Distributions of total and size-fractionated particulate ²¹⁰Po and ²¹⁰Pb activities along the North Atlantic GEOTRACES GA01 transect: GEOVIDE cruise

Vertical distributions of total and particulate polonium-210 (²¹⁰Po) and lead-210 (²¹⁰Pb) activities in the water column were measured at 11 stations in the North Atlantic during the GEOTRACES GA01 transect: GEOVIDE cruise in May–June 2014. Total ²¹⁰Po activity was on average 24 % lower than ²¹⁰Pb activity in the upper 100 m, and it was closer to unity in the mesopelagic (100–1000 m). The partitioning coefficients (K_d) along the transect suggest the preferential association of ²¹⁰Po relative to ²¹⁰Pb onto particles. The prominent role of small particles in sorption was confirmed by the observation that over 80 % of the particulate radionuclide activity was on small particles. To account for the observed surface water ²¹⁰Po ∕ ²¹⁰Pb disequilibria, particulate radionuclide activities and export of both small (1–53 µm) and large (>53 µm) particles must be considered. A comparison between the GEOVIDE total particulate ²¹⁰Po ∕ ²¹⁰Pb activity ratios (ARs) and the ratios in previous studies revealed a distinct geographic distribution, with lower particulate ARs in the high-latitude North Atlantic (including this study) and Arctic in relation to all other samples. For the samples where apparent oxygen utilization (AOU) was calculated at the same depth and time as the ²¹⁰Po ∕ ²¹⁰Pb AR (40 stations including this study), there was a two-phase correlation between the total particulate AR and AOU, likely reflecting the nature of the particles and demonstrating the forces of remineralization and radionuclide decay from particles as they age.

1 Introduction

The major goal of the international GEOTRACES program is to characterize the distributions of trace elements and isotopes (TEIs) in the ocean on a global scale, and to identify and quantify processes that control these distributions (GEOTRACES Planning Group, 2006). The GEOVIDE section was a contribution of the French GEOTRACES program to this global program in the subpolar North Atlantic. The GEOVIDE GA01 cruise was carried out in 2014 in the North Atlantic and consisted of two sections: a section along the OVIDE (Observatoire de la variabilité interannuelle et décennale en Atlantique Nord) line between Lisbon (Portugal) and Cape Farewell (southern tip of Greenland), and a Cape Farewell–St. John's (Canada) section across the Labrador Sea (Fig. 1). Since 2002, the OVIDE section has been occupied biennially to collect physical and biogeochemical data (Mercier et al., 2015). The knowledge of the currents, water masses, and biogeochemical provinces gained from the previous OVIDE campaigns enabled the optimal strategy for TEI sampling and provided help for the interpretation of the distribution of TEIs in the subpolar North Atlantic (García-Ibáñez et al., 2015). In addition to the OVIDE line, the Labrador Sea section provided a unique opportunity to study TEI distributions along the boundary current of the western North Atlantic subpolar gyre (Sarthou et al., 2018).

Polonium-210 (²¹⁰Po, $T_{\mathrm{1}/\mathrm{2}}=\mathrm{138.4}$ days) and its radioactive grandparent lead-210 (²¹⁰Pb, $T_{\mathrm{1}/\mathrm{2}}=\mathrm{22.3}$ years) are two non-conservative ²³⁸U decay series products. The GEOTRACES program has included both radionuclides in its TEI list primarily due to ²¹⁰Po's enhanced bioaccumulation and the use of the ²¹⁰Po–²¹⁰Pb pair as a proxy for assessing particle export in the upper ocean. The distribution of ²¹⁰Po and ²¹⁰Pb has been widely measured over the last several decades in the Atlantic (e.g., Bacon et al., 1976; Sarin et al., 1999; Rigaud et al., 2015; Ceballos-Romero et al., 2016), Pacific (e.g., Nozaki and Tsunogai, 1976; Murray et al., 2005; Verdeny et al., 2008), Indian (e.g., Cochran et al., 1983; Sarin et al., 1994; Subha Anand et al., 2017), Arctic (e.g., Moore and Smith, 1986; He et al., 2015; Roca-Martí et al., 2016), and Southern oceans (e.g., Shimmield et al., 1995; Friedrich and Rutgers van der Loeff, 2002). However, since the data reported by Bacon et al. (1980b) at the Labrador Sea stations (47.8–53.7^∘ N), there have been few studies of ²¹⁰Po and ²¹⁰Pb activity in the North Atlantic at latitudes greater than 40^∘ N. The GEOVIDE cruise, which targeted the North Atlantic from 40 to 60^∘ N, provided an opportunity to fill this data gap.

Besides ascertaining the distribution of the natural radionuclides under specific geographic conditions, this project aimed to answer questions about their biogeochemical behaviors in various marine environments. Owing to the significantly longer half-life of ²¹⁰Pb relative to ²¹⁰Po, the two radionuclides are expected to be in secular equilibrium (total ²¹⁰Po ∕ ²¹⁰Pb activity ratio = 1) in the ocean, assuming no net removal or addition of either radionuclide. A deficit of ²¹⁰Po activity relative to ²¹⁰Pb activity (²¹⁰Po ∕ ²¹⁰Pb activity ratio <1), however, is commonly found in the upper ocean (e.g., Bacon et al., 1976; Nozaki and Tsunogai, 1976; Cochran et al., 1983; Sarin et al., 1999). This has been attributed to a higher particle reactivity of ²¹⁰Po (higher partitioning coefficient, K_d) than ²¹⁰Pb in seawater. Particles, therefore, become enriched in ²¹⁰Po (²¹⁰Po ∕ ²¹⁰Pb activity ratio >1), and their sinking to deeper waters results in a ²¹⁰Po activity deficit relative to ²¹⁰Pb activity in the upper water column where particles are formed.

