AGAGE data for marine N2O studies
The network of Advanced Global Atmospheric Gases Experiment (AGAGE) stations has been quantifying greenhouse gas levels including N2O since the late 1970s27. This global network now produces high frequency, high precision measurements; the data improved substantially in the mid-1990s with the incorporation of higher precision methodologies28. N2O data from the last ~20 years are measured every 40 min (Fig. 1b, Supplementary Figs. 1 and 2) and have a precision on any individual measurement of 0.1 ppb but much higher confidence for the ensemble28,29. Combining these measurements with atmospheric back trajectories generated by the HYSPLIT4 model from monitoring stations (Fig. 1a, Supplementary Data 1), we map the relationship between atmospheric N2O concentrations and marine O2 content. Similar analyses have previously been conducted to investigate recent continuing sources of chlorofluorocarbons from land30. The marine environment is dynamic, and time-resolution is difficult to sample for many biologically-active chemical parameters requiring in situ point measurements. The AGAGE dataset, however, does not suffer the same restrictions, and the high-frequency measurements can be utilized to investigate variability in detail, including the important interannual dynamics that arise due to the influence of El Niño–Southern Oscillation (ENSO) on these regions.
Land sources could obscure the attribution of marine N2O, so all back trajectories that intersect continents over a 20-day model run are excluded in our analysis. This rigid protocol limits the choice of stations and marine areas. Multiple stations were considered for analysis, but eliminated as being ill-suited for identifying features in the OMZs. The land and monsoonal influence in the Arabian Sea obscures any marine signal in that OMZ. Similarly, the available high precision North Pacific atmospheric measurement stations are located along the California coast and in Hawaii, and do not communicate with the eastern tropical North Pacific OMZ before mixing homogenizes any advected signal (Supplementary Fig. 3). The station at American Samoa, however, permits excellent monitoring of the ETSP. The easterly trade winds rapidly transport air parcels intersecting the region of low O2 across the Pacific (scale of days to weeks) before turbulent mixing with other air masses is able to fully mask the OMZ signal (Fig. 2). Moreover, along this trajectory, air tends to remain in the troposphere (Supplementary Fig. 4), minimizing stratospheric impacts on surface conditions. The raw N2O data from Samoa, when de-trended for the long-term anthropogenic effect and de-seasonalized to minimize the intra-annual oscillation driven largely by interhemispheric transport from the north as the Intertropical Convergence Zone moves southward during boreal winter (Supplementary Figs. 1 and 2), reveal striking relationships with the ocean biogeochemical state (Fig. 3).
Locations of air parcels with their de-trended nitrous oxide measurements at Samoa are plotted (a) 5, (b) 10, (c) 15, and (d) 20 days prior to arrival at Samoa. De-trended nitrous oxide measurements are represented by the color of the corresponding dot, as anomalies relative to the station mean. The data maintain a clear spatial delineation for the full 20-day period, with much higher concentrations passing over the eastern tropical Pacific, and much lower concentrations arriving from the west and the Southern Ocean. Overlain on this map are the 10 and 20% oxygen saturation horizons at the 1026.5 kg m−3 potential density level (black contours). The orange circle indicates the location of the Samoa station.
De-trended and de-seasonalized N2O concentrations from Samoa (color) are gridded based on back-trajectory locations 15 days prior, and a mean value is calculated for air passing over each 5° grid cell at that time (color). Overlain on this map are the 10 and 20% oxygen saturation horizons at the 1026.5 kg m−3 potential density level (black lines). These two data sets come from entirely independent sources, and the coincidence between them is striking; high atmospheric N2O concentrations align with the oxygen minimum zone, extending westward into the tropical Pacific and down the Chilean coast of South America. The orange circle indicates the location of the Samoa station.
The seasonal modulation in Samoa N2O has been well-established as an effect of northern influence. However, across the Pacific sites, trajectories remain mostly well-confined to the hemispheres in which they originated (Supplementary Fig. 5). The air parcels that reach Samoa from the northern hemisphere however, do not have markedly higher N2O concentrations than their southern counterparts. Many high concentration trajectories do travel across the Isthmus of Panama from the Caribbean (Supplementary Fig. 6), but these are filtered out by the land filter. Notably, 20 days prior there are still a large number of high concentration trajectories in the southern hemisphere, which travel northward off the coast of South America before transiting to Samoa. Indeed, very few trajectories actually cross over the South American continent; most are deflected either north or south along its western edge. As a result, land emissions of N2O from South America are unlikely to have a large impact on the visible spatial signal.
