210Pb and 137Cs were measured in 25 sediment cores collected during the International Polar Year from a transect spanning the North American Arctic margin, from the North Bering Sea to Baffin Bay/Davis Strait. Profiles and inventories of the radioisotopes were used to determine sediment mixing and accumulation at each site and to assess the intensity of scavenging and burial. Sediment accumulation rates derived from 210Pb and validated using 137Cs are between ≤0.04 and 0.23 g cm−2 yr−1. 210Pb cannot be used to derive sedimentation rates for vigorously biomixed sediments from the North Bering-Chukchi shelf. Elevated 137Cs activities and inventories in recently deposited sediments imply delayed inputs of particle-associated 137Cs to the sediments, likely transported from the watershed to the coast and subsequently redistributed to shelf/slope sediments. Inventories of 210Pbex in all cores meet or exceed the estimated supply of 210Pbex from atmospheric deposition and decay of 226Ra in the water column. This implies that in contrast to the deep Arctic Ocean basin, there is a sufficient supply of suspended particulates along the North American Arctic margin to scavenge the supply of 210Pbex. 210Pbex inventories in sediments are up to 21-fold greater than the in situ supply at some sites. Large inventories of 210Pbex in sediments along the North Bering-Chukchi shelf result primarily from focusing, while those along the north Chukchi slope (Barrow Canyon) and in Baffin Bay/Davis Strait reflect strong boundary scavenging, likely supported by lateral exchanges with deep/interior Atlantic-origin waters.
 The removal of particle-reactive substances from the water column by sinking particles and the subsequent burial in sediments are key processes in ocean biogeochemical cycles [Rutgers van der Loeff and Geibert, 2008]. In world oceans, the continental margins are particularly important for these processes because of generally large particle fluxes supported by proximal terrestrial inputs and high primary production [cf. Chen et al., 2003; Liu et al., 2000]. Enhanced scavenging by particles at ocean margins (boundary scavenging), together with enhanced seawater supply mediated by eddies, advection and upwelling may, furthermore, alter the distribution of particle-reactive substances in the global ocean [cf. Carpenter et al., 1981; Cochran et al., 1990; 1995].
 Radioisotopes, many of which have well-known supply and removal mechanisms, are frequently used as tracers for studying scavenging and burial in marine systems [cf. Baskaran et al., 1996; Cochran et al., 1990; Huh et al., 1997; Lepore et al., 2009; Smith et al., 2003]. 210Pb, one of the most widely used natural radioisotopes, is supplied to the ocean by atmospheric deposition and by decay of 226Ra, which is almost conservative in the ocean [Cochran, 1992; Rutgers van der Loeff and Geibert, 2008]. Dissolved 210Pb in the water column is removed by particle scavenging and radioactive decay (half-life = 22.3 years). In contrast, 137Cs is an artificial radionuclide (half-life = 30.2 years), first released into the environment by atmospheric nuclear weapon tests conducted in the 1950s–1960s. This radionuclide is often used together with 210Pb to validate geochronologies in sediments [Smith, 2001] and as a water-mass tracer because of its largely conservative behavior once it becomes dissolved in seawater [cf. Cochran et al., 1995; Smith et al., 1998, 1999].
 Boundary scavenging is likely to be a particularly important process in the Arctic Ocean partly because the shelves and slopes occupy more than half of the total area [Jakobsson et al., 2004], and partly because the high productivity, particle supply, and sedimentation at the margin provide an exceptional contrast with the sediment-starved, oligotrophic interior ocean [Gosselin et al., 1997; Honjo et al., 2010]. The importance of the Arctic margin as a sink for particle-reactive elements is supported by deficits of dissolved 210Pb relative to 226Ra in the upper halocline in the Canada Basin, which are believed to result from scavenging of 210Pb as waters that form the halocline transit the Chukchi shelf and slope [Moore and Smith, 1986; Smith and Ellis, 1995; Smith et al., 2003]. However, the scavenging history of a given water mass is difficult to infer [Rutgers van der Loeff and Berger, 1993], and thus, 210Pb deficits in other interior water masses (e.g., lower halocline and intermediate waters) suggest other regions or processes are also involved in scavenging [Moore and Smith, 1986; Smith and Ellis, 1995; Smith et al., 2003]. The quantitative importance of boundary scavenging to, for example, the overall 210Pb budget in the Arctic Ocean, is not known. Most of the shelf and slope measurements for the Arctic Ocean have been made in the Chukchi Sea [Huh et al., 1997; Lepore et al., 2009; Smith et al., 2003], which may poorly represent scavenging at other Arctic margins because of the exceptional productivity and consequent vertical particle flux in the Chukchi [cf. Grebmeier et al., 2006]. Arctic margins are diverse, with pronounced differences in seawater sources, water residence times, sea ice cover, degree of terrestrial influence, productivity, and strength of exchanges with the interior ocean [Carmack and Wassman, 2006]. More than 20 ocean domains have been identified along the North American Arctic margin alone, based on oceanographic structure and property distributions [Carmack et al., 2010]. A broader view of the variations in and controls on boundary scavenging along the Arctic margin is needed to improve our understanding of the cycling of particle reactive substances in the Arctic Ocean and the origins of water-mass properties observed offshore. Furthermore, the Arctic Ocean is entering a time of change in its sea ice distribution [Johannessen et al., 2004], river runoff [Arnell, 2005; Lammers et al., 2001], and ocean circulation [Walsh et al., 2011]. Boundary scavenging is, therefore, an important process to understand given that each of these changes will affect distributions of dissolved elements, the supply and transport of particles, and the processes leading to scavenging at the ocean margin.
 In this paper, we report new measurements of 210Pb and 137Cs in 25 sediment cores collected broadly along the North American Arctic margin, from the North Bering Sea eastward to Baffin Bay/Davis Strait (Figure 1). From the sediment profiles and inventories of 210Pb and 137Cs, we determine sediment mixing and accumulation at each site and assess the intensity of scavenging and burial of particle-reactive elements along the North American Arctic margin. Finally, incorporating other published sediment data, we propose how 210Pb and 137Cs can be used together to discriminate between the relative roles played by scavenging and focusing at any given site.
 The 25 sediment cores analyzed in this study were collected during oceanographic missions aboard the Canadian Coast Guard Ships Wilfrid Laurier in July 2007 and Louis S. St. Laurent in July 2008. The core sites span water depths of 50 to 2125 m and were distributed widely along the North American Arctic and sub-Arctic coastline from the North Bering Sea shelf in the west to Baffin Bay/Davis Strait in the east (Figure 1). The North Bering Sea shelf (stations SLIP1, 3, and 4), Chukchi Sea shelf (stations UTN3, 5, and 7), Barrow Canyon area in the northeast Chukchi Sea (stations BC3, 4, 5, and 6), and Beaufort Sea slope (stations CG1, 2, and 3) were sampled in 2007 and the Canadian Arctic Archipelago (CAA) (stations QM1, VS1, FS1, BE2, PS2, PS1, CAA2, CAA1, and BB11) and Baffin Bay/Davis Strait (DS1, 2, and 5) were sampled in 2008.
