Data was collected in the Chatham Rise section of the Southern Ocean, located east of Aotearoa New Zealand, as part of the SalpPOOP (‘Salp Particle expOrt and Ocean Production’) voyage during October to November 2018 ( Figure 1 ). We conducted five Lagrangian experiments (hereafter referred to as “cycles”) that each lasted four to eight days (Décima et al., 2023). Our sampling locations were mainly within the STF, in regions that were predominantly influenced by either ST or SA water masses. Our most coastal cycle, termed SA-SC, was impacted by the Southland Current (SC), a diversion of the STF constituted mainly by SA water (Sutton, 2003). Three cycles were sampled in SA waters (SA-SC, SA-1 and SA-2) and two cycles in ST waters (ST-1 and ST-2) ( Supplementary Table 1 , including station locations and characteristics). Two cycles in SA water (SA-SC and SA-1) and one cycle in ST water (ST-2) occurred within blooms of the salp Salpa thompsoni. In each cycle, satellite-tagged marker buoys were used to track the sampled water mass by tethering subsurface drogues centered at 15-m depth to each buoy to follow the mixed layer (ML) in a Lagrangian frame of reference (Landry et al., 2009; Stukel et al., 2015). The mixed layer depth was identified using CTD downcast data and determined as the depth where the change in density (sigma-t) exceeded the 10 m depth density by 0.125 kg m^-3 (Levitus, 1982) and the deep chlorophyll maximum (DCM) depth, when present, was determined by the depth of subsurface peak in chlorophyll a (Chla) fluorescence. Using this Lagrangian approach allowed us to sample the same water mass over time, to give context to the variability observed in our measurements. We used a 24-place 12-L Niskin bottle rosette equipped with a Seabird (SBE 911plus) CTD and profiling Chla fluorometer to obtain sample water at each station and chose six depths spanning the euphotic zone (surface to the 1% incident light level) based on Chla fluorescence measured during the downcast. The rosetteʻs sensors included SBE 3plus, SBE 4, and SBE 43 dual sensors for temperature, salinity, and dissolved oxygen, a Seapoint fluorescence sensor, and a photosynthetically active radiation (PAR) sensor (Biospherical Instruments QCP‐2300L‐HP). At each location, we conducted a single (~0:200), daily depth-resolved profile from the rosette, sampling for each parameter - epifluorescence microscopy (epi), flow cytometry (FCM), FlowCam, size-fractionated Chla and POC samples to quantify microplankton, nanoplankton and picoplankton biomass and abundance throughout the euphotic zone.

POC was estimated by filtering a seawater sample (2.2 L) onto a pre-combusted 25-mm GF/F filter under low vacuum pressure (5 to 7” Hg), using similar depths as CTD casts for FCM, FlowCam and epi. These filters were deep frozen (-80°C) for later analysis. On shore, samples were thawed, acidified with fuming HCl in a desiccator, excess acid removed under vacuum pressure and finally the samples were placed in a drying oven for 24 h at 40°C. The samples were then packaged into a tin capsule and analyzed at the University of California Davis Stable Isotope Facility for C content measured with an elemental analyzer coupled to an isotope ratio mass spectrometer.

Samples for size fractionated Chla were analyzed by first gravity filtering a sample (250 mL) through a 20 µm 47-mm polycarbonate filter, then with low vacuum pressure, filtering sequentially through a 2 µm and a 0.2 µm polycarbonate filter. Filters were then folded, placed in 1.5 mL cryovials and frozen (-80°C) until analysis (Gutiérrez-Rodríguez et al., 2020). Chla and acidified phaeopigment-a concentrations were measured (within 3 months) using ice-cold 90% acetone extraction by spectrofluorometric methods (APHA 10200 H) on a Varian Cary Eclipse fluorescence spectrophotometer (Rice et al., 2012).

Nutrient samples were gravity filtered using a 0.2 µm Acrocap in-line capsule (Pall-Gelman) connected to a Niskin bottle by an acid-rinsed silicon tube into a 50-mL falcon tube that was first triple-rinsed with filtered seawater, filled halfway, capped and sealed with parafilm and stored at -80°C until analysis. Nutrient samples were analyzed for nitrate (NO₃ ⁻), ammonium (NH₄ ⁺), dissolved reactive phosphorus (DRP) and silicic acid (DRSi) concentrations using an Astoria Pacific API300 micro segmented flow analyzer (Astoria‐Pacific, Clackamas, OR, USA). Analyses were completed by the National Institute of Water and Atmospheric Research (NIWA) in Hamilton (New Zealand) following standard procedures (Law et al., 2011).

