Boundary scavenging in the Pacific Ocean: a comparison of 10Be and 231Pa

Concentrations of U, Th, 231Pa and 10Be were measured in Holocene sediments from two cores collected off the west coast of South America, two cores from the East Pacific Rise, two from the equatorial Pacific and one from the south Pacific central gyre. Our results, together with data from 5 cores reported in the literature, show that boundary scavenging plays a major role in the removal of 10Be from the Pacific Ocean. Deposition rates of 10Be at three margin sites are more than an order of magnitude greater than at sites of red clay accumulation in the deep central Pacific. Deposition of 231Pa is 4 to 5-fold greater at the margin sites. The residence time of 10Be with respect to chemical scavenging, defined as its inventory in the water column divided by its rate of removal to the sediments, varies regionally from > 1000 years at the red-clay sites in the deep central Pacific to ∼ 100 years at the margin sites. Different factors control boundary scavenging of Pa and Be. For example, scavenging of 231Pa is enhanced by metal-oxide coatings of particles, whereas this seems to have little influence on the scavenging of 10Be. © 1990.


Introduction
Several studies have shown that certain insoluble elements are removed from seawater and deposited in sediments at ocean margins at much greater rates than they are deposited in open-ocean sediments. This preferential deposition of insoluble elements at ocean margins has been termed "boundary scavenging" [1], and has been most intensively studied with respect to its effect on the transport and fate of naturally-occurring uranium-series radionuclides such as 21°pb [1][2][3][4] and 231pa [5][6][7][8]. Enhanced deposition of fallout-derived Pu isotopes in ocean-margin sediments has also been reported [9][10][11].
Studies of 23°Th and 231pa have provided the most unequivocal evidence for the existence, and importance, of boundary scavenging. 23°Th and 231pa are introduced into seawater by decay of their respective dissolved uranium progenitors (234U, 235U; the only significant source of 23°Th and 231pa in the oceans) with an initial 231pa/23°Th activity ratio of 0.093. Thorium and protactinium * Present address: Center for Accelerator Mass Spectrometry, Lawrence Livermore National laboratory, L-397, Livermore, CA 94550, U.S.A.
are both very insoluble, being removed to the sediments by chemical scavenging processes on a time scale (10-100 years [5,6]) much less than their radioactive half lives (23°Th=75,200 y; 231pa = 32,500 y). Since decay in the water column is negligible, the ocean-wide average unsupported 231pa//23°Th activity ratio for newly-formed sediment must equal the production ratio of 0.093. Deviations from a ratio of 0.093 in recent sediments reflect a net lateral redistribution of 231pa relative to 23°Th in the water column during the time interval between their production and burial [5][6][7][8]12,13]. 23°Th is deposited at a relatively uniform rate over the sea floor (when normalized to a constant water depth), with little net lateral transport between its production in the water column and burial in marine sediments [7,[12][13][14][15]. In contrast to 23°Th, 231pa is extensively redistributed. Examination of 231pa/23°Th ratios in particulate matter and sediments showed that only about 1/3 of the 231pa produced in central ocean gyres is removed to underlying sediments, with the remaining -2/3 being transported to ocean margins prior to burial [5,6,12,13,15]. This net lateral transport occurs over distances of up to thousands of kilometers on time scales of ~< 100 years [5,6]. Most elements have residence times in the deep ocean > 100 years, so the potential for boundary scavenging to play an important role in their transport and fate in the ocean is not limited by the length of time required for the elements to be transported laterally over basin-scale distances.
The biogeochemical processes that lead to boundary scavenging, and their relative importance, have not been unambiguously defined. Greater total fluxes of particulate matter near ocean margins certainly contribute to enhanced chemical scavenging. The relative time scales of lateral mixing and of the residence times of the scavenged elements is another important factor [12]. It has also been suggested that specific particulate phases such as MnO 2 coating on particles [1,6] or biogenic opal [15], contribute to enhanced scavenging in ocean-margin waters. However, further work is needed to fully understand the importance of individual particulate phases. Until then, it is impossible to predict from fundamental principles the extent to which a particular element will be affected by boundary scavenging. For now, this must be determined empirically by measuring the deposition rate of an element of interest in representative open-ocean and ocean-margin environments.
~°Be is an important tracer for dating deep-sea sediments and Mn nodules [16][17][18]. A thorough understanding of the sources, transport, and fate of ~°Be in the oceans is essential to obtain reliable chronologies from 1°Be and to fully utilize it as a tracer of geochemical process. This need for a better understanding of the marine geochemistry of 1°Be led us to investigate the extent to which ~°Be is preferentially removed from seawater at ocean margins.
Limited evidence suggests that ~°Be removal from the oceans may be influenced by boundary scavenging. 1°Be accumulation rates in margin sediments off Northwest Africa [19] and off California [20] have been found to be greater than the rate of l°Be supply to the earth's surface from the atmosphere. However, this alone does not unequivocally prove that l°Be is preferentially removed at ocean margins because other physical transport processes, such as sediment focussing by bottom currents and gravity flows (e.g., turbidites, slumps), could also lead to local rates of 1°Be accumulation much greater than the ocean-wide average. Some method is needed to correct for the effects of local sediment transport and compare the rates of 1°Be scavenging from the water column at ocean-margin and open-ocean sites. Our solution to this problem is to normalize 1°Be concentrations in sediments to activities of unsupported 23°Th. Bacon and coworkers [21][22][23] have shown that concentrations of trace elements, CaCO3, clay, etc., can be normalized to 23°Th activities to determine regional scavenging rates (fluxes) for the elements (phases) of interest. The relatively uniform deposition of 23°Th makes it an ideal tracer against which the concentrations of other elements such as 1°Be can be normalized to determine whether or not they are preferentially removed at ocean margins.
