Seasonality of phytoplankton growth limitation by iron and manganese in subantarctic waters

Phytoplankton play a major role in the marine carbon cycle by driving the transfer of carbon dioxide from the atmosphere into the ocean through photosynthesis. This process is part of the biological carbon pump, and its strength varies within and between oceanic regions (Lenton et al., 2013; Deppeler and Davidson, 2017). The Southern Ocean is comprised of several biogeochemical regions with contrasting hydrographic and nutrient conditions: the subantarctic zone, the polar front zone, the Antarctic zone, and the seasonal sea ice zone, each delimited by fronts (Orsi et al., 1995). South of the subtropical front, phytoplankton growth is limited mainly by very low concentrations of iron (Fe), a micronutrient essential to many photosynthetic processes (Boyd et al., 2000; Twining and Baines, 2013). Other factors may also limit phytoplankton growth, such as low light and temperature, or low silicic acid concentrations north of the polar front (Boyd, 2002; Bowie et al., 2009; Strzepek et al., 2012). These limiting factors directly impact the strength of the biological carbon pump and hence need to be identified to project changes to the oceanic carbon cycle during the Anthropocene.

The subantarctic zone, the northernmost region of the Southern Ocean, is an important area in terms of biology and carbon uptake. Indeed, this region sustains the strongest carbon sink of all the Southern Ocean biogeochemical regions (Lenton et al., 2013). In the subantarctic zone, phytoplankton communities transition from coccolithophore-dominated populations further north towards diatom-dominated communities in polar waters (Trull et al., 2001). In terms of size fractions, coccolithophores and diatoms overlap within the nanoeukaryote size class (2–20 µm), which also includes nanoflagellates and dinoflagellates (Kopczyńska et al., 2001). Subantarctic microeukaryotes (>20 µm) are usually composed of large diatoms and dinoflagellates (Kopczyńska et al., 2007; Petrou et al., 2011; Eriksen et al., 2018). Smaller phytoplankton, termed picoeukaryotes (0.2–2 µm), are composed of cyanobacteria and very small taxa of the previously mentioned groups, and constitute an important portion of the subantarctic phytoplankton community (Kopczyńska et al., 2001; Cassar et al., 2015). However, high grazing pressure, especially from microzooplankton and heterotrophic nanoflagellates, keeps pico- and nanoeukaryote abundance relatively low with little seasonal variability (Kopczyńska et al., 2001; Pearce et al., 2011; Deppeler and Davidson, 2017).

In addition to Fe, other factors may (co-)limit subantarctic phytoplankton growth. Silicic acid was demonstrated to co-limit subantarctic diatom growth in late summer/autumn (Hutchins et al., 2001). In addition, manganese (Mn) was observed to co-limit phytoplankton growth in both coastal (Wu et al., 2019) and open ocean waters (Browning et al., 2021; Balaguer et al., 2022) of the Southern Ocean in austral summer. Mn is an essential micronutrient used in the oxygen-evolving complex for the water-splitting reaction of photosynthesis (Armstrong, 2008) and in some phytoplankton taxa in the defence against reactive oxygen species (Peers and Price, 2004; Wolfe-Simon et al., 2006). Saito et al. (2008) classified Mn co-limitation as a type II “Biochemical substitution co-limitation,” in which two elements are expected to substitute for each other for the same active site of an enzyme, for example, Fe and Mn within the superoxide dismutase enzyme. However, due to the essential role of Mn during photosynthesis (Armstrong, 2008), Mn may also limit phytoplankton growth as an independent nutrient. Importantly, phytoplankton Mn requirements may vary depending on Fe conditions (Peers and Price, 2004), and if Fe limitation increases the cellular requirement for Mn, Mn (co-)limitation may be expected in Southern Ocean phytoplankton limited by Fe (Boyd et al., 2000; Deppeler and Davidson, 2017). New biogeochemical modelling suggested that extensive Mn limitation may occur in subantarctic waters during the austral spring, resulting from sub-optimal light levels combined with higher Fe availability compared to Mn (Hawco et al., 2022). So far, the study of Fe-Mn co-limitation of phytoplankton growth has been restricted to a few polar Southern Ocean sites (Buma et al., 1991; Scharek et al., 1997; Sedwick et al., 2000; Wu et al., 2019; Browning et al., 2021; Balaguer et al., 2022; Burns et al., 2023), with only Browning et al. (2021) investigating potential co-limitation within subantarctic waters.

South of Tasmania (Australia), concentrations of both dissolved Fe (dFe) and dissolved Mn (dMn) have been observed to remain low in surface subantarctic waters (≤0.3 nM; Sedwick et al., 1997; Bowie et al., 2009; Latour et al., 2021). In this region, Fe, like Mn, may be delivered to the ocean through atmospheric inputs from Tasmania and mainland Australia or sedimentary inputs from the Tasmanian shelf (Bowie et al., 2009; Perron et al., 2020). Southward advection of subtropical waters has also been observed to supply Fe- and Mn-enriched waters to the subantarctic zone (Sedwick et al., 2008; Bowie et al., 2009). To date, no studies have investigated Fe and Mn co-limitation in the Australian sector of the Southern Ocean. Additionally, to our knowledge, there has been no prior study of the seasonality of Mn or Fe-Mn (co-)limitation in any subantarctic region. This study presents the results of three shipboard incubation experiments performed in subantarctic waters in the Australian sector of the Southern Ocean examining Fe-Mn co-limitation in austral spring, summer, and autumn. We test the hypothesis that both Fe and Mn will limit phytoplankton growth in summer and autumn following micronutrient depletion during the growth season.

