N2O air-sea fluxes were continuously measured on a transect crossing the Subtropical South Pacific Gyre (SPG) from its central part toward its coastal boundary (associated with coastal upwelling off central Chile) during the BIOSOPE cruise (austral spring, 2004). Continuous N2O air-sea fluxes in the central part of the SPG (132° to 114°W) were slightly negative (−0.48 ± 0.44 μmol m−2 day−1), whereas in its eastern part (114°W to 81°W), they were slightly positive (0.41 ± 0.34 μmol m−2 day−1), reaching up to 18 μmol m−2 day−1 in the coastal area (130 km from the coast). The transition between oceanic and coastal conditions was characterized by an abrupt increase in N2O emissions from 80°W eastward and was associated with an increase of surface Chl-a contents. This trend corresponded to the change in trophic status from very oligotrophic to eutrophic. The outward (through the air-sea interface) and inward (through the pycnocline) N2O fluxes of the surface layer (SL) were also determined at three representative stations, taking into account turbulent diffusion and vertical advection. The N2O released into the atmosphere from the eastern part of the SPG came largely (70–80%) from the N2O produced in the SL (2.0 × 10−3μmol m−3 day−1). In the coastal area, N2O production in the SL reached up to 1.16 μmol m−3 day−1, and 28% of the N2O released into the atmosphere was upwelled though the pycnocline by Ekman transport. The annual N2O emissions estimated for the eastern South Pacific reach 50 Gg of N2O, confirming the importance of this region for global emissions and reaffirming coastal upwelling centers as areas of strong production and outgassing.
 Nitrous oxide (N2O) is a greenhouse and ozonolitic gas naturally produced by microorganisms involved in the nitrogen cycle. Furthermore, atmospheric N2O concentrations attributable to anthropogenic sources have been reported to increase since about the last century [Kroeze et al., 1999]. Understanding the N2O cycle in the Earth's system demands thorough knowledge of this gas's sources and sinks, their respective dynamics, and the processes involved in N2O cycling.
 The global ocean acts as a net N2O source into the atmosphere, with annual N2O emissions estimated at 4 Tg N a−1 [Nevison et al., 1995] and recently re-estimated at 7 Tg N a−1 [Bange, 2006]. However, regional emissions fluctuate widely among the global oceans. For example, the surfaces of subtropical oceanic gyres are in equilibrium or slightly oversaturated in N2O with respect to the atmospheric concentrations [Nevison et al., 1995], whereas saturation in the Atlantic equatorial belt reaches ∼110% [Oudot et al., 2002]. Eastern boundary coastal upwelling areas are major sources of marine N2O to the atmosphere [Bange et al., 1996; Nevison et al., 2004]. In fact, saturation values as high as 400% with respect to atmospheric concentrations have been recorded in these areas [Cornejo et al., 2006; Nevison et al., 2004].
 In the ocean, N2O is produced by several microbiological processes; the best-known of these are aerobic ammonium oxidation (nitrification) and nitrate reduction to N2O (partial denitrification) [Naqvi, 1991]. Nitrification is a ubiquitous chemoautotrophic process in the oceans, whereas denitrification, a bacterial respiratory process in which nitrate is used as an electron acceptor, is restricted to areas with low oxygen concentrations. Recently, other processes (e.g., nitrifier denitrification, which is nitrite reduction carried out by ammonium oxidizers) have been included as possible sources of N2O in the ocean [Charpentier et al., 2007]. These processes greatly depend on local biogeochemical conditions and are enhanced at low oxygen concentrations, although less than suboxic threshold O2 levels (i.e., <20 μM) drive net N2O reduction to N2 (complete denitrification), as observed in severe oxygen minimum zones [e.g., Castro-González and Farías, 2004]. Hence, most oceanic N2O production takes place in subsurface waters, which accumulate N2O at levels higher than 200% saturation. Most features of subsurface data sets are consistent with N2O sources dominated by nitrification based on a widespread, robust ΔN2O-AOU correlation [Nevison et al., 2003].
 According to these antecedents, the N2O released into the atmosphere can come from two main sources: the N2O produced directly in the surface layer and that produced under and transported through the pycnocline by turbulent mixing or vertical advection (associated mainly with the upwelling process). Upwelling is an important process in eastern boundary regions with equatorward winds, carrying not only nutrients that enhance biological activity (including N2O production processes) but also gases such as N2O. Otherwise, upwelling related to Ekman pumping due to the wind stress curl in open ocean areas, particularly subtropical anticyclonic gyres, is very low or slightly negative [Tomczak and Godfrey, 1994] and the transport of gases through the pycnocline must be mainly driven by turbulent diffusion [Lewis et al., 1984].
