enough NR for early life and subsequent organic deposition, the biotic source previously substantiated by high NH4+ phyllosilicates and isotopic and genomic evidence of nitrogen fixers in the Archean would have been required by the Paleoarchean. This would mean biotic-sourced N that was previously assimilated and deposited as organic matter in lithified sediments would be supplied to the features through hydrothermal remobilization in quantities large enough to produce significant water–air NH3 fluxes. This mechanism creates up to millimolar NH4+ concentrations found in modern hot springs and previous studies date this hydrothermal recycling process of organic bound N back to at least 3.24 Ga42. It must be noted that these Archean studies refer to hydrothermal cycling in the deep ocean so is not directly applicable to terrestrial hot springs but rather provides evidence for hydrothermal recycling of organic bound during this period.

Potential isotopic signature for hydrothermal water–air ammonia flux

The early Earth rock record has a wide range of δ15N (− 10.2 to 50.4‰) compared to the tighter range in modern sediments that is attributed to biological mediation (Supplementary Fig. 1). The wide range is a product of competing abiotic and biotic processes with the higher end values ascribed to incomplete biological N processing and/or NH3 volatilization from alkaline paleolakes20,21. With respect to rock records potentially produced in terrestrial hydrothermal settings, here, I present evidence that a wide range of δ15N can be a result of fractionation effects associated with terrestrial hydrothermal features and processes. For instance, a previous isotopic investigation of modern hot springs reports a similar range of δ15N-NH4+ values in water samples and sediment samples of − 6 to 30‰ and − 10 to 23‰, respectively28. The higher portion of the δ15N range in these modern springs was produced by preferential partitioning and volatilization of 14NH3. Deng et al., state these systems are not static and can be regarded and modeled as open system30. The isotope effects are modeled (Fig. 2B) for a hypothetical Archean hot spring using a starting δ15N-NH4+(aq) value of 0‰ and a temperature-dependent isotope effect of 35.6‰ which corresponds to the average temperature of the modern hydrothermal features (333 K) (SI Methods)43. A starting NH4+ pool of 0‰ was chosen as the δ15N in sedimentary rocks as it would be representative of the N2 fixing organisms present in the early Archean in which a majority of δ15N values fall around 0‰31. Interaction between hot fluids and organic-bound N imparts minimal isotopic fractionation as N is remobilized44. Additional, less significant fractionation, which would further enrich the aqueous NH3, occurs during outgassing (NH3(aq) to NH3(g)) and is not included in the model here but for reference the effect is calculated as 6.5‰ at 333 K30. Model results display that the δ15N of the residual NH4+ which would subsequently be imprinted in the rock record encompasses the higher end positive δ15N represented in the Archean rock record, although it should be stressed that there is no known evidence of hydrothermal activity in the Neoarchean Tumbiana Formation where the highest δ15N values have been recorded. The scenario invoked herein may, however, apply in other, hydrothermally active settings on Archean land surfaces. The higher δ15N values would be concentrated in the immediate depositional environment while the lighter product of this volatilization would be deposited elsewhere, mixing with large pools of N and potentially lowering its δ15N. This scenario was suggested by Holloway et al.28 in which they explain the modern springs with lower δ15N values contain NH4+ derived from recondensation of 15N-depleted NH3(g) that was transported from sites of phase separation and partial NH4+ loss. These isotope effects followed by deposition and mixing with other NH4+ pools would lead to a wide range of δ15N depending on environmental settings and existing N pools.

Hadean and Archean volcano-hydrothermal systems were estimated to be 4 to 5 times more active than present day due to radiogenic heating32 lending the Archean Earth to having more prevalent hydrothermal features. While many of these features may have been submarine due the early Archean Earth resembling a “water world”45,46, modeled NH3 flux results reported here suggest only 0.0092% of the Earth is needed to host terrestrial hydrothermal features to rival the significance of present-day biogenic fixation. While investigations based on siliciclastic sediments propose some continental exposure as early as 3.8 billion years ago47, low but more pronounced exposure in the Meso- and Neoarchean is indicated by paleosol and Sr isotope studies45,48,49. Evidence of continental crust growth throughout the Archean, particularly associated with island arcs50,51, suggests a potential setting for the emergence of associated terrestrial hydrothermal features. These features may have been linked with hydrothermal systems circulating organic bound N in marine sediments, subsequently providing NH3 for water–air flux. A portion of the higher and wide ranging δ15N values in the Archean rock record are observed in the Pilbara Supergroup where hydrothermal features were a common environment including hot springs and hydrothermal vents52,53. While this specific nitrogen isotope fractionation pathway described in this study can so far not be linked to the existing δ15N record from the Archean as evidence of hydrothermal processes is lacking for sites with unusual δ15N enrichments, other formations, for instance the Dresser formation, Pilbara Craton, show evidence of surficial hydrothermal features including geyserite, sinter terracettes and mineralized remnants of hot spring pools/vents54 and may prove to have recorded these processes. However, it should be noted origins of features in this formation are still under debate and could be remnants of deep-sea hydrothermal systems55,56. While the 15N rock record data from the Archean is limited, this eon was affected by widespread volcanism and the associated hydrothermal systems interacting with N enriched lithified sediments could have produced a setting prime for water–air flux of bioavailable N to the Archean atmosphere.

