New nutrients exert fundamental control on dissolved organic carbon accumulation in the surface Atlantic Ocean Cristina Romera-Castilloa,1,2, Robert T. Letscherb, and Dennis A. Hansella a Department of Ocean Sciences, Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, FL 33149; and bEarth System Science, University of California, Irvine, CA 92697
The inventories of carbon residing in organic matter dissolved in the ocean [dissolved organic carbon (DOC)] and in the atmosphere as CO2 are of the same order of magnitude, such that small changes in the DOC pool could have important consequences in atmospheric carbon and thus climate. DOC in the global ocean is largely formed in the sunlit euphotic zone, but identifying predictable controls on that production is an important yet unrealized goal. Here, we use a testable and causative correlation between the net production of DOC and the consumption of new nutrients in the euphotic zone of the Atlantic Ocean. We demonstrate that new nutrients introduced to the euphotic zone by upwelling in divergence zones and by winter convective overturn of the water column, and the primary production associated with those nutrients, are the ultimate driver of DOC distributions across the Atlantic basins. As new nutrient input will change with a changing climate, the role of DOC in the ocean’s biological pump should likewise be impacted. DOC
| new nutrients | net community production | Atlantic | surface ocean
M
arine dissolved organic carbon (DOC) is largely produced in the euphotic zone (0–100 m), representing the fate of up to ∼50% of oceanic primary production (1). Heterotrophic microbes consume most of that material shortly after its production (hours to days), hence its being described as highly biologically labile (2). Only an amount equal to ∼4% of global annual net primary production accumulates as DOC (∼2 Pg C y−1), with much of this material having a lifetime adequate to allow its export to subeuphotic zone depths with water column overturn, thus contributing to the biological pump (3). However, our understanding of the controls underlying the distributions and concentrations of DOC in the ocean is weak. For example, DOC observed in the surface Atlantic (Fig. 1A) shows lowest concentrations (70 μmol kg−1). What controls these distributions, and are they predictable from other variables that may provide a mechanistic rationale? Net DOC production and subsequent distributions in the open ocean have been estimated from modeled primary production (e.g., refs. 3 and 4), but a limitation to this solution is that observations demonstrating a reliable and causative correlation between the two variables over relevant temporal (seasonal through annual) and spatial (basin) scales have been assumed but not made. More complex models have added variables beyond primary production, such as grazing and detritus as well as the fractionation of dissolved organic matter (DOM) according to variable lability (5), but these again are not fully testable in nature. Here we propose to use a correlation that is testable and causative, using oceanic observations: that between accumulation (net production) of DOC and new nutrients consumed in the euphotic zone. Results and Discussion New primary production depends on an external supply of new nutrients reaching the euphotic layer. Although nitrogen fixation www.pnas.org/cgi/doi/10.1073/pnas.1605344113
and atmospheric deposition can be important sources of nutrients in specific regions, two physical processes (Fig. 2) dominate the supply of marine nutrients: wintertime vertical mixing and upwelling in the ocean divergence zones (6). In the northern North Atlantic, strong, cold winds associated with winter storms remove heat from the surface ocean, progressively increasing its density and thus deepening the mixed layer by convective overturn. Reaching mixed layer depths >200 m in some northern locations (7), water column overturn is the main process bringing nutrients to the euphotic zone there. Seasonal warming and restratification supports development of spring phytoplankton blooms (8) with an associated net production of DOC (9); each winter, the system is reset with deep vertical mixing. Upwelling, in contrast, dominates the equatorial system due to easterly trade winds combined with Ekman transport, together producing divergent, poleward flows of near-surface waters that are replaced from below with cold, nutrient-enriched waters. Upwelling is also important in the eastern boundaries of the Atlantic, particularly the Canary (northwest Africa) and Benguela (southwest Africa) upwelling systems. In the Southern Ocean, divergence and subsequent upwelling is driven by the strong, regular westerlies, dominating the system at >45°S (10). Here we associate net DOC production with nitrate deficits observed throughout the surface Atlantic; the relationship (ΔDOC/ΔNO3−) is then applied to climatological nitrate distributions (from which deficits are estimated) to predict DOC concentrations at the surface. Both ΔDOC and ΔNO 3− are calculated as the difference in DOC and NO3−concentrations, respectively, between surface and underlying source waters Significance Oceanic dissolved organic carbon (DOC) is, at 662 ± 32 petagrams of carbon (Pg C), one of Earth’s major, exchangeable carbon reservoirs. With its large inventory and biologic turnover, the pool has been invoked as a principal source/sink of atmospheric CO2 driving paleoclimate variations. Controls on past source/sink functions remain unknown, but insights can be gleaned from the modern ocean. Here, we find that new nutrients introduced to the euphotic zone exert a fundamental and predictable control on the net production of DOC, in turn playing a large part in determining the surface distribution and concentrations of the pool. Climate-driven changes in surface ocean renewal of nutrients will ultimately control the source term, dictating the inventory of carbon residing in marine DOC. Author contributions: C.R.-C. and D.A.H. designed research; C.R.-C. and D.A.H. performed data interpretation; C.R.-C., R.T.L., and D.A.H. contributed new reagents/analytic tools; C.R.-C. analyzed data; R.T.L. ran the model; and C.R.-C. and D.A.H. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. 1
Present address: Department of Marine Biology, University of Vienna, 1090 Vienna, Austria.
2
To whom correspondence should be addressed. Email: [email protected] or [email protected].
PNAS | September 20, 2016 | vol. 113 | no. 38 | 10497–10502
EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES
Edited by David M. Karl, University of Hawaii, Honolulu, HI, and approved July 11, 2016 (received for review April 1, 2016)
(Methods); the effect of lateral circulation on vertical supply is addressed below. ΔNO3− is then converted to units of carbon, thus providing an estimate of the net community production (NCP). The ratio ΔDOC/NCP, hereafter the Net DOC Production ratio (NDPr), was calculated from observations of nitrate and DOC along the lines shown in Fig. 1A. The resulting ratios (Fig. 2C) largely fell between 0.10 and 0.40, as reported previously in a variety of environments (9, 11–13). An exception was the North Atlantic subtropical gyre (NASG) where the ratios were both wide-ranging and high, reaching >0.80 (Fig. 2C). Similarly high values, up to 0.70, were observed in that system during spring blooms (9), but the ratio subsequently decreased during the course of the summer season because net DOC production halted after the spring bloom while export production as sinking particles continued, thus making NDPr in the NASG subject to the time frame considered. Also, in the Arabian Sea NDPr was found to be ∼0.8 during the NE monsoon, after new nutrients reach the euphotic zone through convective overturn of the water column (14). The highest NDPr values (Fig. 2C) were from the westernmost NASG lines (A20 and A22, Fig. 1A), occupied within the time frame of the spring bloom there (March–May 2012). To identify a single NDPr with greatest applicability to the entire Atlantic basin, a sensitivity test was conducted to determine the ratio generating the smallest error between calculated and observed values; a NDPr = 0.17 was the best fit. This value was applied to observed nitrate deficits on the lines in Fig. 1A to obtain the DOC added to the euphotic zone as a result of the consumption of new nutrients (Methods), using ΔDOC = ΔNO3- * 6.6 * 0.17 = NCP * 0.17,
[1]
where 6.6 is the molar conversion from N to C units. To directly compare ΔDOC with the observed DOC concentrations, it was necessary to sum the estimate with the average DOC concentration found in the underlying source waters [i.e., adding the background DOC concentration (DOCsource) to the predicted increase in concentration associated with NCP]; this background value varied with latitude (Table 1 in Methods):
DOCcalculated = ΔDOC + DOCsource .
