A Circum‐Antarctic Plankton Isoscape: Carbon Export Potential Across the Summertime Southern Ocean

The Southern Ocean accounts for ∼30% of the ocean's CO2 sink, partly due to its biological pump that transfers surface‐produced organic carbon to deeper waters. To estimate large‐scale Southern Ocean carbon export potential and characterize its drivers, we measured the carbon and nitrogen isotope ratios of surface suspended particulate matter (δ13CSPM, δ15NSPM) for samples collected in summer 2016/2017 during the Antarctic Circumnavigation Expedition (364 stations). Concurrent measurements of phytoplankton community composition revealed the dominance of large diatoms in the Antarctic and nano‐phytoplankton (mainly haptophytes) in open Subantarctic waters. As expected, δ13CSPM was strongly dependent on pCO2, with local deviations in this relationship explained by phytoplankton community dynamics. δ15NSPM reflected the nitrogen sources consumed by phytoplankton, with higher inferred nitrate (versus recycled ammonium) dependence generally coinciding with higher micro‐phytoplankton abundances. Using δ15NSPM and a two‐endmember isotope mixing model, we quantified the extent of nitrate‐ versus ammonium‐supported growth, which yields a measure of carbon export potential. We estimate that across the Southern Ocean, 41 ± 29% of the surface‐produced organic carbon was potentially exported below the seasonal mixed layer during the growth season, with maximum export potential (50%–99%) occurring in the Antarctic Circumpolar Current's southern Boundary Zone and near the (Sub)Antarctic islands, reaching a minimum in the Subtropical Zone (<33%). Alongside iron, phytoplankton community composition emerged as an important driver of the Southern Ocean's biological pump, with large diatoms dominating regions characterized by high nitrate dependence and elevated carbon export potential and smaller, mainly non‐diatom taxa proliferating in waters where recycled ammonium supported most productivity.

• Phytoplankton community composition influences suspended particulate matter carbon and nitrogen isotope ratios across the Southern Ocean • The nitrogen isotopes of particulate matter and a two-endmember isotope mixing model can be used to estimate carbon export potential • 40% of summertime primary production is potentially exported, with a higher fraction exported near (Sub)Antarctic islands and melting ice

Supporting Information:
Supporting Information may be found in the online version of this article.

Introduction
Phytoplankton are the foundation of marine food webs (Reynolds, 2006) and contribute to climate regulation by fixing CO 2 into organic carbon biomass (i.e., primary production; Falkowski et al., 1998).In the Southern Ocean, iron and light limitation of phytoplankton lead to mixed-layer macronutrient (i.e., nitrate and phosphate) concentrations that are perennially high (Martin et al., 1990;Sunda & Huntsman, 1997).Nonetheless, the Southern Ocean is a major contributor to the global ocean CO 2 sink, accounting for ∼30% of oceanic carbon uptake (Arteaga et al., 2018;DeVries, 2014;Frölicher et al., 2015).A key mechanism that maintains the surface-to-deep CO 2 gradient is the biological pump, which transfers a portion of the photosynthetically produced organic carbon into the ocean interior where it may be stored for hundreds of years (Volk & Hoffert, 1985).
Annually, ∼20% of net primary production (NPP) occurring in the Southern Ocean is exported from surface waters (Arteaga et al., 2018;Schlitzer, 2002).NPP and carbon export are elevated over the Antarctic continental shelf, near the (Sub)Antarctic islands, and in the marginal ice zone (MIZ), with the Ross and Weddell Seas emerging as productivity hotspots (Arrigo et al., 2008a(Arrigo et al., , 2008b;;Pollard et al., 2009;Schlitzer, 2002).Here, carbon production and export are thought to be enhanced because of local increases in the iron (and at times, silicic acid) supply and/or stratification (and thus higher light exposure) driven by seasonal ice melt (e.g., Death et al., 2014;Lannuzel et al., 2016).While experimental analyses, geochemical measurements, satellite data, and biogeochemical models have been used to investigate carbon export in the Southern Ocean (e.g., Fan et al., 2020;Hirawake et al., 2011;Kerkar et al., 2020), there remains a need for additional estimates based on in situ observations.Furthermore, to better predict how the Southern Ocean's biological pump and CO 2 sink may change requires an improved understanding of the drivers of carbon export at large scales (Henley et al., 2020).
Phytoplankton δ 15 N (often approximated by the δ 15 N of SPM; δ 15 N SPM ) is set by the δ 15 N of the nitrogenous nutrients supporting growth (e.g., nitrate vs. ammonium; Treibergs et al., 2014), as well as the kinetic isotopic fractionation occurring during N assimilation (e.g., Granger et al., 2004;Needoba et al., 2004).For example, phytoplankton preferentially consume 14 N-bearing nitrate, causing the residual pool to become progressively enriched in 15 N as nitrate consumption proceeds (Mariotti et al., 1981;Sigman et al., 1999).In the summertime Southern Ocean, variations in surface δ 15 N SPM have been attributed to isotope fractionation during nitrate consumption, with a strong negative correlation observed between surface nitrate concentration and δ 15 N SPM (Altabet & Francois, 1994;Lourey et al., 2003;Smart et al., 2020).However, the δ 15 N of phytoplankton (and thus SPM) is also influenced by other N nutrients that support production, with the assimilation of recycled N (e.g., ammonium) typically yielding a lower δ 15 N SPM than the consumption of nitrate supplied from depth (Altabet, 1988;Fawcett et al., 2011Fawcett et al., , 2014;;Treibergs et al., 2014).Because of the different isotopic signals of nitrate versus ammonium assimilation, δ 15 N SPM can be used to distinguish "new" versus "regenerated" N uptake by phytoplankton (e.g., Fawcett et al., 2011Fawcett et al., , 2014;;Lourey et al., 2003;Van Oostende et al., 2017).Moreover, since the rate of new N uptake by phytoplankton must be balanced by the downward flux of organic matter ("export production"), distinguishing nitrate-from ammonium-fueled NPP using δ 15 N SPM provides a means of estimating carbon export potential (Dugdale & Goering, 1967;Eppley & Peterson, 1979;Fawcett et al., 2011).