In this work, we describe the distributions of total and size-fractionated particulate ²¹⁰Po and ²¹⁰Pb activity along the GEOVIDE cruise in the North Atlantic. These data are a significant contribution to the high-latitude North Atlantic ²¹⁰Po and ²¹⁰Pb activity data set. We present a compilation of particulate ²¹⁰Po ∕ ²¹⁰Pb activity ratios (ARs) from previous studies in the global ocean, and the results are discussed in regards to the aging of water and biochemical processes. We also describe the relationship among small particles, adsorption, and scavenging of radionuclides. These results lead to recommendations for the estimation of particulate organic carbon (POC) export flux based on the ²¹⁰Po ∕ ²¹⁰Pb disequilibrium, a topic that is covered in a companion paper (Tang et al., 2018).

2 Methods

2.1 Sample collection

The French GEOTRACES cruise to the North Atlantic (GEOVIDE, section GA01; 15 May–30 June 2014) was completed on the R/V Pourquoi Pas?. The research vessel departed from Lisbon, Portugal; headed northwest to the Greenland shelf; crossed the Labrador Sea; and ended in St John's, Newfoundland, Canada (Fig. 1). A rosette equipped with conductivity–temperature–depth (CTD) sensors and 12 L Niskin bottles was used to collect 200 seawater samples (5–10 L each) from 10 full-water-column “super” (multicast) stations (16–22 depths per station) and one “XLarge” (five-cast) station to 800 m (station 26, nine depths) for the determination of total ²¹⁰Po and ²¹⁰Pb activity. Upon recovery, seawater samples were transferred to 10 L acid-cleaned containers. In addition, particulate radionuclide activities in two size classes (1–53 and >53 µm) were collected at 3–10 depths per station using large-volume in situ filtration systems (Challenger Oceanic pumps and McLane pumps) equipped with 142 mm filter holders. Each filter head contained a stacked 53 µm PETEX screen followed by a 1 µm pore size quartz fiber QMA filter. The volume filtered was determined via flow meters mounted below each filter head, and the mean volume pumped through each head was 881 L. Once recovered, clear polyethylene caps were placed on the top of the pump heads, and they were brought into a clean laboratory for sub-sampling.

2.2 Total ²¹⁰Po and ²¹⁰Pb

Total ²¹⁰Po and ²¹⁰Pb activities were determined from the seawater samples by the cobalt–ammonium pyrrolidine dithiocarbamate (Co-APDC) technique (Fleer and Bacon, 1984). Samples were acidified to a pH<2 with concentrated HCl immediately after collection and spiked with known amounts of ²⁰⁹Po and stable lead as chemical yield tracers. After vigorous stirring and at least 12 h of isotope equilibration, cobalt nitrate and APDC solutions were added to co-precipitate Po and Pb. Samples were filtered through a 0.45 µm membrane filter, and the filters with the precipitate were placed into clean falcon tubes, sealed with Parafilm, and stored in double-bags. As the delay between sample collection and first Po plating increases, the uncertainty of the calculated ²¹⁰Po activity also increases. In addition, it is necessary to balance counting periods with the number of samples as the uncertainty due to alpha spectrometry counting decreases by increasing the counting time. To limit the delay between sampling and processing and to ensure higher counting statistics by having more alpha spectrometers devoted to this project, sample processing and analyses were split between Universitat Autònoma de Barcelona (UAB; samples from stations 1, 13, and 21) and Queens College (QC; stations 26, 32, 38, 44, 60, 69, and 77). Both laboratories followed the same procedure. Briefly, the filters were digested into a solution of concentrated HNO₃ and HCl; after the solution was evaporated to dryness, the samples were recovered in 1 and 0.5 M HCl solution at UAB and QC, respectively (a 0.5–2 M HCl solution is recommended; Rigaud et al., 2013). A polished pure silver disc (Flynn, 1968) with one side covered by enamel paint was placed into the weak acid solution and heated so that the polonium nuclides were spontaneously plated onto only one side of the disc. The activities of both Po nuclides on the disc were measured by alpha spectrometry. Any ²¹⁰Po and ²⁰⁹Po remaining in the plating solution were removed using AG 1-X8 anion exchange resin, and the final solution was re-spiked with ²⁰⁹Po and stored for more than 6 months to allow ingrowth of ²¹⁰Po from the decay of ²¹⁰Pb.

The ²¹⁰Pb activity was then determined by re-plating the solutions using silver discs and measuring the ingrown ²¹⁰Po. Two aliquots of the plating solutions for each sample were taken before the first and second platings for the measurement of total Pb concentration by inductively coupled plasma mass spectrometry (ICP-MS) to determine sample recovery during processing. The average recoveries produced by UAB and QC were 83±11 % (n=54) and 76±14 % (n=144), respectively. The activities of ²¹⁰Po and ²¹⁰Pb at the time of collection were determined by a series of corrections, including nuclide decay, ingrowth, chemical recoveries, detector backgrounds, and blank contamination following the methods in Rigaud et al. (2013). The activity uncertainties from UAB were on average 8 % for both ²¹⁰Po and ²¹⁰Pb activity, while the QC uncertainties were on average 13 % for ²¹⁰Po activity and 16 % for ²¹⁰Pb activity. The greater uncertainties of ²¹⁰Po and ²¹⁰Pb activities in the samples processed at QC were due to the longer delay between sampling and first plating (68 vs. 50 days) and higher uncertainties in the determination of the recovery of lead.