[2]
DOCcalculated showed significant agreement (R2 = 0.64, P < 0.001) with observed DOC (Fig. 3) in all regions of the Atlantic with the exception of the western North Atlantic, with that discrepancy being strongest in the southern sector of the NASG (considered further below; those anomalous data are not included in the regression statistics given above). If the DOCsource is not added to ΔDOC, the regression between DOCcalculated and DOCobserved is a bit lower but still high and significant (R2 = 0.525; P < 0.0001; n = 268), indicating that the contribution of ΔDOC to Eq. 2 is 82.2% and, therefore, that new nutrients, as they control NCP, are the main driver of DOC accumulation. Given validation of our approach using the observed nutrient and DOC fields (Fig. 3), we applied the same calculations to climatological NO3− data taken from the World Ocean Atlas (see details in Methods) to develop a basin-wide view of the DOC distribution (Fig. 1B). In general, DOCcalculated (Fig. 1B, color background) matches well with the observations (Fig. 1B, dots), with an error of 8.94% (root-mean-square error, rmse). The highest DOCcalculated concentrations are in the tropics, with maxima north and south of the equatorial upwelling (∼76 μmol kg−1) and at the eastern boundary upwelling systems (Angola and Guinea Domes ∼78 μmol kg−1). In the northern North Atlantic, observed DOC values ranging from 60 to 65 μmol kg−1 are also well reproduced. In the Southern Ocean (>45°S), strong water column vertical mixing precludes a high accumulation of DOC, resulting in low concentrations (≤ 50 μmol kg−1) both in the observed and calculated fields. Applying a 17% NDPr to the total organic carbon export in the Atlantic (export is taken to be equivalent to basin-wide NCP), with estimates for total export ranging from 4.15 Pg C y−1 to 4.3 Pg C y−1 (15, 16), net DOC production of 0.70–0.75 Pg C y−1 occurs over the whole basin. Thus, the Atlantic contributes ∼36% of the global net DOC production of ∼2 Pg C y−1 (3). Some areas of the tropical Atlantic are permanently stratified (e.g., southwestern NASG), thus negating the validity of our approach in those areas. Also, the season in which the observations were made could cause a mismatch to calculated DOC, such as
Fig. 1. DOC concentrations (micromoles per kilogram) in the surface Atlantic Ocean: (A) Observed DOC concentrations; (B) Observed DOC (colored dots) underlain by DOCcalculated; (C) Observed DOC (colored dots) underlain by DOCcalculated subjected to circulation and remineralization, and with the addition of terrigenous DOC. Cruise season: A20 (April–May 2012), A22 (March–April 2012); A16N (August–October 2013); A16S (December 2013–February 2014); A10 (August–October 2011); A13.5 (March–April 2010); A05 (January–February 1998); CAIBOX (July–August 2009); OVIDE (June–July 2002); Good Hope (November–December 2004).
10498 | www.pnas.org/cgi/doi/10.1073/pnas.1605344113
Romera-Castillo et al.
EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES
Fig. 2. Distributions of (A) nitrate (micromoles per kilogram) and (B) DOC (micromoles per kilogram) along a meridional section in the Atlantic Ocean (A16, located from 63°N–60°S and 20–25°W). Two potential density contours, 25.5 and 26.5 kg m−3, are shown. Blue and red arrows schematically depict the divergent (upwelling) and convergent (downwelling) systems, respectively; the black arrow depicts wintertime convective overturn. (C) Meridional distribution of the calculated NDPr using data from the lines shown in Fig. 1A, with the color bar depicting longitude.
may have existed in the NASG above. But, the calculation results in reasonable agreement between observed and calculated variables throughout the vast majority of the ocean basin. Considering the mismatches between the calculated and observed DOC concentrations in the southwest sector of the NASG, advection is an important process not considered in our static view of the system (Fig. 1B); neither are terrestrial DOC inputs accounted for. To Romera-Castillo et al.
overcome these shortcomings, we added the climatological ΔNO3− values to a steady-state ocean circulation model including terrestrial DOC inputs and DOC remineralization (Methods). The zonal advection of DOC in the tropics is evidenced in the model result by enrichment in, for example, the Caribbean Sea (Fig. 1C; color background shows improved consistency with observations) relative to that system absent advection (Fig. 1B). Inputs PNAS | September 20, 2016 | vol. 113 | no. 38 | 10499
Table 1. Source depth and DOC source concentration added to ΔDOC to obtain DOCcalculated Latitude >23.5°N 23.5°N–38°S >38°S
Source depth
DOCsource, μmol kg−1
200 m σ26.5 100 m
52 52 45
of terrestrial DOM are evident near the outflow of the Amazon River (Fig. 1C). However, mismatches between the model (Fig. 1C) and observations (Fig. 1A) in the NASG remain, with the overall error (rmse = 8.71%) only slightly improved relative to the values in Fig. 1B. The exclusion of southwestern NASG data improved the fit neither in DOCcalculated (rmse = 8.91%) nor in DOCmodel (rmse = 8.62%). Whereas observations made as a snapshot at sea (Fig. 1A) were compared here to estimates based on nutrient climatology (Fig. 1B) and steady-state ocean circulation (Fig. 1C), we do not find that advection or terrigenous DOC explains the remarkable DOC anomaly in the western North Atlantic, with observed DOC exceeding DOCcalculated by up to 13 μmol kg−1 near, e.g., 20–27°N, 43–61°W (Figs. 1 and 3). Certainly, applying a higher NDPr ratio to the ΔNO3 in that area will give closer DOCcalculated to the observations, but as this system is more permanently stratified, that solution is likely artificial. This enhanced DOC feature is present in field observations made both in 2003 (17) and 2010 (data used in this analysis). Two alternative sources are (i) local production due to new nitrogen additions beyond local vertical mixing and (ii) allochthonous inputs. Possible sources of new nitrogen are diazotrophic N2 fixation, atmospheric deposition, and river runoff, although none of these appear to explain the anomalously high DOC. N2 fixation is higher in the North than in the South Atlantic likely because of the higher dust inputs (ref. 18 and references therein). However, DOC flux inferred from the reported dissolved organic nitrogen (DON) released by N2 fixation is far too low to contribute to the DOC excess observed [up to 0.09 nmol N L−1·h−1 DON released (19), converted to carbon at a ratio of 7:1 (0.6 nmol C L−1·h−1), and assuming recalcitrance to biological consumption, would require several years
to reach the excess 13-μmol kg−1 accumulation]. Atmospheric deposition, such as of North African dust, is another source of new nutrients in the North Atlantic (20). However, aerosol optical depth data suggest higher deposition in the eastern than in the western North Atlantic (19), where an excess of DOC is not observed. Therefore, deposition likely does not appreciably contribute to the anomalous DOC signal. Advection of DOC from the closest major rivers, the Amazon and Orinoco, does not reach as far north as the region of excess DOC, according to surface salinity distributions [e.g., World Ocean Atlas (WOA) 2013]. Finally, evaporative concentration of the DOC signal is not the cause of the elevated values, given that salinity normalization does not erase the signal. Interestingly, the anomalous DOC coincides with a maximum in DON and a light-stable isotope composition (δ15N avg. −30‰) in the particulate organic nitrogen fraction (21); the source of those signals too is unknown. Concluding Remarks New nutrients (here as NO3− consumption) fundamentally control net DOC accumulation in the surface Atlantic Ocean, so climate-driven changes in ocean dynamics will in turn affect the supply of those nutrients to the euphotic zone, thus varying DOC inventory. Intensification and spatial homogenization of coastal upwelling systems due to climate change have been predicted (22, 23), which should in turn increase the vertical nitrate flux and net DOC production. In contrast, ocean warming should also intensify thermal stratification, reducing nutrient flux by vertical mixing (24, 25) in regions away from coastal upwelling systems. It is also possible that climate change could affect the NDPr value because it likely depends on the autotrophic and heterotrophic community compositions, although outside the NASG the variability in that ratio was modest (Fig. 2C). The uncertain poise set by these counter influences will impact the future course of the oceanic DOC inventory. Methods Data. The observed data used in this work were collected from several cruises within the US Global Ocean Carbon and Repeat Hydrography: A20 (cdiac.ornl. gov/ftp/oceans/CLIVAR/A20_2012/) and A22 (cdiac.ornl.gov/ftp/oceans/ CLIVAR/A22_2012/); A05 (cdiac.ornl.gov/ftp/oceans/ar01woce/); A16N (cdiac. ornl.gov/ftp/oceans/CLIVAR/A16N_2013/); A16S (cdiac.ornl.gov/ftp/oceans/
Fig. 3. Relationship between surface DOCobserved and DOCcalculated from nitrate observations along the lines in Fig. 1A. The z axes (color bar) indicate the (A) latitude (°N) and (B) longitude (°E) of each data pair. Data from the NASG, as outliers from the correlation, are encircled. The linear regression (black line) was calculated excluding NASG data (from 20 to 50°N); Y = 13.7835 (±2.3324) + 0.7663 (±0.035) · X ; R2 = 0.638; P < 0.001; n = 268, rmse = 5.8%.
