State-of-the-art global nutrient deposition fields are
coupled here to the Pelagic Interactions Scheme for
Carbon and Ecosystem Studies (PISCES) biogeochemistry model to investigate their effect
on ocean biogeochemistry in the context of atmospheric forcings for
pre-industrial, present, and future periods. PISCES, as part of the European Community Earth system model (EC-Earth)
model suite, runs in offline mode using prescribed dynamical fields as
simulated by the Nucleus for European
Modelling of the Ocean (NEMO) ocean model. Present-day atmospheric deposition fluxes
of inorganic N, Fe, and P into the global ocean account for
Marine primary production is a critical component of the global carbon cycle
and important for sustaining the habitability on Earth, although it is vulnerable
to environmental changes (e.g.,
Steinacher et al., 2010). For example, an estimated decline in subarctic
productivity has been reported to accompany the warming of the last 150 years (Osman et al., 2019). Global warming
induced by greenhouse gas emissions has increased ocean stratification,
reducing the supply of nutrients from subsurface waters and inhibiting the
growth of phytoplankton in the surface ocean
(Behrenfeld et al., 2006). Thus, the role of
nutrient supply by atmospheric deposition will likely be more important in a
warmer climate. Several studies have documented the importance of primary
production on the surface ocean CO
During primary production, the growth of the phytoplankton functional types (e.g., diatoms and nanophytoplankton) results in a newly formed particulate organic matter within the euphotic zone. These processes are limited however by light, temperature, and nutrients' availability. Nutrient inputs to the euphotic upper ocean result from internal ocean dynamics, such as upwelling or external sources, i.e., input by rivers and atmospheric deposition. The effects of riverine inputs have been, however, widely investigated and are found mostly restricted along the coasts or in marginal shelf basins (e.g., Behrenfeld et al., 2006; Gröger et al., 2013; Holt et al., 2012). Hence, the atmospheric deposition is the only external supply that can reach distal open-ocean regions far remote from land.
Human activities have heavily perturbed the atmospheric chemical composition and thus the nutrient deposition inputs to the ocean (e.g., Mahowald et al., 2017), but their impact on marine biogeochemistry, and consequently on the oceanic carbon cycles and nutrient cycles, is rather complex and still not fully understood. Among other species deposited into the open ocean, nitrogen (N), iron (Fe), phosphorus (P), and silicon (Si) are the nutrients that significantly limit the marine phytoplankton growth rates and thus directly impact on ocean–atmosphere carbon fluxes, in particular where nutrients are the growth-limiting factor for phytoplankton.
Atmospheric nitrogen inputs to the global ocean are mainly derived from
anthropogenic combustion and agricultural sources over densely populated
regions (Duce et
al., 2008). So far, it is widely accepted that the marine biota primarily
utilizes the inorganic nitrogen both in its oxidized (i.e., nitrogen oxides
(NO
Present-day atmospheric nitrogen input to the oceans is estimated to be
roughly 39–68 Tg N yr
The most important atmospheric source of marine nutrients, such as Fe, P,
and Si, in the open ocean is the mineral dust deposition. Dust aerosols are
usually subject to intensive atmospheric processing during their long-range
transport over remote oceanic regions. Changes in the properties of mineral
aerosols during atmospheric transport involve chemical interactions with air
masses (i.e., aerosol aging) that lead to different coatings of dust
particles by sulfate (SO
Iron is primarily utilized by marine phytoplankton in its dissolved form,
although the actual bioavailability may substantially differ from the
soluble forms of the deposited nutrients into the ocean
(Meskhidze et al., 2019). For example,
Rubin et al. (2011) showed that in the
tropical and subtropical Atlantic Ocean, some marine organisms (e.g., the
Present-day global atmospheric DFe and dissolved P (DP) deposition fluxes
into the ocean are calculated in the range of 0.2–0.4 Tg Fe yr
The primary production is strongly linked to both nitrogen and iron
acquisitions by marine biota. However, surface oceanic nutrient
concentrations are strongly impacted by atmospheric deposition on a regional
scale. Nitrogen atmospheric inputs were shown to have a significant effect
on marine productivity, export production, and carbon uptake in
low-nutrient low-chlorophyll (LNLC) regions. For example, the impact of N
and P atmospheric deposition on strong oligotrophic regions, such as the
Eastern Mediterranean, may lead to an increase in present-day primary
production by up to 35 % (Christodoulaki et al., 2013),
resulting overall in a total phytoplanktonic biomass increase by 16 %
since the pre-industrial era
(Christodoulaki et al., 2016). The
global marine primary production rates are currently estimated on the order
of 44–67 Pg C yr
A large portion of the global ocean, especially the subtropical gyres, is depleted in nitrate and phosphate, and consequently sustains low productivity (e.g., C. M. Moore et al., 2013). About 40 % of the global ocean is estimated to be N limited (Krishnamurthy et al., 2009; Wang et al., 2015a), with most of the remaining area to be Fe limited. The relatively larger increase in N than P deposition in many oceanic regions of the globe causes shifts from N to P limitation (Krishnamurthy et al., 2009; C. M. Moore et al., 2013). On the other hand, many studies suggest that anthropogenic Fe deposition is the most important factor for carbon uptake (Krishnamurthy et al., 2010; Okin et al., 2011), mainly due to its positive effect on productivity in high-nutrient low-chlorophyll (HNLC) regions. Accordingly, the essential role of iron in oceanic productivity is currently well established (Tagliabue et al., 2017) and routinely included in marine biogeochemistry models (Aumont and Bopp, 2006; Hajima et al., 2020; Hamilton et al., 2020; Ito et al., 2019b; Moore et al., 2001; Tagliabue et al., 2014, 2016).