Implications for early Earth NR processing

Hydrothermal features (e.g. hot springs) provide a concentrated source of diverse chemical species and an expansive range of physical environments to support a wide array of microbial communities57. These characteristics have led many researchers to investigate hot springs as an ideal setting for the origin of life. Experiments have shown key feedstock molecules for prebiotic chemistry can be produced in abundance in shallow and surficial hydrothermal systems58. This coupled with evidence that hydrothermal fluids actively leached Mo, a strict trace metal requirement for Mo-nitrogenase, from volcanic rocks in the Archean support hydrothermal environments as a potential origin of biological N2 fixation15,59,60,61,62. Regardless of whether life originated in these hydrothermal features, the NH3 flux from hot NH3-supersaturated springs to nearby cooler, NH3-undersaturated springs may have supplied the essential bioavailable N in these cooler settings to spur and sustain life. Characterizing the NH4+ source and transport mechanisms is fundamental to reconstructing the N cycle during the Archean, a period before widespread established oxygenic photosynthesis meant the cycle was driven by the interplay between diazotrophy, ammonium regeneration and assimilation7. Well preserved Archean sediments are not common, and of the few Archean rock record studies, the majority have been concentrated on low energy, deeper water settings. Future sampling focused on paleo-hydrothermal settings, including those younger than the Archean will be valuable in assessing the validity of these hydrothermal feature effects on the range of δ15N in the rock record. Including corroborating data (e.g., trace metals, Fe speciation redox indicators, C/N ratio, δ13Corg and δ13Ccarb and biomarker δ15N) in these studies will be essential to predict paleo-depositional environments and potential mechanisms creating the δ15N rock signatures63. These additional diagnostic tools could provide localized evidence of settings for significant water–air NH3 flux on early Earth.

I do recognize, as with most studies attempting to predict the production, transport and processing of N on early Earth, the Archean water–air flux estimates are based on the assumption that modern analogs are a reasonable proxy. The multiple lines of evidence presented in this work attempt to support this modern analog and suggest that the water–air NH3 flux was a prominent mechanism providing bioavailable N to neighboring NH3-undersaturated environments as part of the early Earth’s chemical reactor and may even be underestimated as the presence of O2 in the modern environment would reduce the amount of NH3 available for outgassing. Potential applications outside the scope of this current work include investigating whether this flux is substantial enough to produce a greenhouse gas effect to counteract the Faint Young Sun and whether remote sensing could be used to detect this potential biogenic signal on exoplanets64. The proposed δ15N rock signature originating from terrestrial hydrothermal features holds significant potential for the identification of analogous past and current features on other planets and their satellites. Geologically active satellites in our solar system, such as Europa, Ganymede, Enceladus and Titan, provide intriguing possibilities of hydrothermal systems and the presence of past water on Mars, combined with compelling geomorphological and mineral evidence, indicates the occurrence of previous hydrothermal activity65,66. For instance, the morphology of opaline silica deposits located near the Home Plate feature in the Columbia Hills of Gusev crater on Mars strongly suggests they are the product of sinter deposits resulting from hot spring activity67. However, the origin of these deposits remains a subject of ongoing debate. Researchers should consider this hydrothermal flux mechanism along with other hypothesized abiotic and biotic processes when applying the δ15N record to unravel the development of Earth’s biogeochemical N cycle and explore potential biosignatures on other planets.

Methods

A literature search was conducted via google scholar to obtain hydrothermal feature data which included the NH4+ concentration, salinity, temperature and pH needed to model the NH3 water–air flux. These prerequisite data were met for 1022 features (data and references in Supplementary Table 1) and flux calculations were based on previous models15,16,17,18,19. The direction and magnitude of the water–air flux of NH3 can be determined from the atmospheric concentration of NH3, a calculated atmospheric equilibrium concentration of NH3 and an exchange velocity. The difference between the atmospheric and equilibrium NH3 concentrations provides the direction of flux, with a positive value denoting water-atmosphere NH3 emission from the feature and a negative value denoting NH3 deposition to the feature. By multiplying that difference by an air-side exchange velocity, a rate of water-atmosphere NH3 flux can be determined19 (Eq. 1):

$${\text{F}}_{{{\text{NH3}}}} = {\text{ kg }} \times \, \left[ {{\text{NH}}_{{3}} \left( {{\text{eq}}} \right) \, {-}{\text{ NH}}_{{3}} \left( {\text{g}} \right)} \right].$$
(1)

FNH3 is the water-atmosphere NH3 flux (ng m−2 s−1), kg is the air-side exchange velocity (m/s), NH3(eq) is the calculated atmospheric equilibrium NH3 concentration (µg/m3), and NH3(g) is the atmospheric NH3 concentration (µg/m3). Full details about how each term was determined are located in SI.