The residence time of Be in the deep sea is long enough for lateral mixing to produce a nearly constant 1°Be concentration of deep waters within an ocean basin [24], although interocean differences in the concentration of l°Be have been reported [25]. The uniform concentration of 1°Be within the deep waters of an ocean basin can, in effect, be treated as a uniform source for chemical scavenging, much like the uniform production of 23°Th and 231pa by U decay. Therefore, 1°Be/ 23°Th ratios in marine sediments will reflect the extent to which ~°Be is preferentially removed from seawater at ocean margins. This is essentially the same approach that was used to identify boundary scavenging of 231pa [5][6][7][8].

Sample locations
23°Th, 231pa and 1°Be concentrations were measured in Holocene sediments from several sites representing open-ocean and ocean-margin environments in the Pacific Ocean (Table 1). Sites were selected to examine the effect on boundary scavenging of particle composition, as well as the magnitude of particle flux. RCll-210 and V28-238 are from equatorial areas with relatively high fluxes, for open-ocean regions, of biogenic partitles. V18-299 is from the south central Pacific, representing an area more typical of central ocean gyres having a lower particle flux. Although particulate matter fluxes have not been measured near V18-299, studies have shown that primary productivity and the ensuing particle flux are higher at the equator than in central ocean gyre regions. For example, Betzer et al. [26] measured primary production and particle flux along a N-S transect across the equator at 153°W and showed that primary productivity and particle flux were about four times higher at the equator than at 12°N. Similarly, comparing fluxes measured at MANOP Site C (01°03'N; 138°56'W [27]) and at PARFLUX P1 (15°21'N; 151°29'W [28]) shows about fourfold higher fluxes of total particulate matter and of organic carbon near the equator (I°N). If the pattern of poleward decrease in particle flux is symmetrical about the equator, then the particle flux near V18-299 would be approximately a quarter of that near RCll-210 and V28-238.
It has been suggested that the presence of manganese (and possibly iron) oxide coatings on particulate matter contributes to enhanced scavenging in ocean-margin regions [1,6]. We have analyzed sediments from two cores collected on the East Pacific Rise (TT154-10 and V19-55) which are known to be rich in Fe and Mn oxyhydroxides derived from venting of metal-rich hydrothermal solutions at the ridge crest [29,30]. Two cores from the equatorial Pacific off Equador (V19-28 and V19-29) were selected to represent an oceanmargin environment.

Methods
Prior to dissolution of the samples (about 1-2 g), 1.88 mg of 9Be (as BeSO 4 solution) was added as a chemical yield monitor along with 236U, 229Th and 233pa. Measurement of U, Th and Pa followed approximately the method described before [31]. Following the separation of U, Th and Pa, Be remained with A1 and other major elements in a small volume (15-20 ml) of dilute HC1. Be was then purified as follows.
Ten normal NaOH was added to the HC1 solution to bring the pH up to 14, which precipitated Mn and residual Fe. After centrifugation, the supernate (containing A1 and Be) was decanted and saved. The precipitate was resuspended in a NaOH solution (pH = 14) and centrifuged. The two supernate fractions were combined and then acidified with HC1 to pH = 1. A1 and Be hydroxides were precipitated overnight by adding NH4OI-I to pH between 8 and 9.
The precipitate containing A1 and Be was separated from the supernate by centrifugation and then was dissolved by 20 ml 1.2N HCI and loaded to a 20 cm column (i.d. 1 cm) of cation exchange resin (Dowex Ag 50W × 12) preconditioned in 1.2N HC1. The column was washed with 20 ml of a Average Holocene accumulation rates of the cores are from the following sources: Radiocarbon (AMS) ages of TT154-10 and V28-238 are from [67]. The chr0nostratigraphy of RCll-210 is based on the 81So record of Globoratalia tumida [52]. The time scale of V19-28 is from a reinterpretation of the 81SO record by Lyle et al. [51]. For V18-55 and V19-29, the core depths of the last glacial maximum (18 ka B.P.) were given by Moore et al. [68]. The ages of the samples in these two cores are derived assuming constant sediment accumulation rates since 18 ka B.P. For V18-299, we measured downcore U-series isotopes and used the excess 23°Th and 231pa to construct the sediment accumulation rate of the core. 8180 of the bulk carbonate was also measured from which the 18 ka horizon was assigned to a depth of 16 cm. The lower accumulation rate is the long term average based on U-series results while the higher value is obtained by assuming constant accumulation since 18 ka. 1.2N HC1 and eluant was discarded. Be was eluted from the column with an additional 50 ml of 1.2N HC1. A clean teflon beaker (250 ml) was used to collect Be which was ready for solvent extraction. If the sample was carbonate rich, i.e., low A1 content, then the above coprecipitation and cation exchange steps were omitted, and Be purification began with the following solvent extraction.