 The goal of this work is to evaluate the contribution made by the central and eastern South Pacific to atmospheric N2O. We also attempt to determine how much the surface (above the pycnocline) and subsurface (under the pycnocline) layers contribute to oceanic N2O emissions. To achieve this goal, we estimate nitrous oxide fluxes across the air-sea interface and pycnocline, along the track of the BIOSOPE cruise in austral spring (October to December 2004; see Figure 1), covering areas with extreme trophic levels, ranging from oligotrophic to eutrophic conditions. Moreover, we evaluate N2O production and transport under three different oceanic regimes in the South Pacific Ocean. The first two oceanic regimes studied are located in the center and eastern part of the central South Pacific Gyre (SPG). This is the largest permanent anticyclonic oceanic gyre and one of the least-studied areas of the ocean [Tomczak and Godfrey, 1994]. This area has been described as the most oligotrophic zone in the world ocean [Claustre and Maritorena, 2003]. Studies carried out on the same cruise as this work show surface chlorophyll (Chl-a) concentrations as low as 0.019 mg m−3 [Claustre et al., 2008] and extreme nutrient (N, Fe) limitations [Bonnet et al., 2008]. Satellite images clearly show a large low-Chl-a zone in the SPG [Claustre et al., 2008]. This low productivity area can be associated with hydrological conditions. Due to its anticyclonicity, the SPG has Ekman transport to the gyre's core, which promotes downwelling, supporting a very thick permanent pycnocline that is deeper than 200 m [Fiedler and Talley, 2006] and indicating an extremely weak supply of nutrients from deep waters. The SPG also has one of the lowest (almost zero) atmospheric dust depositions [Mahowald et al., 2005], contributing to its extreme oligotrophy.
 The third oceanic study area is the central Chile upwelling region, one of the most productive marine environments in the world [Carr, 2001; Daneri et al., 2000]. This area's wind pattern is highly favorable to upwelling (mainly SW winds), especially in spring-summer [e.g., Cornejo et al., 2007]. The upwelling regime ensures a high supply of nitrate and other nutrients that enhance both primary productivity and bacterial activity [Vargas et al., 2007]. Furthermore, Silva and Neshyba  reported very low oxygen waters (<1 ml L−1) associated with Equatorial Subsurface Waters (ESSW) moving southwards through the Peru Chile Undercurrent (PCU) off Peru-Chile (5° to 45°S). This undercurrent has been described as having a denitrification regime, at least off Peru and northern Chile [Strub et al., 1995]. These reports suggest complex dynamics for the factors controlling N2O production in this area.
2.1. Study Area
 All samples and data were collected during the BIOSOPE cruise (October–December 2004, austral spring) on board the French ship R/V L'Atalante, covering an extension of 6000 km, approximately. The cruise was divided into two parts: Leg 1 from Marquises Islands (∼140°W, 8°S) to Easter Island (∼109°W, 27°S) and Leg 2 from Easter Island to Talcahuano (∼73.1°W, 36.7°S), Chile. N2O determinations were made continuously at the surface (from 132°W, 13°S to 73.5°W, 33.5°S, 130 km offshore) and by discrete profiles. For discrete profiles, three stations were chosen: GYR (26.07°S, 113.99°W), EGY (31.90°S, 91.41°W), and UPX (34.58°S, 72.43°W). These stations represented different oceanographic regimes: GYR and EGY are at the middle and on the eastern border of the South Pacific Gyre, respectively, and UPX is in the coastal upwelling zone (33 km offshore). Furthermore, we made shallow (0 to 200 m) profiles at selected stations (Sta18, Sta19, Sta20, Sta21) in order to compare continuous with discrete N2O determinations. The continuously sampled transect and the location of the discrete sampling stations are shown in Figure 1.
2.2. Oceanographic Data
 Temperature and salinity profiles were obtained with a Seabird SBE 911 + CTD, every three hours for six days at the stations GYR and EGY and for three days at the UPX station.
 Herein, the surface layer (SL) was defined as by Sprintall and Roemmich ; the base was defined as the depth at which the temperature is equal to the coldest sea surface temperature (SST) observed locally in data from a long time series. In this case, we used an approximation of the results of these authors for the Subtropical South Pacific, this is, 180 m between 140°W and 90°W (which includes GYR and EGY), and 40 m at 72°W. Thus defined, the SL was used to balance the N2O passing through the top and the bottom of this layer.
 Current velocities were measured with two vessel-mounted, acoustic Doppler current profilers (ADCP) from RD Instruments. At GYR and EGY, a 75-KHz, long-range, 16-m depth resolution ADCP was used, whereas the ADCP used at UPX had 300 KHz, was short-range, and had a depth resolution of 4 m.
 Surface Chl-a was determined as total Chl-a (Chl-a plus divinyl Chl-a) as described by Raimbault et al.  and averaged for the first 200 m of the water column. Oxygen was obtained with a Seabird SBE43 oxygen sensor mounted in a rosette system and calibrated with Winkler titration.