Global Biogeochemical Cycles
STIRNIMANN ET AL. 10.1029/2023GB007808 3 of 31 measured environmental variables to predict δ 13 C SPM and δ 15 N SPM for the entire Southern Ocean.The authors found that δ 13 C SPM and δ 15 N SPM generally decreased southwards, coincident with changes in sea surface temperature (SST) and N nutrient availability.Additionally, δ 13 C SPM and δ 15 N SPM increased near Antarctica, the (Sub) Antarctic islands, and within the MIZ, as has been observed by others (Espinasse et al., 2019;Smart et al., 2020;Trull et al., 2008).While this meta-analysis represents a significant improvement in Southern Ocean isoscape coverage, particularly for regions where sample collection has been limited, more direct observations are needed to fill sampling gaps and validate isoscape models (St John Glew et al., 2021).In addition, because SPM includes diverse living and dead autotrophic and heterotrophic material, measurements of δ 13 C SPM and δ 15 N SPM that are not accompanied by phytoplankton assemblage data may yield a limited view of the biogeochemical functioning of a system (Falkowski, 1991).Another shortcoming of δ 15 N SPM isoscapes in particular is that they often do not consider variations in the δ 15 N of the nitrate supply as a potential driver of δ 15 N SPM variability (cf.Van Oostende et al., 2017).However, the δ 15 N of the nitrate source to surface waters (e.g., Subantarctic Mode Water; SAMW) can change by >2‰ across the Southern Ocean (Fripiat et al., 2021;Rafter et al., 2013), such that for the same amount of nitrate consumption, and without varying any other environmental parameter, phytoplankton biomass δ 15 N (and thus, δ 15 N SPM ) could also differ by >2‰.
Here, we present circum-Antarctic δ 13 C SPM and δ 15 N SPM isoscapes and phytoplankton community composition data collected during the summer 2016/17 Antarctic Circumnavigation Expedition (ACE).The ACE cruise provided a unique opportunity to collect samples across all sectors of the Southern Ocean during a single season.As such, a major advantage of our dataset is its high spatial resolution, which means that very little interpolation is required (Brault et al., 2018).Below, we discuss possible drivers of isotopic variability among and within the Southern Ocean's hydrographic zones and, by applying a two-endmember isotope mixing model to our δ 15 N SPM data, constrain phytoplankton reliance on new versus recycled N, from which we can estimate relative summertime carbon export potential.
Seawater samples were collected from the ship's underway system (∼4.5 m intake) every one to three hours while the ship was steaming (330 stations) and from Niskin bottles triggered between the surface and 200 m during conductivity-temperature-depth (CTD) hydrocast deployments (34 stations).In total, we collected samples at 142, 132, and 90 stations during Leg1, Leg2, and Leg3, respectively.
At the underway stations, samples for phosphate ( ), and nitrite (  NO2 − ) concentrations were collected in sample-rinsed 50 mL Falcon tubes that were immediately frozen at −20°C.Samples for ammonium (NH 4 + ) concentrations were collected in 50 mL "aged" high-density polyethylene (HDPE) bottles (Smith et al., 2022) and analyzed on board within 24 hr of collection (Leg2 and Leg3) or frozen and measured on land (Leg1).Nutrient samples were also collected at the CTD stations (Hassler & Ellwood, 2020;Janssen et al., 2020) and iron concentrations were measured for samples collected at 17 CTD stations during Leg1 and Leg2 (Janssen et al., 2020).
To capture the bulk SPM, duplicate 2L seawater aliquots were collected at all the underway and all but two of the CTD stations in sample-rinsed opaque HDPE bottles, then filtered through pre-combusted (450°C for 8 hr) glass fiber filters (GF-75s; 0.3 μm pore size).Filters were stored in pre-combusted foil envelopes at −80°C until analysis.

10.1029/2023GB007808
4 of 31 Samples for micro-phytoplankton (20-200 μm) taxonomy were collected at 83 underway stations.Seawater was filtered through a 20 μm nylon mesh with the volume filtered varying according to the concentration of cells in the water (range of 4.5-60 L).The mesh filters were suspended in 5 mL of 0.2 μm-filtered seawater, fixed via the addition of 5 μL of 25% glutaraldehyde, and stored at 4°C in the dark.
The SPM filters were dried for 24 hr at 40°C, with cored subsamples then packaged into tin capsules that were analyzed for carbon (C) and N content and δ 13 C and δ 15 N using a Flash 2000 elemental analyzer coupled to a Delta V Plus isotope ratio mass spectrometer (IRMS) (detection limit of 2 μg C and 1 μg N, precision of ±0.2‰ for δ 13 C and δ 15 N).Three in-house standards calibrated against certified reference materials were included in each IRMS run along with numerous blanks (unused pre-combusted filter + tin capsule).Particulate organic C and N (POC and PON) concentrations were calculated by normalizing C and N content to seawater volume filtered.The station-specific δ 13 C SPM and δ 15 N SPM reported hereafter are the POC-and PON concentration-weighted averages of duplicate samples.
Phytoplankton taxonomic identification involved gently homogenizing each preserved sample, then transferring 0.08 mL onto two clean microscope slides.Cells with intact chloroplasts (i.e., alive at the time of sampling) were counted at 630× magnification using a Zeiss Axioscope A1 light microscope; all live cells on both slides were counted.The remaining volume of preserved sample was treated with 10% hydrochloric acid and 37% hydrogen peroxide to remove carbonate particles and organic matter, respectively.Permanent slides were prepared by pipetting distilled-water-cleaned material onto acid-washed coverslips that were air-dried overnight, then mounted onto glass slides using Naphrax® mountant (refractive index of 1.7).The permanent slides were examined using a Zeiss Axioscope A1 light microscope equipped with differential interference contrast at 1,000× magnification under oil immersion.Stubs were also prepared from the cleaned material for examination using a JEOL JSM-7001F field emission scanning electron microscope to visualize the morphological features not evident under the light microscope.

Supporting Data Products
Satellite data were used to supplement the ship-based surface hydrography measurements (Text S1 in Supporting Information S1).Daily high-resolution SST (GHRSST) Level-4 data on a global 0.054° grid from the Operational Sea Surface Temperature and Ice Analysis database (https://podaac.jpl.nasa.gov/dataset/UKMO-L4HRfnd-GLOB-OSTIA;UKMO, 2005) were used to determine the positions of the major oceanic fronts following Orsi et al. (1995): The Subtropical Zone (STZ) north of the Subtropical Front (STF) is characterized by an SST of 20°C-10°C.Between the STF and Subantarctic Front (SAF) is the Subantarctic Zone (SAZ) with an SST of 10°C-6°C, while the Polar Frontal Zone (PFZ) between the SAF and Antarctic Polar Front (APF) has an SST of 6°C-4°C.The PFZ is characterized by high physical variability and is often considered the transition zone between Antarctic and Subantarctic waters (Bowie et al., 2011;Orsi et al., 1995); here, we take the APF as the boundary between these two major regions.South of the APF is the Southern Antarctic Circumpolar Current Zone (SACCZ), which remains ice-free year-round.The SACCZ has an SST of 4°C-0.7°Cand is bounded to the south by the Southern ACC Front (SACCF), the southern edge of the ACC core (Orsi et al., 1995).South of the SACCF is the southern Boundary Zone (sBZ), which experiences seasonal sea-ice cover (Squire, 2022).We group all stations south of the SACCF, including in the Antarctic Continental Zone, into the sBZ.

An Isotopic Framework for Assessing Phytoplankton Dependence on Nitrate Versus Ammonium
Since δ 15 N SPM is strongly influenced by the N nutrients consumed by phytoplankton (e.g., Fawcett et al., 2011;Lourey et al., 2003) and given that phytoplankton reliance on new N is quantitatively related to export production (Dugdale & Goering, 1967;Eppley & Peterson, 1979), our SPM data can be used to estimate relative carbon export potential across the summertime Southern Ocean.
While the 0.3 μm filters likely trapped some non-phytoplankton material, including bacteria, we assume that the surface SPM was dominated by actively growing phytoplankton (e.g., Popp et al., 1999;Trull & Armand, 2001), a notion supported by the measured POC:PON ratios (Section 3.2).As such, δ 15 N SPM can be described by a two-endmember isotope mixing model (Fawcett et al., 2011) as: where   15 NSPM new and   15 NSPM RN are the δ 15 N values of phytoplankton biomass produced from the assimilation of new and regenerated N, respectively, and f new and f RN are the fraction of phytoplankton growth fueled by new and regenerated N (with f new + f RN = 1).The fraction of new production can thus be computed as: should be low given the isotope effects associated with its production (Checkley & Miller, 1989;Macko et al., 1986;Möbius, 2013;Silfer et al., 1992), its short residence time in the surface layer in summer (days; Smith et al., 2022), and the fact that the flux of low-δ 15 N remineralized  NH4 + should be fairly high (Smith et al., 2022).We estimate   15 N NH 4 + as: where ε regen is the isotope effect associated with  NH4 + regeneration, set here to 1.5‰.While ε regen has been estimated under steady state conditions to be ∼3‰ (Möbius, 2013), we use a lower value because (a) processes other than regeneration are coincidentally acting on the SPM, violating the steady state assumption and altering δ 15 N SPM ; (b) while the primary fate of regenerated  NH4 + is consumption by phytoplankton, some NH 4 + will have accumulated over a variable time period and as such, is unlikely to have a δ 15 N equal to that of the NH 4 + instantaneously produced from the in situ SPM; and (c) (Hoch et al., 1992;Liu et al., 2013;Pennock et al., 1996) (Granger et al., 2004;Mariotti et al., 1981).The effect of this fractionation can be characterized using the Rayleigh model, which describes a unidirectional reaction (i.e.,  NO3 − assimilation into SPM) that proceeds with a constant isotope effect (ɛ assim ) under conditions where the reactant  NO3 − is neither resupplied (e.g., by mixing) nor lost by any mechanism other than phytoplankton assimilation