To the 50 ml Be solution was added 6 ml of Na2EDTA solution (20g Na2EDTA + 110 ml 1N NH4OH ) and a magnetic stir bar. The pH was raised to 6 with NH4OH and then 6 ml of acetyl acetone was added. The mixture was stirred for 2 hours, after which 20 ml of chloroform (CHC13) was added to the solution and stirring continued for about 20 minutes. Then the solution was transferred to a teflon separatory funnel and shaken for 5 minutes. After the phases separated, the CHC13 phase (containing Be) was drained into a clean teflon beaker. The aqueous phase in the separatory funnel was further extracted with an additional 10 ml of CHC13. The two fractions of CHC13 (30 ml) were combined and transferred to a clean teflon separatory funnel and 10 ml of 7N HC1 was added and shaken for 5 minutes. The CHC13 was drained into the used beaker and the HC1 (containing Be) into a clean teflon beaker. The CHC13 was returned to the separatory funnel and extracted with additional 10 ml of 7N HC1. The combined HC1 solution was heated to dryness 3 times in conc. HNO 3. Then conc. HC1 was added and heated to dryness 2 times.
The residue was dissolved in 10 ml of 0.12N HC1 and loaded to a second cation column which was identical to the first one except that the cation exchange resin was conditioned in 0.12N HC1. The column was first washed with 30 ml of 0.12N HC1 and then 35 ml of 1.2N HC1 and all the eluant up to this point was discarded. 50 ml of 1.2N HC1 was used to elute Be out of column (residual A1 remained in the resin). The Be solution was collected in a clean teflon beaker and heated. When near dryness, 1 ml of conc. HNO 3 was added and then evaporated to dryness. The residue was taken up with 2 ml of conc. HC1 and then evaporated to 0.5 ml. 5 ml of Milli-Q H20 was added to the 0.5 ml HC1 solution and heated gently for a few minutes. NH4OH was added to precipitate Be(OH)2 at pH = 8 -10 overnight. The precipitate was filtered onto an ashless combustible filter (Whatman #42) and combusted in a Pt crucible at 900 °C for 4 hours. The Be oxide was then analyzed by accelerator mass spectrometer [32].

Results and discussion
Measured 23°Th and 231pa activities (Table 2) consist of three components: that which was scavenged from seawater, the principal component of interest, that which was supported by U in the detrital minerals, and that which was produced by decay of authigenic U found in the sediments. Authigenic U is present in most of the sediments analyzed, as can be seen by comparing activities of 238U with those of 232Th. Lithogenic material has a 238U/232Th activity ratio of about unity, reflecting the average crustal abundances of U and Th [33]. Organic-rich margin sediments (V19-28 and V19-29) and metalliferous sediments from the East Pacific Rise (TT154-10 and V19-55) both have 238U/232Th activity ratios greatly in excess of one (Table 2). This authigenic uranium must be taken into account in calculating unsupported 23°Th and 231pa activities. Based on surveys of the literature, and on other work done in our laboratory, an average 238U/232Th activity ratio for the lithogenic fraction of marine sediments is estimated to be 0.8 + 0.2. This factor was used along with the measured 232Th activity to correct for the lithogenic component (i.e., the detrital U-supported fraction) of 23°Th and 231pa. It was further assumed that authigenic uranium accumulation was synchronous with deposition of the sediments, so simple ingrowth equations could be used to evaluate the production of 23°Th and 231pa by decay of authigenic U. The component of Z3°Th scavenged from seawater, hereafter referred to as the unsupported fraction, was then calculated as: a Sources for sediment chronologies are given in Table 1. b Subscript "xs" designates unsupported value. Subscript "0" indicates that the measured activities of xs (unsupported) 23°Th and 23:Pa have been decay corrected to the time at which the sediments were deposited. c Deposition Indices represent the measured (N/xs23°Th)0 ratio, where N =231pa or 1°Be, divided by the ocean-wide average value for the ratio. Values < 1.0 indicate net loss of N by lateral transport while values > 1.0 reflect a net lateral import of the isotope relative to 23°Th. See text. and for 231pa: where subscripts "xs" and "m" refer to unsupported and measured activities, respectively, t is the time since deposition of the sediments, and X is the appropriate radioactive decay constant. The factors 1.14 and 0.046 represent the 234U/238U and the 235U/23SU activity ratios in seawater, respectively. Equation (1) fails to correct for decay of unsupported authigenic 234U, but this correction is negligible for the results reported here because all of the samples are of Holocene age. Uncertainties derived from counting statistics for each nuclide were propagated along with the 25% uncertainty in the lithogenic U/Th ratio to obtain the reported uncertainties for the unsupported 23°Th and 231pa activities. Unsupported 23°Th and 231pa activities at the time of sediment deposition (Table 3) were calculated from the average age of each sample, t, as: where ,sN is the unsupported 23°Th or 231pa activity calculated from equation (1) or (2), respectively. For V28-238 and TT154-10, measured 14C ages were used for t. For the other open-ocean cores, a bioturbated mixed layer of 5 or 6 cm was assumed in assigning an age to surface samples (see Table I for sources of core chronologies). Sediment accumulation rates at the margin sites were sufficiently rapid that the effect of bioturbation on the age correction is negligible.

23°Th // 232 Th ratios in the South Pacific
Before discussing the evidence for boundary scavenging, note that the 23°Th/E32Th ratios of the East Pacific Rise sediments ( Table 2; TT154-10 and V19-55)are the highest ever reported for marine samples other than corals. The large uncertainties result from the very small number of 232Th counts detected by alpha spectrometry. Reanalysis of the two samples from T1~154-10 by mass spectrometry gave 23°Th/E32Th activity ratios of 790_+ 10 (1.2-2.4 cm) and 800_+ 17 (4.2-6.0 cm), confirming the high ratios obtained by alpha spectrometry. Thorium is not significantly removed from basalts by hydrothermal solutions, and the Th which is injected into seawater from mid-ocean ridge vents has a 23°Th/232Th activity ratio certainly < 100 and probably < 10 [34,35]. Therefore, seawater is the only reasonable source for Th with a 23°Th/232Th activity ratio of -800. These high ratios are, in fact, a lower limit for the isotopic composition of Th dissolved in South Pacific deep water because the EPR sediments must contain some lithogenic material, which has a 23°Th/232Th activity ratio of about unity, as noted above.
