The response of marine carbon and nutrient cycles to ocean acidification: Large uncertainties related to phytoplankton physiological assumptions
2011; Wiley; Volume: 25; Issue: 3 Linguagem: Inglês
10.1029/2010gb003929
ISSN1944-9224
AutoresAlessandro Tagliabue, Laurent Bopp, Marion Gehlen,
Tópico(s)Marine Biology and Ecology Research
ResumoGlobal Biogeochemical CyclesVolume 25, Issue 3 Free Access The response of marine carbon and nutrient cycles to ocean acidification: Large uncertainties related to phytoplankton physiological assumptions Alessandro Tagliabue, Alessandro Tagliabue [email protected] Laboratoire des Sciences du Climat et de l'Environnement, IPSL-CEA-CNRS-UVSQ Orme des Merisiers, Gif sur Yvette, FranceSearch for more papers by this authorLaurent Bopp, Laurent Bopp Laboratoire des Sciences du Climat et de l'Environnement, IPSL-CEA-CNRS-UVSQ Orme des Merisiers, Gif sur Yvette, FranceSearch for more papers by this authorMarion Gehlen, Marion Gehlen Laboratoire des Sciences du Climat et de l'Environnement, IPSL-CEA-CNRS-UVSQ Orme des Merisiers, Gif sur Yvette, FranceSearch for more papers by this author Alessandro Tagliabue, Alessandro Tagliabue [email protected] Laboratoire des Sciences du Climat et de l'Environnement, IPSL-CEA-CNRS-UVSQ Orme des Merisiers, Gif sur Yvette, FranceSearch for more papers by this authorLaurent Bopp, Laurent Bopp Laboratoire des Sciences du Climat et de l'Environnement, IPSL-CEA-CNRS-UVSQ Orme des Merisiers, Gif sur Yvette, FranceSearch for more papers by this authorMarion Gehlen, Marion Gehlen Laboratoire des Sciences du Climat et de l'Environnement, IPSL-CEA-CNRS-UVSQ Orme des Merisiers, Gif sur Yvette, FranceSearch for more papers by this author First published: 25 August 2011 https://doi.org/10.1029/2010GB003929Citations: 51AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Abstract [1] Alongside climate change, anthropogenic emissions of CO2 will cause ocean acidification (OA), which will impact upon key biogeochemical processes in the ocean such as net primary production (NPP), carbon export (CEX), N2 fixation (NFIX), denitrification (DENIT), and ocean suboxia (SOX). However, appraising the impact of OA on marine biogeochemical cycles requires ocean general circulation and biogeochemistry models (OGCBMs) that necessitate a number of assumptions regarding the response of phytoplankton physiological processes to OA. Of particular importance are changes in C:N:P stoichiometry, which cannot be accounted for in current generation OGCBMs that rely on fixed Redfield C:N:P ratios. We developed a new version of the PISCES OGCBM that resolves the cycles of C, N, and P independently to investigate the impact of assumptions that OA (1) enhances NPP, (2) enhances losses of fixed carbon in dissolved organic forms, and (3) modifies the uptake of nutrients by phytoplankton. In total, six simulations were performed over the period 1860–2100. We find that while the prescribed "CO2 sensitivity" of rate processes explains the NPP response, there are large uncertainties in the response of CEX, NFIX, DENIT, and SOX related to assumptions regarding the fate of fixed carbon and nutrient uptake. The overall responses of NPP and CEX are opposite and of similar magnitude to those predicted to occur from climate change alone, suggesting that changes in stoichiometry and NPP in response to OA (and probably also climate change) need to be evaluated in non-Redfield coupled-climate OGCBMs. Using a recent synthesis of OA experiments, it was not possible to evaluate whether one or more of our scenarios was most likely. Future coupled experimental modeling approaches are necessary to better understand the impact of OA on ocean biogeochemistry. Key Points Biogeochemical impact of acidification sensitive to physiological assumptions OA induced variable stoichiometry yields negative feedbacks on climate change Impact of variable stoichiometry must be included and tested in climate models 1. Introduction [2] The increases in atmospheric CO2 (CO2atm) concentrations recorded since the industrial revolution are predicted to continue over the coming century and will likely have profound impacts on the Earth's climate [Denman et al., 2007]. In this