Transport and burial rates of 10Be and 231Pa in the pacific-ocean during the holocene period

An ocean-wide study of the rates of removal of10Be and231Pa in the Pacific Ocean has identified intensified scavenging of the10Be and231Pa in several ocean margin areas, including the Northeastern and Northwestern Pacific, the Bering Sea, the Eastern Equatorial Pacific and the South Pacific Ocean. Scavenging rates of10Be and231Pa are clearly correlated to particle flux. Principal component analysis further suggests that scavenging of10Be and231Pa may be related to opal productivity in surface waters. A simple box model was constructed to partition the deposition of230Th,231Pa and10Be between open ocean and ocean margin sediments. Model parameters were constrained using measured values of230Th and231Pa, which have a common source, and then applied to10Be. An average Holocene10Be deposition rate for the entire Pacific Ocean is estimated to be ∼ 1.5 × 106 atoms/cm2 yr−1, with ∼ 70% of the total10Be supplied to the Pacific being deposited in margin sediments underlying only 10% of the ocean. The short residence times of10Be in ocean margin regions (from < 100 to ∼ 200 yr) compared to the long10Be residence time in the central open Pacific Ocean ( ∼ 1000 yr) reflects the intensified scavenging of10Be in ocean margin waters. The results of this study suggest that the Pacific Ocean acts as a relatively closed basin with respect to the transport and burial of10Be; therefore, the average10Be deposition rate in the Pacific Ocean can be used as an estimate of the global average production rate of10Be in the atmosphere during the Holocene period.


I. Introduction
l°Be (tl/2 = 1.5 × 106 yr) is produced in the upper atmosphere by spallation of oxygen and nitrogen atoms by galactic cosmic rays [1]. Many applications of l°Be as a geochemical tracer require the atmospheric production rate of l°Be to be accurately known (see [2] for a review). One of the ways of estimating the global average 1°Be production rate involves measurement of l°Be in deep-sea sediments. Initial estimates of the deposition rate of l°Be from deep-sea cores in the open Pacific Ocean were in close agreement with theoretical calculations, ranging mostly from about 0.4 to 1.2× 106 atoms/cm 2 yr -1 [3][4][5]. Studies of deep-sea cores in the open Atlantic Ocean [6,7] yielded a rate of ~ 0.6 × 106 atoms/cm 2 yr-1 which is on the lower side of the theoretical calculations. As more and more results emerged, tremendous differences among l°Be deposition rates at different locations in the oceans became apparent. A deposition rate of 8.3 x 106 atoms//cm 2 yr-1 was detected in sediments accumulated off Africa [8]; about the same rate of 11 × 106 atoms/cm 2 yr -1 is derived from l°Be results in sediments off the coast of California [9] and in Northwest Pacific margin sediments [10]. The highest l°Be deposition rate (~ 60 × 106 atoms/cm 2 yr -1) was observed in the Zaire deep-sea fan in the Angola Basin [11], while an extremely low deposition rate (~ 0.02 × 10 6  TABLE 1 Locations, sediment types and sedtmentation rates of the cores in the Pacific Ocean. i The se&mentation rates for most of the cores were from literature (see detailed hst of references in [26]). Those for the following cores were based on this work: for the Point Sur cores (BCl16, 133, 150, and 151) they were estimated to be the same as that of several cores in a nearby area dated by the 14C method (C. Reimers, unpublished data); San Clemente QP2 was dated by 14C counting of total carbon extracted from the core; for V18-299, tt was obtained by plotting unsupported 23°Th and 231pa against depth of the core; for RC8-81, it was uncertain and tentatively assigned as 0.2 cm/kyr (close to those of red clay cores m the North Pacific) based on the great water depth and the red clay content of the core.
atoms/cm 2 yr -1) was reported in the Arctic Ocean [12]. It is obvious that the deposition rate of l°Be at a single site, or even an average of a few sites in the ocean, can not provide a reliable estimate of the global average production rate of l°Be. A more comprehensive understanding of the marine geochemistry of l°Be is required before its deposition rate in the ocean may be used as an estimate of its global average production rate.
