Marine Chemistry 258 (2024) 104346 Contents lists available at ScienceDirect Marine Chemistry journal homepage: www.elsevier.com/locate/marchem Carbon and silica fluxes during a declining North Atlantic spring bloom as part of the EXPORTS program Samantha J. Clevenger a, Claudia R. Benitez-Nelson b, Montserrat Roca-Martí c, d, e, Wokil Bam c, Margaret Estapa f, Jennifer A. Kenyon a, Steven Pike c, Laure Resplandy g, Abigale Wyatt g, Ken O. Buesseler c, * a MIT-WHOI Joint Program in Oceanography, Applied Ocean Science and Engineering, Cambridge, MA, United States b School of the Earth, Ocean and Environment, University of South Carolina, Columbia, SC, United States c Department of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, Woods Hole, MA, United States d Department of Oceanography, Dalhousie University, Halifax, Nova Scotia, Canada e Institut de Ciència i Tecnologia Ambientals (ICTA-UAB), Universitat Autònoma de Barcelona, 08193 Cerdanyola del Vallès, Barcelona, Spain f Darling Marine Center, School of Marine Sciences, University of Maine, Walpole, ME, United States g Department of Geosciences, Princeton University, Princeton, NJ, United States A R T I C L E I N F O A B S T R A C T Keywords: The goal of NASA’s EXport Processes in the Ocean from RemoTe Sensing (EXPORTS) project is to develop a Biological carbon pump predictive understanding of the fate of global ocean primary productivity and export of carbon from the surface Carbon export to the deep ocean. Thorium-234 (234Th, t1/2 = 24.1 d) was used to measure sinking particle export from an Thorium anticyclonic eddy during the EXPORTS North Atlantic cruise (May 2021) at the Porcupine Abyssal Plain. The Radionuclides four-week sampling period was broken into three time periods (“epochs”) where 800 234Th seawater samples North Atlantic Ocean spring bloom Diatom bloom were collected from over 50 CTD casts with high depth resolution over the upper 500 m. Size-fractioned par­ Biogenic silica export ticulate samples were collected to determine particulate organic carbon (POC) and biogenic silica (bSi) to 234Th Porcupine Abyssal Plain ratios using in situ McLane pumps. A 234Th non-steady state model shows an eddy center epoch average pro­ gression of increasing 234Th export (~2800 ± 300 (Epoch 1; standard deviation) to 4500 ± 700 (Epoch 3) dpm m− 2 d− 1) out of the top 110 m of the water column over the course of the cruise (29 d). This translates into an epoch average progression of ~11 ± 1 to 14 ± 2 mmol C m− 2 d− 1 of sinking POC flux, and ~ 3 ± 1 to 6 ± 1 mmol bSi m− 2 d− 1 of sinking bSi flux to deeper waters at 110 m. The overall efficiency of the biological carbon pump (amount of net primary production reaching 100 m below the euphotic zone) increases from ~10% to ~30% throughout the sampling period. The temporal trends discussed extensively in this paper show that POC and bSi export increase during diatom bloom evolution. 1. Introduction cold, dissolved inorganic carbon-rich water to depth at high latitudes, the BCP is responsible for export of particulate organic carbon (POC) to The biological carbon pump (BCP) is an important component of the depth, largely via gravitational settling referred to as the biological natural global carbon cycle, as it helps control and regulate carbon di­ gravitational pump (BGP) (Sarmiento and Gruber, 2006; Buesseler et al., oxide (CO2) uptake from the atmosphere into the ocean. It is estimated 2007; Boyd et al., 2019). More recently, active injection of particles to that the ocean acts as a sink for up to one-third of global anthropogen­ depth via physical mediation (e.g., subduction) or biological mediation ically emitted CO2 (Friedlingstein et al., 2022), with approximately 1% (e.g., diel vertical migration) has been considered as an additional ultimately reaching the seafloor on average (Feely et al., 2001). The pathway of carbon sequestration and burial, termed the particle injec­ processes controlling the magnitude of sequestration and burial canon­ tion pump (PIP) (Lévy et al., 2013; Omand et al., 2015; Jónasdóttir et al., ically include both the BCP and the solubility pump (Volk and Hoffert, 2015; Dall’Olmo et al., 2016; Stukel et al., 2017; Boyd et al., 2019). 1985; McKinley et al., 2016). While the solubility pump mostly transfers Despite the seemingly small total burial of ~1%, atmospheric CO2 * Corresponding author at: Woods Hole Oceanographic Institution, Mail Stop 25, 266 Woods Hole Rd, Woods Hole, MA, United States. E-mail address: kbuesseler@whoi.edu (K.O. Buesseler). https://doi.org/10.1016/j.marchem.2023.104346 Received 27 May 2023; Received in revised form 23 November 2023; Accepted 3 December 2023 Available online 12 December 2023 0304-4203/© 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by- nc-nd/4.0/). S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 relative to preindustrial levels would be ~50% higher than it is now if predominantly in the dissolved form as UO2(CO3)4− 3 and is conserva­ not for the current strength of the BCP (Sarmiento and Toggweiler, tively distributed throughout the water column with respect to salinity 1984; Parekh et al., 2006; Sanders et al., 2014). In the total absence of (Langmuir, 1978; Santschi et al., 2006; Owen et al., 2011). Given their the BCP-mediated BGP, models predict that atmospheric CO2 could be half-lives, if there were no particles in the ocean, 238U and 234Th would nearly twice as high (Maier-Reimer et al., 1996). Even subtle changes in be in secular equilibrium in seawater (i.e., their activities would be the magnitude of the BCP and the depth of remineralization – and equal). However, when particles are present, primarily through bio­ therefore the magnitude of the downward transport of carbon from the logical activity, 234Th is rapidly scavenged. When those particles sink, surface to the deep ocean – can have a large impact on global CO2 levels they remove 234Th from the water column causing disequilibrium with 238 (Kwon et al., 2009). The North Atlantic in particular is believed to be a U. The net deficit of 234Th can therefore be used to measure the flux of 234 critical component of the global BCP due to its extensive productivity Th on passively sinking particles (i.e., biogenic material) out of a and bloom dynamics (e.g., Buesseler et al., 1992; Robertson et al., 1993; given ocean depth (e.g., Cochran and Masqué, 2003; Verdeny et al., Alkire et al., 2012; Henson et al., 2019). Thus, the study of the North 2009). Studies using 234Th to determine POC flux have been carried out Atlantic BCP system is crucial to understanding the current state of the since the early 1990’s (seminally in Buesseler et al., 1992) following global carbon cycle and its future changes. Here, we use the radioactive initial studies focusing exclusively on rates of particle cycling beginning isotope thorium-234 (234Th, t1/2 = 24.1 d) to quantify the downward in the 1960’s (e.g., Bhat et al., 1969; Li et al., 1979; Kaufman et al., 1981; flux of POC and biogenic silica (bSi) associated with sinking particles Bacon and Anderson, 1982; Bacon et al., 1985; Coale and Bruland, 1987; throughout an anticyclonic eddy during the North Atlantic bloom, as Nozaki et al., 1987; Huh et al., 1989). Thorium-234 use has more part of NASA’s EXport Processes in the Ocean from RemoTe Sensing recently been extended to allow for flux analyses of other species beyond (EXPORTS) program. POC, including pollutants, metals, and bioelements (e.g., Gustafsson The spring bloom in the Northeast Atlantic Ocean is an annually et al., 1997; Weinstein and Moran, 2005; Black et al., 2018; Black et al., recurrent event, hypothesized to be initiated by three main factors: (1) 2020; Tesán Onrubia et al., 2020; Roca-Martí et al., 2021). the critical depth hypothesis wherein springtime stratification of the upper ocean shoals the mixed layer (Savidge et al., 1995; Bury et al., 2. Study overview and analysis context 2001; Henson et al., 2009; Martin et al., 2011), thereby fueling pro­ duction that exceeds loss (Riley, 1942; Sverdrup, 1953); (2) critical Samples were collected during the EXPORTS 2021 field campaign in turbulence hypothesis wherein a bloom is initiated in an arbitrarily deep the vicinity of the PAP site (approximately 48.5◦ N, 16.3◦ W) in the larger layer with changes in mixing intensity (Huisman et al., 1999), and; (3) context of a North Atlantic spring bloom. This served as the “strong” BCP the dilution-recoupling hypothesis wherein decreasing grazing pressure endmember compared to the EXPORTS 2018 campaign in the North controls the bloom onset (Behrenfeld, 2010; Behrenfeld et al., 2013). Pacific Ocean Station Papa (OSP) site (Siegel et al., 2016; Buesseler Reviews of these three hypotheses at our study site, Porcupine Abyssal et al., 2020b). Sampling followed an anticyclonic eddy during late Plain (PAP), can be found in Rumyantseva et al. (2019) and Binetti et al. spring 2021, capturing a transition in the spring bloom. The goal of (2020), both showing that the critical depth hypothesis is most following a Lagrangian (i.e., retentive) eddy, as opposed to random responsible for bloom initiation. In the North Atlantic Ocean, production sampling of an eddy field, was to capture a time series-style analysis of is at first dominated by fast-growing diatoms and coccolithophores, until the biological and chemical dynamics – and ultimately carbon export – surface silicate is depleted, at which point the system shifts to smaller occurring in an eddy. Details on finding and tracking a strong, retentive flagellates and picophytoplankton (Lochte et al., 1993; Sieracki et al., eddy for this field study can be found in Erikson et al., 2022. Samples 1993; Leblanc et al., 2009). Our study site, PAP, is considered repre­ were classified as coming from the eddy center (EC, < 15 km from sentative of the overall open ocean within the North Atlantic Drifting autonomous asset-defined eddy center, see diamonds in Fig. 1) and non- Province (Longhurst, 1991), and the PAP Sustained Observatory has eddy center (NEC, > 15 km from eddy center, see circles in Fig. 1). Data produced much of our contemporary understanding of temporal varia­ will be presented in the context of these two groups. A 15 km radius was tion in the deep sea, as well as links between the surface ocean and identified by Johnson et al. (2024) as the radius wherein 86% of casts seabed, over the past three decades (Hartman et al., 2021). Prior studies taken had a stable and consistent temperature and salinity regime, and is have shown that during the sampling time period at the late stages of the therefore what the majority of the EXPORTS program considers the eddy spring bloom, the PAP site is dominated by diatoms and haptophytes, center in analysis. although the overall phytoplankton community during the spring bloom Extensive sampling of 234Th occurred on the RRS Discovery (DY131; is highly dynamic (Leblanc et al., 2009; Smythe-Wright et al., 2010). 01–29 May; hereafter referred to in text as the “Survey” ship or data), Much of the organic carbon at PAP is exported as large, fast sinking both in and outside of the eddy center and on the edges of the eddy particles accessible to large zooplankton, although prokaryotes are (Fig. 1). The Survey cruise took place as part of a three-ship operation, estimated to be responsible for 70 to 92% of estimated remineralization also including the RRS James Cook (JC214; NASA EXPORTS; 04–29 May; below the surface (~50 m) (Giering et al., 2014). Consideration of hereafter referred to as the “Process” ship), and the R/V Sarmiento de community composition to understand surface carbon export efficiency Gamboa (SG2105; WHOI Ocean Twilight Zone Project; 05–19 May; is important, as export is highly variable depending on the coupling of hereafter referred to as the “Trophic” ship). The Process ship sampled primary producers and upper ocean remineralization by zooplankton exclusively within the EC, while the Trophic ship sampled a wider sur­ and bacteria (Henson et al., 2019; Koski et al., 2020). Indeed, at PAP vey of the eddy (both EC and NEC locations) similar to the Survey ship. biological productivity and temperature account for about 77% of the The EXPORTS 2021 field study was broken into three “epochs,” with annual change in pCO2 (Macovei et al., 2020). These biological dy­ sampling dates of: 04–10 May (Epoch 1; E1), 11–20 May (Epoch 2; E2), namics have been shown in prior studies to lead to high export effi­ and 21–29 May (Epoch 3; E3). These epochs repeated sampling over a ciencies, with an Ez-ratio (FluxEz/NPP) ranging from 0.2 to 0.4, and flux period of time and are representative of sinking particle propagation transmission to 100 m below the euphotic zone (FluxEz+100/FluxEz) down to 500 m assuming a particle sinking rate of 50–100 m d− 1 (i.e., ~ ranging from 0.5 - > 1.0, where NPP is net primary productivity, Ez is 5–10 d; Siegel et al., 2016). Epochs of approximately these lengths were the depth of the euphotic or sunlit zone of the ocean, and the FluxEz is delineated by two large storms that disrupted sampling. While the flux at the Ez (Buesseler and Boyd, 2009; Buesseler et al., 2020a). storms changed the sampling system, to be discussed extensively, the The 234Th approach to measuring particle export is based upon the beginning of the large storms marked the ends of each epoch as sampling disequilibrium between the particle-reactive 234Th and its soluble, was paused. Therefore, E2 and E3 are representative of the system as naturally occurring radioactive parent uranium-238 (238U, t1/2 = altered by the most recent large storm. 4.47⋅109 y) in seawater. In oxygenated seawater, 238U exists Vertical mixing in response to storms was apparent in potential 2 S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 Fig. 1. Composite daily chlorophyll (mg m− 3 ) maps are shown from 3 days representing Epochs 1 (2021-05-01), 2 (2021-05-18), and 3 (2021-05-27). The purple square is the eddy center as determined by autonomous assets on that particular day. Symbols for casts (circles = NEC, diamonds = EC) and pumps (X’s, concurrent with thorium casts) are shown, as a function of distance in degrees from the eddy center on the day that they were taken. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) density, saturated oxygen (O2 sat), nitrate (NO3) and dissolved silica 3. Methods (dSi) data in the eddy center (Johnson et al., 2024). Mixed layer O2 sat throughout the cruise was >100%, consistent with supersaturation and 3.1. Sample collection and processing biological production. With each storm, surface water was mixed with deeper, less oxygenated waters. Nitrate levels in the mixed layer 3.1.1. Total 234Th seawater samples increased after each storm period, but decreased between storms, sug­ During the cruise, 800 234Th seawater samples were collected over gesting that storms generally mixed nutrients upwards where they were 50 CTD casts with high depth resolution (12–17 samples) from the then depleted. Mean NO3 values with standard deviation per epoch were surface to 500 m. Sampling and analysis for total (dissolved + particu­ 4.86 ± 0.11 (E1), 4.74 ± 0.38 (E2), and 5.52 ± 0.27 μmol L− 1 (E3). late) 234Th was carried out in accordance with the method described in Mean SiO4 values with standard deviation per epoch were 0.24 ± 0.11 Clevenger et al., 2021. Seawater was collected directly from the ship­ (E1), 0.52 ± 0.31 (E2), and 1.45 ± 0.21 μmol L− 1 (E3). Nutrient data board rosette Niskin bottles equipped with a Seabird CTD system. They cited here are from autonomous gliders and seawater measurement, and were then immediately acidified to pH = ~ 1 and spiked with 1 mL 230 detailed further in Johnson et al. (2024), and are consistent with Th (~ 50 dpm g− 1, dpm = disintegrations per minute) to be used as a nutrient budget findings in Meyer et al. (2023). Data from deployed yield monitor. Following a 6-h equilibration time, sample pH was gliders show a deepening of the mixed layer on average throughout the increased to a pH = 8–9 and reagents were added to form a manganese cruise, wherein the E1 average was 26 m, E2 was 40 m and E3 was 50 m oxide precipitate, which preferentially scavenges 234Th and 230Th while for casts considered in this study. Furthermore, concentrations of high leaving 238U in solution. Samples were filtered 8 h after precipitation chlorophyll in the eddy center (1.04 ± 0.14 mg m− 3) deepened after E1 onto 25 mm quartz microfiber filters (QMA), and mounted for beta and the first storm, but dissipated in intensity throughout E2 (0.93 ± counting at sea once dry. 0.10 mg m− 3) to ultimately be <1 mg m− 3 throughout the water column Mounted samples were analyzed to ≤3% counting error on one of 6 by E3 (0.66 ± 0.06 mg m− 3). anticoincidence, 5-position, low-level Risø beta counters with low The relatively high nitrate concentrations but low silicate concen­ background activity (average 0.25 counts per minute, cpm). Beta trations indicate that the bloom was dominated by siliceous organisms, counters were calibrated using low-energy 238U standards, rotated with nitrate remaining to support an additional bloom of non-silicified through each of the 30 detectors to obtain detector-by-detector effi­ phytoplankton (Sieracki et al., 1993; Johnson et al., 2024). Although ciencies. Six of the 30 detectors differed from an average normalized siliceous organisms other than diatoms exist, three key findings suggest detector efficiency by 4 to 5%, and were adjusted accordingly (Cle­ that we were indeed sampling the demise of a diatom-dominated bloom: venger et al., 2021). For further calibration, 5 deep water (2500 m) (1) the low proportion of PIC (< 10%) in the overall carbon pool (this samples were rotated through each detector to obtain an overall counter study), (2) HPLC results show diatom-associated fucoxanthin dominated efficiency. Thorium-234 and 238U are assumed to be in secular equilib­ the community at the beginning of the sample period (Meyer et al., rium by this depth. The 5 deep water samples, counted on all 30 de­ 2023), and (3) diatom representation in the organism pool viewed via tectors, had an average activity of 2.44 ± 0.13 dpm L− 1. onboard imaging (H. Sosik, personal comm.). Later in the cruise, how­ After samples were allowed to decay for >6 half-lives (at least 144 ever, a modified community of smaller, more diverse phytoplankton days), they were recounted to obtain a final sample count rate (cruise emerges. This includes dinoflagellates and haptophytes (Meyer et al., average = 0.50 ± 0.06 cpm), which was subtracted from the first count. 2023; H. Sosik, personal comm.), with the ratio of small cell to large cell The final background represents the counter background as well as the phytoplankton equalizing (Cohn et al., 2023), consistent with previous activity from non− 234Th beta-emitting radionuclides carried by the studies of bloom progression (e.g., Lochte et al., 1993; Henson et al., manganese oxide precipitate. This net cpm value was corrected for 234Th 2012; Sundby et al., 2016). The demise of a diatom bloom leading into a decay and ingrowth, as well as sample volume and method yield to shift to a smaller, more diverse community structure (Cohn et al., 2023; obtain a 234Th activity in dpm L− 1. The yield monitor correction was Meyer et al., 2023) is the biological context driving the evolution of carried out on each sample in accordance with the method detailed in export to be discussed here. Clevenger et al., 2021, adapted from Pike et al., 2005. The cruise average yield monitor recovery value was 92 ± 8%. Errors reported include first counting error, recovery error, and final counting error. All resulting 234Th activity values are reported in the SeaBASS Data Re­ pository (https://seabass.gsfc.nasa.gov/archive/WHOI/buesseler/ EXPORTS/EXPORTSNA/archive/). 3 S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 3.1.2. Thorium-234, carbon and biogenic silica on size-fractionated HCl; Brzezinski and Nelson, 1989) and spectrophotometric detection particles collected using in situ pumps (UV–Visible Spectrophotometer, Genesys 10S, Thermo Scientific). In order to obtain element to 234Th ratios (hereafter referred to as The average of all process blanks was subtracted from total 234Th, element/Th), size-fractionated particles were collected using battery- PC, PIC, and bSi measurements. Process blanks were < 3% of the total 234 powered in situ pumps (Large Volume Water Transfer System, McLane Th, PC and bSi measured on the filters (Ag filters: 0.07 ± 0.10 dpm 234 Industries) coupled to two 142-mm diameter mini-MULVFS (Multiple Th; 1/3 Ag filters: 0.58 ± 0.49 μmol C, 0.04 ± 0.06 μmol Si; QMA Unit Large Volume in-situ Filtration System) filter holders (Bishop et al., punches: 0.06 ± 0.09 dpm 234Th, 0.56 ± 0.45 μmol C; 1/8 Supor filters: 2012; Lam et al., 2015). One filter holder was loaded with three Nitex 0.02 ± 0.02 μmol Si). Particulate inorganic carbon process blanks screens (335, 51 and 5 μm nominal pore size) followed by a pre- accounted for 15% of the measured PIC on 1/3 Ag filter (i.e. screen combusted QMA filter (~1 μm nominal pore size). Size-fractionated rinses; 0.70 ± 0.37 μmol C) and 28% of PIC on the QMA punches (0.25 particles from this 4-tier filter system were analyzed for 234Th, partic­ ± 0.03 μmol C). Triplicate QMA punches from three samples (20, 175, ulate carbon (PC), particulate inorganic carbon (PIC), bSi, as well as 500 m) were analyzed for PC and PIC, obtaining a relative standard other analytes (i.e., particulate nitrogen, phosphorus, pigments, deviation of 4–6% and 4%, respectively. These values indicate that 210 Po/210Pb, 228Th, Ba isotopic composition, RNA, and other organic particle distribution was relatively even across the QMA filters and compounds). The other filter holder was equipped with a 51 μm Nitex support the reproducibility of the subsampling method (Maiti et al., screen followed by a polyethersulfone membrane filter (Supor, 0.8 μm 2012). Particulate organic carbon concentrations were obtained from nominal pore size) used for bSi (analyzed only on the Supor filters), and the difference between blank-corrected PC and PIC results (PIC averaged other analytes (i.e., trace metals and organic compounds that will be ≤10% of PC in all size fractions). Biogenic silica for the 1–5 μm size reported elsewhere). Nitex screens and Supor filters were acid-leached fraction was obtained from the difference between blank-corrected bSi prior to the cruise. from the 0.8–51 μm (Supor filters) and the 5–51 μm (5 μm Nitex screens) Seven to nine pumps were each deployed for a total of eight casts (1 size fractions. Data for POC and bSi have associated uncertainties in E1, 3 in E2, and 4 in E3). In E1 and E2, pumps were deployed at 20, derived from the blank corrections, and particulate 234Th from the blank 50, 75, 125, 175, 330 and 500 m, and in E3 at 20, 50, 95, 145, 195, 330, corrections and counting. Size-fractionated particle data from the 2021 and 500 m (in addition to 35 and 70 m in two casts in E3). The deepest 5 EXPORTS expedition are found in the SeaBASS Data Repository https:// depths in all epochs coincided with the depths of sediment traps seabass.gsfc.nasa.gov/archive/WHOI/buesseler/EXPORTS/EXPORTS deployed in the eddy core by the Process ship (Estapa et al., 2023). A NA/archive/). pressure sensor attached to the deepest pump confirmed depths. Pumps were programmed to sample for four to five hours at a starting flow rate 3.2. Thorium-234 data analysis of 6 L min− 1 and pumped on average 1000 L. Approximately 60 samples for each size fraction were collected. Additional filter holders containing 3.2.1. Thorium-234 activity and steady state flux analysis a full set of filters, but without any active water pumping, were mounted Generally, 234Th flux can be calculated using a 234Th activity balance on the shallowest (20 m) or deepest (500 m) in situ pump to obtain described by the following equation: process (or dipped) blanks (n = 7; Lam et al., 2015). On the process blanks, a prefilter (0.2 μm Supor) was placed on top of the 335 μm Nitex d 234 Th 238 screen to avoid the collection of particles while filters were exposed to = Uλ234 Th − 234 Thλ234 Th − E + V (1) dt seawater. After recovering the in situ pumps, the prefilter was discarded and the process blanks were processed in the same manner as the sam­ where 234Th (dpm L− 1) is the total 234Th activity, 238U (dpm L− 1) is the 238 ples. Immediately after recovering the pumps, residual water from each U activity, λ234 Th (0.0288 d− 1) is the decay constant of 234Th, E (dpm filter holder was removed by vacuum to avoid particle loss. L− 1 d− 1) is the net export of 234Th from adsorption onto sinking parti­ Filter processing was conducted in a clean-air bench placed in a cles, and V (dpm L− 1 d− 1) is the sum of advective and diffusive fluxes of HEPA-filtered work space following the protocols used during the EX­ 234 Th. If we assume a steady state (SS) case where d 234 Th/dt = 0 and PORTS North Pacific cruise (Buesseler et al., 2020b; Roca-Martí et al., assume that advective and diffusive fluxes are negligible compared to 2021). Zooplankton that were not part of the passive sinking flux production and decay rates (V = 0), the downward flux of 234Th on (“swimmers”) and were visible to the naked eye were handpicked from particles can then be determined by integrating the deficit of 234Th the 335 μm screens using forceps. Particles were gently rinsed off the relative to 238U over a given depth (z), in accordance with the following Nitex screens onto 25 mm diameter 1.2 μm pore-size silver (Ag) filters equation: using 0.1 μm prefiltered refrigerated seawater from 500 m depth. At five ∫z( stations the entire screens were rinsed, whereas at three stations the ( )) E= λ234 Th 238 U − 234 Th dz (2) screens were cut in half using a stainless-steel rotary blade (45 mm) to 0 allow for further analyses of geochemical constituents (Cutter et al., 2017). QMA filters were subsampled for 234Th, PC, and PIC using cir­ 3.2.2. Thorium-234 non-steady state analysis cular punches of 22–25 mm diameter. QMA and Ag filters were dried in In some situations, when repeat sampling allows for time-series a 50–60 ◦ C oven, beta counted at sea for 234Th activities and recounted measurements of the same water mass, a non-steady state (NSS) case six months later on shore for background activities. After counting, must be considered where d 234 Th/dt ∕ = 0 (Savoye et al., 2006). Note screen samples that had been rinsed onto Ag filters were split by weight that in these cases, advective and diffusive fluxes are still most often (Lamborg et al., 2008), and ~ 1/3 of each sample was analyzed for PC, considered negligible. A NSS model is needed when there are increases PIC and bSi. Biogenic silica was also measured on a subsection (1/8) of or decreases in export that cause a change in 234Th activities versus time the Supor filters (kept frozen after collection until processing). Partic­ within the time frame of 234Th’s half-life (24.1 d). The SS assumption ulate carbon was analyzed using high-temperature combustion on a can underestimate 234Th removal when there is intense scavenging, or Costech 4010 Elemental Analyzer. Particulate inorganic carbon was can overestimate 234Th removal during post-bloom conditions (Buess­ determined by coulometric titration using a CO2 coulometer (model eler et al., 1992; Ceballos-Romero et al., 2018). Based on both the CM5014, UIC Inc.). Each sample was placed in a glass vial secured to a sampling scheme of the 2021 EXPORTS cruise discussed in Section 2, CM5130 acidification module and treated with 2 mL of 1 N H3PO4 to and the rapidly changing bloom conditions during the sampling period, liberate the CO2 into a closed system to be measured (Honjo et al., the data presented here is an ideal candidate for NSS analysis. 1995). Biogenic silica was determined by following the 0.2 M NaOH A number of factors specific to this study’s data and previous liter­ digestion method (40 min at 95 ◦ C followed by neutralization with 1 M ature support the use of a NSS model over a SS model. The North Atlantic 4 S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 bloom represents a dynamic biological ecosystem with changes on a the specific depth of each sample had its own d 234 Th/dt value derived timescale less than the 234Th half-life (Brown et al., 1985; Esaias et al., from linear fits between the overall bin d 234 Th/dt values (see Supple­ 1986; Buesseler et al., 1992; Savoye et al., 2006; Ceballos-Romero et al., mental Fig. 1 for depth profiles of NSS terms and resulting linear fits). 