context, the ocean is a key regulator of CO2atm concentrations that takes up one quarter of anthropogenic CO2 emissions and air-sea CO2 exchange will respond to both a changing air-sea CO2 gradient and any associated climate change [e.g., Le Quéré et al., 2009]. Modifications to climate will also impact key biogeochemical processes in the ocean, such as net primary productivity (NPP), the export of organic carbon to the ocean interior (CEX, which is responsible for maintaining the ocean's vertical dissolved inorganic carbon (DIC) gradient) and the cycling of oxygen and nutritive elements (nitrogen, phosphorous, silicic acid and iron). Recent studies by state-of-the-art coupled-climate models suggest that climate change will reduce oceanic NPP and CEX by 7–13% and 6–20%, respectively, by 2100 [Steinacher et al., 2010] primarily due to the lesser vertical nutrient supply that results from climate-induced changes to stratification (similar to earlier studies [e.g., Bopp et al., 2001, 2005]). [3] Ocean biogeochemistry will not only be impacted by climate change over the coming century, but also by progressive ocean "acidification" (OA). As CO2atm rises, the absolute uptake of CO2 by the ocean will increase and, as a result of oceanic carbon chemistry, the pH will decline, i.e., become more acidic [Zeebe and Wolf-Gladrow, 2001]. This process is ongoing, with ocean pH already having declined by ∼0.1 pH units or a 30% increase in the concentration of H+ ions [Caldeira and Wickett, 2003]. Modeling studies suggest potential declines in mean surface pH of 0.3–0.4 pH units by 2100 [e.g., Orr et al., 2005], which have the potential to dramatically alter ocean biogeochemical cycles [e.g., Finkel et al., 2010; Hutchins et al., 2009; Riebesell et al., 2009]. Leaving aside the potential impact of OA on calcifying marine organisms [e.g., Fabry, 2008; Ridgwell et al., 2009 and references therein], a number of CO2 perturbation experiments have suggested CO2atm impacts on the biology of noncalcifiers. For example, the consumption of DIC during photosynthesis has been shown to increase with rising CO2 levels [e.g., Raven and Johnston, 1991; Riebesell et al., 1993; Hein and Sand-Jensen, 1997; Gordillo et al., 2003; Hutchins et al., 2007; Riebesell et al., 2007], as has the fixation of N2 by diazotrophs [e.g., Hutchins et al., 2007; Barcelos e Ramos et al., 2007; Levitan et al., 2007], although nutrient limitation effects can also be important [e.g., Gordillo et al., 2003; Fu et al., 2008]. One recent laboratory study suggests OA might reduce Fe bioavailability [Shi et al., 2010], while a mesocosm experiment showed the opposite response [Breitbarth et al., 2010]. There is little evidence of impacts on P cycling in the very few studies that have been performed thus far [e.g., Fu et al., 2007; Tanaka et al., 2008], with similarly inconclusive results for silicic acid (Si) [Milligan et al., 2004; Bellerby et al., 2008]. Finally, changes to the composition of the phytoplankton in response to OA (as well as climate change) could also impact the cycling of "taxon-specific" nutrients, such as Si for diatoms [e.g., Tortell et al., 2002, 2008]. [4] Importantly, there have also been a number of experimental studies that have suggested modifications to the stoichiometry of C to other nutrients (generally N and P) in response to OA. Classically, the cycling of C, N and P has been conceptually coupled by the "Redfield" ratio [Redfield, 1934] and most global models of ocean biogeochemistry consider that (aside from N2 fixers in some cases) the relative contribution of C, N and P to organic matter remains constant [e.g., Aumont and Bopp, 2006; Moore and Doney, 2007]. In a mesocosm experiment, Riebesell et al. [2007] found significant increases in the C/N uptake ratio due to additional C fixation in response to OA that could act as a negative feedback on climate change. Other studies of natural populations, as well as culture experiments suggest modifications to the C/N and N/P ratio of organic matter in response to OA are likely, but that their nature is complex [Hutchins et al., 2009; Finkel et al., 2010]. Moreover, floristic shifts in phytoplankton with species-specific C/N/P ratios [e.g., Arrigo et al., 1999; Mills and Arrigo, 2010] in response to OA might also be important [e.g., Feng et al., 2010]. [5] Assessing the impact of OA on ocean biogeochemistry necessitates the use of global ocean general circulation and biogeochemical models (OGCBMs) that resolve the relevant biogeochemical cycles and processes in a prognostic fashion. Since all coupled-climate OGCBMs assume "Redfield" stoichiometry for organic matter, climate or CO2atm induced variability in C/N/P ratios are not considered when estimates are made of NPP and CEX over the coming century [Steinacher et al., 2010]. An early study by Schneider et al. [2004] suggested that CO2atm related changes to the C/N ratio could act as a weak negative feedback on CO2atm (∼70 PgC), but biological production and remineralization were not treated explicitly. More recently, a climate coupled OGCBM study suggests that CO2-induced increases in the C/N ratio are indeed a weak feedback on CO2atm (relative to the size of anthropogenic emissions), but large impacts on CEX and the cycling of O2 and N were found [Oschlies et al., 2008]. Cumulative CEX increased by over 100 Pg C and there was a 50% increase in the volume of suboxic water by 2100 [Oschlies et al., 2008]. However, estimating the impact of OA-induced changes in rates or stoichiometry in OGCBMs is not straightforward and requires various assumptions to be made. Most clear-cut is that OA enhances CO2 fixation, which raises C/N uptake ratios and that these changes are preserved in sinking organic matter [Riebesell et al., 2007; Oschlies et al., 2008]. On the other hand, it is plausible that the additional fixed DIC is lost as dissolved organic carbon (DOC) with no change in particulate organic matter ratios, with increased production of so-called transparent exopolymer particles (TEPs) [Engel, 2002] able to enhance CEX by aggregation and coagulation of DOC [Riebesell et al., 2007; Arrigo, 2007]. In this case, spatial variability in bacterial limitation will be important in governing the turnover of any additional DOC [Thingstad et al., 2008] and the impact on CEX. Additionally, if additional DOC is subducted away from the sea surface, then the depth of remineralization to DIC will dictate its influence on air-sea CO2 exchange. Regardless of the fate of the extra DIC fixation, assumptions made about the transport into the phytoplankton cell (termed "uptake" hereafter) of other nutrients (P, Si and Fe) might also be important. If uptake of P, Si and Fe were unchanged, then C/X ratios (where X is either P, Si, or Fe) of organic matter would rise alongside C/N ratios, with N/X ratios unchanged. Conversely, if uptake of P, Si and Fe were upregulated in response to OA, C/X ratios would remain unchanged in organic matter and decreasing N/X ratios would accompany the change in C/N ratios. Finally, modifications to N/P ratios could result from changes to the uptake of N or P. Assessing these factors requires multinutrient OGCBMs that are able to decouple the cycles of N and P from C in dissolved/particulate organic and inorganic pools, with most global OGCBMs unable to respond to these questions [e.g., Anderson and Mitra, 2010]. [6] In this study, we modified the PISCES OGCBM to treat the cycling of N and P independent of C in all organic and inorganic pools and examined the impact of OA-induced changes to C fixation, the loss of fixed carbon as DOC, as well as modifications to the uptake of other nutritive elements between 1860 and 2100. Specifically, we conducted four model experiments: (1) the C fixation rate and thus C/N ratio of organic matter increases with CO2atm via enhanced C fixation, but the uptake of P, Si and Fe was unchanged (raising C/X and C/N, but leaving N/X unchanged, ExpPP); (2) the increase in C fixation with rising CO2atm is offset by increased exudation of DOC (ExpPP-DOC), resulting in no change to stoichiometry of sinking organic matter; (3) the C fixation rate and thus the C/N ratio of organic matter increases with CO2atm, as well as uptake of P, Si and Fe (raising C/N, reducing N/X, but leaving C/X unchanged, ExpPP-PSiFe); and (4) the C/N ratio of organic matter increases with CO2atm via reduced N uptake, with no direct change