Global Biogeochemical Cycles
STIRNIMANN ET AL.
10.1029/2023GB007808 7 of 31 (Mariotti et al., 1981).In the Southern Ocean in summer, the N isotope dynamics of  NO3 − assimilation are reasonably well described by the Rayleigh model since  NO3 − is supplied to the surface mainly during winter mixing, then assimilated in spring and summer following surface-layer stratification (Sigman et al., 1999).
The Rayleigh model describes the isotopic evolution of the reactant pool (δ 15 N reactant ; Equation 4a), the instantaneously generated product pool (δ 15 N instantaneous ; Equation 4b), and the accumulated product pool (δ 15 N accumulated ; Equation 4c) (Mariotti et al., 1981)  ).The δ 15 N instantaneous is the SPM produced from  NO3 − at each moment of its consumption, while the δ 15 N accumulated is the sum of all the SPM produced from  NO3 − since consumption began (Mariotti et al., 1981).While some phytoplankton biomass accumulates in surface waters during the growth season (e.g., Swart et al., 2015;van Leeuwe et al., 2020), a significant fraction will also be exported (indeed, we estimate that 60 ± 40% of the SPM produced since the start of the growth season was exported prior to our sampling; Text S2 in Supporting Information S1 We note that if   15 NSPM new is approximated by either the instantaneous or the accumulated product equations (Equations 4b and 4c), our definition of   15 NSPM new (Equation 6) will still return the correct value of δ 15 N instantaneous or δ 15 N accumulated (Figure S2 in Supporting Information S1).
In applying Equation 6to our dataset, we assume that  NO3 − source is the  NO3 − that was present in the mixed layer directly following winter convection, which can be approximated by the  NO3 − measured between the base of the summer mixed layer and the maximum depth of the winter mixed layer; we refer to the waters between the winter and summer mixed layer depths (MLDs) as W-S.We use the biogeochemical-Southern Ocean State Estimate (Mazloff et al., 2010;Verdy & Mazloff, 2017)  While estimates of ɛ assim for the summertime Southern Ocean have been observed to vary considerably (DiFiore et al., 2010;Fripiat et al., 2019;Karsh et al., 2003;Lourey et al., 2003;Sigman et al., 1999), DiFiore et al. (2010) showed a strong linear correlation between the summer MLD and ɛ assim (Equation 8).We use this relationship to derive ɛ assim at each station before calculating a median ɛ assim for each hydrographic zone: Via the approach outlined above, we can estimate   15 NSPM RN and   15 NSPM new , which we then incorporate into Equation 2 to estimate f new from our SPM data.

Statistical Analyses
We used R for statistical analysis of our data (R Core Team, 2018) and Python (van Rossum, 1995) and Ocean Data View (Schlitzer, 2021) for data visualization.Results are reported as mean ± 1 standard deviation,  median Q3

Q1
, where Q 1 = first quartile and Q 3 = third quartile, or median ± interquartile range (IQR), where Shapiro-Wilk tests, quantile-quantile plot analysis, and Levene's tests were performed to assess the normality and homogeneity of variance of data distributions.Data were sub-grouped according to geographical region (hydrographic zones = STZ, SAZ, PFZ, SACCZ, and sBZ; oceans = Subantarctic and Antarctic).While parametric assumptions were satisfied when data were grouped as Subantarctic or Antarctic, the equal variance assumption was violated when data were sub-grouped into the five hydrographic zones.Non-parametric analyses were thus used to test for differences among zones.Parametric analyses included analysis of variance and t-tests, while non-parametric analyses included the Kruskal-Wallis test and multiple pairwise-Wilcoxon test.When multiple statistical tests were performed on a single dataset, the probability values were adjusted using the Bonferroni correction (Chen et al., 2017).Hereafter, the probability values obtained from the parametric and non-parametric tests are denoted as p-value(s) and p-value(s)*, respectively.
The lm R-function was used for linear regression model analysis of the surface SPM data, with the p-value and R 2 reported for each model.When the assumption of equal variance was violated, we applied a weighted least squares model and report the p-value WLS and  R 2 WLS .Spearman's rank correlation (ρ) was used as a non-parametric measure of how well the relationship between two variables was described by a monotonic function.A Principal Component Analysis (PCA) was performed to investigate how various parameters (e.g., SST, nutrient concentrations, δ 15 N SPM ) influenced data variance and distribution, with clustering based on observational similarity.Biodiversity and evenness were determined for the taxonomic counts using the Shannon-Wiener index (H′).

POC and PON Concentrations
The POC and PON concentrations showed similar trends along the transect, ranging from 1.9 to 70.7 μM (  7.4 9.9 5.8 μM) and from 0.3 to 7.7 μM (  1.0 open STZ south of South Africa and in the sBZ near Siple Island (30.0-70.7 μM and 4.0-7.7 μM, respectively), with elevated concentrations also observed near Kerguelen and Bouvet Islands, the Mertz Glacier, the mouth of the Ross Sea, the WAP, and off South America.The POC:PON ratio ranged from 3.5 to 25.1 (median of  7.2 7.8 6.7 ; Figure 3c), with median ratios in the Subantarctic (  7.4 8.0 6.8 ) that were significant different from those in the Antarctic (  6.9 7.4 6.5 ; p-value <0.01).

