Surface Ocean Biogeochemistry Regulates the Impact of Anthropogenic Aerosol Fe Deposition on the Cycling of Iron and Iron Isotopes in the North Pacific.

Distinctively-light isotopic signatures associated with Fe released from anthropogenic activity have been used to trace basin-scale impacts. However, this approach is complicated by the way Fe cycle processes modulate oceanic dissolved Fe (dFe) signatures (δ56Fediss) post deposition. Here we include dust, wildfire, and anthropogenic aerosol Fe deposition in a global ocean biogeochemical model with active Fe isotope cycling, to quantify how anthropogenic Fe impacts surface ocean dFe and δ56Fediss. Using the North Pacific as a natural laboratory, the response of dFe, δ56Fediss, and primary productivity are spatially and seasonally variable and do not simply follow the footprint of atmospheric deposition. Instead, the effect of anthropogenic Fe is regulated by the biogeochemical regime, specifically the degree of Fe limitation and rates of primary production. Overall, we find that while δ56Fediss does trace anthropogenic input, the response is muted by fractionation during phytoplankton uptake, but amplified by other isotopically-light Fe sources.

Introduction

Aeolian deposition is an important source of nutrients to the open surface ocean (Hamilton et al., 2022) especially in remote areas where other sources are limited. In the case of iron (Fe), an essential micronutrient for phytoplankton growth, this atmospheric supply is of particular importance as surface ocean dissolved Fe (dFe) concentrations can be extremely low. Consequently, much research has been devoted to understanding sources, character, transport and dissolution of Fe‐bearing aerosols, with the main focus historically being on Fe input from desert dust (e.g., Baker et al., 2016). By fertilizing past and present surface ocean ecosystems, such dust‐sourced Fe deposition is thought to induce substantial carbon sequestration and to be partly responsible for reduced atmospheric CO2 on glacial and millennial timescales (Martin, 1990; Martínez‐Garcia et al., 2011, 2014). However, wildfires and anthropogenic sources (combustion, biomass burning) also release Fe‐bearing aerosols and while the total Fe supply from these pyrogenic sources is lower than for dust (which contributes ca. 95% to total emissions), Fe solubility can be orders of magnitude higher (Ito et al., 2021). This increased solubility may partly be related to the co‐emission of acidic species, which work to enhance the bioaccessibility of this Fe source for phytoplankton (Li et al., 2017).

Isolating the impact of anthropogenic aerosol Fe (anthro‐Fe) on marine biogeochemical cycles requires the disentangling of different aerosol Fe sources. This is challenging as both dust and pyrogenic Fe emissions exhibit high temporal and spatial variability (Hamilton et al., 2020). Furthermore, aerosols often mix during transport, and the atmospheric processing of insoluble minerals to a “soluble” form (referred to as “dFe” hereafter, as it corresponds to oceanic dFe) adds another degree of complexity (e.g., Meskhidze et al., 2019). Mechanistic atmospheric aerosol Fe modeling (Myriokefalitakis et al., 2018), can help trace the origin of marine dFe deposition, whereas geochemical methods allow a more direct assessment of aerosol Fe impacts in the ocean. A recent approach exploits the variable isotopic Fe signatures (δ56Fe) of aerosol Fe from different sources, namely the distinctively‐light δ56Fe linked to certain anthropogenic combustions processes (up to −4‰, Kurisu et al., 2019), which is in stark contrast with the crustal δ56Fe (ca. +0.1‰) observed in dissolvable desert‐dust Fe (Conway et al., 2019; Waeles at al., 2007). These differences in source “endmember” δ56Fe have been used to infer the anthropogenic contribution to marine aerosol Fe (Conway et al., 2019; Kurisu et al., 2021) and surface ocean dFe (Pinedo‐González et al., 2020). However, surface ocean dFe isotopic signatures (δ56Fediss) are most likely driven by the complex interplay of external sources and fractionation during internal cycling (König et al., 2021), which, together with atmospheric processing, raise uncertainties for using simple mass balance approaches to constraining natural and anthropogenic dFe deposition.