2018), suggesting that an NSS model should be used. Previous studies of In this study, we derived NSS flux according to the following modi­ this PAP site during a bloom (e.g. Buesseler et al., 1992; Le Moigne et al., fication of Eq. 1: 2013) show that, generally, 234Th flux values are >800 dpm m− 2 d− 1, [ ( 234 ) ] which is the flux threshold over which Savoye et al., 2006 suggests ( ) d ThSpecific FluxNSS = 238 U − 234 Th λ234 Th − *1000 *depth integral reoccupation of a sample site for NSS analysis. Indeed, analysis of the dt data in this experiment revealed 234Th flux values that are significantly (3) lower than previous studies when calculated using the simple SS model. Further, 234Th activities when considered in distinct depth bins clearly where the depth integral is in (m) units, and the d 234 Th/dt term is the decrease with time (example shown in Fig. 2) indicating that the NSS sample-specific term described above. The above equation allows for analysis of this data is more appropriate than SS analysis. derivation of NSS fluxes for distinct depth bins, and then derivation of It was decided to break casts into two groups for a Lagrangian time- NSS fluxes throughout the entire water column. This results in high series analysis: eddy center (EC; < 15 km from eddy center; n = 23), and resolution, sample-by-sample derivation of NSS fluxes. non-eddy center (NEC; > 15 km from eddy center; n = 27), although a number of iterations of groupings were considered (see Supplemental 234 234 Information). Once groups were chosen, 234Th activities from each cast 3.3. Physical transport effects on Th fluxes to validate Th analysis were averaged within 9 depth bins to 500 m – the same bins for each of the two groups – and d 234 Th/dt values were calculated using a linear fit The SS equation described in Section 3.2.1 (Eq. 2) is a 1-D accounting of 234Th activity values over time from the first cast. Fits and corre­ that ignores the impacts of physical transport and temporal variability in 234 sponding errors (uncertainty within one standard deviation) across the 9 Th activities by assuming these terms are relatively small. However, depth bins for both groups can be found in Table 1. An example of the to assess the uncertainties tied to this assumption, we examined 234Th 234 export inside a simulated tracked eddy using an idealized double gyre, Th activity versus time plot used to calculate d 234 Th/dt values is submesoscale permitting physical model coupled with a simplified shown in Fig. 2, using a set of data (20–50 m depth bin in the EC biogeochemical model with explicit 234Th cycling (Resplandy et al., grouping) that is representative of the trend found in all depth bins for 2012; Resplandy et al., 2019). In-brief, the physical model was run at a each of the two groups, particularly in the upper water column where 1/54 degree (2 km) horizontal resolution and forced using climatolog­ the relationship is the strongest. Beyond the derivation of discrete ical atmospheric forcing. A lower resolution output (1/9◦ ) from this d 234 Th/dt values in the 9 depth bins chosen, linear fits between the physical model was used to run the offline biogeochemical model d 234 Th/dt values at the midpoint depth of each of the 9 depth bins were (LOBSTER) to calculate the biogeochemical cycling (temporal compo­ produced, and the resulting linear equations were used to define a nent) of nitrate, ammonium, one phytoplankton, one zooplankton, and specific d 234 Th/dt for each sample at any given depth. In other words, two size-classes of detritus. This lower resolution of the physical model was previously shown to have an almost indistinguishable impact on the Fig. 2. An example of the 234Th activity versus day from the first sample, used to derive dTh/dt for non-steady state calculations. This specific set of data is the 20–50 m depth bin for the eddy center (EC) grouping. The linear decrease in activity versus time in a particular depth bin is indicative of a system that is not in steady state. 5 S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 Table 1 Non-steady state (NSS) terms at 9 depths bins for the EC and NEC groups, derived from linear fits of 234Th activity (dpm L− 1) over time (d) for the entire duration of the sampling period. Depth profiles of NSS terms can be seen in Supplemental Fig. 1. EC Depth Bins (m) 0–20 20–50 50–75 75–95 95–125 125–150 150–200 200–300 300–500 dTh/dt (dpm L-1 d-1) − 0.0254 − 0.0287 − 0.0278 − 0.0231 − 0.0223 − 0.0077 − 0.0063 − 0.0064 − 0.0045 Error 8% 6% 7% 10% 8% 28% 30% 33% 45% R2 0.54 0.58 0.49 0.42 0.39 0.34 0.23 0.13 0.1 NEC Depth Bins 0–20 20–50 50–75 75–95 95–125 125–150 150–200 200–300 300–500 dTh/dt (dpm L-1 d-1) − 0.0217 − 0.0224 − 0.0239 − 0.0352 − 0.0368 − 0.0142 − 0.0109 − 0.0029 − 0.0013 Error 12% 9% 9% 8% 6% 19% 22% 87% 194% R2 0.41 0.45 0.38 0.50 0.50 0.22 0.32 0.07 0.01 biological tracers. Explicit 234Th cycling is included per Resplandy et al. the center of a coherent eddy. [2012], which results from the processes of radioactive decay, particle The real-world cruise system was influenced by a number of storms adsorption/desorption, ecosystem cycling, detritus aggregation/disag­ (storm days making up 42% of cruise period), causing entrainment ve­ gregation, and remineralization. The model solution is not specific to the locity (wE = ∂∂Ht where H is MLD) to range from 10 to 35 m d− 1, and cruise period (forcing is climatological) but it captures the North surface core waters (defined by the depth of the 27.1 isopycnal) to ac­ Atlantic bloom dynamics and reproduce the range of 234Th-based export count for 40% of the T and S variability observed by autonomous assets in the region (see further details and evaluation in Resplandy et al., 2012 within the eddy center (Johnson et al., 2024). Analysis of tracer flux and Resplandy et al., 2019). suggests that advection played a larger role in particulate movement From the model output, a mesoscale, anticyclonic simulated-eddy dynamics than entrainment did, whereas entrainment played a larger (sim-eddy) similar to the cruise eddy and located in the subpolar gyre role in nutrient concentrations (Johnson et al., 2024; Meyer et al., was identified and tracked within the model using closed contours of sea 2023). Although the physical effects of advection and entrainment surface height (SSH) anomalies. The sim-eddy was roughly ½ degree (50 certainly contribute to uncertainty and variability of the results of this km) in diameter and thus comparable in size with the cruise eddy (core study, particularly in the surface waters, this sim-eddy comparison is diameter 30 km; Johnson et al., 2024). In addition, modeled sub-surface useful to obtain a baseline assessment of the impact of physical transport (~75 m depth) horizontal 234Th gradients in the vicinity of the sim-eddy in an ideal system within the EC waters. reached ~0.3 dpm km− 1, similar to the observed cruise eddy gradients The physical transport terms around the edges, outside of the 15 km (~0.2 dpm km− 1), calculated between 75 m-depth cruise samples radius EC boundary, are expected to be somewhat larger (Resplandy separated by a minimum of 1 km and a maximum of 15 km, and taken et al., 2012). Given that derivation of these terms specific to the cruise- within two days of each other (similar to the model output frequency). eddy is outside of the scope of this paper, discussion will focus largely on The sim-eddy core was defined as a 2 × 2-degree box at the center of the data from the EC to minimize the uncertainties associated with physical SSH contour, which captures a radius of 12.5 km from the eddy center transport on our estimates. Presentation and interpretation of NEC data (similar to the 15 km radius used to define EC samples in the cruise is a focus of future work. eddy). The 30-day period of the sim-eddy lifespan that most closely paralleled cruise observations was chosen to quantify the contribution of physical transport to the 234Th export flux and evaluate the potential 3.4. Analysis of biological carbon pump characteristics uncertainty associated with the assumption that physical transport is small in the eddy center. Upper-ocean BCP efficiency was assessed by the ratio between POC From model output specific to the eddy center, we determined the export out of the Ez relative to integrated NPP in the Ez (Ez-ratio = percentage of the total simulated 234Th downward flux that was due to FluxEz/NPP) and the transfer efficiency of POC through the upper twi­ physical transport and compared it to the percentage of the NSS term light zone (T100 = FluxEz+100m/FluxEz) (Buesseler and Boyd, 2009). The determined from Eq. 3 above (Table 2). The total modeled physical extent to which POC that is produced is then exported out of the sunlit transport was 11% of the total NSS 234Th flux at 75 m, increasing to 27% zone, and then how much reaches deeper in the water column, is of the total NSS 234Th flux at 550 m. These percentages are generally controlled at least by the following factors: the amount of primary within the error of the total flux measurements presented in this study, production (De La Rocha and Passow, 2007), particle sinking velocity supporting the assumption that the physical transport term is small in (De La Rocha and Passow, 2007), the rate of remineralization (Le Moigne et al., 2016), mineral ballasting of particles that influences particle sinking rates (De La Rocha and Passow, 2007; Le Moigne et al., Table 2 2014), injection of POC to depth by vertical migration among other Comparison of data from a simulated eddy model, to actual data from this study processes (Boyd et al., 2019), and zooplankton feeding (Cavan et al., (EC group, average over all epochs). 2015; Le Moigne et al., 2015). Model Data Study We defined a unique Ez base depth for each cast, rather than a fixed Data depth, allowing for a cast-specific assessment and a more informative comparison of BCP efficiencies (Buesseler et al., 2020a). This is because Depth Total Phys Trans dTh/dt % Depth Total dTh/dt % Flux % of Total of Total Flux of Total the Ez base depth often differs between sampling sites. Even small dif­ ferences in Ez depths can affect comparisons of BCP efficiencies relative 75 2721 11 43 75 2903 78 91 3133 11 44 90 3309 81 to organic matter production and gradients in POC flux just below the Ez 135 3739 14 39 125 3720 88 (Buesseler et al., 2020a). Here, in order to compare data to previous 170 4044 15 36 175 3721 97 work and conduct a cast-specific analysis, we calculated the BCP 335 5013 21 37 330 3786 80 characterizing-metrics, the Ez-ratio and T100, using both a fixed 100 m 550 5398 27 43 500 3795 73 Ez depth and cast-specific Ez depth definitions based on in situ 6 S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 fluorescence data. For depth-integrated NPP estimates, the depth of the If we applied a SS 234Th model for all data, regardless of location, base of the Ez was defined as 0.1% of the maximum photosynthetically epoch average 234Th flux would range between approximately − 3200 ± available radiation (PAR0.1) (J. Fox, personal communication). Depth- 700 to 1700 ± 200 dpm m− 2 d− 1 (second row, Fig. 3 and Supplemental integrated NPP values were calculated using discrete carbon uptake Fig. 2; Supplemental Table 1). To find such an excess of 234Th measurements from 14C analysis of multi-day incubations, and are as throughout the subsurface would mean there is some constant source of follows: 90 ± 10 (E1), 82 ± 13 (E2), and 54 ± 9 mmolC m− 2 d− 1 (E3). 234 Th that is unsustainable given the lack of an equal deficit above. The Depth profiles of NPP were integrated using trapezoidal integration to SS component in this case describes an imbalance between surface 0.1% light levels using Compact-Optimal Profiling System (C-OPS) es­ export and mesopelagic accumulation of 234Th. Using an NSS model timates of PAR % depths. However, 234Th sample collection was often accounts for variations over the course of the survey, where mesopelagic done during the night with significant cloud cover. Sample timing accumulation is low from E1 to E3, and the excess of 234Th observed at combined with sensor malfunctions resulted in only 46% of sampled E1 gradually decreases towards secular equilibrium. It could be hy­ casts having a known PAR0.1 value. Therefore, as a proxy for system­ pothesized that the SS results observed in E1 are inherited from export atically defining the Ez for each cast we used the primary production and accumulation of 234Th that may have occurred before our cruise zone (PPZ), defined as the depth at which the signal from in situ fluo­ period. We provide average SS flux values per epoch in Supplemental rescence sensors dropped to 10% of the maximum above that depth Table 1, however, this is for comparative reasons only and should not be (Owens et al., 2015). The 10% definition was chosen by Owens et al. viewed as a lower limit on fluxes during EXPORTS. (2015) as a conservative estimate that delineates between surface water When the NSS model is applied, 234Th fluxes increase between where chlorophyll pigment is present, and deeper water where not epochs, and generally increase from cast to cast within each epoch enough light is present to maintain chlorophyll. Table 3 shows NPP and (second row, Fig. 3 (EC) and Supplemental Fig. 2 (NEC)). Cast-to-cast depth horizon (MLD, PPZ, PAR1%, PAR0.1%) values with standard increase for EC casts can be seen in Supplemental Fig. 3, which shows deviations of the all-cast (NEC and EC group inclusive) means. We that variability in the dataset within epochs is typically temporal in considered the PPZ depth to be a boundary below which there is no nature. Epoch-average, maximum 234Th fluxes with standard deviation further production of fresh POC due to photosynthesis. The use of PPZ are as follows: 2900 ± 890 (E1; 125 m), 4100 ± 1760 (E2; 410 m), and allows for comparisons of BCP efficiencies between sites that have 5400 ± 1680 dpm m− 2 d− 1 (E3; 500 m) for EC; and 3000 ± NA (E1; 150 different particle sources, remineralization efficiencies, and weather m), 5400 ± 440 (E2; 225 m), and 6000 ± 800 dpm m− 2 d− 1 (E3; 225 m) patterns. The PPZ is found to be several meters to tens of meters below for the NEC. Average fluxes with standard deviation at 500 m are: 2200 PAR1.0, but is typically closer to PAR0.1 (Marra et al., 2014; Owens et al., ± 1400 (E1), 3800 ± 2000 (E2), and 5400 ± 1680 dpm m− 2 d− 1 (E3) for 2015). As chlorophyll pigments that are used to define PPZ have a EC; and 2000 ± NA (E1), 4500 ± 1680 (E2), and 5400 ± 900 dpm m− 2 lifetime of multiple days (e.g., Lacour et al., 2019), PPZ measurements d− 1 (E3) for NEC. Average fluxes and standard deviations at the all- should be seen as a multiday average boundary for organic matter epoch average PPZ depths (closest depth sampled) are as follows: production (Buesseler et al., 2020a). 