High oxygen saturation % in the upper euphotic zone during the SA-SC cycle relative to other cycles, along with higher fluorescence seen in Chla relative to other SA cycles, suggest that this cycle had higher autotroph biomass typical of coastal upwelling and/or mixing associated with the SC ( Table 1 ; Figure 2 ). SA-SC, SA-1 and SA-2 displayed traits characteristic of colder, Southern Ocean water masses with lower average temperature as a function of depth, higher oxygen saturation % and lower salinity concentrations relative to the ST cycles ( Figure 2 ). The water parcel in SA-SC traveled up the eastern coast of Southern New Zealand and diverged when meeting the STF. While both of our SA cycles were located further offshore than our SA-SC cycle, SA-2 is a better representation of typical SA waters while SA-1 is likely influenced by mixing with the STF as indicated by high fluorescence and oxygen saturation % ( Figure 2 ; Table 1 ). ST-1 and ST-2 cycles displayed overall lower oxygen saturation %, higher temperatures, and higher salinity throughout the water column ( Figure 2 ). SA cycles generally had higher surface concentrations of DRP, NO₃ ^- and NH₄ ⁺ and DRSi compared to ST cycles while surface Chla measurements were higher in ST cycles compared to SA cycles ( Table 1 ).

Phytoplankton biomass and abundance data provided additional evidence of the distinct separation of SA and ST water masses. Abundance data suggests that nanoplankton had higher counts in ST (mean ± 1 standard error; 973.69 ± 112.27 cells mL^-1) compared to SA (604.43 ± 46.91 cells mL^-1) ( Figure 3 ). Diatoms showed the opposite pattern with higher abundances in SA (12.7 ± 2.54 cells mL^-1) compared to ST (0.17 ± 0.06 cells mL^-1), although diatoms were also rare in SA-2. Microplankton showed no significant difference in abundance between SA (8.43 ± 0.61 cells mL^-1) and ST (8.88 ± 1.05 cells mL^-1). Overall, when analyzing the vertically averaged biomass, which is calculated by taking the average biomass across all euphotic depths, picoplankton (as a summed group) exhibited the highest biomass in the two offshore SA cycles averaged together (18.5 ± 0.9 µg C L^-1) compared to nanoplankton (16.6 ± 1.3 µg C L^-1) and microplankton (13.0 ± 1.1 µg C L^-1) ( Figure 4 ). However, nanoplankton biomass (35.2 ± 4.6 µg C L^-1) was highest in ST cycles compared to picoplankton (28.9 ± 1.1 µg C L^-1) and microplankton (14.4 ± 1.6 µg C L^-1). Diatoms had higher vertically averaged biomass in all SA cycles (0.8 ± 0.1 µg C L^-1) compared to ST cycles (0.1 ± 0.04 µg C L^-1) ( Figure 4 ) and higher abundance in SA (12.7 ± 2.5 cells mL^-1) compared to ST (0.9 ± 0.51 cells mL^-1) ( Figure 3 ). Although diatoms had similar abundances in the upper water column for SA-SC and SA-1 ( Figure 3 ), vertical biomass distributions suggested that diatoms had higher biomass in SA-SC compared to SA-1 ( Figure 4 ). Generally, microplankton biomass did not differ between SA and ST regions (16.2 and 14.4 µg C L^-1, respectively) and had the highest vertically averaged biomass at the SA-SC water parcel (21.7 ± 2.2 µg C L^-1) ( Figure 4 ). The abundance of microplankton was also comparable between the SA and ST cycles (8.43-8.88 cells mL^-1) ( Figure 3 ). In SA cycles, nanoplankton vertically averaged biomass (15.6 ± 0.9 µg C L^-1) was comparable to microplankton biomass (16.2 ± 1.2 µg C L^-1) but was nearly two-fold higher in ST cycles (35.2 ± 4.6 µg C L^-1 and 14.4 ± 1.6 µg C L^-1, respectively) ( Figure 4 ).