Oxygen minimum zones emerge as N2O hotspots
A divergence in concentrations is visible just one-day prior, with higher concentrations arriving from the north than from the south (Supplementary Fig. 3). This pattern remains moving further back in time; the highest concentrations pass over the eastern tropical Pacific, while the lowest come from the west passing over the Southern Ocean. The pattern becomes slightly less defined between 15 and 20 days-prior, as parcels with high concentrations diverge again along the western coasts of North and South America. Given the velocity of the trade winds and the distance between Samoa and the ETSP, ~8 m s−1 and 104 km, respectively, this time scale is reasonable. Certainly the observed atmospheric concentration anomalies of N2O at Samoa are a combined result of intra-hemispheric biological production with interhemispheric transport, but the analysis presented herein aims to tease apart the biological impact from the tropical Pacific ocean from the additional seasonal supply from the N2O-enriched northern hemisphere (Supplementary Fig. 7).
In defining the OMZ boundary by the 20% O2 saturation contour on the 1026.5 kg m−3 potential density surface, the 15-day back trajectories identify higher than average N2O among the air parcels passing over the OMZ (Fig. 3). This isopycnal layer, generally at depths of ~200–500 m (shallower in the eastern Pacific), tends to co-locate the minimum O2 concentration within the water column31. Whereas water this deep does not tend to equilibrate rapidly with the atmosphere, this density surface reflects broader features in the water column, such as the existence of a shallower N2O maximum driven by microbial processes18. This shallower N2O maximum can be more easily transmitted to the surface waters from the interplay of upwelling and lateral and vertical mixing32. Mesoscale and sub-mesoscale eddies in particular, below the spatial resolution that can be resolved by these data, are likely pathways that promote mixing of deeper N2O into the surface waters33,34.
The air masses reaching Samoa that pass over the OMZ are ~0.4 ppb higher than the remainder of the South Pacific. This intense localization to the lowest O2 waters in the eastern tropical Pacific, extending southward along the Peruvian and Chilean margin, suggests a disproportionate role of OMZs and coastal upwelling waters in generating atmospheric N2O relative to the open ocean. Extension of higher N2O anomalies along the equator outside of the 20% O2 contour also remain visible west to 140° W (Fig. 3), albeit with reduced magnitude, indicative of a separate likely N2O source from equatorial upwelling and subsequent outgassing35,36. Further, the air overlying the coastal Pacific along the South American upwelling region is especially high in its N2O content, indicative of the heterogeneities that exist even within suboxic waters17. The coastal-most waters are the most productive and the resulting organic matter fuels greater microbial generation of N2O. Moreover, the depth of the anoxic onset and coincident N2O maximum is generally shallower toward the coast, increasing the communication between ocean and air. Narrow, shallow shelves within the OMZs in particular are important producers of N2O17,23,37,38, but local sources with such spatial resolution cannot be identified via our methods.
While the absolute difference between N2O mixing ratios in air that has passed over the OMZ versus those that did not is small, the abundances above the OMZs are likely much higher but are diluted to the observed anomalies after 15 days of transport and mixing. Notably, the air parcels will continue to gain N2O as they travel over supersaturated water masses or lose N2O over undersaturated areas. The coherence of the 15-day backtracked Samoan N2O signal and its clear demarcation with the ETSP highlight the significance of the OMZs for the marine N2O efflux. Indeed, the offset between the air passing over the OMZ grid cells and the rest of the South Pacific matches the average seasonal variability in the Samoan data. The seasonal cycle, which includes terrestrial sources and is heavily influenced by seasonal shifts in the intertropical convergence zone, peaks in Austral winter (January/February) with minima 0.50 ± 0.13 ppb (se) lower in summer (July/August). The rapid N2O cycling rates inherent to the highly productive OMZs cause these regions to respond quickly to time-dependent shifts in the underlying biogeochemistry and physical structure of the water column18,24.