10500 | www.pnas.org/cgi/doi/10.1073/pnas.1605344113
Romera-Castillo et al.
Sampling and Analysis. Descriptions of analytical methods and sampling during US Global Ocean Carbon and Repeat Hydrography program cruises are summarized in the data reports cited above. DOC samples from the OVIDE, CAIBOX, and Good Hope sections were filtered through precombusted GF/F Whatman filters using a glass filtration system. After acidification to pH < 2, the samples were stored in heat-sealed glass ampoules and stored at 4 °C. DOC was analyzed by high-temperature combustion using a Shimadzu TOC5000 analyzer. Potassium hydrogen phthalate was used as standard. Deep seawater and low carbon references waters as provided by the Hansell CRM Program (29) were measured regularly to assess day-to-day and instrumentto-instrument variability. Nutrient analyses were performed on board by standard colorimetric methods using an Alpkem auto analyzer (30). Calculation of Net DOC Production. To determine the correlation between net DOC production (ΔDOC) and deficits in nitrate (ΔNO3−), differences in observed concentrations of those variables between the underlying source and surface waters were calculated. The selected underlying source water varied with the latitude (Table 1): depth source = σ0 26.5 isopycnal for tropical and coastal areas experiencing upwelling, representative of the water upwelled in equatorial upwelling systems (31); in the northern North Atlantic, depth source = 200 m because winter vertical mixing commonly reaches that depth (7); for the Southern Ocean, depth source = 100 m was based on the NO3− observations (Fig. 2A) and WOA (2013) climatology. Several combinations of NDPr (between 10% and –25%), DOC backgrounds and source depths were tested and the best fits determined by minimizing the model-data misfit for the Atlantic basin, using the mean-normalized rmse. Eq. 2 was also applied to the nitrate deficits from the WOA climatology, calculated using the same source depths explained above.
production, (ii ) addition of terrigenous DOC via rivers, (iii ) transport with the ocean circulation, and (iv) DOC remineralization. The governing equations for DOC are ∂DOCNR + TDOCNR = f NCP − κDOCNR , ∂t
[3]
DOC = DOCNR + DOCR + DOCT ,
[4]
where DOCNR = nonrefractory DOC, DOCR = refractory DOC, and DOCT = terrigenous DOC. The steady-state, annual ocean circulation was represented as a tracer transport matrix (T) created from output of the Parallel Ocean Program 2 ocean circulation model using CORE-II forcing (32), including parameter value changes to the isopycnal diffusivity to improve the simulation of unresolved equatorial jets. The model has a nominal 1°×1° horizontal resolution with 60 vertical levels. The source for DOCNR production was a variable fraction, f, of annual NCP. Annual NCP was estimated using the spatial map of (NO3−source-NO3−surface) from the WOA climatology (interpolated onto the circulation model grid and converted to C units with multiplication by 6.6) and multiplied by the depth of the annual average mixed layer or euphotic zone depth, whichever is shallower, diagnosed from a recent simulation of the Community Earth System Model-Biogeochemical Elemental Cycling model (4), to obtain the upper ocean depth integrated rate of NCP. DOCNR was remineralized with a uniform decay rate, κ. DOCR was estimated at 52 μmol kg−1 for latitudes north of 38 °S and 45 μmol kg−1 south of 38 °S, based on an empirical linear regression of the DOC observations for the Atlantic basin (Table 1). DOCT was estimated from year 2000 fluxes of river DOC export from the Global Nutrient Export from WaterSheds 2 model (33) and assumes a 99% nonrefractory, 1% refractory ratio with 7- and 16,000-y lifetimes, respectively (4). The linear equation 3 was solved for DOCNR by direct matrix inversion for a range of parameter values for f and κ in the range f = 0.10–0.25, κ -1 = 1, 2, 3, 4, 5 y. DOCNR was substituted in Eq. 4 and the resulting modeled DOC distribution was compared against the DOC observational dataset for the Atlantic basin. The best-fit values of f and κ were determined by minimizing the model-data misfit for the Atlantic basin, using the mean-normalized, rmse between corresponding DOC observation–DOC model estimate pairs found in the upper five levels (50 m) of the model. All of the plots were performed using Ocean Data View software (34).
Modeling the Atlantic DOC Distribution with an Ocean Circulation Model. The Atlantic DOC distribution including advection and terrigenous inputs was estimated using an offline model of DOC cycling that considers (i ) the production of nonrefractory DOC from NO3-diagnosed net community
ACKNOWLEDGMENTS. We thank the crews and officers of the research vessels carrying out the sample collection, as well as the many scientists involved in those cruises and in subsequent data processing. We also thank Dr. X. A. Álvarez-Salgado for the use of DOC data he collected during cruises supported by the Spanish government. We thank two anonymous reviewers for their valuable suggestions for improving the manuscript. Data collection on US CLIVAR sections and involvement by C.R.-C. and D.A.H. were supported by US National Science Foundation OCE1436748. R.T.L. acknowledges support from the US Department of Energy, Office of Science, Biological and Environmental Research, under SciDAC Award DE-SC0012550.