Picoplankton groups suggested that hbact water column biomass was 2-fold higher in ST (22.0 ± 1.5 µg C L^-1) relative to SA cycles (12.4 ± 0.5 µg C L^-1). Pro was found in SA cycles (0.6 ± 0.1 µg C L^-1), particularly in the cycles conducted further offshore and away from the STF, but it was not present in ST cycles ( Figure 4 ). Syn had slightly higher overall vertically averaged biomass in ST cycles (6.5 ± 1.2 µg C L^-1) compared to SA cycles (4.5 ± 0.5 µg C L^-1) ( Figure 4 ). Peuk were present at very low concentrations in all regions with slightly higher vertically averaged biomass in the ST (0.4 ± 0.04 µg C L^-1) compared to the SA (0.2 ± 0.002 µg C L^-1). Average C biomass and abundance for each group and water parcel can be found in Supplementary Table 3 .

Normalized size spectra allow us to visualize the abundance and total biomass of cells in a specific size range. The biomass size spectrum reflects the energy flow within the system, with the slope providing information on growth and mortality dynamics and trophic efficiency within the community. This can also signify trophic recycling processes within the system (Zhou, 2006). The relative shape of the curve describes the degree to which community composition is dominated by small or large phytoplankton, with a less negative slope indicating a relatively greater importance of large cells (microplankton dominance) and vice versa (note that the smallest size class in both NBSS ( Figure 5 ) and NASS ( Figure 6 ) represent flow cytometric picoplankton in the size range of 0.5 to 2 µm that contains the averaged sum of hbact, Syn, Pro and peuk). The overall average slope of NBSS of all ST cycles was -0.78, which is less negative than the overall average slope of SA cycles (-1.00) and suggests a greater relative proportion of larger cells in ST relative to SA cycles ( Supplementary Table 4 ). Similarly, the average slope of NASS of all SA cycles was -3.26 and slightly more negative compared to ST cycles (-3.04). A table that contains all slope values can be found in the Supplementary Material ( Supplementary Table 4 ). While NBSS did not show monotonic changes with depth, the NBSS was typically more negative at the DCM than in the mixed layer suggesting that small cells contributed proportionally more to biomass in the deep euphotic zone than the shallow euphotic zone. The NBSS plots also show consistently low biomass of small nano cells (2 – 4 µm cells) across all cycles and depths.

Vertically averaged summed biomass, calculated by taking the vertical integrations of biomass above the ML and below the ML and then diving by the two depths, showed differences between picoplankton, nanoplankton and microplankton within the ML compared to below the ML ( Figure 7 ). ML depths can be found in Table 1 . Picoplankton had slightly lower vertically averaged biomass within the ML (Hbact 16.5 ± 3.2 mg C m^-3, Syn 4.5 ± 1.9 mg C m^-3, Pro 0.2 ± 0.1 mg C m^-3, peuk 0.2 ± 0.1 mg C m^-3) compared to below the ML (Hbact 17.6 ± 3.3 mg C m^-3, Syn 7.2 ± 1.7 mg C m^-3, Pro 0.4 ± 0.3 mg C m^-3, peuk 0.3 ± 0.04 mg C m^-3) ( Figure 7A ), which is more clearly seen in relative proportion with picoplankton contributing to more vertically averaged biomass below the ML (42% on average) compared to the ML (30% on average) ( Figure 7B ). Conversely, nanoplankton and microplankton had higher biomass in the ML (nanoplankton 29.1 ± 7.0 mg C m^-3, microplankton 18.4 ± 2.1 mg C m^-3) compared to below the ML (nanoplankton 19.3 mg C m^-3 ± 2.4, microplankton 15.3 ± 2.6 mg C m^-3). Diatoms had similar vertically averaged biomass in the ML (0.9 ± 0.6) and below (1.1 ± 0.6). Picoplankton as a group contributed to more vertically averaged biomass below the ML while the larger plankton groups (nanoplankton, diatoms and microplankton) contributed to more biomass in the ML ( Figure 7B ). ST water samples had higher vertically average summed biomass in the ML compared to SA water samples (92.7 ± 3.2 mg C m^-3 and 54.7 ± 5.4 mg C m^-3, respectively) and below the ML (ST: 69.6 ± 11.5 mg C m^-3, SA: 55.3 ± 3.9 mg C m^-3).

