Summertime productivity and carbon export potential in the 1 Weddell Sea , with a focus on the waters adjacent to Larsen C 2 Ice Shelf 3 4

16 The Weddell Sea represents a point of origin in the Southern Ocean where globally-important water masses form. 17 Biological activities in Weddell Sea surface waters thus affect large-scale ocean biogeochemistry. During 18 January/February 2019, we measured net primary production (NPP), nitrogen (nitrate, ammonium, urea) uptake, 19 and nitrification in the western Weddell Sea at the Antarctic Peninsula (AP) and Larsen C Ice Shelf (LCIS), in the 20 southwestern Weddell Gyre (WG), and at Fimbul Ice Shelf (FIS) in the south-eastern Weddell Sea. The highest 21 average rates of NPP and greatest nutrient drawdown occurred at LCIS. Here, the phytoplankton community was 22 dominated by colonial Phaeocystis antarctica, with diatoms increasing in abundance later in the season as sea-ice 23 melted. At the other stations, NPP was variable, and diatoms known to enhance carbon export (e.g., Thalassiosira 24 spp.) were dominant. Euphotic zone nitrification was always below detection, such that nitrate uptake could be 25 used as a proxy for carbon export potential, which was highest in absolute terms at LCIS and the AP. Surprisingly, 26 the highest f-ratios occurred near FIS rather than LCIS (average of 0.73 ± 0.09 versus 0.47 ± 0.08). We attribute 27 this unexpected result to partial ammonium inhibition of nitrate uptake at LCIS (where ammonium concentrations 28 were 0.6 ± 0.4 μM, versus 0.05 ± 0.1 μM at FIS), with elevated ammonium resulting from increased heterotrophy 29 following the accumulation of nitrate-fuelled phytoplankton biomass in early summer. Across the Weddell Sea, 30 carbon export appears to be controlled by a combination of physical, chemical, and biological factors, with the 31 highest potential export flux occurring at the ice shelves and lowest in the central WG. 32 33


39
The Southern Ocean is an important driver of Earth's climate as it transports large quantities of heat and dissolved 40 gases, and supplies 65-85% of the global ocean's nutrients (Keffer and Holloway, 1988   sensor. Density (sigma-theta; σθ) was derived from CTD measurements of temperature, salinity, and pressure, and

171
Seawater was collected from discrete depths using a rosette of twenty-four 12 L Niskin bottles. At each station, 172 seawater samples for nutrient analysis were collected throughout the water column (typically at 15 discrete 173 depths), while samples for phytoplankton taxonomy and rate experiments were taken from 3-6 depths (see below) 174 that were selected based on profiles of temperature, chlorophyll-a fluorescence, and PAR measured during the 175 CTD down-casts.

177
Simulated in situ experiments were conducted to determine the rates of net primary production (NPP), N uptake 178 (as nitrate (NO3 -), ammonium (NH4 + ), and urea-N), and nitrite (NO2 -) oxidation (a measure of nitrification). For 179 NPP and N uptake, seawater was collected from three depths coinciding with the 55%, 10%, and 1% PAR levels, 180 then pre-screened through 200 μm mesh to remove large grazers and transferred to six 1 L and six 2 L 181 polycarbonate bottles per depth. 15 N-labeled NO3 -, NH4 + , or urea-N was added to four of the twelve bottles (i.e., 182 two 1 L and two 2 L bottles per N species) and NaH 13 CO3 was added to the bottles amended with 15 N-NH4 + . The  analysis platform following published auto-analysis protocols (Diamond, 1994;Grasshoff, 1976) in a 202 configuration with a detection limit of 0.5 µM. Duplicate samples were measured for NO3 -+NO2and Si(OH)4 on 203 different days, and the standard deviation for duplicates was <0.5 µM, with a lower standard deviation for lower-204 concentration samples. NO3concentrations were determined by subtraction of NO2from NO3 -+NO2 -.

205
Concentrations of phosphate (PO4 3-) and NO2were measured shipboard by standard benchtop colourimetric  a Turner Designs Trilogy fluorometer equipped with a UV module. The detection limit was <0.05 µM and the 213 standard deviation for duplicate samples was ≤0.05 µM. The matrix effect (ME) that results from the calibration 214 of seawater samples with Milli-Q water standards was calculated using the standard addition method (Saxberg 215 and Kowalski, 1979). All samples were corrected for the ME (Taylor et al., 2007), which was always <10% and 216 typically ≤5%. Urea-N concentrations were measured via the colourimetric method of Revilla et al. (2005) using 217 a Thermo Scientific Genesis 30 Visible spectrophotometer equipped with either a 1 cm-or 5 cm-pathlength cell.

