Geophysical Research Letters

Influence of marine denitrification on atmospheric N2O variability during the Holocene

Authors


Abstract

[1] Oceanic denitrification centers are thought to be important marine sources for atmospheric N2O. To consider Holocene variability in this source, we reconstruct the Holocene paleo-denitrification history of the Peru margin, a major marine denitrification center, using high-resolution sedimentary δ15N data. This record along with a contemporaneous one from the western Arabian Sea (Altabet et al., 2002) shows similarities with the recently available high-resolution atmospheric N2O record (Flückiger et al., 2002). While the role of terrestrial processes in the observed N2O changes remains uncertain, these results suggest that variability in marine denitrification in major upwelling centers such as the Peru margin and the Arabian Sea contributed significantly to atmospheric N2O evolution during the Holocene.

1. Introduction

[2] Nitrous oxide (N2O) (atmospheric lifetime ∼120 years) [Intergovernmental Panel on Climate Change, 2001] is the fourth most important greenhouse gas (after water vapor, CO2 and CH4). The major pre-industrial sources of atmospheric N2O are nitrification and denitrification occurring in terrestrial soils and the ocean. N2O is released as a by-product of microbially mediated nitrification and is an intermediate for denitrification [Naqvi et al., 1998]. Upwelling regions such as the Arabian Sea (AS) and the equatorial tropical north and south Pacific (ETNP & ETSP) are sites of high N2O release to the atmosphere via upwelling of low O2 waters that are relatively rich in N2O, produced by these processes. Major sinks are its photo-disassociation and reaction with excited oxygen in the stratosphere. Recent studies in heavily human-impacted coastal areas of the eastern AS have indicated global expansion of sub-surface denitrification and in turn, increasing N2O fluxes to the atmosphere, with a possible influence on global climate [Naqvi et al., 2000]. To assess future N2O evolution in the anthropogenic era, it is important to assess the sensitivity of the natural atmospheric N2O variability recorded in ice cores with respect to the variability of marine denitrification occurring in major upwelling centers during the Holocene. Evidence for its natural Holocene variability became recently available from the Antarctica ice core ‘EPICA Dome C’, which revealed a significant (∼14%) fluctuations on millennial time scales [Flückiger et al., 2002]. While terrestrial soil sources at middle to high latitudes of northern hemisphere have been proposed to have played an important role in regulating atmospheric N2O inventory [Flückiger et al., 2002], it is important to explore oceanic contributions as well, to understand overall controlling factors responsible for N2O variability. Recent studies [Altabet et al., 1995, 2002; Ganeshram et al., 1995] from the AS and ETNP have established a coupling between past variations in marine denitrification intensity and late-Quaternary climate changes. Plausible linkages between marine denitrification in the AS and atmospheric N2O variability during late-Quaternary were suggested by Suthhof et al. [2001]. The ETSP/Peru margin (PM) is another such region with high surface productivity and an intense oxygen minimum zone (OMZ) (Figure S1). This region is responsible for 10–25% of global marine denitrification [Codispoti et al., 1986]. The ETSP/PM also lies in the heart of El Niño Southern Oscillation (ENSO), where short term ENSO cycles coupled with movements of the Intertropical convergence zone (ITCZ) (controlled by solar insolation) are known to be capable of influencing global climate on millennial time scale [Clement et al., 1999]. Past variations (millennial to centennial scale) in sub-surface denitrification in this region could thus have a significant effect on global inventory of atmospheric N2O.

[3] Until recently, obtaining well-dated core material from the PM has been difficult for high-resolution paleo-reconstructions. Poor preservation of foraminiferal CaCO3 due to high organic carbon respiration rates severely limits conventional dating schemes under the highly productive PM upwelling cell. However, 14C dating of extracted sedimentary alkenones of 95–99% purity along with bulk organic carbon has been proven an alternative dating tool in this region [Higginson and Altabet, 2004], validating the use of bulk organic matter dating for paleo-reconstructions along the PM. Here, we present high resolution Holocene data of sedimentary δ15N from three sediment cores underlying the OMZ off Peru. A close inspection of these records together with available Holocene high-resolution marine denitrification variability inferred from the western AS [Altabet et al., 2002] reveals similarities on millennial time scale. Overall marine denitrification variability from both of these major upwelling regions exhibit similarity with Holocene atmospheric N2O variability preserved in the ice core record from Antarctica [Flückiger et al., 2002].

