Isotopic Evidence for an Aerobic Nitrogen Cycle in the Latest Archean

See allHide authors and affiliations

Science  20 Feb 2009:
Vol. 323, Issue 5917, pp. 1045-1048
DOI: 10.1126/science.1165675


The nitrogen cycle provides essential nutrients to the biosphere, but its antiquity in modern form is unclear. In a drill core though homogeneous organic-rich shale in the 2.5-billion-year-old Mount McRae Shale, Australia, nitrogen isotope values vary from +1.0 to +7.5 per mil (‰) and back to +2.5‰ over ∼30 meters. These changes evidently record a transient departure from a largely anaerobic to an aerobic nitrogen cycle complete with nitrification and denitrification. Complementary molybdenum abundance and sulfur isotopic values suggest that nitrification occurred in response to a small increase in surface-ocean oxygenation. These data imply that nitrifying and denitrifying microbes had already evolved by the late Archean and were present before oxygen first began to accumulate in the atmosphere.

All living organisms require fixed nitrogen for the synthesis of vital biomolecules, such as proteins and nucleic acids. Under low-oxygen conditions, nitrogen-fixing organisms meet this need by reducing dinitrogen gas (N2) to ammonium (NH4+), which is readily incorporated into organic matter. All other organisms rely upon the degradation of N2 fixers through ammonification to fulfill their nitrogen requirements. Although the evolved NH4+ is stable under anoxic conditions, the presence of O2 in the surface ocean promotes nitrification, the microbial oxidation of NH4+ to nitrite (NO2) or nitrate (NO3). These oxidized species are either assimilated by organisms or, under low-oxygen conditions, biologically reduced and ultimately released to the atmosphere. The latter process provides a conduit for loss of fixed N from the ocean and proceeds via denitrification, the stepwise reduction of NO3 or NO2 to NO, N2O, and finally N2, or by anammox, the coupling of NH4+ oxidation to NO2 reduction. Because unique isotopic fractionations are imparted during many of these transformations, the nitrogen isotopic composition δ15N (1) of organic matter preserved in ancient sediments provides information about the evolution of the N cycle. Here we report δ15N and total nitrogen (TN) measurements, as well as δ13Corg and total organic carbon (TOC) data, obtained at a resolution of approximately one data point per meter from ∼100 m of continuous drill core through the ∼2.5–billion-year-old Mount McRae Shale, Hamersley Group, Western Australia (2).

If the fixed N reservoir is a steady-state system, the δ15N of fixed N input and output will be equal (δ15Ninput = δ15Noutput). The isotope effect ϵ (3) imparted during fixed N input is approximated by the difference in δ15N between the atmospheric N2 source and the fixed N product (ϵinput = δ15Ndinitrogen – δ15Ninput, where δ15Ndinitrogen = 0‰), whereas that of the fixed N output is approximated by the difference in δ15N between the oceanic fixed N source and the N2 product (ϵoutput = δ15Nfixed N – δ15Noutput). Thus, the mean δ15N of oceanic fixed N is roughly equal to the difference in ϵ between the output and input processes (δ15Nfixed N = ϵoutput – ϵinput). In the modern ocean, N2 fixation is the primary source of fixed N, and denitrification the primary sink. Although fractionation during N2 fixation is minimal, with ϵN2 fixation = –3 to + 4‰ (4, 5), the isotope effect of denitrification from both NO3 and NO2 is large, with ϵdenitrification = +20 to +30‰ (4). This results in a residual NO3 pool substantially enriched in 15N. Thus, the +5‰ mean isotopic value of modern deep-ocean NO3 is attributed to the fractionation imparted during denitrification. Furthermore, because fixed N upwelled to the surface ocean is completely consumed by primary producers under most conditions (6), the δ15N value of organic matter in well-preserved ancient sediments should record the mean δ15N of oceanic fixed N.

The organic-rich Mount McRae Shale was deposited at 2.5 billion years ago (Ga), shortly before the main rise in O2 2.45 to 2.22 Ga (7). Its geological setting, the sampling procedure, and analytical methods are described in the Supporting Online Material. In the lower portion of the sampled section (Fig. 1), δ15N values average around +2.5‰ with relatively little variation (+1.3 to +3.8‰). They rise from +1.0‰ at ∼161 m to a peak of +7.5‰ at 139 m, then fall back to +2.5‰. The rise in δ15N roughly correlates with an enrichment in TOC between 153 and 126 m, from 3 to 16 weight percent (wt %). Additionally, atomic C/N ratios increase from ∼150 to ∼65 from base to top. δ13Corg values vary little below 150 m, consistently falling between –36 and –42‰. There is no δ13Corg excursion corresponding to the δ15N spike. Instead, δ13Corg values rise from –42 to –34‰ above 134 m, corresponding to a decrease in δ15N values.