Intensified scavenging in ocean margin areas (i.e., boundary scavenging) was shown to greatly influence the removal of particle-reactive chemical substances such as 21°pb [13][14][15][16], Pu isotopes [17] and 231pa [18][19][20][21][22] in the ocean. The results of previous studies [9,10] and our work [23][24][25] have shown that l°Be is also preferentially removed from the oceans in some margin areas. In order to evaluate the source of l°Be, and to understand its transport within the oceans, we need to conduct an ocean-wide study: (1) to estimate the extent to which l°Be deposition is enhanced in different ocean margin regions; and (2) to examine what factors influence the scavenging of l°Be from seawater to sediments. After the behavior of l°Be in various marine environments is well understood, we will be in a better position to use l°Be as a geochemical and geophysical tracer. This paper presents an overview of Pacific-wide boundary scavenging of 1°Be and 231pa during the Holocene period (i.e., the last 10,000 yr).

Core selection and results
Sediments from deep-sea cores in the Pacific were analyzed for U, 230Th, 231pa and l°Be nuclides and major and trace elements [26]. 23°Th and 231pa are radioactive nuclides (tl/2 = 75,200 and 32,500 yr, respectively) which decay with time. Therefore, it is necessary to make decay corrections on cores that have age information to obtain the initial unsupported concentrations for 230Th and 231pa (i.e., the fractions of 23°Th and 231pa that are only produced by decay of dissolved 234U and 235U, respectively, in the water column). Since the half-life of 1°Be is relatively long (1.5 × 10 6 yr) compared to that of 230Th and 231pa, it is unnecessary to make decay corrections for 1°Be for the purpose of studying deep-sea sediments accumulated during the last 10,000 years. Cores that had stratigraphic controls such as lSo, 14C and 13C isotope results were preferred.
In some areas where stratigraphic information was unavailable but otherwise useful cores could be obtained, the sedimentation rates of the cores were estimated in two ways. One was to measure downcore unsupported 2a°Th and 231pa concentrations and then construct an average sedimentation rate for the core of interest. The second was by comparing the core of interest to nearby cores for which stratigraphic information was available. The principal criteria for core selection were that the chronologies of the cores should be of good quality, representative of a variety of marine environments, and readily accessible [26]. Thirty-two cores that were believed to best serve our purposes were selected (Table 1, Fig. 1).
Five cores (V20-122, RC14-105, V21-146, V32-126 and V32-128) were chosen from the Northwest Pacific where work has been carried out to study eolian dust input from Asia [27,28]. A vast area in the central North Pacific gyre region is covered by red clay sediment which originates from eolian dust, and sedimentation rates in such an area are usually < 5 mm/ky [29,30]. It is important to know what role the area of very low sedimentation plays in Zalpa and l°Be deposition. Four cores (V20-85, V20-88, MANOP Site R and Site S) that have sedimentation rates of~ 2 mm/ky were thus included. Intensified scavenging of Z31pa (relative to 2a°Th) was found in the Northeastern Pacific [20] where we selected eight Cores were also selected to include sediments rich in Mn and Fe oxides, opal and CaCO3, which may help to examine the effect of particle composition on scavenging of the nuclides. Enhanced scavenging of 23°Th and 231pa was suggested to be related to Mn coating of particles in the East Pacific Rise [31] where two cores rich in Mn and Fe were chosen (TT154-10 and V19-55). The major phase scavenging l°Be from water column to the sea floor has been postulated to be aluminosilicate rather than carbonate [6,32]. The relatively high carbonate content in sediments in five cores (V28-238, RCll-210, MANOP Site C, V19-28 and V19-29) along the equator may allow us to test this hypothesis. Scavenging of 231pa was suggested to be enhanced by high opaline silica flux [33,34]. A role for opal in the scavenging of l°Be has also been inferred from the similarity of Be and Si profiles in sediment pore waters [35]. Since opal is one of the two most important biogenic components (the other being CaCO 3) in marine sediments in terms of quantity, the influence of opal flux on scavenging of 231pa and l°Be is worth close evaluation. In addition to one core in the Bering Sea (RC14-121) where the Holocene sediment contains a very high diatom content [36], we included two cores (E15-6 and E17-9) in the South Pacific (near the Antarctic) where the opal deposition rate is very high [37,38].