Another effect of iron is that it stimulates the nitrogen fixation
(Camarero and Catalan, 2012; Schulz et al.,
2012), because N
The present study aims to analyze the impact of a comprehensive representation of atmospheric inputs to the oceanic productivity. For this, a state-of-the-art marine biogeochemical model is used to integrate the recent knowledge of the atmospheric nutrient deposition fluxes into the ocean, driven by natural and combustion emissions, along with further processing during atmospheric transport. The outlined variable composition and varying sources of the deposited nutrients (i.e., N, Fe, and P) used in this work have been recently modeled with a state-of-the-art atmospheric chemistry and transport model based on pre-industrial, present, and future emissions. The description of the biogeochemical model and the parameterizations used in the atmospheric chemistry transport model, which determines the atmospheric deposition fields of this work, are presented in Sect. 2. A detailed description of the regional changes in deposition fluxes and the linked atmospheric processes controlling them is also provided. In Sect. 3, the modeled nutrient oceanic concentrations are presented, and the relevant biogeochemical processes, such as the nitrogen fixation and the primary production, are discussed and compared to estimates from observations and other modeling studies. The role of present-day air pollutants in nutrients' atmospheric deposition is also discussed, via comparison of experiments forced from atmospheric inputs of pre-industrial and projected anthropogenic and biomass burning emission scenarios. The impact of the atmospheric nutrients' organic fraction on the global oceanic productivity is assessed in Sect. 4. Moreover, the implications of our findings concerning the above biogeochemistry parameters are discussed in Sect. 5. Finally, the main conclusions are summarized in Sect. 6.
The state-of-the-art Pelagic Interactions Scheme for
Carbon and Ecosystem Studies (PISCES) biogeochemistry model
(Aumont et al., 2015),
enabled within the framework of the European Community Earth system model
(EC-Earth;
The dynamical physical outputs used to force PISCES for this study were produced by the physical ocean model NEMO, following the OMIP (Ocean Modelling Intercomparison Project; Orr et al., 2017) protocol. OMIP aims at harmonizing forcing fields of boundary conditions, as well as validation and analysis procedures among different ocean models. Atmospheric forcing fields are from the CORE II (Coordinated Ocean-ice Reference Experiments – phase II; Large and Yeager, 2009) forcing. CORE II provides a 62-year interannual forcing for the period 1948–2009. The physical model is initialized with gridded observational data from the World Ocean Atlas 2013 (Locarnini et al., 2013; Zweng et al., 2013) and then run for 310 years by repeating the 62-year CORE II forcing. The necessary physical variables to force the offline PISCES biogeochemical model (see Table S1 in the Supplement) were taken from the last 62-year iteration. However, to avoid any long-term trends from the spin-up, the multi-year (1948–2009; i.e., the fifth iteration of the 310-year run) mean of daily forcing fields was calculated. The resulting mean 1-year forcing thus contains the mean seasonal cycle and is applied (repeatedly) to drive all simulations with the biogeochemical PISCES offline model. All biogeochemical simulations are initialized and forced with the same physical fields from the average 1-year forcing derived from the OMIP run. Thus, all PISCES offline simulations are drift free in physical variables. More details of the OMIP protocol can be found in Orr et al. (2017) and a first validation of the OMIP run is provided by Skyllas et al. (2019).
For this study, PISCES uses a
In contrast to previous studies
(e.g., Aumont et al.,
2015), the new N, Fe, and P atmospheric deposition fields considered here
(see Sect. 2.2) are all calculated based on
emissions of natural and nutrient-containing combustion aerosols, detailed atmospheric gas- and aqueous-phase chemical schemes, and mineral dissolution processes due to atmospheric acidity and organic ligands
in aerosol water and cloud droplets.
Note that, as for the default PISCES configuration, the Si deposition fluxes
into the ocean are only based on the new dust deposition fields coupled in
the model. Moreover, for this work, no extra optimizations for the iron
scavenging parameters have been applied, since the default PISCES
configuration already considers a variable iron solubility on the dust
deposition inputs (Aumont
et al., 2015). The simple chemistry scheme available in PISCES is here used,
which is based on one ligand (L) of dissolved inorganic Fe and one dissolved
complexed iron (FeL). The ligand concentration in the ocean is kept
constant, equal to 0.6 nmol L
Two transient simulations from 1651 to 2100 are performed here to study the
impact of nutrients' atmospheric input on global marine productivity:
a standard (STD) simulation accounting for the inorganic fractions of the
deposited atmospheric nutrients (N, P, and Fe) into the global ocean, and a sensitivity (ORG) simulation, as for STD but also accounting for the
organic fractions of the deposited atmospheric nutrients (N, P, and Fe).