Chlorophyll-a Concentrations and Phytoplankton Community Composition From Pigment Data
Total Chl-a concentrations ranged from 0.01 to 3.7 mg m −3 (median of  0.2 0.4 0.1 mg m −3 ) for the entire transect and were significantly higher (p-value* <0.001) in the sBZ and the STZ ( 10.1029/2023GB007808 13 of 31 >0.05;Table 3 and Figure 5a).Low Chl-a was recorded in the open ocean (0.01-0.3 mg m −3 ) while elevated concentrations (>3 mg m −3 ) occurred near Heard Island, Siple Island, and South America.Elevated Chl-a (1-2 mg m −3 ) was also measured at the mouth of the Ross Sea and near the Mertz Glacier.
Total micro-phytoplankton abundances ranged from 2 to 12,056 cells L −1 (average of 1,773 ± 2,393 cells L −1 and median of  880 2022 217 cells L −1 for the transect), with 119 species of diatoms, 21 species of dinoflagellates, and 2 species of silicoflagellates identified.Diatom abundances ranged from 0 to 11,972 cells L −1 , with diatoms dominating at most stations and accounting for an average of 82.5 ± 30.6% of the micro-phytoplankton (Figure 6a).Maximum diatom abundances occurred between South Sandwich and Bouvet Islands, with high abundances also observed near the PEIs, Kerguelen, and Heard Islands (Figures 6b and 7a).Diatom abundances were particularly low (0-200 cells L −1 ) in the open ocean over the Indian archipelago, at the end of Leg1, and for most of Leg2.Dinoflagellate abundances ranged from 0 to 762 cells L −1 , with dinoflagellates accounting for 17.1 ± 30.7% of the micro-phytoplankton (Figure 6c).Near the end of Leg1 and east of South America, the dinoflagellate contribution increased to 70%-100%, with abundances of >200 cells L −1 (and a maximum abundance near Tasmania of 762 cells L −1 ), while for much of Leg1 and Leg2, dinoflagellate abundances were ≤20 cells L −1 .Silicoflagellates were rare, averaging 4 ± 23% of the total micro-phytoplankton and reaching a maximum abundance near the PEIs of 55.6 cells L −1 (data not shown).In terms of micro-phytoplankton species, Fragilariopsis kerguelensis (Figure S9a in Supporting Information S1) dominated the diatom communities (average of 56 ± 28% of total diatoms), reaching particularly high abundances near the PEIs and Kerguelen Island (1,000-5,556 cells L −1 ; Figures 7a-7c).By contrast, the high diatom abundances near Heard Island (three stations) were due to Odontella weissflogii and Eucampia antarctica var.antarctica (Figure S9c in Supporting Information S1) (1,895 and 1,519 cell L −1 , respectively), while the high abundances between the South Sandwich and Bouvet Islands were largely due to Chaetoceros atlanticus (3,264 ± 2,212 cells L −1 , ∼40%) and F. kerguelensis (1,861 ± 1,254 cells L −1 , ∼20%).
Micro-phytoplankton species richness was higher in the SAZ, PFZ, and SACCZ (108, 101, and 72 species, respectively) than in the STZ and sBZ (9 and 8 species), noting that of the 83 stations sampled for micro-phytoplankton, there were 5 in the STZ, 33 in the SAZ, 26 in the PFZ, 18 in the SACCZ, and one in the sBZ (Table S1 in Supporting Information S1).Excluding the sBZ (one station only), the most diverse micro-phytoplankton communities occurred in the PFZ (H′ of  1.2 1.4 0.9 ) and SACCZ (H′ of  1.0 1.5 0.7 ), followed by the SAZ (H′ of  0.9 1.2 0.6 ) and STZ (H′ of  0.7 0.9 0.6 ).

Estimates of f new
For the 269 observations to which we applied our isotope model (Text S2 in Supporting Information S1), f new ranged from −7.0 to 35.2, with 54 values ∉ (0, 1).Most of these values occurred in the STZ and near or   4).Values of f new > 0.7 were estimated near the PEIs, at the mouth of the Ross Sea, at Scott and Siple Islands, between the South Sandwich and Bouvet Islands, and downstream of South America, with f new > 0.9 near Kerguelen and Heard and to the east

Principal Component Analysis
The first and the second principal components (PC1 and PC2) together explained 52.8% of the variance, with the data clustering according to hydrographic zone (Figure 9).Observations in the STZ and SAZ were associated with higher values of SST, δ 13 C SPM , δ 15 N SPM ,   15 NSPM new , POC:PON, δ 15 N of the NO 3 − source (W-S and 200-300 m; δ 15 N source ), and with a higher relative contribution to the Chl-a biomass of nano-and pico-phytoplankton (F-nano and F-pico, respectively).The SACCZ and sBZ stations were negatively associated with these variables.The PCA revealed a generally positive relationship for the SAZ and PFZ with pCO 2 and a slight negative relationship with f new .Data from the SACCZ and sBZ were associated with higher concentrations of  SiO 4− 4 , PO4 3− , NH4 + , and  NO3 − , and higher relative contributions of micro-phytoplankton (F-micro) to the Chl-a biomass.Observations at the extreme latitudes (a few from the STZ and many from the sBZ) were associated with higher concentrations of POC, PON, and Chl-a, high values of δ 15 N SPM ,   15 NSPM new , fCO 2 , and pH, and low pCO 2 .Many observations from the sBZ and some from the SACCZ were slightly positively correlated with f new .Data from stations near the islands and continents did not show particular patterns, except for stations at Siple Island, which were positively associated with fCO 2 , pH, δ 15 N SPM ,   15 NSPM new , and Chl-a, POC, and PON biomass, with a slight positive relationship with f new and the contribution of micro-phytoplankton to Chl-a (F-micro).

