Response of Southern Ocean Resource Stress in a Changing Climate

Phytoplankton underpin ocean net primary production (NPP) and Southern Ocean phytoplankton display different ecological‐biogeochemical traits, compared to temperate species. Climate models currently forecast consistent across‐model NPP increases due to climate change, yet neglect specific aspects of the Southern Ocean ecological‐biogeochemical system. We conducted experiments to evaluate how key regional traits, including multiple limiting nutrients, unique photophysiology and differential resource acquisition, drive changes in the projected response of resource stress, NPP and export production under a high emissions scenario. Although Southern Ocean iron limitation is widespread, it declines in the future and is replaced by growing manganese limitation, as concentrations cannot support increasing growth rates. Distinct phytoplankton traits either amplify or dampen climate‐driven changes, depending on whether they are those typical of Antarctic or temperate phytoplankton, respectively. Overall, future Southern Ocean NPP trends may be more uncertain than currently assumed and future efforts should focus on accounting for regional ecological‐biogeochemical differences.


Introduction
The Southern Ocean is rich in upwelled macronutrients (such as nitrogen, phosphorus, and silica), however phytoplankton net primary production (NPP) does not drive full nutrient depletion, resulting in inefficient carbon removal (Sigman et al., 2010).This arises because insufficient availability of other bio-essential resources, such as the micronutrient iron (Fe), curtails phytoplankton growth (Moore et al., 2013).Macronutrients not consumed in the Southern Ocean are then transported to low latitudes via mode and intermediate waters to fuel NPP therein.Fe is mostly required by phytoplankton for photosynthesis, especially in photosystem, PSI (12 atoms) and PSII (3 atoms), as well as a lower requirement being associated with nitrate reductase and respiratory processes (Raven, 1990).In addition, another bio-essential micronutient, manganese (Mn), is also associated with PSII, where four atoms are involved in water-splitting reaction (Raven, 1990) as well as in the antioxidant superoxide dismutase (SOD) (Peers & Price, 2004;Wolfe-Simon et al., 2006).At present, results from both CMIP5 and CMIP6 models show strong across-model agreement in their projections of increasing Southern Ocean NPP in response to climate change (CC), due to the alleviation of Fe limitation (Leung et al., 2015;Misumi et al., 2014;Tagliabue et al., 2021).
Current CC projections for Southern Ocean NPP may be affected by the neglect of the role of other limiting nutrients and the unique traits of the resident phytoplankton.Previous work, based on field, experiment and modeling efforts, has highlighted a potential role for Mn limitation of NPP in the region, especially under changing Fe supply (Browning et al., 2021;Browning & Moore, 2023;Hawco et al., 2022;Middag et al., 2011Middag et al., , 2013)).Moreover, Southern Ocean phytoplankton are known to employ a unique set of physiological adaptations that allow more efficient use of Fe, relative to temperate groups, by both increasing their light-harvesting antenna size (Strzepek et al., 2019) and exploiting organically complexed Fe pools more efficiently (Strzepek et al., 2011).Adjustments to photophysiology are thought to affect both Fe and Mn demands (Hawco et al., 2022), and which affect NPP and export production (Anugerahanti & Tagliabue, 2023).However, we lack an assessment of how Fe and Mn impact regional NPP and export production in a CC scenario and the potential role played by distinct Antarctic and temperate phytoplankton traits.As these aspects of the Southern Ocean biogeochemical system have not been accounted for in NPP projections conducted to date, their contribution to uncertainty in a changing climate remains unknown.
To improve assessments how Southern Ocean NPP might be impacted by CC, we need to quantify at first order whether the specificities of the region are relevant for simulating the evolution of NPP in a changing climate.Key issues to address include (a) the response of Fe and Mn limitation to changing Fe and Mn supply and demand in the Southern Ocean and (b) if locally adapted traits make a significant contribution to NPP and export production trends.Here we address these challenges using state-of-the-art global model simulations under a high-emission scenario, where the role of Fe and Mn, as well as physiological traits in shaping the response of Southern Ocean NPP and export production are assessed using a suite of sensitivity studies.