Here, we present results for the pre-industrial (PAST: 1851–1870 average),
present-day (PRESENT: 2001–2020 average), and projected future (FUTURE:
2081–2100 average) periods. For all PISCES simulations, the first 200 years
(i.e., 1651–1850) are not interpreted but considered as a spin-up to reach a
quasi-equilibrium state in the model with a well-ventilated upper ocean.
Moreover, the atmospheric CO
All atmospheric nutrient inputs coupled to PISCES are derived from the
offline global atmospheric chemistry–transport model (CTM) TM4-ECPL. The CTM
is driven by the ECMWF (European Centre for Medium-Range Weather Forecasts)
Interim reanalysis project (ERA-Interim) meteorology
(Dee et al., 2011) for the year 2010 and uses a
horizontal resolution of 3
Simulations with the atmospheric transport and chemistry model are, nonetheless, extremely expensive. Therefore, limitations in available computational resources made it necessary to reduce the CTM simulations to representative single years for (1) the pre-industrial state (before 1850), (2) the present-day state (representing the year 2010), and (3) a mid-century (2050), as well as an end-of-century (2100) state. However, as the typical residence time of tropospheric aerosols is on the order of days, the atmospheric depositional fields used in PISCES represent a well-equilibrated atmospheric chemistry and deposition flux, without the need of time-transient simulations.
For the ocean biogeochemistry model spin-up (i.e., from 1651 to 1850), the
pre-industrial field (the year 1850) was applied. After the 200-year spin-up
period, the atmospheric deposition input data for the STD and ORG
simulations were linearly interpolated from pre-industrial to present-day
conditions (i.e., the year 2010) to smoothly capture the transition from
past to the modern conditions (e.g., Krishnamurthy et al., 2009).
Respectively, the deposition data from the present day were linearly
interpolated to the projected estimates (i.e., the years 2050 and 2100).
Note that for all temporal and spatial interpolations of this work (as well
as for the drift corrections), the Climate Data Operators (CDO v1.9.8)
software, as provided by the Max Planck Institute for Meteorology, is used here
(
For the calculation of the atmospheric nitrogen deposition fluxes, the CTM
uses primary emissions of NO
Atmospheric deposition fluxes into the ocean (kg m
Nutrients' (N, Fe, P) atmospheric inputs (Tg yr
Figure 1b presents the annual mean spatial distribution of dissolved
nitrogen deposition as considered in PISCES for the STD simulation. The
present-day DIN (oxidized and reduced dissolved inorganic nitrogen)
deposition fluxes into the global ocean are estimated to be
Compared to the present day, almost all ocean basins (except some parts of
the South Indian and the South Pacific oceans) display a substantially lower
(
The global atmospheric Fe deposition fluxes are parameterized in the CTM considering primary Fe emissions associated with minerals in dust and combustion processes. The Fe content of dust minerals is based on detailed mineralogy maps (Nickovic et al., 2012), accounting also for an initial soluble Fe content in the mineral emissions (Ito and Xu, 2014), overall resulting in a mean Fe content of about 3.2 % in dust emissions. For the Fe-containing combustion aerosols, the CTM accounts for emissions from biomass burning, coal, and oil combustion (Ito, 2013; Luo et al., 2008) with dissolved Fe content of 12 %, 8 %, and 81 %, respectively. The CTM further accounts for acid and organic ligand solubilization of dust aerosols, in both aerosol water and cloud droplets, as well as for the aging (i.e., the conversion of insoluble to soluble) of the Fe-containing combustion aerosols via atmospheric processing. More details on the atmospheric Fe-cycle setup can be found in Myriokefalitakis et al. (2015, 2018), along with updates from Kanakidou et al. (2020).
Figure 1e presents the annual mean spatial distribution of dissolved iron
atmospheric deposition fluxes, as considered in PISCES for the STD
simulation. DFe deposition fluxes into the ocean present strong spatial
variability (Fig. 1e) with an annual flux of
The CTM calculates increases in DFe deposition rates since the pre-industrial
era, as stronger Fe combustion emissions and more efficient dust
dissolution rates due to a more acidic environment occur in the modern era;
overall, this accounts for about 1.5 times higher DFe atmospheric input to the
global ocean for the present day (Table 1). On the other hand, the derived DFe
in the ocean for the future atmosphere is calculated to be
The atmospheric P cycle is calculated based on emissions of insoluble mineral P, phosphate, and insoluble and soluble organic P (OP), with the resulted DP deposition fluxes in the CTM being driven by natural (i.e., dust, bioaerosols, sea spray, and volcanic aerosols) and combustion P-containing aerosol emissions. Acid solubilization of dust particles (i.e., conversion from mineral P to phosphate) takes place in both aerosol water and cloud droplets, along with the aging of OP-containing aerosols during atmospheric processing. The CTM accounts for two P-containing insoluble minerals (the fluorapatite and the hydroxyapatite) in dust based on soil mineralogy maps (Nickovic et al., 2012), as well as for OP present in the soil's organic matter (Kanakidou et al., 2012). A solubility of 10 % is applied to all P-containing dust emissions in the CTM. For P-containing combustion aerosols, the CTM accounts for anthropogenic (i.e., for fossil fuel, coal, waste, and biofuel) and biomass burning emissions, based on observed P : BC mass ratios (Mahowald et al., 2008). Sea spray and volcanic aerosols account for a rather low DIP global source, in contrast to bioaerosols which are estimated to contribute significantly to DOP. The naturally emitted OP by bioaerosols is overall represented by bacteria, fungi, and pollen (Myriokefalitakis et al., 2017). More details on the P-cycle representation in the CTM can be found in Myriokefalitakis et al. (2016).
