Low rates of nitrogen fixation in eastern tropical South Pacific surface waters

Significance

We present direct, field-based measurements of low nitrogen fixation rates in the eastern tropical South Pacific (ETSP) Ocean demonstrating that N₂ fixation plays a minor role supporting export production regionally. These results are in contrast to indirect estimates that the highest global rates of N₂ fixation occur in the ETSP. The low N₂-fixation rates occur in a region with relatively high surface ocean phosphate concentrations (and low nitrate concentrations) but where atmospheric iron deposition rates are diminishingly low. Consequently, these results indicate that the ETSP hosts a minor fraction of global N₂-fixation fluxes and that low nitrate to phosphate concentration ratios alone are insufficient to support high N₂-fixation fluxes.

Abstract

An extensive region of the Eastern Tropical South Pacific (ETSP) Ocean has surface waters that are nitrate-poor yet phosphate-rich. It has been proposed that this distribution of surface nutrients provides a geochemical niche favorable for N₂ fixation, the primary source of nitrogen to the ocean. Here, we present results from two cruises to the ETSP where rates of N₂ fixation and its contribution to export production were determined with a suite of geochemical and biological measurements. N₂ fixation was only detectable using nitrogen isotopic mass balances at two of six stations, and rates ranged from 0 to 23 µmol N m⁻² d⁻¹ based on sediment trap fluxes. Whereas the fractional importance of N₂ fixation did not change, the N₂-fixation rates at these two stations were several-fold higher when scaled to other productivity metrics. Regardless of the choice of productivity metric these N₂-fixation rates are low compared with other oligotrophic locations, and the nitrogen isotope budgets indicate that N₂ fixation supports no more than 20% of export production regionally. Although euphotic zone-integrated short-term N₂-fixation rates were higher, up to 100 µmol N m⁻² d⁻¹, and detected N₂ fixation at all six stations, studies of nitrogenase gene abundance and expression from the same cruises align with the geochemical data and together indicate that N₂ fixation is a minor source of new nitrogen to surface waters of the ETSP. This finding is consistent with the hypothesis that, despite a relative abundance of phosphate, iron may limit N₂ fixation in the ETSP.

Rates of marine photosynthesis significantly influence the concentration of carbon dioxide in the atmosphere, particularly on glacial–interglacial time scales (1). In broad regions of the low latitude ocean, rates of photosynthesis are thought to be limited by the availability of fixed nitrogen (N), thus restricting the biological sequestration of carbon in the deep ocean (2). The inventory of fixed N in the ocean is regulated by the balance between the principal, biologically mediated N sources and sinks, di-nitrogen (N₂) fixation, and denitrification. Despite the critical importance of the fixed N inventory in regulating photosynthesis, the spatial distribution of N fluxes to the ocean remain poorly constrained.

For decades, both geochemical and biological analyses (3–6) have pointed to a spatial decoupling of N fluxes to and from the ocean, with the highest rates of N₂ fixation documented in the tropical North Atlantic and the largest N losses concentrated in the Eastern tropical Pacific and Arabian Sea. This spatial decoupling of N inputs and outputs corresponds to a temporal decoupling, requiring the time scale of ocean circulation for N₂ fixation to respond to changes in rates of denitrification and vice versa. Despite this apparent decoupling in the modern ocean, paleoceanographic evidence indicates that N fluxes to and from the ocean have been closely balanced at least over the past 20 kyr (7, 8), although more recent work suggests transient imbalances may have occurred (9). Although N loss in the ocean is largely constrained to the sediments and water column oxygen deficient zones (ODZs) in the Eastern Pacific and Arabian Sea, similar constraints on the location of N₂-fixation fluxes to the ocean are lacking and thus the degree to which marine N sources and sinks have been coupled through time remain uncertain.

Both remote sensing (10) and biogeochemical modeling (11) have suggested that the highest rates of N₂ fixation may occur in the surface waters of the Eastern and Central tropical Pacific, which carry the geochemical signatures of N loss occurring in nearby sediments and water column ODZs (5, 6, 11). This spatial coupling of N inputs and losses in the ocean would provide the proximity necessary to allow feedbacks within the N cycle to respond to the changes in the N inventory on the decade- to hundred-year time scales apparently required by the paleoceanographic record (8, 11). Furthermore, these findings are consistent with paleoceanographic evidence for denitrification influencing the location of N₂-fixation fluxes (12). Nonetheless, these indirect estimates of high N₂-fixation rates occurring in the Eastern tropical Pacific have remained controversial in part because they challenge the well-documented high iron requirement of the enzyme nitrogenase, which carries out N₂ fixation (13, 14). The surface waters of the Central and Eastern tropical South Pacific (ETSP) receive some of the lowest atmospheric dust fluxes (15), raising the question how the largest global marine N₂-fixation fluxes (11) could occur in a region where iron sources are scarce. Indeed, previous field studies found N₂-fixation rates that were at or below detection limits and inferred that iron limits primary productivity in the region (16). Other models of the global distribution of marine N₂-fixation fluxes emphasize the iron requirements of diazotrophs and shift the bulk of predicted N₂-fixation fluxes to the Southwest Pacific Ocean proximal to larger atmospheric dust sources (17–19).

Given the uncertainty in the spatial distribution of N₂ fixation in the global ocean, field campaigns making direct measurements of N fluxes to the ocean remain a priority, in particular in the expansive ETSP gyre, where few such measurements have been made (20). Here, we present a suite of measurements from the ETSP gyre, collected between 80°W to 100°W along 20°S (Fig. S1), evaluating N₂ fixation on ∼1- to 3-d time scales, including short-term (i.e., 24 h) ${}^{15}\text{N}_2$ incubation-based fixation rates. N₂-fixation rates and their contribution to export production were also evaluated with a N isotope mass balance which compared the isotopic composition of sinking particulate N (PN_sink) collected for 35–70 h in floating sediment traps with that of subsurface ${\text{NO}}_3^{-}.$ Together with a parallel analysis of the abundance and expression of the dinitrogenase reductase iron protein encoded by the NifH gene from the same cruises (21), as well as measurements of the isotopic composition of PN_sink collected over 13 mo in a deep-moored sediment trap (22), these datasets constrain the role of N₂ fixation in the ETSP gyre.

Fig. S1.