Methods and Model Metrics
We used the PISCES-QUOTA global ocean biogeochemical ocean model (Aumont et al., 2015;Kwiatkowski et al., 2018), with the inclusion of the biogeochemical Mn cycle and phytoplankton Mn limitation (Anugerahanti & Tagliabue, 2023;Hawco et al., 2022).The model has three phytoplankton functional types (PFTs); diatoms, nanophytoplankton, and picophytoplankton, two zooplankton and multiple limiting nutrients (nitrogen, phosphorus, iron, manganese, and silica for diatoms).PISCES-QUOTA has a fully flexible stoichiometry, with each elemental pool tracked in living biomass and two size classes of sinking particles.External inputs of iron and manganese from aerosols, sediments, rivers, and hydrothermal vents are included.Our model explicitly calculates the uptake of Fe and Mn, independently of growth and carbon fixation; with the resource requirements of Fe set by the sum of those associated with photosynthesis, respiration and nitrate reductase and those for Mn associated with PSII and a basal Mn demand linked to growth rate, following (Hawco et al., 2022).
Our control experiment uses the default Fe-Mn limitation parameterization scheme, as described in Anugerahanti and Tagliabue (2023); where growth is limited Mn, and the Mn uptake can be affected by zinc (Zn), as it has higher affinity for the Mn transporter (Sunda & Huntsman, 1996, 1998).We conducted experiments using offline physical fields from IPSL-CM5A climate model under the high emissions RCP8.5 scenario.Each experiment consists of three simulations; from 1801 to 1851 we used a preindustrial control with atmospheric CO 2 concentrations fixed to 285 ppm.From 1851 to 2005, CO 2 concentrations and associated climate are varied according to the historical pathway, and then follow the high emission RCP8.5 scenario (Riahi et al., 2011) from 2006 to 2100.To quantify the difference between present and future simulation, we define the 1986-2005 average as the reference (REF) and 2091-2100 average as the CC periods.
We quantify the response of Fe and Mn deficiency to changing resource availability by calculating the total area deficient in a given resource.We define deficiency as when cellular demand for the given resource exceeds the available internal cellular concentration (Moore, 2016;Saito et al., 2008).To calculate deficiency for either Fe or Mn, we use the ratio between the overall Fe and Mn requirements for growth (Q Fe/Mn,req ) and the realized internal cellular concentration of each resource (or quota (Q Fe/Mn ).We can also define the actual resource limitation of growth by either Fe or Mn, depending on which element is most deficient (Moore, 2016).By isolating the individual contributions to overall deficiency in our model, we also assess whether changes in deficiency are driven by either altered cellular requirements for a given element (Q Fe/Mn,req ) or its acquisition (Q Fe/Mn ).
To address how different traits affect Southern Ocean resource deficiency, NPP and export production in a changing climate, we performed additional experiments where we varied the key traits that have been documented to differ between "Southern Ocean" and "temperate" phytoplankton.In our control model we assume an "average" Chl:PSU ratio of 1000:1 (Hawco et al., 2022;Strzepek et al., 2019) and represent the speciation of DFe assuming a prognostic Fe-binding ligand scheme (Völker & Tagliabue, 2015).Organically complexed Fe is assumed to be 75-fold less bioavailable than "free" Fe by default (Shaked et al., 2021), by applying a higher half saturation constant for organically complexed Fe uptake.We conducted three additional sensitivity experiments, that span different ecological assumptions associated with temperate or Southern Ocean traits as summarized in table 1.

