238U disequilibrium

ARTICLE IN PRESS

Deep-Sea Research I 53 (2006) 1742–1761 www.elsevier.com/locate/dsr

A study of particle exchange at the sediment–water interface in the Benguela upwelling area based on 234Th/238U disequilibrium Maik Inthorna,, Michiel Rutgers van der Loeffb, Matthias Zabelc a

SINTEF Petroleum Research, S.P. Andersens vei 15b, 7465 Trondheim, Norway Alfred-Wegener-Institute for Polar and Marine Research, Am Handelshafen 12, 27570 Bremerhaven, Germany c University of Bremen, FB5-Geosciences, Klagenfurter Street, 28359 Bremen, Germany

b

Received 15 August 2005; received in revised form 11 August 2006; accepted 14 August 2006 Available online 24 October 2006

Abstract The natural isotope 234Th is used in a small-scale survey of particle transport and exchange processes at the sediment–water interface in the Benguela upwelling area. Results from water and suspended particulate matter (SPM) samples from the uppermost and lowermost water column as well as the underlying sediment of three stations are compared. The stations are situated in different sedimentological environments at 1200–1350 m water depth at the continental slope off Namibia. Highly differing extent and particle content of the bottom nepheloid layer (BNL) are determined from transmissometer data. Three models are presented, all explaining the 234Th depletion of the BNL and 234 Th excess of the surface sediment that were observed. While the first model is based solely on local resuspension of surface sediment particles, the second evaluates the influence of vertical particle settling from the surface waters on the 234 Th budget in the BNL. The third model explains 234Th depletion in the BNL by sedimentation of particles that were suspended in the BNL during long-range transport. Particle inventory of the BNL is highest at a depocenter of organic matter at 25.51S, where strong deposition is presently taking place and lateral particle transport is suggested to predominate sediment accumulation. This is supported by the high settling flux of particles out of the BNL into the sediments of the depocenter, exceeding the vertical particle flux into sediment traps at intermediate depth in the same area by up to an order of magnitude. High particle residence/removal times in the BNL above the depocenter in the range of 5–9 weeks support this interpretation. Comparison with carbon mineralization rates that are known from the area reveals that, notwithstanding the large fraction of advected particles, organic carbon flux into the surface sediment is remineralized to a large extent. The deployment of a bottom water sampler served as an in situ resuspension experiment and provided the first data of 234Th activity on in situ resuspended particles. We found a mean specific activity of 86 disintegrations per minute (dpm) g
1 (39–339 dpm g
1), intermediate between the high values for suspended particles (in situ pump: 580–760 dpm g
1; CTD rosette: 870–1560 dpm g
1) and the low values measured at the sediment surface (26–37 dpm g
1). This represents essential information for the modeling of 234Th exchange processes at the sediment–water interface. r 2006 Elsevier Ltd. All rights reserved. Keywords: Nepheloid layer; Suspended particulate matter; Sediment–water interface; Thorium; Resuspension; Lateral transport; Regional index terms: Atlantic; Benguela; Namibia

Corresponding author. Tel.: +47 73591257.

E-mail address: [email protected] (M. Inthorn). 0967-0637/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.dsr.2006.08.004

ARTICLE IN PRESS M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

1. Introduction Increasing attention has been paid to lateral particle transport in subsurface nepheloid layers at continental margins. Lateral transport can decouple areas of high surface-water productivity from centers of organic matter burial in the sediments and influences quality and preservation of organic matter (Biscaye and Anderson, 1994; Jahnke et al., 1990). This has serious implications for organic matter flux models, which often assume vertical particle settling through the water column when using chlorophyll distribution of the surface waters derived from satellite images and sediment trap data to estimate carbon burial rates in the sediments (Seiter et al., 2005; Usbeck et al., 2003; Wefer and Fischer, 1993). Quantitatively important transport processes are assumed to occur in the bottom nepheloid layer (BNL), where concentrations of suspended particulate matter (SPM) are enhanced up to several hundreds of meters above the seafloor (Bacon and Rutgers van der Loeff, 1989; McCave, 1986). In the BNL, particles can be transported laterally over large distances by cyclic sedimentation and resuspension (Thomsen and van Weering, 1998). Because of long oxygen exposure time, aggregation, disaggregation, and associated microbial degradation processes, the organic fraction of the SPM alters continuously (Boetius et al., 2000; Thomsen and McCave, 2000). Nevertheless, studies that quantify particle transport in the BNL and exchange with the surface sediment are still few. An appropriate tool to assess these processes on a short time scale is the radioactive nuclide 234Th (e.g., Aller and DeMaster, 1984; Bacon and Rutgers van der Loeff, 1989; Muir et al., 2005; Rutgers van der Loeff and Boudreau, 1997; Santschi et al., 1999). It is produced from decay of 238U in seawater and has a half-life (t1/2) of 24.1 d (decay constant l: 0.0288 d
1). Compared to its parent, 234Th is very particle reactive and therefore a good tracer for particle dynamics in the water column on a time scale of weeks up to about 100 d. When no 234Th is removed from a given parcel of water by processes other than radioactive decay, total 234Th (dissolved+particulate, 234Tht) is in secular equilibrium with 238U, meaning that 234Th decay is balanced by production from 238U. This is the case in most of the deep oceanic water column below a depth of 200 m. However, in waters with throughput of freshly produced or old, resuspended particles, scavenging

1743

causes depletion of dissolved 234Th with respect to its parent (234Thdepl ¼ 238U–234Tht). From this disequilibrium, transport rates of the particulate and dissolved phases relative to each other can be derived. Because of the fast decay of 234Th, it is also well suited to the estimation of particle exchange between the sediment and the lowermost water column (Rutgers van der Loeff et al., 2002; Turnewitsch and Springer, 2001; Waples et al., 2006) and to the determination of bioturbation rates in the surface sediments (Aller and DeMaster, 1984). The Benguela upwelling area is the most productive eastern boundary current system of the ocean (Carr, 2002). For this area, Inthorn et al. (2006a) report predominant significance of lateral particle transport relative to vertical settling based on 14C ages of 1800–3500 yr and qualitative properties of surface sediments and suspended particles from the BNL. This confirms results of previous studies in the same area by Giraudeau et al. (2000) and Mollenhauer et al. (2003). The transport takes place in intermediate and BNLs extending from the highly productive inner shelf towards a major depocenter of organic matter at the upper to central slope off Namibia (Inthorn et al., 2006b). Here we provide 234 Th data from the Benguela upwelling area using a set of water and sediment samples from three stations at about 1300 m water depth on the Namibian continental slope (Fig. 1). One station is directly above the depocenter of organic matter, while the two other stations are north and south of it, where organic carbon (Corg) contents are

Fig. 1. Bathymetric map of the southwest African continental margin displaying the positions of the stations sampled for 234Th (solid circles) and 14C (open circles). Main surface and subsurface currents are indicated by solid black and dashed gray arrows, respectively.