Fractional contribution of subsurface NO₃+NO₂ and N₂ fixation to export production in the ETSP based on the results of the δ¹⁵N budgets. Station numbers are overlain upon Moderate Resolution Imaging Spectroradiometer (MODIS) surface ocean chlorophyll for February 2010 (A) and March 2011 (B). The relative magnitude of the PN_sink flux is represented by the area of the circle, with the fraction of the PN_sink flux supported by subsurface ${\text{NO}}_3^{-}+{\text{NO}}_2^{-}$ in green and the fraction supported by N₂ fixation in red.

Results

N₂-Fixation Rates and Their Contribution to Export Production in the ETSP.

N₂-fixation rates calculated with geochemical methods track the distinct imprints that diazotrophs leave on water column nutrient stoichiometry and isotopic composition. The approach used here relies on the unique isotopic signature $\mathrm{(}{\mathrm{\delta }}^{15}\text{N}=\left\{\left[{{\left({}^{15}\text{N}/{}^{14}\text{N}\right)}_{\text{sample}}/\left({}^{15}\text{N}/{}^{14}\text{N}\right)}_{\text{reference}}\right]\mathrm{-}1\right\}\times 1\text{,}000\text{,}$ with atmospheric N₂ as the reference) of diazotrophic biomass [δ¹⁵N_Nfix = −2–0‰ (23, 24)], which is low compared with that of mean ocean nitrate [δ¹⁵N_NO3, ∼5‰ (25)]. Because the δ¹⁵N of the two dominant sources of new N to surface waters [subsurface ${\text{NO}}_3^{-}$ + nitrite $\left({\text{NO}}_2^{-}\right)$ and N₂ fixation] have distinct δ¹⁵N values, the δ¹⁵N of these two sources can be used together with the δ¹⁵N of PN_sink (δ¹⁵N_PNsink) to construct a two end-member mixing model (δ¹⁵N budget) to estimate the relative importance of each for supporting export production (26–28) (Fig. S2). The fractional importance of N₂ fixation for supporting export production (x) can be expressed as

$${\text{δ}}^{15}{\text{N}}_{\text{PNsink}}=\text{x}\left({\text{δ}}^{15}{\text{N}}_{\text{Nfix}}\right)+\left(1-\text{x}\right)\left({\text{δ}}^{15}{\text{N}}_{\text{NO}3\mathrm{+}\text{NO}2}\right).$$ [1]

Rearranging and solving for x yields

$$\text{x}=\left({\text{δ}}^{15}{\text{N}}_{\text{NO}3\mathrm{+}\text{NO}2}-{\text{δ}}^{15}{\text{N}}_{\text{PNsink}}\right)/\left({\text{δ}}^{15}{\text{N}}_{\text{Nfix}}+{\text{δ}}^{15}{\text{N}}_{\text{NO}3\mathrm{+}\text{NO}2}\right).$$ [2]

Multiplying x by the PN_sink flux provides a time-averaged N₂-fixation rate that can be compared with short-term N₂-fixation rate measurements.

Fig. S2.

Idealized δ¹⁵N budget. In a δ¹⁵N budget the isotopic composition of the quantitatively dominant export flux from the euphotic zone (“δ¹⁵N_PNsink”) reflects the proportional contributions of the two dominant sources of new nitrogen to surface waters, N₂ fixation (“N_fix”) and subsurface nitrate+nitrite (NO₃+NO₂), which in this example have been assigned δ¹⁵N of −1‰ and 9‰, respectively. In the case where the δ¹⁵N_PNsink = 1.5‰, 75% of export production is supported by N₂ fixation and 25% from NO₃+NO₂ (A); in the case where the δ¹⁵N_PNsink = 6.5‰, 75% of export production is supported by NO₃+NO₂ and 25% from N₂ fixation (B).

Floating sediment trap δ¹⁵N_PNsink and water column δ¹⁵N_NO3+NO2 measurements from two cruises to the ETSP (January to February 2010 and March to April 2011) were used to construct the δ¹⁵N budgets (Table 1 and Fig. S1). This approach depends on the careful characterization of both the δ¹⁵N_Nfix as well as the δ¹⁵N_NO3+NO2 end-members. The δ¹⁵N_Nfix end-member is constrained by field measurements of diazotrophic biomass δ¹⁵N that range between approximately −2‰ and 0‰ (23, 24); lower δ¹⁵N_Nfix values result in a lesser contribution of N₂ fixation to export production (Eq. 2). Here, we take δ¹⁵N_Nfix = −1‰. Uncertainty in the δ¹⁵N_NO3+NO2 end-member largely depends on the degree to which δ¹⁵N_NO3+NO2 varies with depth in the upper thermocline and whether or not there is unconsumed ${\text{NO}}_3^{-}+{\text{NO}}_2^{-}$ in surface waters. At these stations, mixed layer concentrations of ${\text{NO}}_3^{-}+{\text{NO}}_2^{-}$ were typically at or below detection limits (0.05 µM) (except in 2011 at the more northern Station 13, where the mixed-layer ${\text{NO}}_3^{-}+{\text{NO}}_2^{-}$ concentration was between 0.3 and 0.6 µM) and increased rapidly below the base of the euphotic zone (Fig. 1 and Table S1). Effectively complete consumption of surface water ${\text{NO}}_3^{-}+{\text{NO}}_2^{-}$ in this region of the ETSP alleviates concerns with the choice of isotope effect used for ${\text{NO}}_3^{-}$ assimilation in δ¹⁵N budget calculations (29, 30). Profiles at eastern stations (i.e., Stations 1 and 13) show isotopic features indicative of dissimilatory ${\text{NO}}_3^{-}$ reduction occurring in the nearby ODZs of the ETSP (Fig. 1) (6, 31–33). Additional increases in δ¹⁵N_NO3+NO2 coincident with the deep chlorophyll a maximum (Stations 1, 3, and 13; Fig. 1) indicate that ${\text{NO}}_3^{-}$ assimilation elevated the δ¹⁵N_NO3+NO2 within and immediately below the base of the euphotic zone (34, 35).

Table 1.