While our epi measurements enabled high vertical resolution of C:Chla variability, they should be treated as upper bounds for the actual C:Chla ratios of nanoplankton and microplankton because they include heterotrophic protists (i.e., due to overstaining with proflavine, we were unable to use Chl autofluorescence to discriminate phototrophs from obligate heterotrophs). Our C:Chla values are calculated by taking the ratio of mean carbon divided by mean Chla for each depth and cycle. Picoplankton consistently exhibited lower average C:Chla ratios in the ML compared to the nanoplankton and microplankton in both the ST and SA regions ( Figure 8A ). Microplankton generally had higher C:Chla ratios compared to nanoplankton, although this relationship varied depending on location and depth, which is likely due to size fractionated Chla differences ( Figure 8A ). Picoplankton had the highest C:Chla signal at 30 m (SA: 48.9 ± 12.1; ST: 19.1 ± 7.4), nanoplankton at 20 m (SA: 354.3 ± 67.2; ST: 168.6 ± 89.9) and microplankton at 12 m (SA: 1044.6 ± 474.1; ST: 501.2 ± 241.7) ( Figure 8A ). In general, SA cycles had higher overall average C:Chla for all groups (e.g., at 12 m: picoplankton 18.4 ± 4.8, nanoplankton 319.6 ± 49.0, microplankton 1044.6 ± 474.1) compared to ST regions (12 m: picoplankton 12.1 ± 3.7, nanoplankton 156.4 ± 26.0, microplankton 501.2 ± 241.7) ( Figure 8A ).

We employed FlowCam samples to calculate C:Chla ratios as an alternative method of estimating C biomass, which mitigates biases associated with including heterotrophs compared to the epi data ( Figure 8B ). However, FlowCam data was limited to 2 depths (ML and DCM) and does not include ≤4 µm cells. FlowCam data revealed higher C:Chla ratios in SA compared to ST waters for both nanoplankton and microplankton and at both depths. Nanoplankton C:Chla was highest in the ML for SA-2 (73.5 ± 11.1) and lowest in the DCM of ST-2 cycle (23.1 ± 2.7) ( Figure 8B ). Microplankton C:Chla was highest in the ML of SA-2 (219.2 ± 43.9, Figure 8B ); a peak in C:Chla ratio that is also seen in epi-based C:Chla estimated for microplankton (1044.6 ± 474.1, Figure 8A ), and was lowest in the DCM of ST-1 (19.2 ± 3.6). In the ML of the SA region, the C:Chla ratio for microplankton was approximately twice as high as that for nanoplankton (120.3 ± 21.6 and 61.7 ± 5.8, respectively) although nanoplankton and microplankton C:Chla were similar in the ST regions (microplankton, 46.2 ± 7.5 and nanoplankton, 42.3 ± 5.1).

To allow comparison to other studies that lack size-resolved C:Chla ratios, we also present phytoplankton C:Chla ratios as the sum of epi (microplankton, nanoplankton and diatoms) and FCM (Syn, Pro and peuk) biomasses divided by total Chla ( Figure 8C ). We also compared C:Chla ratios determined from POC:total Chla to determine how well our estimations of phytoplankton C paired with overall POC to Chla depth profiles ( Figure 8C ). A point to consider when examining different methods used to calculate C:Chla is the use of different types of filters (e.g., polycarbonate vs. GF/F filters), which may influence the derived ratios due to variations in pore size and protein-binding activity, especially in samples with high dissolved organic matter (Chow et al., 2005). This variation in filter type is often unavoidable when sampling for POC, silicate, and other parameters, yet it may impact the results for bulk C:Chla values. Both summed phytoplankton and POC C:Chla had much higher average surface (~5 m) values in SA (144.0 ± 18.6 and 267.5 ± 20.8, respectively) compared to ST (70.9 ± 6.8 and 173.1 ± 4.1, respectively). Overall, POC C:Chla was higher than summed phytoplankton C:Chla, which is expected as POC includes hbact and detritus whereas summed phytoplankton contains only autotrophic C (and, in the case of our epi samples, also includes some heterotrophic protists). A detailed summary table of our various methods of estimating C:Chla can be found in Supplementary Table 5 .

To quantify the proportion of the seston that is alive versus dead (i.e., detritus, including dead cells, fecal pellets, etc.), living biomass (microplankton, diatoms, nanoplankton, Pro, Syn, hbact and peuk) was summed and compared to bulk POC measurements. Across all locations and 6 depths, the median ratio of living:total C was 0.67 ± 0.03, which suggests that roughly 33% of the seston is detrital ( Figure 9 ). We had initially hypothesized that salps might increase the ratio of living:dead organic C, because their high filtration rates and rapidly sinking fecal pellets could efficiently remove detritus from the euphotic zone. We used a Mann-Whitney U test to compare the differences in living and total C between water parcels with and without salps. The results indicated a possibly meaningful decrease in the proportion of living seston in the presence of salps, although the difference was not statistically significant at the 95% confidence level (p-value = 0.48). In contrast to differences between salp-influenced and non-salp water parcels, we found no meaningful difference in percent living seston between SA (0.70 ± 0.04) and ST samples (0.66 ± 0.05; p-value = 0.84).