2.3 Particulate ²¹⁰Po and ²¹⁰Pb

After collection via in situ pumping, one quarter (equivalent to ∼220 L) of the PETEX screen containing >53 µm, or “large”, particles was processed for radionuclide activity. Swimmers were carefully removed from all samples. The QMA filters containing 1–53 µm, or “small”, particles were sub-sampled (two to four punches of 12 mm diameter), achieving a mean effective volume of ∼66 L. The screens and punches were stored in double bags at −80 ^∘C until the analyses onshore. The particulate samples were split between the two laboratories in parallel to the seawater samples. The filters were spiked with ²⁰⁹Po tracer solution and stable lead, digested using a mixture of concentrated HF, HNO₃, and HCl at UAB, but only HNO₃ and HCl at QC. After multiple rounds of digestion and evaporation to near dryness, the samples were recovered in 0.5 M HCl solution. Any remaining pieces of filter which were not completely digested were carefully removed, rinsed with 0.5 M HCl solution several times, and then discarded. The analyses of the particulate radionuclide activities were identical to those for the seawater samples described in Sect. 2.2.

2.4 Concentration of suspended particulate matter (SPM)

The Helene Planquette group (University of Brest, co-authors in this issue) collected subsamples from the same screens and filters that were sampled previously for radionuclides to determine major phase composition (particulate organic matter (POM), lithogenic material, calcium carbonate (CaCO₃), opal, Fe(OH)₃, and MnO₂) (Lam et al., 2015, and references therein). The mass concentration of SPM was calculated as the sum of the chemical dry weight of the major particulate phases.

The calculated SPM concentration was compared to the in situ transmission data obtained from the rosette CTD sensor (Fig. S1 in the Supplement). The overall negative relationship was statistically significant (R²=0.7, n=53, p<0.0001), suggesting that the SPM concentrations determined were reasonable estimates of particle concentration in the water column. We used the SPM values to determine the partitioning coefficient, K_d, for ²¹⁰Po and ²¹⁰Pb in Sect. 4.4.

2.5 Satellite-based data

The 8-day composites of surface chlorophyll-a concentration for each station were retrieved from NASA's MODIS products (https://oceancolor.gsfc.nasa.gov; last access: 6 September 2018) for the period from January to July 2014. The time-series chlorophyll-a concentrations were used to show the development of a phytoplankton bloom over time along the transect.

2.6 Historical values

The historical data of the particulate ²¹⁰Po and ²¹⁰Pb activity, and the hydrological parameters (pressure, temperature, salinity, and dissolved oxygen) were obtained from databases and publications. The location, date, database address or publication name, and type of data (particulate ²¹⁰Po and ²¹⁰Pb activity or hydrological parameters) from all other studies are listed in Table S1 in the Supplement.

Figure 2 The depth profiles of total ²¹⁰Po (²¹⁰Po_t, red circles) and ²¹⁰Pb activities (²¹⁰Pb_t, grey squares) along the GEOVIDE section. The horizontal blue line is the bottom depth, which coincided with the deepest water sample except for station 26, which was sampled only down to 1000 m. Note that the depth scale for each plot may be different. The profiles are shown in the order of sampling date with the region indicated on the top left: West European Basin (WEB), Iceland Basin (IB), Irminger Sea (IS), Labrador Sea (LS).

2.7 Apparent oxygen utilization

Apparent oxygen utilization ($\mathrm{AOU}={\mathrm{O}}_{\mathrm{2}\mathrm{saturated}}-{\mathrm{O}}_{\mathrm{2}\mathrm{measured}}$) is defined as the difference between the saturated oxygen at a given temperature and salinity and the measured in situ oxygen concentration (Ito et al., 2004; Duteil et al., 2013). A positive AOU indicates either water mass aging and outgassing of oxygen or biological activity, namely respiration (e.g., Keeling et al., 1998; Boyer et al., 1999). Negative AOU, indicating that the water is oversaturated with dissolved oxygen, can appear under the conditions of an intense bloom (e.g., Coppola et al., 2017).

The dissolved oxygen concentration was measured by Winkler titration, and the saturated oxygen concentration was calculated as a function of in situ temperature and salinity, and 1 atm of total pressure based on the built-in function in Ocean Data View (https://odv.awi.de; last access: 6 September 2018).

2.8 Statistical analyses

Statistical analyses were carried out in R Studio version 3 using fitting linear models and Welch two-sample t tests. Linear regression analysis was used to investigate the relationship between total particulate ²¹⁰Po ∕ ²¹⁰Pb AR and AOU. The Welch two-sample t test was applied to assess whether the mean of the total particulate ²¹⁰Po ∕ ²¹⁰Pb AR was the same as the mean of the small particulate ²¹⁰Po ∕ ²¹⁰Pb AR. It was also applied to investigate the means of the total ²¹⁰Pb activity in the western and eastern sections along the transect.

3 Results

3.1 Total ²¹⁰Po and ²¹⁰Pb activities

Total ²¹⁰Po activities (²¹⁰Po_t) in all samples ranged from 2.2 to 16.4 dpm 100 L⁻¹, and the mean ²¹⁰Po_t was 8.8±2.4 dpm 100 L⁻¹ (n=198, Fig. 2). ²¹⁰Po_t activities were generally low within the mixed-layer and euphotic zone (15–47 m), slightly increased or remained relatively constant in the depth range between the mixed layer and 250 m, and then decreased with water depth at most of the stations except station 26. Near the seafloor, stations 1, 13, and 44 had a slight increase of ²¹⁰Po_t activity.

Total ²¹⁰Pb activities (²¹⁰Pb_t) were between 2.1 and 20.6 dpm 100 L⁻¹ with a mean value of 10.0±3.0 dpm 100 L⁻¹ (n=198, Fig. 2). ²¹⁰Pb_t activities were low in the surface, slightly increased in the subsurface, and decreased with water depth. Stations 1, 13, 44, and 60 exhibited an increase near the seafloor.

Figure 3 The upper 250 m of the depth profiles of total ²¹⁰Po (²¹⁰Po_t, red circles) and ²¹⁰Pb activities (²¹⁰Pb_t, grey squares) along the GEOVIDE section. The horizontal orange and magenta lines denote the mixed-layer depth (MLD) and the base of the euphotic zone (Z_1 %), respectively. The depth profiles are shown in the order of sampling and grouped by region (refer to Fig. 2 for the text abbreviations).