ARTICLE IN PRESS M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

1744

significantly lower. The results of this small-scale study are the first data set to quantify lateral transport processes in the BNL of the Benguela upwelling area. They provide insight into the relative significance of vertical and lateral particle transport, as well as Corg remineralization. Additionally, specific 234Th activities (234Thspec, as 234Th activity per unit mass of dry particles) of different particle types and from an in situ resuspension of sediment-surface particles are determined that will help to calibrate existing models for 234Th exchange at the sediment–water interface. 2. Methods 2.1. Sampling On the M57/2 expedition of R.V. Meteor from 27 February to 8 March 2003, three stations on the continental slope off Namibia (between 241240 and 261440 S, 1231–1353 m water depth; see Fig. 1 and Table 1) were sampled for 234Th analysis. At every station, 20 l of surface water were obtained from the ship’s seawater supply (7 m water depth). A total of 20 l of water from 10, 40, and 100 m above the seafloor were obtained with a Hydro-Bios 12-bottle rosette sampler (CTD) with 5-l free-flow water sample bottles (Hydro-Bios) and by closing four bottles at a time. The device was equipped with an SBE 911+CTD for salinity, temperature, and pressure deteminations and a bottom sensor. Bottles were closed after bottom contact during the heave of the instrument; pressure differences were used to determine depth above seafloor as accurately as possible (71 m). Additionally, a bottom water sampler (BWS) was used to retrieve water from distinct depth intervals above the seafloor. The BWS consists of a threefooted steel frame tethered to the ship’s line via a swivel. Both the frame and the inner central axis have current-sails and an acoustic bottom sensor.

This permits a waiting time just before deployment of the device at the seafloor and ensures an untwisted wire connection and best possible alignment with the bottom currents. At 0.25, 0.4, 0.6, 0.9, and 1.1 m above the seafloor, 5-l free-flow water sample bottles (Hydro-Bios) are horizontally attached to a central axis. These sampling bottles were closed about 30 min after the deployment of the device. At stations GeoB8463 and GeoB8467, SPM was additionally collected with a McLane in situ pump attached to the BWS. The in situ pump was equipped with a 143 mm pre-weighed Nuclepore polycarbonate filter, and 160–170 l of water were filtered in about 60 min of pumping time. The pumping was programmed to begin about 20 min after the deployment of the BWS. Unfortunately, the pump did not work properly at station GeoB84101. The BWS was equipped with an 0.25 m pathlength Seatech transmissometer measuring beam attenuation by suspended particles over the full deployment, including the descent through the water column. At GeoB8463 and GeoB84101, surficial sediment was sampled for 234Th analyses with a multicorer system. Regrettably, surface sediments of station GeoB8467 were not sampled because of limited ship time. Surface sediments at stations GeoB8418 and GeoB8451 were sampled for radiocarbon dating. These stations are located at about 1000 m water depth at 24.251S and 25.51S, less than 10 km from stations GeoB84101 and GeoB8463, respectively (Table 1 and Fig. 1). 2.2. Water samples Water samples were treated according to the method of Rutgers van der Loeff and Moore (1999). Particles were removed for determination of particulate 234Th (234Thp) by filtration through 143 mm nuclepore polycarbonate filters with a pore size of 0.4 mm with an air pressure pump and a pressure of

Table 1 Positions of the stations that were sampled for 234Th and 14C analysis during M57/2 with the ship’s seawater supply (SW), a CTD rosette (CTD), the bottom water sampler (BWS) with the in situ pump (IP) and a multiple corer system (MC) Type

Station no.

Latitude (1S)

Longitude (1E)

Devices

Depth (m)

234

GeoB84101 GeoB8463 GeoB8467 GeoB8418 GeoB8451

241 24.00 251 36.70 261 44.80 241 21.90 251 28.90

131 02.50 131 16.90 131 25.30 131‘08.20 131 21.60

SW, CTD, MC, BWS SW, CTD, MC, BWS, IP SW, CTD, BWS, IP MC MC

1231 1335 1353 1006 1028

Th Th 234 Th 14 C 14 C 234

ARTICLE IN PRESS M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

0.5 bar directly after sample retrieval. All filters were pre-weighed on a balance equipped with a polonium static eliminator to minimize electrostatic charging of the polycarbonate. In the filtrate, dissolved 234Th (234Thd) was precipitated quantitatively with MnO2 and filtered. All filters were folded in the same fashion, dried, and kept in small plastic boxes for transport. Because of the comparatively small amount of water sampled with the BWS bottles, only 234Tht was determined in these samples by precipitating MnO2 directly in the unfiltered sample. By subsequent filtration, particulate and dissolved 234Th were recovered as 234Tht on one filter. Filters from the in situ pump were used to collect corresponding 234 Thp from the lowermost water column and dried and folded as described above. Beam attenuation (c) profiles were used to derive SPM values for those depths where 234Th-samples were taken from the BNL and the surface nepheloid layer (SNL), which corresponds to the particle-rich euphotic zone. This was done according to calibration algorithms derived from a set of particle samples and the measurement of corresponding beam attenuation profiles on the Namibian continental margin during M57/2 (Inthorn et al., 2006b), following procedures described by Gardner et al. (1993): SNL : SPM ½mg l
1  ¼ 1:13 c ½m
1 
0:01 ðR2 ¼ 0:74; n ¼ 20Þ,

ð1Þ

BNL : SPM ½mg l
1  ¼ 1:77 c ½m
1 
0:37 ðR2 ¼ 0:65; n ¼ 14Þ.

ð2Þ

Because of the high particle load on BWS and in situ pump filters compared to the CTD samples, SPM in the former samples could be determined directly by reweighing of the filters. For the BWS bottle samples, the weight of the MnO2 particles was approximated from the average weight of MnO2 on particle-free filters and subtracted. This procedure results in a relative error of up to 35% for SPM determination on the BWS filters with lowest particle load (Table 2). 2.3. Sediment samples The sediment cores were sampled at high resolution (one sample every 2 mm in the uppermost centimeter, increasing sample distances up to 0.8 mm further below; see Fig. 4) and treated

1745

according to Rutgers van der Loeff et al. (2002). Approximately 1 ml of sediment was resuspended in a few milliliters of distilled water. Subsamples were filtered through preweighed 143 mm nuclepore polycarbonate filters and handled like the particle filters described above. After drying, filters were reweighed to determine sample weight. Radiocarbon analyses of bulk organic carbon was performed with an Accelerator Mass Spectrometer at the Leibniz Laboratory for Radiometric Dating and Isotope Research (University of Kiel, Germany) on sediment from stations GeoB8418 (KIA 22694-22697) and GeoB8451 (KIA 2270122705; precision: 1%). Sediment accumulation rates were obtained by linear interpolation between the 14 C age of the sediment surface sample and a deeper sample at 10–20 cm. 2.4.