Spatial Variability in δ 13 C SPM
During photosynthesis, phytoplankton preferentially fix 12 C-bearing CO 2 , producing biomass (SPM) that is lower in δ 13 C than the CO 2 substrate (O' Leary, 1981).The remineralization of this SPM returns low-δ 13 C CO 2 to the dissolved inorganic carbon (DIC) pool (e.g., McCorkle et al., 1985) such that upwelling of deeper waters lowers the δ 13 C of surface DIC (Gruber et al., 1999).The extent to which phytoplankton discriminate against 13 CO 2 is strongly driven by their growth rate and the ambient pCO 2 (François et al., 1993;Goericke, 1994;Laws et al., 1995).For the same initial pCO 2 , a higher growth rate yields less isotopic fractionation and thus a higher δ 13 C SPM , while for phytoplankton growing at the same rate, higher surface-water pCO 2 is associated with higher fractionation and a lower δ 13 C SPM (Gruber et al., 1999;Rau et al., 1991).The δ 13 C of the DIC available to phytoplankton will also affect δ 13 C SPM as it sets the isotopic baseline for the system (e.g., Tamelander et al., 2009).
We observed a general northward increase in δ 13 C SPM , of ∼12‰ between the Antarctic and the STZ, with a higher median (±IQR) δ 13 C SPM in the Subantarctic than the Antarctic (−25.2 ± 2.1‰ vs. −28.4± 2.0‰; Figures 4a, 4c and 4e; Figure S15a in Supporting Information S1).This meridional trend has been reported previously and attributed to a northward decrease in isotope fractionation during photosynthesis driven by a decline in surface pCO 2 and higher phytoplankton growth rates (Arteaga et al., 2018;Espinasse et al., 2019;François et al., 1993;Popp et al., 1999).Across the Subantarctic, δ 13 C SPM was negatively correlated with pCO 2 (ρ = −0.40;Figures 10a and 10b) and (relatedly) positively correlated with pH, fCO 2 , and SST (ρ = 0.50, 0.53, 0.37; p-values <0.05; Figure S17 in Supporting Information S1), which confirms that pCO 2 was a dominant driver of Subantarctic δ 13 C SPM .However, we also observed strong deviations from the expected pCO 2 -driven latitudinal trend in δ 13 C SPM .For instance, δ 13 C SPM reached a minimum in the southern PFZ and northern SACCZ (55-60°S; median of −29.3 ± 3.2‰, reaching −33.7‰).Furthermore, incidences of anomalously high δ 13 C SPM relative to expectations from pCO 2 were apparent near and/or downstream of some of the islands (e.g., the PEIs) and off South America, while in the SACCZ and sBZ where pCO 2 was the lowest, we observed anomalously low values of δ 13 C SPM (Figures 2f  and 3d).
The higher δ 13 C SPM near the Subantarctic islands and continents could be due to the coincidence of low pCO 2 and the composition and growth rates of the in situ phytoplankton.During fast-growing blooms, large diatoms have been observed to have a higher δ 13 C SPM than smaller cells (Fry & Wainright, 1991), with Nakatsuka et al. (1992) reporting a positive correlation between phytoplankton growth rate and δ 13 C SPM .However, Berg et al. (2011) observed a low δ 13 C SPM for diatoms growing rapidly following iron fertilization in the open PFZ, with δ 13 C SPM increasing later as the post-bloom phytoplankton community became more diverse.In our study, phytoplankton biomass at Heard Island was over six-times higher than over the Atlantic archipelago and the rest of the Indian archipelago (Figure 5a).High phytoplankton growth rates at Heard Island, stimulated by an island mass effect-related input of nutrients (e.g., Jena, 2016) (Figures 2a-2e), may have reduced net isotope fractionation, yielding substantially higher values of δ 13 C SPM (  − 19.7 −19.6 −20.7 ‰) than at the Atlantic archipelago (  − 29.1 −28.5 −29.8 ‰) and over the rest of the Indian archipelago (  − 25.0 −24.4 −25.9 ‰).The δ 13 C SPM differences among the islands were unlikely due only to pCO 2 and growth rate, however.The δ 13 C SPM at the Indian archipelago (excluding Heard Island) was higher than at the Atlantic archipelago, despite both regions having similar pCO 2 and phytoplankton biomass.Micro-phytoplankton (presumably mainly diatoms) were the main contributors to total Chl-a at the Atlantic archipelago (73 ± 10%; Figure 5) while at the Indian archipelago, the micro-phytoplankton contribution was lower, 49 ± 15% and 31 ± 6% at the PEIs and Crozet (increasing to 94 ± 3% at Heard Island).Two distinct diatom communities were present at the Indian archipelago, one dominated by F. kerguelensis (most of the archipelago) and the second dominated by O. weissflogii and E. antarctica var.antarctica (Heard Island).
Assuming that iron was not limiting near the islands (e.g., Schallenberg et al., 2018), the distinct phytoplankton communities of the Atlantic and Indian archipelagos (excluding Heard Island) may be analogous to the two assemblages described by Berg et al. (2011).In that study, rapidly growing Chaetoceros spp.dominated the assemblage shortly after iron addition, with a δ 13 C SPM that was slightly higher than prior to fertilization.The phytoplankton community then evolved into a more diverse assemblage of smaller cells including haptophytes.This change coincided with a further ∼1.5‰rise in δ 13 C SPM , attributed to a higher ratio of PEPc to RuBisCO ), and ammonium (  NH4 + ), the partial pressure and air-sea flux of CO 2 (pCO 2 and fCO 2 ), the concentrations of particulate organic carbon and nitrogen (POC and PON), their molar ratio (POC:PON) and isotopes (δ 13 C SPM and δ 15 N SPM ), total chlorophyll-a (Chl-a), and the contributions of micro-, nano-, and pico-phytoplankton to Chl-a (F-micro, F-nano, F-pico) determined from the High Performance Liquid Chromatography data.Modeled parameters are also included: the winter and summer mixed layer depths (MLD winter and MLD summer ), the concentration and δ 15 N of the NO 10.1029/2023GB007808 20 of 31 activity and thus a decline in net isotopic fractionation.Our data may be similarly explained-we measured a higher δ 13 C SPM in waters dominated by a smaller, more diverse phytoplankton community (the Indian archipelago excluding Heard Island) and a lower δ 13 C SPM coincident with the dominance of large diatoms including Chaetoceros spp.(Atlantic archipelago).
As outlined above, Berg et al. (2011) reported a lower δ 13 C SPM for large, iron-replete diatoms than for the more mixed and smaller phytoplankton cells that succeeded them.This dynamic may explain the anomalously low δ 13 C SPM at low pCO 2 observed near the Mertz Glacier, Ross Sea mouth, and Siple and Peter 1 st Islands (red ellipse in Figures 10c and 10d).Here, biomass was high (Figures 3a-3b, 5 3a-3b , and 10d) and the phytoplankton community comprised near-exclusively large diatoms (mostly F. kerguelensis; Figure 7).However, the rapid growth rates typically associated with high biomass production occur with relatively low isotopic fractionation (Laws et al., 1995), yielding higher values of δ 13 C SPM , as at Heard Island (Figures 5 and 6).The δ 13 C SPM difference between Heard Island and the Mertz Glacier/Ross Sea/Siple/Peter 1 st Island region may thus involve the rate of CO 2 resupply to surface waters rather than being predominantly due to growth rate.At Heard Island, CO 2 depletion by phytoplankton may have been faster than its rate of resupply, causing δ 13 C SPM to increase due to 13 C-enrichment of the in situ CO 2 (Deuser, 1970;Villinski et al., 2008).Such 13 C-enrichment is typically negligible because rapid equilibration of CO 2 and HCO 3 − essentially renders the entire (large) DIC pool available to phytoplankton (Tortell et al., 1997(Tortell et al., , 2008)).However, if phytoplankton growth is particularly rapid, as is possible near Heard Island where conditions are extremely favorable (Mongin et al., 2008(Mongin et al., , 2009)), CO 2 can be so quickly depleted as to raise its δ 13 C (Deuser, 1970).In contrast, the Mertz Glacier/Ross Sea/Siple/Peter 1 st Island region was characterized by the highest fCO 2 of the Antarctic Ocean (>10 mg m −2 h −1 from atmosphere to ocean; Figure S5 in Supporting Information S1), likely induced by high biological activity (Tagliabue & Arrigo, 2016) and intense CO 2 dissolution driven by low SSTs and enhanced ice melt (Fransson et al., 2011;Tortell et al., 2012).The resultant high rate of CO 2 resupply would negate any 13 C-enrichment of the in situ CO 2 and thus, the SPM (Freeman & Hayes, 1992;Rau et al., 1989).
The anomalously high-pCO 2 and high-δ 13 C SPM (yellow ellipse in Figures 10c and 10d) occurred far downstream of Bouvet and Heard Islands in waters that were influenced by the APF.Phytoplankton biomass was low and comprised ∼50% nano-phytoplankton (mostly haptophytes; Figure 5).We suggest that these phytoplankton were mainly a post-bloom, iron-limited community expressing a higher ratio of PEPc to RuBisCO activity, which would have caused lower net isotope fractionation and thus a higher δ 13 C SPM (Descolas-Gros & Fontugne, 1985, 1990).
To statistically validate the influence of phytoplankton community composition, alongside that of pCO 2 , on the spatial distribution of δ 13 C SPM , we conducted a Generalized Additive Model analysis (see Text S7 in Supporting Information S1).The results confirm that community composition strongly influences surface δ 13 C SPM across the Southern Ocean, with elevated fractions of nano-phytoplankton clearly associated with a higher δ 13 C SPM and higher proportions of micro-phytoplankton corresponding to a lower δ 13 C SPM.

Spatial Variability in δ 15 N SPM
In our dataset, δ 15 N SPM decreased by ∼3‰ from the STZ (2.1 ± 5.4‰ [median ± IQR]) to the SAZ and PFZ where it reached a minimum (−0.1 ± 3.1‰ and −0.8 ± 2.2‰, respectively), before increasing again into the Antarctic (SACCZ = 0.3 ± 2.8‰ and sBZ = 0.6 ± 1.3‰; Table 2; Figures 3e, 4b, 4d, and 4f).For the entire transect, δ 15 N SPM was positively correlated with POC, PON, and Chl-a (ρ = 0.48, 0.44, and 0.45; p-value <0.001), consistent with the expectation that higher biomass accompanies higher rates of productivity and greater  NO3 − utilization, the latter yielding higher values of δ 15 N SPM (Altabet & Francois, 1994).Generally, the latitudinal trends in δ 15 N SPM agree well with previous observations (e.g., Espinasse et al., 2019;St John Glew et al., 2021).Using a Bayesian hierarchical spatial model applied to data collected over 50 years, St John Glew et al. (2021) determined that SST and MLD were good predictors of δ 15 N SPM in the Southern Ocean; however, we observe a positive relationship between δ 15 N SPM and SST only in the Subantarctic (ρ = 0.37, p-value < 0.001; Figure S17 in Supporting Information S1) and a positive correlation with MLD only in the Antarctic (ρ = 0.37, p-value < 0.001; Figure S18 in Supporting Information S1).The inconsistency of our findings with those of St John Glew et al. (2021) highlights our lack of mechanistic understanding of the synergistic drivers of δ 15 N SPM across the Southern Ocean.
The δ 15 N of phytoplankton biomass depends on the δ 15 N of the N sources consumed, the extent of consumption of those N sources, and the isotope effect(s) expressed during consumption (Altabet & Francois, 2001;Fawcett et al., 2011;Granger et al., 2004;Sigman et al., 1999).The δ 15 N of subsurface  NO3 − (i.e.,   15 NNO 3 − (source) ) is influenced by exchange with underlying deep-water  NO3 − and by N-cycle processes occurring in the surface and subsurface during water mass circulation (Rafter et al., 2013;Sigman et al., 2000).South of the SACCF,

Global Biogeochemical Cycles
STIRNIMANN ET AL.

