Quantifying 210Po/210Pb Disequilibrium in Seawater: A Comparison of Two Precipitation Methods With Differing Results

The disequilibrium between lead-210 (210Pb) and polonium-210 (210Po) is increasingly used in oceanography to quantify particulate organic carbon (POC) export from the upper ocean. This proxy is based on the deficits of 210Po typically observed in the upper water column due to the preferential removal of 210Po relative to 210Pb by sinking particles. Yet, a number of studies have reported unexpected large 210Po deficits in the deep ocean indicating scavenging of 210Po despite its radioactive mean life of ∼ 200 days. Two precipitation methods, Fe(OH)3 and Co-APDC, are typically used to concentrate Pb and Po from seawater samples, and deep 210Po deficits raise the question whether this feature is biogeochemically consistent or there is a methodological issue. Here, we present a compilation of 210Pb and 210Po studies that suggests that 210Po deficits at depths >300 m are more often observed in studies where Fe(OH)3 is used to precipitate Pb and Po from seawater, than in those using Co-APDC (in 68 versus 33% of the profiles analyzed for each method, respectively). In order to test whether 210Po/210Pb disequilibrium can be partly related to a methodological artifact, we directly compared the total activities of 210Pb and 210Po in four duplicate ocean depth-profiles determined by using Fe(OH)3 and Co-APDC on unfiltered seawater samples. While both methods produced the same 210Pb activities, results from the Co-APDC method showed equilibrium between 210Pb and 210Po below 100 m, whereas the Fe(OH)3 method resulted in activities of 210Po significantly lower than 210Pb throughout the entire water column. These results show that 210Po deficits in deep waters, but also in the upper ocean, may be greater when calculated using a commonly used Fe(OH)3 protocol. This finding has potential implications for the use of the 210Po/210Pb pair as a tracer of particle export in the oceans because 210Po (and thus POC) fluxes calculated using Fe(OH)3 on unfiltered seawater samples may be overestimated. Recommendations for future research are provided based on the possible reasons for the discrepancy in 210Po activities between both analytical methods.

The disequilibrium between lead-210 ( 210 Pb) and polonium-210 ( 210 Po) is increasingly used in oceanography to quantify particulate organic carbon (POC) export from the upper ocean. This proxy is based on the deficits of 210 Po typically observed in the upper water column due to the preferential removal of 210 Po relative to 210 Pb by sinking particles. Yet, a number of studies have reported unexpected large 210 Po deficits in the deep ocean indicating scavenging of 210 Po despite its radioactive mean life of ∼ 200 days. Two precipitation methods, Fe(OH) 3 and Co-APDC, are typically used to concentrate Pb and Po from seawater samples, and deep 210 Po deficits raise the question whether this feature is biogeochemically consistent or there is a methodological issue. Here, we present a compilation of 210 Pb and 210 Po studies that suggests that 210 Po deficits at depths >300 m are more often observed in studies where Fe(OH) 3 is used to precipitate Pb and Po from seawater, than in those using Co-APDC (in 68 versus 33% of the profiles analyzed for each method, respectively). In order to test whether 210 Po/ 210 Pb disequilibrium can be partly related to a methodological artifact, we directly compared the total activities of 210 Pb and 210 Po in four duplicate ocean depth-profiles determined by using Fe(OH) 3 and Co-APDC on unfiltered seawater samples. While both methods produced the same 210 Pb activities, results from the Co-APDC method showed equilibrium between 210 Pb and 210 Po below 100 m, whereas the Fe(OH) 3 method resulted in activities of 210 Po significantly lower than 210 Pb throughout the entire water column. These results show that 210 Po deficits in deep waters, but also in the upper ocean, may be greater when calculated using a commonly used