2300 ± 270 (E1; PPZ = 67 m), 3300 ± 700 (E2; PPZ = 90 m), 4600 ± 680 dpm m− 2 d− 1 (E3; PPZ = 133 m) for EC; 1300 ± NA (E1; PPZ = 53 4. Results m), 3800 ± 500 (E2; PPZ = 94 m), and 5400 ± 530 dpm m− 2 d− 1 (E3; PPZ = 122 m) for NEC). Average EC NSS flux values per epoch can be 4.1. Thorium-234 flux results found in Table 4, average NEC NSS flux values per epoch can be found in Supplemental Data Table 2, and all NSS flux data per cast can be found Thorium-234 activity with depth by epoch is shown in the top row of in Supplemental Data Table 3. Fig. 3 (EC) and Supplemental Fig. 2 (NEC). Average 234Th activities below the epoch-specific PPZ were consistently larger than 238U activ­ 4.2. Particulate organic carbon and biogenic silica ratios ities in all epochs regardless of location. In a typical SS system, 234Th activity profiles would show a deficit relative to 238U at depth in the Ez The use of POC/Th and bSi/Th ratios in pump samples to determine indicative of export (234Th < 238U), excess of 234Th (234Th > 238U) near element fluxes from 234Th fluxes assumes that the particles in the pump the base of the Ez, and equilibrium (234Th = 238U) in deeper waters. An samples represent the majority of sinking particles at the depth and time excess of 234Th compared to 238U is indicative of a source of 234Th, such of 234Th flux measurement. This assumption fails only if a significant as respiration or remineralization of sinking particles. The overall tem­ fraction of the sinking flux is carried by particles that differ in POC/Th poral trends of average 234Th activities with depth show that zones of and bSi/Th, which we assess by comparing results from pumps and export and remineralization change by epoch with the deepening of the sediment traps deployed at the same depths and times during EXPORTS PPZ. In each epoch average activity profile (first row, Fig. 3 and Sup­ (75–500 m, Fig. 3; shallow pumps were excluded from this analysis). plemental Fig. 2), export (activity deficit) can be seen until reaching The sediment trap data can be found in the SeaBASS Data Repository equilibrium and remineralization (activity excess) starts at approxi­ (https://seabass.gsfc.nasa.gov/archive/Maine/Estapa/EXPORTS/E mately the epoch average PPZ depth. It is noteworthy that this excess XPORTSNA/). The >5 μm pump size fraction (i.e., combination of 5–51, persists through 500 m, the base of the measurement zone, rather than 51–335, and > 335 μm size fractions) rather than the >1–5 μm size reaching equilibrium in a canonical manner. This is indicative of a fraction or a single size fraction (Fig. 4) was used to derive POC and bSi system that is out of steady state. fluxes for the following reasons: (1) bSi/Th ratios in the smallest size fraction (1–5 μm) were an order of magnitude lower than in traps (0.14 ± 0.10 vs. 0.9 ± 0.2 μmol dpm− 1, respectively), whereas 5–51, 51–335, Table 3 and > 335 μm (1.1 ± 0.2, 1.3 ± 0.6, 1.3 ± 0.2 μmol dpm− 1, respec­ Net primary productivity epoch averages. Epoch average depth horizons (mixed tively) were similar to traps (Fig. 4); (2) POC/Th ratios in the >5 μm size layer depth, primary production zone, 1% photosynthetically available radia­ fraction better matched traps relative to the >1 μm size fraction in tion, 0.1% photosynthetically available radiation), averaged over all casts. epochs 1 and 2, and the >5 μm fraction captured the decreasing tem­ Standard deviations of all measurements included. NPP (mmol m− 2 d− 1)is in­ poral trend observed in traps; and (3) while the POC/Th and bSi/Th in tegrated to PAR0.1%. the >5 and > 51 μm size fractions were very similar (Fig. 3), we chose NPP MLD PPZ PAR1% PAR0.1% the >5 μm as it represented a larger fraction of the particulate material. E1 90 ± 10 26 ± 6 65 ± 7 37 ± 1 56 ±2 The pump casts showed temporal variability in POC/Th and bSi/Th E2 82 ± 13 40 ± 17 93 ± 18 51 ± 12 76 ± 17 ratios across epochs (Fig. 5). In studies with less temporal variability in E3 54 ± 9 50 ± 29 127 ± 22 51 ± 4 76 ± 7 the system over the sampling period (e.g., Buesseler et al., 2020b), it has 7 S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 Fig. 3. The 6 panels in this figure show: (Top row) thorium-234 and uranium-238 activity with depth, error bars represent counting and processing error, and (bottom row) thorium-234 flux from both steady state (SS) and non-steady state (NSS) models, error bars represent cumulative error from thorium measurements and error from the NSS analysis. Data is from casts in the eddy center (EC). In all subplots, individual data points show all measurements taken in that epoch, while the line and error bars represent the epoch average. Horizontal dotted lines show the average PPZ of all EC casts for that particular epoch. been possible to use an overall average of all pump casts to obtain depth- deepening of the PPZ and MLD. specific POC/Th ratios. Due to the temporal variability in this system, In the same manner as POC/Th ratios, bSi/Th ratios were derived however, epoch-specific POC/Th and bSi/Th ratios from pump casts from pump data. These bSi/Th ratios (> 5 μm size fraction) can be seen were applied to NSS 234Th flux values to estimate POC fluxes. in Fig. 5, and are notable in that the slope of bSi/Th linear attenuation in Fits and corresponding errors were derived in two parts through the the surface waters becomes less steep from E1 to E2, and bSi/Th ratios depth profiles: (1) at shallower pump depths, ratios decrease linearly remain constant throughout the upper 500 m by E3. Surface bSi/Th (except for bSi/Th in E3), therefore a slope is found with a simple linear ratios decrease by >50% between E1 and E3, while mesopelagic ratios fit and a percent error is calculated from the average individual un­ increase by nearly 50% in that same time. Based on derived linear fits, certainties of the measurements, and (2) at deeper pump depths, data the bSi/Th ratios that were applied to 234Th flux values fall in the were averaged across depths, where error is applied as the standard following ranges: 2.6 ± 0.5 to 0.9 ± 0.1 μmol dpm− 1 (E1), 2.1 ± 0.5 to deviation between those data points. Differences in these can be seen in 1.0 ± 0.1 μmol dpm− 1 (E2), and consistently 1.3 ± 0.1 μmol dpm− 1 Fig. 5. This is consistent with how element/Th ratio analysis has been (E3). carried out in previous 234Th studies (Black et al., 2018; Buesseler et al., Sediment trap-derived ratios are included for comparison purposes 2020b; Kenyon, 2021). Only one set of epoch-by-epoch POC/Th and (Fig. 5, magenta diamonds). The 234Th flux tracer method relies on the bSi/Th values were applied to 234Th flux profiles from the corresponding assumption that pump-derived data is representative of overall sinking epoch, regardless of location in relation to the eddy center. It should be particles, and sediment traps are an alternative means of analyzing noted that ratios from above the MLD are extrapolated from fewer sinking particles. A t-test confirms that at a 5% significance level, the pumps in a regime with complex mixing dynamics, and thus the stron­ POC/Th ratio data is statistically the same in two out of three epochs (E1 gest application of this method is below the PPZ. and E3), and the bSi/Th ratio data is statistically the same in one out of Epoch-by-epoch temporal variation in POC/Th ratios (> 5 μm size three epochs (E2). More importantly for the application of this data, the fraction) derived from McLane pump sampling is most apparent in the largest deviation between corresponding pump and trap values was 39% shallower pump depth linear fits (Fig. 5). The sharp ratio attenuation in of the average value of the two, with the average deviation of all 6 being the upper water column goes from the surface to ~50 m in E1, ~ 75 m in 23%. The differences in pump and sediment trap data could be due to E2, and ~ 95 m in E3, before becoming constant beneath those depths to the breakup of marine snow going through pumps, causing the loss of the deepest sampled depth at 500 m. Further, there are differences in the carbon and thorium associated with smaller particles, thereby rate of decrease in POC/Th versus depth in the upper water column in increasing the bSi/Th ratio compared to trap data. This would affect the three epochs. The fastest decrease versus depth is observed in E2 carbon and thorium, so would not be apparent in the POC/Th ratios (slope = − 6.63), then E1 (− 8.60), then E3 (− 9.28 m dpm μmolC− 1). from pumps. If we consider sediment trap data as alternative bSi/Th This is consistent with the deepening of the MLD and remineralization ratios, bSi fluxes derived from pumps could be overestimated by an all- layer throughout the cruise and supports the decision not to apply a epoch average of 27% with a standard deviation of 17%. However, with single cruise-average POC/Th fit. Based on the derived linear fits, the large errors these results are not unexpected nor are they necessarily POC/Th ratios with error that were applied to 234Th flux values fall in indicative of a failed sinking particle composition assumption. the following ranges: 10 ± 2 to 4.0 ± 0.3 μmol dpm− 1 (E1), 14 ± 3 to 3.2 ± 0.3 μmol dpm− 1 (E2), and 12 ± 2 to 3.1 ± 0.5 μmol dpm− 1 (E3). Note that in E3, the position of the pump targeting the base of the Ez changed 4.3. Particulate organic carbon flux results from 75 m (the depth in E1 and E2) to 95 m to capture the observed Epoch-average particulate organic carbon fluxes with a standard 8 S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 Table 4 Epoch average measurements (non-steady state 234Th fluxes, POC, and bSi fluxes) for the eddy center (EC) group, and associated errors. Epoch Averages EC Depth NSS Flux Error SD POC Flux Error SD bSi Flux Error SD (m) dpm m-2 d- dpm m-2 d- dpm m-2 d- mmol m-2 d- mmol m-2 d- mmol m-2 d- mmol m-2 d- mmol m-2 d- mmol m-2 d- 1 1 1 1 1 1 1 1 1 5 369 69 59 3.6 0.9 0.5 0.9 0.3 0.1 20 853 107 125 6.7 1.4 1.0 2.0 0.5 0.3 35 1243 123 99 7.8 1.5 0.6 2.7 0.6 0.2 50 1698 151 261 7.0 0.9 1.1 3.2 0.7 0.5 60 1747 146 104 7.0 0.8 0.5 3.2 0.7 0.2 75 2292 183 266 9.1 1.0 1.1 3.6 0.8 0.4 90 2607 198 330 10.4 1.2 1.3 3.4 0.8 0.4 110 2810 216 333 11.2 1.2 1.3 3.0 0.7 0.3 125 2911 232 326 11.6 1.3 1.3 2.6 0.3 0.3 150 2884 263 408 11.5 1.4 1.6 2.6 0.3 0.4 175 2787 295 366 11.1 1.5 1.5 2.5 0.3 0.3 195 2689 363 499 10.7 1.7 2.0 2.4 0.4 0.4 225 2709 360 595 10.8 1.7 2.4 2.4 0.4 0.5 260 2855 436 887 11.4 2.0 3.5 2.5 0.4 0.8 275 2256 529 580 9.0 2.2 2.3 2.0 0.5 0.5 330 2478 624 841 9.9 2.6 3.4 2.2 0.6 0.7 410 2402 731 1429 9.6 3.1 5.7 2.1 0.7 1.3 Epoch 1 500 2220 773 1394 8.9 3.2 5.6 2.0 0.7 1.2 5 583 48 101 8.1 1.7 1.4 1.2 0.3 0.2 20 1219 72 222 14.2 2.8 2.6 2.2 0.6 0.4 35 1800 90 336 17.0 3.3 3.1 2.9 0.7 0.5 50 2307 106 484 16.3 3.2 3.6 3.2 0.8 0.7 60 2777 109 590 15.7 3.1 3.0 3.5 0.9 0.7 75 3003 130 680 9.8 1.2 2.3 3.2 0.4 0.7 90 3265 150 709 10.4 1.0 2.3 3.4 0.3 0.7 110 3486 163 716 11.2 1.1 2.3 3.7 0.3 0.7 125 3619 179 758 11.6 1.1 2.4 3.8 0.3 0.8 150 3610 207 767 11.6 1.2 2.5 3.8 0.4 0.8 175 3655 229 829 11.7 1.2 2.7 3.8 0.4 0.9 195 3683 277 922 11.8 1.3 3.0 3.9 0.4 1.0 225 3619 301 1099 11.6 1.4 3.5 3.8 0.4 1.2 260 3778 354 1658 12.1 1.6 5.3 4.0 0.5 1.7 330 3818 510 1491 12.2 2.0 4.8 4.0 0.6 1.6 410 4087 582 1761 13.1 2.2 5.6 4.3 0.7 1.8 Epoch 2 500 3726 643 1949 11.9 2.3 6.2 3.9 0.7 2.0 5 580 54 104 7.1 1.3 1.3 0.8 0.1 0.1 20 1243 83 191 13.4 2.2 2.1 1.6 0.2 0.3 35 1924 104 245 17.6 2.8 2.3 2.5 0.2 0.3 50 2670 122 647 19.9 3.1 4.5 3.5 0.3 0.8 60 2832 130 243 18.5 2.9 1.7 3.7 0.3 0.4 75 3413 147 255 16.7 2.6 1.1 4.5 0.3 0.3 90 4053 172 497 12.7 2.1 1.3 5.4 0.4 0.7 110 4488 187 673 13.8 2.4 2.1 5.9 0.4 0.9 125 4629 203 674 14.2 2.5 2.1 6.1 0.5 0.9 150 4702 234 647 14.4 2.6 2.0 6.2 0.5 0.9 175 4722 259 670 14.5 2.6 2.1 6.2 0.5 0.9 195 4785 306 783 14.7 2.7 2.4 6.3 0.6 1.0 225 4690 323 388 14.4 2.7 1.2 6.2 0.6 0.5 260 4662 408 633 14.3 2.8 1.9 6.2 0.7 0.8 275 5238 441 1270 16.1 3.1 3.9 6.9 0.7 1.7 330 5061 536 1115 15.5 3.2 3.4 6.7 0.8 1.5 410 5054 663 1126 15.5 3.4 3.5 6.7 1.0 1.5 Epoch 3 500 5439 696 1676 16.7 3.6 5.1 7.2 1.0 2.2 deviation range calculated using the NSS are shown in Fig. 6 (EC) and and magnitude of POC flux per epoch, with epoch averaged peak POC the third row of Supplemental Fig. 2 (NEC), and epoch average POC flux fluxes and standard deviations of approximately 12.1 ± NA (E1; 150 m), data can be found in Table 4 (EC) and Supplemental Table 2 (NEC). 17.4 ± 1.4 (E2; 225 m), and 18.3 ± 2.8 mmolC m− 2 d− 1 (E3; 225 m). Within the EC group, epoch averaged maximum POC fluxes with stan­ The depth of the peak for E1 is approximately the same as in the EC dard deviation increased by epoch from approximately 11.6 ± 1.3 (E1; group but becomes deeper in subsequent epochs, as opposed to shoaling 125 m), 17.0 ± 3.1 (E2; 35 m), to 19.9 ± 4.5 mmolC m− 2 d− 1 (E3; 50 m). observed in the EC group. The epoch averaged flux values and standard The depth of this peak flux becomes notably shallower after E1. The deviations at the sample depths closest to the epoch averaged PPZ are as epoch averaged flux values with standard deviation at the sample depths follows: 6.2 ± NA (E1; PPZ = 53 m), 12.0 ± 1.6 (E2; PPZ = 94 m), and closest to the epoch averaged PPZ are as follows: 9.1 ± 1.1 (E1; PPZ = 16.6 ± 1.6 mmolC m− 2 d− 1 (E3; PPZ = 122 m). Data for all casts can be 67 m), 10.4 ± 2.3 (E2; PPZ = 90), and 14.2 ± 2.1 mmolC m− 2 d− 1 (E3; found in Supplemental Table 3. PPZ = 133 m). The NEC group follows approximately the same temporal increase 9 S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 4.5. Biological carbon pump efficiency assessment The Ez-ratio (defined in Section 3.4 as FluxEz/NPP), gives insight into the strength of the BCP. Depth-integrated NPP results with standard deviation are averaged over epoch and are as follows: 90 ± 10 (E1, n = 2), 82 ± 13 (E2, n = 4), 54 ± 9 mmolC m− 2 d− 1 (E3, n = 5), with all samples taken in the eddy center on the Process ship (J. Fox, personal communication; seen in Table 3). Results for Ez-ratio calculations using the PPZ show that the Ez-ratio increases from epoch to epoch regardless of the depth horizon considered, PPZ or 100 m, with the most significant change occurring between E2 and E3 (Fig. 7, Table 5). Results calculated using PPZ for the EC group show an approximate doubling of the Ez- ratio between epoch, and are as follows: 0.08 ± 0.01 (E1), 0.14 ± 0.03 (E2), and 0.26 ± 0.04 (E3). These results indicate that throughout the sampling period, the BCP becomes more efficient at exporting POC out of the PPZ. Considering NPP values and flux values, this increase in efficiency could be a function of both less NPP (decreasing values) and more export (increasing values), both driving greater export versus production over time. However, the Ez-ratio should be considered with caution, as the temporally variable Ez-ratios could also be reflective of a decoupling between export and production, which has been