to the uptake of C, P, Si, or Fe (raising C/N, reducing N/X, but leaving C/X unchanged, ExpPP-N). The model experiments are summarized in Figure 1 and Table 1 and two control experiments were also performed. With these experiments we can assess the impact of assuming (1) DOC losses also increase, (2) concomitant decreases to N/P ratios, and (3) the processes behind N/P ratio decreases (N or P uptake) on ocean biogeochemistry. Figure 1Open in figure viewerPowerPoint A schematic representation of the simulations conducted in this study, with a focus on the changes to the transport of DIC, DIN, P, Fe, Si, the production of DOC, and the associated changes to C/N, C/P, and N/P ratios. In the bottom right panel we present the evolution of atmospheric CO2 and the "CO2 sensitivity" (that impacts the rate processes illuminated by red arrows in the other panels) between 1860 and 2100. Table 1. A Summary of the Model Experiments That Describe if There is a Change in Atmospheric CO2, Which Element's Biotic Uptake is Impacted, if the C/N, C/X, and N/P Ratios of Organic Matter Change (Relative to the Control), and Whether Increased C Uptake is Offset by Increased DOC Productiona Exp CO2 Impact C/N C/X N/P DOC CTL No None No No No No CO2-CTL Yes None No No No No ExpPP Yes C + + No No ExpPP-DOC Yes C No No No Yes ExpPP-PSiFe Yes C, X + No − No ExpPP-N Yes N + No − No a A "+" signifies that the ratio increased, while a "−" denotes a decrease in the given ratio. The letter X denotes P, Fe and Si. In all cases the rate process increases (or decreases for N uptake) at a CO2-sensitive rate (0.002714 ppm CO2atm−1) taken from the experiments of Riebesell et al. [2007]. 2. Methods PISCES-NP [7] The PISCES model simulates nanophytoplankton, diatoms, two size classes of organic matter that sink and are remineralized differentially, two zooplankton grazers, DIC, DOC, oxygen (O2), nitrate (NO3), ammonium (NH4), phosphate (PO4), Si, Fe, and resolves the ocean carbon system chemistry (see Aumont and Bopp [2006] for a complete description). Like its contemporary OGCBMs, the standard version of PISCES rests on "Redfield" nutrient stoichiometry and "Monod" limitation of phytoplankton growth rates [Aumont and Bopp, 2006]. Therefore N and P concentrations in organic matter (phytoplankton, zooplankton, sinking particles, etc.) are not prognostically computed and transported by the model, but are assumed to exist in a constant ratio to C (the C/N/P ratio is 122/16/1 [Takahashi et al., 1985]). While this is advantageous in some respects, it clearly precludes a full assessment of the potential impact of CO2-induced changes in organic matter stoichiometry, particularly in terms of the unconstrained impacts on other nutrients. Nevertheless, the standard PISCES model does include substantial variability in C/Fe and C/Si ratios as a function of the phytoplankton functional group (diatoms or nanophytoplankton) and environmental conditions (e.g., nutrient concentrations or light levels) and already treated Fe and Si independently [Aumont and Bopp, 2006]. For this study, we revised the PISCES model to treat the uptake, assimilation, sinking and remineralization of N and P independent of C, which permits the model to prognostically compute the N and P content in all C pools. In total 14 additional tracers were added, including dissolved organic N and P (DON and DOP), which increases the total number of variables from 24 (standard PISCES) to 38 (PISCES-NP) to be transported by the physical model. [8] The nutrient limitation philosophy of PISCES-NP is unchanged from that described by Aumont and Bopp [2006]. It assumes Monod limitation of growth, where a temperature-dependent maximum growth rate is scaled by a "light limitation" term and then by the most limiting nutrient (N, P, and Fe, as well as Si for diatoms). Half-saturation constants are fixed for NO3, NH4, PO4 and DOP (which is assumed to be available to phytoplankton). On the other hand, the half-saturation constant for Fe increases as a function of the total carbon biomass for each group. This is an attempt to account for the noted effect of cell size on phytoplankton affinity for Fe. Smaller cells are assumed to have lower half-saturation