234

Th measurement

All 234Th samples were measured with a gas-flow anti-coincidence beta counter with 10 cm lead shielding (Risø National Laboratory, Roskilde, Denmark) at the AWI Bremerhaven within 10 weeks after sampling. The count rates of the samples were in the ranges of 0.49–2.42 counts per minute (cpm) for the particulate CTD samples, 1.80–6.55 cpm for the CTD samples of dissolved 234 Th, 1.82–17.02 cpm for the BWS samples, 5.15–8.20 for the in situ pump samples and 2.29–7.49 cpm for the sediment samples. All these values are well above the detector background of 0.18 cpm as determined from blank filters. For the particle filters, determination of background count rates from other beta emitters was done 9 months after the cruise according to Rutgers van der Loeff and Moore (1999). Calibration was performed with spiked filters (Rutgers van der Loeff and Moore, 1999). For the CTD samples 234Tht was calculated as the sum of corresponding 234Thp and 234Thd values. 238U concentration in the seawater was calculated from salinity after Chen et al. (1986). The propagated measurement error (including sampling and counting error) was determined to be 5% (1
s) for 234Tht in CTD/SW samples and 6% in BWS bottle samples. For 234Thd in CTD/SW samples the error is 6%, and for 234Thp it is 4%. Because of varying thickness of the sediment filters, 234Th activities were corrected for selfabsorption according to Rutgers van der Loeff and Moore (1999) by applying a calibration algorithm derived from measuring strong beta

ARTICLE IN PRESS M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

1746

Table 2 234 Th and SPM data of water-column samples at the three stations 234

Thp (dpm l
1)

234

Thd (dpm l
1)

234

Tht (dpm l
1)

234

Thdepl (dpm l
1)

234

0.7370.13 0.3170.04 0.3970.05 0.3970.05 8.071.0 9.071.0 6.271.0 9.971.0 7.771.0

0.3270.01 0.3670.01 0.4370.02 0.4070.02 0.5770.19 1.2370.21 0.9770.20 1.0670.20 0.7970.19

0.6170.04 1.9170.10 1.5770.09 1.7470.13 n/d n/d n/d n/d n/d

0.9370.04 2.2770.10 2.0070.09 2.1470.13 2.3170.14 2.9670.16 2.7170.15 2.7970.16 2.5370.14

1.4870.04 0.1170.10 0.3870.09 0.2470.13 0.0770.14
0.5970.16
0.3370.15
0.4270.16
0.1570.14

450780 11607160 11007150 10207140 717 25 137728 156741 107723 103729

1328 100 40 10 1.1 0.9 0.6 0.4 0.25 1.1

0.9270.17 0.4670.06 0.3870.05 0.4970.06 3.171.0 3.371.5 16.271.0 122.671.5 119.571.0 0.4170.01

0.2370.01 0.4970.02 0.5870.02 0.6770.03 0.9470.19 1.1370.19 2.1270.23 5.5570.68 16.171.4 0.3970.01

0.5670.04 1.6270.12 1.5270.11 1.3670.10 n/d n/d n/d n/d n/d n/d

0.7970.04 2.1070.12 2.1070.11 2.0370.10 2.3070.16 2.4970.16 3.4870.21 6.9170.68 17.471.4 n/d

1.6270.04 0.2770.12 0.2770.11 0.3570.10 0.0970.16
0.1070.16
1.1070.21
4.5370.68
15.171.4 n/d

250750 10507150 15607210 13607190 3047120 3397117 131716 4576 134712 584739

1346 100 40 10 1.1 0.9 0.6 0.4 0.25 0.3

0.4770.08 0.2770.04 0.1970.03 0.2570.03 7.371.0 10.871.5 11.271.0 30.871.5 20.671.0 0.2970.01

0.3470.01 0.2470.01 0.2370.01 0.3170.01 0.6070.18 0.8370.21 0.4470.18 3.1470.26 2.0270.23 0.2370.01

0.9270.06 2.0570.14 2.0570.14 1.9070.10 n/d n/d n/d n/d n/d n/d

1.2770.06 2.2970.14 2.2870.14 2.2270.10 2.5070.15 2.7370.19 2.3470.15 5.0570.14 3.9370.21 n/d

1.1570.06 0.0970.14 0.1070.14 0.1670.10
0.1270.15
0.3570.19 0.0470.15
2.6670.14
1.5470.21 n/d

7307140 8707120 12207170 12607180 82727 77723 39716 102710 98712 764735

Station

Type

h.a.s. (m)

GeoB84101

SW CTD CTD CTD BWS BWS BWS BWS BWS

1224 100 40 10 1.1 0.9 0.6 0.4 0.25

GeoB8463

SW CTD CTD CTD BWS BWS BWS BWS BWS IP

GeoB8467

SW CTD CTD CTD BWS BWS BWS BWS BWS IP

SPM (mg l
1)

Thspec (dpm g
1)

For explanation of sample type abbreviations see Table 1. Note that SPM for the SW and CTD samples was determined from beam attenuation according to Eqs. (1) and (2), respectively, while SPM was determined from particle weight for the BWS and in situ pump samples. 234Thp data of the BWS samples is not measured but calculated as difference between the respective 234Tht value and 234Thd of the lowermost CTD sample (10 m a.s.) at the respective station. h.a.s. ¼ height above seafloor, n/d ¼ no data. Propagated measurement errors (including sampling and 2s counting error) are given.

sources below a subset of samples representative of sample weight. Because of high particle load on the BWS filters at 0.25 and 0.4 m above the seafloor at station GeoB8463, the correction for self-absorption was applied to these two filters as well. When the relatively low beta activities of 234Th in the sediment are measured, supported 234Th and the influence of other beta emitters, such as 40K, 226Ra daughters, and 210Pb, must be considered. Determination of these background count rates was done by recounting of all samples 9 month after the cruise, the difference of the two measurements giving 234Th excess (234Thxs) provided by recent deposition of

particles to the sediment. Thereby, we assume that the activities of the other beta emitters do not change over time. The major advantage of this procedure is its simple and quick sample handling, because it does not involve any chemical preparation of the sediment. It suffices in the uppermost few centimeters of sediment, but with increasing depth the uncertainties in the determination of background beta activity make the error in unsupported 234 Th large. Some nuclides may not have been in equilibrium or may have escaped as a gas phase prior to the first count after sampling. We estimate the error in the 234Thxs inventories to be about 25%.