The techniques used for analyses of the radionuclides have been described in detail elsewhere [24]. The initial (i.e., decay corrected) unsupported concentrations of 23°Th and 231pa (des-230 231~ ignated as xs Tho and xs rao, respectively; Table  2), were calculated based methods described previously [23,24]. The basic principle of our approach is to use 23°Th, a nuclide that is produced uniformly throughout the ocean by decay of dissolved 234U and deposited to the sea floor at a rate nearly equal to its production rate in the overlying water column, as a tracer against which 231pa and l°Be are normalized so that 231pa/Ea°Th and l°Be/23°Th ratios in the sediments can be used as indicators of the intensity of scavenging of 231pa and l°Be (see [23] for a description of the normalization and the assumptions implicit therein; and also discussions in Section 3 below).
For the sake of convenience, "Pa/Th" and "Be/Th" will be used throughout the text to designate the ratios of xs(Ealpa/23°Th)o and l°Be/23°Th respectively.

Intensified scavenging of 231pa and l°Be in ocean margins
The 231pa/23°Th production ratio in the water column is known to be a constant value (0.093) because of the fixed ratio of 235U to 234U (the progenitors of 231pa and 23°Th, respectively) in the ocean. Since the rate of removal of 23°Th from seawater is relatively uniform throughout the ocean [18,19,22], a Pa/Th ratio in the sediment greater than the production ratio of 0.093 indicates that the site is a sink for 231pa (relative to 23°Th), receiving an extra amount of 231pa in addition to the 23apa produced in the overlying water column. Conversely, if the ratio is less than the production ratio, the site acts as a source with a certain fraction of 231pa produced in the overlying water column being laterally exported (relative to 230Th) to other area(s) in the ocean [23,24]. Similar to Pa/Th ratios, Be/Th ratios can be used as indicators of the intensity of scavenging of l°Be relative to 23°Th [23].
Notes to Table 2: 1 The errors include propagation of lo-counting statistics. Sources for the ages are given in [26] 2 The core-top age with a question mark indicates that the mixed layer depth of the core was not known, therefore an age was assigned. 3 Indicates duplicates of the same sample. 4 There was not enough BeO for good l°Be measurement due to loss during handling 5 The samples were taken from different secUons at the same depth level, rather than duplicates of the sample. For most of the cores, two samples were taken from the Holocene section for the radiochemical analysis. Where this is the case the averages of the two Pa/Th and Be/Th ratios are used to represent the Holocene results. As will be demonstrated below, the overall results indicate that both 231pa and 1°Be are preferentially removed to sediments in ocean margin regions (as shown by higher Pa/Th and Be/Th ratios), i.e., boundary scavenging has a great influence on removal of these two nuclides from the ocean.
The lowest Pa/Th ratio (0.03) among the open ocean sites is at MANOP Site S whose sedimentation rate is < 0.1 cm/ky [39]; the Pa/Th ratios in some margin areas are around 0.2 ( Fig. 2): two times higher than the production ratio, clearly showing a pattern of boundary scavenging. On average, the Pa/Th ratio in ocean margin sediments (~ 0.15 to 0.2) is about 4-5 times that in areas of red clay accumulation in the open ocean (~ 0.03-0.04). The pattern of boundary scavenging of 231pa (relative to 23°Th) in this study is in agreement with published results summarized in [20] and our data fill in the Eastern Pacific where some of the 231pa concentrations in [20] were inferred.