Previously reported values for the 23°Th/232Th ratio of Th dissolved in seawater have generally been less than 200 [5,6,[36][37][38][39][40]. Most of these ratios were measured in samples from the North Pacific. 23°Th/232Th ratios of -800 in EPR sedime'nts indicate one of two things: either the 230 Th/232 Th ratio of Th dissolved in South Pacific deep waters is much greater than in North Pacific waters, or, the North Pacific samples suffered from some form of 232Th contamination. The latter seems unlikely because the North Pacific samples were collected and analyzed by a variety of methods and were processed by investigators in at least 4 different laboratories. It is more likely that the 23°Th//232Th ratios reflect regional variability in the supply of 232Th to the oceans. The residence time of Th in the deep sea is only 20-40 years [5,6], too short to permit intrabasin homogenization of 232Th concentrations by lateral mixing. Therefore, it is not unreasonable that regional variations in the 232Th content of deep ocean water should exist which reflect a balance between local supply and chemical scavenging.
The East Pacific Rise is situated to receive minimal supply of continental lithogenic material, the leaching of which provides the predominant source of dissolved 232Th in the oceans. Lithogenic material is supplied to the central Pacific primarily by atmospheric transport of continental dust [41]. Eolian transport of dust to the equatorial Pacific is one to two orders of magnitude less than the rate of supply to the central North Pacific gyre at about 38°N [41][42][43]. The region around the EPR at 10-17 ° S, where our cores were collected, should therefore have received a substantially lower eolian input of lithogenic dust (i.e., 232Th), than is supplied to regions of the North Pacific from which previous Th results were obtained. The continental margins are themselves sources of lithogenic particles, and hence 232Th, to the deep sea [40]. The Peru-Chile trench would tend to trap lithogenic material moving downslope from the continent before it reaches the EPR. Furthermore, the EPR rises above the surrounding abyssal plain where fine-grained lithogenic particles could be transported in nepheloid layers. Consequently, there is little opportunity for supply of lithogenic particles to deep waters surrounding the crest of the EPR. This very weak source of 232Th leads to extremely low concentrations of 232Th in deep south Pacific waters which, in turn, are reflected in 23°Th/232Th ratios in EPR sediments much higher than in the North Pacific or elsewhere.

230Th and 231pa results in the context of earlier studies
Unsupported 231pa//E3°Th ratios measured in this survey are generally consistent with results from earlier work. Results are presented here as Pa/Th ratios, rather than the more traditional Th/Pa ratios, because it is normalization to 23°Th that permits lateral transport to be evaluated. 231pa and 23°Th are produced by U decay in seawater at an initial 231pa/E3°Zh activity ratio of 0.093. Sediments from V18-299, where the lowest particulate flux occurs, have an initial unsupported 231pa/E3°Zh activity ratio of 0.044 (Table  3) which is even a little high for central ocean gyres [7]. Sediments from V19-28 and V19-29 have ratios of 0.18-0.20, which are typical of ocean margins [6,7]. East Pacific Rise sediments also have 231pa//E3°Th ratios (0.14-0.16) greater than the ocean-wide average (0.093), reflecting the enhanced scavenging of Pa from bottom waters by metal oxyhydroxide phases [30]. Enrichment of 231pa in metalliferous sediments is consistent with the hypothesis that Mn-oxide coatings on particles contribute to the enhanced scavenging of Pa at (some) ocean margins [6].
One new feature of the distribution of 23°Th and 231pa in the Pacific is that 231pa/23°Th ratios in equatorial sediments (0.069 in RCll-210 and 0.062-0.067 in V28-238) are nearly a factor of two greater than ratios typically found in central ocean gyres (often < 0.033 [7]), indicating an enhanced scavenging of Pa from the equatorial Pacific water column. The eolian input of aluminosilicate phases at the equator is one to two orders of magnitude less than the North Pacific central gyre [41][42][43], and there should not be a significant source of metalliferous particulate phases at these equatorial sites. Something other than lithogenic particles or metal oxide coatings must be responsible for the enhanced scavenging of Pa. Bruland and Coale [44] showed that higher biological productivity at the equator leads to enhanced scavenging of 234Th from surface waters compared to scavenging rates in regions to the north and south. Apparently the relatively greater flux of particles generated by the higher biological productivity in equatorial surface waters also leads to enhanced scavenging throughout the deep water column where virtually all of the 23°Th and 231pa in the sediments originates. Note, however, that the 231pa/23°Th ratios in equatorial sediments are less than the production ratio of 0.093. Although scavenging is enhanced relative to central ocean gyres, there is still a net loss of 231pa from equatorial Pacific waters by lateral transport to the margins.