218
The detection limit was 0.05 µM and the standard deviation for duplicate samples was ≤0.05 µM. Hereafter, we 219 use "urea" when referring to urea-N.

286
To determine relative carbon export potential at each station, we calculated the f-ratio (a measure of 287 new production relative to total (i.e., new+regenerated) production) using the absolute N uptake and 288    conical plankton net (r = 12.5 cm; h = 50 cm) with a mesh size of 55 μm. Samples were transferred to 50 mL 302 centrifuge tubes, fixed with 10 μL of 25% glutaraldehyde, and stored at room temperature in the dark until later 303 analysis via light and scanning electron microscopy. Additionally, samples for flow cytometry were collected in 304 9 50 mL centrifuge tubes from Niskin bottles fired at the 55%, 10%, and 1% PAR depths. These samples were fixed 305 with 10 μL of 25% glutaraldehyde and stored in the dark at 4°C until analysis.

307
Onshore, each preserved net-sample was homogenized, and one drop (40 μL) was wet mounted on a slide. All the 308 cells on the slide with intact chloroplasts (i.e., alive at the time of sampling) were counted at 400x or 630x 309 magnification using a Zeiss AxioScope A1 light microscope (LM). The number of cells mL -1 was calculated as:

317
An aliquot of 5 mL from each preserved sample was cleaned by removing carbonate particles and organic matter 318 using 10% hydrochloric acid and 37% hydrogen peroxide, respectively. After thorough rinsing with distilled 319 water, permanent slides were prepared by pipetting the cleaned material onto acid-washed coverslips, air drying 320 them overnight, and mounting the cover slips onto glass slides using Naphrax® mountant (refractive index = 1.7).

321
The permanent slides were examined using a Zeiss AxioScope A1 LM equipped with differential interference

327
The average size (μm) and carbon content (pg C cell -1 ) of each identified diatom species was taken from Leblanc   The MLD appeared most strongly controlled by salinity at all stations and was always shallower than the depth 383 of the euphotic zone (Zeu; MLD of 13.9 ± 5.9 m and Zeu of 28.5 ± 9.1 m; n = 10) ( Figure 2d; Table 1). The rates of NPP, N uptake and 387 nitrification were therefore trapezoidally-integrated to Zeu rather than to the MLD since we assume that 388 phytoplankton were active at least to the depth of 1% PAR.

403
Urea concentrations were more variable, likely due to variability in the processes that produce this N form (e.g.,

417
The concentrations of NO2were generally low throughout the euphotic zone, and decreased to below detection

446
(1999) observed a NO3 -:PO4 3depletion ratio of ~20:1, while in areas dominated by iron-deplete diatoms, this 447 ratio was ~10:1. The NO3 -:PO4 3depletion ratios can thus also yield insights into the dominant phytoplankton 448 species active in the upper water column. In our study, the average euphotic zone Si(OH)4:NO3depletion ratios 449 ranged from 0.5 to 6.1 (Table 1), with the highest ratios estimated for the WG stations (average of 5.4 ± 5.5) and 450 at FIS in late summer (average of 2.3 ± 0.5). The euphotic zone average NO3 -:PO4 3depletion ratios were more 451 variable, ranging from 3.7 ± 1.5 to 48.6 ± 11.5, with the lowest ratios computed for the WG stations (average of 452 4.1 ± 1.5) and the highest for FIS in early summer (average of 33.7 ± 3.6). In the latter case, the degree of Si(OH)4 453 and PO4 3depletion was extremely low (Table 1)    to the lower end of the rates observed at LCIS, while NPP along the AP increased shoreward (i.e., the lowest rates 505 were observed at AP1 and the highest at AP3) to values similar to those observed at LCIS. The highest euphotic 506 zone-integrated rates of NPP were observed at AP3 (65.0 ± 0.1 mmol m -2 d -1 ) and L5 (61.0 ± 0.7 mmol m -2 d -1 ), 507 while the lowest occurred at L10 (1.8 ± 0.04 mmol m -2 d -1 ) ( Table 2).

509
As per NPP, the rates of ρNO3decreased towards Zeu at all stations (Figure 6b, f and j), as did the extent of NO3 -510 depletion (Figure 4a). The depth-specific rates of ρNO3were highest at LCIS and lowest in early summer at FIS.