INTRODUCTION
The biological carbon pump is a major mechanism for removing carbon dioxide from the atmosphere principally mediated by the transfer of organic particles from the surface to the deep ocean by different export pathways (Boyd et al., 2019). The flux of particulate organic carbon (POC) varies strongly across regions and time (Buesseler and Boyd, 2009), hindering the estimation of POC export in the global ocean (ranging from 5 to >12 Pg C yr −1 ; Boyd and Trull, 2007;Henson et al., 2011;Siegel et al., 2014). The naturally occurring radionuclides lead-210 ( 210 Pb, half-life = 22.3 years) and polonium-210 ( 210 Po, half-life = 138 days) have been widely used as particle tracers in the marine environment for decades (e.g., Bacon et al., 1976Bacon et al., , 1988Cochran and Masqué, 2003) and, in the last years, especially, to quantify POC export (e.g., Tang and Stewart, 2019). Indeed, the combination of 210 Po/ 210 Pb with other particle export methods, such as sediment traps or the thorium-234/uranium-238 ( 234 Th/ 238 U) pair, has given a broader perspective on downward export fluxes integrating time scales ranging from a few days to several months (Stewart et al., 2007(Stewart et al., , 2011Buesseler et al., 2008;Verdeny et al., 2009;Wei et al., 2011;Le Moigne et al., 2013;Ceballos-Romero et al., 2016;Maiti et al., 2016;Roca-Martí et al., 2016;Anand et al., 2018;Hayes et al., 2018).
The use of 210 Po/ 210 Pb disequilibrium as a proxy for POC export relies on the assumption that sinking of organic particles generates a deficit of 210 Po with respect to 210 Pb in the upper water column due to preferential scavenging of Po compared to Pb (Friedrich and Rutgers van der Loeff, 2002;Verdeny et al., 2009). Unlike 210 Pb (or 234 Th), Po can be assimilated into cells, possibly as an analog of sulfur, and subsequently cycled as the organic matter is regenerated (Stewart et al., 2008). In the particle-poor deep ocean, radioactive equilibrium between 210 Pb and 210 Po is anticipated because of the long scavenging residence times and short half-life of 210 Po (e.g., Bacon et al., 1976;Cochran et al., 1983). Yet, large deficits of 210 Po have been observed in the mesopelagic (∼ 100-1000 m) and bathypelagic (>1000 m) zones in different regions of the world ocean, including the North Atlantic (e.g., Kim and Church, 2001;Hong et al., 2013;Rigaud et al., 2015), the North, Equatorial and South Pacific (e.g., Thomson and Turekian, 1976;Nozaki et al., 1990Nozaki et al., , 1997Chung and Wu, 2005;Hu et al., 2014), the Arctic Ocean (e.g., Smith et al., 2003;Roca-Martí et al., 2018), and the Southern Ocean (e.g., Friedrich and Rutgers van der Loeff, 2002). Disequilibrium in deep waters has been commonly attributed to the scavenging of 210 Po by particles from the local upper water column or from the shelves due to high biological productivity (Hu et al., 2014;Rigaud et al., 2015;Ma et al., 2017). Recently, a model study by De Soto et al. (2018) has argued that the disequilibrium between 210 Pb and 210 Po at depth in the Porcupine Abyssal Plain, NE Atlantic, could be explained by a significant adsorption of 210 Po onto particles as they sink through the water column concurrent with negligible desorption. An alternate explanation is that this 210 Po deficit reflects a missing sink of 210 Po in deep waters by means, for instance, of 210 Po uptake by bacteria and its transfer to higher trophic levels in the oligotrophic ocean (Kim, 2001). Low 210 Po/ 210 Pb activity ratios at depth have also been associated with hydrothermal activity (Kadko et al., 1987) and the focusing of atmospherically derived 210 Pb (with low 210 Po) by isopycnal transport (Nozaki et al., 1990).
One hypothesis that has not been addressed yet is that these 210 Po deficits could also be influenced by an analytical bias. Church et al. (2012) pointed out that there is the potential for differential extraction of the Po spike used as a chemical yield tracer (usually 209 Po, although 208 Po has also been used) versus in situ 210 Po, depending on the precipitation method used. Two main methods have been used to pre-concentrate Pb and Po from seawater samples: the iron hydroxide [Fe(OH) 3 , Thomson and Turekian, 1976] and the cobalt ammonium pyrrolidine dithiocarbamate (Co-APDC, Fleer and Bacon, 1984). The Fe(OH) 3 method is the most used analytical procedure for the extraction of 210 Pb and 210 Po from seawater likely because it is less time-consuming than the Co-APDC procedure. The latter method involves adding APDC to the sample to chelate Pb and Po and subsequent flocculation of the colloidal chelate by adding excess cobalt, followed by sample filtration. To date, studies have shown that both methods are effective in coprecipitating stable Pb and 209 Po added to seawater samples as yield monitors, as shown by recoveries of more than 70% of these spikes (Matthews et al., 2007;Rigaud et al., 2013). However, to our knowledge, there has been no systematic evaluation of whether the Fe(OH) 3 and Co-APDC methods produce comparable 210 Po results.
Here, we present a compilation of 210 Pb and 210 Po studies classified according to the precipitation method used and whether 210 Po deficits at depth were observed. In addition, we directly compare the total activities of 210 Pb and 210 Po in four duplicate ocean depth-profiles determined by using both precipitation methods, Fe(OH) 3 and Co-APDC, on unfiltered seawater samples. We discuss the implications of the results of this comparison for using the 210 Po/ 210 Pb pair as a tracer of particle export in the oceans and provide recommendations for future research.