found to cause large variations in Ez-ratios (e.g., Wassmann, 1998; Henson et al., 2015). Transfer efficiencies (T100) are used to examine the extent of remi­ neralization in the first 100 m below the Ez, where prior studies suggest that flux gradients and differences between sites can be large (e.g., Maiti et al., 2010). Transfer efficiencies were calculated using the specific PPZ of each cast, as well as using a fixed 100 m depth for comparison to other studies (Fig. 7, Table 5). The T100 calculated using PPZ depths in the EC group are as follows: 1.55 ± 0.35 (E1), 1.07 ± 0.15 (E2), and 1.07 ± 0.11 (E3). Had we used a 100 m depth to calculate T100, results would be the same within uncertainty with the exception of E1, which is signifi­ Fig. 4. Element/Th (particulate organic carbon, POC, and biogenic silica, bSi) cantly lower (see Fig. 7, Table 5). Transfer efficiency values >1 are ratios are presented here, showing variation of the 75–500 m depth range per indicative of flux increases below the Ez, which could be due to in situ epoch (colors, see legend) and sample size class (x-axis). Similarities between aggregation or repackaging processes occurring in the upper twilight pump-collected size class (> 1, > 5, > 51 μm) compared to ratios from sediment zone, diel vertical migration and active transport injecting particles traps were considered in order to make a decision on which size class is most representative of sinking particles. Whiskers represent the minimum and beneath the Ez, or a temporal lag of earlier higher fluxes (i.e., remnant maximum, boxes represent the first and third quartile, and the line represents high flux at depth is not related to low flux out of surface at the time of the median. Both POC/Th and bSi/Th ratios were considered in this decision, sampling). Regardless of the reason, the transfer of POC downwards and ultimately the >5 μm size class was applied to determine both POC and once it reaches the PPZ is highly efficient and indicates limited impact of bSi fluxes. remineralization below the PPZ. Considering these two metrics, Fig. 8 shows T100 (x-axis) against the 4.4. Biogenic silica flux results Ez-ratio (y-axis), with contour lines showing overall BCP efficiency (i.e., the fraction of NPP that reaches a depth of 100 m below the Ez) and Biogenic silica fluxes were calculated to help assess bloom charac­ circle radii being proportional to the magnitude of NPP. This figure, first teristics, as it was known from various community composition ana­ published in Buesseler and Boyd (2009), has been used in a number of lyses, sediment trap content, and nutrient analyses that diatoms were an subsequent studies (e.g., Siegel et al., 2016; Buesseler et al., 2020a; important component of the bloom and related export fluxes. Siegel et al., 2021) to illustrate variations in BCP efficiencies across Results from bSi flux analysis show a general trend of increasing flux different regions and seasons. As observed in Fig. 8, in this study, the throughout the cruise sampling period (Fig. 6, epoch average with overall BCP efficiency varied between epochs, but generally not between standard deviation range). The maximum fluxes with standard deviation the EC and NEC. In other words, E1 is characterized by ~10% of surface of the bSi epoch averages in the EC are as follows: 3.6 ± 0.4 (E1; 75 m), POC reaching the T100 boundary, increasing to ~13% in E2, and ~ 30% 4.3 ± 1.8 (E2; 410 m), and 7.2 ± 2.2 mmolbSi m− 2 d− 1 (E3; 500 m). The in E3. For the EC group in particular, the BCP efficiency change between epoch averaged flux values at the sample depths closest to the epoch E2 and E3 is driven by the Ez-ratio, possibly indicating a relatively more averaged PPZ are as follows: 3.6 ± 0.4 (E1; PPZ = 67 m), 3.4 ± 0.7 (E2; significant change in surface ecology versus ecology below the Ez. This PPZ = 90 m), and 6.1 ± 0.9 mmolbSi m− 2 d− 1 (E3; PPZ = 133 m). temporal change highlights the importance of high-frequency, time-se­ Similarly, the maximum epoch averaged flux values with standard de­ ries style measurements during periods of dynamic system change such viations for NEC are as follows: 3.2 ± NA (E1; 75 m), 5.7 ± 0.5 (E2; 225 as a bloom scenario. Had the sampling period been shorter, the temporal m), and 7.9 ± 1.2 mmolbSi m− 2 d− 1 (E3; 225 m). At the epoch average evolution of BCP efficiency would not have been captured. PPZ in the NEC, flux values and standard deviations are: 2.6 ± NA (E1; PPZ = 53 m), 3.9 ± 0.5 (E2; PPZ = 94 m), and 7.1 ± 0.7 mmolbSi m− 2 5. Discussion d− 1 (E3; PPZ = 122 m). Epoch averaged data can be found in Table 4 (EC) and Supplemental Table 2 (NEC). Data for all casts can be found in 5.1. Study and diatom bloom overview Supplemental Table 3. The North Atlantic EXPORTS 2021 program was carried out with the goal of better elucidating links between the chemical and biological 10 S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 Fig. 5. POC/Th (top row) and bSi/Th ratios (bottom row) per epoch, as collected by McLane pumps and representative of the >5 μm particle size class, are shown here in μmol dpm− 1 . Large symbols are representative of depth averages over all pump casts, while small symbols are individual data points. Dashed lines around the shallow depths linear fits are the average percent error range of the measurements, while dashed lines on deeper depths vertical fits are standard deviation of the individual measurements. Included for each epoch are the average PPZ (⋅⋅⋅) and MLD (− ⋅ -) depths for all casts in that epoch. Diamond symbols represent sediment trap data, included for general comparison. Fig. 6. Average flux profiles for biogenic silica (bSi, circles) and particulate organic carbon (POC, diamonds) from eddy center (EC) casts per epoch are presented, with shading indicating the standard deviation of the average. Horizontal lines show the epoch-average PPZ (dotted) and MLD (dashed) for corresponding casts. Evident here is the temporal evolution of fluxes of both species: bSi flux becomes more apparent in the mesopelagic (below the PPZ), and POC flux increases generally but particularly in the euphotic zone (above the PPZ). 11 S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 Antia, 1997; Leblanc et al., 2005). Prior studies in the North Atlantic that are critical to understanding export have shown that the amount of POC reaching depth is controlled heavily by diatom presence (Briggs et al., 2011; Rynearson et al., 2013; Le Moigne et al., 2014). Analysis of the system silica cycle and bSi flux is a natural complement to the carbon export dataset presented in this study, as it supports the hypothesis that the observed temporal carbon flux increase is indeed consistent with the demise of a diatom bloom. The sampling period began in a state of low silicic acid (dSi) concen­ tration (< 1 μmol L− 1) and low Si/NO3, indicating that substantial growth of diatoms had already occurred. Low dSi concentrations per­ sisted through the cruise, but increased in response to storm-driven entrainment that replenished surface concentration to 1–2 μmol L− 1. However, despite increased available nutrient concentrations, surface bSi stocks continued to decline while mesopelagic bSi stocks increased – indicating diatom export through E2 and into E3 (M. Brzezinski, per­ sonal communication). These trends, along with overall decreasing NPP throughout the cruise, suggest that diatoms contributed to the observed trend of increasing POC flux over the sampling period, particularly into the mesopelagic. 5.2. Temporal evolution of carbon and biogenic silica fluxes Fig. 7. Metrics classifying the biological carbon pump (BCP) are shown per There is significant temporal evolution of the diatom bloom and its epoch (x-axes). All flux data is particulate organic carbon (POC) flux, and for the eddy center (EC) group. Calculations on (A) and (B) are done using both chemical and biological characteristics over the sampling period. PPZ and a standard 100 m as a proxy for Ez (see bar colour and labels). (A) Although there are slight differences in bSi and POC flux data between shows Ez-ratio (Flux the EC and NEC groups (Table 4 and Supplemental Table 2), specifically NPP ), classifying the strength of flux relative to production in Ez as it relates to patterns with depth, we focus the following discussion on the Ez; (B) shows transfer efficiency (T100; Flux FluzEz ) classifying flux attenuation Ez+100 the EC flux data as eddy center casts are theoretically less influenced by below the Ez. physical transport than casts at the edge (Resplandy et al., 2012). Average EC flux profiles of POC and bSi are shown in Fig. 6 where two dynamics within an eddy during the demise of the spring bloom and the large patterns are particularly evident: (1) in E1, bSi fluxes above the efficiency of carbon export and sequestration. Diatom blooms in the PPZ are much closer in magnitude to POC fluxes than in the other North Atlantic have been hypothesized to be initiated by deep winter epochs, and (2) while both cumulative POC and bSi fluxes ~100 m mixing that dilutes predators and therefore decreases loss to grazing below the PPZ increase with time, the rate of increase is faster for bSi. In (Behrenfeld, 2010). The bloom accelerates as the mixed layer shoals in other words, bSi fluxes become comparatively closer to POC fluxes in E2 the spring and increased irradiance supports growth rates that continue and E3 compared to in E1. to exceed predator grazing. These conditions combine to produce a These features are further delineated by cast to cast temporal trends system that quickly exhausts silica concentrations, leading to rapid in bSi and POC fluxes at distinct depths, which reveal stark changes in changes in diatom growth rates, community structure and export bSi and POC export over time (Fig. 9). For POC (which encompasses (Gardner et al., 1993; Sieracki et al., 1993; Behrenfeld, 2010; Alkire diatom carbon + other carbon), there is a general increase in export over et al., 2012). Export driven by diatoms has been described as “pulse- the sampling period at all depths. Biogenic silica flux similarly increases like” sedimentation events, and these contribute a large fraction of throughout the sampling period. However, these increases occur largely overall carbon export to the deep ocean (Lampitt, 1985; Smetacek, below the 50 m depth bin, lying below the mixed layer and close to the 1985; Berger and Wefer, 1990; Honjo and Manganini, 1993; Lampitt and depth of the PPZ. The magnitude of the bSi fluxes relative to the POC Table 5 Ez-ratio (FluxEz/NPP), and T100 (FluxEz+100m/FluxEz)for each epoch and group. Ez-ratios and T100 were calculated both using primary production zone (PPZ) and 100 m as the threshold for the Ez base. Epoch average standard deviations are included. Note that the Ez-ratio for the NEC is calculated using NPP from the EC. Therefore, this calculation is for general comparison purposes only. Ez-Ratio PPZ 100 M EC (+/− ) NEC (+/− ) EC (+/− ) NEC (+/− ) E1 0.08 0.01 0.06 NA E1 0.12 0.01 0.11 NA E2 0.14 0.03 0.15 0.03 E2 0.13 0.03 0.16 0.02 E3 0.26 0.04 0.29 0.05 E3 0.24 0.03 0.26 0.03 Universal Avg Universal Avg 0.16 0.09 0.16 0.06 T100 PPZ 100 M EC (+/− ) NEC (+/− ) EC (+/− ) NEC (+/− ) E1 1.55 0.35 2.18 NA E1 1.01 0.12 1.08 NA E2 1.07 0.15 1.31 0.17 E2 1.09 0.11 1.25 0.11 E3 1.07 0.11 1.16 0.11 E3 1.14 0.07 1.30 0.07 Universal Avg Universal Avg 1.39 0.43 1.15 0.11 12 S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 Fig. 8. Plot adapted from Buesseler and Boyd, 2009 and Buesseler et al. 2020, depicting transfer efficiency (T100 = Flux FluzEz ) on the x-axis, and the export ratio (Ez- Ez+100 ratio = Flux NPP ) on the y-axis. Contour lines are overall BCP efficiency percentages, and symbol area is proportional to NPP. The filled bright green symbol corresponds Ez to EXPORTS 2018 Ocean Station Papa (Buesseler et al. 2020). The filled blue symbols correspond to this study, as detailed in the legend, with 1–3 denoting epochs. Note that NEC data from E1 is omitted, as it was derived from one profile only. Open symbols are from previous studies (details corresponding to numbers in Supplemental Information), and are colour coded corresponding to region: North Atlantic Ocean (blue), Pacific Ocean (green), Southern Ocean (pink). (For inter­ pretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) FluxbSi fluxes are shown as the average total bSi flux divided by POC flux (Flux ) ratio with depth as particles move through the water column. The effi­ POC per epoch, which is equivalent to the ratio of bSi to POC on >5 μm ciency of transport of bSi versus POC is driven largely by preferential particles (Fig. 9, grey shaded area corresponding to third y-axis). The remineralization of POC over bSi dissolution, along with other processes balance between silica production, bSi export and phytoplankton com­ including herbivore grazing on diatoms within the mesopelagic (Nelson munity structure is included in separate studies, and therefore the spe­ et al., 1995; Ragueneau et al., 2002; Thunell et al., 2008). The trends cific contribution of diatom carbon to the total carbon pool is not described above of increasing bSi/POC ratio in the mesopelagic over discussed further. time could be simply from POC remineralization becoming compara­ As both the numerator and denominator in the bSi/POC ratio tively more preferential than bSi dissolution over time. However, changes, it is helpful to focus on temporal evolution of the ratio. The assuming that remineralization/dissolution of POC and bSi change general stability of bSi flux compared to the increase in POC in the top roughly in parallel over time, the increase in the bSi/POC ratio in the 50 m of the water column causes the bSi/POC ratio to approximately mesopelagic throughout the sampling period temporally indicates a halve from E1 to E2 in the 5–50 m depth bins: 0.26 to 0.15 (5 m), 0.29 to relative increase in the fraction of carbon carried to the mesopelagic 0.16 (20 m), and 0.46 to 0.20 (50 m) (Fig. 8). This could indicate that zone by diatoms. This hypothesis parallels that derived from Marine after E1 diatom carbon contributes a decreasing portion of overall POC Snow Catcher data (increasing bSi/POC ratio at depth over time; E. flux in the upper water column. However, in deeper waters (e.g. 150 m) Romanelli, pers. communication). These patterns are consistent with a the decreasing bSi/POC trend reverses to a temporal increase: 0.22 (E1), progressive increase in the contribution of diatoms to flux deeper in the to 0.33 (E2), to 0.43 (E3). Presenting these numbers differently, the water column during what appears to be the demise of a diatom bloom. euphotic (5 m) and mesopelagic (150 m) ratios are approximately the Such changes occurring over a timescale of days to weeks contrast with same in E1 (0.26 vs. 0.22), double in E2 (0.15 vs. 0.33), and are four diatom bloom termination via cell aggregation that often leads to an times higher in E3 (0.11 vs. 0.43) – showing an opposite temporal abrupt increase in diatom export, removing considerable biomass from change in the flux composition between euphotic and mesopelagic surface waters and truncating quickly over the timescale of days (e.g., depths over the sampling period. These trends could indicate that Billett et al., 1983; Thiel et al., 1989). While the observation presented diatom flux contributes to an increasing portion of overall POC flux over here is specifically focused on termination and export, mixing caused by time in deeper waters. In other words, both POC and bSi fluxes increase high winds in the North Atlantic during the spring (autumn observations over time, but, particularly in the mesopelagic, bSi flux increases faster are different) are known to slow down the development and progression than POC flux. of blooms by decreasing light-dependent mean specific growth rates of Evaluation of the contribution of diatoms to POC flux considering communities (Waniek, 2003; Ueyama and Monger, 2005; Henson et al., bSi/POC ratios must account for the known trend of increasing bSi/POC 2009; Rumyantseva et al., 2019). The system studied here was 13 S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 Fig. 9. Depth-specific temporal evolution of particulate organic carbon (POC, blue diamonds, left y-axis) and biogenic silica (bSi, red circles, right y-axis) flux values, over 5 to 500 m. Here, the temporal evolution of POC and bSi fluxes throughout the sampling period can be directly compared at 8 distinct depths. The third (black) y-axis represents the epoch average bSi/POC ratio. This is indicated on the plot with grey shading, and ratio values are written at the top of the plot for clarity. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) interrupted by four storms encompassing 42% of the field program declined (Lemaitre et al., 2018). Similar POC export values were found (Johnson et al., 2024). This could have simply resulted in slower on the same cruise derived from 210Pb–210Po disequilibrium (a longer biomass accumulation and thus bloom termination. Furthermore, storm- integration time than the 234Th method), ranging from negligible to 10 driven enhanced nutrient entrainment from surrounding waters (John­ mmolC m− 2 d− 1, with the highest fluxes and highest Ez-ratios found at son et al., 2024) could have also prolonged the bloom. sampling sites where diatoms were the dominant phytoplankton group (Tang et al., 2019). Following a diatom bloom in the subpolar North Atlantic, Martin 5.3. Comparison to similar studies in bloom systems et al., 2011 measured downward POC flux following a Lagrangian float via the 234Th disequilibrium method. They found that POC flux during 5.3.1. Studies in the North Atlantic the first three weeks of the diatom bloom was as low, < 1–2.5 mmolC Biological carbon pump characteristics and POC export in the PAP m− 2 d− 1, and rose to 30–50 mmolC m− 2 d− 1 in the top 100 m, 20 days area have long been a focus of study. During late April through May, after the bloom began. Data from co-occurrent sediment traps at 600 and Buesseler et al. (1992) applied a similar NSS 234Th model to the one used 750 m suggested that 25–43% of the 100 m POC export sank below 750 here to estimate 234Th uptake and removal rates, and ultimately POC m, indicating a strong BCP system. In a study at PAP that sampled in the export dynamics as part of the JGOFS program. In that study, they found summer (July–August) after a bloom that had been observed in that throughout the one-month occupation (only 4 casts) of the bloom May–June (Le Moigne et al., 2013), POC export fluxes derived using the progression, 234Th export flux developed a “two-layered system” in 234 Th method were found to progress over time from 1.6 to 15 mmolC which the mixed layer had POC export of 5–40 mmolC m− 2 d− 1, with m− 2 d− 1 at 50 m and 1.7 to 17 mmolC m− 2 d− 1 at 150 m. This range, increased export continuing below the mixed layer, resulting in a T100 > statistically the same between the 50 and 150 m depths, shows a system 1. Notably, the highest POC export values were found during the phase similar to this study, in that there is no significant POC export attenu­ of the bloom where diatoms dominated the community structure, just ation below the Ez. This study also measured bSi fluxes, ranging from before their demise (Buesseler et al., 1992). This is consistent with the 0.008 to 0.08 mmolbSi m− 2 d− 1 at 150 m, showing low post-bloom data presented here, in which POC export peaks above the base of the export (Le Moigne et al., 2013). Information on studies cited here can euphotic zone, but then has a secondary increase in deeper waters. be found in Table 6. Overall, previous studies of diatom blooms in the Similarly, the data consistently shows a T100 at or > 1, and a diatom- North Atlantic show similar trends with respect to a temporal increase in dominated community structure. Buesseler et al. (1992) also found an POC export based on the presence and phase of the diatom bloom Ez-ratio of 0.05–0.42, consistent with results found in this study. A sampled. number of other studies in the North Atlantic region have found persistent export at depths deeper than the mixed layer, including the 5.3.2. Beyond the North Atlantic Iceland Basin (Martin et al., 2011), and PAP (Le Moigne et al., 2013; The Southern Ocean is an area which presents an interesting coun­ Ceballos-Romero et al., 2016) where disequilibria was present as deep as terpart to assessing BCP characteristics in a diatom bloom system. 400 m. The GEOTRACES cruise along the North Atlantic GA01 section During the onset of a diatom-dominated bloom in the Kerguelen area of (May–June) showed that surface POC export flux ranged from 1.4 the Southern Ocean as part of the KEOPS2 study, POC export at 100 m mmolC m− 2 d− 1, at stations where the bloom was close to its peak, to 12 varied from 3.5 to 11.8 mmolC m− 2 d− 1 over a 19-day sampling period mmolC m− 2 d− 1, at stations where the bloom (with a mixed phyto­ (Planchon et al., 2015). This early-bloom flux was generally plankton community with diatoms and coccolithophores) had already 14 S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 Table 6 2 Comparison studies referred to in Discussion Section 5.3. Export ranges are POC export in mmolPOC m− d− 1. Citation Location Time Bloom Stage POC Export Depth This Study N. Atlantic (PAP) May Peak to demise 11 to 14 110 m Buesseler et al., 1992 N. Atlantic (PAP) April–May Onset to peak 5 to 40 Mixed layer depth Lemaitre et al., 2018 N. Atlantic (GEOVIDE track) May–June Onset to peak 1.4 to 12 Surface equilibrium depth Martin et al., 2011 N. Atlantic (Iceland Basin) May Onset 1 to 2.5 100 m Martin et al., 2011 N. Atlantic (Iceland Basin) May Onset to peak 30 to 50 100 m Le Moigne et al., 2013 N. Atlantic (PAP) July–August Demise 1.6 to 15 50 m Le Moigne et al., 2013 N. Atlantic (PAP) July–August Demise 1.7 to 17 150 m Planchon et al., 2015 Southern Ocean (Kerguelen) October–November Onset 3.5 to 11.8 100 m Savoye et al., 2008 Southern Ocean (Kerguelen) January–February Peak to demise 9 to 38 100 m Roca-Martí et al., 2017 Southern Ocean (Atlantic) January–March Peak to demise 11 to 67 100 m Buesseler et al., 2005 Southern Ocean Artificial (SOFEX) Onset 1 to 11 100 m significantly lower than the KEOPS1 study that occurred at the peak and Interrogation of differences between the temporal trends of bSi and decline of a bloom in the same region. During the KEOPS1 study, POC POC fluxes is critical to understanding bloom dynamics and progression, fluxes at 100 m ranged from 9 to 38 mmolC m− 2 d− 1, with export largely thereby better understanding ocean carbon sequestration as a whole. mediated by fast-sinking aggregates of silicified diatoms (Savoye et al., Comparison of bSi and POC flux profiles, temporal trends, and bSi/POC 2008). Following a vast diatom bloom during its decline in the Atlantic ratios evidenced that bSi-driven flux out of the Ez into the mesopelagic sector of the Southern Ocean, using 234Th and sediment traps, Roca- increases over the sampling period, with a stronger increase for bSi than Martí et al., 2017 showed high POC fluxes at 100 m (11 to 67 mmolC for POC. This indicates the progression of the demise of a diatom bloom. m− 2 d− 1), but relatively low Ez-ratios compared to other diatoms blooms Further, this study suggests that the diatom export in the period of (< 0.2, Ez-ratios defined at 100 m). Taken together, these studies span sampling occurred more gradually than many diatom bloom demise different stages of a diatom bloom in the Southern Ocean, show POC studies propose. Aggregation dynamics together with the ballasting ef­ fluxes that are generally low in the beginning of the bloom, and increase fect of bSi fosters sinking of individual cells, and aggregates constitute a significantly at the peak and through the demise of the bloom. This is very efficient and robust biological carbon pump capable of sequestering consistent with the study presented here, with a 30-day period of the carbon in the deep ocean. The more prolonged export of diatom standing demise of the bloom showing increasing POC flux driven by aggregates stock observed in this sampling period may alter this dynamic, allowing of sinking diatoms. Information on studies cited here can be found in more time for the remineralization of diatom POC relative to bSi Table 6. possibly, and modulating the efficiency of diatom carbon export. The Southern Ocean Iron Fertilization Experiment (SOFeX) followed Tracking fluxes of multiple elements with the 234Th method gives an artificial, iron fertilization-fueled bloom event, and offers an inter­ extremely high-resolution direct comparison of species. This is critical esting comparison to this work (Buesseler et al., 2005). Both studies for increased confidence in analysis of complex, dynamic environments. have T100 values consistently ≥1, resulting from continued increase in During the sampling period, the overall efficiency of the BCP (as POC export beneath the base of the Ez (In Fig. 8, see numbers 11–13). quantified by the amount of NPP reaching 100 m below the euphotic The introduction of iron during the SOFeX study caused enhanced POC zone) increased from 10% in E1, to 13% in E2, and to 30% in E3. This and bSi export out of the Ez as compared to a control outside of the study observed tripling of BCP efficiency over a four-week sampling period is a site, where POC export at 100 m in the study site from the beginning to case for high temporal resolution measurements tracking dynamic en­ the end of the sampling period ranged from 1 to 11 mmolC m− 2 d− 1. vironments. Had the sampling period been shorter or shifted to another Therein, iron addition in SOFeX created diatom bloom dynamics and time period, an extremely different picture of the North Atlantic would resulting BCP characteristics that were similar to the North Atlantic have emerged. Thus, the analysis of the BCP as part of the global carbon bloom presented here. The induced diatom bloom at the point of sam­ cycle must be done through large scale sampling and observation pling caused an increase in BCP efficiency of 1–10%, while the natural schemes as conducted herein. Snapshots of systems fundamentally bloom presented here at the point of sampling shows an efficiency in­ misrepresent the complexity of dynamic processes, and therefore skew crease from 10 to 30%. However, the timeline of sampling between science’s view on the ocean’s role in climate. these two studies was different, which is reflected in study results: SOFeX sampled the initiation of the bloom while this study largely Funding sampled the demise of the bloom. It would be a useful area of further study to follow the entire timeline of an artificial bloom with high- Authors would like to acknowledge support from the National resolution 234Th sampling. Regardless, these two, time series-style Aeronautics and Space Administration (NASA) as part of the EXPORTS studies offer a comparison that can be informative in future iron addi­ program awards 80NSSC17K0662, 80NSSC21K0015, 80NSSC17K0555; tion and BCP enhancement experiments. MRM was supported by funding from the Ocean Frontier Institute In­ ternational Postdoctoral Fellowship Program and the Beatriu de Pinós 6. Conclusions Postdoctoral Program (2021-BP-00109); KOB and WB were supported by funding from the Woods Hole Oceanographic Institution’s Ocean This study represents an important contribution not only to the Twilight Zone study. literature assessing diatom blooms in the North Atlantic and in eddies, but also in the field’s understanding of NSS systems and the 234Th CRediT authorship contribution statement method application within them. A high spatial and temporal resolution 234 Th sampling scheme totaling 800 samples collected, in combination Samantha J. Clevenger: Conceptualization, Data curation, Formal with POC and bSi/Th ratio determination on particles, allowed for the analysis, Investigation, Methodology, Visualization, Writing – original NSS derivation of POC and bSi fluxes throughout the upper 500 m of the draft, Writing – review & editing. Claudia R. Benitez-Nelson: water column. The unprecedented high resolution of this NSS study is Conceptualization, Data curation, Funding acquisition, Project admin­ unique in its scale, and both validates and expands on previous uses of istration, Supervision, Writing – review & editing. Montserrat Roca- an NSS model in 234Th analysis. Martí: Data curation, Writing – original draft, Writing – review & 15 S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 editing, Formal analysis. Wokil Bam: Data curation. Margaret Estapa: Boyd, P.W., Claustre, H., Levy, M., Siegel, D.A., Weber, T., 2019. Multi-faceted particle pumps drive carbon sequestration in the ocean. Nature 568, 327–335. https://doi. Data curation, Funding acquisition, Writing – review & editing. Jenni­ org/10.1038/s41586-019-1098-2. fer A. Kenyon: Data curation. Steven Pike: Data curation. Laure Briggs, N., Perry, M.J., Cetinić, I., Lee, C., D’Asaro, E., Gray, A.M., Rehm, E., 2011. High- Resplandy: Conceptualization, Funding acquisition, Supervision, resolution observations of aggregate flux during a sub-polar North Atlantic spring Writing – review & editing. Abigale Wyatt: Formal analysis, Writing – bloom. Deep-Sea Res. I Oceanogr. Res. Pap. 58, 1031–1039. https://doi.org/ 10.1016/j.dsr.2011.07.007. original draft. Ken O. Buesseler: Conceptualization, Funding acquisi­ Brown, O.B., Evans, R.H., Brown, J.W., Gordon, H.R., Smith, R.C., Baker, K.S., 1985. tion, Investigation, Supervision, Writing – review & editing, Project Phytoplankton blooming off the U.S. East Coast: A satellite description. Science 229, administration. 