10.1029/2023GB007808
22 of 31 lower circumpolar deep water with a δ 15 N of 4.8‰ is the ultimate source of  NO3 − to the mixed layer, while north of the SACCF, UCDW supplies  NO3 − with a δ 15 N of ∼5‰ (Fripiat et al., 2019;Rafter et al., 2013;Sigman et al., 2000;Smart et al., 2015).Partial consumption of  NO3 − in the northward-flowing Antarctic Surface Water that derives from UCDW raises its δ 15 N, such that the Antarctic Intermediate Water and SAMW that form in the PFZ and SAZ are higher in   15 NNO 3 − than UCDW (5.7‰ and >6.3‰, respectively; Rafter et al., 2013;Sigman et al., 1999).As these water masses subduct and flow northwards, they become the sub-mixed layer  NO3 − source across much of the Subantarctic.At the same time, surface processes continue to modify subsurface  NO3 − as sinking particles recording the δ 15 N of mixed-layer N cycling are remineralized below the mixed layer (Rafter et al., 2013;Sigman et al., 1999).
Our estimates of f new indicate that the degree of  NO3 − consumption increased poleward (Figure 8b).Such a trend could be related to latitudinal changes in nutrient concentrations, light and MLD, and/or phytoplankton community composition.Indeed, f new was (weakly) positively correlated with the relative contributions of micro-phytoplankton to Chl-a, likely due to the dominance of the micro-phytoplankton by diatoms (Figure 5), which are  NO3 − specialists.Indeed, diatoms have been shown to take up  NO3 − faster than other phytoplankton groups at comparable substrate concentrations (Eppley et al., 1969;Hildebrand & Dahlin, 2000;Paasche et al., 1984), often achieving higher  NO3 − uptake and reduction rates than required for growth, especially in cold,  NO3 − -rich environments (Lomas & Glibert, 1999, 2000).f new was negatively correlated with the contribution to Chl-a of nano-phytoplankton, which typically show a higher affinity for  NH4 + than micro-phytoplankton (Probyn, 1985;Stirnimann et al., 2021;Wafar et al., 2004).The apparent preference of nano-phytoplankton for  NH4 + over  NO3 − can be explained by their higher surface-area-to-volume ratio, which makes them more competitive for scarce nutrients than the larger diatoms (Marañón, 2015).
Of our 269 measurements of δ 15 N SPM , 54 yielded estimates of f new that fell outside the range of 0-1 (gray symbols in Figures S19b and S19c in Supporting Information S1).These samples were mainly collected near the islands, the continents, and at the mouth of the Ross Sea (Figure S14 in Supporting Information S1).Here, processes such as bathymetrically induced mixing, ice melt, and/or terrestrial N inputs may have altered the mixed-layer  NO3 − and/or  NH4 + pools (e.g., Cavagna et al., 2015;Fripiat et al., 2014;Shatova et al., 2016) , such as terrestrially derived N, which can be very high in δ 15 N (Erskine et al., 1998;Wainright et al., 1998), and/or non-assimilation N-cycle processes such as bacterial decomposition of SPM, which raises its δ 15 N (Möbius, 2013).Since our sampling was conducted during a period of active phytoplankton growth when SPM production would have exceeded its degradation, we suggest that the consumption of external N accounts for most of the anomalous δ 15 N SPM values.We measured high concentrations of  NH4 + near the islands and Antarctica (Figure 2d), which may have been partly supplied via terrestrial run-off and/or melting sea-ice and glaciers (Fripiat et al., 2014;Otero et al., 2018).Land-derived  NH4 + is generally high in δ 15 N due to trophic enrichment and/or volatilization of ammonia gas from guano and other organic matter deposited by sea birds and mammals (i.e., the δ 15 N of guano can be >12‰; Mizutani et al., 1986;Wainright et al., 1998).Phytoplankton consumption of this high   15 N NH + 4 could yield higher values of δ 15 N SPM than is possible from the assimilation of subsurface  NO3 − .
The high δ 15 N SPM at the mouth of the Ross Sea and near Siple Island (Figure 3e) may have been influenced by the release of nutrients from melting sea-ice and/or coastal glaciers.Surface  NH4 + concentrations were relatively high almost everywhere during Leg2 (  0.9 1.3 0.5 μM), particularly between the Mertz Glacier and Scott Island and downstream of Peter 1 st Island (>1.5 μM) (Figure 2d).Remineralization of organic N produced by ice algae and excretion by zooplankton feeding under the sea-ice produce  NH4 + and other reduced N forms that accumulate in and under the ice from winter to early summer (Fripiat et al., 2014(Fripiat et al., , 2017;;Louw et al., 2022;Roukaerts et al., 2016).This  NH4 + appears to be relatively low in δ 15 N (<0‰; Fripiat et al., 2014), as would be expected given the N isotope effects associated with its production (Macko et al., 1986;Möbius, 2013;Silfer et al., 1992).However, there may be times when co-occurring  NH4 + assimilation, which occurs with little to no fractionation (Hoch et al., 1992;Liu et al., 2013;Pennock et al., 1996), and  NH4 + oxidation, which is associated with a large isotope effect (14-19‰; Casciotti et al., 2003), cause the δ 15 N of the sea-ice  NH4 + pool to rise significantly.In contrast to  NH4 + , algal  NO3 − consumption is a fractionating process, which in the spring/summer sea-ice yields a relatively low-concentration NO 3 − pool that can be very high in δ 15 N (Fripiat et al., 2014;Roukaerts et al., 2016).When the ice melts, the N species are released into the water column, along with high concentrations of iron (Lannuzel et al., 2016).This iron stimulates phytoplankton  NO3 − consumption (Timmermans et al., 1994), including of the high-δ 15 N  NO3 − released from the ice, resulting in the production of higher δ 15 N SPM than predicted by a Rayleigh model initialized with the underlying   15 NNO 3 − (source) .