2.3. Continuous N2O Sampling
 N2O surface concentrations were measured continuously along the track of the cruise with a Shimadzu Gas Chromatograph equipped with a 63Ni electronic capture detector (ECD) and a Pora PLOT-Q capillary column from J&W Scientific. The gas chromatograph system was attached to an automated control system (SIMON) developed in the Laboratory of Oceanographic Processes and Climate, University of Concepción, Chile, based on Ray Weiss's (Scripps Institution of Oceanography, La Jolla, Ca.) scheme for a control-sampling unit and equilibrator (Weiss equilibrator). The equilibrator design is described by Johnson . Surface water samples were pumped continuously from a depth of 3 m into the Weiss equilibrator at a water flow of 20 L min−1 and showered into the equilibrator; the flow was maintained constant during the entire cruise. The sample gas obtained from the equilibrator (headspace) was carried to the sampling unit attached to an automatic control unit. The sampling unit had two electronically controlled valves and a 1-mL sampling loop that contained the gas sample finally injected into the gas chromatograph. The system also took air samples directly from the bow of the ship, moving them to the working laboratory through Dekabon tube. The automatic control system alternated as follows: standard, equilibrator sample, standard, air sample. Each cycle took about 30 min. The response of the chromatograph to the N2O concentration was tested with three Micromat standard gas mixtures (100 ppb, 500 ppb, 1000 ppb) and was linear within this concentration range. The air concentration was determined along the cruise track, being 317 ± 2 ppb (n = 124, 95% confidence); this concentration was used to continuous N2O flux calculations. The amount of the sample gas in the loop depended directly on the temperature of the sampling loop, limiting the accuracy of the results for these injection quantities; however, the temperature of the sampling unit was held constant.
2.4. Discrete N2O Sampling
 N2O samples were always collected one hour before sunrise with 12-L Niskin bottles attached to a CTD-O rosette. The samples for N2O analyses were transferred directly into 125-mL glass flasks (duplicate), preserved with HgCl2, and sealed with butyl rubber stoppers. The N2O was extracted from the sample by sparging with helium and then carried to a pre-concentration loop, which is a stainless steel tube (4.35 mm internal diameter) packed with glass beads (Flushing GH, 60/80 mesh). Finally, the N2O was determined with an HP 6890 gas chromatograph equipped with a Pora PLOT-Q capillary column and attached to a Finnigan MAT252 mass spectrometer. A detailed description of the analytical procedure is given by Yamagishi et al. . The analytical accuracy of the N2O concentration analyses was better than 1%.
2.5. Air-Sea N2O Exchange
 Continuous sampling data were used to determine N2O fluxes at the air-sea interface, Fair (μmol day−1 m−2), using the relation
where kw (m s−1) is the gas transfer velocity depending on wind speed, Cw is the N2O concentration (nM), and Csat is the N2O concentration at relative equilibrium with the atmospheric concentration according to the solubility parameterization of Weiss and Price . The temperature of the seawater in the equilibrator was recorded continuously, and the N2O concentration was corrected to the sea surface temperature using
where F(eq) and F(SST) are the solubility factor of N2O at the equilibrator and sea surface temperatures, respectively, and Ceq is the N2O concentration in the equilibrator's seawater. The calculation of kw was made using the parameterization of Nightingale et al. 
where u is the wind speed (m s−1) and Sc is the Schmidt number for N2O; that is, the relationship between viscosity and the diffusion coefficient of N2O in water that depends on the temperature and salinity of the seawater. For N2O, the Schmidt number as a function of temperature (T in °C) is given by Wanninkhof 
The wind speed was measured on board and normalized to 10 m height by using the relationship of Garratt .
 The surface concentrations obtained by SIMON were compared with the discrete surface samples in order to test the results obtained with the continuous measurements. The standard deviation for the averaged surface concentrations obtained by SIMON at these points (when the ship was static) was 2%. The averaged difference between the concentrations obtained by the two methods was less than 5%, bestowing reasonable confidence in the data obtained by SIMON.
 The air-sea fluxes (Fair) at the stations GYR, EGY, and UPX were calculated following the same procedure as the continuous fluxes but using N2O sea-surface concentrations obtained from discrete samples as the average SL N2O concentrations.
2.6. Diffusive Cross-Pycnocline N2O Exchange
 N2O fluxes through the pycnocline due to turbulent diffusivity (Fdeep) at GYR, EGY, and UPX were estimated as follows:
where K(z) is the diffusion coefficient at depth z. The K(z) was determined using the parameterization of Pacanowski and Philander  (hereinafter PP81), which is based on the bulk Richardson number (balance between stabilizing and turbulent forces)
where N is the buoyancy frequency and S is the magnitude of the vertical gradient of the horizontal velocity. The buoyancy frequency was determined using temperature and salinity data from CTD measurements
where g is the gravity acceleration, ρ0 is the reference density, and −ρ(z) the mean density profile. The magnitude of the vertical gradient of the horizontal velocity was determined as follows:
where u and v are the eastward and northward components of the velocity measured by the ADCP.
The CTD and ADCP data used for the determination of turbulent diffusivity were averaged over 6-day (GYR, EGY) and 3-day (UPX) periods in order to filter out tides, inertial waves, and other higher-frequency oscillations.
2.7. Determination of Ekman Transport
 Ekman transport in the upwelling zone (UPX station) was determined using 72-h averaged wind data from the shipboard measurements to calculate the alongshore wind stress (τy)
where ρair is the typical air density at sea level (1.2 kg m−3), Cd is a drag coefficient (0.0013), ∣U∣ is the magnitude of wind velocity, and v is the alongshore component of the wind. The alongshore wind stress allows calculating the offshore Ekman transport, Mx, in m2 s−1
where ρss is the representative water density at sea surface (1025.5 kg m−3) and f is the Coriolis parameter. The N2O offshore flux due to Ekman transport (Foff) was calculated as follows:
where the differential term represents the offshore gradient of the surface N2O concentration. For this, we used the average surface concentration at the stations UPX and Sta21, which are separated by 311.44 km. Sta21 is considered to be close to the limit of the coastal-influence zone.