The bioassay experiments were performed onboard the RV Investigator during three voyages, IN2018_V04 in September 2018 (austral spring), IN2019_V02 in March 2019 (austral autumn) and IN2020_V08 in December 2020 (austral summer). The first experiment was conducted at Process Station 2 (PS2) on the East Australian Current (EAC) voyage IN2018_V04 (45.44°S, 153.31°E), and the following two experiments at the Southern Ocean Time Series (SOTS) station (46.80°S, 141.884°E; Figure 1). Both sites are within the subantarctic zone to the southeast and southwest of Tasmania, respectively (Bowie et al., 2011).

Prior to starting the bioassays, initial oceanographic settings were studied at each site through the deployment of a 36-bottle conductivity-temperature-depth (CTD) rosette, also measuring oxygen, fluorescence and photosynthetically active radiation (PAR; Sea-Bird Electronics sensors: SBE4C, SBE3 T, SBE9plus, SBE43, FLBBNTU, QCP–2300 HP). Seawater used for the bioassay experiments was collected at 15 m depth for the spring and autumn experiments and at 20 m for the summer experiment using a polyurethane powder-coated aluminium rosette, or “trace metal rosette” (TMR) (Sea-bird Scientific, USA; Holmes et al., 2020). Samples for macronutrients, flow cytometry and photophysiology analyses were collected from the TMR to characterise the initial phytoplankton communities. Polycarbonate bottles used for the incubations were washed with Neutracon detergent for 48 h and then in 10% HCl for 7 days to remove trace metal contamination. After multiple Milli-Q water rinses, bottles were dried overnight in an ISO Class 5 laminar flow hood before being double-bagged in plastic. Onboard, the bottles were rinsed three times with the incubation seawater and then filled inside an ISO Class 5 containerized clean room. The seawater was unamended (control) or spiked with a solution of Fe, Mn or a combination of both. The Fe and Mn spikes were prepared in 0.01 M Ultrapure HCl using ultrapure salts of FeCl₃ and MnCl₂. For the spring experiment, an Fe(NO₃)₃ solution was used, prepared in 0.30 M nitric acid. Triplicates were used for each treatment, resulting in 12 bottles for 4 treatments, named hereafter: control, +Fe, +Mn, and +FeMn. Concentrations of Fe and Mn were adjusted to reach a final concentration of at least 2 nM, which we considered as nutrient-replete conditions (Browning et al., 2021). The bottles were then incubated in deck-board incubators inside neutral-density mesh bags to reproduce the light penetrating the surface ocean, at approximately 15–20 m: 25.5% of incident irradiance for spring and autumn and 12.4% for the summer experiment. The average daily PAR (± SD, n = 7) intensity within the incubators was 120 ± 95, 72 ± 66 and 97 ± 86 µmol photons m⁻² s⁻¹ for spring, summer and autumn, respectively (described in further detail in Figure S1). Deck-board incubators allowed the algal communities to follow their regular diel light:dark cycles, which were approximately 13:11, 16:8 and 12:12 (hours:hours) for spring, summer and autumn, respectively. The temperature of the incubators was maintained by a continuous flow of seawater, keeping the bottles at the same temperature as the surrounding surface (about 7 m depth) seawater.

Bottles were sampled at day 7 for macronutrients, flow cytometry and photophysiology analyses for each experiment. Flow cytometry samples were fixed using 2% (v/v) glutaraldehyde (Electron-microscope grade, 25%) for phytoplankton samples collected during the second voyage in the autumn of 2019. For the summer 2020 voyage, a mixture of formaldehyde-hexamine (18%:10% v/w) was used to preserve phytoplankton samples (Veldhuis et al., 2001). Due to a technical issue, flow cytometry samples from the spring 2018 voyage were lost. All bacterial samples were fixed using 2% glutaraldehyde (Electron-microscope grade, 25%). All flow cytometry samples were held at 4°C in the dark for 25–30 min after being fixed, then flash-frozen in liquid nitrogen and stored in a −80°C freezer until analysed onshore (Marie et al., 1999).

Following the subsampling, a portion of the remaining seawater was dispensed into 300 mL acid-washed polycarbonate bottles and spiked with 16–20 µCi of sodium ¹⁴C-bicarbonate (NaH¹⁴CO₃; specific activity 1.85 GBq mmol⁻¹; PerkinElmer, USA) and 0.2 nM of an acidified ⁵⁵Fe solution (⁵⁵FeCl₃ in 0.1 M Ultrapure HCl; specific activity 30 MBq mmol⁻¹; PerkinElmer; Ellwood et al., 2020). Bottles were then incubated in the deck-board incubators for another 24 h, under the same conditions as the bioassay experiments. The spiked samples were then filtered sequentially through 20, 2 and 0.2 µm polycarbonate filters (47 mm diameter; Poretics, USA), separated by 200 µm nylon mesh spacers. The filters were washed with Titanium(III) EDTA-citrate reagent for 5 min to dissolve Fe (oxy)hydroxides and remove extracellular particle-bound ferric ions and rinsed three times with 15 mL of 0.2 µm-filtered seawater (Hudson and Morel, 1989). Finally, filters were placed in 20 mL glass vials (Wheaton Industries, USA) and acidified with 200 µL of 1.2 M HCl. These filters were then stored at room temperature for Fe and carbon uptake determination onshore.