ARTICLE IN PRESS M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

3. Results 3.1. Nepheloid layer distribution in the water column Surface and BNLs are well developed over the continental slope off Namibia, as described by Inthorn et al. (2006b) and visualized by the beam attenuation profiles (Fig. 2a). At GeoB8463, beam attenuation is highest over almost the entire water column. A quite intensive SNL is distinguishable in the topmost 60 m, but also the BNL is most prominent and about 220 m thick. The northernmost station, GeoB84101, takes an intermediate position in beam attenuation. The SNL is 70 m thick, but less intensive. The BNL extends 96 m above the seafloor with increasing beam attenuation from 96 to 40 and constant intensity in the bottommost 40 m. At the southernmost station GeoB8467, particle content is comparatively low. A weak SNL covers the topmost 70 m, and only in the lowermost 20 m is a weak BNL distinguishable. Additionally, an intermediate nepheloid layer is discernible at 700–900 m above the seafloor. For a general description of the nepheloid layer distribution off Namibia, see Inthorn et al. (2006b). 3.2.

234

234

Th in the water column

Thp values of the surface-water samples are considerably lower at GeoB8463 compared to GeoB84101 and GeoB8467 (Fig. 2b and Table 2). But at this station, 234Thp values increase considerably towards the seafloor and reach 0.67 disintegrations per minute (dpm) l
1. At station GeoB84101, 234 Thp stays at a constant level of 0.4 dpm l
1 throughout the water column, increasing slightly towards the seafloor. At GeoB8467, 234Thp is higher in the surface waters than in the lowermost 100 m above the seafloor, with highest 234Thp of 0.31 dpm l
1 10 m above the seafloor. 234 Thd is generally depleted in the SNL and the bottommost samples compared to the samples 100 m above the seafloor (Fig. 2c and Table 2). Throughout the water column, 234Thd values are highest at station GeoB8467 and lowest at GeoB8463. Fig. 2d illustrates the overall depletion of 234Tht with respect to the supply by its parent 238U. At all stations, surface-water samples are significantly depleted in 234Tht, the depletion being more intense at GeoB8463 and GeoB84101 compared to GeoB8467. In the samples from the lowermost 100 m above the

1747

seafloor, 234Tht is generally higher. Nevertheless, there is significant depletion of 234Th at all three stations. At GeoB8463, 234Th is depleted at all three sampling depths, while the depletion is quite small in the sample 100 m above the seafloor at GeoB84101, and in the 100 and 40 m samples of GeoB8467. 3.3. BWS data The BWS bottle samples at GeoB8463 and GeoB8467 reveal a strong increase in SPM towards the seafloor (Fig. 3a and Table 2). The increase is exceptionally strong at station GeoB8463, where SPM concentrations reach more than 100 mg l
1 at the two lowermost bottles. Only at GeoB84101, SPM values stay rather constant at a level of 6.24–9.89 mg l
1. It is striking that SPM obtained with the in situ pump from the same deployments at stations GeoB8463 and GeoB8467 is considerably lower (0.41 and 0.29 mg l
1, respectively). SPM data, calculated from beam attenuation 0.25 m above the seafloor, are in the range of the in situ pump results (0.68–0.22 mg l
1 at stations GeoB8463 and GeoB8467, respectively) but considerably below the bottle data. Fig. 3b shows 234Tht of the BWS bottle samples. Corresponding to the SPM data, 234Tht activity strongly increases towards the seafloor at GeoB8463 and GeoB8467. In the lowermost two samples, 234 Tht clearly exceeds the level of supported 234Th activity from 238U. Only at station GeoB84101 does the 234Tht activity stay around equilibrium at all five sampling heights. Compared to 234Tht data of the lowermost CTD samples (Fig. 2d), 234Tht activities at all three stations are significantly higher. In contrast to this, 234Thp of the in situ pump filters (Fig. 3b) is in better agreement with the low values of the CTD particle samples. 3.4. Sediment data At GeoB8463, the maximum in 234Thxs is not at the sediment surface but 0.6–0.8 cm and no excess 234 Th was found at 2.0 cm depth. There are two deeper maxima, at 2.8–3.6 and 5.2–6.4 cm depth (Fig. 4). At station GeoB84101, the maximum is at 0.2–0.4 cm, and 234Thxs reaches zero at 0.6 cm depth. 234 Thxs scatters around zero down to 3.0 cm, and shows unexpected negative values below this depth. Calculated linear sediment accumulation rates are 260 mg m
2 d
1 for station GeoB8418 and 707 mg m
2 d
1 for GeoB8451.

ARTICLE IN PRESS 1748

M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

Fig. 2. Vertical profiles of (a) beam attenuation, (b) 234Thp, (c) 234Thd, and (d) 234Tht activity against sample height above the seafloor at stations GeoB84101, GeoB8463, and GeoB8467. Legend from graph (b) is also valid for graphs (c) and (d). Topmost samples are from the ship’s seawater supply; samples from 10, 40 and 100 m above the seafloor are from the CTD rosette. The hatched box in graph (d) corresponds to the mean activity of 238U in seawater. Horizontal error bars indicate propagated errors and are sometimes smaller than symbol.

ARTICLE IN PRESS M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

1749

Fig. 3. (a) SPM values and (b) 234Tht activities of BWS bottle samples from stations GeoB84101, GeoB8463 and GeoB8467 against height above seafloor. In addition, SPM values and 234Thp from the in situ pump attached to the BWS are given for stations GeoB8463 and GeoB8467. The dashed line corresponds to the mean activity of 238U. Horizontal error bars are smaller than symbols.