1. Nagata T (2008) Organic matter-bacteria interactions in seawater. Microbial Ecology of the Oceans, ed Kirchmann DL (John Wiley & Sons, Hoboken, NJ), pp 207–241. 2. Kirchman DL, Suzuki Y, Garside C, Ducklow HW (1991) High turnover rates of dissolved organic carbon during a spring phytoplankton bloom. Nature 352(6336): 612–614. 3. Hansell DA, Carlson CA, Repeta DJ, Reiner S (2009) Dissolved organic matter in the ocean. Oceanography (Wash DC) 22:202–211. 4. Letscher RT, Moore JK, Teng YC, Primeau F (2015) Variable C: N: P stoichiometry of dissolved organic matter cycling in the Community Earth System Model. Biogeosciences 12:209–211. 5. Anderson TR, Christian JR, Flynn KJ (2015) Modeling DOM biogeochemistry. Biogeochemistry of Marine Dissolved Organic Matter, eds Hansell DA, Carlson CA (Academic, Oxford), pp 635–667. 6. Chavez FP, Toggweiler IR (1995) Biological and chemical consequences of open ocean upwelling. Upwelling in the Ocean: Modern Processes and Ancient Records, eds Summerhayes CP, et al. (John Wiley, New York), pp 313–320. 7. Glover DM, Brewer PG (1988) Estimates of wintertime mixed layer nutrient concentrations in the North Atlantic. Deep-Sea Res. Part A 35(9):1525–1546. 8. Siegel DA, Maritorena S, Nelson NB, Hansell DA, Lorenzi-Kayser M (2002) Global distribution and dynamics of colored dissolved and detrital organic materials. J Geophys Res 107:C12. 9. Hansell DA, Carlson CA (1998) Net community production of dissolved organic carbon. Global Biogeochem Cycles 12(3):443–453. 10. Talley LD, Pickard GL, Emery WJ, Swift JH (2011) Descriptive Physical Oceanography: An Introduction (Academic, San Diego).
11. Álvarez-Salgado XA, Gago J, Miguez BM, Perez FF (2001) Net ecosystem production of dissolved organic carbon in a coastal upwelling system: The Ria de Vigo, Iberian Margin of the North Atlantic. Limnol Oceanogr 46(1):135–147. 12. Mathis JT, Hansell DA, Kadko D, Bates NR, Cooper LW (2007) Determining net dissolved organic carbon production in the hydrographically complex western Arctic Ocean. Limnol Oceanogr 52(5):1789–1799. 13. Fassbender AJ, Sabine CL, Cronin MF (2016) Net community production and calcification from seven years of NOAA Station Papa mooring measurements. Global Biogeochem Cycles 30(2):250–267. 14. Hansell DA (2013) Dissolved Organic Carbon in the Carbon Cycle of the Indian Ocean. Indian Ocean Biogeochemical Processes and Ecological Variability, eds Wiggert JD, et al. (American Geophysical Union, Washington, DC), pp 217–230. 15. Laws EA, Falkowski PG, Smith WO, Ducklow H, McCarthy JJ (2000) Temperature effects on export production in the open ocean. Global Biogeochem Cycles 14(4):1231–1246. 16. Dunne JP, Sarmiento JL, Gnanadesikan A (2007) A synthesis of global particle export from the surface ocean and cycling through the ocean interior and on the seafloor. Global Biogeochem Cycles 21:GB4006. 17. Carlson CA, et al. (2010) Dissolved organic carbon export and subsequent remineralization in the mesopelagic and bathypelagic realms of the North Atlantic basin. Deep Sea Res Part II Top Stud Oceanogr 57(16):1433–1445. 18. Benavides M, Voss M (2015) Five decades of N2 fixation research in the North Atlantic Ocean. Front Mater Sci, 10.3389/fmars.2015.00040. 19. Benavides M, et al. (2013) Longitudinal variability of size-fractionated N2 fixation and DON release rates along 24.5°N in the subtropical North Atlantic. J Geophys Res: Oceans 118(7):3406–3415.
Romera-Castillo et al.
PNAS | September 20, 2016 | vol. 113 | no. 38 | 10501
EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES
CLIVAR/A16S_2013/); A13.5 (cdiac.ornl.gov/ftp/oceans/CLIVAR/A13.5_2010. data/); and A10 (cdiac.ornl.gov/ftp/oceans/CLIVAR/A10_2011/); as well as three Spanish cruises: OVIDE (26); Good Hope (27); and CAIBOX (28). Most data were collected during the local summer–fall season. DOC and NO3− data were made on full water column conductivity-temperature-depth casts at ∼60nautical mile resolution. Sample analyses from A16, A20, A22, A13.5, and A05 were performed by the D. A. Hansell laboratory at the University of Miami. Data from section A10 were analyzed by the C. A. Carlson laboratory at the University of California, Santa Barbara; OVIDE, Good Hope, and CAIBOX data were analyzed by the X. A. Álvarez-Salgado Laboratory at the Instituto de Investigaciones Marinas-Consejo Superior de Investigaciones Científicas, Vigo, Spain. Climatological nitrate data were downloaded from WOA13 (https:// www.nodc.noaa.gov/cgi-bin/OC5/woa13/woa13oxnu.pl). Data climatologies providing the best fit between DOCcalculated and DOCobserved (lowest rmse) were WOA annual climatology for >23.5 °N and WOA summer climatology for < 23.5 °N.
20. Zamora LM, Prospero JM, Hansell DA (2011) Organic nitrogen in aerosols and precipitation at Barbados and Miami: Implications regarding sources, transport and deposition to the western subtropical North Atlantic. J Geophys Res: Atmospheres 116(D20):D20309. 21. Landolfi A, Dietze H, Volpe G (2016) Longitudinal variability of organic nutrients in the North Atlantic subtropical gyre. Deep Sea Res Part I Oceanogr Res Pap 111:50–60. 22. Sydeman WJ, et al. (2014) Climate change. Climate change and wind intensification in coastal upwelling ecosystems. Science 345(6192):77–80. 23. Wang D, Gouhier TC, Menge BA, Ganguly AR (2015) Intensification and spatial homogenization of coastal upwelling under climate change. Nature 518(7539):390–394. 24. Cermeño P, et al. (2008) The role of nutricline depth in regulating the ocean carbon cycle. Proc Natl Acad Sci USA 105(51):20344–20349. 25. Bopp L, et al. (2013) Multiple stressors of ocean ecosystems in the 21st century: Projections with CMIP5 models. Biogeosciences 10:6225–6245. 26. Álvarez-Salgado XA, et al. (2013) New insights on the mineralization of dissolved organic matter in central, intermediate, and deep water masses of the northeast North Atlantic. Limnol Oceanogr 58(2):681–696.
10502 | www.pnas.org/cgi/doi/10.1073/pnas.1605344113
27. Guerrero-Feijóo E, Nieto-Cid M, Álvarez M, Álvarez-Salgado XA (2014) Dissolved organic matter cycling in the confluence of the Atlantic and Indian oceans south of Africa. Deep Sea Res Part I Oceanogr Res Pap 83(0):12–23. 28. Lønborg C, Álvarez-Salgado XA (2014) Tracing dissolved organic matter cycling in the eastern boundary of the temperate North Atlantic using absorption and fluorescence spectroscopy. Deep Sea Res Part I Oceanogr Res Pap 85(0):35–46. 29. Hansell DA (2005) Dissolved organic carbon reference material program. Eos Trans AGU 86:318. 30. Hansen HP, Koroleff F (1999) Methods of Seawater Analysis, eds Grasshoff K, Kermling M, Ehrhardt M (Wiley-VCH, Weinheim, Germany), pp 159–226. 31. Kawase M, Sarmiento JL (1985) Nutrients in the Atlantic thermocline. J Geophys Res: Oceans 90(C5):8961–8979. 32. Bardin A, Primeau F, Lindsay K (2014) An offline implicit solver for simulating prebomb radiocarbon. Ocean Model 73:45–58. 33. Mayorga E, et al. (2010) Global Nutrient Export from WaterSheds 2 (NEWS 2): Model development and implementation. Environ Model Softw 25(7):837–853. 34. Schlitzer R (2015) Ocean Data View. Available at https://odv.awi.de. Accessed March 20, 2015.