Seawater temperature was recorded by the CTD sensor, and practical salinity was measured onboard using a conductive cell (Kawano, 2010). Dissolved macronutrients were also analysed onboard using segmented flow analysis (Rees et al., 2018). In spring, the use of an Fe nitrate solution led to the addition of >18 µM of nitrate. Final nitrate concentrations can be found in the supplementary material (Figure S2). As nitrate concentrations were replete in spring, we do not expect such nitrate amendment to interfere with the interpretation of our results. One silicic acid measurement was removed from the analysis due to an inconsistent result (autumn experiment, in the +Mn treatment). In summer, final silicic acid concentrations were all below the detection limit (<0.2 µM), precluding statistical analyses. Initial dissolved trace metal concentrations (Table 1) were measured through high-resolution inductively coupled plasma mass spectrometry (HR-ICP-MS; Element XR; ThermoScientific) after pre-concentration at the Australian National University (Canberra, Australia) (Ellwood et al., 2019). The detection limit was 0.06 and 0.07 nM for dFe and dMn (n = 3), respectively. Dissolved Fe and Mn concentrations were used to estimate Mn deficiency relative to Fe as Mn* = dMn − dFe/R_Fe:Mn, where R_Fe:Mn (2.67) is the typical Fe:Mn molar ratio of phytoplankton, calculated across a range of taxa (Moore et al., 2013; Browning et al., 2021). Values of Mn* ≥ 0.16 suggest Mn-replete conditions (Browning et al., 2021).

Fast Repetition Rate Fluorometry was used to determine the maximum photochemical efficiency (F_v/F_m) and functional absorption cross section (σ_PSII) of photosystem II (PSII) using a Light-induced Fluorescence Transients Fast Repetition Rate fluorometer (Soliense, USA). After low light (2 µmol photons m⁻² s⁻¹) acclimation for approximately 30 min, samples were exposed to 140 flashes of light every 2.5 μsec (saturation sequence) to saturate PSII and the first stable electron acceptor, Q_A, after which the time interval between flashes was increased exponentially (relaxation sequence) for 90 flashes. F_v/F_m (where F_v = F_m − F_o) was calculated from F_o and F_m, which refer to the minimum and maximum fluorescence, respectively, in the dark-acclimated state. F_v/F_m and σ_PSII were determined from the mean of 200 iterations of the fluorescence induction and relaxation protocol measured at 470 nm. At least 10 acquisitions were measured for each sample and used to calculate the average value of F_v/F_m and σ_PSII.

Flow cytometry samples were analysed at the Menzies Institute for Medical Research (University of Tasmania, Hobart), using an Aurora Cytek flow cytometer (Cytek Biosciences). This instrument can measure particles ranging from 200 nm to at least 60 µm, although the maximum measurable size for this instrument remains unknown. Briefly, frozen samples were thawed at 37°C for 5–10 min before running 500 µL of unstained samples at flow rates of approximately 50 µL min⁻¹, using Milli-Q water as sheath fluid. Violet and blue excitation lights were used to differentiate the main phytoplankton groups through their fluorescence pigments: chlorophyll with red fluorescence and phycoerythrin with orange fluorescence, respectively, against forward scatter (FSC). All scatter and fluorescence parameters were analysed based on values from the integrated area of the excitation peak. Results obtained from both the summer and autumn voyages were analysed using SpectroFlo software. For an overall comparison between the two seasons, phytoplankton communities were divided into three gates: picoeukaryotes, nanoeukaryotes and microeukaryotes, identified on the violet channel (V12, 405 nm excitation, 692 nm emission) against FSC. If the signal from V12 was saturated, we used another excitation wavelength (B7, 488 nm excitation, 661 nm emission). Picocyanobacteria were identified on another fluorescence channel (B4, 488 nm excitation, 581 nm emission) due to the presence of phycoerythrin (Marie et al., 1999). Cell counts per unit volume were determined from the instrument through the known volume analysed. We then used the cell counts to calculate the relative importance of each group in terms of biovolume by comparing their biovolume and abundance, using the following equation from Bach et al. (2018):

where F_pop represents the fraction of biovolume of a specific phytoplankton population (pop) relative to total phytoplankton biovolume and N represents the abundance via cell count of a specific population or all phytoplankton cells (all). Biovolume was calculated using area-integrated size (FSC-A) corrected with a power regression (Selfe, 2022) as follows:

Heterotrophic bacterial counts were performed on thawed fixed samples after the addition of SYBR Green I stain (1,000-fold dilution). Samples were incubated with the stain for 15 min at room temperature in the dark. Then, a 50 µL aliquot of stained sample was run on the instrument at a high flow rate. Bacteria were identified using blue excitation and green fluorescence (B2, 488 nm excitation, 525 nm emission). Cell counts were determined as described above for phytoplankton.

Iron (⁵⁵Fe) uptake and net primary productivity (¹⁴C uptake) were determined by measuring disintegrations per minute on a liquid scintillation counter (PerkinElmer Tri-Carb 2910 TR). Filters were incubated at least 24 h prior to analysis in 10 mL of Ultima Gold liquid scintillation cocktail (PerkinElmer). Daily carbon incorporation rates were estimated following Hoppe et al. (2017). The uptake of ⁵⁵Fe and ¹⁴C were corrected for ambient dFe and dissolved inorganic carbon concentrations.

Statistical analyses were performed in R (R “stats” package; R Core Team, 2020). Datasets were separated per season and initially examined for homogeneity of variance using a Levene’s test, and normality using a Shapiro-Wilk test. Where data were both normally distributed and homoscedastic, significant differences between treatments were investigated using a one-way analysis of variance (ANOVA) with a Tukey’s HSD post hoc test. When the homogeneity or normality was violated, a Kruskal-Wallis test was performed followed by a Wilcoxon signed-rank test where the former result was significant. Significance was accepted at p < 0.05. For the tests on flow cytometry data (population biovolume) and uptake rates (Fe and carbon uptake), each size class was tested separately. When the one-way ANOVA indicated significant differences between two results appearing near-identical, we performed a linear regression model. We compared model fits between a simple linear regression model and a linear mixed effect model using the Akaike information criterion. Lower values for this criterion were observed for the linear model, which we used for two specific analyses (see section on “Iron uptake and net primary productivity”).