The mean ²¹⁰Po_t ∕ ²¹⁰Pb_t AR of all samples was 0.92±0.28 (n=198, Fig. 2). When considering different basins separately, there is a tendency of decreasing ²¹⁰Po_t ∕ ²¹⁰Pb_t AR from the West European Basin (1.10±0.35) westwards to the Iceland Basin (0.90±0.19) and the Irminger Sea and the Labrador Sea (0.80±0.18 and 0.83±0.21, respectively).

Figure 4 Vertical profiles of the particulate ²¹⁰Po and ²¹⁰Pb activity in the small size fraction (1–53 µm, ²¹⁰Po_s, ²¹⁰Pb_s). Note that the depth scale may differ among plots, and the activity scale at station 44 differs from the scale in all other plots. The horizontal blue line represents the bottom depth at that station.

For all regions, significant deficits of ²¹⁰Po_t (0.80±0.20, n=40) were observed within the mixed-layer and euphotic zone (Fig. 3). Secular equilibrium was also observed at some shallow depths (i.e., 80 m at station 44) and even in surface waters (i.e., 15 m at station 38). ²¹⁰Po_t excesses relative to ²¹⁰Pb_t, which were larger than ²¹⁰Po_t surface depletions at the same stations, were observed below the surface at some depths at stations 1, 13, and 21 in the West European Basin (Fig. 2). At depths below the surface to ∼1500 m in the Iceland Basin, the Irminger Sea, and the Labrador Sea, the water samples still indicated a ²¹⁰Po deficiency (AR: 0.84±0.17; n=27). Secular equilibrium was generally reached near the bottom depths in all basins except at stations 13 and 60, where the water samples were enriched in ²¹⁰Po_t (²¹⁰Po_t ∕ ²¹⁰Pb_t AR = 1.58±0.16) and depleted in ²¹⁰Po_t (²¹⁰Po_t ∕ ²¹⁰Pb_t AR = 0.50±0.12), respectively.

Figure 5 The vertical profiles of the particulate ²¹⁰Po and ²¹⁰Pb activity in the large size fraction (>53 µm, ²¹⁰Po_l, ²¹⁰Pb_l) in the top 800 m. Note that the activity scale at station 26 differs from the scale in all other plots.

3.2 Particulate ²¹⁰Po and ²¹⁰Pb activities

Small particulate ²¹⁰Po (²¹⁰Po_s) activities varied in a wide range from 0.08 to 4.82 dpm 100 L⁻¹ (mean: 0.76±0.63 dpm 100 L⁻¹; n=81); about 83 % of the values in the small particles were lower than 1.0 dpm 100 L⁻¹, with higher ²¹⁰Po_s values generally observed in the surface samples (Fig. 4, Table S2). The range of small particulate ²¹⁰Pb (²¹⁰Pb_s) activities was 0.07 to 2.89 dpm 100 L⁻¹ (mean: 0.56±0.46 dpm 100 L⁻¹; n=81). The vertical profiles of ²¹⁰Pb_s were generally similar to those of ²¹⁰Po_s, with relatively high activity in the surface, lower activity in the subsurface, and increasing activity with depth (Fig. 4). This has been seen in the North Atlantic along the GEOTRACES GA03 transect (Rigaud et al., 2015). The mean ²¹⁰Po_s ∕ ²¹⁰Pb_s AR was 1.43±0.96 in the surface waters (n=14, ≤47 m) and 1.57±0.90 with all samples included (n=81, 8–3440 m). While most surface observations had an AR of ²¹⁰Po_s ∕ ²¹⁰Pb_s higher than unity, five surface samples at stations 69 and 77 showed an enrichment of ²¹⁰Pb activity over ²¹⁰Po (²¹⁰Po_s ∕ ²¹⁰Pb_s AR: 0.62±0.18).

Large particulate ²¹⁰Po (²¹⁰Po_l) activities ranged from 0.01 to 0.83 dpm 100 L⁻¹ with a mean of 0.10±0.12 dpm 100 L⁻¹ (n=59, Fig. 5, Table S2). The range of ²¹⁰Pb activity in the large particles (²¹⁰Pb_l) was from 0.02 to 0.67 dpm 100 L⁻¹ (mean: 0.12±0.14 dpm 100 L⁻¹; n=59). The highest ²¹⁰Po_l and ²¹⁰Pb_l values were found at 30 m at station 26. The mean ²¹⁰Po_l ∕ ²¹⁰Pb_l AR was 1.09±1.54 in the surface waters (n=14, ≤47 m) and 1.06±0.86 when all data were considered (n=59, 8–800 m). There were 17 % of the samples with a depletion of ²¹⁰Po activity relative to ²¹⁰Pb activity in large particles (mean AR: 0.49±0.23), particularly in surface waters from the western section. We address this issue further in Sect. 4.2 and 4.3.

The percentages of total ²¹⁰Po activity in the small and large particles ranged from 0.9 % to 46.7 % (mean: 8.0±6.7 %) and from 0.1 % to 8.9 % (mean: 1.2±1.5 %), respectively. The percentage of total ²¹⁰Pb activity ranged from 0.7 % to 21.4 % (mean: 4.9±3.8 %) and from 0.2 to 5.9 % (mean: 1.1±1.2 %) in the small and large particulate phase, respectively. These values revealed that both radionuclides were predominantly present in the dissolved phase along this transect, as is commonly found in the ocean. The particulate percentages reported here are similar to the values reported from the FS “Meteor” cruise 32 in the North Atlantic (Bacon et al., 1976) and along the North Atlantic GA03 transect (Rigaud et al., 2015).