Similarly, the pattern of boundary scavenging of 1°Be is shown by higher Be/Th ratios in margin sediments (Fig. 3). The very low Be/Th ratios in the pelagic sediment in the open ocean in this study are in agreement with the Be/Th ratios measured by other investigators in the same type of sediment (see + symbols in Fig. 3). The lowest Be/Th ratio (0.03 × 109 atoms/dpm) in this study is also at MANOP Site S. In margin areas, the Be/Th ratios range from 0.3 to 1.5 × 109 atoms/dpm. The range in the Be/Th ratio between ocean margin sediments and deep open ocean red clay sediments exceeds a factor of 10, i.e., approximately twice the range seen for Pa/Th ratios.
One core (PS BC133 from the coastal area of California) has exceptionally high ratios of Pa/Th (average 0.85) and Be/Th (average 14.2 × 109 atoms/dpm; Table 2). The very high contents of Fe and K (20% and 4.2%, respectively [26]) of this core suggest that the samples from this core might be rich in glauconite, a mineral that usually forms in shallow margin areas [40] and has been reported to be present in sandy sediments in this area, with a maximum concentration of 35% of the total sediments [41]. Since such a result is so rare in our data set, we do not consider it representative of average ocean margin environments, but it apparently suggests the preferential uptake of 1°Be and 231pa, relative to 23°Th, by the Fe-rich minerals in these sediments.
The ocean cores may be more easily seen in Fig. 4, in that the cores are so arranged that those in the upper part of the figure are from open ocean areas (i.e., sedimentation rates are relatively low) and those in the lower part are from margins (i.e., sedimentation rates are relatively high). A general correlation is seen between sedimentation rate and the Pa/Th ratio (Fig. 4a), and between sedimentation rate and the Be/Th ratio (Fig. 4b).
Also shown in the figure are the ratios of Be/Th normalized to water depth ([Be/Th]r~; Fig. 4b).
The vertical distribution of dissolved 1°Be in the open Pacific exhibits a depletion in surface waters (~ 600-1000 atoms/g) and enrichment to relatively constant values at depth (~ 2000 atoms/g), which is suggested to reflect scavenging of 1°Be from the surface waters followed by regeneration at depth [42]. The production of 230Th is directly proportional to water depth and the source of 23°Th is, therefore, greater at deeper sites. Consequently, if 23°Th is removed through- out the water column [18,19,24] and 1°Be is only removed from surface waters [42], then, among the sites of similar scavenging intensity, the Be/Th ratios of sediments in shallow ocean areas will be higher than that in deep ocean areas. For the purpose of comparing 1°Be scavenging rates at different sites in the ocean, Be/Th ratios should, therefore, be normalized to a constant water depth. The mean depth of 4200 m of the Pacific [43] was chosen against which the Be/Th ratios were normalized: [Be/Th]~v = (Be/Th) x Z/4200 (1) where (Be/Th) is the measured ratio and Z is the water depth (in meters) above the core. Normalization of the Be/Th ratio to a standard water depth leads to a minimum estimate for the extent to which l°Be is influenced by boundary scavenging. Implicit in this approach is the assumption that 1°Be is scavenged only from surface waters [42]. This may not be valid in ocean is the mean depth of the Pacific adopted from [43]). The results of PS BC133 are not included (see text). margin regions where fluxes of ~°Be collected by sediment traps indicate that scavenging takes place throughout the water column [24], in which case the depth normalization will underestimate the actual enhancement of ~°Be deposition. Another factor which causes estimates of the extent of boundary scavenging of 23~pa and 1°Be, based on Pa/Th and Be/Th ratios, to be minimum values is the assumption that the deposition rate of 23°Th is everywhere equal to its production rate in the overlying water column. Boundary scavenging also exerts some influence on 23°Th [24], so that the actual enhancement of the deposition of 231pa and 1°Be in ocean margin sediments should be obtained by multiplying the Pa/Th and Be/Th ratios by the extent to which 23°Th deposition is enhanced at margins (note that enhancement of 23°Th deposition at margins is generally unknown; see modeling results in Section 5). Yet, despite these conditions, which lead to minimum estimates of enhanced scavenging at margins, the depth normalized Be/Th ratios ([Be/Th] N) still span a full order of magnitude (Fig. 4b), providing unequivocal evidence for greatly enhanced scavenging of 1°Be in ocean margin areas.