Oceanographic conditions differed between the three experiments across temperature, salinity and silicic acid profiles (Figure 2). In spring, the surface ocean was characterized by a deep mixed layer depth (MLD), down to 200 m. Temperature, salinity and silicic acid concentrations were constant within the mixed layer, with values at about 10.5°C, 34.9 and <3 µM, respectively. In summer, stronger stratification was observed, with the MLD reaching just below 100 m. Summer surface temperatures were similar to spring, but cooler below 25 m (approximately 10°C). In summer, salinity was lower than in spring (<34.6). Similarly, summer silicic acid concentrations were low, with 0.86 µM measured in surface waters. In autumn, the MLD reached 100 m, where the temperature was ≥11°C and the salinity close to summer salinity. Silicic acid concentrations were the lowest in autumn, with 0.80 µM measured in surface waters.

Initial dFe and dMn concentrations present in the incubated seawater differed between seasons (Table 1). Both dFe and dMn concentrations were highest during the summer, with intermediate values in spring and lowest concentrations in autumn (Table 1). For all seasons, calculated Mn* values were well above the 0.16 nM deficiency threshold suggested by Browning et al. (2021).

For each experiment, nitrate, phosphate and silicic acid concentrations were measured initially and after 7 days of incubation (Figures 3 and S2). Initial concentrations for all three macronutrients were highest in spring and lowest in autumn. Initial nitrate concentrations ranged from 8.4 µM to 10.8 µM, phosphate concentrations from 0.71 µM to 0.82 µM, and silicic acid concentrations from 0.80 µM to 2.8 µM (Figures 3 and S2). Nitrate, phosphate and silicic acid concentrations decreased over the 7-day incubation, across all seasons and treatments (Figures 3 and S2), although with seasonal variability.

In spring, no significant differences in phosphate and silicic acid concentrations were observed between treatments by day 7 (ANOVA; Figure 3). However, the greatest drawdown of both nutrients was observed under combined +FeMn addition. Final nitrate concentrations could only be compared between the control and +Mn treatments, which were found not to differ significantly (ANOVA; Figure S2). In summer, relative to the control, final nitrate and phosphate concentrations were significantly lower in the treatments where Fe was added (+Fe and +FeMn; p < 0.05, Tukey’s HSD). Final summer silicic acid concentrations were all below the instrument detection limit (0.2 µM). In autumn, no significant differences were observed in either phosphate or silicic acid concentrations between treatments (ANOVA; Figure 3). Autumn final nitrate concentrations were lower in the +FeMn treatment compared to the +Mn addition alone (p < 0.05; Tukey’s HSD; Figure S2).

The photochemical efficiency of PSII (F_v/F_m) differed between treatments and seasons (Figure 4A). In spring, no significant differences in final F_v/F_m values were measured between treatments (ANOVA). In summer, only the treatments with Fe additions (+Fe and +FeMn) maintained F_v/F_m values as high as the initial community, and significantly higher than the control and +Mn treatments (p < 0.05, Tukey’s HSD). In autumn, we measured significantly higher F_v/F_m values in both treatments with Fe additions (+Fe and +FeMn) compared to the +Mn treatment only (p < 0.05, Tukey’s HSD). However, F_v/F_m values measured in both Fe treatments were not significantly higher than the control (p > 0.05, Tukey’s HSD).

The functional absorption cross-section of PSII (σ_PSII) also differed between treatments and seasons (Figure 4B). The initial value was higher in summer compared to spring and autumn. In spring, we observed a significant decrease in σ_PSII only in the +FeMn treatment, compared to the other treatments (p < 0.05, Tukey’s HSD). In summer, both treatments with Fe additions (+Fe and +FeMn) were characterized by a decrease in σ_PSII compared to the control and +Mn treatments (p < 0.05, Tukey’s HSD). In autumn, no significant differences in σ_PSII were observed between treatments (p > 0.05, ANOVA).

Different initial phytoplankton communities were observed between the summer and autumn experiments. In summer, picoeukaryotes dominated the cell counts at the start of the experiment (Table 2). However, nanoeukaryotes dominated community biovolume, as defined by Equation 1 (Figure 5). In autumn, cyanobacteria showed the greater abundance through cell counts, although picoeukaryotes were nearly as abundant (Table 2). In comparison to the summer experiment, the importance of picoeukaryotes and picocyanobacteria in terms of population biovolume was much larger, although nanoeukaryotes still dominated the community biovolume (Figure 5).

After 7 days of incubation, no significant differences in cell counts were observed between treatments in summer and autumn (ANOVA or Kruskal-Wallis test). Similarly, no significant differences in population biovolume were observed across treatments for microeukaryotes. In summer, nanoeukaryotes showed higher population biovolume under +FeMn compared to the +Mn treatment, while picocyanobacteria showed a significant increase in population biovolume under +Mn addition compared to +Fe (p < 0.05, Tukey’s HSD). In autumn, picoeukaryotes were observed to have higher population biovolume under +Mn compared to both Fe treatments (p < 0.05, Tukey’s HSD). Despite an increase in the population biovolume of picocyanobacteria in the +Mn treatment compared to the control (more than doubled), no significant differences were observed between treatments. Between initial conditions and the control treatment there was a large decrease in cell count and population biovolume for picoeukaryotes in summer and both picoeukaryotes and picocyanobacteria in autumn. Such a decrease suggests a loss of viable cells due to grazing, also supported by higher ammonia concentrations measured in summer, and especially in autumn (Figure S3; Mengesha et al., 1998).

Iron uptake rates varied between seasons and size fractions (Figure 6). In spring, the small size class (0.2–2 µm) showed higher Fe uptake rates in the +Fe treatment compared to the control and +FeMn treatments (p < 0.05, Tukey’s HSD). The highest Fe uptake rates were recorded during the spring experiment within nanoeukaryotes (the 2–20 µm size class) and microeukaryotes (>20 µm) characterised by significantly higher Fe uptake rates in the +Fe treatment compared to the control and +Mn treatments (up to over 6-fold higher; p < 0.05, Tukey’s HSD; Figures 6 and S4).