The present-day global annual DIP deposition flux into the global ocean for
the STD simulation accounts for
An increase in the global DIP deposition of
The simulated annual mean nitrate surface concentrations in the seawater for the present day and the relative differences for past and future eras are presented in Fig. 2. The present-day surface nitrate distribution shows high concentrations along the equatorial divergence where nutrients are upwelled, and solar insolation supports good light conditions throughout the year. In the high latitudes, cooler water temperatures and seasonally damped light conditions reduce nutrient consumption by biological productivity, resulting overall in elevated annual mean nutrient concentrations. High surface concentrations are also calculated in the high-latitude Southern Ocean where the deep convection around Antarctica maintains high nutrient transport to the surface and productivity is limited by a thick mixed layer, lower water temperatures, and reduced light conditions. Elevated nutrient concentrations are likewise simulated in the eastern equatorial Pacific, and the subarctic North Pacific, i.e., the well-known HNLC regions. All in all, this reflects a reasonable simulation of the abiotic oceanographic drivers in the model. A comparison between the simulated present-day surface nitrate concentrations and the compiled data from WOA is given in Fig. S4.
Surface oceanic concentrations (mmol m
Increased atmospheric nitrogen deposition fluxes from the pre-industrial to
the modern era (Table 1) result in a respective increase in surface nitrate
concentrations in almost all oceanic regions, with some exceptions in the
eastern equatorial and the subpolar Pacific Ocean (Fig. 2a). In remote
oceanic areas, far from any coastal or riverine nutrient supply and thus
strongly nutrient limited, the higher present-day inorganic nitrogen and
iron atmospheric inputs compared to the pre-industrial era (Fig. 1a, d)
increase primary production (see also Sect. 3.3). This may, overall, lead to
increased export of the surface seawater nitrogen to the deeper ocean in the
form of sinking biogenic particles in these oligotrophic oceanic regions
(e.g., Krishnamurthy et al., 2007, 2010). For the
future conditions, however, the model calculates both negatives and positive
changes in the surface nitrate concentrations (Fig. 2c), resulting from an
overall decrease (
Figure 2 also presents the annual mean surface concentrations of iron for the present day in the STD simulation, along with the respective past and future relative differences. The present-day surface iron distribution (Fig. 2e) shows high concentrations in the subtropical North Atlantic Ocean and the Arabian Sea, with the lowest surface concentrations being calculated in the eastern equatorial Pacific and the Southern Ocean. A secondary maximum in iron concentrations is calculated in the subarctic North Pacific. In general, iron concentrations in the model are low, especially in the Southern Ocean, the eastern equatorial Pacific, and the subarctic North Pacific. Higher concentrations, however, are found along the coasts or over the continental margins. For the PAST (Fig. 2d) and FUTURE (Fig. 2f) conditions, the model generally calculates lower iron surface concentrations, reflecting overall the respective decreases in DFe deposition fluxes into the ocean (Fig. 1d, f, respectively). Consequently, the strongest declines are found in the Northern Hemisphere, especially in the mid- to high-latitude Pacific and Atlantic. An exception is the NW Pacific where higher iron input in FUTURE (Fig. 1f) results in elevated oceanic iron concentrations. A comparison of the simulated present-day surface oceanic iron concentrations with available DFe oceanic observational data (Tagliabue et al., 2012) is given in Fig. S5.