Our results suggest that ST cycles were dominated by nanoplankton, which is in agreement with previous research finding that nanoplankton comprised ~30% of total phytoplankton biomass in STF locations, compared to ~20% of total phytoplankton biomass in SA locations (Hall et al., 1999). Notably, our cycles sampled the STF zone, which is characterized by active mesoscale processes that mix ST and SA waters. Consequently, when appropriate, our samples are considered ST-influenced or SA-influenced within the STF when comparing our data to previous published research. Offshore SA cycles were dominated by picoplankton, while the SA-SC cycle, which was impacted by the Southland Current, was dominated by microplankton. High microplankton biomass in the SA-SC likely resulted from potential mixing of elevated nitrate and silicate SA waters with iron-enriched waters from the coastal domain and/or Southland Current, a subantarctic current that transports mainly SA (90%) but also ST (10%) waters (Sutton, 2003). In contrast, SA-1 and SA-2, which were located further offshore, were more oligotrophic than SA-SC and were dominated by picoplankton thriving in a high-nutrient, low-chlorophyll (HNLC) region, where iron is likely the limiting factor to phytoplankton growth (Boyd et al., 2010; Gutiérrez-Rodríguez et al., 2020). Although past research in this region of the SW Pacific suggests that microplankton biomass varies between oceanographic regions/water masses and can contribute ~5% of total cell C in STF regions and ~45% of total cell C in SA regions (Hall et al., 1999), during our study, microplankton did not vary greatly between regions (SA: 13.0 ± 1.1 µg C L^-1; ST: 14.4 ± 1.6 µg C L^-1) despite having the highest biomass in our SA-SC cycle (21.7 ± 2.2 µg C L^-1).

Our analysis of the NBSS and NASS spectra further elucidates the structure within these plankton communities by highlighting the distributions of biomass and abundance across different plankton size classes. For example, the lower NBSS slope observed in ST cycles (-0.78) compared to SA cycles (-1.00) suggests that there is a greater contribution of larger cells which potentially indicates higher trophic efficiency and a reduced loss of energy through trophic transfers in the ST cycles (Zhou, 2006). This aligns with our observations of increased nanoplankton dominance in ST cycles and picoplankton dominance in SA cycles, further supported by the more negative NASS slope in SA (-3.26) compared to ST (-3.04). The slope at the DCM can also provide insights into the vertical community structure. We observed that, apart from ST-2, the NBSS slopes tend to be more negative at the DCM which may suggest that there is a greater proportion of smaller cells at the DCM compared to the surface. This depth-related shift in size structure may suggest that smaller cells are better adapted to lower light levels found at the DCM, a pattern that we found in the vertical distribution of our picoplankton biomass as well. These insights from NBSS and NASS deepen our understanding of the community composition dynamics within each region and can illustrate how the size structure may influence carbon cycling and energy transfer across trophic levels.

Diatoms were present at low abundance in both regions with slightly higher biomass in SA cycles. Our results are in agreement with low values of diatoms reported in the SA region (Kopczynska et al., 2001; Peloquin et al., 2011; Cassar et al., 2015). In contrast, Chang and Gall (1998) found that vertically averaged diatom biomass was much higher in ST compared to SA (ST: 2804 μg C L^-1; SA: 13 μg C L^-1), unlike our study which found 10-fold lower average vertically averaged diatom biomass in ST compared to SA (ST: 0.1 ± 0.1 µg C L^-1; SA: 1.4 ± 0.6 µg C L^-1). Chang and Gall (1998) study occurred during a stronger spring bloom, with diatom biomass comprising 68% of total plankton biomass in the ST region, and their DRSi ambient concentration was 6-fold higher than in our study. The draw-down of DRSi and the relatively low abundance of diatoms during our study indicates a fading diatom bloom similar to that reported by Chang and Gall (1998) and Twining et al. (2014). The ST-1 average DRSi concentration of 0.09 ± 0.01 µM is below what is considered DRSi limiting (i.e., <0.2 µM), which is rare to find anywhere in the ocean (Nelson and Dortch, 1996). Despite the high concentration of Chla in ST-2 (2.21 ± 0.20 µg C L^-1, Table 1 ), diatoms were not abundant unlike previous reports (e.g. Chang and Gall, 1998; Twining et al., 2014) and instead the community was dominated by small flagellates and dinoflagellates. This highlights temporal variability in spring bloom dynamics and plankton community composition, particularly the diatom contribution, in this dynamic region (Chiswell et al., 2019; Gutiérrez-Rodríguez et al., 2022).