In summer, picoeukaryotes (0.2–2 µm) did not show any significant difference in Fe uptake rates between treatments, while nanoeukaryotes (2–20 µm) showed enhanced Fe uptake rates under +Fe, with a mean value four times higher than in the control (p < 0.05, Tukey’s HSD; Figures 6 and S4). The combined +FeMn addition did not stimulate increased Fe uptake compared to the control (p = 0.06, Tukey’s HSD). Microeukaryotes (>20 µm) had Fe uptake rates up to six times higher in +Fe and +FeMn treatments compared to the control and +Mn treatments (p < 0.05, Tukey’s HSD; Figures 6 and S4).

In autumn, despite a visual trend suggesting an increase in Fe uptake rates under both +Fe and +FeMn additions (up to 2–6-fold; Figure S4), no significant differences in Fe uptake rates were observed across treatments or size classes (p > 0.05, Kruskal-Wallis test). This absence of significant differences is likely due to the small dataset (only 2 data points for the +Fe treatment).

Net primary productivity, measured through ¹⁴C uptake, also varied between seasons and size fractions (Figures 7 and S5). During the spring experiment, no differences in these carbon uptake rates were recorded within picoeukaryotes (0.2–2 µm) or microeukaryotes (>20 µm). Conversely, nanoeukaryotes (2–20 µm) displayed higher carbon uptake rates in the +FeMn treatment compared to the control (p < 0.05, Tukey’s HSD; Figure 7). Due to very similar results in the +Fe treatment, we performed a linear regression model which showed significantly higher carbon uptake rates in both Fe treatments compared to the control (p < 0.05).

In summer, picoeukaryotes (0.2–2 µm) had higher carbon uptake rates in the +Fe treatment compared to the control (p < 0.05, Tukey’s HSD). In addition, both Fe treatments (+Fe and +FeMn) had significantly higher carbon uptake rates than the +Mn treatment (p < 0.05, Tukey’s HSD). Within nanoeukaryotes (2–20 µm), a significant difference between carbon uptake rate was only observed between the +Fe and +Mn treatments (p < 0.05, Tukey’s HSD), with higher carbon uptake rate in the +Fe treatment. Carbon uptake by microeukaryotes (>20 µm) was significantly higher only in the +FeMn treatment compared to the control (p < 0.05, Tukey’s HSD). The carbon uptake rates measured in the +Fe treatment, while elevated, were not significantly different than the control according to the one-way ANOVA (p = 0.05, Tukey’s HSD). Again, to verify this result, a linear regression model was performed, which showed significantly higher carbon uptake rates in both Fe treatments compared to the control (p < 0.05). In autumn, lower net primary productivity was recorded (Figure 7). Despite visual trends suggesting a treatment effect (Figures 7 and S5), no significant differences were observed between treatments or size classes.

Iron to carbon (Fe:C) uptake ratios differed between seasons and treatments, with overall higher ratios measured in autumn (Table 3). In spring, the Fe:C ratio ranged between 16 and 112 µmol mol⁻¹, with the minimum value observed for picoeukaryotes (0.2–2 µm) and the maximum value measured for nanoeukaryotes (2–20 µm). In summer, microeukaryotes (>20 µm) had the lowest Fe:C uptake ratio (2–9 µmol mol⁻¹), while picoeukaryotes (0.2–2 µm) had the highest ratio, up to 33 µmol mol⁻¹. In autumn, we also measured the lowest Fe:C uptake ratio for microeukaryotes (>20 µm), with a minimum of 39 µmol mol⁻¹, while picoeukaryotes showed the highest Fe:C ratio of 914 µmol mol⁻¹.

Spring Fe:C ratios displayed higher values under the +Fe treatment compared to the control in all size classes (p < 0.05, Tukey’s HSD). In addition, nanoeukaryotes were characterised by higher Fe:C ratios in the +Mn treatment compared to the control (p < 0.05, Tukey’s HSD). Summer picoeukaryotes (0.2–2 µm) had no differences in Fe:C uptake ratios between treatments, while the ratios for nano- (2–20 µm) and microeukaryotes (>20 µm) were higher within +Fe and +FeMn treatments compared to the control and +Mn treatments (p < 0.05, Tukey’s HSD). However, higher Fe uptake is expected within phytoplankton cells for an Fe addition treatment (Strzepek et al., 2005). In autumn, no significant differences in Fe:C uptake ratios were observed across treatments or size classes (Kruskal-Wallis test), but again this lack of significance may result from the small sample size.

Contrasting hydrographical conditions may be expected between experiments due to the different location of the spring experiment, conducted southeast of Tasmania at station PS2, relative to the summer and autumn experiments which were performed at the SOTS station, southwest of Tasmania. The intrusion of warmer and saltier waters from the subtropical zone is commonly observed in the northern part of the subantarctic zone near SOTS and can originate from either mixing with waters from the Zeehan Current, or mixing with waters and eddies from the EAC (Bowie et al., 2011). Compared to the SOTS site, waters eastern of Tasmania will be heavily influenced by the EAC (Ridgway, 2007; Bowie et al., 2009). This influence likely explains the difference in salinity observed in spring compared to the two other experiments (Figure 2). However, autonomous seasonal records of phytoplankton communities from the SOTS station revealed no change in community composition due to the input of subtropical waters in the subantarctic zone (Eriksen et al., 2018). Hence, we suggest that the results of the three experiments are comparable, despite the influence of subtropical waters at PS2 during the spring experiment.