Biogeochemical Context and the Extent of Fe and Mn Limitation
CC drives alterations to the physical and biogeochemical dynamics of the Southern Ocean under under a high emissions scenario.In our simulations, there is an increase in the minimum and maximum sea surface temperatures (SST) and shallower maximum mixed layer depths (MLD) in winter south of 40°S.The greatest increase in minimum SSTs occur mostly in the sub-Antarctic Atlantic, where they warm by ∼4°C, while maximum SST on average increase by ∼2°C, and warming at the same region (Figures S1a-S1f in Supporting Information S1).Shallower winter MLDs are found throughout the subantarctic Pacific and Indian sectors, as well as along the coast of Antarctica (Figures S1d-S1f in Supporting Information S1).Some regions to the south of Tasmania and New Zealand show deeper maximum MLDs, while most of the Atlantic sector of the Southern Ocean shows little change (Figures S1 in Supporting Information S1).For more discussion of MLD changes across CMIP5 models see Sallée et al. (2013).
Southern Ocean Fe limitation prevails in the control model during the reference period, but Mn limitation increases by ∼2.2% by the end of the century.This occurs because growth rates increase in response to greater DFe availability, and draw down NO 3 concentrations, but Mn concentration do not increase sufficiently (see Figure S2c in Supporting Information S1).Mn limitation increases because of the greater DZn concentrations in some regions (e.g., near West Antarctica (see Figure S2d in Supporting Information S1)), as Zn hinders Mn uptake (Hawco et al., 2022;Sunda & Huntsman, 1998).Overall, only 0.46% (3.59 × 10 5 km 2 ) of the Southern Ocean surface area is more deficient in Mn than Fe during the reference period in our model (Figures 1a-1e), similar to previous modeling studies (Anugerahanti & Tagliabue, 2023;Hawco et al., 2022) and consistent with the limited observational evidence for direct Mn limitation in the region (Browning & Moore, 2023).Fe limitation covers 99.5% of the Southern Ocean (7.73 × 10 7 km 2 ) and is most widespread for diatoms and least extensive for picophytoplankton (98.7% and 87.6% of the total area of limitation, Figures 1b and 1d, respectively) because of the higher modeled Q Fe but similar magnitude Q Fe,req .In the reference period, picophytoplankton have the strongest Mn limitation (Figure 1h), consistent with results of shipboard experiments conducted in the sub-Antarctic region of the Pacific sector (Latour et al., 2023).In our model, picophytoplankton have different pattern of Fe and Mn limitation (Figures 1d-1h), because of their small size that lessens Fe limitation and opens up the scope for Mn regulation of their increased growth rates.While Fe limitation declines, Mn limited area increases two-fold between the reference and CC periods (Figure 1e), with the largest relative changes in Mn limitation seen for diatoms and nanophytoplankton (Figures 1f and 1g).
The Southern Ocean-specific Large Antenna experiment reduces overall regional limitation for Fe and Mn compared to the control model (Figures 1a-1e, Figures S3a and S3b in Supporting Information S1).Fe and Mn limitation declines by 4.2 × 10 5 km 2 (0.54%) and 2.2 × 10 5 km 2 (60%), relative to the control model, during the reference period, respectively.Fe limitation continues to decline in this experiment in response to CC, but the relative change is 25% smaller than that seen in the control model (Figure S3c in Supporting Information S1).Note.We consider Southern Ocean traits to be a larger photosynthetic antenna (Strzepek et al., 2019).Temperate phytoplankton would have 10× lower bioavailability of organically complexed Fe (Strzepek et al., 2011) and smaller antenna (Strzepek et al., 2019).
Although both diatoms and nanophytoplankton show lower Fe limitation with a larger antenna, relative to the control model, for picophytoplankton, assuming a larger antenna increases their Fe limited area (Figure 1d).This is because picophytoplankton Mn limitation declines markedly (Figure 1h) in this experiment due to reduced Mn demands, which is then replaced by growing Fe limitation.
Both temperate trait experiments increase the extent of Fe limitation, but distinct trends emerge for Mn limitation.
In the Small Antenna experiment, Fe limitation increases by 0.6% (4.6 × 10 5 km 2 ) above that in the control model (Figure 1a), while for Mn limitation grew (8.5% more than the control model, averaged over the reference period).