Romera-Castillo et al.
The inventories of carbon residing in organic matter dissolved in the ocean [dissolved organic carbon (DOC)] and in the atmosphere as CO2 are of the same order of magnitude, such that small changes in the DOC pool could have important consequences in atmospheric carbon and thus climate. DOC in the global ocean is largely formed in the sunlit euphotic zone, but identifying predictable controls on that production is an important yet unrealized goal. Here, we use a testable and causative correlation between the net production of DOC and the consumption of new nutrients in the euphotic zone of the Atlantic Ocean. We demonstrate that new nutrients introduced to the euphotic zone by upwelling in divergence zones and by winter convective overturn of the water column, and the primary production associated with those nutrients, are the ultimate driver of DOC distributions across the Atlantic basins. As new nutrient input will change with a changing climate, the role of DOC in the ocean’s biological pump should likewise be impacted. DOC
| new nutrients | net community production | Atlantic | surface ocean
M
arine dissolved organic carbon (DOC) is largely produced in the euphotic zone (0–100 m), representing the fate of up to ∼50% of oceanic primary production (1). Heterotrophic microbes consume most of that material shortly after its production (hours to days), hence its being described as highly biologically labile (2). Only an amount equal to ∼4% of global annual net primary production accumulates as DOC (∼2 Pg C y−1), with much of this material having a lifetime adequate to allow its export to subeuphotic zone depths with water column overturn, thus contributing to the biological pump (3). However, our understanding of the controls underlying the distributions and concentrations of DOC in the ocean is weak. For example, DOC observed in the surface Atlantic (Fig. 1A) shows lowest concentrations (70 μmol kg−1). What controls these distributions, and are they predictable from other variables that may provide a mechanistic rationale? Net DOC production and subsequent distributions in the open ocean have been estimated from modeled primary production (e.g., refs. 3 and 4), but a limitation to this solution is that observations demonstrating a reliable and causative correlation between the two variables over relevant temporal (seasonal through annual) and spatial (basin) scales have been assumed but not made. More complex models have added variables beyond primary production, such as grazing and detritus as well as the fractionation of dissolved organic matter (DOM) according to variable lability (5), but these again are not fully testable in nature. Here we propose to use a correlation that is testable and causative, using oceanic observations: that between accumulation (net production) of DOC and new nutrients consumed in the euphotic zone. Results and Discussion New primary production depends on an external supply of new nutrients reaching the euphotic layer. Although nitrogen fixation www.pnas.org/cgi/doi/10.1073/pnas.1605344113
and atmospheric deposition can be important sources of nutrients in specific regions, two physical processes (Fig. 2) dominate the supply of marine nutrients: wintertime vertical mixing and upwelling in the ocean divergence zones (6). In the northern North Atlantic, strong, cold winds associated with winter storms remove heat from the surface ocean, progressively increasing its density and thus deepening the mixed layer by convective overturn. Reaching mixed layer depths >200 m in some northern locations (7), water column overturn is the main process bringing nutrients to the euphotic zone there. Seasonal warming and restratification supports development of spring phytoplankton blooms (8) with an associated net production of DOC (9); each winter, the system is reset with deep vertical mixing. Upwelling, in contrast, dominates the equatorial system due to easterly trade winds combined with Ekman transport, together producing divergent, poleward flows of near-surface waters that are replaced from below with cold, nutrient-enriched waters. Upwelling is also important in the eastern boundaries of the Atlantic, particularly the Canary (northwest Africa) and Benguela (southwest Africa) upwelling systems. In the Southern Ocean, divergence and subsequent upwelling is driven by the strong, regular westerlies, dominating the system at >45°S (10). Here we associate net DOC production with nitrate deficits observed throughout the surface Atlantic; the relationship (ΔDOC/ΔNO3−) is then applied to climatological nitrate distributions (from which deficits are estimated) to predict DOC concentrations at the surface. Both ΔDOC and ΔNO 3− are calculated as the difference in DOC and NO3−concentrations, respectively, between surface and underlying source waters Significance Oceanic dissolved organic carbon (DOC) is, at 662 ± 32 petagrams of carbon (Pg C), one of Earth’s major, exchangeable carbon reservoirs. With its large inventory and biologic turnover, the pool has been invoked as a principal source/sink of atmospheric CO2 driving paleoclimate variations. Controls on past source/sink functions remain unknown, but insights can be gleaned from the modern ocean. Here, we find that new nutrients introduced to the euphotic zone exert a fundamental and predictable control on the net production of DOC, in turn playing a large part in determining the surface distribution and concentrations of the pool. Climate-driven changes in surface ocean renewal of nutrients will ultimately control the source term, dictating the inventory of carbon residing in marine DOC. Author contributions: C.R.-C. and D.A.H. designed research; C.R.-C. and D.A.H. performed data interpretation; C.R.-C., R.T.L., and D.A.H. contributed new reagents/analytic tools; C.R.-C. analyzed data; R.T.L. ran the model; and C.R.-C. and D.A.H. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. 1
Present address: Department of Marine Biology, University of Vienna, 1090 Vienna, Austria.
2
To whom correspondence should be addressed. Email: [email protected] or [email protected].
PNAS | September 20, 2016 | vol. 113 | no. 38 | 10497–10502
EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES
Edited by David M. Karl, University of Hawaii, Honolulu, HI, and approved July 11, 2016 (received for review April 1, 2016)
(Methods); the effect of lateral circulation on vertical supply is addressed below. ΔNO3− is then converted to units of carbon, thus providing an estimate of the net community production (NCP). The ratio ΔDOC/NCP, hereafter the Net DOC Production ratio (NDPr), was calculated from observations of nitrate and DOC along the lines shown in Fig. 1A. The resulting ratios (Fig. 2C) largely fell between 0.10 and 0.40, as reported previously in a variety of environments (9, 11–13). An exception was the North Atlantic subtropical gyre (NASG) where the ratios were both wide-ranging and high, reaching >0.80 (Fig. 2C). Similarly high values, up to 0.70, were observed in that system during spring blooms (9), but the ratio subsequently decreased during the course of the summer season because net DOC production halted after the spring bloom while export production as sinking particles continued, thus making NDPr in the NASG subject to the time frame considered. Also, in the Arabian Sea NDPr was found to be ∼0.8 during the NE monsoon, after new nutrients reach the euphotic zone through convective overturn of the water column (14). The highest NDPr values (Fig. 2C) were from the westernmost NASG lines (A20 and A22, Fig. 1A), occupied within the time frame of the spring bloom there (March–May 2012). To identify a single NDPr with greatest applicability to the entire Atlantic basin, a sensitivity test was conducted to determine the ratio generating the smallest error between calculated and observed values; a NDPr = 0.17 was the best fit. This value was applied to observed nitrate deficits on the lines in Fig. 1A to obtain the DOC added to the euphotic zone as a result of the consumption of new nutrients (Methods), using ΔDOC = ΔNO3- * 6.6 * 0.17 = NCP * 0.17,
[1]
where 6.6 is the molar conversion from N to C units. To directly compare ΔDOC with the observed DOC concentrations, it was necessary to sum the estimate with the average DOC concentration found in the underlying source waters [i.e., adding the background DOC concentration (DOCsource) to the predicted increase in concentration associated with NCP]; this background value varied with latitude (Table 1 in Methods):
DOCcalculated = ΔDOC + DOCsource .