Surface water nutrient concentrations (setting our initial conditions) varied seasonally. Higher phosphate, nitrate and silicic acid concentrations were observed at the beginning of the spring experiment, which are characteristic of the early season following winter mixing of surface waters (Rintoul and Trull, 2001). In contrast, macronutrient concentrations were lowest in autumn, likely resulting from biological consumption following the growth season (spring/summer blooms; Boyd, 2002; Bowie et al., 2009). Phosphate and nitrate concentrations decrease during the summer due to biological uptake but are expected to remain higher than limiting levels throughout the year (Hutchins et al., 2001; Rintoul and Trull, 2001). On the other hand, subantarctic silicic acid concentrations severely decrease during the growth season, due to consumption from silicifying phytoplankton (Hutchins et al., 2001; Eriksen et al., 2018). The incubated water of the summer and autumn experiments had very low initial silicic acid concentrations of 0.86 µM and 0.80 µM, respectively, which is sufficiently low to limit the growth of some silicifying phytoplankton (Paasche, 1973; Hutchins et al., 2001; Westwood et al., 2011). Initial trace metal concentrations were highest in summer. To observe higher summer dFe and dMn concentrations compared to the spring experiment is surprising. Usually, higher concentrations are recorded in spring, prior to biological consumption and resulting from i) aerosol depositions coming from proximal land masses (Perron et al., 2020), ii) southern advection of Fe- and Mn-enriched subtropical waters from the EAC (Sedwick et al., 2008; Bowie et al., 2009), and/or iii) replete trace metal levels present after the winter season associated with wind-mixing of Fe- and Mn-enriched subsurface waters (Bowie et al., 2009). However, both aerosol depositions and advection of trace metal-enriched waters may be intensified seasonally due to the dust storm season and the Australian fire season (spring/summer) or the peak season (summer) of the EAC intensity (Ridgway, 2007; Perron et al., 2020; Traill et al., 2022). Such intensification may explain the higher dFe and dMn concentrations observed in summer at the SOTS station.

Initial summer and autumn phytoplankton biovolume was dominated by nanoeukaryotes, as previously observed in this subantarctic region (Kopczyńska et al., 2007; Pearce et al., 2011). However, the importance of picoeukaryotes and picocyanobacteria was much greater in the water incubated during the autumn experiment (Table 2 and Figure 5), characteristic of a regenerative system (Kopczyńska et al., 2001). Phytoplankton communities dominated by small taxa are heavily controlled by microzooplankton and nano-heterotrophs, as previously shown (Pearce et al., 2011). We observed this top-down control on small cells in both summer and autumn (Figure 5), associated with elevated ammonia concentrations (Figure S3), characteristic of enhanced zooplankton activity (Mengesha et al., 1998). In all seasons, in-situ light limitation of phytoplankton growth is expected due to deep mixed layer depths (≥100 m; Figure 2). Indeed, Rintoul and Trull (2001) previously observed that a mixed layer depth of 75–100 m, common in summer for this region, was deep enough to light-limit phytoplankton growth. Despite this limitation, initial physiological measurements indicated that the bulk phytoplankton communities were relatively healthy (F_v/F_m ≥ 0.5) in all three seasons (Figure 4), suggesting that Fe concentrations were high enough to support the photosynthetic requirements of these light-limited communities (Strzepek et al., 2012). However, our data indicated various degrees of Fe limitation resulting from increased light conditions during the incubation experiments.

Phytoplankton growth in subantarctic waters is usually assumed to be Fe-limited (Boyd et al., 1999; Sedwick et al., 1999; Hutchins et al., 2001; Petrou et al., 2011), especially in spring and summer (Boyd, 2002). However, our experiments demonstrate that the degree to which Fe limits phytoplankton photophysiology and productivity can differ across seasons. For example, moderate signals of Fe limitation were observed in spring, with only changes in σ_PSII and carbon uptake rates, compared to the summer experiment. The high spring F_v/F_m values measured in all treatments (Figure 4A) suggested efficient light utilization in PSII (Greene et al., 1992; Hopkinson and Barbeau, 2008). Unfortunately, the lack of flow cytometry data for the spring experiment precludes an assessment of the spring phytoplankton community composition or how it may have evolved with Fe additions. Previous reports showed this subantarctic region to be characterized by a succession from large diatoms in spring towards smaller, weakly silicified diatoms in summer/autumn (Eriksen et al., 2018). In spring, we observed that most of the Fe and carbon uptake was associated with nano- and microeukaryotes (Figures 6 and 7). Hence, the spring experiment may have taken place during the transition from large diatoms (>20 µm) toward smaller (2–20 µm) and more weakly silicified diatoms in response to decreasing ambient dFe and silicic acid concentrations (Eriksen et al., 2018).

The strongest signal of Fe limitation was observed during the summer experiment. This observation was supported by i) the larger drawdown of phosphate and nitrate concentrations in both treatments where Fe was added (Figures 3A and S2), ii) the increase in F_v/F_m and decrease in σ_PSII with Fe additions (Figure 4), and iii) the increase in carbon uptake rates observed with Fe additions (Figure 7). These results suggest that the addition of Fe alleviated growth limitation (Greene et al., 1992; Petrou et al., 2011) and agree with the previous suggestion of Fe limitation in summer (Boyd, 2002). Although nitrate levels were drawn down greatly by the end of the experiment within both Fe treatments (Figure S2), co-limitation by Fe and silicic acid may more likely occur toward the end of the experiment (Figure 3B) as well as under natural conditions, due to the lower initial silicic acid levels. This interpretation is supported by the final depletion in silicic acid concentrations compared to nitrate levels in all treatments (Figure 3B and S2). Flow cytometry results indicated that nanoeukaryotes (2–20 µm) dominated population biovolume and remained the dominant group throughout the experiment in all treatments (Figure 5). Combined with the high uptake of silicic acid observed in summer (Figure 3B), these results suggest the growth stimulation of relatively small diatoms, within the nanoeukaryote size range (2–20 µm), in agreement with previous results (Eriksen et al., 2018). Despite the overall dominance of nanoeukaryotes in terms of cell numbers, microplankton (>20 µm) dominated primary productivity (Figure 7). Microeukaryotes comprised about 15% of the population biovolume of all size classes (Figure 5), with very low Fe:C ratios (Table 3), implying that large phytoplankton were able to sustain growth and substantial carbon assimilation with very low Fe requirements. Similar carbon uptake rates were recorded within nanoeukaryotes and more surprisingly within picoeukaryotes, despite grazing pressure on the latter (Figures 5 and 7). Relatively higher Fe uptake rates observed in the 0.2–2 µm size class may indicate higher efficiency in Fe uptake from small cells, possibly due to the higher surface area to volume ratio of these organisms (Sunda and Huntsman, 1995; Strzepek et al., 2011). Notably, this size fraction also includes Fe uptake by heterotrophic bacteria, but their contribution to Fe uptake was not determined.