Pro was present in SA cycles but absent in all ST cycles ( Figure 3 ). This finding is unexpected because Pro is typically observed in warmer, oligotrophic regions, although it has been found to occur in colder waters, down to ~10°C (Doolittle et al., 2008). More often though, Pro is found in warmer ST waters in the Tasman Front, suggesting that its abundance in SA waters may be constrained by temperature (Dubreuil et al., 2003; Yang et al., 2010; Ellwood et al., 2013). In SA cycles, Pro typically had higher biomass deeper in the water column compared to the surface, e.g., for SA-2, average Pro biomass was ~5.6-fold lower at 5 m compared to 46 m (0.3 ± 0.02 and 1.4 ± 0.3 µg C L^-1, respectively).

Macronutrient limitation may explain the lower microplankton biomass seen in ST cycles, particularly nitrogen as indicated by the NO₃ ^-:PO₄ ⁺ ratios in Table 1 which are well below the Redfield ratio and suggest nitrogen limitation may be occurring in this region. In contract, SA cycles are likely limited by trace metals, such as iron, which is characteristic of SA water masses (Banse, 1996; Sedwick et al., 2000; Nodder et al., 2007; Ellwood et al., 2013) and is likely responsible for the reduced biomass of microplankton seen in the offshore SA cycles. Additionally, potential silica limitation in both ST and SA waters, as seen in low silica concentrations ( Table 1 ), could impact diatom productively offshore. It was only in the SA-SC cycle, where macronutrients were relatively high and trace metals were likely available due to proximity to the shelf, that microplankton became dominant contributors to the community. This is in contrast to evidence of iron limitation that was found in offshore SA cycles, as suggested by high macronutrient concentrations ( Table 1 ) and evidenced by photosystem II maximum photochemical efficiency (Fv/Fm) and Q_A reoxidation kinetics (Q_A ^-lifetime), which are indicative of micronutrient-induced physiological stress (Gorbunov and Falkowski, 2021; Décima et al., 2023).

Vertically resolved biomass profiles show distinct differences with depth. Picoplankton (<2 µm) comprised a consistently higher proportion of biomass beneath the surface ML than within the ML; the opposite was true for nanoplankton, diatoms and microplankton ( Figures 7A, B ). This may result from light limitation, because larger cells often have lower intracellular Chla concentrations than small cells to offset the “package effect”, thus potentially making them inferior competitors under light limitation (Finkel, 2001). Laboratory culture experiments also show that diatoms have higher C:Chla during nitrogen starvation, with smaller prasinophytes exhibiting a more muted response to nutrient limitation (Liefer et al., 2018). Previous research in the West Antarctica Peninsula has indicated that large cells, particularly diatoms, are better suited for adapting to higher irradiance levels and that light limitation controls phytoplankton biomass rather than nutrient limitation (Carvalho et al., 2020), although we note that both likely play a role in our study region.

These differences between samples taken within or below the ML highlight the importance of depth-resolved sampling as opposed to just sampling surface waters (Peloquin et al., 2011). Our results show multiple trends with depth, including differences in community composition ( Figure 7 ), size spectra ( Figure 5 and Figure 6 ), and C:Chla ratios ( Figure 8 ). The latter result is not unexpected, as surface waters contain phytoplankton acclimated to high-light environments as compared to low light acclimated plankton found in greater depths (Ryther and Menzel, 1959).

Water column stratification also plays a role in light limitation and cell size distribution. Under strong stratification, typically characterized by low nutrient availability and warmer temperatures, the plankton community tends to shift towards smaller cells. As stratification weakens, the community may transition from smaller to larger cells (Li, 2002). A potential explanation for the dominance of medium-sized cells from a top-down perspective is grazing pressure (Chen and Liu, 2010). Microzooplankton grazing is the main source of phytoplankton mortality globally, accounting for 67% of phytoplankton daily growth (Calbet and Landry, 2004). Safi et al. (2023) showed a tight coupling between picoplankton growth and microzooplankton grazing in the STF, suggesting that the low biomass of picoplankton in this region was to some extent due to the effective grazing of microzooplankton. Strom et al. (2007) in the Gulf of Alaska showed that grazing rates varied according to phytoplankton size classes, with highest grazing rates on phytoplankton <5 µm (0.48 d^-1) and the lowest grazing rates on phytoplankton >20 µm (0.17 d^-1). They hypothesized that variations in mortality on larger phytoplankton arose mainly through variations in biomass of the larger microzooplankton (>40 µm, ciliates and dinoflagellates). Thus grazing may play a role in size spectra variability ( Figures 5 , 6 ) as grazing on smaller cells is often tightly coupled to the growth rate of these cells.