4. Discussion The organic-rich sediments at productive continental margins are a sink for uranium (Cochran, 1992). If this significantly changes the 238U content in the bottom waters, it should be taken into account in our calculation of the 234Th depletion. First, we assume that the ventilation of the water masses over the Namibian slope should be sufficient to prevent a significant depletion of uranium in the mid water column (Inthorn et al., 2006b). As we did not measure 238U directly, we estimated the uranium concentration gradient in the BNL based on normal values for deep-sea mixing and uranium accumulation. Uranium accumulation rates, as derived from pore water profiles of uranium, range from 0 to 16 mg cm
2 kyr
1 in suboxic hemipelagic sediments and up to 70 mg cm
2 kyr
1 in anoxic sediments, in which sulfate reduction occurs (Barnes and Cochran, 1990; Cochran, 1992). A sink of this magnitude would at steady state be resupplied by a gradient in dissolved U over the BNL. In the BNL we can assume enhanced vertical mixing rates as a

result of turbulent shear over the seafloor. Taking an eddy diffusion coefficient of 10 cm2 s
1, the concentration drop over a BNL of 100 m thickness would amount to 2.1 ng kg
1, less than 1% of ambient U concentrations (3.2 mg kg
1 at 35% salinity; Chen et al., 1986). We conclude that the uranium removal by accumulation in the sediments does not affect our procedure to derive the uranium in the water column from salinity. 4.1. The in situ resuspension experiment It is improbable that the high SPM and 234Tht values of the BWS bottle samples (Fig. 3 and Table 2) result solely from particles originally suspended in the bottom waters. After deposition of the BWS at the seafloor, measured beam attenuation far exceeded the values in the nepheloid layers (Fig. 5). High amounts of particles must have been resuspended from the surface sediment during the deployment. Some of these particles apparently entered the bottles and were deposited on their bottoms, where they were inhibited from flushing by

ARTICLE IN PRESS 1750

M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

Fig. 4. 234Thxs in surficial sediments of stations GeoB84101 and GeoB8463 (see also Table 3). Gray areas are used for calculation of the inventory of 234Thxs in the sediments. Horizontal error bars indicate propagated errors and are sometimes smaller than symbol.

Fig. 5. Water depth (dashed line) and beam attenuation (solid line) measured by the bottom water sampler during the deployment at station GeoB8463 illustrating intense particle resuspension after landing of the device at the sea floor.

the boundary layer within the bottles themselves. The SPM values of the BWS samples are much higher than the results from the in situ pump filters

and the beam attenuation profiles (data prior to deployment). The in situ pump filters were not affected by the resuspension because the pumping

ARTICLE IN PRESS M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

interval was programmed to start about 20 min after deployment, while the deployment-induced resuspension lasted only a couple of minutes until the particles had settled again or drifted away with the bottom current (Fig. 5). Differences in the particulate organic carbon concentrations measured in bottle samples compared to in situ pump samples are described in the literature (Bishop, 1999; Buesseler et al., 2006; Gardner et al., 2003; Liu et al., 2005; Moran et al., 1999). But large differences in the SPM concentrations measured with pumps or bottles have, to the best of our knowledge, never been described before and are very improbable. We therefore explain the high 234 Tht values of the BWS bottle samples by artificial resuspension of additional 234Th on particles from the sediment surface. The difference between 234Tht of the BWS bottle samples and 234Thd of the lowermost CTD samples provides low estimates of the particulate 234Th in the BWS samples, because previous studies (Rutgers van der Loeff et al., 2002; Turnewitsch and Springer, 2001) showed that 234Thd generally decreases with decreasing distance to the sediment surface, where particle throughput and, as a result, scavenging is most intensive. The resulting 234Thp values are correlated with the respective particle concentrations (Fig. 6): 234


1


1

Thp ½dpm l  ¼ 0:086 SPM ½mg l  þ 0:27

ðR2 ¼ 0:73; n ¼ 15Þ.

ð3Þ

GeoB8463 and GeoB8467 are much more affected by this artificial resuspension compared to GeoB84101 (Fig. 3b). The difference may be due to varying grain size composition of the sediments or the impact of the BWS during deposition. SPM values strongly decrease from the lowermost to the uppermost bottle at GeoB8463 and GeoB8467, while all five bottles show comparatively low values at GeoB84101. Although the sampling of bottom water with bottles of Niskin type horizontally fixed to a frame is comparatively easy and has proven to be useful for the sampling of fluids (Mau et al., 2006; Sauter et al., 2005), we show that such a deployment may not obtain representative samples of the naturally suspended particles—at least for deployments in areas with soft, easily resuspendable surface sediments. Nevertheless, this artificial resuspension event of the BWS gives particularly interesting results, as described below.

1751

Fig. 6. 234Thp activities of BWS bottle samples, as difference of 234 Tht of the BWS bottle samples to 234Thd of the lowermost CTD sample, from stations GeoB84101, GeoB8463, and GeoB8467 against SPM values. Dashed line is from linear regression of all data. Vertical error bars indicate propagated errors and are sometimes smaller than symbol.

4.2. Specific 234Th activity of different particle sample types Up to now, 234Th models using particle exchange budgets between the seafloor and the BNL have only used rough estimates of the specific 234Th activities of resuspended particles, based on 234Th activity of sediment surface samples (e.g., Aller and DeMaster, 1984; Bacon and Rutgers van der Loeff, 1989; Rutgers van der Loeff et al., 2002). The results of the in situ resuspension experiment, as sampled with the BWS, should provide better data. 234Thspec was calculated from 234Thp and SPM data for the water samples. For the sediment surface samples, representing the uppermost 0.2 cm of the sediment, 234 Thspec was determined from sample weight and 234 Thxs, because in the sediments contribution

ARTICLE IN PRESS 1752

M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

of supported Th has to be considered (Fig. 7 and Table 2). The linear trend (Eq. (3)) of the BWS samples corresponds to average 234Thspec of the resuspended particles of 86 dpm g
1. Nevertheless, the samples cover a wide range of 234Thspec values (39–339 dpm g
1). The two BWS samples with high activities of about 300 dpm g
1 at GeoB8463 are from the uppermost bottles and thought to contain very little artificially resuspended material, but a higher fraction of long-time suspended material. All

other BWS samples show values below 160 dpm g
1. 234 Thspec of the BWS bottle samples is intermediate between the high values from the in situ pump (580–760 dpm g
1) and the CTD rosette (870–1560 dpm g
1), and the low values measured at the sediment surface (26 dpm g
1 at GeoB8463 and 37 dpm g
1 at GeoB84101). The low 234Thspec data of the surface sediments result from intensive dilution through non-local mixing and are in the same range as sediment data from the Middle Atlantic bight (10–40 dpm g
1, Santschi et al., 1999), and somewhat lower than at the Fram Strait (40–80 dpm g
1, Rutgers van der Loeff et al., 2002). According to our data, an intensive resuspension event, as it can be induced by storm events or internal wave action, is thought to preferentially resuspend particles with higher 234Thspec compared to bulk sediment surface samples. From a sediment core, it is virtually impossible to sample only the very thin layer of these easily resuspendable particles. This is the advantage of our resuspension experiment. On the other hand, 234Thspec of the BWS bottle samples is low compared to particles suspended in the BNL otherwise. The high 234Thspec of the in situ pump and CTD samples, representing the ‘‘background’’ situation, indicate fine particles that stay in suspension or directly at the sediment– water interface for longer time intervals, and hence are strongly enriched in 234Th. Particle size and composition are other parameters that are known to influence 234Thspec values (Buesseler et al, 2006), but they have not been studied in this survey. However, for the BWS samples the effect of the strong resuspension event overwhelms the influence of these parameters. 4.3. Particle dynamics at the sediment– water interface on the Namibian continental slope