2. Material and Methods

[4] Figure 1 shows sediment core locations off Peru used in this study. The age-depth models (Figure 1, inset) of these cores were established through a combination of compound specific alkenone 14C dates and previously published 14C dates of bulk organic carbon [Remiers and Suess, 1983]. The sediment core W7706-40 (W40, 11°15′S, 77°58′W; water depth ∼186 m) covers the last ∼5000 yr over its 1.85 m length and ODP Site 1228D (1228D, 11°3.85′S, 78°4.67′W; ∼273 m) covers ∼14 to ∼0.8 ka BP in its upper ∼2.8 m. The location of core MW8708-PC2 is further south (PC2, 15.1°S, 75.7°W; ∼270 m) and it covers the time span from ∼8 to ∼2 ka BP based on linear interpolation of two 14C dates of bulk organic carbon. All 14C ages were corrected with a reservoir age correction of 420 years, and have a 1σ error ≤ ±250 yr during the Holocene. An additional age constraint for the northern cores W40 and 1228D comes from the presence of a correlative ash layer at 147–151 cm and 120–122 cm respectively, corresponding to 3.2 ± 0.1 ka BP. All sediment cores were sampled every cm and freeze dried prior to analysis of chemical and isotopic proxies. Average temporal resolutions provided by the cm scale sampling for the studied cores W40, 1228D and PC2, are ∼22, 55 and 25 years per cm respectively. Measurements of N content (wt. %) and δ15N were made on ∼10–25 mg aliquots of homogenized dried bulk sediment. Use of bulk sedimentary δ15N to infer paleo-denitrification follows Altabet et al. [1995, 2002]. Details of analytical methods applied are published elsewhere [Altabet et al., 2002]. All δ15N data are relative to atmospheric N2 and overall reproducibility of δ15N measurements is better than ±0.2‰.

Figure 1.

Sediment core locations on the Peru margin. Inset shows age-depth models used for establishing chronostratigraphies of the sediment cores. All 14C ages have a 1σ error of <±250 yr.

[5] To compare Holocene denitrification variability in the PM with that of the AS, we used δ15N data of two sediment cores RC27-23 (17°59.6′N, 57°39.3′E; ∼820 m) and RC27-14 (18°15.2′N, 57°35.4′E; ∼596 m) from the intense upwelling region of the Oman margin. Both cores are at water depths within the OMZ (∼200 to 1200 m), thereby ensuring excellent preservation of sedimentary organic matter and δ15N signals with minimal diagenetic artifacts. Low bioturbation and high accumulation rates provide very high temporal resolution as compared to other contemporaneous records (Figure 2c) [Altabet et al., 2002]. Likewise, our PM denitrification record is based on three sediment cores from <300 m water depths within the core of OMZ (∼150 to 400 m). Available temporal resolution in Holocene δ15N records from the ETNP is not sufficient at present to directly compare paleo-denitrification of the ETNP, the third major marine denitrification center.

Figure 2.

(a and b) Sedimentary δ15N in cores W40, 1228D and PC2 from PM along with (c) those from AS [Altabet et al., 2002]. (d) Holocene atmospheric N2O variability [Flückiger et al., 2002]. (e and f) Visual matching of PM and AS denitrification reveal close similarity with atmospheric N2O variability during the Holocene. (g and h) Millennial scale trends using a low pass filter (i.e., 5 pt running average). Overall uncertainty in δ15N measurements is ≤0.2‰.