3. Boundary scavenging of lOBe
Lateral transport of 231pa was quantified by comparing 231pa/23°Th ratios of sediments and particulate matter from different ocean regions [6,13]. Dividing decay-corrected 231 Pa/23°Th ratios of sediments by the production ratio of 0.093 produces a simple deposition index, ID[Pa], which reflects the net lateral import (ID[Pa] > 1.0) or export (ID[Pa]< 1.0) of 231pa at each site examined. A similar deposition index for l°Be can be defined by dividing decay-corrected ~0 Be/23°Th ratios measured at specific sites by the ocean-wide average value for this ratio:  (4) This index is not quite identical to ID[Pa] because of the different nature of the sources of l°Be and 231pa, but it is a useful parameter that permits easy visualization of the relative scavenging intensity of l°Be in different ocean regions. For now, we adopt a value of 0.122 x 10 9 (at dpm -1) for (l°Be/xs23°Th)avg, which is obtained by dividing the global 1°Be production rate (1.21 × 10 9 at cm -2 ka -1 [45]) by the rate of 23°Th production in an average 3800 m deep ocean (9.88 dpm cm -2 ka-l). Although the value of (l°Be/xsE3°Th)avg for the Pacific may be revised when more sedi- ment data become available, 10 values calculated from our data can still be used to compare the relative rates of l°Be scavenging among these sites.
Deposition index values for 231pa and l°Be calculated from our data are presented in Table 3. Additional It)[Be] values calculated from 1°Be and 23°Th results reported in the literature are listed in Table 4. Ku et al. [25] measured a l°Be/xs23°Th ratio of 0.061 (10 9 at dpm -1) in core-top sediments of V28-258 (11°52'S, 165°45'W, 5528 m), about half our open-ocean ratio. Complete 1°Be and 23°Th profiles were reported for this core. The surface-most 23°Th activity, from which the above ~°Be/xs23°Th ratio was calculated, is below that expected from the trend established by deeper samples. Extrapolating the complete profiles to the surface, which corrects for decay in the mixed layer, produces a (l°Be/ xs23°Th)0 ratio of only 0.021 (10 9 at dpm-l). Both measured and extrapolated ratios are reported in Table 4. l°Be and 23°Th results from four red clay cores studied by Amin et al. [46] were decay corrected using their reported values for sediment accumulation rate at each site. A 5 cm thick mixed layer was assumed when decay corrections required a mixed layer thickness. Nuclide activities from DSDP site 580 [47] and MANOP Sites H and M [48,49] were used without decay correction because the former were from rapidly accumulating sediments and the latter from sediment trap sam-pies, neither of which required significant decay corrections.
Amin et al. [46] also calculated the xsZ3°Th inventories for their cores and found that, with one exception, they approximately balanced 23°Th production in the overlying water column. This is consistent with other results indicating little lateral redistribution of 23°Th in the deep sea. More important, it indicates that recent sediments have not been lost from these sites due to erosion by bottom currents, nor were significant amounts of surface sediments lost during the coring process. Loss of surface sediments under either of these conditions would lead to high apparent values for l°Be/xs23°Th0 ratios at these red-clay sites.
Deposition index values for 231pa and l°Be are summarized in Figs. 1 and 2. Although the number of data points is still relatively small, the patterns of scavenging intensity indicated by these 1 D values is consistent with previous results for Th and Pa summarized by Yang et al. [7]. It) values much less than 1.0, indicating net lateral export, occur at open-ocean sites. I D values are much greater than 1.0 at the margin sites, indicative of enhanced scavenging which in turn leads to net lateral import of 231pa and 1°Be to these regions.
Enrichment of 1°Be in ocean-margin sediments suggests that boundary scavenging plays an important role in the removal of Be from the ocean. Fig. 1, Deposition Index values for 231pa. A value less than 1.0 indicates that 231pa is removed from the water column to the sediments at a rate less than its rate of production in the water column while the converse is true for values greater than 1,0. ID[Pa] is a measure of the net lateral export or import of Pa at each site. ID[Pa] values for our cores are from Table 3. ID[Pa] values for the region around DSDP 580 are calculated from [7]. Values for the MANOP sites are from [49] while Pa and Th data from [72] were used to calculate ID[Pa ] for V28-258.
Another factor to consider before accepting this conclusion is whether or not the excess 1°Be in margin sediments could reflect local supply from the continents. The 1°Be concentrations of river sediments range world-wide from I to 4 x 108 at g-1 [50], much too low to have contributed significantly to the l°Be content of the sediments at the two margin sites off South America (3-5 × 109 at g-1; Table 2). These margin sediments are about 80% marine biogenic material [51], further reducing the contribution that continental detritus could make to the 1°Be content of sediments at these sites. The sites of V19-28 and V19-29 are located seaward of the Peru-Chile trench, which would tend to trap detrital material moving seaward from the continent. No large rivers occur in nearby regions of South America which could introduce significant amounts of dissolved 1°Be. While leaching of aerosol dust can be an important local source of A1 and 9Be in surface ocean waters, this seems not to be the case for 1°Be [25]. Ruling out these potential local sources for the excess l°Be in the sediments from V19-28 and V19-29 leads us to conclude that the higher l°Be/23°Th ratios there must reflect preferential removal of ~°Be from seawater at ocean margins.

Deposition rates and residence times of 1°Be in the Pacific
Normalizing 1°Be concentrations to 23°Th allowed us to establish unequivocally that 1°Be is preferentially removed from the ocean at margins. Normalizing to 23°Th will also permit us to quantify rates of ~°Be removal in different ocean regions and, from that, to evaluate residence times of roBe with respect to chemical scavenging for each region.