 Considering a Rossby (Rd) radius of deformation for this area of about 30 km [Chelton et al., 1998], the vertical upwelling velocity, W (m s−1), can be estimated from
In order to test our results for Mx, we calculated the offshore Ekman transport using QuikSCAT wind data averaged for the spring-summer period; the difference with the previously calculated value was less than 10%. N2O fluxes due to Ekman transport (FEk) at the UPX station were calculated as follows:
where [N2O]deep is the concentration representative of the layer located immediately below the SL and [N2O]s is the concentration in the SL.
2.8. Contribution of Surface and Subsurface Waters to Air-Sea N2O Flux
 We used the fluxes described above to test the contribution of the ocean under and within the SL to the N2O air-sea flux, given different oceanographic regimes. For this, we made a simple balance between the inward N2O flux though the pycnocline and the outward flux though the air-sea interface (Figure 2a). The balance is represented as follows:
where PI (μmol m−2 d−1) is the integrated N2O production in the SL. Along the coast, we considered the effect of upwelling due to Ekman offshore water transport; thus, our balance was rearranged (Figure 2b) and two new terms (Foff and FEk) were introduced
Although, due to alongshore transport, the meridional N2O flux was considered and estimated, its magnitude can be neglected and it is not included in this balance, as is discussed in section 4.2.
 Also the surface N2O production P (μmol m−3 d−1) was calculated as the integrated surface N2O production (PI) divided by SL thickness.
3.1. Continuous Air-Sea Exchange
 Continuous N2O fluxes are shown in Figure 3. Fluxes were slightly negative during most of Leg 1, from 132°W to 114°W, averaging −0.48 μmol m−2 day−1 (SD 0.44, n = 297) in this section. Eastward from GYR and during most of Leg 2, from 113°W to 81°W, the N2O fluxes were slightly positive, averaging 0.41 μmol m−2 day−1 (SD 0.34, n = 291). From 80°W eastward, an abrupt change was observed in N2O emissions: at 75°W (270 km offshore), the flux rose up to 5 μmol m−2 day−1 and at 73°W (130 km from the coast, where the last continuous data were taken), it reached levels up to 18 μmol m−2 day−1. N2O concentrations determined by SIMON incremented gradually from 130°W eastward until 75°W, where the increase became abrupt.
Figure 3a compares the results of SIMON and the surface temperature and salinity. As the surface waters cooled toward the east (and south), the N2O concentration gradually increased as a result of the inverse solubility-temperature relationship, but air-sea fluxes remained low. The minor variability in temperature and salinity seemed to be related to the diurnal irradiation cycle, but eastward of 75°W, the surface waters cooled abruptly, indicating the offshore influence of coastal upwelling. Continuous N2O fluxes were highly correlated with average surface Chl-a concentrations (R2 = 0.994) (Figure 3b) along 114°S to 73°S, but were less correlated with surface contents of phosphate (R2 = 0.776) and nitrate (R2 = 0.648) (data not shown). These results indicated that the N2O flux was directly related to the phytoplankton biomass. Nutrient levels in the surface water were extremely low in the central part of the gyre, increasing gradually toward 93°W [Raimbault et al., 2008]. These distributions not only reflected the supply of nutrients from subsurface waters but also the regenerated fraction from the SL.
3.2. N2O and O2 Contents Along the Water Column
 Vertical N2O and O2 distributions are shown in Figure 4 (top). N2O was slightly oversaturated at GYR and EGY from the surface to ∼200 m. Oxygen concentrations in this layer at both oceanic stations were quite homogeneous (Figures 4a and 4b).
 Temperature, salinity, and density profiles are shown Figure 4 (bottom). Both oceanic stations had shallow, very light thermoclines (and concomitant pycnoclines) from 10 to 20 m. This surface density gradient seemed to be related to daily irradiation (and wind conditions). A second, deeper temperature gradient (permanent thermocline-pycnocline) was observed from 180 m to ∼450 m. Both the thermocline and pycnocline had steeper gradients at EGY than at GYR (Figures 4d and 4e). The SL, defined as in section 2.2, was useful for our purposes because it was analogous to a local ‘ventilation’ depth, the deepest surface at which the atmospheric influence can be felt through local wind stirring and heat loss [Sprintall and Roemmich, 1999]. In fact, both gases (N2O, O2) varied little over the SL. The average N2O concentrations in the SL (shaded area in Figure 4) were 7.09 nM (SD = 0.32, n = 10) at GYR and 8.07 nM (SD 0.70, n = 10) at EGY, representing 105.1% and 101.4% saturation, respectively. Below the SL, N2O was highly oversaturated (∼230%) at both stations. The N2O profiles at GYR and EGY were almost mirrored by the O2 profiles, indicating a close relationship between N2O production and O2 consumption in the water column.