We then combined radionuclide activity on the small and large particles from the same depth as the total particulate activity. There were 56 samples in total (surface to 800 m), and 41 of them were from the upper 200 m. Most of the total particulate ²¹⁰Po (²¹⁰Po_p) and ²¹⁰Pb (²¹⁰Pb_p) activity was on the small particles, with 86 % of ²¹⁰Po_p and 80 % of ²¹⁰Pb_p on the small size fraction (data not shown). The total particulate ²¹⁰Po and ²¹⁰Pb AR (²¹⁰Po_p ∕ ²¹⁰Pb_p) had the same mean as that of the small particulate ²¹⁰Po and ²¹⁰Pb AR (²¹⁰Po_s ∕ ²¹⁰Pb_s) (Welch two-sample t test, n=56, p=0.1), indicating that the values of the ²¹⁰Po_p ∕ ²¹⁰Pb_p activity ratios were driven by the small particles. While the majority of particulate matter was enriched in ²¹⁰Po (²¹⁰Po_p ∕ ²¹⁰Pb_p AR > 1), there were 13 out of 56 total samples from various depths that were depleted in ²¹⁰Po relative to ²¹⁰Pb. The ²¹⁰Po_p ∕ ²¹⁰Pb_p activity ratios from this study are compared to the results from previous studies in various oceanic regimes in Sect. 4.2.

4 Discussion

4.1 Total ²¹⁰Po and ²¹⁰Pb activities

The overall profiles of ²¹⁰Po_t and ²¹⁰Pb_t activities were different among basins (Fig. 2). The deficiencies of ²¹⁰Po_t activities with respect to ²¹⁰Pb_t activities in the surface samples from the Iceland Basin, the Irminger Sea, and the Labrador Sea were generally greater than those from the West European Basin. Such disequilibria generally extended to the deep waters (1700–2950 m). In contrast, ²¹⁰Po_t activities in the West European Basin were generally enriched relative to ²¹⁰Pb_t activities from below the surface to the bottom of the profile. In the West European Basin, the subsurface ²¹⁰Po_t activity excess was much larger than the surface depletion, suggesting that some external source would be needed to maintain this excess ²¹⁰Po activity within the water column. One possible source of these subsurface ²¹⁰Po activity excesses below 2000 m at stations 1 and 13 could be the lower Northeast Atlantic Deep Water (NEADW_L), which was the dominant water mass in the Iberian Basin from 2000 m to the bottom and had a concentration of silicate up to 48 µmol kg⁻¹ (García-Ibáñez et al., 2015). High activity of ²¹⁰Po in deep samples could be due to the dissolution of diatoms or herbivore feces (Cooper, 1952). As these particles sink and dissolve, ²¹⁰Po activity may have been preferentially released to the dissolved phase compared to ²¹⁰Pb activity (Bacon et al., 1976), leading to ²¹⁰Po excess observed in the deep waters at stations 1 and 13. For the subsurface ²¹⁰Po activity excesses at station 1 between 400 and 1000 m, where lateral inputs of particulate Fe (PFe) from the margin was observed (Gourain et al., 2018), the likely process is diffusion of ²¹⁰Po from those particles originated from the margin, and such excess could be transported westwards to station 13 by lateral advection. An alternative source of ²¹⁰Po activity excess between 50 and 250 m at stations 1 and 13 (Fig. 3) could be the eastern boundary upwelling along the coast of the Iberian Peninsula (García-Ibáñez et al., 2015). Even though no strong upwelling events were revealed from temperature and density profiles during the cruise, northerly winds favoring upwelling were recorded 2–3 months before the sampling (Shelley et al., 2017). The deep water may have excess ²¹⁰Po activity due to the remineralization of sinking particles. The upwelling of this water mass prior to the sampling date could maintain such subsurface excess ²¹⁰Po activity. Similar findings have been reported in the Cariaco Trench for the upper 300 m of the water column by Bacon et al. (1980a).

As atmospheric deposition is the main source of ²¹⁰Pb to the water column (e.g., Masqué et al., 2002), we divided the GA01 transect into a western section (stations 44–77) and an eastern section (stations 1–38) based on atmospheric deposition boxes described in Shelley et al. (2017). Total atmospheric deposition fluxes of a suite of aerosol-sourced trace metals (TEs) were reported to be higher in the east than the west for 18 out of 19 TEs (Shelley et al., 2017). However, a two-sample t test revealed a greater mean of ²¹⁰Pb_t activity in surface waters in the western than in the eastern section (p<0.02; mean: 12.1 vs. 10.4 dpm 100 L⁻¹), despite the fact that ²¹⁰Pb is usually associated with aerosols. Even though the direct input of atmospheric ²¹⁰Pb may be larger in the east (assuming it behaves like the other trace metals, but without aerosol ²¹⁰Pb data we cannot confirm this), alternative inputs of ²¹⁰Pb from freshwater (e.g., sea ice processes and meteoric water) could be a greater source of ²¹⁰Pb activity to the west. The freshwater sources over the Greenland shelf and slope have been identified by Benetti et al. (2017) and were believed to be an important source of Fe (Tonnard et al., 2018) and Al (Menzel Barraqueta et al., 2018) off of Greenland during this cruise. This result highlights the need in the future to measure ²¹⁰Pb activity simultaneously in the atmospheric and local freshwater sources in order to account for all source terms.

Figure 6 Time-series (1 January–12 July 2014) chlorophyll-a concentrations (8-day averages) from Aqua MODIS (https://oceancolor.gsfc.nasa.gov; last access: 6 September 2018) at each station along the GA01 transect. The vertical red line denotes the sampling date at each station. The horizontal blue line denotes chlorophyll-a concentration of 0.5 mg m⁻³. The time when chlorophyll-a concentration first exceeded 0.5 mg m⁻³ after the end of the last bloom defines the date when the next bloom began.