3. Sedimentary δ15N as a Water Column Denitrification Index on the Peru Margin

[6] The PM is characterized by high surface productivity induced by coastal upwelling of nutrient rich waters from the poleward flowing sub-surface (Peru) undercurrent. A combination of poor ventilation and organic matter decay produces suboxic conditions at intermediate water depths, supporting substantial denitrification (Figure S1) but not sulfate reduction. This water mass is found between ∼5°S to 15°S contacting the sea floor between 150–400 m water depth [Bruland et al., 2005] where deposition of laminated organic rich diatomaceous sediments can occur [Kudrass, 2000]. During normal/La Niña conditions, intense coastal upwelling of cool nutrient-rich waters fuel high surface productivity in the region. In contrast, during El Niño events, upwelling of nutrient rich water diminishes as the thermocline/nutricline in the region deepens, eventually resulting in a substantial decrease in surface productivity [Arntz and Fahrbach, 1991]. Both the reduced flux of carbon and depression of the thermocline moderate low-oxygen conditions along the PM. Hence, El Niño conditions should result in a suppression of sub-surface denitrification in the PM and a reduction in N2O supply to the surface waters. Since the temporal resolutions of our cm scale sampling of sediment cores do not allow detection of individual ENSO events, we discuss here only evidence for centennial scale denitrification changes, which, we infer, are linked to modulations in ENSO conditions.

[7] Limited supply of dissolved Iron (Fe), a micronutrient for surface ocean production, could result in partial utilization of nutrients by phytoplankton and thereby influence sedimentary δ15N. This is not likely, however, since modern upwelled waters on the central PM have been found to be Fe-rich due to contact with reducing sediments [Bruland et al., 2005]. Moreover, in the core of denitrification zone on the PM at 15 to 16°S, combined water column water NO3 and surficial sediment δ15N data show the same large enrichments (up to 10–11‰) [Liu, 1979; Altabet, 2005] and thus no evidence of influence from partial NO3 utilization. As in the AS and ETNP, the denitrification signal is faithfully transferred to margin sediments and downcore δ15N should mimic the denitrification history of sub-surface waters of the PM given the excellent state of organic matter preservation [Altabet, 2005]. Excellent correlation(s) observed between δ15N and productivity proxies for the late-Holocene period in one of the studied cores (W40) further supports this contention (R. Agnihotri et al., manuscript in preparation, 2006).

4. Results

[8] Figures 2a and 2b show depth-profiles of δ15N in studied sediment cores. Cores W40 and PC2 show high frequency variations in δ15N (from 4.7 to 8.2‰ and 4.4 to 9.2‰ respectively) as compared to 1228D (from 4.3 to 6.8‰) during the Holocene. The lower amplitude of δ15N variations in the core 1228D could be due to signal averaging owing to the lower sedimentation rate (∼55 yrs/cm compare to ∼22 yrs/cm for the core W40) at this location. High frequency oscillations of δ15N in core PC2, suggest higher sensitivity of this southern core-site to rapidly changing hydrographic conditions most probably forced by ENSO. Depth profiles of cores 1228D and PC2 demonstrate a increasing trend in denitrification with arrival of high frequency variability in δ15N after ∼5 to 5.6 ka BP (Figures 2a and 2b), at the same time as terrestrial records suggest the advent of modern ENSO [Rodbell et al., 1999; Moy et al., 2002]. However, onset of high frequency variability in δ15N in core W40 appears to occur after ∼3 ka BP, which might be an artifact of a sudden drop in sedimentation rate (from ∼165 cm to 185 cm interval) (Figure 1, inset). Nonetheless, overlapping time slices of the cores W40 and 1228D show similar variability patterns (Figure 2a), suggesting commonality of the δ15N signal in close proximity areas of the PM, and thereby allowing combining their respective time series to produce a Holocene composite (Figure 2e) for comparison with the observed Holocene variability in the western AS denitrification (Figure 2c) and N2O in Antarctica ice core record (Figure 2d). Southern core PC2 being chronologically constrained by only two 14C dates is excluded from the composite. We compare the PM δ15N, the western AS δ15N and ice core N2O records each on its own independent chronology. Visual inspection clearly shows that Holocene marine denitrification variability on the PM as well as the AS closely follows Holocene evolution of N2O. Despite uncertainties involving individual data points (Figure 2e), broad covariance of marine denitrification at millennial scale at these two major denitrification centers (Figures 2g and 2h) is noteworthy. The PM, the AS and ice core N2O record also reproduce a broad Holocene scale minimum in denitrification and minimum in N2O between about 8 to 6 ka BP. The PM denitrification, not only shows closer covariance with Holocene atmospheric N2O evolution (Figure 2e), it appears to be the only one amongst the three major upwelling/denitrification centers that displays large centennial scale variability and could have directly affected the atmospheric inventory of N2O during the Holocene on this time scale.