Compilation of 210 Pb and 210 Po Studies
A review of 210 Pb and 210 Po studies was conducted in order to identify those that reported depth profiles of total (dissolved + particulate) 210 Pb and 210 Po activities in water depths >300 m, classify them according to the precipitation method used, and determine whether 210 Po deficits at depth were found (Figure 1). The compilation includes a total of 213 depth profiles from 41 studies published between 1976 and 2020.
The precipitation methods were divided into the following categories: (1) Fe(OH) 3 TOT siph , where unfiltered seawater samples were precipitated with Fe(OH) 3 and the supernatant was siphoned off or decanted to allow further processing of the Fe(OH) 3 precipitate; (2) Fe(OH) 3 TOT filt precip , where unfiltered seawater samples were precipitated with Fe(OH) 3 and the precipitate was filtered; (3) Fe(OH) 3 DISS + PART , where prefiltered seawater samples were precipitated with Fe(OH) 3 and the particulate fraction was analyzed separately; (4) Co-APDC TOT , where unfiltered seawater samples were precipitated with Co-APDC; (5) Co-APDC DISS+PART , where prefiltered seawater samples were precipitated with Co-APDC and the particulate fraction was analyzed separately.
In addition, we determined whether 210 Po deficits at depth were found for those profiles that presented total 210 Pb and 210 Po activities at least at two depths from ≥300 m. Profiles were considered to show 210 Po deficits at depth if they presented total 210 Po/ 210 Pb activity ratios <0.8 at least at two depths ≥300 m.

Duplicate Profiles of 210 Pb and 210 Po
Detailed procedures for the methods compared in this study [Fe(OH) 3 TOT siph and Co-APDC TOT in Figure 1] are described below and can be found at the Center for Marine and Environmental Radioactivity website 1 . Additional details for each set of samples processed using Fe(OH) 3 and Co-APDC are provided in Supplementary Table 1 including information on the 209 Po and stable Pb tracers, detector background, blank from the stable Pb tracer, and chemical recoveries.

Sampling and Pre-conditioning
A total of 35 duplicate sample pairs were collected for the determination of total 210 Pb and 210 Po in seawater (6.5-10.4 L, Supplementary Table 1) using Niskin bottles attached to a conductivity-temperature-depth (CTD) rosette at four locations ( Figure 2 and Table 1). Three profiles were collected in the Mediterranean Sea in 2011-2013 (Ionian and Catalano-Balear Seas), while another profile was collected at the Southern Ocean Time Series (SOTS) site in 2018. This allowed to increase the statistical significance and compare results obtained from contrasting oceanographic regimes. Duplicate samples were all 1 https://cmer.whoi.edu/cookbook/ collected from the same CTD cast and the same Niskin bottles, except for one profile (Catalano-Balear South, CBS) where duplicate samples were collected from different Niskin bottles.
Unfiltered seawater samples were acidified immediately after collection to pH 1-2 using HCl (∼ 1 mL per liter of sample), and spiked with known amounts of 209 Po (T 1/2 = 125 years, 2-4 dpm, Supplementary Table 1) and stable Pb (4-42 mg, Supplementary Table 1) to monitor the losses of Po and Pb during the radiochemistry procedure. Samples were vigorously shaken after the addition of HCl and each spike. Two solutions of 209 Po were used, both in acid media (9 M HCl for the Mediterranean Sea samples; 1 M HNO 3 for SOTS) and prepared from standard solutions (Oak Ridge National Laboratory, United States; Eckert & Ziegler, Germany; respectively). The Pb solution (aqueous) was prepared from ancient Pb (>200 years) to minimize 210 Pb and 210 Po contamination. From each pair of duplicates, one sample was processed using Fe(OH) 3 (Thomson and Turekian, 1976;Sarin et al., 1992) and the other using Co-APDC (Boyle and Edmond, 1975;Fleer and Bacon, 1984) as described below. All initial processing, including the precipitation of 210 Pb and 210 Po, was accomplished at sea.

Fe-Hydroxide Method
240 mg of Fe were added to each acidified and spiked sample in the form of FeCl 3 solution. After vigorous shaking, samples were allowed to equilibrate for 9-24 h (Supplementary Table 1). Pb and Po isotopes were then precipitated with Fe(OH) 3 by adjusting the pH to 8-9 with NH 4 OH. The precipitate was allowed to settle for a few hours, and then most of the supernatant visibly free of iron hydroxides was carefully siphoned off. The precipitate was transferred into 250 mL plastic bottles and stored for 12-37 days (Supplementary Table 1) until further processing in land-based laboratories: Mediterranean Sea samples were analyzed at Universitat Autònoma de Barcelona and SOTS samples at the Edith Cowan University. There, samples were centrifuged and the excess supernatant removed. Milli-Q water was then added to the precipitates to dissolve salts and remove them by suction after a second centrifugation. Precipitates were transferred into beakers and dissolved using concentrated HCl. After evaporation to near dryness, samples were re-dissolved with ∼ 80 mL of 1 M HCl and ascorbic acid was added to reduce Fe 3+ to Fe 2+ .