Estimating Relative Carbon Export Potential From δ 15 N SPM
In this study, we aimed to assess Southern Ocean carbon export potential by estimating the fraction of organic matter produced via the uptake of new versus regenerated nutrients (i.e., subsurface  NO3 − vs. recycled  NH + 4 ; Dugdale & Goering, 1967;Eppley & Peterson, 1979).The relative strength of the biological pump can be approximated by the f-ratio, assuming that the surface ocean is at steady state over an annual cycle (Eppley & Peterson, 1979).Here, we derive f new , a measure of the f-ratio, using a two-endmember isotope mixing model, with the δ 15 N of the consumed  NO3 − (i.e.,   15 NSPM new ) estimated via the Rayleigh model.However, instead of assuming that the δ 15 N SPM generated from the assimilation of subsurface  NO3 − can be approximated by the accumulated (or instantaneous) product equation, we account for the fact that some portion of the SPM would have been exported from the surface layer between the start of the growth season and our sampling.Above, we have discussed cases in which our isotope mixing model failed (e.g., due to an input of high-δ 15 N allochthonous N).Below, we discuss our estimates of f new and the implications for carbon export potential in the case where the δ 15 N SPM data are well described by the model (i.e., 68% of the observations).
In a mass balance sense, our data suggest that the Southern Ocean exports >40% of the carbon produced in the summer (f new =  41 60 32 %), with minimal difference between the Subantarctic and Antarctic Oceans (39 ± 20% and 43 ± 32% [median ± IQR], respectively; Figure 8; Table 4).These values generally agree with existing estimates of the f-ratio for the summertime Southern Ocean (which averages ∼50%; e.g., Joubert et al., 2011;Le Moigne et al., 2016;Mdutyana et al., 2020;Prakash et al., 2015;Sambrotto & Mace, 2000;Schlitzer, 2002).Examining our data by zone shows that f new decreased by ∼20% from the high to the lower latitudes (Figure 8b).In the sBZ, 60 ± 21% of phytoplankton carbon was potentially exported, reaching ∼80% near the sea-ice (e.g., at the Ross Sea mouth; Table 4 and Figure 8b).f new was lowest in the STZ, although ∼30% of NPP was still potentially exported.Across the SAZ, PFZ, and SACCZ, f new was highly variable (ranging between 7% and ∼98%), yet similar median values were estimated (39%-44%).High values of f new (>65%) were associated with stations near and downstream of many of the islands, with ∼75% of NPP potentially exported near the PEIs and >80% downstream of Kerguelen Island.Lower values of f new (<50%) were determined for open ocean stations that were distant from the islands or covered with sea-ice (e.g., the open STZ, south of Tasmania, and the SACCZ during the latter half of Leg2; Figure 8a).The apparent latitudinal trend in f new may thus be driven principally by processes that occur at 10.1029/2023GB007808 24 of 31 the regional scale, such as the alleviation of iron limitation that leads to proportionally higher  NO3 − dependence near the islands, at the hydrographic fronts, and off the continental shelves.For instance, the large region between the South Sandwich and Bouvet Islands that was characterized by a high f new (73 ± 33%) may have biased the mean SACCZ estimate upwards, while most of the truly open ocean stations where regenerated production dominated were located at lower latitudes, potentially biasing those f new estimates downwards.

𝐴𝐴
NO3 − uptake by Southern Ocean phytoplankton is typically enhanced by an increase in light, a stable upper water column, and an input of limiting nutrients (i.e., iron and/or  SiO 4− 4 ) via upwelling, terrestrial run-off, and/ or melting sea-ice and glaciers (Cochlan, 2008 and references therein).The island mass effect is generally associated with upwelling, input of land-derived nutrients, and retention and stabilization of surface waters over shallow plateaus, which drive localized increases in phytoplankton biomass and nutrient and CO 2 drawdown (e.g., Holmes et al., 2019;Planquette et al., 2011;Schallenberg et al., 2018).In the Ross Sea, large and persistent polynyas reduce sea-ice cover, increasing stratification and light (Arrigo et al., 2015;et al., 2000), as well as iron availability (Alderkamp et al., 2012).Relief from iron limitation enhances  NO3 − reductase activity in phytoplankton, particularly diatoms, allowing for higher  NO3 − uptake and increased rates of biomass production (Blain et al., 2007;De Baar et al., 1997;Karsh et al., 2003;Timmermans et al., 1994).By contrast, low iron concentrations limit  NO3 − uptake and favor the growth of small phytoplankton reliant on regenerated N (Martin et al., 1990;Sunda & Hardison, 2007).In our dataset, the open ocean regions were characterized by low iron concentrations (<0.05 nmol kg −1 ; Figure 2e) and supported low biomass and a higher proportion of nano-phytoplankton, predominantly haptophytes (Figure 5d; Text S6 in Supporting Information S1), which may be more efficient than diatoms at utilizing bacterially regenerated iron (Fourquez et al., 2022).By contrast, near the South Sandwich Islands, PEIs, and Heard Island, iron concentrations were higher (≥0.2 nmol kg −1 ), biomass and micro-phytoplankton contributions to total Chl-a were elevated, large diatoms were abundant (Figures 5e and 7), and local minima were observed in pCO 2 (Figure 2f).Interestingly, downstream of Crozet Island (50-60°E), the plankton assemblage was dominated by nano-phytoplankton (mainly haptophytes) and diatom abundance was low; here, regenerated N fueled more than half of NPP, limiting carbon export potential.
Across the Southern Ocean, f new was slightly positively correlated with  SiO 4− 4 (ρ = +0.25,p-value <0.001; Figure S16 in Supporting Information S1).Since the availability of  SiO 4− 4 controls the distribution of diatoms (Hoffmann et al., 2008) that typically dominate new production (Dugdale et al., 1995;Egge & Aksnes, 1992), and given that the fraction of micro-phytoplankton was strongly positively correlated with  SiO 4− 4 (ρ = +0.70,p-value <0.001), the relationship of f new to  SiO 4− 4 is not unexpected.Consistently, f new was also positively correlated with micro-phytoplankton biomass, weakly in the Subantarctic and strongly in the Antarctic (ρ = +0.28 and + 0.56, respectively; Figures S17, S18, and S20 in Supporting Information S1).By contrast, low values of f new occurred where nano-phytoplankton (mainly haptophytes) dominated, typically in open ocean waters.Here, low iron availability limits diatom growth and  NO3 − assimilation (Timmermans et al., 2004), which leads to the development of regenerated N-fueled systems dominated by nano-phytoplankton that are more competitive for scarce nutrients (Fourquez et al., 2022;Hare et al., 2007).

Conclusions
In this circum-Antarctic study, we sought to (a) better understand the drivers of δ 13 C SPM and δ 15 N SPM isoscape variability across the summertime Southern Ocean and (b) quantify carbon export potential from measurements of δ 15 N SPM .In general, the latitudinal gradients in δ 13 C SPM and δ 15 N SPM were consistent with previous studies, with clear differences observed between the Subantarctic and Antarctic Oceans.As expected, δ 13 C SPM was highly dependent on seawater pCO 2 , with phytoplankton community composition helping to explain values of δ 13 C SPM that deviated from the expected relationship with pCO 2 .The observed trends in δ 15 N SPM were driven by the δ 15 N of the N nutrients and the extent of phytoplankton reliance on new versus regenerated N. Phytoplankton community composition also played a role as some taxa specialize in  NO3 − assimilation while others favor recycled  NH4 + consumption.In contrast to previous isoscape studies, this work highlights the considerable influence of phytoplankton community dynamics on δ 13 C SPM and δ 15 N SPM .For instance, higher δ 13 C SPM was linked to higher haptophyte abundance, and is thus potentially explained by their carboxylation of  HCO3 − using PEPc.Conversely, higher δ 15 N SPM usually coincided with a greater contribution to the phytoplankton assemblage of diatoms, which specialize in the assimilation of  NO3 − .We employed a novel approach to determine relative carbon export potential using a two-endmember isotope mixing model that incorporated the Rayleigh equations for isotope fractionation during  NO3 − assimilation by phytoplankton.Our derived values of f new are consistent with previous summertime measurements of the f-ratio, which validates our approach, although we note that different methods for estimating the f-ratio integrate over different time scales, making a direct comparison difficult.For example, f-ratios determined from 15 N-tracer-based measurements of  NO3 − and  NH4 + uptake (e.g., Joubert et al., 2011;Mdutyana et al., 2020) reflect the conditions of the water column at the time of sampling while our δ 15 N SPM -based approach integrates over weeks to months, and geochemical estimates (e.g., profiling float-based measurements of  NO3 − consumption; Johnson et al., 2017) can integrate over the annual cycle.We estimate that on average, >40% of the carbon produced in Southern Ocean surface waters was potentially exported, with a higher percentage near the (Sub)Antarctic islands and in regions of melting sea-ice, induced by an increase in phytoplankton growth rates and  NO3 − uptake, likely in response to iron input.The onset of iron limitation should drive phytoplankton to consume proportionally more  NH4 + as the growth season progresses (Cochlan, 2008;Smith et al., 2022).As such, f new is likely to have declined following our sampling.By the same logic, however, f new would have been considerably higher at the beginning of the growth season, with much of the organic N then produced already exported by the time of our sampling.We thus conclude that our estimates of f new provide a reasonable approximation of the fraction of NPP exported from the Southern Ocean surface during the summer growth period.
Bulk SPM is easily sampled and δ 13 C SPM and δ 15 N SPM are already widely analyzed.Here we show how δ 15 N SPM can be used to infer relative carbon export potential, an approach that could be applied to other ocean regions.We find that   15 NNO 3 − (source) is a critical variable that should be measured alongside δ 15 N SPM ; indeed,   15 NNO 3 − (source) varies by >3‰ across the Southern Ocean, meaning that 3‰ of the variability in δ 15 N SPM could be explained by  NO3 − δ 15 N alone, before changes in new versus regenerated N uptake or phytoplankton community dynamics need to be invoked.As such, we recommend that studies using measurements of δ 15 N SPM (e.g., trophic analyses) consider the influence of   15 NNO 3 − (source) .