In response to CC, Fe limitation remained resilient when in the Small Antenna experiment and, unlike the control model, did not decline.The Low Ligand Bioavailability experiment, leads to increased Fe limitation (1.0% or 7.4 × 10 5 km 2 ) during reference period (Figure 1a) and smallest extent of Mn limitation (87% less or 3.2 × 10 5 km 2 , Figure 1e).Although there is no direct effect of changing the bioavailability of the organically complexed DFe pool does not directly affect phytoplankton Mn requirements or acquisition, the Mn limitation declines due to growing Fe limitation.Unlike other experiments, the extent of both Fe and Mn limitation in the Low Ligand Bioavailability experiment increased in response to CC (less than 1%, 6.8 × 10 4 km 2 and 1.1 × 10 5 km 2 , respectively).Diatoms are the most sensitive to any alterations to traits, regarding their Fe limitation, especially when we assume a smaller antenna size (Figure 1f), while other traits showing less (Figures 1g  and 1h).Thus overall, traits typical of Southern Ocean plankton that lead to reduced Fe limitation tend to open a greater scope for Mn limitation, while temperate traits do the opposite.The top and bottom panels show the total extent of Fe and Mn limitation, respectively, with the total phytoplankton community shown in panels a and e, followed by the areas of Fe and Mn limitation for diatoms (b, f), nanophytoplankton (c, g) and picophytoplankton (d, h).Note that the y-axis scale is altered for picophytoplankton, to ranges between 5× 10 7 -7.85 × 10 7 in panel (d), and being set one magnitude higher than for the other PFTs in panel (h).

Drivers of Southern Ocean Fe and Mn Deficiency: The Role of Different Traits
The average change of deficiency in Fe and Mn deficiency may be small at the scale of the entire Southern Ocean (1.3% less and 2.2% more deficient compared to the reference period, respectively south of 40°S), but these signals can be much larger regionally.Fe and Mn can become ∼30% less deficient within the sub-Antarctic, during the most productive, but also the most deficient month (Figures 2a and 2d).In the more sub-Antarctic and polar regions of the Southern Ocean, Fe deficiency increases by more than 30%, while Mn deficiency can increase by up to 60% in the polar Southern Ocean.These changes also coincide with projected alterations to DFe, for Fe deficiency and both DMn, and DZn for Mn deficiency.
The change in Mn and Fe deficiency in our simulations is mostly driven by changes in phytoplankton resource acquisition, with a smaller role for altered resource requirements.From the control model, projected changes in Fe resource acquisition (quantified using the change in cellular quotas) show a very similar pattern to those projected for Fe deficiency (Figures 2a and 2b), and explain 62% of the overall change in Fe deficiency.In the offshore Southern Ocean, over 80% of the projected change in the demand for Fe (Figure 2c) is driven by photosynthesis, as Chl:C ratios incease alongside the increase in Fe concentration which increases growth rate (Figures S5a and S5b in Supporting Information S1).The remaining Fe demands are associated with elevated nitrate reductase requirements, due to the increase in nitrogen uptake, mostly along the Antarctic coastline where NPP rates are projected to increase in response to CC (Figure S5c in Supporting Information S1).For Mn, projected changes in Mn acquisition explained 58% of the overall Mn deficiency change (Figures 2d and 2e).As discussed previously, changes in Mn acquisition are driven by the DMn and DZn availability.Over 80% of the projected change in Mn requirements due to CC is driven by the ∼15% increase in phytoplankton growth rates due to reduced Fe limitation (Figures S5d and S5e in Supporting Information S1), with a much smaller role played by the greater Mn requirements from photosynthesis due to increased Chl:C ratios.The lower Mn requirements near East Antarctica (Figure 2f) are largely due to the decreased requirement in photosynthesis due to reduced Chl:C ratios in this region (Figure S5f in Supporting Information S1).