[2]
DOCcalculated showed significant agreement (R2 = 0.64, P < 0.001) with observed DOC (Fig. 3) in all regions of the Atlantic with the exception of the western North Atlantic, with that discrepancy being strongest in the southern sector of the NASG (considered further below; those anomalous data are not included in the regression statistics given above). If the DOCsource is not added to ΔDOC, the regression between DOCcalculated and DOCobserved is a bit lower but still high and significant (R2 = 0.525; P < 0.0001; n = 268), indicating that the contribution of ΔDOC to Eq. 2 is 82.2% and, therefore, that new nutrients, as they control NCP, are the main driver of DOC accumulation. Given validation of our approach using the observed nutrient and DOC fields (Fig. 3), we applied the same calculations to climatological NO3− data taken from the World Ocean Atlas (see details in Methods) to develop a basin-wide view of the DOC distribution (Fig. 1B). In general, DOCcalculated (Fig. 1B, color background) matches well with the observations (Fig. 1B, dots), with an error of 8.94% (root-mean-square error, rmse). The highest DOCcalculated concentrations are in the tropics, with maxima north and south of the equatorial upwelling (∼76 μmol kg−1) and at the eastern boundary upwelling systems (Angola and Guinea Domes ∼78 μmol kg−1). In the northern North Atlantic, observed DOC values ranging from 60 to 65 μmol kg−1 are also well reproduced. In the Southern Ocean (>45°S), strong water column vertical mixing precludes a high accumulation of DOC, resulting in low concentrations (≤ 50 μmol kg−1) both in the observed and calculated fields. Applying a 17% NDPr to the total organic carbon export in the Atlantic (export is taken to be equivalent to basin-wide NCP), with estimates for total export ranging from 4.15 Pg C y−1 to 4.3 Pg C y−1 (15, 16), net DOC production of 0.70–0.75 Pg C y−1 occurs over the whole basin. Thus, the Atlantic contributes ∼36% of the global net DOC production of ∼2 Pg C y−1 (3). Some areas of the tropical Atlantic are permanently stratified (e.g., southwestern NASG), thus negating the validity of our approach in those areas. Also, the season in which the observations were made could cause a mismatch to calculated DOC, such as
Fig. 1. DOC concentrations (micromoles per kilogram) in the surface Atlantic Ocean: (A) Observed DOC concentrations; (B) Observed DOC (colored dots) underlain by DOCcalculated; (C) Observed DOC (colored dots) underlain by DOCcalculated subjected to circulation and remineralization, and with the addition of terrigenous DOC. Cruise season: A20 (April–May 2012), A22 (March–April 2012); A16N (August–October 2013); A16S (December 2013–February 2014); A10 (August–October 2011); A13.5 (March–April 2010); A05 (January–February 1998); CAIBOX (July–August 2009); OVIDE (June–July 2002); Good Hope (November–December 2004).
10498 | www.pnas.org/cgi/doi/10.1073/pnas.1605344113
Romera-Castillo et al.
EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES
Fig. 2. Distributions of (A) nitrate (micromoles per kilogram) and (B) DOC (micromoles per kilogram) along a meridional section in the Atlantic Ocean (A16, located from 63°N–60°S and 20–25°W). Two potential density contours, 25.5 and 26.5 kg m−3, are shown. Blue and red arrows schematically depict the divergent (upwelling) and convergent (downwelling) systems, respectively; the black arrow depicts wintertime convective overturn. (C) Meridional distribution of the calculated NDPr using data from the lines shown in Fig. 1A, with the color bar depicting longitude.
may have existed in the NASG above. But, the calculation results in reasonable agreement between observed and calculated variables throughout the vast majority of the ocean basin. Considering the mismatches between the calculated and observed DOC concentrations in the southwest sector of the NASG, advection is an important process not considered in our static view of the system (Fig. 1B); neither are terrestrial DOC inputs accounted for. To Romera-Castillo et al.
overcome these shortcomings, we added the climatological ΔNO3− values to a steady-state ocean circulation model including terrestrial DOC inputs and DOC remineralization (Methods). The zonal advection of DOC in the tropics is evidenced in the model result by enrichment in, for example, the Caribbean Sea (Fig. 1C; color background shows improved consistency with observations) relative to that system absent advection (Fig. 1B). Inputs PNAS | September 20, 2016 | vol. 113 | no. 38 | 10499
Table 1. Source depth and DOC source concentration added to ΔDOC to obtain DOCcalculated Latitude >23.5°N 23.5°N–38°S >38°S
Source depth
DOCsource, μmol kg−1
200 m σ26.5 100 m
52 52 45
of terrestrial DOM are evident near the outflow of the Amazon River (Fig. 1C). However, mismatches between the model (Fig. 1C) and observations (Fig. 1A) in the NASG remain, with the overall error (rmse = 8.71%) only slightly improved relative to the values in Fig. 1B. The exclusion of southwestern NASG data improved the fit neither in DOCcalculated (rmse = 8.91%) nor in DOCmodel (rmse = 8.62%). Whereas observations made as a snapshot at sea (Fig. 1A) were compared here to estimates based on nutrient climatology (Fig. 1B) and steady-state ocean circulation (Fig. 1C), we do not find that advection or terrigenous DOC explains the remarkable DOC anomaly in the western North Atlantic, with observed DOC exceeding DOCcalculated by up to 13 μmol kg−1 near, e.g., 20–27°N, 43–61°W (Figs. 1 and 3). Certainly, applying a higher NDPr ratio to the ΔNO3 in that area will give closer DOCcalculated to the observations, but as this system is more permanently stratified, that solution is likely artificial. This enhanced DOC feature is present in field observations made both in 2003 (17) and 2010 (data used in this analysis). Two alternative sources are (i) local production due to new nitrogen additions beyond local vertical mixing and (ii) allochthonous inputs. Possible sources of new nitrogen are diazotrophic N2 fixation, atmospheric deposition, and river runoff, although none of these appear to explain the anomalously high DOC. N2 fixation is higher in the North than in the South Atlantic likely because of the higher dust inputs (ref. 18 and references therein). However, DOC flux inferred from the reported dissolved organic nitrogen (DON) released by N2 fixation is far too low to contribute to the DOC excess observed [up to 0.09 nmol N L−1·h−1 DON released (19), converted to carbon at a ratio of 7:1 (0.6 nmol C L−1·h−1), and assuming recalcitrance to biological consumption, would require several years
to reach the excess 13-μmol kg−1 accumulation]. Atmospheric deposition, such as of North African dust, is another source of new nutrients in the North Atlantic (20). However, aerosol optical depth data suggest higher deposition in the eastern than in the western North Atlantic (19), where an excess of DOC is not observed. Therefore, deposition likely does not appreciably contribute to the anomalous DOC signal. Advection of DOC from the closest major rivers, the Amazon and Orinoco, does not reach as far north as the region of excess DOC, according to surface salinity distributions [e.g., World Ocean Atlas (WOA) 2013]. Finally, evaporative concentration of the DOC signal is not the cause of the elevated values, given that salinity normalization does not erase the signal. Interestingly, the anomalous DOC coincides with a maximum in DON and a light-stable isotope composition (δ15N avg. −30‰) in the particulate organic nitrogen fraction (21); the source of those signals too is unknown. Concluding Remarks New nutrients (here as NO3− consumption) fundamentally control net DOC accumulation in the surface Atlantic Ocean, so climate-driven changes in ocean dynamics will in turn affect the supply of those nutrients to the euphotic zone, thus varying DOC inventory. Intensification and spatial homogenization of coastal upwelling systems due to climate change have been predicted (22, 23), which should in turn increase the vertical nitrate flux and net DOC production. In contrast, ocean warming should also intensify thermal stratification, reducing nutrient flux by vertical mixing (24, 25) in regions away from coastal upwelling systems. It is also possible that climate change could affect the NDPr value because it likely depends on the autotrophic and heterotrophic community compositions, although outside the NASG the variability in that ratio was modest (Fig. 2C). The uncertain poise set by these counter influences will impact the future course of the oceanic DOC inventory. Methods Data. The observed data used in this work were collected from several cruises within the US Global Ocean Carbon and Repeat Hydrography: A20 (cdiac.ornl. gov/ftp/oceans/CLIVAR/A20_2012/) and A22 (cdiac.ornl.gov/ftp/oceans/ CLIVAR/A22_2012/); A05 (cdiac.ornl.gov/ftp/oceans/ar01woce/); A16N (cdiac. ornl.gov/ftp/oceans/CLIVAR/A16N_2013/); A16S (cdiac.ornl.gov/ftp/oceans/
Fig. 3. Relationship between surface DOCobserved and DOCcalculated from nitrate observations along the lines in Fig. 1A. The z axes (color bar) indicate the (A) latitude (°N) and (B) longitude (°E) of each data pair. Data from the NASG, as outliers from the correlation, are encircled. The linear regression (black line) was calculated excluding NASG data (from 20 to 50°N); Y = 13.7835 (±2.3324) + 0.7663 (±0.035) · X ; R2 = 0.638; P < 0.001; n = 268, rmse = 5.8%.