Taking the average deposition rate of 23°Th to be everywhere equivalent to its rate of production in the overlying water column, the known rate of 23°Th production in seawater (PTh; dpm cm -2 ka -1) can be used to quantify regional average 1°Be deposition rates (SBe; at cm -2 ka -1) [21][22][23] as" Sae = P'rh" (l°Be/xsE3°Th)0 (5) 23°Th is produced at a rate of 0.0026Z dpm cm -2 ka -1 where Z is the water depth in meters, roBe deposition rates calculated from equation (5) are presented in Table 5. These results show that the deposition rate of roBe approximately equals its global average rate of production (1.21 x 10 6 at cm -2 a -1 [45]) in open-ocean sediments containing appreciable amounts of CaCO 3 where accumu-  Table 1). 1°Be consistently accumulates at rates much less than its global production rate in deeper abyssal red-clay regions (first 5 sites in Table 4). In ocean-margin regions, on the other hand, 1°Be is scavenged from the water column at rates as much as six times its average rate of production in the atmosphere, the excess ~°Be being supplied to the margins by lateral mixing from the open ocean. If the average 1°Be deposition rate (6.28 × 10 6 at cm -2 a -~) at the three margin sites (V19-28, V19-29 and DSDP 580) is typical of all ocean margin regions, if such regions constitute 10% of the area of the Pacific Ocean, and if 1.21 × 10 6 at era-2 a-a is an appropriate value for the average rate of ~°Be supply to the Pacific, then about 50% of the ~°Be introduced into the ocean is removed at the margins. As the ~°Be deposition rate at margins is based on results from only three sites, there is a large uncertainty in this estimate. Future studies will reduce this uncertainty, but they should not change the basic conclusion that a large pro-portion of the 1°Be supplied to the Pacific Ocean is deposited in margin sediments. For each site, the residence time of 1°Be with respect to chemical scavenging from the water column can be calculated from the inventory of ~°Be in the water column, 1Be, and the 1°Be deposition rate, Sac, as: "rse = IB~/SB~ (6) ~°Be inventories are calculated for each site using the average ~°Be concentration of Pacific seawater of 1700 at g-1 [24]. Values of ~'B, range from > 1000 years in the red clay regions through -500 years at our three "open-ocean" sites to a value of only -100 years for the margin sites (Table 5).
There are too few data at this time to estimate the Pacific-wide average residence time of Be, but a value of -500 years would be obtained for a 3800 m deep ocean with an ~verage ID[Be] of 1.0. More important than the ocean-wide average value of ~ae is the striking difference between l°Be deposition rates and residence times in deep open-ocean waters and the those in ocean-margin regions. The much greater scavenging rate at margins, coupled with lateral mixing, leads to much higher rates of l°Be deposition in oceanmargin sediments (see section 4.8).

Uncertainties associated with boundary scavenging of 23°Th
Because the deposition rates and residence times discussed in the preceding section were obtained by normalizing measured l°Be concentrations in sediments to a tracer with a known production rate (23°Th), they are free from biases caused by local sediment transport. However, this approach is not totally free from uncertainty because boundary scavenging of 23°Th is assumed to be negligible; i.e., 23°Th is assumed to be scavenged from the water column in the region in which it is produced. Particulate 23°Th fluxes measured with sediment traps and model calculations suggest that as much as 20% of the 23°Th produced in the open ocean may be transported to margins prior to burial [5,6,12,13]. Uncertainty in the extent of lateral redistribution of 23°Th limits the precision with which we can assign scavenging rates of l°Be. Some constraints can be placed on 1°Be fluxes, however, because we know that the tendency for redistribution is from the open ocean to margins. Consequently, open-ocean l°Be fluxes estimated by this approach represent upper limits, whereas calculated l°Be fluxes at ocean margins represent lower limits. Regional values of rBe are similarly affected, so 1000 years is a lower limit for open-ocean regions while 100 years is an upper limit for the margin sites. If there is an error in our approach, then it is in the direction that boundary scavenging of l°Be is even more extensive than we have estimated.
Our approach of normalizing 1°Be concentrations to 23°Th has another distinct advantage in that it allows the deposition rate of l°Be at any site to be estimated by analyzing a single sample for roBe and 230Th. With this approach, it will be feasible to survey entire ocean basins to more completely assess the nature of 1°Be removal from the oceans.

Independent estimates of WBe accumulation rates
l°Be deposition rates can be calculated independently using sediment mass accumulation rate (MAR) data where available to test our results and interpretation derived by normalizing l°Be concentrations to 23°Th. At the margin location of V19-28, we have used nuclide activities (Table 3) and MAR [51] from the upper Holocene record (17-19 cm) to calculate accumulation rates of l°Be (10.1 X 10 6 at cm -2 a -1) and 23°Th (16.7 dpm cm -2 ka-1). The accumulation rate of l°Be is 8.3 times its global average production rate whereas unsupported 23°Th is accumulating at a rate equivalent to 2.36 times its rate of production in the overlying water column. The factor of 2.36 seems too high to reflect boundary scavenging of 230 Th, so it may indicate that local sediment transport has focussed sediment to the core site. However, the factor of 2.36 represents an upper limit for sediment focussing, from which a lower limit for the rate of l°Be removal from the water column in this region can be set at 4.3 × 10 6 at cm -2 a -I (10.1/2.36). Scavenging of 1°Be to these margin sediments occurs at a rate no less than 3.5 times greater than its average global rate of production (1.21 × 10 6 at cm -2 a-l).