 Sediment cores were collected using a box corer, which penetrates the seafloor to a maximum of 50 cm and has a collection area of 600 cm2. The cores were sub-sectioned horizontally aboard the ship, generally into 0.5 cm intervals for the top 2 cm, 1 cm intervals for the 2–10 cm depth interval, and 2, 3, or 5 cm intervals for the remainder of the core. Sediment from the outermost 3–5 cm of the box was discarded. Sediment samples were placed in plastic bags or bottles and kept frozen for later chemical analysis.
 At INRS (Institut National de la Recherche Scientifique) in Quebec City, subsamples of sediment were subsampled for moisture content then freeze dried, ground, and placed in sealed vials for at least 1 month to ensure secular equilibrium between 226Ra and 222Rn, and 214Pb and 214Bi. 226Ra activity was obtained from the average of 214Pb and 214Bi. The activities of 137Cs, 210Pb, 214Pb, and 214Bi were measured by gamma spectrometry [Appleby et al., 1986; Schelske et al., 1994] using an ultra-low background Ge well detector and corrected for self-absorption and sample geometry [Appleby and Piliposian, 2004]. Corrections for self-absorption were made by determining counting efficiencies for dry sediment samples (bulk densities from 0.8 to 1.3 g cm−3) labeled with standard solutions of mixed nuclides (AEA Technology; QCY B40), and for sample geometry, we used sample mass. Samples were counted for 85,000 s, which produces counting errors of 5–15% in the upper layers of each core.
 Excess 210Pb (210Pbex) activities were calculated as the difference between the total 210Pb activity and the 226Ra activity in each respective sediment section. 210Pbex activity was corrected to the date of core collection, and the results reported on a dry-weight basis, corrected for salt content assuming a pore water salinity of 34.75. Activities in units of dpm g−1 were converted to units of dpm cm−3 by assuming a density of sediment solids of 2.65 g cm−3. Inventories of 210Pbex and 137Cs in each core were calculated by summing the activities Ci (dpm g−1) down to a depth where the radioisotope was no longer detectable, according to the equation: inventory = ΣCimi, where mi is the mass-depth increment (g cm−2) corresponding to the depth interval.
3 Results and Discussion
3.1 Profiles of Porosity, 226Ra, 210Pb, and 137Cs
 Profiles of sediment porosity and observations at the time of sectioning show generally uniform sediments undergoing compaction with time (not shown). Short cores collected at UTN3 and BC3 were coarse textured with porosity profiles suggesting increased coarseness with depth. This may also apply to the short cores, CAA1 and CAA2, and DS2. Relatively abrupt shifts in sediment texture also occur in core BC6. Sediment 226Ra activity was relatively constant within each core (coefficient of variation 3–23%; Figure 2), consistent with the generally uniform sediment texture. The average 226Ra activity varied from 0.67 ± 0.06 dpm g−1 in core DS5, which was one of the most coarse-grained cores, to between 1.16 dpm g−1 and 3.68 dpm g−1 in the remaining cores.
 In most of the cores, total 210Pb activities in surface sediment sections were 3.5-fold to 7-fold higher than the corresponding 226Ra activities. Total 210Pb activities then decreased down core, reaching levels equivalent to those of 226Ra in deep sediment sections (Figure 2). Cores from both the interior CAA (FS1, VS1, PS2) and the North Bering and Chukchi shelves (SLIP1, 3, 4, UTN3, 5, 7, and BC3) had low 210Pb activities (<3 times 226Ra) throughout. Total 210Pb activities remained above the corresponding 226Ra activities at the bottoms of cores UTN3, BC3, and PS2.
210Pbex activities calculated as the difference between the total 210Pb activity and the 226Ra activity in each respective sediment section varied from 1.6 to 16.8 dpm g−1 among the surface sediment sections of the cores (Figure 3). The vertical profiles of 210Pbex in the cores followed two general patterns: (1) low, nearly constant activity extending down to 10–30 cm or more (SLIP1, 3, 4, UTN3, 5, 7, and BC3) or (2) a two-layer profile with nearly constant 210Pbex activity in the surface layer (1–5 cm thick) and exponentially decreasing activity with depth in the lower layer (all other cores). The 210Pbex profile in core BC6 is anomalous; it exhibits two-layer behavior but the 210Pbex activities are exceptionally low in the surface layer.
aTable gives whole core average values of porosity (φav), thickness of the surface mixed layer (SML), sedimentation velocity (ω), specific activity in depositing material (C0), and order-of-magnitude mixing rate (Kb1) derived from a two-layer advective-diffusive model. The model assumes rapid mixing (Kb1) in SML and low rate of mixing (0.01 cm2/yr) in a deeper layer. ω is estimated from slope of ln (210Pbex) versus depth and then sedimentation rate (r) derived from ω using r = 2.65 × (1 − φav) × ω. Kb1 and C0 are assigned to give best fit to 210Pbex profile. Rates of sedimentation and mixing were then verified using 137Cs profiles (see text for details). For cores with profiles implying deep active mixing (SLIP1,3,4, UTN3,5,7, and BC3), we estimated a mixing rate (Db) assuming negligible sediment accumulation according to the equations given in Anderson et al. .
North Bering Sea shelf
Low total 210Pb and nearly constant with depth
Low total 210Pb and nearly constant with depth
Low total 210Pb and nearly constant with depth
Chukchi Sea shelf
Low total 210Pb and nearly constant with depth
Low total 210Pb and nearly constant with depth
Low total 210Pb and nearly constant with depth
Low total 210Pb, possibly non-steady state (see porosity profile, Fig. 2)
Low 210Pbex in top 1.5 cm
Some variations in porosity
Top 5–6 cm disturbed; rates derived from data over 7–20 cm
Beaufort Sea slope
Low 210Pbex in top 1.5 cm
ω gives good fit to 137Cs onset
ω gives good fit to 137Cs onset
ω gives good fit to 137Cs onset
Very low total 210Pb (<3 times supplied); ω should be viewed as maximum rate
Very low total 210Pb; ω should be viewed as maximum rate
ω gives good fit to 137Cs onset
Very low total 210Pb, deep extent; ω should be considered uncertain
Short core; 137Cs profile muted; ω should be considered uncertain
Low 210Pbex in top 0.5 cm, ω poorly constrained by 210Pbex (3 points)
Baffin Bay/Davis Strait
Possibly nonsteady state
137Cs profile muted; ω should be viewed as maximum rate
 Likewise, the profiles of 137Cs in the sediment cores (Figure 4) followed two general patterns. In the cores from the North Bering and Chukchi shelf (SLIP1, 3, 4, UTN3, 5, 7, and BC3), 137Cs exhibited low and nearly constant activities (~0.2–0.3 dpm g−1) extending down to 10–30 cm or more, generally resembling the 210Pbex profiles. In all other cores, the 137Cs profiles generally exhibited high values at or near the surface, below which activities decreased rapidly with depth, with activities at or below detection (<0.1 dpm g−1) in deep sediment sections. Again, core BC6 appeared to have anomalously low 137Cs activities in the top 5–6 cm of the core (0.29 dpm g−1 on average; Figure 4).