4.2 Total particulate ²¹⁰Po ∕ ²¹⁰Pb AR

A proposed explanation for the depletion of ²¹⁰Po activity relative to ²¹⁰Pb activity (AR < 1) in some particles is effective recycling, commonly characterized by a subsurface excess of dissolved ²¹⁰Po activity released from enriched particles leaving the surface. Bacon et al. (1976) suggested that the efficiency of this recycling could reach up to 50 %, while there is no significant concurrent release of ²¹⁰Pb activity in the water column. Laboratory studies have found the release rate of ²¹⁰Po in marine particulate matter to be significant; for example, 41 % of the ²¹⁰Po activity in euphausiid fecal pellets was released over 5 days as presented in Heyraud et al. (1976). An alternative explanation for the depletion of ²¹⁰Po activity in particles is their lithogenic origin. ²¹⁰Po ∕ ²¹⁰Pb AR in lithogenic particles was reported to be similar to or less than unity (Nozaki et al., 1998; Tateda et al., 2003). In addition, the AR < 1 observed at station 1 (120, 250, and 550 m) could be associated with lithogenic particles from the Iberian Margin, where 100 % of the PFe had a lithogenic origin, while the lithogenic contribution to PFe at other stations was smaller (Gourain et al., 2018).

Figure 7 Depths at which the total particulate (>1 µm) ²¹⁰Po ∕ ²¹⁰Pb activity ratio was lower than unity vs. the time since the last bloom (data are presented in Table 1).

The time-series chlorophyll-a concentrations (8-day composite; https://oceancolor.gsfc.nasa.gov, last access: 6 September 2018) from January to July 2014 at each station revealed bloom conditions about 4 months prior to the sampling time (Fig. 6). We estimated the days since the last bloom began prior to the sampling date for each station (Table 1) and put these data into the context of the low ²¹⁰Po_p ∕ ²¹⁰Pb_p AR (<1) in the total particles >1 µm (Fig. 7). Eight stations had total particulate samples with ²¹⁰Po_p ∕ ²¹⁰Pb_p AR lower than unity from either shallow or deep waters. Specifically, when the time since the last bloom began was relatively short (25–47 days), the samples with ²¹⁰Po_p ∕ ²¹⁰Pb_p AR < 1 were observed in the shallow waters (10–60 m). In contrast, when more time (50–74 days) had passed since the last bloom, the depths at which samples had ²¹⁰Po_p ∕ ²¹⁰Pb_p AR < 1 were found to be much deeper (120–500 m). The results indicated that post-bloom particles could be recycled for weeks in shallow depths and take weeks to months to sink to deeper waters.

Figure 8 Comparison of particulate ²¹⁰Po ∕ ²¹⁰Pb activity ratios in the upper 200 m from this study and 20 previous studies (references in Table 2). Information about the study site, sampling date, method, and particle size of each study is shown in Table 2. The black circles represent data from previous studies, while the blue circles are the results from samples analyzed at QC from three recent GEOTRACES transects (GA03; GP16; and this study: GA01 GEOVIDE). The filled magenta and open circles indicate activity ratios lower and higher than 1, respectively.

The averages of ²¹⁰Po_p ∕ ²¹⁰Pb_p AR within the upper 200 m water column were put into a global context with previously reported results (Fig. 8). Total particulate ²¹⁰Po ∕ ²¹⁰Pb ARs in the open ocean in previous studies (e.g., equatorial western Pacific, Bellingshausen Sea, Bermuda Atlantic Time-series Study (BATS) site, Labrador Sea) were generally greater than unity. In contrast to the open ocean, the data show a distinct trend of depletion of relative ²¹⁰Po activity in marine particles from the shallow seas of the high-latitude Northern Hemisphere. The lowest total particulate ²¹⁰Po ∕ ²¹⁰Pb AR values (Table 2, 0.4–0.5) were found in the Chukchi shelf (He et al., 2015) and other seas from the Eurasian sector (Barents, Kara, and Laptev seas) as well as in central Arctic (Friedrich, 2011). Previous studies have observed depletion of relative ²¹⁰Po activity in nearshore particles in the Yellow Sea (Hong et al., 1999); in the turbid waters off of western Taiwan (Wei et al., 2012); on the shelf of Woods Hole, MA, USA (Rigaud et al., 2015); and now in the margin station off St. John's, Canada (this study). The previous authors attributed the relative depletion of particulate ²¹⁰Po activity in the nearshore waters to the terrestrial origin/riverine input of particles with a low ²¹⁰Po ∕ ²¹⁰Pb AR. This may partially explain low activity ratios in the samples from the shelf of the Arctic Ocean as well, since it receives ∼10 % of global river runoff and is the most riverine-influenced of all of the world's oceans (Opsahl et al., 1999; Carmack et al., 2006). The Arctic Basin, similarly, had widespread deficits of particulate ²¹⁰Po activity in the upper water column during the sea ice minimum in 2007 (Roca-Martí et al., 2018). Besides shelf particles, the authors suggest that other particle types could also play a role in lowering the particulate AR, including sea ice sediments, remineralized material, fecal pellets, and picoplankton aggregates.

Table 1 Biological characteristics of the water column determined by chlorophyll-a concentration (8-day composite) from Fig. 6, including the date when the last bloom began, the difference in chlorophyll-a concentration between the sampling time and last bloom peak, and the days since the last bloom. Activity ratios of ²¹⁰Po_p ∕ ²¹⁰Pb_p<1 and their corresponding depths are also shown. NA indicates that all samples from the corresponding depth range had ²¹⁰Po_p ∕ ²¹⁰Pb_p equal to or greater than 1 (no sample with ²¹⁰Po_p ∕ ²¹⁰Pb_p<1).

Table 2 The compilation of total particulate ²¹⁰Po ∕ ²¹⁰Pb activity ratios (²¹⁰Po_p ∕ ²¹⁰Pb_p) averaged in the upper 200 m, including this study.