Eisenhauer et al. [47] reported a long-term average 1°Be accumulation rate of 11.5 X 10 6 at cm -2 a -1 at the site of DSDP-580 in a margin region site off Japan where Yang et al. [7] documented intense boundary scavenging of 231pa. Brown et al. [20] measured l°Be concentrations in surface sediments off southern California and used average sediment mass accumulation rates to estimate l°Be deposition rates of about 10 × 10 6 at cm -2 a -1 for that ocean-margin region. These results consistently indicate rates of 10 Be deposition in margin sediments much greater than its global average rate of supply.
The MAR for the depth interval of RCll-210 from which our samples were taken was 0.89 g cm -2 ka -1 [52]. Multiplying this MAR by our measured l°Be content gives a l°Be deposition rate of 2.65 x 10 6 at cm -2 a -1, about twice our estimated average rate of 1°Be scavenging from equatorial Pacific waters (Table 5). Similarly, using the initial unsupported 23°Th activity of 23.85 dpm g-1 (Table 3), a 23°Th accumulation rate of 21.2 dpm cm -2 ka-1 is obtained, nearly twice the rate of 23°Th production in the overlying 4420 m water column (11.5 dpm cm -2 ka-1). The extent to which accumulation of 1°Be and 23°Th exceed their respective production rates could reflect lateral transport of dissolved nuclides to a region of enhanced scavenging under the equatorial zone of high productivity. It could also reflect local focussing of sediments to the position where the core was collected, or, it could result from an error in assigning a chronology to the upper portion of the core (M. Lyle, personal communication). At this time there is no way to choose the correct explanation from among these possibilities. This example illustrates the value of normalizing the accumulation rates of elements, nuclides, or solid phases to a tracer such as 23°Th whose known source permits lateral redistribution to be identified and regional burial rates to be evaluated.
Along with their l°Be results, Amin et al. [46] reported sediment accumulation rates based on 23°Th/232Th profiles and in situ sediment densities from which accumulation rates of a°Be can be calculated. These are 0.42, 0.12, 0.47 and 0.81 × 10 6 at cm -2 a -1 for MSN 147G, NOVA 111-13, NOVA 111-16 and MSN 96 respectively. Where inconsistencies exist between these l°Be accumulation rates and those obtained by normalizing to 23°Th (Table 5), the former tend to be lower. This could indicate a net lateral export of 23°Th from these regions, but the inconsistency is not systematic, as would be the case if 230Th export were the cause. This suggests that errors may also exist for the estimated sediment accumulation rates or the sediment densities. In any case, the deposition rate of 1°Be in pelagic red clays is much less than its global average production rate, whereas l°Be is deposited in ocean-margin sediments at rates greatly exceeding its global average production rate.

7. Boundary scavenging of 1°Be in the Atlantic Ocean
By combining results from various sources, we can show that boundary scavenging must also influence the removal of 1°Be from the North Atlantic Ocean. l°Be and 23°Th data are available from two open-ocean sites. Somayajulu et al. [53] studied a core collected at 32°N on the Mid-Atlantic Ridge. A Holocene l°Be/xs23°Th0 ratio of 0.032 × 109 (at dpm -1) can be calculated from their results. At another open-ocean site (40 ° 36' N, 21°42'W; 3485 m), Southon et al. [54] reported a Holocene l°Be/xs23°Tho ratio of 0.070 × 10 9. Resuits are also available for two ocean-margin sites.
Holocene sediments off northwest Africa contain a l°Be//xsE3°Th0 ratio of 0.15 × 10 9 (compare Th results from [55] with Be results from [19] for core 12310: 23°30'N, 18°43'W; 3080 m). We have analyzed surficial sediments from two cores collected on the continental slope south of New England (OCE-152 BC5: 38°09'N, 70°56'W; 2691 m; OCE-152 BC9: 39°42'N, 70°55'W; 1981 m) where we obtained an average l°Be/xsE3°Th0 ratio of 0.40 × 109. The l°Be/xsZa°Tho ratio in margin sediments is as much as an order of magnitude greater than in open-ocean sediments, indicating that preferential removal of l°Be at ocean margins also occurs in the North Atlantic.

Conditions responsible for boundary scavenging
Here the basic principles of boundary scavenging are reviewed to illustrate how certain particlereactive elements (Be or Pa) can follow distinctly different pathways of removal from the ocean compared to other particle-reactive elements such as 23°Th. Bacon [12] reviewed boundary scavenging and provided a quantitative discussion of the processes that lead to the separation of 231pa and Z3°Th between their production in, and removal from, the ocean. A more general illustration is given here of the conditions and processes which lead to boundary scavenging. Important parameters for this illustration include: the reactivity of each element toward scavenging by particles (often expressed as a residence time with respect to scavenging), the magnitude of the particle flux in each region, and the lateral mixing rate in the ocean. For simplicity, it is first assumed that particles have a homogeneous composition throughout the ocean so that differential scavenging reflects only the relative particle flux at each location, and not the surface chemical properties of the particles themselves. The influence of particle composition is discussed below.