To investigate the effect of anthro‐Fe on surface ocean dFe and δ56Fediss, we coupled a novel aerosol dFe deposition scheme to a δ56Fe‐enabled biogeochemical ocean model (König et al., 2021). While our modeling approach is global, we focus on the North Pacific as a natural laboratory, both because it receives substantial anthropogenic and natural aerosol Fe input, and due to its contrasting productivity regimes (Fe‐limited subpolar gyre and nitrogen‐limited subtropical gyre; Longhurst, 2007). We find distinct responses of dFe concentration and δ56Fediss to the anthro‐Fe input flux. However, neither response fully corresponds to the anthro‐Fe deposition pattern, due to the diverse biogeochemical state of the surface ocean across the region.

Methods

We used a version of the PISCES biogeochemical ocean model (Aumont et al., 2015) with variable particle reactivity (Aumont et al., 2017) and a dynamic ligand pool (Völker & Tagliabue, 2015) which incorporates δ56Fe cycling by including two prognostic tracers each (heavy 56Fe and light 54Fe) for dFe, diatom Fe, nanophytoplankton Fe, small and large particulate Fe pools (König et al., 2021). Isotopic fractionation factors (α) are applied to phytoplankton uptake (α of 0.9995), and complexation by organic ligands (α of 1.0006), so that uptake and scavenging (of free dFe) preferentially remove light dFe, and colloidal pumping (of complexed dFe) heavy dFe.

For the three aerosol dFe sources (desert dust, wildfires, and anthropogenic activity), we applied monthly mean dFe deposition fluxes from a 35‐year simulation (1980–2014) of the CAM6 atmospheric model with MIMI Fe mechanism (Hamilton et al., 2020). In addition to tracing the dFe fraction of each Fe source, MIMI accounts for both proton‐ and organic‐ligand dissolution of Fe during transport. Anthro‐Fe emissions are based on an inventory by Rathod et al. (2020), updated to cover the period 1980–2014, and include metal smelting and shipping among other industrial, residential, and traffic Fe. This anthro‐Fe signal neglects dFe released from anthropogenic acidic processing of dust or wildfire Fe, which is instead included in their respective dFe deposition fluxes (assuming no alteration to δ56Fediss during processing). We also included dFe input beneath the surface ocean layer via subsurface dissolution of desert‐dust particles, as described in Aumont et al. (2015), whereas for wildfire and anthro‐Fe only surface dFe deposition is included.

We prescribed the same source δ56Fe endmembers as in König et al. (2021): input of moderately‐light hydrothermal dFe (−0.5‰), light to crustal sedimentary dFe (−1‰ to +0.09‰), neutral riverine dFe (0‰), and crustal dust dFe (+0.09‰). For wildfire dFe, we chose a moderately‐light endmember (−0.5‰), based on the generally light values observed in above‐ground plant tissue (Wu et al., 2019 and references therein). For anthro‐Fe, we applied a light endmember (−1.6‰) based on dFe signatures observed for North Atlantic marine aerosol samples likely of anthropogenic origin (Conway et al., 2019).

To determine the impact of anthro‐Fe both on surface ocean Fe cycling and productivity, experiments were run with (“standard” experiment) and without anthro‐Fe deposition, and compared. We also carried out additional experiments to (a) assess the impact of a very light anthropogenic endmember (−4‰) based on δ56Fe observed by Kurisu et al. (2019) for dFe from fine aerosol particles sampled close to a steel plant; (b) evaluate the role of surface ocean processing, with model experiments with either phytoplankton uptake or complexation fractionation turned off (i.e., α set to 1); and, (c) determine the respective contribution of each external dFe source to surface ocean δ56Fediss using experiments where the isotopic effect of each source was artificially muted by setting their endmember to 0‰. For an overview of all experiments and their rationale see Table S1 in Supporting Information S1.

All experiments were run off‐line (i.e., a repeating climatological annual cycle of ocean physics) and with identical sedimentary, river, and hydrothermal dFe input for each year. This allows us to isolate dFe deposition from dust, wildfires, and anthropogenic sources as the only cause of interannual variability. Each experiment was spun up for 200 years (1780–1980) using an average monthly deposition field (1980–2014 mean) for natural sources (dust and wildfires). Following Krishnamurthy et al. (2009), we applied a linearly‐increasing scaling factor for anthro‐Fe input from zero emissions in 1880–1980 values, based on the quasi‐linear increase in black carbon emissions during this period (Bond et al., 2007). From 1980 to 2014 monthly varying deposition of dust, wildfire and anthropogenic dFe was used.