Co-APDC Method
After 9-24 h of isotope equilibration, 10 mg of Co and 800 mg of APDC were added to each sample as cobalt nitrate and APDC solutions, shaking the samples vigorously after each reagent addition. Samples were allowed to equilibrate for several hours (6-12 h) and then filtered through 0.2 µm pore-size filters (Whatman membrane filters mixed cellulose ester, WHA10401731, 142 mm diameter). Samples were stored for later processing in the land-based laboratories. The filters with the Co-APDC precipitates were digested at <100 • C using concentrated HNO 3 in beakers covered with watch glasses. The solutions were then evaporated to near dryness and HNO 3 was completely eliminated from the samples by addition of 2 mL of concentrated HCl and subsequent evaporation to near dryness FIGURE 1 | Compilation of studies reporting depth profiles (≥300 m) of total 210 Pb and 210 Po activities. Legend shows the precipitation method used in each study: Fe(OH) 3 TOT siph , Fe(OH) 3 TOT filt precip , Fe(OH) 3 DISS + PART , Co-APDC TOT , Co-APDC DISS+PART (see section "Compilation of 210 Pb and 210 Po Studies" for definitions). Many of these studies represent a transect of stations but for clarity the dots represent the location of one of the stations sampled for 210 Pb and 210 Po per study. The number next to each dot corresponds to the study reference and the numbers in parentheses refer to the number of profiles where deficits of 210 Po (total 210 Po/ 210 Pb activity ratios <0.8) at depth were found, as shown by at least two data points from ≥300 m, with respect to the total profiles analyzed per study. The inset plot shows the percentage of profiles with deficits of 210 Po at depth with respect to the total profiles analyzed per method considering all studies compiled. To avoid duplication, data from the D341 cruise presented in Le Moigne et al. (2013)   for three consecutive times. The residues were re-dissolved with ∼ 80 mL of 1 M HCl.

Po Plating and Counting
Silver disks (0.1 mm thick, 25 mm diameter) were suspended in the 1 M HCl solutions using a nylon string to allow the autodeposition of Po isotopes (i.e., plating) at ∼ 80 • C and constant stirring for at least 6 h (Flynn, 1968). One side of the disks was previously coated with urethane to maximize Po plating on the non-coated side and optimize counting statistics. The time elapsed between sampling and the first Po plating was minimized as much as possible to reduce the uncertainty of 210 Po activities (Rigaud et al., 2013). 210 Po and 209 Po emissions were counted by alpha spectrometry (Fleer and Bacon, 1984) using Solutions were re-plated and also passed through an anion-exchange resin (AG 1-X8, Sarin et al., 1992) to ensure the complete elimination of Po from samples (Rigaud et al., 2013). Samples were re-spiked with 2-4 dpm of 209 Po tracer and stored for at least 6 months in 9 M HCl to allow 210 Po ingrowth from 210 Pb. After this time, samples were evaporated to near-dryness and re-dissolved with 1 M HCl to determine 210 Po ingrowth from 210 Pb by re-plating the solutions on silver disks and subsequent measurement of Po isotopes by alpha spectrometry as described above.

Chemical Recoveries of Pb and Data Treatment
Typically, two aliquots from each sample were taken before the first and last platings to determine the chemical recovery of stable Pb by inductively coupled plasma-optical emission spectrometry. However, only the second aliquot was taken from CBS samples. Considering all other samples, the recovery of Pb from the first aliquot was on average 89 ± 7% for the Fe(OH) 3 method and 78 ± 12% for Co-APDC (Mann-Whitney ranksum test, P ≤ 0.001; Supplementary Table 1). These results were not significantly different from those determined from the second aliquot [Mann-Whitney rank-sum test, P = 0.276 for Fe(OH) 3 and P = 0.401 for Co-APDC; Supplementary Figure 1], indicating that Pb losses occurred during the precipitation with Fe(OH) 3 or Co-APDC rather than during the anion-exchange procedure. Therefore, for CBS samples, we assumed that the Pb recoveries at the first plating were the same as those determined at the last plating. The 210 Pb and 210 Po blanks measured were equivalent to 0.014-0.016 dpm for the Ionian, Catalano-Balear North (CBN) and SOTS profiles. They increased to 0.270 dpm for the CBS profile due to the higher amount of stable Pb added. The contamination from the Pb solution contributed on average to <4% of the 210 Po activity at the first and last platings, except for the CBS profile, where it contributed 25-32% (Supplementary Table 1). 210 Pb and 210 Po activities at the time of sampling were carefully calculated applying blank, ingrowth, decay and recovery corrections, as detailed by Rigaud et al. (2013). Overall uncertainties in activity accounting for errors in counting, detector background, 209 Po activity, and the contamination from the Pb solution were on average 7% for 210 Pb (5-10%) and 8% for 210 Po (5-23%) for both methods.
Three-way ANOVA tests were run in order to examine whether Fe(OH) 3 and Co-APDC resulted in comparable 210 Pb and 210 Po results and whether potential methodological differences depended on the region or water depth [see section "Direct Comparison of the Fe(OH) 3 versus Co-APDC Methods"]. The factors considered were: (i) precipitation method [Fe(OH) 3 versus Co-APDC]; (ii) region (Mediterranean Sea versus SOTS); (iii) depth (within the primary production zone versus deeper waters). The base of the primary production zone (PPZ) was defined as the depth where fluorescence declined to 10% of the maximum signal measured in overlying waters (Owens et al., 2015). Statistical analyses were conducted using SigmaPlot 11.0 (Systat Software, Inc., United States) with a significance level set at 0.05.