 At the UPX station, the whole water column was highly N2O-oversaturated, from 230% at the surface to 480% at the N2O maximum (350 m). At this station, the O2 concentrations displayed strong depletion down to ∼10 μmol kg−1 between 150 m and 300 m and did not show a clear relationship with the N2O profile (Figure 4c). A clear pycnocline appeared close to 40 m at UPX (Figure 4f). The density increased almost monotonically until 800 m. The average N2O concentration for the SL (shaded area) defined by this pycnocline was 23.14 nM (SD 1.60, n = 3). No separation of the N2O profile related to stability was observed, indicating different physical behavior than at the oceanic stations.
3.3. Discrete Cross-Pycnocline and Air-Sea N2O Exchange
Table 1 shows the piston velocity (kw), diffusivity coefficient (K(z)), offshore transport (Mx), and upwelling velocity across the pycnocline (W) for the three selected stations. The diffusivity coefficient appeared to be quite constant in the different studied areas. In contrast, the piston velocity increased eastward due to higher wind speeds associated with coastal winds, being about five times larger at UPX than at GYR. Otherwise, our estimation of upwelling velocity (W), in which was on the order of 3 m day−1, appeared to be close to the range of the upwelling velocities previously estimated for summer in this area through a numerical model [Leth and Middleton, 2004].
Table 1. Parameter Obtained for the Surface Layers (SL) at the Three Selected Stationsa
kw (m s−1)
K(z) (m2 s−1)
Mx (m2 s−1)
W (m s−1)
Abbreviations are as follows: SL, surface layers; kw, piston velocity at the sea surface; K(z), vertical diffusivity at the bottom of the surface layer; W, upwelling velocity.
1.02 × 10−5
1.01 × 10−5
2.25 × 10−5
1.00 × 10−5
5.05 × 10−5
1.19 × 10−5
3.15 × 10−5
 N2O fluxes through the boundaries of the SL are listed in Table 2. The air-sea N2O flux in the coastal area greatly exceeded the flux found in the central South Pacific, following the trend of the continuously measured fluxes. Otherwise, cross-pycnocline fluxes due to turbulent diffusion (Fdeep) were smaller than air-sea fluxes (Fair) in all cases. However, at UPX, the air-sea flux was two orders of magnitude higher than the diffusive flux. Part of this difference could be accounted for by upwelling (FEk), which seemed to be more important than diffusivity to cross-pycnocline N2O transport near the coast. As a consequence of offshore transport, we found N2O fluxes of 3.81 μmol m−2 day−1 exported away from the coast. The offshore gradient of the surface concentration (see equation (12)) obtained with discrete samples was 4.6 × 10−5μmol m−4, similar to the gradient calculated using SIMON data (3.9 × 10−5μmol m−4), indicating that our estimation of offshore transport was plausible. The balance between the inward flux through the pycnocline and the outward flux through the air-sea interface and offshore transport (referred to as integrated surface production, PI) is also shown in Table 2. An imbalance was observed at the three selected stations; however, in the coastal area, much higher surface production was concentrated in a smaller SL. In fact, surface N2O production (P) was more than 400 times higher in the coastal area (UPX) that at the oceanic stations (GYR and EGY).
Table 2. N2O Fluxes Though Boundaries of the Surface Layer (Fdeep, FEk, Foff, Fair), Integrated Surface N2O Production (PI), and Surface Production (P)
Fdeep (μmol m−2 day−1)
FEk (μmol m−2 day−1)
Foff (μmol m−2 day−1)
Fair (μmol m−2 day−1)
PI (μmol m−2 day−1)
P (μmol m−3 day−1)
1.53 × 10−3
2.06 × 10−3
 Analytical error propagation was evaluated for the measurements, giving a 25% error for the central gyre's air-sea flux, which was on the same order of magnitude as previous estimations [Bange et al., 2001]. Explanations of the calculation of this error can be found in Text S1.
4.1. Surface N2O Distribution and Ocean-Atmosphere Fluxes
 The subtropical South Pacific Gyre (SPG) is a region of anticyclonic wind stress curl with convergent Ekman transport to the core of the gyre (negative Ekman pumping), a very deep pycnocline, and very low primary productivity [Fiedler and Talley, 2006]. The extreme oligotrophy of this study area, characterized by mean annual Chl-a values of 0.02 mg m−3 [Claustre et al., 2008] and a severe nitrogen limitation [Bonnet et al., 2008], indicates that most of the nutrients are primarily supplied by regeneration. In fact, Raimbault and Garcia  showed that active regeneration processes fuel up to 95% of the biological nitrogen demand. Primary production in this area is very low according to the surface Chl-a concentration (Figure 3b), and most of the biological activity is restricted to the base of the mixed layer (between 200 and 250 m) [Grob et al., 2007]. These characteristics explain N2O surface concentrations in near-equilibrium with atmospheric N2O and concomitant low N2O fluxes in the SPG. Under such extreme biogeochemical conditions, the production of gaseous forms of nitrogen should be limited by organic matter availability, which, in turn, recycles ammonium. Otherwise, the change from slightly negative to slightly positive fluxes westward and eastward from the GYR station does not seem to be correlated with any particular changes in nutrients or organic matter content [see Raimbault et al., 2008]. Negative fluxes found in the center of the gyre initially suggest a sink of atmospheric N2O. Regarding the limited solubility of this gas at high surface temperatures found in this part of the Pacific Ocean, this undersaturation should be explained by an N2O consumption process that acts more quickly than the air-sea exchange in this zone. Recently, N2O fixation (as an alternative to N2 fixation) has been measured as a mechanism that occurs in the eastern South Pacific (ESP) (L. Farías et al., N2O fixation in the eastern South Pacific Ocean, manuscript in preparation, 2010), offering a feasible explanation for the negative fluxes. However, we cannot rule out the possibility that the undersaturation is due to an analytical error associated with the measurements. Standard deviations of the data reach 92% of the averaged value (see section 3.1), and the propagation of the analytical error evaluated for the discrete air-sea flux was 25% of the value (see Text S1), which was on the same order as a previous estimation [Bange et al., 2001].