 The vertically uniform profiles of both 210Pbex and 137Cs in the cores from the North Bering (SLIP1 to 4) and Chukchi shelves (UTN3 to 7) imply active mixing of sediment to depths of between 10 and 30 cm in these regions, presumably as a result of biomixing by benthic foragers. Uniform distributions of 210Pbex to depths of 10–20 cm in sediments from the North Bering-Chukchi margin have previously been attributed to high-density benthic infauna mixing the surface sediments [Baskaran and Naidu, 1995; Clough et al., 1997; Grebmeier, 1993; Lepore et al., 2009]. The generally deep mixing evident in North Bering-Chukchi shelf sediments is consistent with the greater biological productivity of the water column in this region, which supports high fluxes of labile carbon and high benthic biomass compared to other areas along the North American Arctic margin [Grebmeier, 1993; Grebmeier et al., 2006].
3.2 Derivation of Sediment Accumulation and Mixing Rates From Two-Layer 210Pbex Profiles
 The sediment cores exhibiting two-layer 210Pbex profiles may be viewed as a surface mixed layer (SML) overlying accumulating sediments that are subject to little or no mixing [cf. Roberts et al., 1997; Smith et al., 1995]. Here we have used a two-layer advective-diffusive equation to fit the data [Lavelle et al., 1985; Robbins, 1978]:
where ω is sedimentation velocity (cm yr−1, constant), C is 210Pbex activity (dpm cm−3), z is depth (cm, positive downward), Kb is the biomixing coefficient (cm2 yr−1), and λ is the decay constant (0.03114 yr−1). The thicknesses of the SMLs were estimated by eye from the 210Pbex profiles (Figure 3). The mixing coefficient (Kb1) was assumed to be constant in the SML of each core and negligible (Kb2 = 0.01 cm2 yr−1) in deeper layers. Values of ω and C0 (surface activity, dpm cm−3 wet sediment) were determined for each core by varying these coefficients together with Kb to achieve the best least squares fit between the analytical solutions and the data. A value for the mass accumulation rate, r, was then calculated as
where ρ is the density of the solids and φav the average porosity below the SML.
 Sedimentation velocities derived in this manner for cores with two-layer 210Pbex profiles are provided in Table 1. The rates estimated for cores FS1, VS1, and especially PS2 should be viewed as uncertain because of the low total 210Pb activities in these cores, which increase the uncertainty in the 210Pbex values and hence derived sedimentation velocities. In core PS2, total 210Pb also failed to decline to a value in secular equilibrium with 226Ra in the deep sediment sections (Figure 2), which implies that subsurface layers are affected by sediment mixing. The derived accumulation rates are also uncertain for CAA2 due to the shortness of the core and for DS2 because of the variation in porosity implying nonuniform sedimentation. For DS2, downward mixing of a small amount of fine-grained material into glacial substrate could affect both porosity and 210Pbex profiles [see, e.g., Muzuka and Hillaire-Marcel, 1999].
 The unusually low 210Pbex (and 137Cs) activities in the top 5–6 cm of core BC6 (Figure 3), together with the reversal in the sediment porosity gradient in this layer (not shown), imply a rapid deposition event (turbidite) at the sediment surface. This interpretation is supported by the Mn profile [Macdonald and Gobeil, 2012], which shows a maximum at 5–10 cm below the surface implying that a recent rapid influx of sediment has buried the original surface. Below this, the relatively smooth porosity and 210Pbex profiles appear not to have been significantly disturbed. Therefore, a sediment accumulation rate (~0.11 cm/yr) was derived for BC6 from data between 7 and 20 cm. This rate is similar to the 0.12 cm/yr determined for an apparently undisturbed core (also called BC6) collected by Lepore et al.  nearby in 2002 (derived from the data in their Table A2). The 210Pbex activity at the surface of this undisturbed core (31.6 dpm g−1) was similar to that at the buried surface (7 cm) in core BC6.
3.3 Validation of Sediment Accumulation and Mixing Rates Using Penetration Depth of 137Cs
 The two principal sources of 137Cs to the Arctic Ocean have been fallout from atmospheric weapons tests and effluent released from the European nuclear reprocessing plants dominated by Sellafield [Aarkrog, 2003]. The first significant atmospheric fallout of 137Cs occurred in the early 1950s; fallout then peaked in 1963–1964 and subsequently decreased to near-zero values in the early 1980s [Monetti, 1996]. The continued near-zero deposition in the Arctic was interrupted by a short pulse of 137Cs in 1986 consequent to the Chernobyl accident [e.g., see Arctic Monitoring and Assessment Programme (AMAP), 2004; 2010]. However, based on ice core records, fallout from Chernobyl over the North American Arctic margin appears to have been negligible (<2% of fallout from atmospheric weapons tests) [Pinglot et al., 2003]. Sellafield discharges to the Irish Sea began and reached maximum values in the 1970s [Aarkrog, 2003] and have since significantly declined [Povinec et al., 2003].
 The extent to which Arctic margin sediments (or other ocean sediments) [cf. Livingston and Povinec, 2000] capture 137Cs from fallout or Sellafield discharges is an open question because 137Cs has a very low particle reactivity in marine waters (partition coefficients between seawater and sediments for 137Cs are about 100–500) [International Atomic Energy Agency (IAEA), 1985]. Indeed, if radioactive decay is accounted for, 137Cs circulates predominantly as a conservative element in ocean surface waters with, for example, an effective half-life in Atlantic Ocean surface waters of about 25 years [Livingston and Povinec, 2000] meaning that the circulating inventory of 137Cs in the upper ocean has but poor connectivity to sediments. If anything, marine sediments likely exhibit a net loss of the 137Cs initially deposited with particles [cf. Su and Huh, 2002]. Accordingly, only a portion of the direct 137Cs fallout has transferred to Arctic margin sediments, either sequestered in the original aerosol or sorbed to marine clay particles from dissolved 137Cs [Baskaran and Naidu, 1995]. Given that Sellafield and other reprocessing plant discharges have been transported to the Arctic in the dissolved phase, the associated 137Cs has tended to remain in the water providing a strong, quasi-conservative tracer [cf. Smith et al., 2011] with little or no transport to sediments via vertical particle flux.
 Regardless of the extent to which Arctic margin sediments have captured the 137Cs from these sources, the depth of penetration of 137Cs in sediments may still be used as a time-stratigraphic marker of the early 1950s, reflecting fallout of the first significant releases of 137Cs into the environment due to atmospheric weapons testing [Smith, 2001]. Here we make this comparison by taking the record of atmospheric fallout [Monetti, 1996] as an input function and simulating the 137Cs profiles in the sediment cores using the 210Pb-derived sediment mixing and accumulation rates and a numerical advective-diffusive model [Johannessen et al., 2005]. The model runs in Matlab and incorporates mixing in the SML, sedimentation and radioactive decay of 137Cs (λ = 0.0231 yr−1).