511
However, because the euphotic zone was generally shallower at LCIS than at the other stations, the euphotic zone-512 integrated rates of ρNO3were fairly similar across the study region, with the largest variability observed at LCIS 513 (

522
At all stations, rates of ρNH4 + increased with depth, reaching a maximum at Zeu (Figure 6c, g and k). The highest 523 depth-specific rates of ρNH4 + were observed at LCIS and the lowest at FIS in early summer. Euphotic zone-524 integrated rates of ρNH4 + at the AP stations were comparable to those observed at LCIS (regional average of 3.3 525 ± 2.2 mmol m -2 d -1 and 2.5 ± 1.3 mmol m -2 d -1 , respectively), while the rates at the WG stations and at FIS in late 526 summer were comparable to the lower end of the LCIS rates (average of 2.0 ± 0.2 mmol m -2 d -1 at WG and 1.9 ± 527 0.0 mmol m -2 d -1 at FIS). The early-to late-summer rise in the euphotic zone-integrated rates of ρNH4 + at FIS 528 coincided with an increase in the average euphotic zone NH4 + concentration from below detection to 0.2 ± 0.1 529 µM (Figure 3a). At the AP, LCIS, and WG stations, the rates of ρNH4 + were similar to the coincident rates of 530 ρNO3 -, while at FIS, ρNH4 + was less than half of ρNO3 -( Table 2). The highest euphotic zone-integrated rates of 531 ρNH4 + were observed at station AP3 (5.8 ± 0.0 mmol m -2 d -1 ), coincident with a high average euphotic zone NH4 + At the stations where urea uptake was measured (LCIS stations and WG1; 11 out of 19 stations; Figure 6; Table   558 2), ρurea accounted for 8 ± 6% of total N uptake (i.e., ρNO3 -+ ρNH4 + + ρurea). Excluding urea uptake when 559 calculating the f-ratio would therefore overestimate the fraction of potentially exportable carbon by ~8%. We thus 560 estimated urea uptake at the stations where it was not measured as: 561 562 ρurea = (ρNO3 -+ ρNH4 + ) x 0.08 563 564 Equation 7 may overestimate urea uptake at some of the stations, particularly where low urea concentrations were 565 measured. Theoretically, ρurea can also be estimated by assuming that total N uptake should equal NPP/6.63, 566 such that any difference between ρNO3 -+ ρNH4 + and NPP/6.63 is due to urea uptake. However, this approach 567 underestimated urea uptake at all the stations where ρurea was directly measured, probably because the use of a 568 C:N ratio of 6.63:1 assumes balanced phytoplankton growth. We therefore chose to use equation 7 to estimate 569 urea uptake for the stations lacking ρurea measurements as this approach will yield a more conservative (i.e.,

650
For the regions of the Weddell Sea that we sampled in summer 2019, the euphotic zone-integrated rates of NPP 651 and N uptake were generally lower at the OOZ stations than the CCSZ stations, with the highest depth-specific 652 uptake rates observed in surface waters at LCIS (Figure 6a-d; Table 2). The few studies that have previously 653 measured summertime rates of NPP and N uptake in the Weddell Sea report similar results, with rates in the  inter-and intra-regional differences below.

663
Light and water column stability: Surface waters throughout the study region were generally well stratified, with

664
MLDs ranging from 7 to 30 m, except at the early-summer FIS stations where the MLD ranged from 63 to 135 m 665 (   (Table 2). Here, we observed a positive relationship between the rates and SST, with ρNO3increasing 673 at higher SSTs, the latter indicative of increased water column stratification (Figures 10a and S4b; see below).

675
Throughout the sampling region, the average euphotic zone rates of ρNH4 + and ρurea also varied with Zeu which 676 could be taken to indicate that these processes were also light dependent. However, such a finding would be 677 unexpected, as the energy requirement associated with NH4 + and urea assimilation is low (El-Sayed and Taguchi

704
At LCIS, the stations closest to the ice shelf were characterised by low SSTs and low rates of NPP and N uptake 705 (stations L1 and L3; Figures 1 , S4 and S5a ; Table 2). The low SSTs can be attributed either to the formation of 706 sea-ice or to the upwelling of WW along the ice shelf. Sea-ice formation, in addition to decreasing SST, also 707 increases the salinity of ASW due to brine rejection (Gill 1973). While the salinity of ASW at the low-SST stations 708 was indeed elevated, the oxygen concentrations were relatively low (≤300 µM, which is below saturation; Figure   709 S5b-d). In surface waters and sea-ice, oxygen is typically saturated as it rapidly equilibrates with the atmosphere 710 (Gleitz et al., 1995) and is produced by photosynthesizing phytoplankton and sea-ice algae. Sea-ice formation 711 should not, therefore, drive a decrease in the oxygen content of ASW. The low oxygen concentrations at stations 712 25 L1 and L3 were contiguous with those in the underlying WW ( Figure S5d), leading us to conclude that the cool, 713 saline waters along the ice-shelf indicate recent upwelling of WW. Such upwelling could temporarily inhibit 714 productivity by decreasing the stability of the water column and mixing phytoplankton below the euphotic zone.