10500 | www.pnas.org/cgi/doi/10.1073/pnas.1605344113
Romera-Castillo et al.
Sampling and Analysis. Descriptions of analytical methods and sampling during US Global Ocean Carbon and Repeat Hydrography program cruises are summarized in the data reports cited above. DOC samples from the OVIDE, CAIBOX, and Good Hope sections were filtered through precombusted GF/F Whatman filters using a glass filtration system. After acidification to pH < 2, the samples were stored in heat-sealed glass ampoules and stored at 4 °C. DOC was analyzed by high-temperature combustion using a Shimadzu TOC5000 analyzer. Potassium hydrogen phthalate was used as standard. Deep seawater and low carbon references waters as provided by the Hansell CRM Program (29) were measured regularly to assess day-to-day and instrumentto-instrument variability. Nutrient analyses were performed on board by standard colorimetric methods using an Alpkem auto analyzer (30). Calculation of Net DOC Production. To determine the correlation between net DOC production (ΔDOC) and deficits in nitrate (ΔNO3−), differences in observed concentrations of those variables between the underlying source and surface waters were calculated. The selected underlying source water varied with the latitude (Table 1): depth source = σ0 26.5 isopycnal for tropical and coastal areas experiencing upwelling, representative of the water upwelled in equatorial upwelling systems (31); in the northern North Atlantic, depth source = 200 m because winter vertical mixing commonly reaches that depth (7); for the Southern Ocean, depth source = 100 m was based on the NO3− observations (Fig. 2A) and WOA (2013) climatology. Several combinations of NDPr (between 10% and –25%), DOC backgrounds and source depths were tested and the best fits determined by minimizing the model-data misfit for the Atlantic basin, using the mean-normalized rmse. Eq. 2 was also applied to the nitrate deficits from the WOA climatology, calculated using the same source depths explained above.
production, (ii ) addition of terrigenous DOC via rivers, (iii ) transport with the ocean circulation, and (iv) DOC remineralization. The governing equations for DOC are ∂DOCNR + TDOCNR = f NCP − κDOCNR , ∂t
[3]
DOC = DOCNR + DOCR + DOCT ,
[4]
where DOCNR = nonrefractory DOC, DOCR = refractory DOC, and DOCT = terrigenous DOC. The steady-state, annual ocean circulation was represented as a tracer transport matrix (T) created from output of the Parallel Ocean Program 2 ocean circulation model using CORE-II forcing (32), including parameter value changes to the isopycnal diffusivity to improve the simulation of unresolved equatorial jets. The model has a nominal 1°×1° horizontal resolution with 60 vertical levels. The source for DOCNR production was a variable fraction, f, of annual NCP. Annual NCP was estimated using the spatial map of (NO3−source-NO3−surface) from the WOA climatology (interpolated onto the circulation model grid and converted to C units with multiplication by 6.6) and multiplied by the depth of the annual average mixed layer or euphotic zone depth, whichever is shallower, diagnosed from a recent simulation of the Community Earth System Model-Biogeochemical Elemental Cycling model (4), to obtain the upper ocean depth integrated rate of NCP. DOCNR was remineralized with a uniform decay rate, κ. DOCR was estimated at 52 μmol kg−1 for latitudes north of 38 °S and 45 μmol kg−1 south of 38 °S, based on an empirical linear regression of the DOC observations for the Atlantic basin (Table 1). DOCT was estimated from year 2000 fluxes of river DOC export from the Global Nutrient Export from WaterSheds 2 model (33) and assumes a 99% nonrefractory, 1% refractory ratio with 7- and 16,000-y lifetimes, respectively (4). The linear equation 3 was solved for DOCNR by direct matrix inversion for a range of parameter values for f and κ in the range f = 0.10–0.25, κ -1 = 1, 2, 3, 4, 5 y. DOCNR was substituted in Eq. 4 and the resulting modeled DOC distribution was compared against the DOC observational dataset for the Atlantic basin. The best-fit values of f and κ were determined by minimizing the model-data misfit for the Atlantic basin, using the mean-normalized, rmse between corresponding DOC observation–DOC model estimate pairs found in the upper five levels (50 m) of the model. All of the plots were performed using Ocean Data View software (34).
Modeling the Atlantic DOC Distribution with an Ocean Circulation Model. The Atlantic DOC distribution including advection and terrigenous inputs was estimated using an offline model of DOC cycling that considers (i ) the production of nonrefractory DOC from NO3-diagnosed net community
ACKNOWLEDGMENTS. We thank the crews and officers of the research vessels carrying out the sample collection, as well as the many scientists involved in those cruises and in subsequent data processing. We also thank Dr. X. A. Álvarez-Salgado for the use of DOC data he collected during cruises supported by the Spanish government. We thank two anonymous reviewers for their valuable suggestions for improving the manuscript. Data collection on US CLIVAR sections and involvement by C.R.-C. and D.A.H. were supported by US National Science Foundation OCE1436748. R.T.L. acknowledges support from the US Department of Energy, Office of Science, Biological and Environmental Research, under SciDAC Award DE-SC0012550.