 Finally, an abrupt increase in N2O emissions to the atmosphere is observed eastward from 75°W, showing the effect of the highly productive South Pacific eastern boundary. Similar N2O flux patterns were found along a transect (along 32.5°S from Tahiti to Valparaiso) performed with SIMON during Leg 2 of the BEAGLE 2003 cruise (M. Cornejo et al., N2O exchange off central Chile: From strong coastal source to a significant oceanic sink, submitted to Progress in Oceanography, 2010), where N2O fluxes abruptly rose from 74°W eastward. The hydrographic features of the eastern South Pacific revealed the presence of cold, salty, nutrient-rich, oxygen-poor ESSW [Shaffer et al., 1995]. The extent of the apparent influence of cold, coastal, nutrient-rich water that exceeds 250 km cannot be explained just by the local dynamics of coastal upwelling, considering the internal Rossby radius of deformation for this area of about 30 km [Chelton et al., 1998]. The extent of the coastal influence area has been studied through satellite surface Chl-a data [Yuras et al., 2005] and a clear separation exists between the coastal upwelling area (restricted to 20–30 km from the coast) in relation to the spring-summer wind pattern and an offshore regime (50–200 km) that is not directly related to coastal upwelling. This offshore regime may be related to mesoscale dynamics (eddies and meanders) and their interactions with large-scale circulation processes that promote the offshore transport of coastal waters, as proposed for this specific area by Leth and Shaffer . In this sense, N2O production may be driven by the same physical mechanism that drives primary productivity in coastal or near-coastal areas.
 Moreover, increased N2O fluxes are correlated with an intensification of the oxygen minimum in subsurface waters from 75°W to the east, reaching less than 20 μmol L−1 (Figure 5). This relationship is expected because low oxygen levels enhance N2O production processes [Nevison et al., 2004] and subsurface waters can be transported into the mixed layer by advection and mixing processes. However, the evident relationship between N2O emissions and surface primary productivity along the whole track of the cruise (Figure 3a), especially in the area of coastal influence, indicates that photoautotrophic and chemolithotrophic (nitrification linked to organic matter remineralization) activities are closely linked. However, in our opinion, this relationship should be indirect, through low oxygen concentrations associated with the respiration of organic matter as well as the supply of labile nitrogen (likely NH4+) for nitrous oxide production processes. Obviously, the same relationship should be expected between N2O fluxes and nutrients (nitrate and phosphate) according to the distribution of the nutrients found during the cruise [Raimbault et al., 2008], although the correlation between the N2O flux and nutrients (nitrate and phosphate) is less significant than that of the N2O flux and Chl-a. Broadly speaking, the N2O emission is directly related to the trophic status of the surface water.
4.2. Cross-Pycnocline Transport and Surface N2O Production
 Turbulent diffusion is the main mechanism of diapycnal mixing in the world ocean. Table 3 lists diffusivity coefficients for several oceanic regions that were estimated with different techniques. These data were obtained over different timescales, with some long-term averages (e.g., tracer-based estimates) and other ensembles of short-term estimates (e.g., microstructure estimates). Furthermore, some were taken in the permanent thermocline (mainly those obtained at tropical stations) and others in the seasonal pycnocline. The closest estimates of turbulent diffusivity are given by direct measurements through tracer-release experiments and velocity microstructure profiles. However, indirect determinations of turbulent diffusivity are also possible, the most commonly used being the parameterization of PP81. Although this model was developed for the Equatorial Pacific, it has been successfully applied elsewhere, including the Antarctic Polar Front [Cisewski et al., 2005], the Weddel Sea [Muench et al., 2002], and the Southern Ocean [Howard et al., 2004]. Cisewski et al.  compared diffusivity values obtained with the PP81 model, velocity microstructure profiles, and dye dispersion experiments; they found good agreement between the three methods, considering the uncertainties inherent to the estimations. This background encourages us to estimate turbulent diffusivity using the PP81 parameterization in the three oceanic study regions.
Table 3. Cross-Pycnocline Diffusivity Coefficients Found in the Scientific Literaturea
K(z) (m2 s−1)
Method acronyms used are as follows: B.T.S., bomb tritium spreading; T.R.E., tracer release experiment; V.M.P., velocity microstructure profiles; B.M., box model; H71, parameterization of Haney ; PP81, parameterization of Pacanowski and Philander .