Notably, presence or absence of salp blooms during each cycle had a negligible impact on phytoplankton community composition. However, results from a Mann-Whitney U test indicated a possibly meaningful, yet not statistically significant, decrease in the proportion of living seston in the presence of salps that was initially surprising. This result may relate to the different types of pellets produced by salps. While some salps sampled on the cruise did produce large, rapidly sinking fecal pellets, they also produced fragile, flaky pellets that sank more slowly and were easily fragmented (Décima et al., 2023). Décima et al. (2023) also found that the ratio of salp fecal pellet flux into sediment traps to fecal pellet production rates increased during the salp bloom progression, suggesting either a change in the characteristics of fecal pellets over time or that a proportion of pellets sank slowly. Either way, this highlights the fact that salps not only export organic matter via their fecal pellets but can also increase the standing stock of detritus in the epipelagic with potential implications for microbial substrate availability and nutrient regeneration.

We see a wide range of C:Chla (statistics of the ratios are derived from the ratio of mean values for each depth and cycle) in the ML and DCM when utilizing different methods of determining C concentrations against summed size fractionated Chla ( Table 2 ). Although the general trends are in agreement (e.g., various methods of C:Chla show highest values in the SA-2 cycle, Table 2 ), C:Chla ratios can vary up to 6-fold dependent upon which C method is used. Thus it is important to consider which groups are being included in C analysis as, for example, our epi method is likely overestimating C:Chla as it includes heterotrophs, thus FlowCam results may be more accurate, although our FlowCam data also likely includes some smaller heterotrophs that could not be visually distinguished from phototrophs. Our bulk POC measurements serve as an independent accuracy check. On average, 66% of our combined living POC measurements were lower than the corresponding bulk POC measurement ( Figure 9 ). Considering the substantial uncertainty associated with cellular C calculations, this suggests that our methods were likely not substantially underestimating or overestimating C biomass. Although our ML FlowCam C:Chla estimates also included some heterotrophs, they are likely closer to representing autotrophic C observed in this area and were in better agreement with Gutiérrez-Rodríguez et al. (2020) who sampled close to, but south of, our SA-SC location, and reported C:Chla values of 42 and 54 compared to FlowCam SA-SC C:Chla of 53 ± 9.0 for nanoplankton and 96 ± 25 for microplankton. Bradford-Grieve et al. (1999) also had similar results to our ML ST-influenced region, averaging 38.5 ± 3.2 for nano and 37.2 ± 4.8 for microplankton (compared to their average of 36), while the SA-influenced region averaged 58.0 ± 3.6 for nanoplankton and 85.0 ± 12.2 for microplankton (compared to their average of 84).

Another important methodological issue to consider is the difference between FlowCam or epi and size fractionated Chla approximation of cell size. In FlowCam and epi, we used ESD as the proxy for determining cell size whereas size fractionated Chla measurements are operationally defined by the pore sizes of the filters. Given the complex shapes of many nanoplankton and microplankton taxa, these differences in approaches to size determination can add uncertainty to size-based C:Chla estimates. For instance, Landry et al. (2009) noted up to a 2-fold overestimation of picoplankton Chla by size fractionated Chla measurements as cells with flexible walls are able to squeeze through 2 µm filter pores under vacuum pressure during filtration (Psarra et al., 2005). Notably, in the SA cycles ( Figure 8A ), the nanoplankton C:Chla ratio was approximately 3 times lower at 12 m depth compared to micro (nanoplankton 319.6 ± 49.0; microplankton 1044.6 ± 474.1), but approximately 50% higher at 50 m (nanoplankton 219.7 ± 53.4; microplankton 141.2 ± 54.4). The observed variability in size fractionated Chla is a key factor, as highlighted by the substantial differences between the 2 µm and 20 µm size fractions at 12 m in the SA-2 cycle. Specifically, the Chla concentrations were 5 times higher in the 2 µm fraction compared to the 20 µm fraction (0.04 µg Chla versus 0.008 µg Chla), while the C biomass measurements were similar (14.8 µg C L^-1 and 17.5 µg C L^-1, respectively).