 The 137Cs profiles simulated by the model using the 210Pb-derived sediment mixing and accumulation rates generally compare well with the lower (initial) portions of the observed 137Cs profiles, where 137Cs activities show pronounced increases above the low levels present in deep sediment sections (compare points and solid lines in Figure 4). This good agreement between the simulated and observed depth of penetration of 137Cs provides support for the 210Pb-derived sedimentation and mixing rates (Table 1). Furthermore, 137Cs profiles simulated using higher mixing rates and correspondingly lower sedimentation rates do not agree with the observed 137Cs penetration depths nearly as well (e.g., compare solid and dashed lines on plot of CG1 in Figure 4). We conclude that in general, 210Pb and 137Cs can be used together to separate accumulation from mixing in Arctic margin sediment cores. However, separation of mixing and accumulation becomes increasingly difficult at lower sedimentation rates, and in these cases, the sedimentation rates should be viewed as maximum values (Table 1). The low 137Cs activities (within a factor of two to three of the detection limit) present in deep layers of some of the cores imply a low level of post-depositional vertical migration of 137Cs, as previously observed in marine sediments [cf. Oughton et al., 1997].
3.4 Evidence From 137Cs Profiles and Inventories for Inputs From Rivers, Resuspension, and Sea Ice Rafting
 Despite the agreement between observed 137Cs penetration in the cores and penetration predicted using the atmospheric fallout input function (Figure 5) and the sedimentation and mixing rates in the cores, the 137Cs activities in more recently deposited sediments are generally under-predicted by the model (Figure 4). Post-depositional remobilization of 137Cs can be ruled out as an explanation for the elevated 137Cs activities in recently deposited sediments because (1) the 137Cs activities are relatively low in the lower (pre-fallout) portions of the core despite concentration gradients that would tend to enhance downward diffusion; (2) the elevated 137Cs activities do not occur with equal strength in every core, as might be expected for a post-depositional remobilization effect; and (3) although only a small fraction of 137Cs is typically mobile in marine sediments [Oughton et al., 1997], in several cores, the highest 137Cs activities are contained in the recently deposited sediments (Figure 4). Furthermore, many of the cores with the most pronounced discrepancies between the observed and simulated 137Cs activities in recently deposited sediments (BC4 and CG1-3) contain very large 137Cs inventories, which are not explainable by 137Cs remobilization within the sediments. Following Baskaran and Naidu , we have compared the inventories in the cores (0.75–5.7 dpm cm−2; Figure 6a) to the maximum inventory accountable by direct atmospheric fallout (~1.8 dpm cm−2 in 2008). More than half the cores have 137Cs inventories greater than this value (compare bars to horizontal dashed line in Figure 6a). The 137Cs inventories in the cores (0.75–5.7 dpm cm−2) span a similar range to previously reported values for the east Chukchi Sea shelf (0.64–4.09 dpm cm−2) [Baskaran and Naidu, 1995] and the North Bering Sea shelf (0.42-4.0 dpm cm−2) [Oguri et al., 2012]. The 137Cs profiles also are similar (when converted to the same units) to those reported previously for the Bering, Chukchi, and Beaufort Sea shelves and slopes [see, e.g., Pirtle-Levy et al., 2009, Figures 6 and 7].
 The likely explanation for enhanced 137Cs activities in recently deposited sediments is significant recent particulate 137Cs deposition. The “additional” 137Cs inputs necessary to account for the profiles in the cores, estimated by developing the simplest input functions (i.e., fewest modifications to the atmospheric input function) that would provide “good fits” to the observed 137Cs profiles in the cores (compare points and dotted lines in Figure 4), reveal a universal, large input of 137Cs during the 1970s–2000s (as illustrated in Figure 5).
 The recent inputs of 137Cs to the sediments most likely reflect the input of 137Cs associated with particulates from rivers and coastal erosion to the coastal zone and subsequent redistribution of sediments through resuspension, lateral transport, and sea ice rafting [e.g., O'Brien et al., 2006]. In contrast to the weakly particle-reactive behavior of 137Cs in marine waters, 137Cs binds almost irreversibly to clay particles in terrestrial and freshwater environments [Smith and Ellis, 1982; Smith et al., 1987]. The spatial distribution of 137Cs, with largest inventories generally occurring in the western half of the study area (Chukchi and Beaufort Seas; Figure 6a), reflects the location of the major riverine particulate input to the Arctic Ocean (associated with the Yukon and Mackenzie Rivers). Relatively high 137Cs inventories in sediments in the Bering Sea shelf, compared to the slope, have been attributed to 137Cs associated with fine-grained particles delivered by the Yukon River [Oguri et al. 2012]. A minor influence of Yukon River sediments may also extend into the southern Chukchi Sea [cf. McManus et al., 1969]. The influence of the Mackenzie River on 137Cs distribution in the Beaufort Sea has not been studied but is probably significant given that this river dominates the supply of particles and other particle-associated elements in the Beaufort Sea [Leitch et al., 2007; Macdonald and Thomas, 1991; Macdonald et al., 1998; Yunker et al., 1993]. A rough estimate of the supply of 137Cs by the Mackenzie River based on sediment discharge (~124 × 106 t yr−1) [Holmes et al., 2002] and the average 137Cs activity in coastal soils (~3.7 dpm g−1; D. Leitch, 2011, unpublished data) yields an amount roughly equivalent to 40% of the 1.8 dpm cm−2 total fallout, assuming this riverine particulate 137Cs to be distributed evenly over the sediments of the Mackenzie shelf (65,000 km2).
 The apparent timeline of 137Cs deposition in margin sediments (Figures 4 and 5) is also consistent with the delay expected for 137Cs initially deposited on land. Transport of fallout 137Cs through watersheds to coastal sediments depends on residence times specific to each watershed but is typically associated with a delay of the peak levels by a year or two, followed by sustained reduced inputs in the 1970s–1980s [Smith and Ellis, 1982; Smith et al., 1987]. Transport of fallout 137Cs through the Mackenzie River watershed is complex. The eastern part of the basin contains large headwater lakes with long water residence times [Gibson et al., 2006] that likely trap much of the particulate 137Cs and delay the fallout signal into the 1960s–1970s and later, whereas the western basin drains steep terrain that encourages erosion and permits rapid transport to the ocean [Carrie et al., 2012]. In Figure 5, we present a “Mackenzie River input function” using the Smith and Ellis  watershed transport model as a way to evaluate the potential role played by particulates from this river, assuming residence times for 137Cs in the water column and drainage basin of 5 and 1000 yrs, respectively (see Smith and Ellis  and Smith et al.  for details). The process of sequential deposition and resuspension as sediment transports across the shelf [e.g., O'Brien et al., 2006] would further delay the fallout signal at our slope stations, typically by a year or two [Robbins et al., 2000]. The delay associated with 137Cs transport by ice rafting is more difficult to estimate and varies, depending upon whether the rafting occurs locally, regionally, or Arctic wide. Nevertheless, sea ice rafting, as a principal transport mechanism for particle-associated 137Cs within the Arctic Ocean [Cámara-Mor et al., 2010], could account for a redistribution of 137Cs between marginal regions of the Chukchi and Beaufort Seas [Eicken et al., 2005], leading to relatively weak spatial gradients in sedimentary 137Cs inventories (e.g., the similar values in the BC and CG cores, despite the closer proximity of the CG sites to the Mackenzie River). Redistribution of 137Cs by sea ice on a larger scale within the Arctic Ocean, i.e., from the Russian sector (where coastal sediments near rivers contain higher 137Cs activities) [Baskaran et al., 1996] to the North American sector, would also help to account for the large 137Cs inventories and relatively large recent inputs to the North American margin sediments.