Further modulation by El Niño and La Niña
It is known that ETSP biogeochemistry is influenced by the El Niño/La Niña oscillations of the tropical Pacific39,40. La Niña brings enhanced upwelling of nutrients to the surface, thereby increasing primary production, further de-oxygenating the subsurface region, and supplying more organic matter to drive greater denitrification and N2O production13. Conversely, during an El Niño decreased upwelling reduces the surface productivity in the eastern tropical Pacific and deepens the oxycline, thereby contracting the OMZ and decreasing N2O emissions. In this study, the near-continuous data over 21 years permits a more quantitative analysis. The period from 1996 through 2016 included both a number of strong El Niño (1997/98, 2002/03, 2009/10, 2015/16) and La Niña (1998/99, 2007/08, 2010/11) states. Our analysis consists of dividing the N2O dataset among ENSO states in two different ways, both of which produced equivalent results: First, by aligning the N2O oscillation against the Niño 3.4 index, resulting in a time lag of 3 months relative to the ENSO state; second, by producing a running integral of the Niño 3.4 index forward in time, which essentially represents a cumulative effect of prolonged El Niño/La Niña periods (Fig. 1c). The Niño 3.4 index describes a 5-month running mean of sea surface temperature anomalies in the central equatorial Pacific, and was used because it is indicative of the broader Pacific region.
Strikingly, during La Niña, air parcels passing over OMZ grid cells increase on top of their already higher than average concentrations, whereas parcels that do not pass over OMZ grid cells exhibit no systematic change (Fig. 4). This increase with La Niña is not uniform across the OMZ, however, as the lowest O2 waters (<10% saturation) increase by approximately one-third while those of hypoxia (<20% saturation) increase during a strong La Niña by more than double compared with neutral conditions (Table 1, Fig. 5). This occurrence implies the lowest O2 waters already exhibit their maximal production whereas the slightly more oxygenated waters on the periphery have potential for lower O2 states with La Niña to augment their N2O production. This decrease in O2 can amplify both nitrification-based production as well as expand the zone where denitrification-sourced N2O is permitted. During El Niño, while the parcels not passing over the OMZ do not change systematically from their mean state, those passing over the OMZ consistently see reduced N2O concentrations relative to their means (Table 1, Fig. 4) and lower than the long-term Samoan average. In all, the lowest O2 waters are by far the most variable temporally, subject to short-scale dynamics that affect these waters more than the open ocean. Such interplay confirms the importance of OMZs, with their shallow N2O maxima and fast generation times18 to marine N2O efflux. The cutoff of coastal upwelling associated with El Niño and the corresponding dramatic decline of primary productivity supplying organic matter to the interior curtail N2O production and eliminate the unique conditions required for OMZ-based production. The effects are even more pronounced during the strongest of El Niño and La Niña events, including the recent 2015-2016 El Niño, with greater shifts from the mean state during the most intense periods (Fig. 5). The Pacific-wide decrease in N2O observed in the AGAGE back-trajectory analysis is in line with ship-based observations from the eastern41, central35, and western42 Pacific basins.
The time series was split among three states: La Niña, with a Niño 3.4 index < −0.5; Neutral, with an index between −0.5 and 0.5; and El Niño, with an index > 0.5. New gridded maps were calculated for each, by averaging only data points within the corresponding El Niño/Neutral/La Niña states. a The change in each grid cell’s mean value from a Neutral state to a La Niña state. b The change from Neutral to El Niño. During La Niña, the eastern tropical South Pacific oxygen minimum zone (OMZ) displays heightened N2O in air that has passed over the water with no consistent effect in air that has passed over non-OMZ waters. During an El Niño, the concentrations in air that has passed over the OMZ are substantially reduced with again no systematic effects observed in air that has not passed over the OMZ.