Fig. 1.

Geochemistry of the Mount McRae stratigraphic section, including δ15N, TN, δ13Corg, TOC, atomic C/N, Fe and Mo enrichments, wt % S, and δ34S.

Given the low metamorphic grade (prehnite-pumpellyite facies to < 300°C) of the Mount McRae Shale (8), preferential 14NH4+ loss during high-temperature devolatilization should not have greatly affected δ15N preservation (9). The direct relation between TN and δ15N (Fig. 1) further supports minimal metamorphic δ15N alteration because 14NH4+ loss would produce the opposite relation. The relatively high C/N ratios are typical of Precambrian organic matter (10) and reflect preferential degradation of organic N during diagenesis. The C/N decrease upsection may result from increased adsorption of diagenetically produced NH4+ onto clay minerals or its substitution for K+ in K-bearing minerals. If so, the lack of covariation between δ15N values and C/N ratios implies that diagenetic δ15N alteration was minimal. Furthermore, where bottom-water O2 is low in the modern ocean, as it was during Mount McRae Shale deposition, there is little difference between the δ15N of the original organic matter and that of diagenetically produced NH4+ (11). Thus, the δ15N recorded in the Mount McRae kerogenous shales probably reflects the mean δ15N of primary-producer biomass.

The δ15N average of +2.5‰ below 161 m is less than the mean value of fixed N in the modern ocean (+5‰), indicating that little fractionation was imparted during fixed N input and output. This could reflect an environment similar to modern stratified basins where nitrification and denitrification are active and ϵdenitrification is under-expressed because NO3 is completely consumed at the oxic/anoxic interface (12). However, nitrification requires environmental O2, but there is no evidence for its presence in this part of the core. In particular, the low amounts of sedimentary Mo below 161 m (despite high S and TOC contents, which should sequester any Mo if available) suggest that oxidative weathering of Mo-bearing sulfides in continental crust or detrital sediments was not active (13), precluding large amounts of atmospheric or oceanic O2. Thus, it is more likely that the low δ15N values below 161 m represent an anoxic N cycle with little fractionation imparted during microbial N2 fixation and negligible fixed N loss (Fig. 2A).

Fig. 2.

Nitrogen cycle transformations. (A) Hypothesized anaerobic N cycle before Mount McRae δ15N excursion and (B) hypothesized suboxic aerobic N cycle at peak of Mount McRae δ15N excursion. The broken line indicates abiotic processes, and the dotted line indicates plausible but unproven processes.

The rise in δ15N values to modern ocean values between 161 and 139 m indicates that fractionation occurred during fixed N loss. Although NH3 loss to the atmosphere has an equilibrium isotope effect comparable to ϵdenitrification (14), this process should not have been favored because it is unlikely that Archean oceanic pH ever exceeded the limit (pH = 9.34) at which the NH4+-NH3 equilibrium shifts toward gaseous NH3 (14, 15). Thus, increased expression of ϵdenitrification was evidently responsible for the increase in δ15N values. The conversion of NO3 or NO2 to N2 was probably biologically catalyzed because abiotic processes apparently reduce NO3 to NH4+ and not N2 (16). Furthermore, expression of ϵdenitrification requires a large pool of NO3 or NO2 so that not all is consumed by denitrification. Thus, it is implausible that NO2 was a product of lightning combustion (17) because only a small amount of NO2 could have been produced by this process under late Archean conditions (18). Instead, it is most likely that microbial N2 fixation introduced NH4+ to the ocean, and increased O2 promoted the oxidation of NH4+ to NO3 or NO2. Because high activation energy prevents NH4+ from abiotically oxidizing to NO3 or NO2 under Earth's surface conditions, nitrifying microbes evidently produced the NO3 or NO2. Microbial denitrification then imparted a N isotope fractionation to the residual oxidized N pool, recorded in organic N (Fig. 2B)