3.5 210Pbex Inventories in the Sediments and Comparison to Total Supply by Atmospheric Deposition and 226Ra Decay
 In Figure 6b, the inventories of 210Pbex calculated from the profiles in the cores (7.6–61.9 dpm cm−2) are compared to the total supply of 210Pbex to the water above each core by atmospheric deposition and 226Ra decay estimated from the literature (compare solid bars and hatched bars). Atmospheric deposition is probably uniform within the study area within a factor of about 3, with the variation reflecting proximity to land, regional rates of radon emanation from soil, dominance of terrestrial versus marine air masses, rates of precipitation, and extent and persistence of sea ice cover, which inhibits deposition into the water [Hermanson, 1990; Omelchenko et al., 2005]. At the low end, an atmospheric deposition rate of 0.06 dpm cm−2 yr−1 has been used in previous studies of 210Pbex distribution in the interior Arctic Ocean [Chen et al., 2012; Huh et al., 1997; Lepore et al., 2009; Smith et al., 2003]. This estimate is based on the long-term records of 210Pbex deposition on ice sheets in Greenland [Dibb and Clausen, 1997] and Ellesmere Island [Peters et al., 1997], as well as measurements at the CESAR ice station in the central Arctic Ocean [Smith and Ellis, 1995]. The long-term deposition rate derived from High Arctic lake sediment cores is similar (0.07 ± 0.054 dpm cm−2 yr−1, n = 10) [Lockhart et al., 1998; Michelutti et al., 2008]. Therefore, 0.06 dpm cm−2 yr−1 probably represents an appropriate rate (at the low end) for parts of the CAA and Baffin Bay/Davis Strait. At the high end, 210Pb deposition rates up to 0.22 dpm cm−2 yr−1 may be expected for the western part of the study area, partly because the duration of sea ice cover is shorter than in the CAA and partly because the adjacent continental land mass has more extensive soil cover than the Arctic islands. Deposition rates of 0.08–0.22 dpm cm−2 yr−1 have been reported for northern Alaska [Preiss and Genthon, 1997; Weiss and Naidu, 1986] and 0.08–0.14 dpm cm−2 yr−1 for Baffin Island, as determined from lake sediment cores [Michelutti et al., 2008]. Steady state atmospheric deposition rates of 0.06–0.22 dpm cm−2 yr−1 contribute 210Pbex inventories (inventory = deposition rate/λ) of 1.9–7.1 dpm cm−2.
 The supply of 210Pbex from the in situ decay of 226Ra is a function of the inventory of 226Ra in the water column. Data collected previously in the Chukchi Sea [Lepore et al., 2009; Smith et al., 2003] indicate that there is relatively little variation in the average 226Ra activity in the water column (6.9–8.0 dpm/100 L), and thus, most of the variation in the 226Ra inventory in the water column is related to the depth of water column (r2 > 0.997). Assuming an average 226Ra activity of 7.5 (±1.1) dpm/100 L throughout our study area, the 226Ra inventory in the water column at each sampling site may be estimated as 226Ra inventory (dpm cm−2) = 7.5 (±1.1) dpm/100 L × 103 L m−3 × m2/102 cm2 × water depth (m). This approach yields a production of 210Pb from 226Ra decay varying from 0.4 dpm cm−2 at the shallowest site (50 m) to 15.9 dpm cm−2 at the deepest site (2125 m).
 The inventories of 210Pbex in the cores either meet or exceed the estimated total supply of 210Pbex to ocean waters from atmospheric deposition and 226Ra decay (compare solid and hatched bars (plus error bars) in Figure 6b; significant differences are indicated with a star). These results imply that on average, scavenging is capable of transferring all of 210Pbex from the water column to the sediments. In contrast, in the deep Arctic Ocean basin, sediment inventories of 210Pbex are, on average, two thirds of the total supply [Huh et al., 1997; Smith et al., 2003; R.W. Macdonald, unpublished data]. These low inventories likely reflect the exceptionally low vertical particle fluxes characteristic of the interior Arctic Ocean [cf. Honjo et al., 2010], which are not capable of removing all of 210Pbex supplied within the water column by 226Ra decay. The present data provide support for the high scavenging efficiencies (>95% of the total supply) calculated from water column budgets in the highly productive Chukchi shelf area [Chen et al., 2012; Huh et al., 1997; Lepore et al., 2009] and the Beaufort Sea [Smith et al., 2003]. They also indicate that high scavenging efficiencies occur in other parts of the margin, including the interior CAA, where inefficient scavenging might have been predicted due to persistent sea ice inhibiting atmospheric deposition, and low primary production and negligible river inflow limiting vertical particle flux [Howell et al., 2006; Michel et al., 2006; Tremblay et al., 2009]. To the contrary, the 210Pbex-derived sediment accumulation rates in the CAA cores (Table 1) are relatively high by marine standards [cf. Darby et al., 2009] implying a sufficient supply of suspended particulates to scavenge 210Pbex and deposit it at the bottom. (Note that the cores do reflect depositional basins and may not be representative of all other areas.)
3.6 Spatial Variation in the Accumulation of 210Pbex and Inferences About the Strength of Boundary Scavenging
 The accumulation of 210Pbex (Figure 6b) implies that scavenging along the North American Arctic margin is enhanced compared to the interior Arctic Ocean, but with pronounced regional variation. From west to east, inventories generally increase from the northern Bering-Chukchi Sea shelves to Barrow Canyon and the Beaufort slope, decrease in the CAA, and then increase again in the eastern part of the study area (eastern margin of the CAA and Baffin Bay/Davis Strait). Cores containing particularly large inventories of 210Pbex, up to 21-fold greater than can be accounted for by atmospheric deposition and water-column 226Ra decay, occur in the North Bering-Chukchi shelf (UTN7), Barrow Canyon (BC6), and Baffin Bay/Davis Strait (DS5 and DS1) (Figure 6).