Climatological oxygen concentrations are at the density level of 1026.5 kg m−3 and El Niño/Southern Oscillation (ENSO) state is divided by neutral (|index| < 0.5), weak (0.5 < |index| < 1.0), and strong (|index| > 1.0) states. N2O concentrations were separated by ENSO states and gridded as in Figs. 3 and 4, then means were calculated for grid cells of each O2 saturation range. N2O in air that has passed over the oxygen minimum zones with O2 < 20% saturation is greater than average during neutral and La Niña states, with highest N2O values at lowest % O2 saturation. Air parcels that have passed over regions with lowest % O2 saturation have nearly equal N2O values during strong La Niña and neutral states, but parcels that have passed over regions with 10–20% saturation exhibit substantial changes in N2O between the two states. Warmer surface waters across the Pacific (El Niño states), coincide with reduced N2O, reducing more substantially with stronger El Niño conditions, especially evident among the 0–20% O2 saturation waters which even show a sign change in anomaly. This El Niño effect is likely due to increased stratification of the water column isolating the deeper N2O rich waters from the surface. Shown are the mean ± standard error for grid cells within a bin.
ENSO impacts atmospheric circulation patterns that might obscure attribution of the observations to differences in marine production and efflux29,43,44. Notably, N2O concentrations are higher in the northern hemisphere because of greater terrestrial and anthropogenic production. Similarly, sulfur hexafluoride (SF6) is a compound with comparable asymmetry in hemispheric concentration because it is an industrially produced volatile compound linked with anthropogenic emissions overwhelmingly in the northern hemisphere. Indeed, SF6 has an even greater interhemispheric gradient and a steeper growth rate than N2O. In repeating our analysis for SF6 for which AGAGE data at Samoa are available (2005 onwards), no apparent skew toward O2 deficient waters (Supplementary Fig. 8) or ENSO-rooted signal arises (Supplementary Figs. 9 and 10), suggesting that the spatio-temporal relationship in N2O concentrations is tied to autochthonous OMZ production rather than hemispheric exchange. In completing our 15-day back trajectory analysis, very few tracks cross over the inter-tropical convergence zone, largely keeping equatorward surface winds isolated in their respective hemispheres (Supplementary Figs. 3, 6 and 7). Still, during an El Niño, there is less influence of mixing from the northern hemisphere and increased south-easterly winds, resulting in a relative decrease in atmospheric N2O across the southern hemisphere45. Indeed, this broad decrease is observed in the data (Figs. 4 and 5). Increased inter-hemispheric exchange during La Niña events should increase concentrations across the southern hemisphere, but the observed changes are to a great extent localized to air passing over the OMZs. The localization again indicates the dominance of the local biological production term over perturbations in atmospheric transport. Impacts of ENSO are not limited to the Pacific Ocean; the inter-annual variability also causes changes to terrestrial emissions of N2O46,47. Impacts of ENSO on marine biogeochemistry can be substantial sources of this interannual variation in atmospheric N2O45. Importantly, over the period of 1999–2009, El Niño periods were indeed associated with low marine N2O fluxes and La Niña events with high anomalies46.
Summary
In developing models capable of predicting N2O under future climate scenarios, proper attribution of the natural sources and variability of this gas is of significant importance. The effects on these regions of anthropogenic climate change, and whether N2O from the OMZs indeed will increase into the future, remain open questions. Our results indicate that the variability of oceanic sources linked to OMZs can feedback substantially on the climate system as low O2 areas are predicted to expand25,26 and ENSO cycles change in intensity and frequency48. The enormous potential of OMZs to produce N2O and their direct response to El Niño/Southern Oscillation forcing, necessitates their explicit inclusion in N2O budget calculations and better monitoring of these regions from the eastern tropical Pacific to quantify the precise effect on N2O efflux before mixing reduces the local signal. The simple model presented here cannot precisely map or quantify these emissions because it does not take into account mixing or integrative effects along the transects. We also cannot directly attribute the role of transient sub-mesoscale eddies on enhanced N2O production33,49,50. While not individually observable from this dataset, such effects are encompassed in the ensemble average over the 21 years of measurements. Nonetheless, the high-frequency data from Samoa reveal not only that the ETSP is a large source of atmospheric N2O, but also that it is subject to such large changes in short-term budgets by ENSO that they are visible more than half-way across the Pacific Ocean.
Source: Ecology - nature.com