The annual mean present-day phosphate surface
concentrations, as calculated for the STD simulation together with the
respective relative differences for past and future eras, are presented in
Fig. 2g–i. The surface phosphate distribution in the model shows high
concentrations along the equatorial divergence where nutrient-rich deep
water is upwelled, as well as in the high latitudes, with the highest
surface concentrations to be simulated in the Southern Ocean. A secondary
maximum is calculated in the eastern equatorial Pacific and the subarctic
North Pacific; both regions are subject to large-scale upwelling of deep
waters. Note that in general, phosphate concentrations in deep waters are
higher than at the surface where nutrients are removed by biotic
productivity and exported by sinking particulate organic matter to the
deeper ocean. However, due to the constant Redfield ratio
(i.e., C : N : P
Nitrogen-fixation (kg N m
Despite the roughly factor-of-1.7 increase in the phosphate deposition inputs to
the ocean from the pre-industrial to the modern era (Table 1), the pre-industrial
surface phosphate oceanic concentrations are calculated 20 %–50 % higher in
most oceanic regions (Fig. 2g), except for the Southern Ocean where no
significant change is calculated. Accordingly, although there is a projected
decrease of the global phosphate input (Table 1), higher phosphorus surface
oceanic concentrations are simulated for the future, up to
For the STD simulation, the nitrogen fixation is calculated to about
112 Tg N yr
The present-day annual mean primary production together with the relative
differences compared to past and future periods are presented in Fig. 3d–f. The
primary production distribution in the open ocean shows high rates along the
equatorial divergence and in the high latitudes, where nutrient
concentrations are high. The decreased nitrogen deposition during
pre-industrial conditions compared to the present day results in lowered primary
production rates almost in all oceanic regions (Fig. 3d). A projected modest
decrease of primary production rates is also calculated by the model (Fig. 3f),
due to the lower (
The present-day modeled globally integrated production (
Despite the relatively strong changes in total atmospheric nutrient supply from PAST to FUTURE (Table 1), the impact of atmospheric nutrients on the global productivity rates remains low in the model. This is, nevertheless, not unexpected, as the atmospheric nutrient supply constitutes only a small fraction of the total ocean nutrient inventory. In addition, oceanic regions that are not nutrient limited today are less sensitive to external nutrient supply. Finally, a large part of primary production is regenerated by remineralized nutrients from particulate organic matter (mainly detritus) in the upper ocean layer.
Limitation for nanophytoplankton production by nutrients (N and P;
To further identify the oceanic regions that are particularly sensitive to changes in external nutrient inputs, the limiting factors for local productivity in the model are investigated. Figure 4a displays limitations due to nitrogen or phosphorus. High values (indicating low limitation) are seen in regions that are subject to intense upwelling, like in the equatorial divergence zones or the western margins of NW and southern Africa and South America (coastal upwelling). Accordingly, these regions are less sensitive to atmospheric deposition as nutrients are supplied from deep ocean layers. Lower nutrient limitation is likewise seen in the midlatitudes to high latitudes where limitations by temperature and light (Fig. 4b) limit the growth rates. Exceptions are the North Pacific, the Southern Ocean, and the equatorial Pacific where iron limitation matters (Fig. 4c). Consequently, the model's nutrient sensitivity is larger in the subtropics (in particular in the subtropical gyres), where good light conditions and warm waters support high growth rates. Furthermore, these regions are far from land nutrient sources, and so a major part of total primary production relates to regenerated production (i.e., with low rates of external nutrient supply) which is limited by nutrients, as both temperature and light are sufficient. This makes productivity in the subtropical gyres sensitive to changes in the external atmospheric nutrient.
Molar oceanic N : 16P ratios averaged in the upper 20 m for PRESENT,
as calculated by the model
In regions with significant macronutrient limitations, the elemental ratio of deposited N : P can be, however, rather important. To estimate the relative impact of the changes in this ratio, we calculated the modeled nitrogen concentrations relative to the model's Redfield ratio (Fig. 5a). For PRESENT, the model exhibits almost everywhere a deficiency with respect to nitrogen. This is in good agreement with data from WOA, which likewise indicate a predominant nitrogen deficiency (Fig. 5b). Next, the N : P ratio relative to the Redfield ratio as supplied by atmospheric deposition for PRESENT together with the changes in PAST and FUTURE is derived (Fig. S8). Overall, a strong excess of N compared to P for modern times is indicated. As a consequence of the model's nitrogen deficiency (Fig. 5a), the atmospheric nitrogen excess maintains higher productivity than without the atmospheric supply. For the pre-industrial era, however, the atmospheric N : P ratio is reduced almost everywhere, further increasing the N deficiency. Hence, rather the lowered atmospheric nitrogen inputs than the lowered phosphorus inputs in PAST and FUTURE are responsible for the diminished productivity in these experiments. To further demonstrate this, we carried out an additional sensitivity simulation (namely, the PIP simulation) where the phosphorus atmospheric deposition fluxes kept constant at pre-industrial levels, while the other studied atmospheric inputs (i.e., N and Fe) varied as for the STD simulation. As expected, the effect on phosphate concentrations (Fig. S9b) and productivity (Fig. S9d) in this sensitivity simulation remains extremely low, with the relative differences compared to STD being less than 1 % almost everywhere. Overall, this demonstrates that changes in phosphorus atmospheric deposition do not play a significant role in marine productivity from pre-industrial to future periods.