Results and Discussion

Annual dFe Deposition and Impact of Anthro‐Fe

For a representative overview of anthro‐Fe deposition to the North Pacific and the impact on surface ocean biogeochemistry, we focus on model results from 2014. As the last year of our simulations, the accumulated effect on dFe concentrations is highest in 2014, but 2014 signals are very similar to the 2010–2014 average (Figure S1 in Supporting Information S1). As expected, there is a distinct west‐east gradient in aeolian dFe input, dominated by dust (Figures 1a–1c). Nevertheless, the anthropogenic contribution is often around 30% of total aerosol dFe deposition and dominates in the westernmost part of the basin. Wildfire dFe inputs are generally minor in this region (Figure 1c), contributing <20% of total aerosol dFe, but higher in the southwest.

Figure 1

Dust (a), anthropogenic (b), and wildfire (c) dFe deposition fluxes (μmol/m2/year) and standard experiment surface ocean (0–10 m) dFe concentration (d; μmol/m3) and δ56Fediss (g; ‰) for 2014. Absolute (e; μmol/m3) and relative (f; %) change in dFe concentration, in δ56Fediss (h; ‰), and for primary production (i; mmolC/m3/year); calculated by subtracting the experiment without anthro‐Fe from the standard experiment (see Table S1 in Supporting Information S1). Red boxes indicate regions analyzed in Section 3.2.

The impact of anthro‐Fe on surface ocean biogeochemistry depends not only on the magnitude of the deposition flux, but also on the local Fe cycle, including the Fe limitation the biota experience. For example, while absolute dFe concentration changes are largest in the west where deposition is highest (Figure 1e), the largest relative change is observed further south and east (Figure 1f), where dFe concentration is typically much lower (Figure 1d), whereas the highest impact on primary production (Figure 1i), is in the Fe‐limited subpolar region.

The impact of anthro‐Fe on δ56Fediss is distinct compared to dFe changes throughout the North Pacific (Figures 1g and 1h). Again, the response of δ56Fediss does not simply follow the atmospheric footprint of anthro‐Fe deposition, but is instead largest in areas with moderate anthro‐Fe input, associated with either a large relative dFe change or a primary production response. While anthro‐Fe addition lowers surface ocean δ56Fediss throughout the region (by up to 0.5‰), δ56Fediss often remains heavy overall, even in areas with a large response to anthro‐Fe (e.g., the eastern subpolar gyre).

Seasonal and Regional Variability

To illustrate the seasonal impact of anthro‐Fe deposition we chose two regions with distinct biogeochemistry—one subpolar, Fe‐limited (region 1), and one in the oligotrophic subtropical gyre (region 2; Figures 1 and 2). Despite some regional variability, seasonal trends are broadly consistent within our chosen regions (Figure S2 in Supporting Information S1). In both regions, anthro‐Fe deposition (Figure 2a) amplifies the seasonality in dFe concentration, primary production, and phytoplankton Fe uptake (Figures 2b–2d). This effect is most pronounced for Fe‐limited region 1, which has a strong seasonal cycle in light and mixed layer depth, leading to a replenishment of dFe in winter and a summer drawdown associated with primary production and Fe uptake (Figures 2b–2d; black lines). Here, anthro‐Fe input increases dFe concentration (ca. +4–30 nmol/m3; Figure 2c) and Fe uptake rates (ca. +2.7 nmol/m3/day, at maximum; Figure 2d) and enhances the early summer primary production peak (Figure 2b; ca. +250 µmolC/m3/day), although with little effect or even decreased productivity in late summer. In contrast, there is little response to primary production in oligotrophic (nitrogen‐limited) region 2 (ca. +6 µmolC/m3/day, at most; Figure 2b), but, instead, a larger increase in dFe concentration (ca. +50 nmol/m3, on average; Figure 2d). In region 2, despite some seasonal changes in Fe uptake, further enhanced by the addition of anthro‐Fe (+2.1 nmol/m3/day, at maximum; Figure 2c), the reduced mixed layer cycle causes much smaller seasonality overall than in region 1.