Imagine a two-dimensional ocean where each wall represents a margin or boundary. Production of each tracer is uniform throughout the ocean (Fig. 3A). the horizontal mixing coefficient. A typical ocean basin might have X = 3000 km and K H = 3 × 107 cm 2 s-1 [56], in which case T = 95 years. Like the real ocean, the flux of particles is high at the margins and low in the interior (Fig. 3B). Because a homogeneous particle composition is assumed, the rate constant for scavenging of a tracer at any particular site is simply proportional to the particle flux. Reactivity of the tracers increases in the order C < B < A and residence times of the tracers in the ocean interior are set to be: zA = 0.1T, ~' B = 1.0T, and z c = 10T. The relative concentration of each tracer along a transect across the ocean is shown in Fig. 3C, and the relative down-ward particulate flux of each tracer is presented in Fig. 3D. In this illustration, throughout the ocean the flux of tracer A approximately equals its rate of production in the water column because its residence time is too short for lateral mixing to transport significant quantities from interior regions to the margins. The residence time of tracer C, on the other hand, is sufficiently long (10T) that lateral mixing nearly homogenizes its concentration throughout the basin. Therefore, the flux of tracer C is almost directly proportional to the particle flux. Tracer B, for which ~-= T, has an intermediate behavior. Significant lateral transport to margins occurs, but the concentration of B is not uniform throughout the ocean. Consequently, the flux of B is neither uniformly equivalent to its production rate nor is it directly proportional to the flux of particulate matter.
While this simple illustration is primarily intended to help clarify the boundary scavenging concept, the similarity between 23°Th, 231pa and 1°Be and the tracers A, B and C respectively is not entirely unintentional. As pointed out by Bacon [12], the relative time scales of lateral mixing and of chemical scavenging of an element is an important factor governing the extent to which boundary scavenging influences the removal of the element from the ocean. Boundary scavenging exerts a greater influence over elements for which ~-> T than for elements with ~-< T.
Particle composition also influences scavenging intensity. Manganese oxide coatings on particles [1,6], biogenic opal [15], and clays [54] have been implicated as phases having a pronounced influence over the scavenging of 231pa and 1°Be. Although boundary scavenging enhances the deposition of both ~°Be and 231pa at margins compared to central ocean gyres, the influence of particle composition on the scavenging of 231pa and 1°Be seems to be different. For example, whereas metalliferous sediments on the EPR (TT154-10, V19-55) have 231pa/23°Th ratios approaching those of the margin sediments (Table  3), the l°Be//23°Th ratio of the metalliferous sediments is only about 50% greater than the ratio in our other "open-ocean" sites listed in Table 1. Our results for metalliferous sediments are consistent with laboratory studies [57] showing that the partition coefficient (KD) for adsorption of Be a Decay-corrected to time of deposition by extrapolating profiles to nodule surface.
onto marine particles does not depend on the MnO 2 content of the particles and with field studies [6] showing a strong dependence of Kt) for Pa on the Mn content of marine particles. This is also consistent with published results for manganese nodules. Whereas Mn nodules preferentially incorporate Pa relative to Th [58], the l°Be/23°Th ratio at the surface of Mn nodules (Table 6) tends to be less than or equal to the average value for deep-sea sediments. The one exception (TDK 40) is a shallow water nodule from an ocean-margin region. While particle composition may influence the nature and intensity of boundary scavenging, metal oxide coatings probably contribute little, if at all, to the boundary scavenging of l°Be. Whether scavenging of l°Be is predominantly controlled by the flux of clays, as suggested by Southon et al. [54], or by some other factor, remains to be determined.

Conclusions
Enhanced scavenging at ocean margins leads to a preferential deposition of l°Be in ocean-margin sediments, l°Be is scavenged from seawater and deposited in ocean-margin sediments at rates at least an order of magnitude greater than in the open ocean, l°Be shows an even greater enhancement of deposition at margins than occurs for 231pa. However, the processes responsible for boundary scavenging of l°Be and 231pa may be different. Whereas metalliferous oxyhydroxides strongly enhance the scavenging of Pa, they have only a small effect on Be.
Because the rate of l°Be deposition in marine sediments varies significantly from one region to another, it is impossible to define an ocean-wide average rate of 1°Be supply by measuring l°Be accumulation rates in a few cores. An average flux of 1°Be to the ocean can only be obtained by first examining a number of sites representative of different major oceanic environments, and then by normalizing l°Be contents of the sediments at those sites to a tracer with a known source, such as 23°Th, to eliminate biases due to physical transport of sediments, l°Be deposition rates in the Pacific Ocean range from 0.4 to 7.3 × 10 6 at cm -2 a -1 among the sites examined here. Using these deposition rates and an average concentration of l°Be in Pacific Ocean water to estimate inventories, we calculate regional residence times for l°Be in the Pacific ranging from -100 years at ocean margins to > 1000 years in deep open-ocean regions.
Production of l°Be is proportional to the cosmic ray flux, as modulated by solar activity and the Earth's magnetic field. There has been widespread interest in trying to use l°Be deposition records to obtain an indirect measure of how one or more of these parameters (cosmic ray flux, magnetic field strength, solar activity) have changed with time. Several investigators have examined marine sediment records for evidence of temporal variability in the rate of l°Be accumulation, from which variability in the rate of l°Be production could be inferred [53,54,[59][60][61][62][63][64][65][66]. With as much as 50% of the X°Be supplied to the oceans being removed at margins, a major change in the nature of boundary scavenging could lead to a significant redistribution of 1°Be burial within the ocean. A change in a local rate of l°Be deposition in marine sediments could reflect such a redistribution rather than a change in the global rate of l°Be production. Consequently, temporal variability in the rate of l°Be production can never be evaluated unambiguously from the l°Be deposition record in a single core, or even from the records of a few cores. Before temporal variability in the global rate of l°Be production can be assessed from marine sediment records, we will need a better understanding of the processes that lead to the preferential removal of 1°Be at ocean margins and of how these processes have changed with time.