 In the SPG, where wind-driven vertical advection is not expected to play a significant role, diffusivity should govern the contribution made by the N2O produced under the pycnocline to the N2O released into the atmosphere. According to our results (listed in Table 2), at the GYR and EGY stations, the cross-pycnocline flux accounts for close to 20% and 30% of the N2O released into the atmosphere, respectively. Cross-pycnocline N2O fluxes falling between the values that we found at GYR and EGY have also been reported for the Subtropical North Pacific Gyre, specifically at the Aloha Station [Dore et al., 1998; Popp et al., 2002] and in the southern ocean [Law and Ling, 2001]; these results are listed in Table 4. As no significant lateral advection (outward or inward) is expected, the difference between the cross-pycnocline and air-sea fluxes must be supplied by surface N2O production. Our estimates of this shallow N2O source are shown in Table 2 as integrated surface production (PI). It is important to note that the production at both oceanic stations is integrated into the very thick SL (180 m) and that the production per volume unit (P) is on the order of 10−3μmol m−2 day−1. Either way, these results agree with a shallow N2O source found previously for oceanic areas [Dore et al., 1998; Popp et al., 2002] and can be taken as evidence of the occurrence of nitrification in the photic layer, as stated by Yool et al. . Otherwise, isotopomeric N2O measurements made by Charpentier et al.  during the same cruise show a clear change of N2O production processes in the high-stability layer close to 300 m (GYR) and 250 m (EGY). According to the hypotheses of these authors, the N2O produced in this specific layer can be attributed to nitrifier denitrification. If this is true, this process contributes approximately 15% of the N2O released to the atmosphere in this part of the ocean, considering the extent of the nitrifier denitrification in the subsurface layer and the amount of N2O transported through the pycnocline by turbulent diffusion. This represents a significant contribution to the global atmospheric N2O budget given the extension of the subtropical gyres in the word ocean.
Table 4. Cross-Pycnocline N2O Flux Found in the Scientific Literature
 Turbulent diffusion in coastal areas appears to be on the same order of magnitude as oceanic diffusivity [Ledwell et al., 2004; Morell et al., 2001; Oakey and Greenan, 2004], even in environments with a significant upwelling regime like the eastern Equatorial Pacific [Cline et al., 1987]. In fact, the cross-pycnocline and air-sea N2O fluxes in coastal areas without important upwelling are only slightly higher than those found in oceanic regimes (Table 4). The main difference between our coastal (UPX) and oceanic (GYR, EGY) stations, in terms of their N2O profiles, is the lack of relationship with the pycnocline at the coastal station and, consequently, the absence of a clear separation between deep and surface N2O. The obvious conclusion is that the highly dynamic upwelling in this area drives the N2O distribution, and the influence of turbulent diffusivity is overwhelmed by wind-driven vertical advection, a much more important factor for the N2O distribution in this area. To achieve this factor at the coastal station (UPX), we introduced one term (FEk) into our balance that represents the N2O advected upward by upwelling. Although this term is just a coarse approximation of the influence of coastal circulation on N2O dynamics, it gives us a first idea about the relative importance of vertical advection and turbulent diffusion.
 The possibility that N2O contributions to our model were due to alongshore circulation at the UPX station was also examined. Measurements made by our laboratory (M. Cornejo et al., submitted manuscript, 2010) along the Chilean coast showed surface meridional bands of N2O concentrations. Furthermore, data were examined and no significant longitudinal gradient was found in the first 40 m depth between 34°S and 36°S. Alongshore surface waters are dominated by Eastern South Pacific Transition Water (ESPTW) transported northward by the Humboldt Current [Silva and Konow, 1975]; the ESPTW does not come from a high N2O productivity zone. In turn, N2O-rich ESSW are transported southward by the Peru-Chile Undercurrent, which runs under the pycnocline until 40°S and that, at the latitude of the UPX station, is found between 100 and 200 m [Silva et al., 2009]. Hence, this alongshore N2O contribution is included in the upward N2O flux through the pycnocline at UPX station (Fdeep and FEk), a conclusion that is supported by isotopic and isotopomerical data taken during the same cruise [Charpentier, 2008].
 On the other hand, as long as a large part of the N2O released into the atmosphere in the coastal area comes from the thin SL (compared to the thick SL in the gyre region) and considering that coastal upwelling areas are the most important sources of marine N2O to the atmosphere, it can be concluded that a great part of marine N2O emissions are directly produced within the SL of coastal upwelling areas. Our rate estimation of N2O production (1.16 μmol m−3 day−1) suggests elevated microbiological activity related to the high production of organic matter in the SL. Therefore, processes yielding N2O should occur within the photic layer (35 m deep, 1% of surface light); these may be reductive processes associated with microsites in the organic suspended matter (partial denitrification, nitrifier denitrification). The occurrence of N2O-reductive processes in the SL at the UPX station is also supported by isotopomerical data [Charpentier, 2008]. Furthermore, if it is assumed that surface N2O is produced by nitrification, and considering 0.25% N2O leakage from this process [Goreau et al., 1980], a nitrification rate of about 464 nmol l−1 d−1 is necessary to account for the estimated N2O production. However, Raimbault and Garcia  reported surface nitrification rates for the area on the order of 30 nmol l−1 d−1, strongly suggesting the existence of other processes that produce this gas in the mixed layer at the UPX station.