Altering photophysiological traits affects overall resource deficiency by modifying resource requirements, while altering Fe bioavailability impacts resource acquisition.Consistent with the changes in limitation, assuming Southern Ocean traits tends to cause a reduction in overall deficiency, while traits associated with temperate phytoplankton cause an increase (Figures S6a and S6d in Supporting Information S1).For Fe, the large antenna trait of Southern Ocean phytoplankton reduces their Fe requirements by 42% (Figure S6b in Supporting Information S1), which in turn reduces Fe deficiency by 17% (relative to the control model).While for Mn large antenna trait reduces Mn requirement by only 10%, which reduces deficiency by 4%, relative to the control model (Figures S7a and S7c in Supporting Information S1).For both Fe and Mn, the deficiency pattern still largely reflects the resource acquisition.Turning to the temperate traits, assuming a smaller antenna raises Fe requirements by 60%, but Fe deficiency only declines by 18% due to the higher Fe availability in the future (Figures S6e and S6f in Supporting Information S1).For Mn, the small antenna trait increases the projected changes in Mn requirements by 25%, but because Mn deficiency is also controlled by the availability and acquisition of Mn, the Mn deficiency is projected to be two-fold higher (Figures S7d-S7f in Supporting Information S1).When we assume the low ligand bioavailability in temperate phytoplankton, the projected change in Fe deficiency of 5.2% is fully controlled by Fe acquisition via the effect on Fe quotas, which decline by 26% (relative to the projected changes in the control model).Although this trait does not alter Mn deficiency directly, low ligand bioavailability still reduces the Mn acquisition by more than half, and increases Mn deficiency by more than twofold (Figures S7g-S7i in Supporting Information S1).This is because of the increase in both growth rate and Chl:C ratio, lead to an indirect increase in Mn deficiency.

Response of NPP and Export Production to Climate Change and the Influence of Different Traits
In response to CC, the control model shows an increase in NPP almost everywhere in the Southern Ocean, except for parts of the Pacific sector north of the subtropical front (Figure 3a).The projected increase in NPP was more muted in the sub-Antarctic, with a banded structure consistent with other CMIP5 and CMIP6 findings (Leung et al., 2015;Tagliabue et al., 2021).The regions with increased NPP in our model are also generally associated with an increased diatom fraction of total phytoplankton carbon biomass and show enhanced export production (Figures 3b and 3c).This is because diatoms respond most strongly to the greater availability of Fe and are assumed to generate large sinking particles in our model.Hence changes in the modeled ecology is also a contributor to the projected changes in export production (Tréguer et al., 2018).In the reference period, both picophytoplankton and diatoms were the major contributors to NPP in the open ocean and coastal areas, respectively, in the control simulation.In response to CC, picophytoplankton become more dominant in the subantarctic, while diatoms generally became dominant in the Antarctic sector south of ∼50°S (Figure S4 in Supporting Information S1).
During our trait experiments, the climate signals found in the control model were either amplified or dampened, depending on the traits applied.When we assumed Southern Ocean traits, the climate signal from the control model was generally amplified (Figures 3d-3f), while the temperate trait experiments led to a dampening of the response to CC associated with the control model (Figures 3g-3l).More specifically, the increases in NPP, diatom fraction and export production seen in the control model were enhanced by 14.6%, 8.05%, and 12.6%, respectively, when our model assumed a larger antenna size typical of Southern Ocean plankton (Figures 3d-3f).
Assuming the smaller antenna size typical of temperate plankton led to a reduction in the projected stimulation of NPP, diatom fraction and export production due to CC by 2.75%, 24%, and 22%, respectively (Figures 3g-3i).This reduction in the positive effects of CC on Southern Ocean NPP, diatom fraction and export production was even greater when Fe uptake traits of temperate plankton were applied, with reductions in NPP, diatom fraction, and export production of 26.9%, 36.5%, and 33.1%, respectively (Figures 3j-3l).We note that the sensitivity of NPP, diatom fraction and export production can be even higher than these regionally aggregated signals at regional scales (Figures 3d-3l).