Results of the δ¹⁵N budgets

Year and station	δ15NNO3+NO2 (‰)	δ15NPNsink (‰)	fnfix (%)	Rnfix δ15N budget (µmol N m−2 d−1)	Rnfix short-term incub (μmol N m−2 d−1)	Short-term incubation
						fnfix (%)	δ15NPNsink (‰)
2010
Station 1	11.5 ± 0.2	11.7 ± 2.1	0 ± 15	0 ± 48	37 ± 17	12	10.0
Station 3	11.1 ± 0.01	13.7	0	0 ± 0	98 ± 129	362	−1
Station 5	9.8 ± 0.9	11.1 ± 0.1	0 ± 0	0 ± 0	51 ± 20	170	−1
2011
Station 1	11.8 ± 0.1	9.2 ± 1.4	21 ± 10	24 ± 13	23 ± 24	20	9.2
Station 5	10.1 ± 0.0	7.9 ± 2.1	20 ± 19	11 ± 10	86 ± 9	160	−1
Station 13	9.6 ± 0.9	13.2 ± 2.3	0 ± 0	0 ± 0	61 ± 22	39	5.4

Results include δ¹⁵N_NO3+NO2 end-member (±1 SD), measured δ¹⁵N_PNsink (±1 SD), the calculated fractional contribution of N₂ fixation to export production (f_nfix; %) (±1 SD), and N₂-fixation rate (R_nfix δ¹⁵N budget; µmol N m⁻² d⁻¹) (±1 SD), as well as N₂-fixation rates from short-term ${}^{15}\text{N}_2$ incubations (R_nfix short-term incub; μmol N m⁻² d⁻¹) (±1 SD). The last two columns report the fraction of the PN_sink flux that would be supported by N₂ fixation as well as the δ¹⁵N_PNsink that would be expected assuming the short-term N₂-fixation rate; see Discussion for details.

Fig. 1.

Biogeochemical characterization of the ETSP upper thermocline. Water column profiles of potential density (dashed black line), ${\text{NO}}_3^{-}+{\text{NO}}_2^{-}$ concentration (open circles), δ¹⁵N_NO3+NO2 (filled circles), δ¹⁵N_PNsink with error bars (±1 SD) (filled triangles), and CTD fluorescence trace from the 2010 cruise (solid green line) or discrete chlorophyll a measurements from the 2011 cruise (filled diamonds) for 2010 Station 5 (100°W, 20°S) (A), 2010 Station 3 (90°W, 20°S) (B), 2010 Station 1 (80°W, 20°S) (C), 2011 Station 5 (100°W, 20°S) (D), 2011 Station 13 (82°W, 15°S) (E), and 2011 Station 1 (80°W, 20°S) (F).

Table S1.

Summary of station location, sampling depths, and biogeochemical parameters at each station

Depth, m	Sigma theta	[NO3−+NO2−], µM (±1 SD)	NO3−+NO2−δ15N, ‰ vs. air (±1 SD)	[O2], µmol/kg	[PNsusp], µM	PNsusp δ15N, ‰ vs. air	Trap δ15N, ‰ vs. air (±1 SD)	Length of trap deployment, h
Cruise year 2010
Station 1 (20° S, 80° W)
0	24.773	0.04 (0)		231.20	0.54	11.2
20	24.856	0.05 (0)		231.18
40	25.273	0.05 (0)		239.67	0.56	9.5
60	25.627	2.3 (0)	18.1 (0.2)	252.16
80	25.834	4.1 (0.3)	14.8 (0.03)	242.31	0.90	8.8
100	26.074	6.7 (0.02)	13.2 (0.2)	233.47			*11.7 (2.1)	35
120	26.471	12.1 (0.2)	12.3 (0.02)	219.82	0.28	8.1
150	26.871	17.5 (2.0)	11.5 (0.2)	173.97	0.24	6.5
200	27.351	22.3 (1.3)	13.8 (0.2)	51.30	0.21	10.5
250	27.758	28 (0.1)	11.8 (0.2)	16.03	0.25	5.4
300	28.121	32.8 (0.1)	9.7 (0.0)	20.31	0.33	7.3
400	28.75	41.9 (0.1)	8.1 (0.2)	22.17	0.23	10.4
Station 3 (19.59° S, 89.59° W)
0	24.753	0.06 (0.01)		225.76	0.41	8.2
20	24.885	0.04 (0.02)		224.97	0.45	7.4
40	25.026	0.04 (0.0)		225.50	0.36	13.2
60	25.36	0.02 (0.01)		226.79
75	25.494	0.03 (0.0)		238.03	0.32	8.7
100	25.7	0.02 (0.0)		235.59
120	25.824	0.2 (0.0)		230.93	0.41	7.3
148	26	2.2 (0.03)	13.3 (0.1)	223.09
200	26.687	9.4 (0.1)	11.4 (0.05)	203.72	0.26	7.3	*13.7	37
250	27.403	19.3 (0.4)	11.1 (0.01)	127.64	0.30	4.3
300	27.922	23.5 (1.3)	12.3 (0.0)	32.95	0.15	9.9
400	28.661	35.1 (1.0)	9 (0.0)	36.47	0.08	8.5
Station 5 (19.59° S, 100° W)
0	24.671	0.02 (0.0)		217.76	0.27	10.8
25	24.777	0.02 (0.0)		218.00	0.42	7.6
50	24.917	0.02 (0.0)		217.71	0.37	7.8
100	25.497	0.03 (0.01)		229.21	0.48	7.6
120	25.715	0.02 (0.01)		223.23	0.30	7.8
150	25.899	0.3 (0.05)		218.87	0.38	9.1
200	26.44	1.6 (0.03)	9.8 (0.9)	215.80	0.32	6.0	*11.1 (0.1)	65
250	26.977	6.6 (0.1)	10.3 (0.2)	203.17
300	27.639	16.4 (0.02)	10.6 (0.1)	158.73	0.20	9.8
400	28.538	30.9 (0.3)	9.3 (0.02)	47.31	0.85	6.8
Cruise year 2011
Station 1 (20° S, 80° W)
2	24.429	0.1 (0.06)		212.13	0.41	8.7
20	24.501	0.1 (0.01)		212.25	0.36	8.9
35	24.58	0 (0)		214.81
40	25.168	0.06 (0.01)		241.14	0.45	9.5
50	25.412	0.05 (0)		245.60	0.45	8.4
60	25.612	0.3 (0.07)		241.63
70	25.767	0.4 (0.01)	19.8 (0.6)	234.04
80	25.846	1.5 (0.08)	16.2 (0.4)	224.98
90	25.929	2.9 (0.04)	14.2 (0.2)	220.14
100	25.992	3.9 (0.04)	13.7 (0.1)	216.74	0.23	9.5	*9.2 (1.4)	39
110	26.06	4.4 (0)	13.5 (0.2)	213.39
120	26.149	5.1 (0.03)	12.9 (0.1)	207.86
140	26.332	8.6 (0.03)	11.8 (0.1)	196.76
160	26.498	10.4 (0.01)		187.14
180	26.855	16.6 (0.04)	11.8 (0.1)	139.05
200	27.076	21 (0.5)	12.3 (0.1)	90.23	0.13	7.9	*7.5 (3.9)	39
225	27.388	23.7 (0.4)	12.8 (0.02)	41.23
250	27.634	24.7 (0.2)	12.7 (0.1)	21.51
275	27.857	28.9 (0.5)	11.6 (0.1)	9.27
300	28.036	31 (0.2)	11.3 (0.2)	8.89	0.12	10.4
325	28.226	33.4 (0.6)	9.9 (0.2)	9.91
350	28.394	35.4 (0.1)	9.3 (0.2)	14.92
375	28.557	36.1 (0.1)	9.4 (0.2)	20.71
Station 5 (20° S, 100° W)
5	24.572	0.03 (0.04)		213.36	0.49	8.1
20	24.653	0.1 (0.06)		213.71	0.38	4.9
40	24.75	0.04 (0.05)		213.61	0.45	7.4
60	24.975	0.03 (0.04)		218.76
70	25.325			234.21
100	25.745	0.1 (0.05)		234.50	0.33	7.3	*7.9 (2.1)	70
120	25.9	0.2 (0.06)		227.47
130	25.969	0.8 (0.08)	10.6	226.48
140	26.033	1.1 (0.02)	10.1 (0.0)	220.40
150	26.088	1.6 (0.06)	10.2 (0.1)	217.85
160	26.143	1.7 (0.06)	10.3 (0.2)	214.75
170	26.194	1.8 (0.02)	10.4 (0.1)	214.46
180	26.251	2 (0.08)	10.5 (0.3)	214.13
190	26.298	2.2 (0.08)	9.5 (1.2)	213.76
200	26.368	2.6 (0.2)	10.3 (0.5)	211.22	0.22	6.3	*6.3 (1.6)	70
210	26.427	3.4 (0.03)	10.2 (0.4)	211.61
225	26.675	5.6 (0.2)	10.6 (0.5)	205.86
250	26.967	8.3 (0.04)	10.6 (0.4)	194.63
275	27.271	14.5 (0.5)	10.8 (0.2)	169.52
300	27.575	19 (0.1)	10.6 (0.8)	109.36	0.24	5.8
325	27.859	23.4 (0.3)	11.7 (0.1)	52.99
350	28.133	26.6 (0.1)	10.7 (0.1)	65.82
375	28.36	30.6 (0.2)	10 (0.5)	44.12
400	28.555	32 (0.4)		42.62	0.15	6.1
Station 13 (15° S, 82° W)
5	24.223	0.4 (0.1)	16.1 (1.0)	214.52	0.88	8.2
20	24.297	0.6 (0.0)	15.6 (0.9)	215.47	0.56	7.6
35	24.382	7.8 (0.0)	11 (0.3)	215.83	0.53	6.9
40	25.323	8.4 (0.1)	10.7 (0.4)	232.52	0.56	6.2
50	25.798	12.1	9.8 (0.5)	214.67	0.40	7.4
60	25.949	14.2 (0.2)	9.6 (0.9)	192.89
70	26.114	16.3 (0.2)	10.1 (0.2)	166.82	0.24	7.5
80	26.265	18.5 (0.1)	9.8 (0.5)	141.99	0.19	6.5
90	26.446	21 (0.3)	10.2 (0.2)	57.76
100	26.577	24.5 (0.1)	11 (0.3)	13.15			*13.2 (2.3)	38
110	26.704	19.4 (0.5)	14.8 (0.1)	1.63
120	26.808	17.5	15.9 (0.1)	0.97
140	26.999	19.4 (1.4)	13.8	0.67
160	27.134	20 (0.5)	12.7 (0.4)	0.52
180	27.26	23.4 (0.1)	11.3 (0.1)	0.47
200	27.371	25 (0.01)	11.1 (0.4)	0.44			*10.1 (1.0)	38
225	27.524	27.5 (0.1)	10.2 (0.1)	0.40
250	27.672	28.3 (0.2)	10.1 (0.2)	0.34	0.43	6.3
275	27.838	28.7	9.6 (0.2)	0.35
300	27.98	29.2 (0.4)	9.7 (0.4)	0.34
325	28.153	30.9 (0.6)	9.3 (0.4)	0.31
350	28.319	32.5 (0.9)	9.0 (0.5)	0.34
375	28.47	34.2	8.1 (1.3)	0.31
400	28.664	36.2		0.32