Fig. 7. 234Thspec of suspended particles of different sample types from the water column at stations GeoB84101, GeoB8463 and GeoB8467 against height above the seafloor. Samples from lowermost 1.1 m above the seafloor are from the BWS; other samples were taken with the CTD rosette. The arrows at the lower abscissa indicate 234Thspec values of samples from the topmost 0.2 cm of the sediment at stations GeoB84101 (thick arrow) and GeoB8463 (thin arrow). Horizontal error-bars indicate propagated errors and are sometimes smaller than symbol.

In the Benguela upwelling area, high Corg contents of the surface sediments of up to 10% on the continental slope off Namibia indicate a depocenter of organic matter with its center in water depths of 400–1500 m at 25.51S. This is reflected in high sedimentation, mass accumulation (Inthorn et al., 2006a; Mollenhauer et al., 2002), and carbon mineralization rates (Aspetsberger, 2006; Ferdelman et al., 1999; Glud et al., 1994) in this area. Generally, vertical flux of fresh biogenic particles or lateral particle transport in subsurface nepheloid layers are possible organic matter sources. While station GeoB8463 is located directly

ARTICLE IN PRESS M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

over the depocenter (Corg content of the surface sediments is 8.06%; Inthorn et al., 2006a), stations GeoB84101 and GeoB8467 are positioned north and south of it, respectively; Corg content of the surface sediments is lower at these stations (3.30% and 4.08%, respectively; Inthorn et al., 2006a). Up to now, there are few examples of simultaneous determination of 234Th inventories of the BNL and the corresponding surface sediment (e.g., DeMaster et al., 1991; Rutgers van der Loeff et al., 2002; Turnewitsch and Springer, 2001). This study provides the first 234Th data reported from the Benguela upwelling area. 234Th/238U disequilibrium is used to estimate the respective significance of the two sources and to quantify particle deposition, reworking, and transport at and just above the sediment–water interface. At all stations, the water samples taken with the CTD at near-bottom depths are depleted in 234Tht (Fig. 2d and Table 2), while there is excess 234Th in the sediments. This 234Th depletion in the BNL is explainable with three different models. The first is based solely on local resuspension; the second evaluates the influence of vertical particle settling from the productive surface waters on the 234Th budget in the BNL; and the third includes sedimentation of particles that have been transported in the BNL over long distances (Fig. 8). In the following, these models will be discussed and compared to each other.

Loeff, 1989). Accordingly, the depletion in the BNL must be balanced by 234Th excess in the surface sediments. Subsurface 234Thxs is explained solely by non-local mixing without any net sediment accumulation. Good correlation of SPM values derived from beam attenuation profiles, to 234Thdepl and 234Thp (Fig. 9) in the BNL from samples obtained with the CTD rosette is observed, according to the relations: 234

Thdepl ½dpm l
1  ¼ 0:92 SPM ½mg l
1 
0:10

ðR2 ¼ 0:70; n ¼ 9Þ, 234

ð4Þ

Thp ½dpm l
1  ¼ 1:29 SPM ½mg l
1 
0:04

ðR2 ¼ 0:75; n ¼ 9Þ.

ð5Þ

The correlation of 234Thdepl and SPM concentrations supports the model, indicating constant settling rates/residence times of the particles during relatively long resuspension cycles in the BNL. Interestingly, such a relationship between SPM and 234 Thdepl is also reported by Rutgers van der Loeff et al. (2002) for the Fram Strait, but for a much higher number of samples, indicating similar particle exchange processes in the two areas. The sample from 40 m above bottom at station GeoB8463 is taken from an intermediate position between two sublayers of the BNL with higher particle content (Fig. 2a). Because the nepheloid layers are subject to varying current conditions, e.g., tidal influences, this sublayer situation is probably not a permanent feature. This explains the relatively low SPM value compared to relatively high 234 Thdepl, integrating over several days. At GeoB84101, the 100 m sample is positioned above the upper boundary of the BNL. Therefore, it shows a lower SPM concentration and less 234Thdepl

4.3.1. Model 1: local resuspension This model assumes that local cyclic resuspension of sediment surface particles to the BNL by bottom currents and later resedimentation on a time scale of weeks is the source of 234Th depletion in the BNL, resulting from scavenging of 234Thdiss to the suspended particles (Bacon and Rutgers van der

Model 1

1753

Model 2

Model 3

S

BNL

L R

M

R

M

B

Fig. 8. Schematic illustration of the three model scenarios for particle transport in the bottom nepheloid layer (BNL) and particle exchange at the sediment–water interface described in the text. S: settling flux; R: resuspension flux; M: non-local mixing; L: long-distance particle transport; B: burial in the surface sediments.

ARTICLE IN PRESS M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

1754

Fig. 9. (a) 234Thdepl and (b) 234Thp of the CTD rosette samples of GeoB84101, GeoB8463, and GeoB8467 against SPM as calculated from beam attenuation. Dashed lines are from linear regression of all data.

compared to the samples from 10 and 40 m above the seafloor. The weak and shallow BNL at station GeoB8467 is reflected in low SPM and 234Thdepl values. The regression in Fig. 9 gives a background SPM level of about 0.1 mg l
1 for zero 234Thdepl. This is a reasonable number for a ‘‘standing stock’’ of very light particles, not involved in particle exchange with the sediment. The inventories of 234Thdepl (Idepl) and 234Thp (Ip) in the BNL at the three stations are calculated by applying Eq. (2) to derive SPM distribution for the overall height of the BNL from beam attenuation, and Eqs. (4) and (5), respectively, to determine 234 Thdepl and 234Thp (Table 4). Based on the model of Rutgers van der Loeff and Boudreau (1997), the average residence time of particles in the BNL is determined by comparison of the overall inventory of 234Thp in the BNL (Ip) with the fluxes of 234Thp into this reservoir (D: ingrowth from depletion; R: resuspension flux; S: settling flux) according to the equation t¼

Ip . DþRþS

(6)

Eq. (6) is, strictly taken, valid only for a single class of particles in a well-mixed BNL at steady state.