[9] Insufficient signal:noise ratio for the N2O data [Flückiger et al., 2002] as well as age-model uncertainties for both types of time series preclude correlating periodicities observed in the time series of PM denitrification with those which may be present in atmospheric N2O variability. Nevertheless, we report here centennial-scale variability in PM denitrification using spectral analysis [Schulz and Stagger, 1997]. Periodograms reveal significant (≥95%) cycles of ∼1310, 750, 550, 475, 402, 307, 261 and ∼2130, 980, 670, 410, 336, 300, 245, 145, 120 and ∼80 years for cores W40 and PC2 respectively (Figure S2). Core 1228D displays major cycles of ∼1100, 770, 622, 520, 400 and 165 years (Figure S2), in which some periodicities are clearly overlapping with those observed in the core W40 within age uncertainties. We speculate that observed centennial scale periodicities (longer than the atmospheric residence time of N2O, i.e., ∼120 years) in marine denitrification off Peru likely affected global atmospheric N2O inventory.

5. Discussion

[10] Since Holocene atmospheric N2O variations could not be attributed to changes in sink strength, observed variability in N2O record must be due to changes in the terrestrial or oceanic sources [Flückiger et al., 2002]. Holocene N2O variability does not follow the Holocene CH4 record (a proxy for global wetland extent and productivity), instead it closely resembles the CO2 record [Flückiger et al., 2002]. It is therefore possible that terrestrial soil sources from the middle to high northern latitudes might instead be responsible for N2O evolution during the Holocene. Soils in these regions which are, in general, less favorable for N2O production (due to lower temperatures and limited nitrogen availability) [Bouwmann et al., 1993], could have turned favorable for N2O production with moderate warming cycles during the Holocene. However, as the flux of greenhouse gases CO2 and N2O from these regions involves feedback mechanisms that are highly sensitive to changing temperature conditions, the exact role of these regions in governing the global inventory of greenhouse gases is still not clear [Mack et al., 2004]. By contrast, the behavior of oceanic sources (especially from tropical regions) is better understood. It is clear that major marine denitrification centers such as ETNP, ETSP and AS, release large amount of N2O into the atmosphere through a combination of biogeochemical/physical processes. This study reveals that marine denitrification in the PM and the AS varied in tandem during the Holocene. Since ratios of terrestrial and marine N2O sources have remained fairly constant during the last 33 kyr [Sowers et al., 2003], our evidence for concurrent changes in marine denitrification in major upwelling centers possess important implication toward understanding natural Holocene atmospheric N2O variability. The mid-Holocene increase in PM denitrification is correlated with cooling of the ocean surface at ∼5 ka BP and possibly more intense PM upwelling [Andrus et al., 2002]. It follows that enhanced ETSP denitrification occurred between strong El Nino events and during La Nina events. For example, more intense denitrification was reported after the collapse of Peruvian anchoveta fishery and also during 1985 La Nina event [Codispoti et al., 1986, 2001]. Changes in the frequency and intensity of ENSO manifest at the centennial scale and longer likely brought about the observed Holocene variability in PM denitrification that in turn influenced global atmospheric N2O inventory. However, to investigate the exact influence of ENSO forcing on PM denitrification and atmospheric N2O evolution, much higher temporally resolved δ15N data from well-dated sediment cores at suitable locations (such as the core site of PC2) along with more precise N2O data from ice cores are needed.

6. Conclusions

[11] Our paleo-denitrification data from the PM and the AS [Altabet et al., 2002] suggest that marine denitrification in the PM and the AS varied in tandem with atmospheric N2O evolution during the Holocene. The PM having high sensitivity to the ENSO cycles appears to be the only one among three major upwelling/denitrification centers has undergone centennial scale variations in sub-surface denitrification especially after the advent of the modern ENSO regime during mid-Holocene and thereby could have directly impacted the atmospheric N2O inventory.

Acknowledgments

[12] This research is sponsored by the U.S. National Science Foundation (NSF). We acknowledge the technical assistance provided by P. Feng, R. Singh and Matt Higginson for supplying 14C dates. We also acknowledge two constructive anonymous reviews and thank core curators of the ODP and OSU core repository. This manuscript is UMD SMAST contribution 06-0603.

Ancillary