constants for Fe and vice versa [Blain et al., 2004; Timmermans et al., 2004]. In PISCES the assumption is made that greater biomass consists of larger cells, which would then have greater half-saturation constants for Fe (i.e., more Fe limited for a given Fe concentration). The half-saturation constant for Si limitation increases as a function of Si concentrations. Light limitation increases as Chl/C ratios decline and as nutrient limitation increases. The light limitation term is modified by Fe stress, and light also modifies the Fe demand. Moreover, the Si demand (or the degree of silicification) is modulated by the degree of Fe stress such that diatoms are more silicified (have a greater Si/C ratio) when Fe limited. However, there are no interactions between N and P limitation. These aspects of PISCES are retained for PISCES-NP (see Aumont and Bopp [2006] for full equations). [9] The parameterization of N2 fixation has been modified from the original version of PISCES to treat N fixers as an implicit functional group. Our philosophy was to consider N2 fixation as a processes by which algae accumulate PON without the uptake of DIN, alongside P and Fe uptake and C fixation that occurs under certain conditions. As such, N2 fixation increases with light, temperature and decreases as NO3 increases and with increasing limitation by either P or Fe. After C, P and Fe are taken up from the dissolved pool during N2 fixation, they are assumed to follow the fate of C, N, P and Fe in the nanophytoplankton (grazing, mortality, sinking, remineralization etc.). Specifically, we take the temperature-dependent growth rate for the nanophytoplankton group, and this is scaled by limitation by light, temperature, P and Fe, as well as NO3 concentrations to result in the realized rate of N2 fixation (moles N l−1 s−1), with uptake of both PO4 and DOP via a N/P ratio of 50 (see Text S1 for more details). [10] DOC is included within PISCES-NP identically as for the original PISCES and is produced via the exudation of a fixed fraction of the organic carbon fixed during NPP, "sloppy feeding" by zooplankton grazers, and the disaggregation of particles. Bacteria are not treated as a functional group in PISCES, but there are several aspects to how remineralization is simulated. Variability in bacterial biomass is accounted for and parameterized implicitly using the results of a version of PISCES that did include bacteria explicitly (bacterial biomass is scaled to a function containing the abundance of microzooplankton and mesozooplankton and depth [see Aumont and Bopp, 2006]), limitation of bacterial remineralization by DOC and nutrients is included, and bacteria take up dissolved Fe. DOC is lost due to remineralization (as a function of temperature, bacterial limitation and bacterial biomass) and aggregation to POC as a function of the DOC concentration and particle load. Therefore, if extra DOC is produced (ExpPP-DOC) it has various potential fates. First, if bacteria are not nutrient limited, DOC will be remineralized back to DIC at the sea surface. Second, in regions where subduction is rapid or bacterial remineralization is limited by nutrients, DOC might be transported to the ocean interior, where the bulk of the remineralization will occur. Finally, DOC can aggregate/coagulate to form sinking particles that will also be remineralized in the ocean interior. Bacterial remineralization in the model is generally limited by nutrients in the oligotrophic gyres and by DOC elsewhere. [11] Overall, the changes to PISCES-NP, relative to PISCES [Aumont and Bopp, 2006] are the addition of 14 tracers representing N and P in all pools and a reformulation of the N2 fixation parameterization. All other aspects are unchanged. In summary, PISCES-NP can be thought of as a "variable stoichiometry-Monod nutrient limitation" OGCBM and thus an intermediate step between "fixed stoichometry-Monod nutrient limitation" OGCBMs that are commonly used for global simulations and complex "variable stoichiometry-quota nutrient limitation" models that are only used in one dimensional or regional frameworks [e.g., Litchman et al., 2006]. Experimental Design [12] Similar to Oschlies et al. [2008], the experimental