In autumn, Fe limitation was not clearly evident. Photochemical efficiency (F_v/F_m) was higher under Fe additions (Figure 4) only when compared to the +Mn treatment. In contrast to the summer experiment, phosphate and silicic acid were drawn down to a lesser extent in autumn even under Fe additions (Figure 3), indicating that Fe was not the primary limiting factor. Given the low initial silicic acid levels observed (0.80 µM), silicic acid may be the primary variable limiting the growth of silicified organisms (Hutchins et al., 2001; Eriksen et al., 2018) and not dFe concentrations or other macronutrients considering that initial phosphate (0.71 µM) and nitrate (8.4 µM) remained above limiting levels (Sedwick et al., 1999; Rintoul and Trull, 2001). However, the possibility of Fe and silicic acid co-limitation of diatom growth cannot be excluded (Boyd, 2002). The results of a previous subantarctic study suggested a seasonal succession of limiting variables, with both Fe and silicic acid concentrations limiting the growth of heavily silicified diatoms in late summer and autumn (Hutchins et al., 2001; Boyd, 2002). This co-limitation leads to a community shift toward non-silicified organisms such as nanoflagellates and/or lightly silicified diatoms with low Fe requirements (Hutchins et al., 2001; Pearce et al., 2011). In autumn, both picoeukaryotes and picocyanobacteria, most likely Synechococcus sp. (Hutchins et al., 2001; Cassar et al., 2015), were much more important in terms of population biovolume compared to the summer experiment (Figure 5). Previous flow cytometric analyses showed picoeukaryotes can dominate the subantarctic phytoplankton community, with picocyanobacteria representing a significant group, contributing about 20% to total phytoplankton biovolume in mid-late summer (Cassar et al., 2015). Previous bioassays from this region observed little effect of Fe additions on picocyanobacteria growth (Hutchins et al., 2001). Both picoeukaryotes and picocyanobacteria were also heavily grazed during the autumn experiment (Figure 5), likely due to microzooplankton or small heterotrophic flagellates (Pearce et al., 2011). Grazing from microzooplankton and heterotrophic flagellates can remove over 70% of primary production if phytoplankton communities are dominated by nano- and/or picoeukaryotes (Pearce et al., 2011). Relatively high Fe uptake rates were measured in all size classes during the autumn experiment compared to summer (Figure 6), possibly due to an upregulation of Fe acquisition in response to chronic Fe limitation in these late-season phytoplankton communities (Strzepek et al., 2011). In the >20 µm size class, dinoflagellates may have dominated phytoplankton abundance, as silicic acid levels were likely limiting the growth of large diatoms (Eriksen et al., 2018). A more detailed examination of the phytoplankton community composition, such as pigment analyses or microscopy, would be required to confirm this hypothesis. Overall, these results highlight the complexity of identifying nutrient stress conditions from a bulk phytoplankton community dataset, where signals from specific taxonomic groups can be hidden (Suggett et al., 2009). However, our findings provide evidence for a strong seasonality of Fe limitation and a seasonal succession of various phytoplankton groups, associated with their responses to key environmental constraints, particularly dFe and silicic acid concentrations (Eriksen et al., 2018), while signs of Mn limitation were more subtle.

Overall, these seasonal experiments did not show a clear signal of Fe-Mn co-limitation, in comparison to the strong responses observed from Fe additions. This outcome concurred with the high Mn* values calculated for all three seasons (Table 1), fitting within the range of Browning et al. (2021) (0.16–0.31 nM) for which Fe was limiting but not Mn. Still, we noted interesting responses to Mn additions. A first observation of the combined effects of Fe and Mn additions was the decrease in functional absorption cross-section of PSII in spring compared to the control treatment (Figure 4B), a result not previously observed. Only one laboratory study on Fe-Mn co-limitation of Chaetoceros debilis reported σ_PSII values (Pausch et al., 2019), while bioassays rarely report this parameter. Large values for σ_PSII are commonly associated with Fe-stress and subsaturating light conditions (Strzepek et al., 2019). A decrease in σ_PSII resulting from combined Fe and Mn additions suggests that Mn may (co-)influence the antennae size of Southern Ocean phytoplankton, although not alone (Figure 4B). This finding warrants further investigation of the influence of Mn on phytoplankton photophysiology. In addition, this change in σ_PSII was observed concurrently with a drawdown of both phosphate and silicic acid concentrations (Figure 3), suggesting that diatoms may drive this signal.