163–167. Brzezinski, M.A., Nelson, D.M., 1989. Seasonal changes in the silicon cycle within a Gulf Stream warm-core ring. Deep Sea Res. Part A 36, 1009–1030. Declaration of Competing Interest Buesseler, K.O., Boyd, P.W., 2009. Shedding light on the processes that control particle export and flux attenuation in the twilight zone of the open ocean. Limnol. Oceanogr. 54, 1210–1232. The authors have no competing interests. Buesseler, K.O., Bacon, M.P., Cochran, J.K., Livingston, H.D., 1992. Carbon and nitrogen export during the JGOFS North Atlantic Bloom Experiment estimated from 234 Data availability Th:238U disequilibria. Deep-Sea Res. I Oceanogr. Res. Pap. 39, 1115–1137. Buesseler, K.O., Andrews, J.E., Pike, S.M., Charette, M.A., Goldson, L.E., Brzezinski, M. A., Lance, V.P., 2005. Particle export during the Southern Ocean Iron Experiment Data is available in NASA’s SeaBASS repository (SOFeX). Limnol. Oceanogr. 50, 311–327. https://doi.org/10.4319/ lo.2005.50.1.0311. Buesseler, K.O., et al., 2007. Revisiting carbon flux through the ocean’s twilight zone. Acknowledgements Science 316, 567–570. Buesseler, K.O., Boyd, P.W., Black, E.E., Siegel, D.A., 2020a. Metrics that matter for Thank you to Pat Kelly, Colleen Durkin, and Sean O’Neill for assis­ assessing the ocean biological carbon pump. PNAS 117, 9679–9687. https://doi.org/ tance with sediment trap data collection; to Jessica Drysdale for running 10.1073/pnas.1918114117. Buesseler, K.O., Benitez-Nelson, C.R., Roca-Martí, M., Wyatt, A.M., Resplandy, L., PIC samples; to Gretchen Swarr for running the iCAP; to Ivona Cetinic, Clevenger, S.J., Drysdale, J.A., Estapa, M.L., Pike, S., Umhau, B.P., 2020b. High- Dave Siegel, and the NASA EXPORTS Project Office for coordinating the resolution spatial and temporal measurements of particulate organic carbon flux EXPORTS efforts; thank you also to the captains and crews of the RRS using thorium-234 in the Northeast Pacific Ocean during the EXport processes in the ocean from RemoTe sensing field campaign. Elem. Sci. Anth. 8 https://doi.org/ James Cook and RRS Discovery. We also thank the three reviewers, 10.1525/elementa.2020.030. including Frédéric Planchon and David DeMaster, for their generous Bury, S.J., Boyd, P.W., Preston, T., Savidge, G., Owens, N.J.P., 2001. Size-fractionated donation of time that has resulted in an improved manuscript. primary production and nitrogen uptake during a North Atlantic phytoplankton bloom: implications for carbon export estimates. Deep-Sea Res. I Oceanogr. Res. Pap. 48, 689–720. https://doi.org/10.1016/S0967-0637(00)00066-2. Appendix A. Supplementary data Cavan, E.L., Le Moigne, F.A.C., Poulton, A.J., Tarling, G.A., Ward, P., Daniels, C.J., Fragoso, G.M., Sanders, R.J., 2015. Attenuation of particulate organic carbon flux in the Soctia Sea, Southern Ocean, is controlled by zooplankton fecal pellets. Geophys. Supplementary data to this article can be found online at https://doi. Res. Lett. 42, 821–830. https://doi.org/10.1002/2014GL062744. org/10.1016/j.marchem.2023.104346. Ceballos-Romero, E., Le Moigne, F.A.C., Henson, S., Marsay, C.M., Sanders, R.J., García- Tenorio, R., Villa-Alfageme, M., 2016. Influence of bloom dynamics on particle export efficiency in the North Atlantic: a comparative study of radioanalytical References techniques and sediment traps. Mar. Chem. 186, 198–210. https://doi.org/10.1016/ j.marchem.2016.10.001. Alkire, M.B., D’Asaro, E., Lee, C., Perry, M.J., Gray, A., Cetinić, I., Briggs, N., Rehm, E., Ceballos-Romero, E., De Soto, F., Le Moigne, F.A.C., García-Tenorio, R., Villa- Kallin, E., Kaiser, J., González-Posada, A., 2012. Estimates of net community Alfageme, M., 2018. 234Th-derived particle fluxes and seasonal variability: when is production and export using high-resolution Lagrangian measurements of O2, NO−3 , the SS assumption reliable? Insights from a novel approach for carbon flux and POC through the evolution of a spring diatom bloom in the North Atlantic. Deep- simulation. Geophys. Res. Lett. 45, 13414–13426. https://doi.org/10.1029/ Sea Res. I Oceanogr. Res. Pap. 64, 157–174. 2018GL079968. Bacon, M.P., Anderson, R.F., 1982. Distribution of thorium isotopes between dissolved Clevenger, S.J., Benitez-Nelson, C.R., Drysdale, J., Pike, S., Puigcorbé, V., Buesseler, K. and particulate forms in the deep sea. J. Geophys. Res. 87, 2045–2056. O., 2021. Review of the analysis of 234Th in small volume (2-4L) seawater samples: Bacon, M.P., Huh, C.-A., Fleer, A.P., Deuser, W.G., 1985. Seasonality in the flux of improvements and recommendations. J. Radioanal. Nucl. Chem. 329, 1–13. natural radionuclides and plutonium in the deep Sargasso Sea. Deep Sea Res. Part A Coale, K.H., Bruland, K.W., 1987. Oceanic stratified euphotic zone as elucidated by 32, 273–286. https://doi.org/10.1016/0198-0149(85)90079-2. 234 Th:238U disequilibria. Limnol. Oceanogr. 32, 189–200. Behrenfeld, M.J., 2010. Abandoning Sverdrup’s critical depth hypothesis on Cochran, J.K., Masqué, P., 2003. Short-lived U/Th series radionuclides in the ocean: phytoplankton blooms. Ecology 91, 977–989. https://doi.org/10.1890/09-1207.1. tracers for scavenging rates, export fluxes, and particle dynamics. Rev. Mineral. Behrenfeld, M.J., Doney, S.C., Lima, I., Boss, E.S., Siegel, D.A., 2013. Annual cycles of Geochem. 52, 461–492. https://doi.org/10.2113/0520461. ecological disturbance and recovery underlying the subarctic Atlantic spring Cohn, M.R., Stephens, B., Meyer, M.G., Sharpe, G., Niebergall, A.K., Graff, J.R., plankton bloom. Global Biogeochem. Cycles. 27, 526–540. Cassar, N., Marchetti, A., Carlson, C.A., Gifford, S., 2023. Microbial respiration in Berger, W.H., Wefer, G., 1990. Export production: seasonality and intermittency, and contrasting ocean provinces via high-frequency optical assays. preprint. https://doi. palaeoceanographic implications. Palaeogeogr. Palaeoclimatol. Palaeoecol. 89, org/10.1101/2023.07.20.549894. 245–254. https://doi.org/10.1016/0031-0182(90)90065-F. Cutter, G., Casciotti, K., Croot, P., Geibert, W., Heimbürger, L.-E., Lohan, M., Bhat, S.G., Krishnaswamy, S., Lal, D., Rama Moore, W.S., 1969. 234Th/238U ratios in the Planquette, H., van de Flierdt, T., 2017. Sampling and Sample-handling Protocols for ocean. Earth Planet. Sci. Lett. 5, 483–491. https://doi.org/10.1016/S0012-821X(68) GEOTRACES Cruises. Version 3.0. https://doi.org/10.25607/OBP-2. 80083-4. Dall’Olmo, G., Dingle, J., Polimene, L., Brewin, R.J.W., Claustre, H., 2016. Substantial Billett, D.S.M., Lampitt, R.S., Rice, A.L., Mantoura, R.F.C., 1983. Seasonal sedimentation energy input to the mesopelagic ecosystem from the seasonal mixed-layer pump. of phytoplankton to the deep-sea benthos. Nature 302, 520–522. Nat. Geosci. 9, 820–823. Binetti, U., Kaiser, J., Damarell, G.M., Rumyantseva, A., Martin, A.P., Henson, S., De La Rocha, C.L., Passow, U., 2007. Factors influencing the sinking of POC and the Heywood, K.J., 2020. Net community oxygen production derived from Seaglider efficiency of the biological carbon pump. Deep-Sea Res. II Top. Stud. Oceanogr. 54, deployments at the porcupine abyssal plain site (PAP; Northeast Atlantic) in 2012- 639–658. 13. Prog. Oceanogr. 183, 102293. Erikson, Z.K., Fields, E., Omand, M.M., Johnson, L., Thompson, A.F., D’Asaro, E., Bishop, J.K.B., Lam, P.J., Wood, T.J., 2012. Getting good particles: accurate sampling of Carvalho, F., Dove, L.A., Lee, C.M., Nicholson, D.P., Shilling, G., Cetinic, I., particles by large volume in-situ filtration. Limnol. Oceanogr. Methods 10, 681–710. Siegel, D., 2022. EXPORTS North Atlantic Eddy tracking. NASA Technical https://doi.org/10.4319/lom.2012.10.681. Publication. (NASA/TM-20220009705). Black, E.E., Lam, P.J., Lee, J.-M., Buesseler, K.O., 2018. Insights from the 238U-234Th Esaias, W.E., Feldman, G.C., McClain, C.R., Elrod, J.A., 1986. Monthly satellite-derived method into the coupling of biological export and the cycling of cadmium, cobalt, phytoplankton pigment distribution for the North Atlantic Ocean basin. EOS Trans. and manganese in the Southeast Pacific Ocean. Global Biogeochem. Cycles 33, Am. Geo. Union 67, 835–837. https://doi.org/10.1029/EO067i044p00835. 15–36. Estapa, M.L., Durkin, C.A., Slade, W.H., Huffard, C.L., O’Neill, S.P., Omand, M.M., 2023. Black, E.E., Kienast, S.S., Lemaitre, N., Lam, P.J., Anderson, R.F., Planquette, H., A new, global optical sediment trap calibration. Limnol. Oceanogr.: Methods. Planchon, F., Buesseler, K.O., 2020. Ironing out Fe residence time in the dynamic Feely, R.A., Sabine, C.L., Takahashi, T., Wanninkhof, R., 2001. Uptake and storage of upper ocean. Global Biogeochem. Cycles 34. https://doi.org/10.1029/ carbon dioxide in the ocean: the global CO2 survey. Oceanography 14, 18–32. 2020GB006592. Friedlingstein, P., O’Sullivan, M., Jones, M.W., et al., 2022. Global carbon budget 2022. Earth Syst. Sci. Data. 14, 4811–4900. https://doi.org/10.5194/essd-14-4811-2022. 16 S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 Gardner, W.D., Walsh, I.D., Richardson, M.J., 1993. Biophysical forcing of particle Le Moigne, F.A.C., Poulton, A.J., Henson, S.A., Daniels, C.J., Fragoso, G.M., Mitchell, E., production and distribution during a spring bloom in the North Atlantic. Deep-Sea Richier, S., Russell, B.C., Smith, H.E.K., Tarling, G.A., Young, J.R., Zubkov, M., 2015. Res. II Top. Stud. Oceanogr. 40, 171–195. https://doi.org/10.1016/0967-0645(93) Carbon export efficiency and phytoplankton community composition in the Atlantic 90012-C. sector of the Arctic Ocean. J. Geophys. Res.: Oceans 120, 3896–3912. https://doi. Giering, S.L.C., et al., 2014. Reconciliation of the carbon budget in the ocean’s twilight org/10.1002/2015JC010700. zone. Nature 507, 480–507. https://doi.org/10.1038/nature13123. Le Moigne, F.A.C., Henson, S.A., Cavan, E., Georges, C., Pabortsava, K., Achterberg, E.P., Gustafsson, Ö., Gschwend, P.M., Buesseler, K.O., 1997. Using 234Th disequilibria to Ceballos-Romero, E., 2016. What causes the inverse relationship between primary estimate the vertical removal rates of polycyclic aromatic hydrocarbons from the production and export efficiency in the Southern Ocean? Geophys. Res. Lett. 43, surface ocean. Mar. Chem. 57, 11–23. 4457–4466. https://doi.org/10.1002/2016GL068480. Hartman, S.E., Bett, B.J., Durden, J.M., Henson, S.A., Iversen, M., Jeffreys, R.M., Leblanc, K., Leynaert, A., Fernandez, C., Rimmelin, P., Moutin, T., Raimbault, P., Ras, J., Horton, T., Lampitt, R., Gates, A.R., 2021. Enduring science: three decades of Quéguiner, B., 2005. A seasonal study of diatom dynamics in the North Atlantic observing the Northeast Atlantic from the porcupine abyssal plain sustained during the POMME experiment (2001): evidence for Si limitation of the spring observatory (PAP-SO). Prog. Oceanogr. 191, 102508 https://doi.org/10.1016/j. bloom. J. Geophys. Res.: Oceans 110. https://doi.org/10.1029/2004JC002621. pocean.2020.102508. Leblanc, K., Hare, C.E., Feng, Y., Berg, G.M., DiTullio, G.R., Neeley, A., Benner, I., Henson, S.A., Dunne, J.P., Sarmiento, J.L., 2009. Decadal variability in North Atlantic Sprengel, C., Beck, A., Sanudo-Wilhelmy, S.A., Passow, U., Klinck, K., Rowe, J.M., phytoplankton blooms. J. Geophys. Res. Oceans 114, C04013. https://doi.org/ Wilhelm, S.W., Brown, C.W., Hutchins, D.A., 2009. Distribution of calcifying and 10.1029/2008JC005139. silicifying phytoplankton in relation to environmental and biogeochemical Henson, S.A., Lampitt, R., Johns, D., 2012. Variability in phytoplankton community parameters during the late stages of the 2005 North East Atlantic spring bloom. structure in response to the North Atlantic Oscillation and implications for organic Biogeosci. 6, 2155–2179. carbon flux. Limnol. Oceanogr. 57, 1591–1601. https://doi.org/10.4319/ Lemaitre, N., Planchon, F., Planquette, H., Dehairs, F., Fonseca-Batista, D., Roukaerts, A., lo.2012.57.6.1591. Deman, F., Tang, Y., Mariez, C., Sarthou, G., 2018. High variability of particulate Henson, S.A., Yool, A., Sanders, R., 2015. Variability in efficiency of particulate organic organic carbon export along the North Atlantic GEOTRACES section GA01 as carbon export: a model study. Global Biogeochem. Cycles 29, 33–45. https://doi. deduced from 234Th fluxes. Biogeosciences 15, 6417–6437. https://doi.org/ org/10.1002/2014GB004965. 10.5194/bg-15-6417-2018. Henson, S., Le Moigne, F., Giering, S., 2019. Drivers of carbon export efficiency in the Lévy, M., et al., 2013. Physical pathways for carbon transfers between the surface mixed global ocean. Global Biogeochem. Cycles 33, 891–903. https://doi.org/10.1029/ layer and the ocean interior. Global Biogeochem. Cycles 27, 1001–1012. 2018GB006158. Li, Y.-H., Feely, H.W., Santschi, P.H., 1979. 228Th:228Ra radioactive disequilibrium in the Honjo, S., Manganini, S.J., 1993. Annual biogenic particle fluxes to the interior of the New York bight and its implications for coastal pollution. Earth Planet. Sci. Lett. 42, North Atlantic Ocean; studied at 34◦ N 21◦ W and 48◦ N 21◦ W. Deep-Sea Res. II Top. 13–26. Stud. Oceanogr. 40, 587–607. https://doi.org/10.1016/0967-0645(93)90034-K. Lochte, K., Ducklow, H.W., Fasham, M.J.R., Stienen, C., 1993. Plankton succession and Honjo, S., Dymond, J., Collier, R., Manganini, S.J., 1995. Export production of particles carbon cycling at 47◦ N 20◦ W during the JGOFS North Atlantic bloom experiment. to the interior of the equatorial Pacific Ocean during the 1992 EqPac experiment. Deep-Sea Res. II Top. Stud. Oceanogr. 40, 91–114. https://doi.org/10.1016/0967- Deep-Sea Res. II Top. Stud. Oceanogr. 42, 831–870. https://doi.org/10.1016/0967- 0645(93)90008-B. 0645(95)00034-N. Longhurst, A.R., 1991. Modelling the unpredictable. American Association for the Huh, C.-A., Moore, W.S., Kadko, D.C., 1989. Oceanic 232Th: a reconnaissance and Advancement of Science 242. Washington, DC. implications of global distribution from manganese nodules. Geochim. Cosmochim. Macovei, V.A., Hartman, S.E., Schuster, U., Torres-Valdés, S., Moore, C.M., Sanders, R.J., Acta 53, 1357–1366. 