results of Riebesell et al. [2007] were used to prescribe an increase in the C/N ratio as a function of CO2atm. This nondimensional multiplicative factor is called the "CO2 sensitivity," and is shown in the bottom right panel of Figure 1. In ExpPP, the C fixation rate is multiplied by this "CO2 sensitivity" factor. In ExpDOC, the C fixation rate is also multiplied by this factor, but the "extra" C is immediately lost as DOC. In ExpPP-PSiFe the C fixation rate and the uptake rates of PO4, Fe and Si are multiplied by the "CO2 sensitivity." In ExpPP-N, only the N uptake rate is divided by the CO2 sensitivity. For example, the prescribed CO2 sensitivity is 1.1 and 1.2 at 2 and 3 times present-day CO2atm, respectively. The model scenarios are summarized schematically in Figure 1. In order to isolate the impact of OA, we chose to keep a constant ocean circulation during our experiments. Between 1860 and 2100 we forced our model with CO2atm concentrations from the recommended CMIP5 experimental protocol [Riahi et al., 2007], with both the experimental CO2atm concentrations and relative "CO2 sensitivity" for the biogeochemical processes considered are presented in the lower right-hand panel of Figure 1. For reference, our experiments are summarized in Table 1 and Figure 1. We conducted two control simulations, one with CO2atm fixed at 1860 levels (CTL) and another where CO2atm evolves (CO2-CTL), but there is no OA impact on primary productivity, DOC exudation and nutrient uptake rates (i.e., a "Redfield" run). Since our ocean circulation is unchanged for the duration of our experiments, we examine results from the final year of integration. Evaluation of PISCES-NP [13] Since PISCES-NP is by definition a "Redfield" model when CO2atm does not impact rate processes, the CO2-CTL simulation is essentially very similar to the standard version of PISCES [Aumont and Bopp, 2006]. [14] By 1994, the accumulation of anthropogenic carbon in the CO2-CTL simulation (relative to the CTL simulation fixed at 1860 CO2atm levels) is 102.43 Pg, which compares well to the estimate of 119 ± 19 Pg by Sabine et al. [2004] that is referenced to the same year (1994) but also accounts for pre-1860 anthropogenic CO2 uptake. Mean annual air-sea anthropogenic CO2 fluxes (FCant) are 2.248 ± 0.121 Pg C for the 1990s, which also compare well to recent observationally based estimates [e.g., Takahashi et al., 2009; Le Quéré et al., 2009]. Modern NPP and CEX (at 100 m) are 49.1 and 7.15 Pg C a−1, respectively, which are of similar magnitude to estimates derived from satellite models or inverse modeling [e.g., Carr et al., 2006; Laws et al., 2000; Schlitzer, 2000]. In the CO2-CTL simulation there is approximately 50 Tg N a−1 of N2 fixation and 40 Tg N a−1 of denitrification between 1860 and 2100. [15] We compared modeled monthly surface NO3, PO4 and O2 from the CO2-CTL simulation to monthly values from the World Ocean Atlas 2005 climatology [Garcia et al., 2006a, 2006b]. For NO3, the regression coefficient (R) was 0.91, with a slope of 0.87 and the mean observed and modeled values were 6.842 ± 9.390 and 7.699 ± 9.313 μmol L−1, respectively. For PO4, R was 0.93, with a slope of 0.97 and the mean observed and modeled values were 0.675 ± 0.620 and 0.523 ± 0.672 μmol L−1, respectively. For O2, R was 0.98, with a slope of 1.00 and the mean observed and modeled values were 277.53 ± 65.749 and 278.98 ± 65.014 μmol L−1, respectively. Surface dissolved Fe (dFe) also compares well to the observations collected in the database of Moore and Braucher [2008], where R was 0.62, the slope was 0.745 and the mean observed and modeled dFe were 0.319 ± 3.603 and 0.277 ± 4.485 nmol L−1, respectively, between 0 and 100 m. 3. Results and Discussion Carbon Cycling [16] During ExpPP, annual NPP and CEX increase by 24 and 22% in 2100, respectively, when OA increases the rate of primary production and C/N ratios (ExpPP, Table 2), which results in an 8.7% increase in annual FCant, or a 38 Pg C increase in cumulative FCant uptake over the period 1860 to 2100. The cumulative FCant increase is similar to the additional 34 Pg found by Oschlies et al. [2008], which is not surprising since the assumptions of ExpPP most closely