Figure 2

Seasonal variability (year 2014) in dFe deposition (a; nmol/m2/day) for a subpolar, Fe‐limited region (Region 1) and a subtropical, nitrogen‐limited region (Region 2; see Figure 1), and the corresponding simulated surface ocean (0–10 m) primary production (b; mmolC/m3/day), phytoplankton Fe uptake (c; nmolFe/m3/day), dFe concentration (d; μmol/m3), and δ56Fediss (e, f; ‰) for experiments with (red) and without (black) anthro‐Fe deposition (difference in blue). For δ56Fediss, the contribution of δ56Fe fractionation and source endmember effects is shown. See Table S1 in Supporting Information S1 for details on attribution.

In contrast to the other parameters (Figures 2b–2d), the δ56Fediss seasonality in region 1 is muted by the addition of anthro‐Fe (Figure 2f). In part, this is due to the deposition of isotopically‐light anthro‐Fe in late summer, which decreases the previously very heavy δ56Fediss by over 0.5‰ (Figures 2e and 2f; short dashes). However, 20%–40% of the δ56Fediss decrease is, in fact, due to fractionation effects during internal cycling, as the additional (anthropogenic) dFe supply and the generally weaker productivity in late summer relieves some of the low dFe conditions during which uptake fractionation would otherwise cause very heavy δ56Fediss (Section 3.3). In region 2, the δ56Fediss seasonality remains weak, as anthro‐Fe is responsible for a near constant δ56Fediss decrease of ca. −0.3‰, of which 11%–14% is due to fractionation effects.

Overall, our results indicate that the impact of anthro‐Fe, on both annual and seasonal scales, strongly depends on the underlying biogeochemical state of the upper ocean, as illustrated by the differential responses of the two example regions.

Surface Ocean δ56Fediss Disentangled

The dynamics of δ56Fediss in the upper ocean depends on the combination of source δ56Fe endmembers and δ56Fe fractionation during Fe cycling. In general, the main drivers of the modeled surface ocean δ56Fediss distribution are fractionation during phytoplankton uptake, which drives the δ56Fediss toward heavier values, and dFe input from reducing sediments, which has the opposite effect (Figures 3a and 3d). Complexation fractionation and anthro‐Fe deposition are of intermediate or local importance in our model (Figures 3b and 3e), whereas the impact of dust and wildfire dFe on δ56Fediss are broadly negligible in this region (Figures 3c and 3f).

Figure 3

Effect of external dFe sources and fractionating processes on surface ocean (0–10 m) δ56Fediss (a–f; ‰, average for 2014), calculated by subtracting δ56Fediss of experiments with muted δ56Fe effects from standard experiment δ56Fediss (see Section 2 and Table S1 Supporting Information S1). Panels (a–f) sum to the overall δ56Fediss distribution (Figure 1g), with a <0.02‰ discrepancy due to hydrothermal dFe.

The strong effect of uptake fractionation on δ56Fediss results from high Fe uptake rates, independent of the degree of Fe limitation, and is largest where dFe concentrations are very low. The large effect of sedimentary dFe on δ56Fediss is due to its pronounced light endmember at shallow depths (−1‰ in the uppermost ca. 400 m; König et al., 2021) and the large dFe input fluxes in coastal areas. Seasonally, the impact of sedimentary dFe is highest in winter as deeper mixed layers entrain additional sediment‐sourced dFe (Figure S3 in Supporting Information S1). Uptake fractionation dominates in late summer, when uptake rates are high and surface dFe has been drawn down to low levels. The effect of complexation fractionation is negligible over most of the region, or even drives δ56Fediss to lighter values in areas where colloidal pumping is the dominant abiotic dFe removal process (König et al., 2021). An exception is the subpolar North Pacific and coastal regions, where elevated scavenging rates, due to high particle and dFe concentrations, mean that complexation is the dominant driver of δ56Fediss (toward heavier values), as only the isotopically‐lighter, uncomplexed dFe is scavenged.

The muted impact of dust dFe on δ56Fediss is due to its (near‐zero) crustal endmember; therefore, dust dFe acts as a “buffer” on δ56Fediss. The effect of wildfires on δ56Fediss is also limited compared to other dFe sources due to its low deposition flux in the area and years studied and a relatively‐moderate endmember (−0.5‰), which may be heavier still if soil‐Fe entrainment (with crustal δ56Fe) contributes substantially to wildfire Fe (Kurisu & Takahashi, 2019). The impact of the isotopically‐light anthropogenic endmember on δ56Fediss is most pronounced in open ocean areas and in late summer, when the mixed layer is shallowest (Figures 3 and S1 in Supporting Information S1), and increases substantially with a lighter choice of endmember (Section 4.2). This arises from the limited impact of subsurface sedimentary dFe signals in these areas and months. However, even though the impact of anthro‐Fe on δ56Fediss is highest in these summertime open ocean systems, their resultant δ56Fediss is often still heavy (Figure 1g).