Model calculations demonstrate three major oceanic regions where the reductions in productivity are significantly stronger in PAST compared to PRESENT, i.e., the subtropical gyres of the Northern Hemisphere Pacific, the Atlantic Ocean, and the northernmost North Pacific (Fig. 3d). The subtropical gyres, however, are the most sensitive to changes in nitrogen input and clearly show the strongest productivity reduction compared to PRESENT. Indeed, the nitrogen concentrations in the subtropics are not much affected (Fig. 2a). This is because good light conditions and warm waters persistently maintain high rates of nutrient consumption, so nitrogen concentrations are already very low in PRESENT. Thus, a change in external nutrient supply feeds immediately into productivity without a significant imprint on nitrogen concentrations.
In the northernmost Pacific, the strong productivity decline in PAST (Fig. 3d) is primarily related to the lowered availability of iron (Fig. 2d), although the reductions in iron deposition remain below 20 % (Fig. 1d). However, besides the light conditions, iron availability is the most important factor for limiting productivity in this region (Fig. 4c) compared to nitrogen and phosphorus (Fig. 4a). As a consequence, slight changes in iron supply have a strong impact.
In the high latitudes, a large part of productivity is related to siliceous diatoms (e.g., Malviya et al., 2016; Uitz et al., 2010), which is accounted for in the model by the low ratios of nanophytoplankton to diatoms (Fig. 6b). Accordingly, the overwhelming part of productivity reduction in the northernmost Pacific (Fig. 3d) is related to the decline of diatoms. This is well reflected by the increase of the ratio of nanophytoplankton to diatoms for PAST relative to PRESENT (Fig. 6a). In turn, this leads to enhanced silicate concentrations in the North Pacific (Fig. 7a). Part of the unutilized silicate is advected southward via the North Pacific Current and the California Current, leading also to elevated concentrations along the western coast of North America (Fig. 7a). Note, however, that the Si atmospheric inputs here are solely dependent on the dust deposition fluxes, and thus they have no interannual variability in the model.
The oceanic concentration ratio of nanophytoplankton to diatoms averaged
in the upper 100 m for PRESENT
A further consequence of the strongly diminished productivity is an
accumulation of nitrogen in the subpolar gyre of the North Pacific (Fig. 2a).
The nitrogen anomaly is strongest in the southwestern area of the gyre,
and part of the excess nitrogen is injected into the northern California
Current. As a result, a strong positive and wedge-shaped productivity
anomaly develops in front of western Canada in PAST (Fig. 3d). This positive
anomaly is caused by the increased production of nanophytoplankton
productivity (not shown) which dominates in this region, as indicated by
higher ratios of nanophytoplankton to diatoms (Fig. 6b); i.e., north of the
wedge, lowered iron limits productivity, while south of the wedge, nitrate is
limiting it. Altogether, this reflects a slight shift from diatom to
nanophytoplankton production in the eastern Pacific (north of 40
Apart from the northernmost Pacific, the decline in diatom production leads to slightly increased silicate concentrations almost everywhere in PAST (Fig. 7a). Productivity changes in the Southern Ocean, however, remain low for PAST (Fig. 3d). The reason for this is the strong light limitation around Antarctica (Fig. 4b) and the deep mixed layer which suppresses productivity and subsequently builds up a large pool of unutilized nutrients. Part of the unutilized nutrients are advected further north into the Southern Ocean, driving productivity there. Accordingly, the reduced deposition of nitrogen and iron in this area (Fig. 1a, d) has only a slight impact on productivity. Consequently, this region is relatively robust against external nutrient input maintaining stable productivity. A similar effect is seen for the North Atlantic where vigorous exchange with Arctic waters takes place across the Norwegian and Greenland seas. By contrast, in the subpolar North Pacific, the import of unutilized nutrients from the Arctic is hampered, as the water exchange with polar waters is limited by the shallow Bering Strait and the Aleutian Arc. Therefore, the North Pacific appears the most sensitive to external nutrient inputs compared to other oceanic regions.
For most of the world ocean, productivity changes in FUTURE are in qualitative agreement with PAST but less pronounced (Fig. 3d, f). This is mostly because both FUTURE and PAST experiments reflect reduced anthropogenic emissions, and thus the same mechanisms are involved. The only notable exception compared to PAST (Fig. 3d) is demonstrated in the eastern North Pacific where a strong negative wedge-shaped anomaly is seen in FUTURE (Fig. 3f). This opposite response is related to different iron inputs to the North Pacific (Fig. 1d, f). In FUTURE, the reduction of iron atmospheric inputs (Fig. 1f) is by far less strong and in the NE Pacific is even higher than today; thus, the productivity increases in the NE Pacific subtropical gyre (Fig. 3f). As a result, more nitrogen is consumed in the subpolar gyre in FUTURE and no nitrogen accumulation takes place as in PAST (Fig. 2a, c). Accordingly, a strong negative nitrogen anomaly develops in the western North Pacific and nitrogen-depleted waters are advected southward along the California Current (i.e., opposite to PAST). Altogether, these results imply an extreme sensitivity of the North Pacific against changes in atmospheric iron input. By contrast, the North Atlantic, which is less affected by iron limitation, reflects a widespread decline in productivity mainly controlled by the reduced nitrogen inputs.