Average δ¹⁵N_PNsink flux values that include at least one sample that was acidified prior to isotopic analysis (to remove inorganic carbon) are indicated with an asterisk.

Because the ETSP has strong gradients in δ¹⁵N_NO3+NO2 with depth in the upper thermocline (31, 33), we use water column profiles of density, ${\text{NO}}_3^{-}+{\text{NO}}_2^{-}$ concentration, and δ¹⁵N_NO3+NO2, together with fluorescence (2010 cruise) or chlorophyll a concentration (2011 cruise), to select the δ¹⁵N_NO3+NO2 end-member at each station (Fig. 1 and Table S1). The δ¹⁵N budgets were evaluated with three choices of the δ¹⁵N_NO3+NO2 end-member: (i) the deep chlorophyll maximum, (ii) the depth of the floating sediment trap, and (iii) the shallowest δ¹⁵N_NO3+NO2 minimum (Table S2). The lower concentration and elevated δ¹⁵N_NO3+NO2 at the base of the deep chlorophyll maximum relative to depths immediately below (Fig. 1) indicates that ${\text{NO}}_3^{-}$ assimilation has affected the δ¹⁵N_NO3+NO2 at this depth and that the source of ${\text{NO}}_3^{-}+{\text{NO}}_2^{-}$ to the euphotic zone has a lower δ¹⁵N, arguing against this choice for the ${\mathrm{\delta }}^{15}{\text{N}}_{\text{NO}3+\text{NO}2}$ end-member (36). A similar argument can be made against assigning a δ¹⁵N_NO3+NO2 source at the depth of the sediment trap. We argue that the shallowest δ¹⁵N_NO3+NO2 minimum is most representative of the unaltered source of ${\text{NO}}_3^{-}+{\text{NO}}_2^{-}$ to the euphotic zone. Similar to the δ¹⁵N_Nfix end-member choice, lower δ¹⁵N_NO3+NO2 end-member values will result in a smaller role for N₂ fixation in supporting export production (Eq. 2). For comparison, the results of all three approaches are given in Table S2.

Table S2.