Factors influencing scavenging
A regional study in the Northeastern Pacific found that fluxes of 23°Th, 231pa and 1°Be are proportional to particle flux, both with respect to temporal variability at a single location and with respect to mean particle flux along a transect normal to the coastline [24]. It remains to be tested, however, if a general relationship exists between the nuclide scavenging rate and particle flux on a basin-wide scale.
Nine cores (core sites 3, 4, 5, 6, 7, 8, 10, 11, and 12, Fig. 1) along a transect across the North Pacific at ~ 40°N provide a useful subset with which to test the influence of particle flux on scavenging intensity on a basin-wide scale. Sediments across the North Pacific are predominantly lithgenic material transported by westerly winds from source regions in Asia [27,28]. Dust flux, and hence sediment accumulation rate, generally decreases from west to east in the open ocean. As there is no information on particle flux in the water column at the sites where the cores in this study were collected, sedimentation rates may be regarded as the best proxy. Patterns of Pa/Th and [Be/Th]r~ ratios both closely mimic the pattern of sedimentation rate along the 40°N transect (Fig. 5). As has been discussed previously, Pa/Th and [Be/Th] N ratios vary according to the scavenging intensities of 231pa and 1°Be. Therefore the close relationships between sedimentation rate and Pa/Th and [Be/Th] N ratios suggest that particle flux is a major factor influencing scavenging of 231pa and 1°Be, consistent with the findings of an earlier study of 1°Be by Tanaka et al. [44]. However, there are exceptions when the results of all the cores are considered. The deviation from the correlation between sedimentation rate and the Pa/Th and [Be/Th] N ratios could be partly due to errors in the sedimentation rates of the cores, which are difficult to assess at the moment because the chronologies of these cores are mostly taken from literature. Sediment composition, which varies among cores, may also affect the scavenging intensity (Fig. 4). Principal component analysis was applied to the elemental results of the sediments (the element concentrations and mathematical procedures were given in [26]) to look for systematic relationships between sediment composition and the nuclide contents. The factor analysis results indicate that the elements may be divided into three groups (Fig. 6): (1) those normally associated with lithogenic phases (AI, Ti, Fe, K, Mg, V and 232Th); (2) those associated with biogenic carbonates (Ca, Sr and Zn);  (3) 230Th, 231pa ' lOBe (designated by ,~Th, xsPa and Be), Ba, Cu and Mn.
In a broader sense, Na, Ni and U seem to be clustered around the lithogenic elements. 23°Th, 231pa and 1°Be are located far away from Ca and Sr in Fig. 6, which is consistent with the hypothesis that marine carbonates may act as diluting phases and do not contribute significantly to nuclide scavenging [6,32]. Distributions of some trace elements (i.e., Ba, Cu and Mn) may be related to biological productivity, albeit for different reasons. Diagenetic enrichment of Mn in sediments usually occurs in areas of high biological productivity where organic input to the sediments is high (e.g., in the Northeastern Pacific; [24,45]). Cu chemistry in the ocean seems to be substantially influenced, and often dominated, by organic complexation [46,47]. The sedimentary distribution of Ba, an element known to be associated with biogenic opal [48], has been proposed as a paleoproductivity indicator [e.g., 49,50]. The association of Ba with 23°Th, 231pa, 1°Be, Cu and Mn (Fig. 6) therefore suggests that scavenging of these nuclides and elements may be related to biological productivity in surface waters.