In addition, we observed an increase in the population biovolume of picocyanobacteria under Mn addition in summer and autumn and picoeukaryotes in autumn (Figure 5). These increases indicate that Mn may have been limiting the growth of some small phytoplankton taxa. The lower bulk F_v/F_m value observed in the +Mn treatment in autumn may support the hypothesis of a large contribution from cyanobacteria, which often have an intrinsically lower F_v/F_m than eukaryotic algae (Campbell et al., 1998; Suggett et al., 2009). However, this parameter is not a reliable indicator of PSII efficiency in cyanobacteria, as they have more flexible electron transport systems (Campbell et al., 1998) and PSII is poorly excited by the wavelength (470 nm) used in this study. We also measured F_v/F_m at 550 nm, a wavelength that better stimulates photochemistry in cyanobacteria, but the signal-to-noise ratio was too low to interpret our data reliably. Cyanobacterial Mn requirements are still poorly understood. For instance, previous laboratory studies of Synechocystis (a freshwater cyanobacteria) showed that dMn concentrations ≤100 nM reduce oxygen evolution capacity and result in the accumulation of partially assembled PSII systems and changes in the organization of photosystem I complexes (Salomon and Keren, 2011). In their most limiting Mn treatment, Salomon and Keren (2011) measured a background dMn concentration of 1.8 nM, which is much higher than what is commonly observed in Southern Ocean open waters. However, oceanic strains may have adapted to lower surrounding dMn concentrations by lowering their Mn requirements. This effect was previously shown in cyanobacteria regarding adaptation to Fe limitation (Ferreira and Straus, 1994). Overall, there is insufficient information on the Mn requirements of subantarctic cyanobacterial strains to predict if the dMn concentration is low enough to become growth-limiting. However, our results provide evidence that Mn may influence the growth of cyanobacteria and picoeukaryotes, particularly late in the growing season when small phytoplankton contribute significantly to total phytoplankton biovolume.

Such subtle signals of Mn (co-)limitation are surprising, considering that Mn was recently observed to influence phytoplankton growth in the Drake Passage (Browning et al., 2021; Balaguer et al., 2022). This points toward spatial variations in Southern Ocean Mn limitation of phytoplankton growth. The shelf inputs from the Australian continent possibly deliver enough Mn to sustain phytoplankton growth. Still, our results suggest that Mn concentrations may be low enough to limit the growth of a subset of the primary producers, influencing subantarctic phytoplankton community composition (Balaguer et al., 2022). The subtle signal of Mn limitation, easily hidden by responses to Fe additions, highlights the need to develop new tools to identify Mn (co-)limitation within subpopulations of the phytoplankton community. This development would allow for the inclusion of a species-specific component within biogeochemical models examining the evolution of the oceanic carbon cycle.

This study aimed to investigate the seasonality of Fe-Mn co-limitation of subantarctic phytoplankton growth. We found that the signal of Mn (co-)limitation observed during these multi-seasonal experiments was masked by the strong seasonality and responses associated with Fe limitation. The spring experiment was characterised by a moderate signal of Fe-Mn co-limitation, associated with the functional absorption cross-section of PSII. In summer, phytoplankton communities were strongly Fe-limited as shown by the high nutrient drawdown and the enhanced photophysiology and carbon uptake rates under Fe additions. In autumn, we suggest that low silicic acid levels limited diatom growth. However, the possibility that silicic acid and Fe were co-limiting diatom growth cannot be excluded. Manganese additions induced subtle community changes and stimulated cyanobacterial and picoeukaryotic population biovolume in summer and autumn, suggesting that some small phytoplankton taxa may be seasonally Mn-limited. In addition, our results show that the Mn (co-)limitation signal may be hard to capture in conventional bioassays, especially when pronounced Fe responses are observed. This study provides evidence for a seasonality in Fe limitation in subantarctic waters as well as indication that Mn control of phytoplankton growth may be associated with specific taxa.

The data supporting the findings of this study are available in the IMAS Metadata Catalogue (https://doi.org/10.25959/BFF2-RC77).

The supplemental files for this article can be found as follows:

Figures S1–S5. DOCX

The authors would like to acknowledge the officers and crew of the RV Investigator (CSIRO Australian Marine National Facility; https://ror.org/01mae9353) for the deployment of all the instruments during the three voyages (IN2018_V04, IN2019_V02 and IN2020_V08), and the hydro-chemistry team who performed the macronutrient analyses onboard. They thank Pamela Barrett and Robin Grun for the collection and analyses of trace metal samples performed for the first two voyages. PL would like to personally thank Aaron Ferderer for his help on the statistical analyses. This manuscript has been published within the Earth and Space Science Open Archive (ESSOAr) preprint server (June 2022; doi.org/10.1002/essoar.10511502.1).

This work was funded through the Antarctic Climate & Ecosystems Cooperative Centre (ACE CRC) and by the Australian Antarctic Program Partnership (AAPP; ASCI000002). PL was supported by the Australian Research Council Special Research Initiative, Australian Centre for Excellence in Antarctic Science (ACEAS; Project Number SR200100008). LTB acknowledges funding from the Australian Research Council through Future Fellowship (FT200100846).

The authors report no competing interests.

Contributed to conception and design: PL, RFS.

Contributed to acquisition of data: PL, RFS, SE, MJE.

Contributed to analysis and interpretation of data: PL, RFS, KW, PvdM, LTB, PWB, TLP, ARB.

Drafted and/or revised the article: PL, RFS, KW, PvdM, SE, LTB, PWB, MJE, TLP, ARB.

Approved the submitted version for publication: PL, RFS, KW, PvdM, SE, LTB, PWB, MJE, TLP, ARB.

How to cite this article: Latour, P, Strzepek, RF, Wuttig, K, van der Merwe, P, Bach, LT, Eggins, S, Boyd, PW, Ellwood, MJ, Pinfold, TL, Bowie, AR. 2023. Seasonality of phytoplankton growth limitation by iron and manganese in subantarctic waters. Elementa: Science of the Anthropocene 11(1). DOI: https://doi.org/10.1525/elementa.2023.00022

Domain Editor-in-Chief: Jody W. Deming, University of Washington, Seattle, WA, USA

Associate Editor: Kevin R. Arrigo, Department of Earth System Science, Stanford University, Stanford, CA, USA

Knowledge Domain: Ocean Science