Evaluation of δ¹⁵N budgets with alternative δ¹⁵N_NO3+NO2 end-members

δ15N budget variables	Station
	1	3	5	13
2010 δ15NPNsink	11.7 ± 2.1‰	13.7‰	11.1 ± 0.1‰	N/A
NO3−+NO2− source from trap depth	100 m, 10% (32 ± 48)	200 m, 0% (0)	200 m, 0% (0 ± 0)	N/A
NO3−+NO2− source from base of chl max	80 m, 19% (62 ± 43)	200 m, 0% (0)	200 m, 0% (0 ± 0)	N/A
NO3−+NO2− source from depth of shallowest NO3−+NO2− from δ15N minima	150 m, 0% (0 ± 48)	250 m, 0% (0)	200 m, 0% (0 ± 0)	N/A
2011 δ15NPNsink	9.2 ± 1.4‰	N/A	7.9 ± 2.1‰	13.2 ± 2.3‰
NO3−+NO2− source from trap depth	100 m, 31% (36 ± 11)	N/A	100 m, no NO3−+NO2− at 100 m	100 m, 0% (0 ± 0)
NO3−+NO2− source from base of chl max	140 m, 21% (24 ± 13)	N/A	160 m, 21% (11 ± 10)	90 m, 0% (0 ± 0)
NO3−+NO2− source from depth of shallowest NO3−+NO2− from δ15N minima	140 m, 21% (24 ± 13)	N/A	140 m, 20% (11 ± 10)	60 m, 0% (0 ± 0)

δ¹⁵N budget estimates of the fractional contribution of N₂ fixation to export production (%) and corresponding N₂ fixation rates in units of μmol N m⁻² d⁻¹ (in parentheses) with different source δ¹⁵N_NO3+NO2 assumptions. Uncertainty represents ±1 SD of the δ¹⁵N_PNsink measurement. Station 3 had no replicate δ¹⁵N_PNsink measurements.

On the 2010 cruise, which took place during the austral summer from January to February 2010, the δ¹⁵N budgets did not detect N₂ fixation at any of the three stations (Table 1 and Fig. S1). However, at Station 1, the SD associated with the δ¹⁵N_PNsink measurement (11.7 ± 2.1‰) leads to relatively high uncertainty in the N₂-fixation rate, 0 ± 48 µmol N m⁻² d⁻¹, leaving open the possibility that N₂ fixation may have contributed up to 15% of export production (Table 1). At Station 3, no replicate measurements of the δ¹⁵N_PNsink (13.7‰) were made, so no uncertainty in the N₂-fixation rate of 0 µmol N m⁻² d⁻¹ could be estimated. At Station 5, the low SD associated with the δ¹⁵N_PNsink measurement (11.1 ± 0.1‰) corresponds to no range in either the fractional contribution of N₂ fixation to export production (0 ± 0%) or the N₂-fixation rate (0 ± 0 µmol N m⁻² d⁻¹) (Table 1).

In March to April 2011, N₂ fixation was detected in the δ¹⁵N budgets at Stations 1 and 5 (Table 1 and Fig. S1). The δ¹⁵N budget at Station 1 gave an N₂-fixation rate of 24 ± 13 µmol N m⁻² d⁻¹, supporting 21 ± 10% of export production, which appears somewhat high and inconsistent with the lack of molecular evidence for diazotrophs at this station (21). Despite the relatively high SD of duplicate δ¹⁵N_PNsink measurements (9.2 ± 1.4‰), the range of δ¹⁵N_PNsink values within the uncertainty are lower than all of the δ¹⁵N_NO3+NO2 shallower than 350 m (Fig. 1 and Table S1), indicating that N₂ fixation is the likely source of the low-δ¹⁵N N required to balance the δ¹⁵N budget at Station 1. In addition, the δ¹⁵N mass balance-based N₂-fixation rate is indistinguishable from the short-term N₂-fixation rate measured at this station in 2011 (see Discussion below) (Table 1).

The δ¹⁵N_PNsink at Station 5 on the 2011 cruise was 7.9 ± 2.1‰, lower than the δ¹⁵N_NO3+NO2 throughout the upper 400 m, which is relatively constant with depth compared with other stations (Fig. 1). The δ¹⁵N budget at Station 5 yields N₂-fixation rates of 11 ± 10 µmol N m⁻² d⁻¹, supporting 20 ± 19% of export production, although we note that the very low mass flux (0.05 mmol N m⁻² d⁻¹) (Table S3) likely contributes to the large uncertainty in δ¹⁵N budget calculations. The δ¹⁵N_PNsink at Station 13 was 13.2 ± 2.3‰, which is higher than the δ¹⁵N_NO3+NO2 at the shallowest δ¹⁵N_NO3+NO2 minimum, 9.6 ± 0.9‰, and thus implies that N₂ fixation does not support any export production at this station. However, a higher δ¹⁵N_PNsink than the δ¹⁵N_NO3+NO2 end-member indicates that another δ¹⁵N_NO3+NO2 end-member may be more appropriate. For example, up to 16 ± 13% of export production at Station 13 could be supported by N₂ fixation if 100% of the ${\text{NO}}_3^{-}+{\text{NO}}_2^{-}$-fueled export production at this station originated between 100 and 150 m, with a maximum δ¹⁵N_NO3+NO2 of 15.9‰, and where a secondary peak in chlorophyll a was observed (Fig. 1).

Table S3.

Summary of PN_sink flux measurements

Year and station	Trap depth, m	No. of traps deployed per depth (total no. of sample splits)	Length of trap deployment, h	PNsink flux, mmol N m−2 d−1 (±1 SD)	δ15NPNsink, ‰ vs. air (±1 SD)
2010
1	100	1 (2)	35	0.317	11.7 (2.1)
3	200	1 (1)	37	0.027	13.7
5	200	1 (2)	65	0.030	11.1 (0.1)
2011
1	100	2 (4)	39	0.116 (0.05)	9.2 (1.4)
13	100	2 (4)	38	0.153 (0.06)	13.2 (2.3)
5	100	2 (4)	70	0.05 (0.02)	7.9 (2.1)

N₂-Fixation Rates Derived from Productivity Metrics.

In the regions of the ETSP with ${\text{NO}}_3^{-}$-depleted surface waters, the δ¹⁵N_PNsink flux is more similar to the δ¹⁵N_NO3+NO2 in the upper thermocline than to the δ¹⁵N_Nfix end-member (Fig. 1), indicating that ${\text{NO}}_3^{-}+{\text{NO}}_2^{-}$ is the dominant source of new N fueling export production, consistent with results from other regions with ${\text{NO}}_3^{-}$-depleted surface waters (26–28). Although the composition of material captured in surface-tethered floating sediment traps is considered representative of the sinking flux, the magnitude of the export flux determined by sediment traps was lower than that recorded by other metrics of new and export production on these two cruises (37), as has been found previously (38). In cases where the δ¹⁵N budgets fail to detect N₂ fixation, multiplying by alternative productivity metrics would not yield positive N₂-fixation rates, and we expect that the results of the δ¹⁵N budgets that fail to detect N₂ fixation at four of the six stations are robust (see Discussion below). However, when N₂ fixation is detected in the δ¹⁵N budgets, multiplying the fraction of export production supported by N₂ fixation (x in Eq. 1) by an underestimate of the export flux may consequently underestimate N₂-fixation rates.