Comparing results from two end member locations supports the inferences based on factor analysis. Two cores from the South Pacific (E15-6 and E17-9) very rich in opal content (60-80%; [37,38]) serve as an end member to examine the effect of opal flux on scavenging of 231pa and l°Be. MANOP Site R in the central North Pacific is treated here as a pure red clay end member. The average concentrations of initial unsupported 231pa and 1°Be (~ 3.4 dpm/g and ~ 5.7 x 10 9 atoms/g, respectively) in MANOP Site R are close to those (~ 2.4 dpm/g and ~ 6.0 × 10 9 atoms/g, respectively) in E15-6 and E17-9 ( Table  2). The sedimentation rate of E15-6 and E17-9 (~ 3.6 cm/ky) is more than one order of magnitude higher than that of MANOP Site R (~ 0.2 cm/ky). Therefore, accumulation rates of 231pa and 1°Be in the opal-rich sediments of the South Pacific must be more than 10 times their respective accumulation rates in the red clay sediment of MANOP Site R in the central North Pacific, leading us to conclude that opal, as well as clay minerals, must be an important phase scavenging 231pa and l°Be.

Modeling distributions of Z31pa and 1°Be
One of our goals is to evaluate an ocean-wide l°Be deposition rate which, in turn, provides an estimate of global average production rate of 1°Be in the atmosphere. It has been demonstrated above (also see [23,24]) that the removal of Z31pa and l°Be is greatly influenced by boundary scavenging. Furthermore, lateral sediment transport (sediment focussing) can result in misinterpretation of 231pa and 1°Be fluxes in sediments [23]. Therefore, one can not reliably determine the ocean-wide 1°Be deposition rate by measuring its accumulation rate in only a few cores. A better approach is to develop a model that relates the burial of 1°Be to the burial of 23°Zh, whose source is precisely known. In the following, a simple boundary scavenging box model is constructed and the validity of input parameters is first tested using 23°Th and 231pa results.

231pa
If the deposition rate of 23°Th in sediments is exactly equal to its production in the water column, then the flux of nuclide N at a specific site, F(N), could be evaluated as [23,51,52]: (2) where [N/Th]~d is the concentration ratio of nuclide N to the initial unsupported 23°Th in the sediment; PTh is the production rate of 23°Th which is proportional to water depth (i.e., PTh = 0.0026Z dpm/cm 2 ky -1, where depth, Z, is in meters). However, F(23°Th) is not equal to PTh in the real world because boundary scavenging can to some extent affect 23°Th [22,24]. Since the mass budgets of 231pa and 1°Be will be constructed by normalizing to Z3°Th, the boundary scavenging effect on 23°Th must be taken into account. A model will have to be constructed in which open ocean and ocean margin regions are both considered (Fig. 7). When the two-box system is at steady state, the 23°Th deposited in margin sediment is derived both from that produced in the overlying water column, and from that imported from the open ocean. This mass balance requires: Rearranging: where Qo ( = eTh × Vo) and Q~ ( = eTh × Vm) are the rates of in situ production of Z3°Th (1 -fro) where fm = Vm/(Vo + Vm) is the ocean margin volume fraction. Assuming that any 23°Th and 231pa which are not scavenged from the open-ocean box are transported to the ocean margin box where they are homogeneously mixed with the 23°Th and 231pa produced in the waters in the ocean margin box, and then scavenged to the margin sediment, we then have: In modeling the scavenging of 23°Th, K o is first permitted to range from 0.85 to 0.95 based on modeling results in [22], and K m is calculated using eqn. (5), assuming that the fraction of the ocean margin volume is 10% of the total Pacific Ocean (i.e., fm = 0.1). The computed K m values range from 1.45 to 2.35 (Table 3), very close to the flux/production ratio for 23°Th (i.e., the observed K m) for sediment trap samples and Holocene sediments (from 1.7 to 2.5) at a nearshore_~ite in the Northeastern Pacific [24]. Thus, the assumed values for K o (0.85-0.95) lead to computed K m values that are consistent with the best estimates of K m available from measured 23°Th fluxes.