match those applied by their model. CEX increases by 22% (Table 2) or by over 1 gC m−2 yr−1 in some regions, with the smallest absolute changes in the oligotrophic gyres (Figure 2). Cumulative CEX increases from 1716.5 to 1796.0 Pg C, an increase of 69.5 Pg C, which is slightly less than the 100 Pg C found in simpler OGCBMs [Oschlies et al., 2008]. Greater NPP during ExpPP due to CO2 fertilization increases the concentration of phytoplankton carbon (as DOC losses were not enhanced concomitantly in this experiment), which PISCES-NP assumes results in larger cells that are consequently more Fe limited due to their reduced affinity for Fe (see Methods). Assuming slightly more Fe limited larger cells implicitly assumes that the greater NPP due to OA did not lead to more cell division. The greater Fe limitation that results causes increased surface water NO3 and, to a lesser degree, PO4. This effect only acts to reduce CEX in the extreme southwest Pacific Ocean (Figure 2a). Figure 2Open in figure viewerPowerPoint The absolute change in CEX (at 100 m, gC m−2 yr−1) in 2100 for (a) ExpPP, (b) ExpPP-DOC, (c) ExpPP-PSiFe, and (d) ExpPP-N (see Figure 1 and Table 1 for details of the experiments). Table 2. A Summary of the Experimental Resultsa Exp Year CO2 C/N C/P N/P NPP CEX FCant NF DN SOX CO2-CTL 1860 286.2 7.63 122 16 49.1 7.15 - 53.6 38.5 3.80 2000 368.9 7.63 122 16 49.1 7.15 2.34 51.7 39.3 3.80 2100 935.9 7.63 122 16 49.1 7.15 7.64 49.9 40.2 3.80 ExpPP 2000 368.9 2.9 3.3 0 3.1 2.9 3.3 −0.99 8.5 4.6 2100 935.9 22.9 22.9 0 24.0 22.1 8.7 −11.3 56.6 36.2 ExpPP-DOC 2000 368.9 0 0 0 2.7 0.6 1.3 −0.30 5.3 1.76 2100 935.9 0 0 0 20.8 3.9 2.8 −8.18 7.0 10.9 ExpPP-PSiFe 2000 368.9 2.9 0 −3.1 1.1 1.0 −0.3 −15.8 4.8 3.1 2100 935.9 22.9 0 −18.8 18.3 5.6 1.7 −89.5 20.5 15.5 ExpPP-N 2000 368.9 2.9 0 −3.1 1.4 0.1 −0.1 −16.8 3.5 0.1 2100 935.9 22.9 0 −18.8 20.2 −0.7 −1.2 −86.9 −9.3 −6.9 a We present the pCO2atm concentration (ppm) for years 1860, 2000 and 2100, results for the control simulation and then the percentage change in organic matter elemental ratios (C/N, C/P, and N/P), net primary productivity (NPP, PgC yr−1), carbon export (CEX, PgC yr−1 at 100 m), the uptake of anthropogenic CO2 (FCant, PgC yr−1), N2 fixation (NF, TgN yr−1), denitrification (DN, TgN yr−1) and the volume of suboxic waters (SOX, ×1015 m3, defined as O2 < 5 × 10−6 mol L−1) for our model experiments in 2000 and 2100. Values in bold and italics highlight increases and decreases, respectively. See Table 1 and Figure 1 for an explanation of the different model experiments. [17] The impact of OA on CEX and FCant is sensitive to modifying assumptions regarding the fate of fixed C. If we assume all additional carbon fixed is lost as DOC (ExpPP-DOC), then while the change in annual NPP is similar to ExpPP (20.8%, Table 2), the increases in annual CEX and FCant are reduced (increases of 3.9 and 2.8%, respectively, for ExpPP-DOC, Table 2). Although the additional DOC is exported to deeper waters by aggregation/coagulation, the increase in CEX would be over fivefold greater if the additional fixed C were exported as particulate organic carbon (POC, ExpPP). The greatest increases in CEX due to DOC aggregation/coagulation are found away from high-productivity regions (i.e., in and around the oligotrophic gyres) wherein bacterial DOC remineralization rates are retarded due to nutrient limitation (Figure 2). This results in only a 13.2 Pg C increase in cumulative CEX by 2100, which is 5 times less than found during ExpPP. The sensitivity of FCant to increased DOC production is slightly less and FCant increases during ExpPP-DOC by approximately one quarter of what was found for ExpPP, with cumulative CO2 uptake 14 Pg C greater by 2100 (i.e., less than one third of ExpPP). Increased surface production of DOC promotes a more recycling food web (the ratio of NPP to CEX is increased by >15%) as its remineralization produces regenerated nutrients and shifts the phytoplankton community to nanophytoplankton dominance. Indeed, remineralization of DOC at the surface increases by >50% by 2100, which produces nutrients, but also DIC that weakens air-sea CO2 gradients and reduces the change in FCant. DOC is also re
Referência(s)