 The strength of boundary scavenging, as measured by 210Pbex inventory in sediments along the Arctic margin, depends on having supply of both particles and dissolved 210Pb. The latter depends partly on the strength of lateral exchange with deep or interior ocean waters, relatively rich in dissolved 210Pbex as a result of the long residence time of the water in an area with low scavenging intensity [Roy-Barman, 2009; Rutgers van der Loeff and Geibert, 2008]. However, another process may contribute to the variation in 210Pbex inventories among the cores (Figure 6): sediment resuspension and focusing that redistribute surface sediments enriched with 210Pbex [cf. Lepore et al., 2009; Masqué et al., 2003; Smith et al., 2003]. Although resuspended material can also scavenge 210Pb from the water during transport [cf. Radakovitch et al., 2003], the 210Pbex inventories elevated as a result of focusing itself may be misinterpreted as strong scavenging.
 To distinguish variation in the strength of boundary scavenging from the effects of sediment focusing, we exploit the disparity in chemical behavior between 137Cs and 210Pbex, specifically the low particle reactivity of 137Cs in marine waters (roughly 2–4 orders of magnitude lower than that of 210Pb) [IAEA, 1985]. This low reactivity means that 137Cs, in contrast to 210Pbex, is not significantly affected by scavenging processes within the marine environment [cf. Smith et al., 1998, 1999]. In Figure 7, we have plotted the inventories of 210Pbex in the cores, adjusted to remove the potential supply by 226Ra decay (210Pbex-adj = 210Pbex − 226Ra decay), against the inventories of 137Cs. Also shown in Figure 7 (see shaded area) is a zone corresponding to the 210Pbex/137Cs inventory proportions in terrestrial soil along the North American coast. We estimate that the 210Pbex/137Cs inventory ratio in terrestrial soil is in the order of 1.0–3.9, based on a 210Pbex inventory in soil of 1.9–7.1 dpm cm−2 (reflecting atmospheric inputs of 0.06–0.22 dpm cm−2 yr−1) and a 137Cs inventory of ≤1.8 dpm cm−2 (reflecting the decay-corrected fallout signal) [Baskaran and Naidu, 1995].
210Pbex-adj inventories in marine sediments unaffected by either focusing or boundary scavenging should be similar to the terrestrial soil values because in adjusting the 210Pbex inventories, we have accounted for supply of 210Pbex in the marine environment by dissolved 226Ra decay, leaving atmospheric deposition as the only other source. Due to the difference in particle reactivity, boundary scavenging will tend to enrich 210Pbex in margin sediments without significantly affecting 137Cs. Thus, the strength of boundary scavenging is reflected in the angle of the vector (see bold arching arrow) in Figure 7, with stronger boundary scavenging shifting sediments closer to the 210Pbex axis in the plot. On the other hand, focusing will increase the inventories of both 137Cs and 210Pbex without substantially changing their proportions [cf. Masqué et al., 2003]. This process is reflected in the length of the vector in Figure 7, i.e., it has the effect of shifting sediments further away from the origin. All the North American Arctic margin sediments lie above terrestrial soil in Figure 7 (210Pbex-adj/137Cs ratios ≥ 4.5), which confirms that in all cases, boundary scavenging has contributed to the 210Pbex inventories in the cores.
 Most of the North American Arctic margin sediments lie along a central trend line (R = 0.92) in Figure 7, which corresponds to an average 210Pbex-adj/137Cs ratio of about 7.2. The large spread of samples along this trend line, away from the origin, is interpreted as variation in the extent to which the core sites have been influenced by sediment focusing. Thus, core UTN7, which plots farthest from the origin along the central trend line (Figure 7), appears to have been most strongly influenced by focusing. Core UTN7 had one of the largest 210Pbex inventories (48.7 dpm cm−2), which was much greater than the inventories in other cores from the same region (Chukchi shelf; Figure 6). UTN7 was the most northerly of the UTN core sites (Figure 1) and the deepest (58 m), lying just at the southern edge of a shallow depression known as Hope Basin. A plausible scenario, based on previous observations in the area [cf. Chen et al., 2012; Dunton et al., 2005], is that the Hope Basin is a site of deposition and accumulation of particles produced locally or particles resuspended farther south in the Chukchi or North Bering Seas and then transported northward by the general water transport. A core collected from the northern end of Hope Basin in 1994 (S2, 52 m depth; R.W. Macdonald, unpublished data) plots near UTN7 in Figure 7, supporting the notion that the basin generally is an area of focusing.
 To a lesser extent, focusing also explains large 210Pbex inventories observed in several other margin cores, such as QM1, BC4, and CG1, which plot partway along the central trend line in Figure 7. QM1 was collected from a small embayment at the western margin of the CAA (Queen Maud Gulf), in a basin where fine-grained sediment particles resuspended on the surrounding shelves may be expected to focus [cf. Radakovitch and Heussner, 1999]. Core BC4 was collected from the upper portion of Barrow Canyon (599 m), where currents interact with the complex bathymetry, forming eddies that allow for deposition of sediments transported in suspension from the nearby shelf [cf. Darby et al., 2009]. Core CG1 was collected from a water depth of 204 m in the Beaufort Sea. In this area, sediments supplied by the Mackenzie River and coastal erosion are transported across the shelf through cycles of deposition, resuspension, and transport associated with, e.g., fall storms [O'Brien et al., 2006]. All three cores (QM1, BC4, and CG1) had larger radionuclide inventories than the other cores from their respective regions (Figure 6) as well as higher sediment accumulation rates (0.18–0.24 cm yr−1; Table 1), which are consistent with expectations for focusing. In summary, based on Figure 7 and the available supporting evidence, the large 210Pbex inventories in cores QM1, BC4, CG1, and especially UTN7, appear to have been influenced significantly by focusing and thus are not considered representative of the strength of boundary scavenging in their respective regions.
 The remaining cores with large inventories of 210Pbex—BC6, DS1, and DS5 (Figure 6)—are strongly enriched in 210Pbex, relative to 137Cs (210Pbex-adj/137Cs ratios of 18.2–30.9), lying outside the 95% confidence limits of the central trend line for the North American margin sediments (Figure 7). For BC6, this is true whether or not the inventories include the 0–7 cm turbidite layer (Figure 7). Furthermore, core BC6 collected nearby by other investigators [Lepore et al., 2009; Pirtle-Levy et al., 2009] plots close to our BC6. We infer that particularly strong boundary scavenging occurs at these sites, indeed stronger scavenging than occurs over the Chukchi shelf, now that we have taken into account that focusing has influenced the inventories of 210Pbex in the cores from that shelf. This result is striking because not only is the Chukchi shelf highly productive, which will ensure that scavenging is efficient, but it is also the site of Pacific inflow to the Arctic Ocean (via Bering Strait), which provides the principal seawater source (forming the top 200 m or so of the water column) in the western Arctic Ocean and along the North American margin [Jones et al., 1998]. This freshly inflowing water is generally taken to be a significant source of 210Pb for scavenging during transport across the Chukchi shelf [cf. Lepore et al., 2009; Chen et al., 2012]. The particularly strong boundary scavenging at core sites BC6, DS1, and DS5 suggests that they have had access to an additional seawater source that is richer in dissolved 210Pbex than the Pacific inflow, a notion that is also supported by the very high 210Pbex activities in particles being deposited at these core sites, as estimated from the 210Pbex profiles in the cores (C0 = 17–36 dpm cm−3, as compared to ≤11 dpm cm−3 in the remaining cores; Table 1).