Correlative redox-sensitive trace metal (13) and S isotopic (19) data imply that intensification of nitrification and denitrification during the δ15N excursion corresponded to a rise in O2 from <10–6 to ≤10–5 present atmospheric level (PAL). Although increased oxygenation of the surface ocean may have limited N2 fixation, and therefore productivity, during certain periods in Earth history (20), reduced N2 fixation during the δ15N excursion is unlikely. Although nitrogenase, the enzyme responsible for N2 fixation, is irreversibly deactivated in the presence of molecular oxygen, inhibition is not apparent until the partial pressure of oxygen (PO) > 0.2 to 0.5 PAL (21), which is much greater than the inferred maximum PO of 10–5 PAL during the excursion. Furthermore, although Fe is a critical component of nitrogenase and the concentration of Fe2+ dropped during the δ15N excursion (Fig. 1), cell culture experiments have shown that even under the low Fe concentrations characteristic of a fully oxygenated ocean, N2 fixation rates are not lowered (22). Although NO2 may have been the most oxidized form of N attained under such low-oxygen conditions, incomplete nitrification would not have inhibited productivity because many microbes assimilate NO2 as effectively as NO3 (23). Thus, surface-ocean oxygenation during the δ15N excursion was high enough to support nitrification, at least to NO2, but not so high as to limit N2 fixation, and therefore productivity.

The drop in δ15N values above 139 m may indicate an increased proportion of sedimentary organic matter represented by N2-fixing biomass. Alternatively, decreased δ15N values may reflect decreased expression of ϵdenitrification. This could have resulted from diminished nitrification (because denitrification should approach complete consumption as the size of the NO2 or NO3 pool decreases), diminished denitrification, or both diminished nitrification and denitrification. Although the drop in δ15N values is difficult to explain, decreased Mo and Fe abundances during this interval without a corresponding decrease in S concentrations (Fig. 1) indicate that the deep ocean may have become euxinic (anoxic and sulfidic). Because both nitrification and denitrification are deactivated under euxinic conditions (24), the formation of euxinic bottom waters—a plausible response to slight surface-ocean oxygenation and increased [SO 2–4] during the δ15N excursion (20)—may have caused the observed drop in δ15N values.

We conclude that the 2.5-Ga Mount McRae section records an episode of increased nitrification and denitrification in response to slight surface-ocean oxygenation. Thus, an aerobic component to the N cycle was transiently active before the atmosphere became oxygen-rich after 2.45 Ga (7). Previously, Beaumont and Robert (10) observed a range of kerogen δ15N in cherts from –6.2 to +13‰ between 3.5 and 2.8 Ga, and +0.3 to +10.1‰ between 2.1 and 0.7 Ga, which they attributed to a transition from an anaerobic to an aerobic N cycle. More recent δ15N data has complicated this scenario. For example, Jia and Kerrich (25) reported a trend from high δ15N kerogen (+15.3‰) in highly metamorphosed Archean shales to low (+3.5‰) values in Proterozoic counterparts, whereas Shen et al. (26) noted that the largest positive δ15N shift occurs in banded iron formations (BIFs) between 2.7 and 2.6 Ga, and not during the transition to a fully oxygenated atmosphere. These contradictory results illustrate the difficulty of interpreting δ15N data from diverse samples with low stratigraphic resolution. For example, because Archean cherts are often associated with hydrothermal systems where chemolithoautotrophs (δ15Nbiomass = –9.6 to +0.9‰) replace photosynthesizers as the primary producers (26), the 15N-depleted ancient cherts reported by Beaumont and Robert may reflect incorporation of chemosynthetic biomass, and not the mean δ15N of oceanic fixed N.

Although denitrifying microbes are widespread across the Archaea and Bacteria, nitrification is more restricted phylogenetically (Fig. 3). The high productivity indicated by the high TOC and the attendant large positive nitrogen isotope fractionations suggests that nitrification could have been performed only by chemolithotrophic nitrifiers, because heterotrophic nitrifiers are, and presumably always were, minor contributors to the marine pool of oxidized nitrogen species. Because marine chemolithotrophic nitrifiers are apparently restricted to the β-and γ-Proteobacteria and the low-temperature marine group I.1 Crenarchaeota [although there are hints that pSL 12 Crenarchaeota might also nitrify (27)], and as these dominant nitrifying groups are terminally branching subphyla of peripheral clades of the Bacteria and Archaea (Fig. 3), our data imply that most prokaryotic phyla in at least one of these domains must have been extant by the time these groups arose. Thus, the macroevolution of microbes in the bacterial and/or archaeal domains may have been largely complete by the end of the Archean.

Fig. 3.

The Tree of Life based on small subunit ribosomal RNA sequence analysis [modified from (29)]. Lineages in red represent known chemolithotrophic nitrifiers. The terminal position of Marine group 1 agrees with more recent widely accepted Archaeal trees [e.g., (30)].

Supporting Online Material

Materials and Methods

Figs. S1 and S2

Table S1


References and Notes

Stay Connected to Science

Navigate This Article