To test the parameters used to simulate boundary scavenging, an average value for [Pa/Th] m of 0.2 was used as model input based on the facts that the Pa/Th ratios in sediment trap samples from a nearshore site in the Northeastern Pacific are around 0.2 [24], and those in the margin sediments are relatively constant (~ 0.2; Fig. 4a). Equation (6) (Fig. 2). The parameters used (i.e., fm = 0.1, K o = 0.85-0.95, and K m = 1.45-2.35; Table 3) thus give a reasonable partitioning of 23°Th and 231pa between open ocean and ocean margin sediments, and will therefore be used for modeling the distribution of l°Be.

1°Be
The Pacific ocean-wide average 1°Be flux (Fae) in the two-box model (Fig. 7) is given by: where PTh is the average 23°Zh production rate in the Pacific (= 0.0026   [22] and constraints by the measured 23°Th fluxes in the sediment traps and sediments in the Northeastern Pacific [24]. ¢ Krn is the enrichment factor of ~°Th in the margin ocean sediment (1.e., the ratio of actual flux to the production rate in the overlying water column) calculated using eqn. ( (Table 4).
Since this is a Pacific-wide study, and factors such as boundary scavenging and sediment focusing that influence the deposition of 1°Be in the ocean have been taken into account, the average Holocene l°Be flux in the Pacific should reflect the global average production rate of l°Be in the atmosphere during the Holocene period. However, before such a conclusion is reached, possible exchanges of 1°Be between the Pacific and the Atlantic need to be considered in terms of the regional residence times of l°Be and the way of water exchange between oceans.
), and the relationship between K m and Fse (eqn. (7)). a = the relationship between K m and Fse for lower-limit values of Be/Th ratios in open ocean and ocean margin sediments (0 04X109 atoms/dpm and 0.4x109 atoms/dpm, respectwely), b = the same relationship for upper-limit values of Be/Th ratios in open ocean and ocean margin sediments (0.06x109 atoms/dpm and 0.6x109 atoms/dpm, respectively). The lower and upper limits for FBe permitted by the extreme ranges of reasonable values for the parameters are 1 0x 106 atoms/cm 2 yr-i, and 2.0 >< 106 atoms/cm 2 yr-i, respectively, as indicated by the solid arrows.

Regional residence times of Z°Be
The oceanic residence time of 1°Be is an important index for estimating the rate of removal of 1°Be in the ocean. The residence time G-) with respect to removal from seawater may be calculated using the following equation [23]: ~" = I/F(Be) (8) where I is the inventory of 1°Be in the ocean (in the unit of atoms/era 2) and is simply proportional to water depth if we adopt a mean 1°Be concentration in the deep Pacific (~ 1800 atoms/g [42]); F(Be) is the 1°Be deposition rate (atoms/era 2 yr-1) at a specific site as defined in eqn. (2).
The regional residence times of l°Be thus calculated do not take into account the effect of boundary scavenging on 23°Th because there is no independent assessment of the effect of boundary scavenging of 23°Th at individual sites. However, since we know that a certain fraction (e.g., ~ 10%) of 23°Th produced in the open ocean is removed to the margins, l°Be residence times calculated for open ocean areas should be regarded as lower limits (because the deposition rates of 1°Be are upper limits using eqn. (2)); and those in ocean margin areas should be upper limits (because the deposition rates of 1°Be are lower limits). The residence time of 1°Be ranges from < 100 yr in ocean margin areas to > 1000 yr in deep, open ocean with very low accumulation rates (Fig. 9). The trend is obvious: residence times in open ocean regions are generally longer and those in margin areas are shorter, with the latter being at least an order of magnitude lower than the former, which is a consequence of intensified scavenging in ocean margin areas.