Most marine biogeochemistry studies mainly account for the inorganic fraction as the most important pool of nutrients from the atmospheric pathway. On the other hand, state-of-the-art atmospheric chemistry models nowadays not only efficiently calculate the total dissolved nutrient atmospheric deposition fluxes, but they include the organic part as well, which turns out to be rather important for the total magnitude of the atmospheric input to the ocean (Fig. S3). However, great uncertainty still exists concerning the importance of atmospheric nutrients' organic fraction on oceanic productivity. For this, we separated here the inorganic and organic fractions of N, Fe, and P deposition fluxes to investigate the role of their organic components in marine biogeochemistry. The differences in nitrogen-fixation and primary production rates between the STD and ORG simulations are presented in Fig. 8. Note that as for the riverine organic fractions in the model (see Aumont et al., 2015), we also assume here an instant transformation of the atmospheric dissolved organic nitrogen (DON) and DOP inputs to the respective inorganic fractions in the water column.
Nitrogen-fixation (kg N m
When the organic fraction of the atmospheric nutrients is considered in the
model, a modest decrease in the global nitrogen-fixation rates of
In the tropical Pacific Ocean, the nitrogen-fixation rates for the ORG
simulation are significantly more intense compared to STD (up to
Primary production increases almost in all ocean basins for the ORG
simulation (Fig. 8d), except some parts of the subpolar Pacific Ocean. In
particular, higher rates are calculated in the subpolar Atlantic Ocean (up
to 15 %). In the N-limited oceanic regions, the increased atmospheric
nitrogen deposition (Fig. S3b) directly increases the production rates (Fig. 8d).
Such a case is the western subtropical North Pacific, where atmospheric
N deposition supports an extra production of up to 15 %. The production
rates are also increased in the subtropical South Pacific and Atlantic
oceans up to nearly 20 %. In total, the primary production increased from
The main focus of this study is to investigate the effect of nutrient deposition on oceanic primary production. Hence, the presented experiments did not account for the impact of future climate change which could interact or may even mask the effect of changed atmospheric deposition fluxes considered here. Consequently, the here-found effects are subject to some uncertainties related to the potential interaction with climate change. For example, climate-induced changes in the global wind system may not only alter atmospheric pathways for nutrients but also impact on oceanic up- and downwelling. Thus, shifts in the seasonal position of trade winds will likewise force shifts in the position of open-ocean and coastal upwelling. These regions are usually nutrient rich and not particularly sensitive to varying atmospheric nutrient inputs. Displacements of these upwelling positions, as a result of climate change, can increase the sensitivity to external nutrient inputs in regions formerly impacted by upwelling.
Several studies have demonstrated that the midlatitude to high-latitude areas, such as the North Atlantic and the Arctic, will be more stratified in a future warmer climate (Bindoff et al., 2019; Fu et al., 2016; Gröger et al., 2013; Sein et al., 2018; Steinacher et al., 2010), with negative feedback on vertical mixing and marine primary production due to reduced upward transport of nutrients into the photic zone. Accordingly, primary production in these regions will probably be more sensitive to changed atmospheric deposition rates in the future. Our results, overall, imply only marginal effects in polar regions like the Arctic Ocean. This is certainly robust under the present climate when marine productivity is limited by temperature- and sea-ice-reducing light conditions in these regions. However, there is a large agreement that climate change will be most severe in the high latitudes, with strong increases in the water temperatures and substantially diminished sea ice cover in the Arctic (Collins et al., 2013). Temperature- and sea-ice-related light limitation will likely become less important in this region, and thus more nutrients will be recycled in the polar region and less exported equatorward. Consequently, changes in atmospheric transport and deposition of the bioavailable nutrients may play a larger role in a future climate, especially under the high-emission climate scenarios. An example can be seen in the high latitudes of the Southern Ocean around Antarctica where the major amount of surplus DFe is deposited in our FUTURE experiment (Fig. 1). As expected, the additional DFe availability has nearly no effect on productivity (Fig. 3d, f) as convective mixing and extremely low water temperatures maintain sufficient nutrients and support low productivity under the present-day climate. This may, however, change with altered oceanographic conditions under a future warmer climate. In the northernmost Pacific, known as an HNLC region where iron is the limiting factor, the increased supply of DFe clearly stimulates marine productivity in the PRESENT and FUTURE periods compared to PAST. However, this increase in productivity is likely overestimated here, since our experiments lack climate-induced changes in future stratification which would reduce the nutrient supply from the deep ocean.
The impact of atmospheric organic nutrients on the global oceanic
productivity turns out as high (
All changes in nutrient deposition fluxes in the model are solely driven by changes in the anthropogenic and biomass burning emissions, along with the changes in insoluble to soluble conversions rates due to atmospheric processing. Indeed, the atmospheric deposition fields used in this study did not account for any changes in dust (and bioaerosol) emissions. Instead, they were kept constant to the present-day atmosphere (i.e., the year 2010), although several studies suggest that dust fluxes may be sensitive to climate change and the land-use changes (e.g., Ginoux et al., 2012; Mahowald et al., 2010; Prospero and Lamb, 2003) and thus could be an important driver of the atmospheric nutrient cycles.