To address potential underestimation of N₂-fixation rates attributable to undercollection of the export flux by the sediment traps, N₂-fixation rates were also calculated using alternative metrics of new production (O₂/Ar ratios, ¹⁴C uptake) and export production (²³⁴Th deficits) (Table S4). These alternative productivity metrics yielded higher rates of N₂ fixation, with the highest N₂-fixation rates at Station 1 resulting from multiplying the fractional importance of N₂ fixation by the O₂/Ar-based net community production (NCP) estimate (197 ± 59 µmol N m⁻² d⁻¹) and at Station 5 by the ²³⁴Th-deficit estimate of export production (100 ± 30 µmol N m⁻² d⁻¹). Multiplying the fractional contribution of N₂ fixation to export production by these alternate productivity metrics provides a geochemically derived upper bound to regional N₂-fixation rates, which, although lower than rates from the tropical North Atlantic (4), are comparable to the intermediate rates observed in the summer in the North Pacific gyre (39). Although this study did not extend to the regions of the Central Pacific, where N₂ fixation was diagnosed by remote sensing studies (10), the prediction for the favorable surface ocean ${\text{NO}}_3^{-}$ and ${\text{PO}}_4^{3-}$ concentrations and stoichiometry in the ETSP gyre to support the highest global N₂-fixation fluxes (11) is not supported by δ¹⁵N budget calculations.

Table S4.

Rates of N₂ fixation

Year and station	Rnfix PNsink flux	Rnfix NCP (O2/Ar)	Rnfix 234Th	Rnfix 14C uptake
2010
1	0	0	0	N/A
3	0	0	0	N/A
5	0	0	0	N/A
2011
1	24 ± 13	197 ± 59	164 ± 45	111
5	11 ± 10	36 ± 77	100 ± 30	56 ± 17
13	0	0	0	0

Values are in µmol N m⁻² d⁻¹, calculated by multiplying the fractional contribution of N₂ fixation to export production from the δ¹⁵N budgets by different metrics of new and export production, including PN_sink flux (R_nfix PN_sink flux) (±1 SD), NCP from O₂/Ar (R_nfix NCP (O₂/Ar)) (±1 SD), ²³⁴Th deficits (R_nfix ²³⁴Th) (±1 SD), and ¹⁴C uptake rates (R_fnix ¹⁴C uptake) (±1 SD).

Discussion

Despite the δ¹⁵N budgets only detecting N₂ fixation at Stations 1 and 5 on the 2011 cruise, N₂ fixation was detected by short-term rate measurements at all stations on both cruises (Table 1). In 2011, the short-term N₂-fixation rate at Station 1 (23 ± 24 µmol N m⁻² d⁻¹) is indistinguishable from the δ¹⁵N budget-based rate using sediment trap fluxes (24 ± 13 µmol N m⁻² d⁻¹), but lower than the other geochemical N₂-fixation flux estimates, which range from 111 to 197 µmol N m⁻² d⁻¹ (Table S4). At Station 5, the short-term rate (86 ± 9 µmol N m⁻² d⁻¹) is higher than all but the ²³⁴Th-based N₂-fixation rate estimate, 100 ± 30 µmol N m⁻² d⁻¹ (Table S4). To evaluate whether the short-term N₂-fixation rates are consistent with the δ¹⁵N measurements, we (i) compare short-term rates of N₂ fixation with the PN_sink mass flux (Table S3) to evaluate what fraction of the PN_sink flux the short-term rate represents (“f_nfix”) and (ii) calculate what the δ¹⁵N_PNsink would need to be for the δ¹⁵N budget to match the short-term rates (Table 1).

As described above, the δ¹⁵N budget-based N₂-fixation rate is indistinguishable from the short-term incubation results at Station 1 in 2011, resulting in a predicted δ¹⁵N_PNsink identical to what was measured (Table 1). At Station 13 in 2011, the short-term rates correspond to 39% of the export flux and would have resulted in a δ¹⁵N_PNsink of 5.4‰, significantly lower than the measured δ¹⁵N_PNsink of 13.2 ± 2.3‰ (Table 1). At the stations with lower organic matter fluxes, the short-term N₂-fixation rates were several-fold larger than the PN_sink fluxes (2010 cruise, Stations 3 and 5, and 2011 cruise Station 5), corresponding to predicted δ¹⁵N_PNsink values of −1‰ (Table 1). We note that the discrepancy at Stations 3 and 5 on the 2010 cruise is exacerbated by the collection of the PN_sink flux at 200 m; at all other stations, the PN_sink flux was collected at 100 m (Table S3). At Station 5 in 2011, we suspect that the low mass flux at this ultraoligotrophic station (16), with the high SD for δ¹⁵N_PNsink, contributes to the larger discrepancy between the short-term N₂-fixation rate and the PN_sink flux. We also note that the calculation of f_nfix with the PN_sink flux, as opposed to the other metrics of new and export production, represents an upper limit to the fractional importance of N₂ fixation for supporting export production. However, together, these calculations suggest that the short-term N₂-fixation rates should have been apparent in all of the δ¹⁵N budgets.

The larger discrepancy between the δ¹⁵N budget- and short-term incubation-based N₂-fixation rates at stations with lower PN_sink fluxes requires consideration of potential explanations. First, the N₂ fixation measured in the short-term incubations may not have contributed to sinking particles if the diazotrophs did not sink and/or were grazed and not converted to fecal pellets before the sediment traps were recovered. However, this scenario might be expected to correspond to a stronger quantitative PCR-based signal of NifH than was observed at these stations (21) (see below). Additionally, analysis of δ¹⁵N_PNsink collected at Station 5 in a deep (i.e., 3,660 m), moored sediment trap deployed from February 2010 to March 2011 shows that the δ¹⁵N_PNsink does record the input of low-δ¹⁵N material, and thus N₂ fixation’s geochemical signature is most pronounced, during the austral summer, when mass fluxes were lowest (22). Moreover, there is no evidence for preferential remineralization of low-δ¹⁵N material between the surface-tethered and 3,660-m traps (22), suggesting that whatever low-δ¹⁵N PN_sink material was produced in surface waters was not rapidly remineralized to low-δ¹⁵N ${\text{NO}}_3^{-}$ in the upper thermocline, as may occur in the subtropical North Pacific (28) and North Atlantic gyres (35), where N₂-fixation rates are higher. Finally, euphotic zone DON concentration and δ¹⁵N at these stations (∼4.5–5.0 µM and 5.0 ± 0.5‰) were consistent with previous measurements from the North Pacific (40) and did not indicate that newly fixed N accumulated in the surface DON pool.