The effect of applying Southern Ocean and temperate traits on the response of NPP and export production to CC is significant in the context of the current across-model uncertainty.If we use one standard deviation across the CMIP5 ensemble as a quantification of model uncertainty, projected changes in NPP and export production show an uncertainty of ±169 and ± 19.1 TgC year 1 , respectively.Turning to our sensitivity experiments, the change in Southern Ocean NPP has a sensitivity of around ±55 Tg C, equivalent to around one-third of the CMIP5 ensemble uncertainty.Moving to export production, the range of projected changes due to climate across our experiments was around ±15 TgC year 1 , which is almost equivalent to the current ensemble uncertainty.Overall, this shows that accounting for specific aspects of the Southern Ocean biogeochemical system can be significant in the context of the existing across-model uncertainty in the CMIP5 ensemble, especially for export production.Moreover, these importance of these signals are even larger for specific regions of the Southern Ocean (e.g., Figure 3), which may become significant for efforts to forecast changes in specific aspects of Southern Ocean biogeochemistry, ecosystems or the carbon cycle more locally for example, to inform policy responses.Overall, our results imply that better understanding the underlying mechanisms shaping the specifics of the Southern Ocean biogeochemical system is an important part of CC projections in this key region.

Wider Implications and Further Work
Our modeling work has highlighted how some of the unique biogeochemical-ecological aspects of the Southern Ocean revealed by recent field and laboratory work can be important in shaping the response of Southern Ocean NPP and export production in a changing climate.These aspects have been neglected by climate models to date and suggest that the relatively high "across-model" confidence in projections in this critical region may be misleading.However, our model also makes simplifications regarding certain aspects of regional ecology and phytoplankton physiology that warrant increased attention in future efforts.
Experimental studies that focus on the eco-physiological traits of Southern Ocean phytoplankton are needed to provide insight into adaptive strategies and the consequent impact on larger scale biogeochemical cycling and NPP or export production can be assessed by suitably complex ocean biogeochemical models.We applied "Southern Ocean" or "temperate" traits to all three of our modeled PFTs (picophytoplankton, nanophytoplankton, and diatoms) and did not address how PFT-specific traits or competition may modulate CC responses.For example, different PFTs have been shown to have specific abilities to take up organically bound iron (Strzepek et al., 2011(Strzepek et al., , 2019)).Our experiments show that such changes may be important in driving the responses of Southern Ocean NPP and export production to CC.Further adaptive responses by either the temperate or Antarctic phytoplankton groups (e.g., via modifications to their thermal performance curves in response to warming or changing resource deficiency (Boyd, 2019;Thomas et al., 2017)) may be important to consider.
In addition to disentangling key traits, further work is needed to fully explore the physiological feedbacks that may arise due to altered photosynthetic antenna sizes and concomitant irradiance changes.For example, any climate-driven modulation of irradiance, for example, due to altered MLD dynamics, may also drive additional feedbacks via their interaction with differences in antenna size across different phytoplankton groups.This may be important because while having a large antenna may be optimal in terms of alleviating Fe and Mn limitation, it has an impact on photosynthesis, especially when light levels are high (Raven, 1990;Ryan-Keogh et al., 2023;Strzepek et al., 2019).Such trade-offs for photosynthesis linked to altered Chl:PSU ratios in response to optimizing Fe and Mn demands are not accounted for in our model experiments.Moreover, if the prevailing antenna size trait of the overall phytoplankton community is driven by changes in the relative abundance of temperate and Antarctic phytoplankton groups in response to warming via specific thermal performance curves (Boyd, 2019), then the role of both temperature and light in shaping community level photosynthetic performance (and hence NPP) may be important issues to address in future experimental and modeling work of CC impacts.