This study presents the implementation of state-of-the-art monthly mean
atmospheric deposition fields in the global biogeochemistry model PISCES.
The model runs in offline modus here, forced by dynamical physical outputs
from the physical ocean model NEMO. The newly coupled atmospheric deposition
fields considered for this work are all calculated based on a detailed
representation of emissions of natural and nutrient-containing combustion
aerosols, detailed atmospheric gas- and aqueous-phase chemical schemes, and
mineral dissolution processes due to atmospheric acidity and organic
ligands. Another feature tested in the present study is the contribution of
organic components to the atmospheric inputs to the global ocean. Moreover,
to effectively isolate here the impact of atmospheric deposition on the
marine biogeochemistry parameters, the atmospheric CO
For the present day,
This work asserts the importance of an explicit representation of the
atmospheric nutrients in the context of biogeochemistry modeling, providing
also a first assessment of the contribution of another source of atmospheric
nutrients than inorganics and thus highlights the potential importance of
organic nutrients on oceanic productivity. Overall, our main conclusions can
be summarized as follows:
There is a general low impact of atmospheric nutrient deposition scenarios on the
total marine primary production on a global scale. This is because much of
modern productivity is driven by nutrients already recycled in the euphotic
zone or by nutrient import from the deep ocean (such as in upwelling
regions). Additionally, atmospheric transport appears rather important, as a
significant part of nutrient deposition takes place in the northern high
latitudes, where light conditions and temperature further limit
productivity. Accordingly, even substantial reductions of nitrogen,
phosphorus, and iron inputs during the
pre-industrial period, result in an only modest decline in primary production compared to the present day. Substantial local productivity changes of up to 20 % are found in regions
limited by nutrients. The strongest sensitivity to atmospheric
nutrients is found for the oligotrophic subtropical gyres of the North
Atlantic and the North Pacific, where good light conditions and warm
temperatures together with low nutrient concentrations predominate.
Additional atmospheric nutrient input to these regions immediately results
in production by increasing the biogenic turnover. The North Pacific appears more sensitive to the external nutrient
atmospheric deposition compared to other oceanic regions mainly for two
reasons: the strongest deposition changes take place in the northern midlatitudes to
high latitudes, and compared to the Southern Ocean and the North
Atlantic, the exchange with cold and nutrient-enriched polar waters is
limited by land by the shallow Bering Strait and the Aleutian Arc. By
contrast, the southern high-latitude ocean contains a large amount of
unutilized nutrients that are advected further north (to midlatitudes),
making this region more robust against changes in external nutrient input.
In agreement, however, with observational evidence from WOA, PISCES exhibits
a widespread surplus of nitrogen compared to phosphorus and with respect to
the Redfield ratio. Therefore, the applied changes in phosphorus inputs have
nearly no impact on primary production in the model. This applies even to
the warm water regions, where reductions in atmospheric iron supply limit
nitrogen fixation by diazotrophs in both PAST and FUTURE periods. The North Pacific turns out to be the most sensitive ocean to iron
atmospheric deposition changes. For the pre-industrial period, the lowered
input of iron to this region leads to a strong decline in siliceous diatom
production, leading to an enrichment of silicate, nitrogen, and phosphorus.
In turn, this leads to enhanced equatorward transport of nutrients,
resulting in elevated production rates of calcareous nanophytoplankton
further southeast. Finally, the effect of atmospheric organic nutrient deposition fluxes on the
global primary production is calculated to be roughly as strong as the effect of
the present-day increased emissions and atmospheric processing on the
oceanic biogeochemistry since the pre-industrial era when only the inorganic
fraction is considered in the model (
Atmospheric nutrient deposition data used for
this study are available at Zenodo (
The supplement related to this article is available online at:
SM prepared the atmospheric input fields, performed the simulations, and conducted the model evaluation. MG prepared the oceanic model forcing data. SM and MG wrote the manuscript. JH and RD contributed to the manuscript preparation.
The authors declare that they have no conflict of interest.
Stelios Myriokefalitakis acknowledges financial
support for this research from the European Union's Horizon 2020 Framework Programme
under the Marie Skłodowska-Curie Actions (grant agreement
no. 705652 – ODEON). Matthias Gröger and Jenny Hieronymus acknowledge
support from the European Union's Horizon 2020 Framework Programme
(grant agreement no. 641816, “Coordinated Research in Earth
Systems and Climate: Experiments, kNowledge, Dissemination and Outreach
(CRESCENDO)”). Stelios Myriokefalitakis acknowledges support from the Joint
Group of Experts on the Scientific Aspects of Marine Environmental
Protection (GESAMP;
This research has been supported by the H2020 Marie Skłodowska-Curie Actions (ODEON, grant no. 705652), the Horizon 2020 Framework Programme, H2020 Environment (CRESCENDO, grant no. 641816), and the National Observatory of Athens (grant no. 5065).
This paper was edited by Piers Chapman and reviewed by two anonymous referees.