Overall, our findings highlight that directly assessing the extent of anthro‐Fe deposition from surface ocean δ56Fediss requires a careful evaluation of other dFe sources and the internal Fe cycling that operate alongside, particularly dFe input from reducing sediments and fractionation during phytoplankton uptake.

Synthesis and Perspectives

A Mosaic in the Biogeochemical Response to Anthro‐Fe

The model also allows us to assess the varying effect of anthro‐Fe deposition globally (Figure 4). We find that the response of surface ocean systems to anthro‐Fe deposition can broadly be characterized into four different categories based on their underlying biogeochemistry (principally degree of Fe limitation and primary productivity; Figure 4a), as highlighted for the North Pacific (Figure 4b), which receives substantial anthro‐Fe inputs (Figure 1b).

Figure 4

Effect of anthro‐Fe deposition on different ocean biological regimes (a), in the North Pacific (b), and at global scale (c). Note that the depicted regions are illustrative of the present day, and may change if, for instance, nutrient input patterns change (e.g., shifts in aeolian deposition). In hatched areas, the effect of anthro‐Fe is weaker and limited to months of highest productivity.

The effects of anthro‐Fe on surface dFe and δ56Fediss varies significantly by category which highlights the importance of the underlying Fe limitation and productivity regime (Figure 4). The mixed layer dFe inventory increases in all non‐Fe limited regions (Cases 1 and 2), compared to Fe‐limited regions (Cases 3 and 4). Conversely, anthro‐Fe deposition stimulates summertime primary production in all Fe‐limited systems (Cases 3 and 4), with a greater impact in Case 3 high‐productivity regimes, but does not stimulate productivity in non Fe‐limited regions (Case 1 and 2). In Case 4 low‐productivity Fe‐limited regimes, the stimulation of productivity by anthro‐Fe is only moderate, and increased productivity in early summer leads to decreased productivity later in the season. In both Case 3 and 4 regimes, dFe concentrations are low and remain unchanged outside unproductive winter months, whereas δ56Fediss decreases, most prominently in Case 4 regimes due to the absence of light sedimentary dFe in these open ocean, low‐productivity areas (Figure 3d) and a dampening of uptake fractionation in late summer when productivity is decreased.

In high‐productivity, non‐Fe limited regions (Case 1), Fe uptake also responds to anthro‐Fe addition, but as this occurs without any stimulation of net primary production it is “luxury” Fe uptake. The extent of luxury uptake depends on how close phytoplankton are to their maximum Fe quota and where luxury uptake is large, it can strongly dampen the increase in dFe inventory so that the dFe concentration increases less than would be expected from anthro‐Fe deposition rates alone. For both Case 1 and 2 non‐Fe limited systems, the tendency of anthro‐Fe to drive lighter δ56Fediss depends both on the magnitude of the dFe pool and the influence of other light dFe sources (notably reducing sediments).

While our focus has been on the North Pacific as a case study, anthro‐Fe deposition, as simulated up to 2014, clearly elicits a response in other regions (Figure 4c). For example, substantial anthro‐Fe deposition to the Northern Indian Ocean leads to an increase in mixed layer dFe inventory, similar to the subtropical North Pacific (Case 2, although Fe limitation may be underestimated by our model in the Arabian Sea; Moffett & Landry, 2020). Primary production and Fe uptake are stimulated in parts of the equatorial Pacific and in the summer months of parts of Atlantic and Southern Indian Ocean, which represent productive, Fe‐limited systems (Case 3), partly due to downstream effects rather than local anthro‐Fe deposition. Downstream effects can also lead to a decrease in productivity as increased nitrogen consumption stimulated by anthro‐Fe can lead to downstream nitrogen limitation, most prominently in the eastern North Pacific. Overall, anthro‐Fe deposition leads to a small increase in global ocean primary production of 0.1% (+0.3% in the uppermost layer), a global increase in Fe uptake of 1.2% (+1.7%), and a dFe concentration increase of 0.3% (+2.4%).