Co-limitation between Fe and Mn has been recently observed and is beginning to be studied more extensively in the Southern Ocean.In addition to the interactions around growth rate, the interplay between Fe and Mn may also occur through the SOD enzyme (McCain et al., 2021;Niyogi, 1999).We did not include the Mn demand for SOD activation in this work and instead assumed this was accounted for via a basal Mn requirement that scales with growth rate (Hawco et al., 2022).Including MnSOD can exacerbate Mn deficiency greatly in recent modeling (Anugerahanti & Tagliabue, 2023), but poorly constrained because other trace metals, such as Fe, nickel, and copper, can also serve as substitutes for Mn as co-factors (Miller, 2012).The combined use of biogeochemical and -omics data sets may offer additional insights here as they may reveal whether Mn serves as the primary co-factor for SOD in the Southern Ocean and how this varies with altered Fe and Mn availabilities.When combined with modeling, this can quantify the potential additional impact on Mn limitation of NPP in a changing Southern Ocean.
In conclusion, our model experiments have shown that the projected reduction in Fe deficiency due to CC in the Southern Ocean is accompanied by increasing Mn deficiency.This aspect is neglected by all CMIP models that assume a large reservoir of macronutrients can be drawn down in response to increased Fe supply.Moreover, the projected changes in NPP and export production in our model due to CC also show sensitivity to the assumption of photophysiological and Fe acquisition traits typical of Southern Ocean and temperate phytoplankton.This suggests that the uncertainty in the projected impacts of CC on NPP and export production may have been underestimated by CMIP models that assume generic cosmopolitan PFTs with fixed traits.The next generation of ocean biogeochemical models used for CC studies would benefit from exploiting the growing insight from physiology and genomic studies to better capture the trade-offs between different traits in the Southern Ocean from a mechanistic standpoint.Such efforts would help quantify the key processes controlling important ecosystem indicators such as NPP and plankton biomass in the critically important Southern Ocean.

•
The impact of iron and manganese stress and role of unique Southern Ocean traits in a changing climate is assessed • Climate change decreases the area of the Southern Ocean limited by iron and increases the area limited by manganese • Traits typical of Southern Ocean phytoplankton amplify, while temperate traits reverse, the expected climate-driven trends Supporting Information: Supporting Information may be found in the online version of this article.

Figure 1 .
Figure 1.Time series of the total surface area of the Southern Ocean limited by Fe and Mn (with a 10 years smoothing applied) for the control model (black), the Southern Ocean specific trait experiment that assumes a large antenna (blue), and two temperate trait experiments that assume a small antenna (orange) and low ligand availability (red).The top and bottom panels show the total extent of Fe and Mn limitation, respectively, with the total phytoplankton community shown in panels a and e, followed by the areas of Fe and Mn limitation for diatoms (b, f), nanophytoplankton (c, g) and picophytoplankton(d, h).Note that the y-axis scale is altered for picophytoplankton, to ranges between 5× 10 7 -7.85 × 10 7 in panel (d), and being set one magnitude higher than for the other PFTs in panel (h).

Figure 2 .
Figure 2. The overall Fe and Mn deficiency change of the Southern Ocean (a and d) and the contributions from resource acquisition and availability, Q Mn/Fe (b and e) and resource demand, Q Mn/Fe,req (c and f) during seasonal minimum in December.Red and blue denote more deficient and less deficient area, respectively.Contributors that looks more similar to the deficiency plot means that it is the main contributor.The term CC and REF denote the climate change and reference period, while μ and Q Fe,opt denote the growth rate and optimum Fe quota, respectively.

Figure 3 .
Figure 3. Changes in annually averaged NPP (a,d,g,j), diatom biomass fraction (b,e,h,k) integrated over the upper 100 m and export production across 100 m (c,f,i,l) south of 40°S.Panels a-c show the difference between climate change and reference period from the control run.Panels d-l then show the difference in the response to climate change between the experiments and control model, which are separated into the Southern Ocean (d-f) and two temperate traits; antenna small (g-i) and low bioavailable ligand.

Table 1
Control Model and Experiments to Assess Resource Deficiency, NPP, and Export Production at the Southern Ocean