1. Nagata T (2008) Organic matter-bacteria interactions in seawater. Microbial Ecology of the Oceans, ed Kirchmann DL (John Wiley & Sons, Hoboken, NJ), pp 207–241. 2. Kirchman DL, Suzuki Y, Garside C, Ducklow HW (1991) High turnover rates of dissolved organic carbon during a spring phytoplankton bloom. Nature 352(6336): 612–614. 3. Hansell DA, Carlson CA, Repeta DJ, Reiner S (2009) Dissolved organic matter in the ocean. Oceanography (Wash DC) 22:202–211. 4. Letscher RT, Moore JK, Teng YC, Primeau F (2015) Variable C: N: P stoichiometry of dissolved organic matter cycling in the Community Earth System Model. Biogeosciences 12:209–211. 5. Anderson TR, Christian JR, Flynn KJ (2015) Modeling DOM biogeochemistry. Biogeochemistry of Marine Dissolved Organic Matter, eds Hansell DA, Carlson CA (Academic, Oxford), pp 635–667. 6. Chavez FP, Toggweiler IR (1995) Biological and chemical consequences of open ocean upwelling. Upwelling in the Ocean: Modern Processes and Ancient Records, eds Summerhayes CP, et al. (John Wiley, New York), pp 313–320. 7. Glover DM, Brewer PG (1988) Estimates of wintertime mixed layer nutrient concentrations in the North Atlantic. Deep-Sea Res. Part A 35(9):1525–1546. 8. Siegel DA, Maritorena S, Nelson NB, Hansell DA, Lorenzi-Kayser M (2002) Global distribution and dynamics of colored dissolved and detrital organic materials. J Geophys Res 107:C12. 9. Hansell DA, Carlson CA (1998) Net community production of dissolved organic carbon. Global Biogeochem Cycles 12(3):443–453. 10. Talley LD, Pickard GL, Emery WJ, Swift JH (2011) Descriptive Physical Oceanography: An Introduction (Academic, San Diego).
11. Álvarez-Salgado XA, Gago J, Miguez BM, Perez FF (2001) Net ecosystem production of dissolved organic carbon in a coastal upwelling system: The Ria de Vigo, Iberian Margin of the North Atlantic. Limnol Oceanogr 46(1):135–147. 12. Mathis JT, Hansell DA, Kadko D, Bates NR, Cooper LW (2007) Determining net dissolved organic carbon production in the hydrographically complex western Arctic Ocean. Limnol Oceanogr 52(5):1789–1799. 13. Fassbender AJ, Sabine CL, Cronin MF (2016) Net community production and calcification from seven years of NOAA Station Papa mooring measurements. Global Biogeochem Cycles 30(2):250–267. 14. Hansell DA (2013) Dissolved Organic Carbon in the Carbon Cycle of the Indian Ocean. Indian Ocean Biogeochemical Processes and Ecological Variability, eds Wiggert JD, et al. (American Geophysical Union, Washington, DC), pp 217–230. 15. Laws EA, Falkowski PG, Smith WO, Ducklow H, McCarthy JJ (2000) Temperature effects on export production in the open ocean. Global Biogeochem Cycles 14(4):1231–1246. 16. Dunne JP, Sarmiento JL, Gnanadesikan A (2007) A synthesis of global particle export from the surface ocean and cycling through the ocean interior and on the seafloor. Global Biogeochem Cycles 21:GB4006. 17. Carlson CA, et al. (2010) Dissolved organic carbon export and subsequent remineralization in the mesopelagic and bathypelagic realms of the North Atlantic basin. Deep Sea Res Part II Top Stud Oceanogr 57(16):1433–1445. 18. Benavides M, Voss M (2015) Five decades of N2 fixation research in the North Atlantic Ocean. Front Mater Sci, 10.3389/fmars.2015.00040. 19. Benavides M, et al. (2013) Longitudinal variability of size-fractionated N2 fixation and DON release rates along 24.5°N in the subtropical North Atlantic. J Geophys Res: Oceans 118(7):3406–3415.
Romera-Castillo et al.
PNAS | September 20, 2016 | vol. 113 | no. 38 | 10501
EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES
CLIVAR/A16S_2013/); A13.5 (cdiac.ornl.gov/ftp/oceans/CLIVAR/A13.5_2010. data/); and A10 (cdiac.ornl.gov/ftp/oceans/CLIVAR/A10_2011/); as well as three Spanish cruises: OVIDE (26); Good Hope (27); and CAIBOX (28). Most data were collected during the local summer–fall season. DOC and NO3− data were made on full water column conductivity-temperature-depth casts at ∼60nautical mile resolution. Sample analyses from A16, A20, A22, A13.5, and A05 were performed by the D. A. Hansell laboratory at the University of Miami. Data from section A10 were analyzed by the C. A. Carlson laboratory at the University of California, Santa Barbara; OVIDE, Good Hope, and CAIBOX data were analyzed by the X. A. Álvarez-Salgado Laboratory at the Instituto de Investigaciones Marinas-Consejo Superior de Investigaciones Científicas, Vigo, Spain. Climatological nitrate data were downloaded from WOA13 (https:// www.nodc.noaa.gov/cgi-bin/OC5/woa13/woa13oxnu.pl). Data climatologies providing the best fit between DOCcalculated and DOCobserved (lowest rmse) were WOA annual climatology for >23.5 °N and WOA summer climatology for < 23.5 °N.
20. Zamora LM, Prospero JM, Hansell DA (2011) Organic nitrogen in aerosols and precipitation at Barbados and Miami: Implications regarding sources, transport and deposition to the western subtropical North Atlantic. J Geophys Res: Atmospheres 116(D20):D20309. 21. Landolfi A, Dietze H, Volpe G (2016) Longitudinal variability of organic nutrients in the North Atlantic subtropical gyre. Deep Sea Res Part I Oceanogr Res Pap 111:50–60. 22. Sydeman WJ, et al. (2014) Climate change. Climate change and wind intensification in coastal upwelling ecosystems. Science 345(6192):77–80. 23. Wang D, Gouhier TC, Menge BA, Ganguly AR (2015) Intensification and spatial homogenization of coastal upwelling under climate change. Nature 518(7539):390–394. 24. Cermeño P, et al. (2008) The role of nutricline depth in regulating the ocean carbon cycle. Proc Natl Acad Sci USA 105(51):20344–20349. 25. Bopp L, et al. (2013) Multiple stressors of ocean ecosystems in the 21st century: Projections with CMIP5 models. Biogeosciences 10:6225–6245. 26. Álvarez-Salgado XA, et al. (2013) New insights on the mineralization of dissolved organic matter in central, intermediate, and deep water masses of the northeast North Atlantic. Limnol Oceanogr 58(2):681–696.
10502 | www.pnas.org/cgi/doi/10.1073/pnas.1605344113
27. Guerrero-Feijóo E, Nieto-Cid M, Álvarez M, Álvarez-Salgado XA (2014) Dissolved organic matter cycling in the confluence of the Atlantic and Indian oceans south of Africa. Deep Sea Res Part I Oceanogr Res Pap 83(0):12–23. 28. Lønborg C, Álvarez-Salgado XA (2014) Tracing dissolved organic matter cycling in the eastern boundary of the temperate North Atlantic using absorption and fluorescence spectroscopy. Deep Sea Res Part I Oceanogr Res Pap 85(0):35–46. 29. Hansell DA (2005) Dissolved organic carbon reference material program. Eos Trans AGU 86:318. 30. Hansen HP, Koroleff F (1999) Methods of Seawater Analysis, eds Grasshoff K, Kermling M, Ehrhardt M (Wiley-VCH, Weinheim, Germany), pp 159–226. 31. Kawase M, Sarmiento JL (1985) Nutrients in the Atlantic thermocline. J Geophys Res: Oceans 90(C5):8961–8979. 32. Bardin A, Primeau F, Lindsay K (2014) An offline implicit solver for simulating prebomb radiocarbon. Ocean Model 73:45–58. 33. Mayorga E, et al. (2010) Global Nutrient Export from WaterSheds 2 (NEWS 2): Model development and implementation. Environ Model Softw 25(7):837–853. 34. Schlitzer R (2015) Ocean Data View. Available at https://odv.awi.de. Accessed March 20, 2015.
Romera-Castillo et al.