Compilation of 210 Pb and 210 Po Studies
Two-thirds of the 210 Pb and 210 Po studies in the literature compilation (Figure 1)  In addition to classifying the studies according to the method used, we also determined whether 210 Po deficits (total 210 Po/ 210 Pb activity ratios <0.8) were found. Interestingly, this compilation shows that 210 Po deficits at depths ≥300 m are found in 65-69% of the profiles analyzed using Fe(OH) 3 , while they are found in only 33% of the profiles analyzed using Co-APDC (Figure 1). This finding is independent of whether 210 Pb and 210 Po were analyzed on unfiltered or prefiltered seawater samples.

Duplicate Profiles of 210 Pb and 210 Po 210 Pb and 210 Po Activities
The profiles of 210 Pb and 210 Po determined using [Fe(OH) 3 Fe(OH) 3 TOT siph ] and Co-APDC (Co-APDC TOT ) in duplicate samples collected from the same CTD cast are shown in Figure 3 (see data in Table 1). 210 Pb activities were not statistically different between both methods, ranging from 6.0-6.2 to 10.7-10.8 dpm 100 L −1 in the Mediterranean Sea (t-test, P = 0.122), and from 10.4-10.9 to 14.5-16.7 dpm 100 L −1 at SOTS (ttest, P = 0.572). In contrast, 210 Po activities were significantly different. In the Mediterranean Sea, 210 Po activities ranged from 1.5 to 6.9 dpm 100 L −1 for Fe(OH) 3 and from 3.1 to 9.6 dpm 100 L −1 for Co-APDC (t-test, P < 0.001; Ionian, CBN and CBS in Figure 3B and Table 1). In the Southern Ocean, at SOTS, 210 Po activities ranged from 5.0 to 13.5 dpm 100 L −1 for Fe(OH) 3 and from 7.6 to 16.2 dpm 100 L −1 for Co-APDC (t-test, P < 0.001). The higher activities of both radionuclides at SOTS reflect the higher levels of their grandparent, 226 Ra, in the Southern Ocean compared to the Mediterranean Sea (Ku and Lin, 1976;van Beek et al., 2009).
The Fe(OH) 3 results showed deficits of 210 Po at all sites and throughout the entire profiles. Below the PPZ, activities of 210 Po were lower than 210 Pb on average by a factor of 1.9 and equilibrium between both radionuclides was only observed at 1600 m at SOTS (Figure 3B). On the contrary, results from Co-APDC showed net removal of 210 Po confined in the upper 75-100 m (Figure 3A), and similar 210 Pb and 210 Po activities at deeper depths. In general, 210 Pb and 210 Po from Co-APDC reached equilibrium around the PPZ depth ( Figure 3A). This is in line with recent studies that found a close overlap between the PPZ depth and the horizon where the 234 Th/ 238 U radionuclide pair reaches equilibrium Roca-Martí et al., 2017;Lemaitre et al., 2018;Buesseler et al., 2020a,b). There is, however, one exception to this pattern. At CBS, 210 Po activities were lower than 210 Pb activities throughout the entire profile for both methods ( Figure 3B). Consequently, 210 Po/ 210 Pb activity ratios were lower than 1.0 within the PPZ and in deeper waters, averaging 0.37 ± 0.14 for Fe(OH) 3 and 0.60 ± 0.10 for Co-APDC (Figure 4 and Table 1). In this area of the Mediterranean Sea, turbidity spikes were observed throughout the entire water column below the surface maximum (Supplementary Figure 2), suggesting a possible net removal of 210 Po by particles below the PPZ in such conditions. At the other stations, turbidity spikes below the surface were not as pronounced as at CBS.
The 210 Po/ 210 Pb activity ratios (Figure 4) from both methods were statistically different (t-test, P < 0.001) with mean values from surface waters to the PPZ depth of 0.53 ± 0.10 using Fe(OH) 3 and 0.84 ± 0.16 using Co-APDC, and 0.66 ± 0.17 and 0.96 ± 0.10, respectively, below the PPZ excluding the CBS profile. Figure 5 shows a cross-plot of the 210 Pb and 210 Po activities obtained from the Fe(OH) 3 and Co-APDC methods. 210 Pb results from Co-APDC and Fe(OH) 3 were statistically similar (ANOVA test, P = 0.418 for the entire dataset, 0.263 for the Mediterranean and 0.849 for SOTS; Figure 5) with differences between mean activities of only 0.2-0.4 dpm 100 L −1 for the entire dataset. There was no significant difference between methods either when comparing samples within the PPZ or below (ANOVA test, P = 0.151 and P = 0.536, respectively). This agrees with the experiments conducted by Chung et al. (1983) in which Fe(OH) 3 and Co-APDC produced identical 210 Pb results. Chung et al. (1983) also tested the effect of equilibration times between 210 Pb and the added stable Pb carrier for times ranging from 1 to 330 days. Their results showed no discernible effect on the measured 210 Pb activities.