4.3 Relationship between total particulate ²¹⁰Po ∕ ²¹⁰Pb AR and AOU

AOU is a time-integrated measure of the amount of oxygen removed during the biogeochemical processes (e.g., respiration, remineralization, oxidation) in the ocean interior. Therefore, AOU is a product of apparent oxygen utilization rate (AOUR) and the age of water mass (e.g., Stanley et al., 2012); i.e., high AOU could be due to either intense biogeochemical processes that have occurred in a short period of time (young water mass) or weaker processes over a longer period of time (old water mass). Consequently, the rate of these biogeochemical processes and time (water mass age) would have different or similar impacts on the ²¹⁰Po_p ∕ ²¹⁰Pb_p AR value depending on the initial AR in the particles and the natural of the particles. For example, the ²¹⁰Po_p ∕ ²¹⁰Pb_p AR would tend to increase with time if the initial AR were <1 because particulate ²¹⁰Po activity would increase from the decay of ²¹⁰Pb and trend towards secular equilibrium (²¹⁰Po_p ∕ ²¹⁰Pb_p AR = 1), and to decrease with time if the initial AR were >1 as the original excess of particulate ²¹⁰Po activity would disappear after seven half-lives of ²¹⁰Po. In contrast, oxygen consumption due to bacterial remineralization would preferentially release ²¹⁰Po activity from particles into the dissolved pool (e.g., Stewart et al., 2008), leading to a lower ²¹⁰Po_p ∕ ²¹⁰Pb_p AR in those particles.

Figure 9 The relationship between AOU (µmol kg⁻¹) and total particulate ²¹⁰Po ∕ ²¹⁰Pb activity ratio (²¹⁰Po_p ∕ ²¹⁰Pb_p) from the upper 200 m in the Northern Hemisphere (>22^∘ N) investigated by a linear regression model (red and blue lines). The 40 stations include data from previous studies – ARK-XXII/2 (77.38–87.83^∘ N, n=15) in the Arctic and BOFS (48.89–49.87^∘ N, n=7), GA03 (22.38–39.70^∘ N, n=7), and GA01 (this study; 40.33–59.80^∘ N, n=11) – in the North Atlantic. The horizontal dashed line represents ²¹⁰Po_p ∕ ²¹⁰Pb_p AR = 1, and the vertical dashed line represents AOU = 25 µmol kg⁻¹. Blue circles denote AOU < 25 µmol kg⁻¹, while red circles denote AOU > 25 µmol kg⁻¹.

The combination of average ²¹⁰Po_p ∕ ²¹⁰Pb_p AR and their corresponding average AOU in the upper 200 m at 40 stations from four independent studies – including ARK-XXII/2 (77.38–87.83^∘ N, n=15) in the Arctic and BOFS (48.89–49.87^∘ N, n=7), GA03 (22.38–39.70^∘ N, n=7), and GA01 (this study; 40.33–59.80^∘ N, n=11) in the North Atlantic (see map in Fig. 8) – suggests two distinct linear trends (Fig. 9). When AOU was lower than 25 µmol kg⁻¹, the ²¹⁰Po_p ∕ ²¹⁰Pb_p AR was found to be greater than unity, together with a linear negative relationship (n=27, R²=0.5, p<0.001) towards the AOU at 25 µmol kg⁻¹. In contrast, AOU values greater than 25 µmol kg⁻¹ were coincident with a ²¹⁰Po_p ∕ ²¹⁰Pb_p AR < 1, and a linear positive relationship (n=12, R²=0.4, p=0.03) towards the highest AOU values was measured. The two contradictory linear trends likely reflect the nature of the particles. For example, the observation of ²¹⁰Po_p ∕ ²¹⁰Pb_p AR>1 with AOU<25 µmol kg⁻¹ may suggest relatively fresh/organic particles in the young water mass. When AOU increases due to either water mass aging or higher AOUR, the ²¹⁰Po_p ∕ ²¹⁰Pb_p AR decreases with a slope of $-\mathrm{0.17}\pm \mathrm{0.04}$. On the other hand, refractory/lithogenic particles may be suggested by the observation of ²¹⁰Po_p ∕ ²¹⁰Pb_p AR<1 with AOU>25 µmol kg⁻¹. For those particles, increasing in AOU due to either water mass aging or higher AOUR would change the ²¹⁰Po_p ∕ ²¹⁰Pb_p AR to a much lesser degree than that for organic particles with a slope of 0.008±0.003. This explanation, however, appears to only hold for the high-latitude Northern Hemisphere, where ²¹⁰Po_p ∕ ²¹⁰Pb_p activity ratios were generally lower than those in the other oceanic settings (Fig. 8). In the high-latitude Southern Hemisphere near Antarctic (e.g., ANT-X/6), for example, there is no apparent relationship between ²¹⁰Po_p ∕ ²¹⁰Pb_p activity ratios and AOU. This relationship (or lack thereof) deserves more study in the future.

4.4 Relationship among small particles, adsorption, and scavenging

The partitioning coefficient, K_d (L kg⁻¹), has been used to describe the particle adsorption behavior of radionuclides. It is defined as the ratio of the adsorbed radionuclide activity (A_p, dpm 100 L⁻¹) to the dissolved radionuclide activity (A_d, dpm 100 L⁻¹), normalized by the SPM concentration (µg L⁻¹):

$$\begin{array}{}\text{(1)}& K_{\mathrm{d}}={\displaystyle \frac{A_{\mathrm{p}}}{A_{\mathrm{d}}}}\times {\displaystyle \frac{\mathrm{1}}{\mathrm{SPM}}}{\mathrm{10}}^{\mathrm{9}}.\end{array}$$

Owing to the different biological and chemical behaviors of ²¹⁰Po and ²¹⁰Pb, the interpretation of measured K_d for ²¹⁰Po (K_d(Po)) may not be as clear as that for ²¹⁰Pb (K_d(Pb)). As claimed previously in Tang et al. (2017), K_d(Po) is complicated because it appears to reflect both the surface adsorption and potential bioaccumulation.