Another potential explanation for the discrepancy between the geochemical and short-term N₂-fixation rates comes from recent work documenting the presence of ¹⁵N-labeled N-oxides and/or ammonium ${(\text{NH}}_4^{+})$ in several brands of ¹⁵N₂-labeled gas (41), which could lead to an overestimate of N₂-fixation rates using ${}^{15}\text{N}_2$ tracer techniques, especially when N₂-fixation rates are low, as in the ETSP. Short-term N₂-fixation rates in this study were determined using two (2010 cruise) or three (2011 cruise) different brands of ${}^{15}\text{N}_2$ labeled gas, including those identified as more and less contaminated (41), and showed largely comparable rates and many instances of N₂-fixation rates at or near the limit of detection, arguing against this explanation as the cause of the discrepancy. Indeed, other work argues that short-term rates determined with ${}^{15}\text{N}_2$ gas may underestimate N₂-fixation rates if ${}^{15}\text{N}_2$ gas is not properly equilibrated before incubation (42), which, if correct, would exacerbate the difference between short-term and geochemical N₂-fixation rate estimates in this study.

The δ¹⁵N-budget– and short-term incubation-based N₂-fixation rate estimates integrate over 24- to 70-h periods; in the context of the offset between the geochemical and short-term N₂-fixation rates, it is useful to consider the results from a study of NifH gene expression on these same cruises, which should reflect diazotrophic activity over comparable time scales (21). Although diazotrophs could be detected, their abundance and activity in the ETSP was low compared with studies from other oligotrophic regions (21). Moreover, the heterotrophic diazotrophs that dominated the diaozotrophic community in the ETSP gyre typically have significantly lower per-cell N₂-fixation rates than the cyanobacterial diazotrophs common where higher N₂-fixation rates are found (4, 21, 39), and these low per-cell rates could not be reconciled with previously reported short-term N₂-fixation rates in the ETSP gyre (21, 43). Thus, the molecular results from these cruises are consistent with the results of the δ¹⁵N budgets that fail to detect significant rates of N₂ fixation at four of the six stations in the ETSP. The discrepancy between the NifH results and the short-term N₂-fixation rates suggests two possibilities: that there are unrecognized diazotrophs contributing to the short-term N₂-fixation rate measurements and/or that the short-term rates are artificially high and that actual N₂-fixation rates are closer to the geochemical estimates, in particular at stations where the δ¹⁵N budgets did not detect N₂ fixation. If the short-term rates are correct then the geochemical data implies that those rates do not persist for long periods of time, because the PN_sink flux and δ¹⁵N required by the short-term rates are not consistent with the δ¹⁵N budgets (Table 1).

Finally, we expect that the higher proportion of export production supported by N₂ fixation in 2011 compared with 2010 is related to the relative timing of the two cruises, with 2011 sampling later during the austral summer (March to April 2011 vs. January to February 2010). In particular, the results from a deep-moored sediment trap deployed at Station 5 from February 2010 to March 2011 records a minimum in δ¹⁵N_PNsink between March to May 2010, when organic matter fluxes were also lowest (22), suggesting that the 2010 cruise may have preceded the peak regional significance of N₂ fixation. The deep trap results also indicate that N₂-fixation rates are unlikely to be significantly higher throughout the rest of the year, consistent with prior observations from adjacent waters sampled during the austral spring (16).

Summary

Here, we present an investigation of N₂ fixation using biological and geochemical tools to evaluate the quantitative significance of N₂ fixation in the ETSP. Although short-term N₂-fixation rates integrated over the euphotic zone averaged 60 µmol N m⁻² d⁻¹, roughly half the rates measured in the North Pacific subtropical gyre (39), geochemical N₂-fixation rates, together with a parallel analysis of NifH, suggest N₂-fixation rates are lower still, from 11 to 24 µmol N m⁻² d⁻¹, when detected at all, and play a minimal role in supporting export production. When scaled to surface waters in the South Pacific with similar nutrient characteristics (i.e., ${[\text{NO}}_3^{-}+{\text{NO}}_2^{-}]$ < 1.0 µM and [${\text{PO}}_4^{3-}$] > 0.1 µM; 24 × 10¹² m²), the δ¹⁵N budget-based N₂-fixation rate of 10 µmol N m⁻² d⁻¹ corresponds to an annual flux of ∼1.2 Tg N y⁻¹, whereas an intermediate short-term N₂-fixation rate (or δ¹⁵N budget-based rate evaluated with an alternate productivity metric) of 75 µmol N m⁻² d⁻¹ corresponds to an annual flux of 9 Tg N y⁻¹. Given that the moored trap δ¹⁵N_PNsink data indicate that N₂-fixation fluxes are unlikely to be significantly higher throughout the rest of the year (22), 9 Tg N y⁻¹ can be considered an upper bound for regional N₂-fixation fluxes. These rates are a small fraction of the 95 Tg N y⁻¹ predicted for the whole Pacific basin from geochemical modeling studies (11), and also challenge the predictions (10, 11) that the highest global N₂-fixation rates are found in the Central and Eastern tropical South Pacific. This work demonstrates that low ${\text{NO}}_3^{-}$ and relatively high ${\text{PO}}_4^{3-}$ concentrations alone are insufficient to support significant N₂-fixation fluxes, perhaps because of the scarcity of iron. The findings are also consistent with recent suggestions that higher N₂-fixation rates might be found in the surface waters of the Southwest Pacific (18, 19), which maintains similarly favorable nutrient concentrations and ratios and receives relatively higher dust fluxes (15).

Materials and Methods

Water Column Sample Collection.

Samples were collected on the research vessel (R/V) Atlantis in January to February 2010, and the R/V Melville in March to April 2011 on a zonal transect along 20° S between 80° W and 100° W, with exact station locations and sample depths, nutrient concentrations, and isotopic compositions reported in Table S1. Water column samples were collected by Niskin bottles deployed on a rosette equipped with conductivity temperature depth (CTD), as well as dissolved oxygen Sea-Bird Electronics sensors. All samples were collected into acid-washed, sample-rinsed high-density polyethylene bottles, and samples from the upper 400 m passed a 0.2-µm filter before collection. All samples were stored at −20 °C until analysis on land.