Direct Comparison of the Fe(OH) 3 versus Co-APDC Methods
In contrast, 210 Po activities from samples processed using Co-APDC were significantly higher than those obtained by using Fe(OH) 3 (ANOVA test, P < 0.001; Figure 5). The difference in the mean 210 Po activities observed between the two precipitation methods was 3.0 dpm 100 L −1 for the entire dataset [Co-APDC: 8.6 ± 3.6 dpm 100 L −1 ; Fe(OH) 3 : 5.6 ± 2.3 dpm 100 L −1 ], which corresponds to 35% of the 210 Po activity obtained with the Co-APDC method. The same statistical analysis was applied separately for the Mediterranean Sea profiles and SOTS (ANOVA test, P = 0.003 and P < 0.001, respectively), showing that the difference in absolute and relative terms between the two methods was larger in the Southern Ocean. At SOTS the difference was 5.1 dpm 100 L −1 , equivalent to 41% of the 210 Po activity obtained with the Co-APDC method, while in the Mediterranean Sea these values were 1.8 dpm 100 L −1 and 28%, respectively. Further, the use of Co-APDC resulted in higher 210 Po activities both within the PPZ and in deeper waters (ANOVA test, P < 0.001 for both; Figure 5), obtaining a difference between methods equal to 32% and 36% of the 210 Po activity obtained with the Co-APDC method, respectively.
Our results from four duplicate 210 Pb and 210 Po profiles support that either scavenging method can be used for reliably extracting 210 Pb from seawater, but the Fe(OH) 3 TOT siph method underestimates 210 Po activities.