In this study, the size-fractionated data of both radionuclide activity and SPM allowed us to calculate the partitioning coefficients for both radionuclides on small and total particles. The dissolved radionuclide activity was calculated as the difference between total and particulate activity. The coefficients for the small particulate and the total particulate phases were normalized by the SPM in the small and total particulate phases, respectively. We present the coefficients for the small particulate phases (K_d(Po)_s, K_d(Pb)_s) and the total particulate phases (K_d(Po)_p, K_d(Pb)_p) because most of the particulate activity (>80 %) was associated with the small particles along the GEOVIDE transect, and most conceptualized scavenging models consider either the two-box model (dissolved–total particulate phases, i.e., K_d(Po)_p) or the three-box model (dissolved–small–large, i.e., K_d(Po)_s) (Clegg and Whitfield, 1990, 1991; Rigaud et al., 2015), and thus activity is concentrated from the dissolved phase to the total or small particles.

Figure 10 Comparison of the partitioning coefficient (K_d, L kg⁻¹) between the dissolved and small particulate phases (K_d(Po)_s, K_d(Pb)_s) vs. between the dissolved and total particulate phases (K_d(Po)_p, K_d(Pb)_p) for (a) ²¹⁰Po and (b) ²¹⁰Pb. The 1 : 1 line is indicated as the solid line in each plot.

The average values of K_d(Po) were 1.6 times of those of K_d(Pb) in both small and total particulate phases, suggesting a higher affinity with particles for ²¹⁰Po with respect to ²¹⁰Pb, which is commonly observed in the global ocean (Bacon et al., 1988; Hong et al., 1999; Masqué et al., 2002; Wei et al., 2014; Tang et al., 2017). The K_d values for the small particulate phase were slightly higher than those for the total particulate phase, but overall these values were very similar for both radionuclides (Fig. 10), suggesting that adsorption/scavenging of radionuclides was driven by small particles along the transect. In addition, there are increasing studies which argue that small particles can form aggregates that sink, and their contribution to carbon export could be larger than previously thought (e.g., Richardson and Jackson, 2007; Lomas and Moran, 2011; Amacher et al., 2013; Puigcorbé et al., 2015). We, therefore, recommend combining the activities of both small and large particles into a total particulate fraction in order to explain total ²¹⁰Po ∕ ²¹⁰Pb disequilibria in the surface waters, and utilizing the characteristics of the total particles (instead of just the large particles) in the estimation of the POC export fluxes (Tang et al., 2018).

Traditionally, large particles collected by in situ filtration with pumps, most commonly defined as particles larger than 53 or 70 µm, were assumed to dominate the sinking flux (Dugdale and Goering, 1967; Bishop et al., 1977; Fowler and Knauer, 1986; Honjo et al., 1992; Walsh and Gardner, 1992) such that the composition (POC ∕ ²¹⁰Po) of the large particle size class was used to convert ²¹⁰Po fluxes into POC export (e.g., Friedrich and Rutgers van der Loeff, 2002; Cochran and Masqué, 2003; Murray et al., 2005; Stewart et al., 2010; Roca-Martí et al., 2016). Given that the true size spectrum of sinking particles for the timescale relevant to the ²¹⁰Po ∕ ²¹⁰Pb method is unknown and the POC flux estimates are sensitive to the particulate POC ∕ ²¹⁰Po ratio, both small and large particles should be sampled for POC ∕ ²¹⁰Po due to the variability in the POC ∕ ²¹⁰Po ratio in different size classes (Hayes et al., 2018).

5 Conclusions

In this study, we reported the vertical distribution of total and size-fractionated particulate ²¹⁰Po and ²¹⁰Pb activities in the North Atlantic during the GEOVIDE GA01 cruise. More than 90 % of the radionuclide activity was found in the dissolved phase, while a small proportion was associated with particles in this transect. Total ²¹⁰Po activity was generally depleted relative to total ²¹⁰Pb activity in the upper 100 m due to the preferential adsorption of ²¹⁰Po activity by particles. Such deficiencies of ²¹⁰Po activities generally extended to the deep waters at most of the stations. In the West European Basin, the excess of ²¹⁰Po activities at stations 1 and 13 in the Northeast Atlantic Deep Water was attributed to the release of ²¹⁰Po during dissolution of sinking biogenic particles.

There appear to be geographic differences in particulate ²¹⁰Po ∕ ²¹⁰Pb activity ratios measured during GEOVIDE and previous studies, with particularly low values in the high-latitude North Atlantic and Arctic. While this observation deserves more attention, we support previous suggestions that this is due to the terrestrial origin/riverine input of particles with a low ²¹⁰Po ∕ ²¹⁰Pb AR into the river-dominated shallow seas of the Arctic. The age of the particles and water masses as well as the importance of biogeochemical processes (e.g., respiration, remineralization) may also explain some of these observations, as there was a significant relationship between the total particulate activity ratio and AOU when both were measured in the North Atlantic (>20^∘ N) and Arctic oceans.

Over 80 % of the particulate radionuclide activity was on small particles, indicating that the scavenging of both radionuclides was driven by small particles. Therefore, we suggest considering the activities of ²¹⁰Po and ²¹⁰Pb from both small and large particles in order to study the water column ²¹⁰Po ∕ ²¹⁰Pb disequilibria and quantify POC export along the GA01 transect. This has been addressed in a companion paper in this issue. We recommend that both small and large particles should be sampled for POC ∕ ²¹⁰Po estimates for the application of the ²¹⁰Po ∕ ²¹⁰Pb method in future studies of POC export.