2020. Impact of physical and biological processes on temporal variations of the Huisman, J., van Oostveen, P., Weissing, F.J., 1999. Critical depth and critical ocean carbon sink in the mid-latitude North Atlantic (2002-2016). Prog. Oceanogr. turbulence: two different mechanisms for the development of phytoplankton blooms. 180. Limnol. Oceanogr. 1781–1787. Maier-Reimer, E., Mikolajewicz, U., Winguth, A., 1996. Future Ocean uptake of CO2: Johnson, L., Siegel, D.A., Thompson, A.F., Fields, E., Erickson, Z.K., Cetinic, I., Lee, C.M., interaction between ocean circulation and biology. Climate Dynam. 12, 711–722. 2024. Assessment of oceanographic conditions during the North Atlantic EXport Maiti, K., Benitez-Nelson, C.R., Buesseler, K.O., 2010. Insights into particle formation Processes in the Ocean from RemoTe Sensing (EXPORTS) field campaign. Prog. and remineralization using the short-lived radionuclide, Thoruim-234. Geophys. Oceanogr. 220, 103170. Res. Lett. 37 https://doi.org/10.1029/2010GL044063. Jónasdóttir, S.H., Visser, A.W., Richardson, K., Heath, M.R., 2015. Seasonal copepod Maiti, K., Buesseler, K.O., Pike, S.M., Benitez-Nelson, C.R., Cai, P., Chen, W., Cochran, K., lipid pump promotes carbon sequestration in the deep North Atlantic. Proc. Natl. Dai, M., Dehairs, F., Gasser, B., et al., 2012. Intercalibration studies of short-lived Acad. Sci. U. S. A. 112, 12122–12126. thorium-234 in the water column and marine particles. Limnol. Oceanogr. Methods Kaufman, A., Li, Y.-H., Turekian, K.K., 1981. The removal rates of 234Th and 228Th from 10, 631–644. https://doi.org/10.4319/lom.2012.10.631. waters of the New York Bight. Earth Planet. Sci. Lett. 54, 385–392. Marra, J.F., Lance, V.P., Vaillancourt, R.D., Hargreaves, B.R., 2014. Resolving the Kenyon, 2021. Anthropogenic and Natural Radioisotopes as Tracers for Contaminant ocean’s euphotic zone. Deep-Sea Res. I Oceanogr. Res. Pap. 83, 45–50. https://doi. Sources and Particulate Fluxes [Unpublished Doctoral Dissertation]. Massachusetts org/10.1016/j.dsr.2013.09.005. Institute of Technology and Woods Hole Oceanographic Institution. Martin, P., Lampitt, R.S., Perry, M.J., Sanders, R., Lee, C., D’Asaro, E., 2011. Export and Koski, M., Valencia, B., Newstead, R., Thiele, C., 2020. The missing piece of the upper mesopelagic particle flux during a North Atlantic spring diatom bloom. Deep-Sea mesopelagic carbon budget? Biomass, vertical distribution and feeding of aggregate- Res. I Oceanogr. Res. Pap. 58, 338–349. https://doi.org/10.1016/j.dsr.2011.01.006. associated copepods at the PAP site. Prog. Oceanogr. 181, 102243. McKinley, G.A., et al., 2016. Timescales for detection of trends in the ocean carbon sink. Kwon, E.Y., Primeau, F., Sarmiento, J.L., 2009. The impact of remineralization depth on Nature 530, 469–472. the air-sea carbon balance. Nat. Geosci. 2, 630–635. Meyer, M.G., Brzezinski, M.A., Cohn, M.R., Kramer, S.J., Paul, N., Sharpe, G., Lacour, L., Briggs, N., Claustre, H., Ardyna, M., Dall’Olmo, G., 2019. The intraseasonal Niebergall, A.K., Gifford, S., Cassar, N., Marchetti, A., 2023. Primary production dynamics of the mixed layer pump in the subpolar North Atlantic Ocean: a dynamics during the decline phase of the North Atlantic annual spring bloom. biogeochemical-Argo float approach. Global Biogeochem. Cycles 33, 266–281. preprint. https://doi.org/10.1101/2023.05.18.541304. https://doi.org/10.1016/j.dsr.2013.09.005. Nelson, D.M., Tréguer, P., Brzezinski, M.A., Leynaert, A., Quéguiner, B., 1995. Lam, P.J., Ohnemus, D.C., Auro, M.E., 2015. Size-fractionated major particle Production and dissolution of biogenic silica in the ocean: revised global estimates, composition and concentrations from the US GEOTRACES North Atlantic Zonal comparison with regional data and relationship to biogenic sedimentation. Global Transect. Deep Sea Res Part II Top Stud Oceanogr 116, 303–320. https://doi.org/ Biogeochem. Cycles 9, 359–372. https://doi.org/10.1029/95GB01070. 10.1016/J.DSR2.2014.11.020. Nozaki, Y., Yang, H.-S., Yamada, M., 1987. Scavenging of thorium in the ocean. Lamborg, C.H., Buesseler, K.O., Valdes, J., Bertrand, C.H., Bidigare, R., Manganini, S., J. Geophys. Res. 92, 772–778. Pike, S., Steinberg, D., Trull, T., Wilson, S., 2008. The flux of bio- and lithogenic Omand, M.M., et al., 2015. Eddy-driven subduction exports particulate organic carbon material associated with sinking particles in the mesopelagic “twilight zone” of the from the spring bloom. Science 348, 222–225. northwest and north Central Pacific Ocean. Deep-Sea Res. II Top. Stud. Oceanogr. Owen, S.A., Buesseler, K.O., Sims, K.W.W., 2011. Re-evaluating the 238U-salinity 55, 1540–1563. https://doi.org/10.1016/J.DSR2.2008.04.011. relationship in seawater: implications for the 238U-234Th disequilibrium method. Lampitt, R.S., 1985. Evidence for the seasonal deposition of detritus to the deep-sea floor Mar. Chem. 127, 31–39. and its subsequent resuspension. Deep Sea Res. Part A 32, 885–897. https://doi.org/ Owens, S.A., Pike, S., Buesseler, K.O., 2015. Thorium-234 as a tracer of particle dynamics 10.1016/0198-0149(85)90034-2. and upper ocean export in the Atlantic Ocean. Deep-Sea Res. II Top. Stud. Oceanogr. Lampitt, R.S., Antia, A.N., 1997. Particle flux in deep seas: regional characteristics and 116, 42–59. https://doi.org/10.1016/j.dsr2.2014.11.010. temporal variability. Deep-Sea Res. I Oceanogr. Res. Pap. 44, 1377–1403. Parekh, P., Dutkiewicz, S., Follows, M.J., Ito, T., 2006. Atmospheric carbon dioxide in a Langmuir, D., 1978. Uranium solution-mineral equilibria at low temperatures with less dusty world. Geophys. Res. Lett. 33 https://doi.org/10.1029/2005GL025098. applications to sedimentary ore deposits. Geochim. Cosmochim. Acta 42, 547–569. Pike, S., Buesseler, K.O., Andrews, J., Savoye, N., 2005. Quantification of the 234Th https://doi.org/10.1016/0016-7037(78)90001-7. recovery in small volume seawater samples by inductively coupled plasma mass Le Moigne, F.A.C., Villa-Alfageme, M., Sanders, R.J., Marsay, C., Henson, S., García- spectrometry. J. Radioanal. Nucl. Chem. 263, 355–360. Tenorio, R., 2013. Export of organic carbon and biominerals derived from 234Th and Planchon, F., Ballas, D., Cavagna, A.-J., Bowie, A.R., Davies, D., Trull, T., Laurenceau- 210 Po at the porcupine abyssal plain. Deep-Sea Res. I 72, 88–101. Cornec, E.C., Van Der Merwe, P., Dehairs, F., 2015. Carbon export in the naturally Le Moigne, F.A.C., Pabortsava, K., Marcinko, C.L.J., Martin, P., Sanders, R.J., 2014. iron-fertilized Kerguelen area of the Southern Ocean based on the 234Th approach. Where is mineral ballast important for surface export of particulate organic carbon in Biogeosciences 12, 3831–3848. https://doi.org/10.5194/bg-12-3831-2015. the ocean? Geophys. Res. Lett. 41, 8460–8468. https://doi.org/10.1002/ Ragueneau, O., Dittert, N., Pondaven, P., Tréguer, P., Corrin, L., 2002. Si/C decoupling in 2014GL061678. the world ocean: is the Southern Ocean different? Deep Sea res. Part II 49, 3127–3154. https://doi.org/10.1016/S0967-0645(02)00075-9. 17 S.J. Clevenger et al. Marine Chemistry 258 (2024) 104346 Resplandy, L., Martin, A.P., Le Moigne, F., Martin, P., Aquilina, A., Mémery, L., Lévy, M., Siegel, D.A., et al., 2021. An operational overview of the EXport Processes in the Ocean Sanders, R., 2012. How does dynamical spatial variability impact 234Th-derived from RemoTe sensing (EXPORTS) Northeast Pacific field deployment. Elem. Sci. estimates of organic export? Deep Sea res. I 68, 24–45. Anth. 9 https://doi.org/10.1525/elementa.2020.00107. Resplandy, L., Lévy, M., McGillicuddy, D.J., 2019. Effects of eddy-driven subduction on Sieracki, M.E., Verity, P.G., Stoecker, D.K., 1993. Plankton community response to ocean biological carbon pump. Global Biogeochem. Cycles 33, 1071–1084. sequential silicate and nitrate depletion during the 1989 North Atlantic spring Riley, G.A., 1942. The relationship of vertical turbulence and spring diatom flowerings. bloom. Deep-Sea Res. II Top. Stud. Oceanogr. 40, 213–225. https://doi.org/ J. Mar. Res. 5, 67–87. 10.1016/0967-0645(93)90014-E. Robertson, J.E., Watson, A.J., Langdon, C., Ling, R.D., Wood, J.W., 1993. Diurnal Smetacek, V.S., 1985. Role of sinking in diatom life-history cycles: ecological, variation in surface pCO2 and O2 at 60◦ N, 20◦ W in the North Atlantic. Deep-Sea Res. evolutionary and geological significance. Mar. Biol. 84, 239–251. II Top. Stud. Oceanogr. 40, 409–422. https://doi.org/10.1016/0967-0645(93) Smythe-Wright, D., Boswell, S., Kim, Y.-N., Kemp, A., 2010. Spatio-temporal changes in 90024-H. the distribution of phytopigments and phytoplanktonic groups at the porcupine Roca-Martí, M., Puigcorbé, V., Iversen, M.H., Rutgers van der Loeff, M., Klaas, C., abyssal plain (PAP) site. Deep-Sea Res. II Top. Stud. Oceanogr. 57, 1324–1335. Cheah, W., Bracher, A., Masqué, P., 2017. High particulate organic carbon export Stukel, M.R., Song, H., Goericke, R., Miller, A.J., 2017. The role of subduction and during the decline of a vast diatom bloom in the Atlantic sector of the Southern gravitational sinking in particle export, carbon sequestration, and the Ocean. Deep-Sea Res. II Top. Stud. Oceanogr. 138, 102–115. https://doi.org/ remineralization length scale in the California current ecosystem. Limnol. Oceanogr. 10.1016/j.dsr2.2015.12.007. 63, 363–383. Roca-Martí, M., Benitez-Nelson, C.R., Umhau, B.P., Wyatt, A.M., Clevenger, S.J., Pike, S., Sundby, S., Drinkwater, K.F., Kjesbu, O.S., 2016. The North Atlantic spring-bloom system Horner, T.J., Estapa, M.L., Resplandy, L., Buesseler, K.O., 2021. Concentrations, – where the changing climate meets the winter dark. Front. Mar. Sci. 3, 28. https:// ratios, and sinking fluxes of major bioelements at Ocean Station papa. Elem. Sci. doi.org/10.3389/fmars.2016.00028. Anth. 9, 00166. https://doi.org/10.1525/elementa.2020.00166. Sverdrup, H.U., 1953. On conditions for the vernal blooming of phytoplankton. ICES J. Rumyantseva, A., Henson, S., Martin, A., Thompson, A.F., Damarell, G.M., Kaiser, J., Mar. Sci. 18, 287–295. https://doi.org/10.1093/icesjms/18.3.287. Heywood, K.J., 2019. Phytoplankton spring bloom initiation: the impact of Tang, Y., Lemaitre, N., Castrillejo, M., Roca-Martí, M., Masqué, P., Stewart, G., 2019. The atmospheric forcing and light in the temperate North Atlantic Ocean. Prog. export flux of particulate organic carbon derived from 210Po/210Pb disequilibria Oceanogr. 178, 102202. along the North Atlantic GEOTRACES GA01 transect: GEOVIDE cruise. Rynearson, T.A., Richardson, K., Lampitt, R.S., Sieracki, M.E., Poulton, A.J., Biogeosciences 16, 309–327. https://doi.org/10.5194/bg-16-309-2019. Lyngsgaard, M.M., Perry, M.J., 2013. Major contribution of diatom resting spores to Tesán Onrubia, J.A., Petrova, M.V., Puigcorbé, V., Black, E.E., Valk, O., Dufour, A., vertical flux in the sub-polar North Atlantic. Deep-Sea Res. I Oceanogr. Res. Pap. 82, Hamelin, B., Buesseler, K.O., Masqué, P., Le Moigne, F.A.C., Sonke, J.E., Rutgers van 60–71. https://doi.org/10.1016/j.dsr.2013.07.013. der Loeff, M., Heimbürger-Boavida, L.-E., 2020. Mercury export flux in thr Arctic Sanders, R., Henson, S.A., Koski, M., Rocha, C.L.D.L., Painter, S.C., Poulton, A.J., Ocean estimated from 234Th/238U disequilibria. ACS Earth Space Chem. 4, 795–801. Riley, J., Salihoglu, B., Visser, A., Yool, A., Bellerby, R., Martin, A., 2014. The https://doi.org/10.1021/acsearthspacechem.0c00055. biological carbon pump in the North Atlantic. Prog. Oceanogr. 129, 200–218. Thiel, H., et al., 1989. Phytodetritus on the deep-sea floor in a central oceanic region of https://doi.org/10.1016/j.pocean.2014.05.005. the Northeast Atlantic. Biol. Oceanogr. 6, 203–239. Santschi, P.H., Murray, J.W., Baskaran, M., Benitez-Nelson, C.R., Guo, L.D., Hung, C.-C., Thunell, R., Benitez-Nelson, C.R., Muller-Karger, F., Lorenzoni, L., Fanning, K., Lamborg, C., Moran, S.B., Passow, U., Roy-Barman, M., 2006. Thorium speciation in Scranton, M., Varela, R., Astor, Y., 2008. Si cycle in the Cariaco Basin, Venezuela: seawater. Mar. Chem. 100, 250–268. Seasonal variability in silicate availability and the Si:C:N composition of sinking Sarmiento, J.L., Gruber, N., 2006. Ocean Biogeochemical Dynamics Ch, vol. 8. Princeton particles. Global Biogeochem. Cycles 22. https://doi.org/10.1029/2007GB003096. Univ. Press, Princeton. Ueyama, R., Monger, B.C., 2005. Wind-induced modulation of seasonal phytoplankton Sarmiento, J.L., Toggweiler, J.R., 1984. A new model for the role of the oceans in blooms in the North Atlantic derived from satellite observations. Limnol. Oceanogr. determining atmospheric PCO2. Nature 308, 621–624. 50, 1820–1829. Savidge, G., Boyd, P., Pomroy, A., Harbour, D., Joint, I., 1995. Phytoplankton production Verdeny, E., Masqué, P., Garcia-Orellana, J., Hanfland, C., Cochran, J.K., Stewart, G.M., and biomass estimates in the Northeast Atlantic Ocean, may-June 1990. Deep-Sea 2009. POC export from ocean surface waters by means of 234Th/238U and 210 Res. I Oceanogr. Res. Pap. 42, 599–617. https://doi.org/10.1016/0967-0637(95) Po/210Pb disequilibria: a review of the use of two radiotracer pairs. Deep Sea Res. 00016-Y. 56, 1502–1518. Savoye, N., Benitez-Nelson, C.R., Burd, A.B., Cochran, J.K., Charette, M., Buesseler, K.O., Volk, T., Hoffert, M., 1985. In: Sundquist, E.T., Broecker, W.S. (Eds.), In the Carbon Cycle Jackson, G.A., Roy-Barman, M., Schmidt, S., Elskens, M., 2006. 234Th sorption and and Atmospheric CO2: Natural Variations Archean to Present, vol. 32. American export models in the water column: a review. Mar. Chem. 100, 234–249. Geophysical Union, Washington, DC, pp. 99–110. Savoye, N., Trull, T.W., Jacquet, S.H.M., Navez, J., Dehairs, F., 2008. 234Th-based export Waniek, J.J., 2003. The role of physical forcing in initiation of spring blooms in the fluxes during a natural iron fertilization experiment in the Southern Ocean (KEOPS). Northeast Atlantic. J. Mar. Syst. 39, 57–82. Deep-Sea Res. II Top. Stud. Oceanogr. 55, 841–855. https://doi.org/10.1016/j. Wassmann, P., 1998. Retention versus export food chains: processes controlling sinking dsr2.2007.12.036. loss from marine pelagic systems. Hydrobiologia 363, 29–57. Siegel, D.A., Buesseler, K.O., Behrenfeld, M.J., Benitez-Nelson, C.R., Boss, E., Weinstein, S.E., Moran, S.B., 2005. Vertical flux of particulate Al, Fe, Pb, and Ba from the Brzezinski, M.A., Burd, A., Carlson, C.A., D’Asaro, E.A., Doney, S.C., Perry, M.J., upper ocean estimated from 234Th/238U disequilibria. Deep-Sea Res. I Oceanogr. Res. Stanley, R.H.R., Steinberg, D.K., 2016. Prediction of the export and fate of global Pap. 52, 1477–1488. https://doi.org/10.1016/j.dsr.2005.03.008. ocean net primary production: the EXPORTS science plan. Front. Mar. Sci. 3 https:// doi.org/10.3389/fmars.2016.00022. 18