Possible Reasons for the Discrepancy in 210 Po Between Methods
The compilation of 210 Pb and 210 Po studies summarized in Figure 1 suggests that the Fe(OH) 3 and Co-APDC methods may yield disparate results for 210 Po. As these studies are from different oceanographic regimes and were conducted at different times, a direct comparison of methods is difficult. However, the total 210 Pb and 210 Po activity results presented here from four duplicate profiles indicate that the Fe(OH) 3 TOT siph method underestimated 210 Po activities throughout the entire water column. 210 Po activities are calculated from the ratio between the count rate of 210 Po to 209 Po multiplied by the known activity of 209 Po added to the samples, and applying appropriate ingrowth and decay corrections (Rigaud et al., 2013). Therefore, while similarly high 209 Po recoveries were obtained for both methods [78 ± 8% for Fe(OH) 3 and 63 ± 16% for Co-APDC], 210 Po was scavenged differently. Below we discuss two possible hypotheses that could explain how the Fe(OH) 3 method may have resulted in a higher extraction of 209 Po than 210 Po from unfiltered samples and, in turn, led to the calculation of lower 210 Po activities compared with the Co-APDC method: (1) the Fe(OH) 3 protocol did not quantitatively extract all of the dissolved 210 Po from seawater due to organic complexation; (2) siphoning of the supernatant from unfiltered samples precipitated with Fe(OH) 3 resulted in a loss of particles and associated 210 Po activity.
First, different chemical speciation between the natural 210 Po present in seawater and the artificial 209 Po added to the samples (in acid media) may prevent a complete equilibration between the isotopes over 9-24 h (Supplementary Table 1) and result in a differential extraction when precipitating with Fe(OH) 3 . Little is known about the speciation of Po in seawater, but it may behave similarly to other group 16 metalloids, such as selenium (Stewart et al., 2008), which is predominantly found in organic form in surface seawater (Cutter and Cutter, 2001). Organic speciation of Po in seawater may arise through its distinctive biogeochemistry in which it is assimilated into organic matter and recycled with it. Previous studies have shown that 210 Po penetrates into the cytoplasm of bacteria and phytoplankton and associates with proteins and sulfur containing compounds in bacteria, phytoplankton and zooplankton (Fisher et al., 1983;Cherrier et al., 1995;Stewart and Fisher, 2003a,b). Therefore, at least some of the dissolved 210 Po atoms present in seawater, especially in the upper water column, would have likely been recycled and perhaps present as organic species. Further, Chuang et al. (2013) showed that Po is particularly prone to chelation by organic ligands like hydroxamate siderophores. Such organic speciation of 210 Po in seawater may reduce its adsorption onto iron hydroxides, whereas 210 Po would be effectively co-precipitated as a dithiocarbamate chelate with the Co-APDC method (Boyle and Edmond, 1975). If that was the case, regional differences in seawater chemistry and Po speciation would result in site-specific discrepancies between the Fe(OH) 3 and Co-APDC methods, in line with the results obtained in this study. In contrast, the 209 Po spike added to the acidified samples would not be speciated in the same way as natural 210 Po. After acidification and tracer addition, samples in the present study were allowed to equilibrate for 9-24 h before precipitation. This equilibration time, although typical of many studies using the Fe(OH) 3 method, may have been too short to destroy organic ligands. As a consequence, the 209 Po spike may not have completely equilibrated with natural 210 Po, leading to less scavenging of 210 Po than of 209 Po on iron hydroxides. Another possible explanation of the difference between the methods may be related to the fact that total (dissolved + particulate) 210 Pb and 210 Po activities were measured. Total seawater samples in the present study were acidified to pH 1-2 immediately after collection (see section "Duplicate Profiles of 210 Pb and 210 Po"). Marine biogenic particles typically have 210 Po/ 210 Pb activity ratios >1 (Cochran and Masqué, 2003), due both to adsorption of 210 Po onto particle surfaces and its incorporation into the particles. In contrast, 210 Pb is only adsorbed onto particle surfaces. As a consequence, 210 Po may not solubilize in an acidified sample over the 9-24 h allowed before precipitation. Samples processed with the Co-APDC method were filtered through 0.2 µm filters after precipitation and were subsequently digested with concentrated HNO 3 , which would have effectively dissolved any particles in the sample along with the Co-APDC precipitate. For the samples precipitated with the Fe(OH) 3 method, most of the supernatant water was siphoned off at sea and the precipitate was returned to the laboratory with residual supernatant. Despite Fe(OH) 3 TOT siph being a common procedure for the determination of total 210 Pb and 210 Po (see Figure 1), we suggest that this method may result in a loss of particles and associated 210 Po activity from the samples. If that was the case, the methodological offset between the total Fe(OH) 3 and Co-APDC protocols would depend on the particulate 210 Pb and 210 Po activities in seawater and vary as a function, for example, of place, time of year and phytoplankton biomass. Indeed, previous studies in the Mediterranean Sea and the Southern Ocean have reported that a significant fraction of the total 210 Po activity is associated with the particulate phase. For example, in the NW Mediterranean Sea, particulate 210 Po (>0.2 µm) activities in surface waters were reported to amount to 21% of the total activities (Masqué et al., 2002b). This agrees with an average of 19% of particulate 210 Po (>0.7 µm) measured at the DYFAMED site (NW Mediterranean Sea) from surface waters to >2400 m (unpublished results, P. Masqué). Similarly, in the Antarctic Circumpolar Current, the relative importance of particulate 210 Po (>1 µm) to total activities was on average 14% in the upper 600 m of the water column, including the sampling of phytoplankton blooms (Friedrich and Rutgers van der Loeff, 2002). These levels of particulate 210 Po comprise a significant fraction of the difference observed here between the Fe(OH) 3 and Co-APDC methods [28% for the Mediterranean Sea and 41% for SOTS, see section "Direct Comparison of the Fe(OH) 3 versus Co-APDC Methods"]. It is important to emphasize that this potential bias would probably be unnoticeable for 210 Pb because particulate 210 Pb activities only amounted to 3-8% of the total activity in these studies (Friedrich and Rutgers van der Loeff, 2002;Masqué et al., 2002b), which are within the uncertainties associated with the 210 Pb measurements. However, the results from Friedrich and Rutgers van der Loeff (2002) and the DYFAMED site obtained by using the Fe(OH) 3 DISS + PART method showed a significant disequilibrium between total 210 Pb and 210 Po at depths ≥300 m (average total 210 Po/ 210 Pb activity ratio = 0.63 ± 0.05 and 0.79 ± 0.13, respectively). This suggests that a possible loss of particles when siphoning the supernatant from unfiltered samples precipitated with Fe(OH) 3 may not fully explain the differences observed between the two precipitation methods.
The compilation of 210 Pb and 210 Po studies shows that deficits of 210 Po at depth are more often observed when using Fe(OH) 3 versus Co-APDC, regardless of whether 210 Pb and 210 Po were analyzed on unfiltered or prefiltered seawater samples (Figure 1). This observation suggests that the organic complexation hypothesis may be the major explanation for the difference between methods. We acknowledge, however, that further experiments are needed in order to test other Fe(OH) 3 protocols and elucidate the underlying reasons behind the mismatch observed in this study. In particular, samples should be processed identically with respect to the treatment of the precipitate [filtered for both the Fe(OH) 3 and Co-APDC methods]. Thus, at this point, our findings cannot be extrapolated to other Fe(OH) 3 protocols, such as that used in the GEOTRACES program for the determination of 210 Pb and 210 Po in filtered seawater samples (Cutter et al., 2017).