$\mathbf{N}{\mathbf{O}}_{\mathbf{3}}^{\mathbf{-}}+\mathbf{N}{\mathbf{O}}_{\mathbf{2}}^{\mathbf{-}}$ Concentration and Isotopic Composition Analysis.

The concentration of ${\text{NO}}_3^{-}+{\text{NO}}_2^{-}$ ${\left(\right[\text{NO}}_3^{-}+{\text{NO}}_2^{-}\left]\right)$ was determined using chemiluminescent analysis (44) in a configuration with a detection limit of 0.05 µM and ±0.1 µM 1 SD. The δ¹⁵N_NO3+NO2 was determined using the “denitrifier” method (45, 46) with modifications (47) on samples with ${[\text{NO}}_3^{-}+{\text{NO}}_2^{-}]$ > 0.3 µM (typically ≤0.2‰ 1 SD).

Sediment Trap Sample Collection and Analysis.

Sinking particulate material was collected using surface-tethered floating particle-interceptor traps (PIT) (48, 49) equipped with 12 polycarbonate cylinders. Floating sediment traps were deployed at 200 m at Stations 3 and 5 on the 2010 cruise; at all other stations traps were deployed at 100 m. Sediment trap samples were collected into a carbonate buffered brine solution and then split into replicate samples. In 2010, two splits were collected from each sediment trap, one of which was acidified to remove inorganic carbon. In 2011, two sediment traps were deployed at each station, each with an acidified and unacidified split (Table S3). In 2010, filtration through a 1-mm mesh removed “swimmers” from sediment trap material. On the 2011 cruise, swimmers were identified by dissecting microscopy and handpicked from sediment trap samples using sterilized micropipettors and forceps. Mass and isotopic fluxes were determined by combustion analysis of trap samples at the University of California Davis Stable Isotope Facility. The limit of detection for combustion analysis was 1.5 µmol of N. The mass-weighted average δ¹⁵N of acidified and nonacidified splits of trap material captured in the sediment traps is reported in Table 1 and Table S3.

¹⁵N₂-Incubation Rate Measurements.

Short-term N₂-fixation rate measurements incubated with ${}^{15}\text{N}_2$ gas from Sigma Aldrich (lot nos. SZ1670V and MBBB0968V) were carried out in triplicate in acid-washed, sample-rinsed, light-transparent 4-L polycarbonate bottles amended with 1.5 mL of 99 atom% ${}^{15}\text{N}_2$ and 1.5 mL of 0.5 M NaH¹³CO_3. A third batch of ¹⁵N₂ gas (Eurisotop) was used to make short-term rate measurements reported elsewhere (43). Incubations were performed with water collected from four depths within the euphotic zone under simulated in situ conditions of temperature and light and run for staggered periods (e.g., 0, 12, 24, and 48 h). Incubations were terminated by filtration of the 4-L sample onto a precombusted 25-mm Whatman glass fiber filter, which was then analyzed by mass spectrometry at the University of Southern California for δ¹⁵N.

The δ¹⁵N of suspended particulate N (PN_susp) (Table S1) averaged over the upper water column (∼150 m) was used as the time zero δ¹⁵N for determining the atom % excess for rate calculations. Trapezoidal integrations of discrete N₂-fixation rate measurements extended down to 150 m, except on the 2010 cruise at Station 3 (180 m) and Station 5 (175 m). Limits of detection were 10–14 µg of N, and reproducibility (±1 SD) was ±0.3‰ for δ¹⁵N and ±0.1–0.2 µg for mass determinations.

Productivity Measurements.

²³⁴Th-based export fluxes were reported elsewhere (37, 50), and rates of O₂/Ar-based NCP (reported in mmol C m⁻² d⁻¹) were determined from the stoichiometric equivalent of net oxygen production (NOP) [NCP = NOP/PQ, where PQ is a photosynthetic quotient (O₂/C) taken here to have a value of 1.4 (51)]. NOP fluxes were calculated assuming steady-state O₂ mass balance within the mixed layer: $\text{NOP}=\mathrm{\Delta }{\text{O}}_{2\text{Bio}}{\text{k}}_{\text{v}}{\text{O}}_{2\text{eq}}$ (52, 53), where k_v is piston velocity (rate of gas exchange, determined as described in ref. 54), O_2eq is the concentration of dissolved O₂ in equilibrium with the atmosphere at the in situ temperature and pressure, and ΔO_2Bio is the biological O₂ supersaturation, determined as $\mathrm{\Delta }{\text{O}}_{2\text{Bio}}={\text{O}}_2/{\text{Ar}}_{\text{supersat}}=\left(\frac{{\text{O}}_2/{\text{Ar}}_{\text{dis}}}{{\text{O}}_2/{\text{Ar}}_{\text{eq}}}-1\right)$ to factor out the effect of temperature/salinity variability on O₂ solubility (52). The effect of bubble injection on dissolved O₂/Ar is typically small enough to be neglected (55). Additional O₂/Ar NCP methodological details are given in SI Materials and Methods. Finally, ¹⁴C uptake and chlorophyll a concentration methods used the Bermuda-Atlantic Time-Series Station protocols (56).

SI Materials and Methods

Mixed-layer dissolved O₂/Ar ratios were measured continuously through the duration of both cruises as ratios of 32/40 masses with a quadrupole mass spectrometer (Prisma Plus 200; Pfeiffer) equipped with an equilibrator inlet (EIMS) following the methodology described in ref. 57. The science seawater intake was 4–5 m below the waterline of the R/V Atlantis (in 2010) and 3–4 m below the waterline of the R/V Melville (2011). Measurements were taken every 500 ms, and averaged over 2 min. The averaging time of the signal of gases supplied by the equilibrator is ∼15 min. At a ship speed of 5 kn, the EIMS O₂/Ar ratio represents the signal averaged over ∼2.5 km. The 32/40 ion current ratio was calibrated every 6–12 h by admitting for 10 min ambient air into the mass spectrometer through a silica-fused capillary of dimensions identical to the equilibrator inlet capillary and interfaced with the inlet through a valve (Valco, HPLC Stream selector valve) (57). The NCP on station was calculated using O₂/Ar values binned in 0.25° zonally (∼28 km at 20 °S latitude), with bins centered on the station locations.