 We propose that interior/deep Atlantic origin water provides the source of 210Pbex and that strong lateral exchange with interior/deep Atlantic origin waters is a key determinant in the sites of strong boundary scavenging along the North American margin. Pacific inflow to the Arctic Ocean must pass over the productive Bering Shelf and is thus prone to scavenging before it enters the Arctic Ocean across the 50 m sill at Bering Strait. Once it enters the Arctic Ocean, Pacific water is further scavenged during transit across the southern Chukchi shelf [Chen et al., 2012; Huh et al., 1997; Lepore et al., 2009; Oguri et al., 2012; Smith et al., 2003]. The transit time for Pacific waters flowing eastward along the North American margin and out through the CAA (in the order of 5–10 years) [Michel et al., 2006] is less than the ~22 year half-life of 210Pb, which means that dissolved 210Pbex concentrations do not have time to grow back into secular equilibrium with 226Ra. Indeed, the generally decreasing 210Pbex-adj/137Cs ratios from the Chukchi (8.5–9.9) to the upper Barrow Canyon and Beaufort slopes (6.4–7.2), and then to the interior CAA (5.2–7.4) suggest that there is no significant recovery of dissolved 210Pbex concentrations in the Pacific water mass during its eastward transit. In contrast, the Atlantic layer, recognized by a Tmax at about 400 m in the Beaufort Sea, enters the Arctic Ocean through the deep passage at Fram Strait. This water clearly contains more dissolved 210Pb than Pacific waters, as evidenced by the very large 210Pbex inventories in margin sediments from the Barents Sea (Figure 7) [Maiti et al., 2010; Zaborska et al., 2010]. The particularly strong 210Pbex enrichment relative to 137Cs in the Barents Sea sediments [and see also Aliev et al., 2007] is striking in view of the relatively strong influence of Sellafield 137Cs discharges in the Atlantic domain of the Arctic Ocean [Smith et al., 1998]. Clearly, 137Cs from reprocessing plants is not significantly particle reactive. Furthermore, the 210Pb in the inflowing water mass might be elevated because atmospheric deposition is higher in the North Atlantic than in the North Pacific [Cochran et al., 1990]. Given that Atlantic inflow has the capacity to deliver dissolved contaminant Pb deposited in waters off Europe [Gobeil et al., 2001], it should follow that it also has the capacity to deliver 210Pb. 210Pb-rich Atlantic origin waters in the Atlantic layer clearly have the potential to supply 210Pbex for scavenging at deep slope sites such as BC6 [see also Smith et al., 1999, 2003]. In Baffin Bay/Davis Strait, where cores DS1 and DS5 were collected, there is probably an analogous supply of poorly scavenged waters from the Atlantic Ocean arriving via the Labrador and Irminger Seas. Atlantic waters flow north at depth along the east side of the strait, bringing in 210Pbex that has accumulated in the water column (from atmospheric deposition and 226Ra decay) during the recent history of this water mass in the Labrador and Irminger Seas, where scavenging rates are low (cf. the high 210Pb activities found in upper portions of the water column in the interior Labrador Sea, approximately 10 dpm/100 L [Bacon et al., 1980] as reported in Moore and Smith ).
 What appears to be the key factor at the BC6, DS1, and DS5 core sites is that lateral exchanges with deep/interior Atlantic-origin waters, which contain relatively high concentrations of 210Pbex, coincide with sufficient particle fluxes to accomplish scavenging. The sediment accumulation rate in core BC6 and previous cores from similar depths in the Barrow Canyon area [Darby et al., 2009; Lepore et al., 2009] exceed those at many sites farther upslope and on the adjacent shelf, where larger organic carbon fluxes would be expected as a function of the shallower water depths alone [Christensen et al., 2008]. Lepore et al.  observed a plume of suspended particulate matter (concentrations >0.3 mg L−1) extending out from the shelf at approximately 100 m water depth over the slope toward the interior basin in the Barrow Canyon area, and there are other mechanisms like brine-enriched density flows [Weingartner et al., 2005; Williams et al., 2008] or eddies [O'Brien et al., 2011] that provide mechanisms to enhance particulate transport of shelf sediment out over the slope. The sources of particulate matter for scavenging at core sites DS1 and DS5 in Baffin Bay/Davis Strait may include biogenic materials transported from the relatively productive west Greenland shelf [Ribergaard et al., 2004] or sediment released from sea ice that has drifted south from the Arctic and melted in Baffin Bay/Davis Strait [cf. Darby et al., 2009]. There are two other processes that probably also influence spatial variation in the accumulation of 210Pbex along the Arctic margin: primary production and sea ice rafting. Because organic matter scavenges 210Pb more efficiently than other particulate material (e.g., clays) [Robbins and Edgington, 1975], sites of enhanced primary production and hence marine biogenic carbon fluxes to the seafloor will likely show greater accumulation of 210Pbex. Along the North American margin, these sites include the Chukchi shelf, the Lancaster Sound polynya in northwest Baffin Bay, and the nearby North Water polynya, which is one of the most productive regions in the Arctic [Michel et al., 2002]. Sea ice-rafted sediment provides a possible source of 210Pbex to sites where ice melts [Chen et al., 2012]. First-year sea ice, redistributing sediment on a local scale, would affect sedimentary 210Pbex and 137Cs inventories in much the same way as focusing (i.e., causing an increase in both inventories at the site where melting occurs without altering their proportions). Thus, as suggested by Chen et al. , release of sea ice-rafted sediments transported from the Beaufort shelf into the western Chukchi Sea might contribute, together with sediment resuspension and focusing, to the large 210Pbex inventories in Chukchi shelf sediments. On the other hand, multiyear sea ice that has been circulating in the Arctic Ocean for ~5–7 years appears to preferentially accumulate 210Pbex via atmospheric deposition and possibly also adsorption of dissolved 210Pbex from the water column onto the sea ice sediments [Baskaran, 2005; Masqué et al., 2007]. Therefore, multiyear sea ice imported from the Beaufort Gyre and melted along the North American margin would contribute particulate 210Pb and particles that could potentially scavenge more 210Pb from the water. However, the amount of multiyear sea ice imported to the North American margin from the Beaufort Gyre appears small [cf. Cámara-Mor et al., 2010].