Although the 1°Be residence time varies greatly from area to area in the Pacific, an average 1°Be residence time is useful in evaluating the behavior of l°Be in the ocean. Using an average l°Be flUX of 1.5 X 10 6 atoms/cm 2 yr-1 calculated from the above modeling, a mean l°Be concentration of 1800 atoms/g [42] and a mean depth of the Pacific of 4200 m [43], the calculated mean 1°Be residence time in the Pacific is ~ 500 yr, consistent with the recent estimate of Kusakabe et al. [53].  (Fig. 9) and in the Atlantic, the net change in the inventory of l°Be due to transport of 1°Be between the Atlantic and the Pacific is expected to be small because of the effects of high scavenging intensity in the South Pacific and a larger influence of boundary scavenging in the smaller Atlantic compared to the Pacific. Another cause of the small partitioning of l°Be between the Pacific and the Atlantic is the way of water exchange between the Pacific and the Atlantic. The advection flow is from surface North Atlantic to deep North Atlantic to Antarctic Circumpolar Currents to deep Pacific and to surface Pacific, and the surface Pacific water is exported to the Indian Ocean via the Indonesian Seas, and then around the southwest of Africa, into South Atlantic Ocean [55]. Considering that the 1°Be concentration in the surface Pacific water (~ 850 atoms/g [42]) is about the same as that in the deep Atlantic [54], mixing and exchange between the Atlantic and the Pacific oceans is unlikely to influence the partitioning of 1°Be between the two oceans significantly. This suggests that the Pacific acts as a relatively closed basin with respect to l°Be supplied from the atmosphere and the average flux of l°Be in the Pacific can be regarded as a good estimate of the global average production rate of l°Be.

Conclusions
Boundary scavenging plays an important role in removal from the ocean of both 231pa and l°Be. Particle flux appears to be the major factor influencing scavenging of 231pa and l°Be in the ocean. Principal component analysis of the chemical data suggests that biological opal productivity may also influence scavenging of 231pa and l°Be.
Modeling results indicate that about 70% of the l°Be directly supplied to the ocean from the atmosphere is accumulated in sediments in ocean margins, which constitute only 10% of the total Pacific volume. The regional residence times of l°Be in the Pacific range from < 100 yr in margin areas to > 1000 yr in the deep open ocean, with a mean residence time of ~ 500 yr.
The results of the modeling allow us to place the lower and upper limits for the range of the global average production rate of l°Be at 1.0 × 106 atoms/cm 2 yr -1 and 2.0 × 106 atoms/cm 2 yr -1, respectively, averaging (1.5 _+ 0.5) × 106 atoms/ em 2 yr -1.
The results of this study have several important implications: (1) 1°Be has been used to trace the cycling of ocean floor sediments at convergent tectonic plates (island arc volcanic rocks; e.g., [56,57]). The influence of boundary scavenging on the deposition of 1°Be in the specific areas of the ocean margins should be well understood before l°Be can be used to model the amount of sediment that was incorporated in the magmatic process accurately.
(2) The production rate of l°Be in the atmosphere reflects the intensity of the cosmic ray flux and the strength of the Earth's magnetic field. The approach described in this paper can be used to detect changes in t°Be production rate [58], from which information about cosmic ray flux and magnetic field strength in the past may be inferred.
(3) Application of 1°Be and 231pa to dating deep-sea sediments requires a knowledge of the source, transport and fate of l°Be and 231pa in the ocean. Possible changes in the nature and intensity of boundary scavenging of l°Be and 231pa may cause changes in deposition rates of l°Be and 231pa in deep-sea sediments at specific sites, thereby invalidating the assumption of a steady supply of l°Be and 231pa with time to the sediments. This assumption needs to be examined before a precise chronology can be developed for deep-sea cores based on measurements of l°Be and 231pa.