715
This mechanism can explain the low uptake rates and weak nutrient depletions observed at the low-SST stations.

717
Relatively cold, saline surface waters have previously been observed at the ice-edge off Larsen A and B Ice

718
Shelves and shown to hinder NPP (Cape et al., 2014). In that case, the dense surface waters were surmised to 719 result either from offshore wind stress at the inshore region that induced localised mixing, or from the advection 720 of surface waters offshore by coastal upwelling. Both mechanisms would decrease water column stability, and by

769
the Si:NO3depletion ratios were low (generally <1) and regenerated N uptake was high relative to the other 770 stations (Figure 11a and d). Under favourable nutrient and light conditions, diatoms will typically consume NO3 -771 over NH4 + as i) NO3is usually present in substantially higher concentrations than NH4 + and ii) the lower surface 772 area-to-volume ratio of (larger) diatoms makes it harder for them to compete with smaller cells for a less abundant

774
The average Si:NO3depletion ratio of 1.0 ± 0.2 at LCIS can therefore be attributed almost entirely to diatoms.

775
When total N uptake is considered, the Si:N depletion ratios decrease to 0.3 ± 0.1, indicating the consumption of 776 three-times more N than Si(OH)4. We attribute this decline to regenerated N uptake by P. antarctica, a 777 phytoplankton group that is known to preferentially consume NH4 + when it is available due to the lower energy

786
We can also use the NO3 -:PO4 3depletion ratios to better understand the iron conditions and relative importance diatoms contributed 6.47 x 10 -3 pg C mL -1 to biomass while P. antarctica only contributed 0.07 x 10 -3 pg C mL -794 1 ), the NO3 -:PO4 3depletion ratios were low (13.0 ± 0.6; Figure 11b). In contrast, at the stations where P. antarctica 795 were numerically dominant (e.g., L6; where P. antarctica constituted 90% of the microphytoplankton) and 796 contributed more to biomass (0.17 x 10 -3 pg C mL -1 ), elevated NO3 -:PO4 3depletion ratios were measured (20.4 ± 797 0.3; Figure 11b; Table 1). Furthermore, high rates of NH4 + uptake were measured at LCIS, equivalent to and at 798 times greater than the coincident NO3uptake rates ( Figure 6; Table 2) , particularly at the stations with the highest 799 relative abundance of P. antarctica. In general, the relative contribution of diatoms versus P. antarctica therefore 800 appears to control the nutrient depletion ratios on a variety of scales in the Weddell Sea.  to sea-ice melt will likely also experience large diatom blooms. We thus conclude that the dominance of diatoms 844 over P. antarctica at the non-LCIS stations was influenced by local hydrodynamic processes that rapidly induce 845 water column stability, and increase light availability (e.g., in areas of recent sea-ice melt). By contrast, P.

846
antarctica dominates under low-light, such as in the deep mixed layers that initially characterize coastal polynyas.

847
Eventually, diatoms will succeed P. antarctica in these polynyas as conditions become favourable for their

862
That said, accounting for urea uptake decreased the average f-ratio by very little, from 0.57 ± 0.15 to 0.52 ± 0.14.

864
Estimates of the f-ratio and carbon export potential can be complicated by euphotic zone nitrification, which

882
Although the highest f-ratios were estimated for the FIS stations, the highest rates of ρNO3were observed at LCIS 883 and along the AP (Figure 8; Table 2). FIS was thus characterised by the highest carbon export potential relative 884 to NPP, while the N cycle data imply that the absolute potential carbon export flux was highest at LCIS and the 885 AP. The maximum extent of nutrient depletion was also observed at LCIS (NO3depletion of 57-428 mmol m -2 886 and PO4 3depletion of 5.8-18.7 mmol m -2 ). Assuming Redfield C:N and C:P stoichiometry of 6.63:1 and 106:1, 887 respectively, the seasonal NO3depletion equates to a carbon export flux of 0.

897
Throughout the Weddell Sea, NH4 + and urea uptake were coupled with substrate availability, while NO3uptake 898 was not. Instead, NO3uptake appeared to vary with light (see above) and as a function of the ambient NH4 + 899 concentration (Figure 12a). At LCIS where NH4 + was elevated throughout the mixed layer at all stations, NO3 -900 uptake and NO3depletion decreased with increasing NH4 + (Figure 12), which we attribute to NH4 + inhibition of 901 NO3uptake (Goeyens et al., 1995 (Figure 12), ρNO3was on average as high as ρNH4 + and was never zero (