Figure 6 .
Figure 6.Surface distribution of the percent contribution (a, c; %) to the micro-phytoplankton and total abundances (cells L −1 ; b, d) of (a)-(b) diatoms and (c)-(d) dinoflagellates.The islands visited during the cruise are indicated on panel (a): PEIs = Prince Edward Islands, Crzt = Crozet Island, Krgln = Kerguelen Island, Hrd = Heard Island, DR = Diego Ramírez Island, SG = South Georgia, and SS = South Sandwich Islands.

Figure 7 .
Figure 7. (a) Abundance (cells L −1 ) and (b)-(c) contribution (%) of different micro-phytoplankton species at each station (b) and in each hydrographic zone (c): STZ = Subtropical Zone, SAZ = Subantarctic Zone, PFZ = Polar Frontal Zone, SACCZ = Southern Antarctic Circumpolar Current Zone, and sBZ = southern Boundary Zone.Vertical arrows indicate islands, with PEIs = Prince Edward Islands, and horizontal lines denote the cruise legs.The percent contribution in panel (c) was estimated from the average abundance of the different micro-phytoplankton species counted in each zone.

Figure 8 .
Figure 8.(a) Fraction of phytoplankton growth fueled by new NO 3 − (f new ) and (b) boxplots of f new for each hydrographic zone: STZ = Subtropical Zone (yellow), SAZ = Subantarctic Zone (orange), PFZ = Polar Frontal Zone (red), SACCZ = Southern Antarctic Circumpolar Current Zone (dark blue), sBZ = southern Boundary Zone (light blue).The islands visited during the cruise are indicated on panel (a): PEIs = Prince Edward Islands, Crzt = Crozet Island, Krgln = Kerguelen Island, Hrd = Heard Island, DR = Diego Ramírez Island, SG = South Georgia, and SS = South Sandwich Islands.The boxplots (panel (b)) show medians (thick horizontal lines), interquartile range (IQR) (50% of the data; box within first and third quartiles), and whiskers (1.5 times the IQR).Numbers at the top of the boxplots are the median NO 3− assimilation isotope effects derived for each zone (ɛ assim ; Equation8) while the numbers at the bottom are the median δ 15 N computed for regenerated N (   15 NSPM RN ; Equation3).

Figure 9 .
Figure 9. Principal Component Analysis of our dataset (246 observations).Surface parameters considered include observations of sea surface temperature, the concentrations of nitrate (  NO3 − ), nitrite (  NO2 − ), phosphate (  PO4 3− ), silicic acid (  SiO 4− 4 ), and ammonium (  NH4 +), the partial pressure and air-sea flux of CO 2 (pCO 2 and fCO 2 ), the concentrations of particulate organic carbon and nitrogen (POC and PON), their molar ratio (POC:PON) and isotopes (δ 13 C SPM and δ 15 N SPM ), total chlorophyll-a (Chl-a), and the contributions of micro-, nano-, and pico-phytoplankton to Chl-a (F-micro, F-nano, F-pico) determined from the High Performance Liquid Chromatography data.Modeled parameters are also included: the winter and summer mixed layer depths (MLD winter and MLD summer ), the concentration and δ 15 N of the NO 3 − source (NO 3 Figure 9. Principal Component Analysis of our dataset (246 observations).Surface parameters considered include observations of sea surface temperature, the concentrations of nitrate (  NO3 − ), nitrite (  NO2 − ), phosphate (  PO4 3− ), silicic acid (  SiO 4− 4 ), and ammonium (  NH4 +), the partial pressure and air-sea flux of CO 2 (pCO 2 and fCO 2 ), the concentrations of particulate organic carbon and nitrogen (POC and PON), their molar ratio (POC:PON) and isotopes (δ 13 C SPM and δ 15 N SPM ), total chlorophyll-a (Chl-a), and the contributions of micro-, nano-, and pico-phytoplankton to Chl-a (F-micro, F-nano, F-pico) determined from the High Performance Liquid Chromatography data.Modeled parameters are also included: the winter and summer mixed layer depths (MLD winter and MLD summer ), the concentration and δ 15 N of the NO 3 − source (NO 3 − source and   15 NNO 3 − source ), the theoretical δ 15 N of phytoplankton biomass produced from the assimilation of NO 3 − (   15 NSPM new ), and the fraction of phytoplankton growth fueled by NO 3 − (f new ).The symbol colors indicate the hydrographic zones: STZ = Subtropical Zone (yellow), SAZ = Subantarctic Zone (orange), PFZ = Polar Frontal Zone (red), SACCZ = Southern Antarctic Circumpolar Current Zone (dark blue), and sBZ = southern Boundary Zone (light blue), while the symbol shapes indicate stations from the open ocean (circles) versus near the islands and continents (diamonds).

Table 2
Medians (±IQR) of the Surface Particulate Organic Carbon (POC) and Nitrogen (PON) Concentrations (μM), POC:PON Ratios, and δ 13 C SPM and δ 15 N SPM (‰) for the Different Regions of the Southern Ocean Sampled During ACE: "Subantarctic" Includes the STZ = Subtropical Zone, SAZ = Subantarctic Zone, PFZ = Polar Frontal Zone, While "Antarctic" Includes the SACCZ = Southern Antarctic Circumpolar Current Zone and sBZ = Southern Boundary Zone Note.In parentheses are the numbers of observations used to calculate the medians."Subantarctic" Includes the STZ = Subtropical Zone, SAZ = Subantarctic Zone, and PFZ = Polar Frontal Zone, While 'Antarctic" Includes the SACCZ = Southern Antarctic Circumpolar Current Zone and sBZ = Southern Boundary Zone.

Table 4
Inputs to and Outputs of the Rayleigh and Isotope Mixing Models , violating the Rayleigh model assumption of a closed system and rendering f new invalid.Mathematically,   15 NSPM new must be higher than the measured δ 15 N SPM to produce an acceptable f new ; if could underpin the low values of f new .At the same time, a large increase in