 It must be noted that the estimated magnitude of the N2O fluxes at the air-sea interface are strongly model-dependant. The most commonly used models for estimating air-sea gas fluxes are those proposed by Liss and Merlivat  and Wanninkhof . Observational radio carbon data show that the Liss and Merlivat  model greatly underestimates the gas fluxes [Sarmiento and Gruber, 2006]. On the contrary, the model proposed by Wanninkhof  is a quadratic curve fitted such that, when averaged over the global wind speed, it is in agreement with the global mean transfer velocity determined from the oceanic uptake of nuclear bomb-derived radio carbon [Broecker et al., 1985]. The parameterization of Nightingale et al. , which is based on a dual (volatile and non-volatile) tracer, shows a wind speed dependence between those of Liss and Merlivat  and Wanninkhof  and can account for non-wind-related factors such as dispersive dilution. Either way, we estimate the air-sea flux at the three stations using both alternative methods [Liss and Merlivat, 1986; Wanninkhof, 1992] in order to test the results of our balance for SL (Table 5). The two models at the three stations yield air-sea fluxes higher than our estimated cross-pycnocline fluxes, indicating that our basic conclusion – that there is a shallow N2O source – is robust and not model-dependant.
Table 5. N2O Fluxes Estimated Using Alternative Modelsa
 Increased N2O fluxes toward the coast and their close relationship with average Chl-a (Figure 3b) suggest that N2O emissions are related to the mechanism that enhances photoautotrophic productivity and, therefore, that high organic matter production implies high N2O production, at least in the surface ocean. The relationship between organic matter production and N2O production is related to the supply of reduced forms of labile nitrogen though organic matter remineralization to ammonium oxidizers, the most probable candidates for N2O production in the SL.
 Based on a simple box model, we can establish that, in the SPG, roughly 70% to 80% of the N2O released into the atmosphere by the ocean is produced in situ in the mixed layer (despite the low N2O production per volume unit) as a part of the biological recycling of nitrogen. Thus, the amount of N2O produced at the surface is not driven by factors that have traditionally been mentioned as inhibitors of microbial activity in the ocean surface (e.g., light or oxygen). Typical nitrification rates in the SL of the SPG during the same cruise were up to 5 μmol N m−3 day−1 [Raimbault and Garcia, 2008]; if we take this value and the N2O production rate in the SL at the GYR station (center of SPG), we can estimate that 0.06% of the nitrogen taken up during nitrification is released as N2O, similar to the yield found in ammonium-oxidizing bacteria cultures at high oxygen concentrations [Goreau et al., 1980].
 As expected, we found very small N2O emissions from the subtropical South Pacific; nonetheless, given the large area of the SPG, it can be considered to be an important source of N2O to the atmosphere. Given the agreement between our discrete and continuous air-sea flux estimations, a representative N2O air-sea flux of 0.4 μmol day−1 m−2 (average of air-sea fluxes found at GYR and EGY) can be assumed for a square subtropical area crossed diagonally by the cruise's track between 114°W, 25°S and 80°W 35°S (Figure 1), where the N2O flux remained nearly constant. With this, the annual N2O emission estimated for this area reaches 50 Gg of N2O (32 Gg N a−1). According to the recent oceanic N2O source of Bange , which reaches 7 Tg N a−1, our estimation represents 4.6% of the annual global emission.
 In the upwelling zone, high N2O production can be seen throughout the entire water column, and wind-driven vertical mixing and offshore water transport make it difficult to evaluate separately the N2O that is produced above or under the pycnocline. As an exercise, we estimated the upwelling velocity necessary to account for the N2O exported through offshore transport and atmospheric outgassing, assuming that there was no surface N2O production. The upwelling velocity (W) obtained from this estimation was ca. 11 m day−1, which is rather high for W and even higher than the typical predictions of W for the study area [Leth and Middleton, 2004]. Thus, a surface N2O source must be present in order to explain the N2O emissions in the coastal region. According to our estimation, at least 60% of the N2O emitted to the atmosphere is produced in the first 40 m (the depth of the mixed layer), which gives an approximation to the importance of the ocean for the atmospheric greenhouse gases budget.
 The authors thank Gerard Eldin for allowing us to use his ADCP data and Patrick Raimbault for allowing us to use surface Chl-a data. Narin Boontanon is greatly thanked for the analysis of N2O samples. Thanks to Osvaldo Ulloa, Gadiel Alarcon, Mauricio Gallegos, and the crew of the R.V. L'Atalante for their help during the BIOSOPE Cruise. José Charpentier thanks to Verónica Molina, Dernis Mediavilla, and Hermann Bange for their scientific discussions that greatly improved the manuscript. The authors thank Ray Weiss for his advice in the building of SIMON; and Victor Villagrán, Mauricio Gallegos, and Osvaldo Ulloa for their efforts to build and test SIMON. Dominique Tailliez and Claudie Bournot are warmly thanked for their efficient help in CTD rosette management and data processing. Hervé Claustre and Antoine Sciandra are also greatly thanked for inviting us to participate in the BIOSOPE cruise. Partial financial assistance was provided by the FONDAP-COPAS center (Project 150100007). J. Charpentier was supported by a grant from the MECESUP UCO002 project. This work was supported by the Chilean National Commission for Scientific and Technological Research through FONDECYT grant 1070518 (Laura Farías).