Does Light δ56Fediss Trace Anthro‐Fe Input?

In systems that are not Fe‐limited and show little productivity (due to other limiting factors), light δ56Fediss may be a useful indicator of anthro‐Fe deposition. However, light δ56Fediss can also be related to sedimentary dFe inputs (e.g., Homoky et al., 2009; Severmann et al., 2010), which then need to be thoroughly accounted for when attempting to isolate anthro‐Fe inputs. In higher‐productivity systems, fractionation during Fe uptake drives δ56Fediss toward heavier values in general, so that the “light” δ56Fediss associated with anthro‐Fe input may be masked. Enhanced fractionation in systems where Fe is the limiting nutrient further complicates simple association of δ56Fediss signals with anthro‐Fe deposition, especially as the system responds to the extra dFe supply. Hence, an assessment of productivity and nutrient limitation status is necessary to more fully link surface ocean δ56Fediss signals and anthro‐Fe input. For this purpose, complementary observations may be useful, such as cellular Fe quotas (e.g., Twining et al., 2021), shipboard experiments of Fe uptake (e.g., Boyd et al., 2012) or omics‐based measurements of Fe stress‐induced proteins (e.g., Caputi et al., 2019) or proteomic Fe stress biomarkers (e.g., Saito et al., 2014).

Finally, the assumption of the anthropogenic δ56Fe endmember is crucial in estimating anthro‐Fe input based on δ56Fediss. Here, we opted for a value (−1.6‰) that, while lighter than any other dFe source in the model, is rather conservative, as some marine‐aerosol δ56Fediss observations are lighter than the combined aeolian‐aerosol δ56Fediss (Figure S4 in Supporting Information S1). Prescribing an even lighter endmember (−4‰) strongly amplifies the effect of anthro‐Fe on surface ocean δ56Fediss, rivaling or even exceeding that of sedimentary dFe (Figure S5 in Supporting Information S1), and, for some locations, compares better to the North Pacific surface ocean and aerosol δ56Fediss observations (Figures S4 and S6 in Supporting Information S1). To better constrain the anthro‐Fe endmember, and thus isolate the impact of surface ocean processes, ocean δ56Fediss measurements should be combined with parallel quantification of aerosol δ56Fe (e.g., Conway et al., 2019; Kurisu et al., 2021). This would also help in assessing the natural variability of anthropogenic aerosol δ56Fe, for instance, between different aerosol size fractions (Kurisu, Sakata, et al., 2016; Kurisu, Takahashi, et al., 2016; Kurisu et al., 2019) and/or effects of atmospheric processing on aerosol δ56Fe (Mulholland et al., 2021).

Future Importance of Anthro‐Fe

While the global response to anthro‐Fe deposition in our simulation period (1980–2014) is relatively modest and local, the importance of anthro‐Fe may increase in the future, also depending on changes to other Fe sources. Anthro‐Fe emissions are expected to rise across Asia even if fossil fuel emissions are replaced, as metal smelting already dominates much of the Fe emission source from China and India (Rathod et al., 2020) and is predicted to proliferate as global demand increases. The North Pacific is thus likely to remain a key region in understanding the impact of human activity on ocean biogeochemical cycles. The impact of natural emissions, however, will depend more on climate and human land‐use factors, such as changes in temperature and precipitation which alter aridity, or agricultural expansion which alters vegetation distributions and fire spread. In particular, wildfire activity is generally predicted to increase in extra‐tropical regions (Bowman et al., 2020 and references therein), which provide Fe to the North Pacific, and, as seen for the South Pacific, such changes can have large impacts on ocean biogeochemistry (Tang et al., 2021). Finally, all aeolian dFe deposition is linked with changes to atmospheric circulation and rain patterns (Letelier et al., 2019), as well as air pollution, which affects solubility (Hamilton et al., 2020), whereas entrainment of dFe from other sources, such as sediments, may be linked to changes in ocean circulation and stratification. Thus, the impact of anthro‐Fe is also linked with ongoing natural and anthropogenic climate variability, such as the Pacific Decadal Oscillation in the North Pacific or, on a larger scale, global warming.

Supporting Information S1

Click here for additional data file.