Time of sampling may also play a role in C:Chla as our Chla data was collected at approximately 02:00, during a period with no natural light. This nighttime sampling may have implications for the interpretation of C:Chla ratios due to diel variations in chlorophyll content within phytoplankton cells. Studies have shown that Chla concentrations in many species, including diatoms, can vary throughout the day, often peaking during daylight and decreasing at night (Owens et al., 1980). This diel cycle is attributed to Chla synthesis being closely linked to photosynthetic activity, which is reduced in the absence of light. Consequently, our nighttime sampling may result in elevated C:Chla ratios, as chlorophyll levels could be lower compared to daytime measurements, while carbon content should remain relatively consistent. The magnitude of diel variation in C:Chla will depend on the relationship between photosynthesis rates during the light period, overall growth rates, and the extent to which energy storage products are mobilized during darkness to support cell synthesis (Geider, 1987).

When comparing regional C:Chla, ST regions had higher summed plankton C biomass compared to SA regions, consistent with prior studies showing highest cellular plankton C concentrations in the STF region, intermediate concentrations in the ST regions and the lowest C in the SA region (Bradford-Grieve et al., 1997). When comparing these C measurements to Chla, we find that our epi C:Chla ratios are higher than other bulk phytoplankton ratios observed in this region. For example, our data suggest average C:Chla is 260.0 ± 40.5 for nanoplankton and 859.2 ± 311.9 for microplankton at 12 m. These ratios are much higher than the bulk average C:Chla recorded for this region by Bradford-Grieve et al. (1999) of 36 in the STF and 84 in SA at 10 m. However, it is important to note the limitations of Bradford-Grieve’s dataset, as their sampling was restricted to a depth of 10 meters at just two locations in the STF and two in the SA, yet it is one of the few studies conducted in the same region as ours. In contrast, our overall averaged C:Chla for picoplankton (15.8 ± 3.2) was in range to that found in the euphotic zone of the Alboran Sea in the Mediterranean where Arin (2002) found C:Chla of 15 to 59 for cells <2 µm. However our C:Chla is higher for nanoplankton and microplankton compared to this study which ranged from 24 to 57 for cells >2 µm. Laboratory studies have found lower C:Chla ranging from 26-113 for various sized plankton including cyanophytes, chlorophytes, diatoms and dinoflagellates (Yacobi and Zohary, 2010). C:Chla tends to increases with nutrient limitation (Laws and Bannister, 1980; Liefer et al., 2018), suggesting that decreases in Chla concentrations are responsible, rather than an increase in cellular C. In our study, the entire community C:Chla was 123.8 ± 18.9 overall which is higher than previous measurements in the region (Bradford-Grieve et al., 1997; Zeldis, 2002). However, higher C:Chla has been recorded (ranging from 150 to 200) in the upper euphotic zone in the Gulf of Mexico (Selph et al., 2021), while modeling studies have predicted C:Chla to be as high as 160 in the upper euphotic zone in latitudes ranging from 0° to 35° N (Taylor et al., 1997) and 200:1 in surface waters in the California Current Ecosystem (Li et al., 2010).

Light, temperature and nutrients have a combined effect on cellular C:Chla ratios, which are dynamically adjusted to optimize the physiological state of phytoplankton (MacIntyre et al., 2000). C:Chla increases linearly with increased light at a constant temperature, yet decreases exponentially with increased temperature at constant light (Geider, 1987). C:Chla is lowest at high temperature, low irradiance and nutrient replete conditions and is maximal at low temperature, high light and nutrient limiting conditions (Geider et al., 1997). In continuous culture experiments with diatoms, C:Chla was found to be low at low irradiance and nutrient limitation (Laws and Bannister, 1980). Disentangling these competing influences can be difficult, particularly in the field as nutrient limitation, including macronutrients and trace-metals, may also impact C:Chla (Taylor et al., 2015) with iron limitation reported to greatly impact C:Chla in the Southern Ocean (Strzepek et al., 2019; Ye et al., 2023) which likely underpin the higher C:Chla observed in SA cycles. A decrease in cellular Chla in cells has been linked to iron deficiency (Glover, 1977; Boyd et al., 2001; Sunda and Huntsman, 1997) and in-situ experiments in the Southern Ocean have shown that iron fertilization led to an increase in cellular Chla concentrations and a decrease in C:Chla (Gall et al., 2001).