Possible Overestimation of 210 Po-Derived Fluxes Using Fe(OH) 3
The comparison of methods presented in this study reveals that while Fe(OH) 3 and Co-APDC yield comparable results for 210 Pb, up to 40% lower 210 Po activities can be measured when using the Fe(OH) 3 method on unfiltered seawater samples. Moreover, unlike the iron hydroxide method, samples processed using Co-APDC showed radioactive equilibrium between 210 Pb and 210 Po at depth (see section " 210 Pb and 210 Po Activities"), consistent with the long scavenging residence times in the deep ocean compared with the mean life of 210 Po.
The lower 210 Po/ 210 Pb activity ratios measured by using the Fe(OH) 3 method are apparent not only in deep samples, where large 210 Po deficits have been reported in multiple studies [most of them using Fe(OH) 3 , Figure 1], but are also evident in samples from the euphotic zone and the upper twilight zone (Figure 4). This observation has important implications with respect to calculations of the export flux of POC (or other elements of interest) associated with sinking particles in the upper ocean. To evaluate these implications, we present the 210 Po-derived fluxes from a 1-D steady-state model (Equation 1), integrating the 210 Po FIGURE 6 | 210 Po fluxes calculated from integrating the 210 Po deficit down to the base of the primary production zone using the Fe(OH) 3 (yellow) and Co-APDC (green) precipitation methods. Labels indicate the ratio between the Fe(OH) 3 -derived to the Co-APDC-derived fluxes.
deficits observed down to the PPZ depth for the duplicate samples processed by using Fe(OH) 3 and Co-APDC (Figure 6): where ( 210 Pb -210 Po) is the integrated 210 Po deficit with respect to 210 Pb down to the depth of the PPZ (dpm m −2 ) and λ 210Po is the decay constant of 210 Po (0.0050 d −1 ). In addition to the PPZ depth, a relative light depth of 0.1% photosynthetically available radiation (PAR) could also be chosen as a reference depth to compare particle flux estimates from different sites (Buesseler et al., 2020b), but we only used the PPZ depth because we lack PAR data for some of the profiles. The 210 Po fluxes obtained from our four duplicate profiles are shown in Figure 6. At CBS, the fluxes derived from Fe(OH) 3 and Co-APDC are similar within uncertainties (34 ± 4 and 26 ± 4 dpm m −2 d −1 , respectively). In contrast, for the Ionian Sea, CBN and SOTS profiles, export estimates from the Fe(OH) 3 method are a factor of 2 to 8 higher compared with Co-APDC estimates. This comparison clearly reveals how different conclusions can be drawn solely depending on the method used, where the Fe(OH) 3 method as applied here can lead to overestimated 210 Po fluxes. This suggests that 210 Po fluxes estimated from a commonly used Fe(OH) 3 protocol may be compromised to different degrees depending on the study area. The resulting exaggerated 210 Po flux would cause a proportional overestimation of the POC fluxes when multiplying the 210 Po flux by the POC/ 210 Po ratio associated with sinking particles.

CONCLUSION AND RECOMMENDATIONS
This study highlights that two commonly used methods for extracting 210 Pb and 210 Po from seawater can produce different activities for 210 Po. On unfiltered seawater samples, precipitating 210 Pb and 210 Po with Fe(OH) 3 and siphoning off the supernatant shows total 210 Po activities up to 40% lower than those obtained with the Co-APDC method in which the precipitate is filtered. Deficits of 210 Po can be used to quantify POC fluxes and, therefore, the Fe(OH) 3 method may lead to artificially high 210 Po-derived POC fluxes. This finding has also important implications for understanding the behavior of Po in marine systems and defining possible new applications of this element to study biogeochemical cycles (e.g., sulfur).
Possible explanations for the lower 210 Po activities observed with the Fe(OH) 3 method include complexation of dissolved 210 Po in seawater preventing complete equilibration with the 209 Po tracer added to the samples, or the loss of particles when siphoning the supernatant from the samples. The compilation of 210 Pb and 210 Po studies presented here suggests that the former may be the major explanation for the difference between methods. Future research is needed to investigate whether longer sample storage after acidification and spiking allows more complete equilibration between natural 210 Po and the 209 Po tracer and gives results more comparable to those from the Co-APDC method. For these tests, we recommend using filtered seawater samples so that only the dissolved fraction is involved, and unfiltered seawater samples with filtration of the Fe(OH) 3 and Co-APDC precipitates. Processing a series of identical samples with increasing times for isotope equilibration before Fe(OH) 3 co-precipitation would allow determination of how quickly equilibration between 210 Po and 209 Po is reached. Laboratory experiments using natural seawater and seawater treated with ultra-violet irradiation (as done for trace metals), together with dissolved organic matter measurements on the samples, may also be useful to test the hypothesis that organic complexation of Po leads to differential extraction of the 209 Po spike and the in situ 210 Po when precipitating with Fe(OH) 3 .

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.