3.7 Schematic of Processes Controlling 210Pb Accumulation Along the North American Arctic Margin
 In Figure 8, we propose a simple schematic of the processes controlling 210Pbex distribution patterns in sediments along the North American Arctic margin by placing our observations in the context of the literature and what is known about water mass circulation pathways, sea ice cover, degree of terrestrial influence, productivity, and strength of lateral exchanges with the deep/interior ocean in this broad area. In keeping with Carmack et al. , we consider the boundary along the North American continent as a transporting system that conducts water of Pacific origin from the Chukchi Sea to Baffin Bay. We highlight those parts of the margin where our data suggest a strong lateral exchange with deep/interior, 210Pb-rich Atlantic-derived waters.
 Beginning in the west (Figure 8, right side of top panel) and following the general track drawn on the top panel, Pacific-origin water flows northward through Bering Strait and into the southern Chukchi Sea. As the water mass flows north across the highly productive North Bering-southern Chukchi Sea shelves, 210Pbex is intensively scavenged by particulate organic matter (POM), which settles through the shallow water column to the seafloor (Figure 8). Because they receive high fluxes of labile carbon, these sediments are vigorously biomixed, with the consequence that 210Pb cannot be used to determine sedimentation rate. Storms resuspend sediments, allowing transport northward with Bering Sea inflow, and some focusing into local depressions [Naidu et al., 2004]. Net deposition increases toward the northern part of the shelf, focusing materials into topographic lows such as Hope Basin (cf. the large inventories of both 210Pbex and 137Cs in cores UTN7 and S2; Figure 7).
 After crossing the Chukchi Shelf, Pacific water has undergone intensive scavenging which removes much of its dissolved 210Pbex, a situation from which it does not fully recover during the remainder of its Arctic Ocean passage. Phytoplankton blooms occur over the shelf break along the Alaskan margin due to upwelling [Christensen et al., 2008; Pickart et al., 2011], and the produced POM provides the means to transport 210Pbex to the seafloor, where it is mixed into the sediments presumably by benthic organisms (e.g., the deeply mixed 210Pbex profile in core BC3; Figure 3). Barrow Canyon and possibly other canyons provide the conduit to enhance transport of resuspended sediment and POM to the basin [cf. Cooper et al., 2009; Darby et al., 2009; Pickart, 2004], as evidenced by the inventory of 210Pbex at BC6 (Figure 7). Further transport from margin to interior is accomplished by other processes like resuspension, nepheloid layers, eddies [O'Brien et al., 2006, 2011], and brine-enriched density flows [Weingartner et al., 2005; Williams et al., 2008]. As these shelf-derived particles settle over the slope through the Atlantic layer, deeper than ~200 m, they scavenge 210Pbex (Figure 8). Subsequently, 210Pbex-rich sediments are deposited onto the seafloor and accumulate near the base of the slope.
 Sediments supplied especially by the Mackenzie River, but also by other smaller rivers and erosion along the coast of the Beaufort Sea, provide another source of particles to scavenge dissolved 210Pbex and deliver it to the seafloor. Surface sediments and associated 210Pbex are moved across the shelf via repeated cycles of deposition, resuspension, and lateral transport [O'Brien et al., 2006] and focused into small-scale depressions (e.g., sites such as CG1; Figure 7).
 Continuing east, the CAA is fed mostly by water of Pacific origin. Sediment supplied by small- and medium-sized rivers (e.g., Back and Coppermine Rivers) is focused into the deeper basins within the CAA (e.g., especially QM1 (Figure 7) and to a lesser extent PS2, BE2, FS1, VS1, and PS1). With the limited fetch, local sea ice rafting may provide the primary mechanism of offshore sediment transport. However, even where the particle supply by ice rafting is high (e.g., at core site PS2; Figure 3), the dissolved 210Pbex available for scavenging remains low, there having been insufficient time to re-grow 210Pbex from dissolved 226Ra during the approximately 5–10 year transit time for seawater [Michel et al., 2006].
 At the eastern margin of the CAA, in Lancaster Sound and Nares Strait, eastward flowing Pacific-origin waters overflow and mix with water in Baffin Bay that is of Atlantic-origin via Davis Strait [Melling et al., 2001; Michel et al., 2006]. Exchange with the latter water mass supplies dissolved 210Pb to Baffin Bay shelf and slope areas, where it is efficiently scavenged by biogenic POM produced in the area's polynyas (Figure 8). The predominantly Pacific-derived water mass in northern Baffin Bay, which contains low levels of dissolved 210Pbex as a result of scavenging in Lancaster Sound and nearby polynya areas, then flows southward along the west side of Baffin Bay. On the east side of the bay, the west Greenland current transports Arctic outflow via Fram Strait northward [Cuny et al., 2005]. Particles transported from the west Greenland shelf or released by melting sea ice scavenge dissolved 210Pbex from the Atlantic-origin waters and deposited onto the seafloor.
 Most of the sediments along the North American Arctic margin contain an actively mixed surface layer, but for many of the cores, a subsurface layer with negligible mixing provides an opportunity to estimate sedimentation rates using 210Pb. An exception is found in the North Bering-Chukchi shelf, where the fluxes of labile carbon support biomixing that is too vigorous and too deep to permit derivation of sedimentation rates with 210Pb. In the former cores, the depth of penetration of 137Cs provides a time-stratigraphic marker of the early 1950s, reflecting fallout of the first significant releases of 137Cs into the environment due to atmospheric weapons testing. However, fallout cannot account for significant recent particulate 137Cs deposition in margin sediments; this likely reflects a delayed transport of particle-associated 137Cs from watersheds to the coastal marine environment (e.g., particulates from rivers and coastal erosion) and subsequent redistribution of shelf sediments through resuspension, lateral transport, and sea ice rafting. In contrast, sedimentary 210Pbex inventories reflect dissolved 210Pbex, supplied to the water column by atmospheric deposition and decay of dissolved 226Ra, scavenged from the water column by suspended particulate matter at all stations. The interplay of two factors at different spatial scales—large-scale 210Pb supply and local particle flux/depositional setting—controls regional variations in 210Pbex accumulation. Sites of strong boundary scavenging along the margin are characterized by the opportunity to exchange with deep/interior Atlantic waters coincident with sufficiently large particle fluxes to scavenge the associated relatively high concentrations of dissolved 210Pb. To avoid incorrectly assigning elevated 210Pbex inventories as products of scavenging when they are actually produced by focusing, we propose the tandem use of 210Pb and 137Cs. This approach may be helpful in studies seeking to use marginal sediments to understand environmental change.
 We gratefully acknowledge the Canadian IPY Program and the Natural Sciences and Engineering Research Council of Canada for their financial support. We thank the captains and crews of the Canadian Coast Guard Ships Wilfrid Laurier (July 2007) and Louis S. St. Laurent (July 2008). We also thank S. Jobidon for support in the laboratory, D. Dubien for technical assistance at sea, L. Rancourt for research assistance, and P. Kimber for drafting the diagrams. R. Schlitzer is acknowledged for Ocean Data View software. Suggestions from Michiel Rutgers van der Loeff improved the manuscript.