In model 1, we assume that the settling flux from the surface waters is negligible and that the particle resuspension flux FR alone supports the inventory of SPM in the BNL (ISPM, as calculated from beam attenuation according to Eq.(2); Table 4). If the settling flux S is neglected, we can write I SPM t¼ (7) FR and Eq.(6) can be reduced to IP t¼ . DþR

(8)

The combination of Eqs. (6) and (7) gives I SPM FR ¼ ðD þ RÞ. IP

(9)

If AR is the specific activity on resuspended particles, the resuspended 234Th flux R is R ¼ F R nAR

(10) 234

Thspec of the and by applying the average resuspended particles as sampled with the BWS of 86 dpm g
1 for AR, FR can be calculated according to the equation: I SPM D FR ¼ . (11) I P
I SPM AR

ARTICLE IN PRESS M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

The calculated particle resuspension fluxes are significantly higher at GeoB8463 and GeoB84101 than at GeoB8467 (Table 4), where there is little particle exchange at the seafloor. However, the 234 Th resuspension flux R is very low compared to the ingrowth from depletion in the BNL itself (Table 4), because of the high 234Thspec of suspended particles in the BNL compared to 234Thspec of resuspended particles as sampled with the BWS (AR). Average residence times of the particles in the BNL can now be calculated as well (Eq. (8)) and are rather high at all three stations, ranging from 8 to 11 weeks (Table 4). These values are relatively high compared to 6 weeks at the Fram Strait (Rutgers van der Loeff et al., 2002) and 1–3 weeks at the Middle Atlantic Bight (Santschi et al., 1999). According to the model, Idepl should be balanced by an equivalent sedimentary 234Thxs inventory (Ised). 234Thxs profiles at GeoB84101 and GeoB8463 reach zero activity at 0.6 and 2.0 cm sediment depth, respectively, but reveal considerable variation and even unexpected negative values below these depths (Fig. 4). This is attributable to errors related to increasing influence of other beta emitters with lower 234Thxs content of the sediment. For calculation of Ised, only the uppermost samples down to the depth where zero activity is reached for the first time have been considered (Table 4). At GeoB84101, Ised is equivalent to Idepl (2.4 and 2.2 dpm cm
2, respectively) whereas at GeoB8463, Ised is somewhat larger (10.5 and 6.1 dpm cm
2, respectively). Given the large errors involved (25% error in Ised, 20–60% error in Idepl, cf Table 2), this is barely significant. Moreover, a perfect match of Idepl and Ised is not to be expected. Because of the short half-life of 234Th, patchy deposition of sedimenting material at the sediment–water interface and bioturbation may cause considerable horizontal heterogeneity. 4.3.2. Model 2: local resuspension plus vertical particle settling In this second model, we estimate the possible influence of the settling of 234Th-bearing particles through the water column (flux S, Eq. (6)) on the 234 Th budget in the BNL. no complete depth profiles of 234Th distribution in the water column were determined during our survey, so we do not know the flux of 234Th settling into the BNL. Even if complete, 234Th profiles had been available for the water column, the uncertainty in S would be large (Rutgers van der Loeff et al., 2002; Buesseler et al., 2006). However, 234Th data of the surface-water

1755

samples provide a first approximation of the intensity of 234Th and particle export from the surface waters (Coale and Bruland, 1985). By assuming that total 234Thdepl is homogenously distributed over, and nil below, the mixed layer, a rough estimate can be made. The three stations GeoB84101, GeoB8463, and GeoB8467 show 234 Thdepl of 1.48, 1.62, and 1.15 dpm l
1 and mixed-layer depths of approximately 46, 52, and 60 m, respectively (Fig. 2a, d and Tables 2 and 3), resulting in rather similar steady-state export fluxes of 1942, 2403, and 1968 dpm m
2 d
1 (Table 4). As the opposite extreme to model 1, representing the no-settling flux situation, in a first scenario (model 2a) the full export flux from the surface waters was injected into the BNL. However, it is to Table 3 234 Th excess in the surface sediments at stations GeoB84101 and GeoB8463 Station

Sediment depth (cm)

234 Thxs (dpm cm
3)

GeoB84101

0.1 0.3 0.5 0.7 0.9 1.25 1.75 2.25 2.75 3.25 3.75 4.25 4.75

4.5071.98 5.0371.30 2.4871.07
0.3970.88
2.0171.60
1.0871.15
0.6170.41
0.5270.39
0.6470.42
4.0871.86
4.2571.34
3.9171.63
2.7371.45

GeoB8463

0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.8 2.2 2.6 3 3.4 3.8 4.2 4.6 5 5.4 6 6.8

3.1571.60 6.6171.47 7.3271.49 15.3372.93 5.9371.20 5.1371.05 2.3870.77 3.7370.99 1.0671.21 0.2970.59
0.1170.46 1.5470.60 2.5870.73 0.0870.57
0.6870.59
0.3870.54
1.0670.81 4.6571.18 4.3471.12 0.8570.93

ARTICLE IN PRESS 1756

M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

Table 4 Model input parameters and results GeoB84101

GeoB8463

GeoB8467

Surface water Mixed layer depth (m) Export flux S (dpm m
2 d
1)

46 19427218

52 24037239

60 19687194

Sediment trap fluxes Sediment trap flux in 2000 (mg m
2 d
1)

206

316

n/d

BNL BNL height (m) 234 Thdepl inventory BNL Idepl (dpm cm
2) 234 Thp inventory BNL Ip (dpm cm
2) SPM inventory BNL ISPM (mg cm
2) 234 Thxs inventory sediment Ised (dpm cm
2)

96 2.270.6 4.271.3 3.571.1 2.470.6

220 6.171.3 1173.3 9.172.7 10.572.6

20 0.270.1 0.570.2 0.570.2 n/d

Model 1. (neglecting S) Ingrowth from depletion D (dpm m
2 d
1) Particle resuspension flux FR (mg m
2 d
1) 234 Th resuspension flux R (dpm m
2 d
1) Particle residence/removal time t (d)

6287157 5637286 48.4726 62724

17407365 15507758 133.3770 59721

57736 62748 5.474 80752

Model 2a (including settling particle flux) Settling particle flux S (dpm m
2 d
1) Specific activity of settling particles AS (dpm g
1) Settling particle flux FS (mg m
2 d
1) Specific activity of settling particles AE (dpm g
1) Particle resuspension flux FR (mg m
2 d
1) 234 Th resuspension flux R (dpm m
2 d
1) Particle residence/removal time t (d) Expected sediment inventory (dpm cm
2)

1942 1092 1778 1200 391 33.7 16.1 9.1

2403 1323 1816 1209 1735 149.2 25.6 15.0

1968 1117 1763 1000 287 24.7 2.4 7.2

Model 2b (including smaller settling particle flux) Settling particle flux S (dpm m
2 d
1) specific activity of settling particles AS (dpm g
1) Settling particle flux FS (mg m
2 d
1) Specific activity of settling particles AE (dpm g
1) Particle resuspension flux FR (mg m
2 d
1) 234 Th resuspension flux R (dpm m
2 d
1) Particle residence/removal time t (d) Expected sediment inventory (dpm cm
2)

583 1092 533 1200 512 44.0 33.5 4.4

721 1323 545 1209 1605 138.0 42.3 9.1

590 1117 529 1000 130 11.2 7.6 2.3

Model 3 (long-range lateral transport) Flux of 234Th to the sediment FE (dpm m
2 d
1) Specific activity of settling particles AE (dpm g
1) Particle settling flux Fsed (mg m
2 d
1) Carbon content of surface sediments (%) Carbon settling flux FCorg (mg m
2 d
1) Particle residence/removal time t (d)

6857171 12007200 5707171 4.8 68721 61726

29957749 12007200 24787743 7.9 297789 37716

As model 2 is a sensitivity analysis, no uncertainties are given for these parameters. n/d ¼ no data. Surface sediments at station GeoB8467 were not sampled for 234Thxs.

be expected that most of the biogenic particle flux settling out of the mixed surface layer is remineralized on its way through the more than 1200 m deep water column resulting in release of the adsorbed thorium. Therefore, in the second scenario (model 2b), the settling flux reaching the BNL was reduced to 30% of the initial flux.

Considering a steady state situation for the budget of the BNL, we can write F E nAE ¼ D þ S þ R,

234

Th

(12)

where FE is the flux of particles with the specific activity AE out of the BNL towards the sediment. We derived AE from ISPM and Ip in the BNL giving

ARTICLE IN PRESS M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

values of 1000–1209 dpm g
1 (Table 4). Ingrowth from depletion D is similar to that in model 1. The settling flux S reaches the BNL from the water column as a particle flux FS with the specific activity AS. Similarly, a resuspended particle flux of FR with specific activity of AR carries an activity flux of R. For AS we applied the average specific activity of the respective CTD samples and for AR the activity of the resuspended particles sampled with the BWS. In steady state, FS and FR together equal FE. We can then write ðF S þ F R ÞnAE ¼ D þ S þ F R nAR ,

(13)

and calculate the resuspension flux FR for the two scenarios: FR ¼

D þ S
F S AE . AE
AR

(14)

For GeoB84101, FR is somewhat lower (394 and 514 dpm cm
2 d
1 in model 2a and b, respectively; Table 4) than in model 1 (566 dpm cm
2 d
1; Table 4). For GeoB8463 the scenarios including settling flux give somewhat higher FR values (1735 and 1605 compared to 1550 dpm cm
2 d
1; Table 4). Only for GeoB8467 does the consideration of a settling flux increase FR significantly, by a factor of 2–4 (290, 133 and 66 dpm cm
2 d
1, respectively; Table 4). In all three scenarios, FR at GeoB8463 is higher by a factor of 3–23 compared to the other two stations, confirming particularly intense particle exchange at the sediment–water interface at this station, positioned in the heart of the depocenter. The residence time of particles with respect to their removal out of the BNL can now be calculated: t¼

I SPM . F S þ F Res

intensive oxygen minimum zone at 200–400 m water depth (Bailey, 1991; Inthorn et al., 2006b). It is likely that part of the 234Th exported from the euphotic zone is released in this zone of mineralization, as was shown previously by Usbeck et al. (2002) and Savoye et al. (2004). This zone was not sampled for 234Th in this study. If only part of the exported thorium reaches the BNL (model 2b), the residence times become intermediate (7–42 d; Table 4). Model results for station GeoB8467, which has the lowest particulate 234Th inventories, are most affected by the varying strength of the settling flux S, which adds much uncertainty to the results from this station. If the settling flux is set to 0, model 2 becomes equivalent to model 1. 4.3.3. Model 3: sedimentation after long-range lateral transport in the BNL The third model explains 234Th depletion in the BNL by sedimentation of particles that have stayed in suspension during long-range transport. There are some indications supporting the significance of the processes involved in this model:





 (15)

If the settling particle flux from the surface waters is completely added to the BNL budget without any mineralization in the mid water column (model 2a), it overwhelms all other fluxes at GeoB84101 and GeoB8467 and is of similar magnitude as D at GeoB8463. This results in much shorter calculated particle residence times in the BNL compared to model 1 (2–25 d compared to 59–76; Table 4). However, from our general understanding of particle fluxes at continental margins and previous results from the study area, we believe this represents a gross overestimation. The vertical flux in the water column is subjected to effective mineralization, as indicated by an

1757



The excess, although barely significant, of Ised relative to Idepl at GeoB8463 indicates lateral particle transport to this location, positioned directly over the depocenter. Mollenhauer et al. (2003) present long-term current measurements from slope-depth bottom waters off Namibia, revealing tidally driven changes in current direction, but offshore-directed mean flow, and velocities up to 25 cm s
1, high enough to keep fine particles in suspension for a long period of time. Even the calculated long residence times of models 1 and 2 corroborate the high significance of long-range advective transport. Net advection during a 50-d period can easily be up to 100 km (cf. Mollenhauer et al., 2003). High 14C-based sediment accumulation rates for the stations at 1000 m water depth at 24.251S and 25.51S support the role of these parts of the Namibian slope as a particle depocenter.

In model 3, we assume that lateral input is the only supply route and disregard both resuspension and vertical settling from the surface water. The 234 Th depletion in the BNL is then maintained by a continuous settling of particles out of the BNL. The flux of 234Th to the sediment can in steady state be derived from the inventory of 234Th in the sediment

ARTICLE IN PRESS M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

1758

Ised (or from the similar inventories of depletion in the BNL Idepl, see discussion above): F E ¼ I sed l,

(16)

(Biscaye and Anderson 1994), yielding a sedimentation rate: F sed ¼ 234

FE I sed l ¼ , AE Thspec

(17)

where AE1200 dpm g
1 is again the specific activity of the particles settling out of the BNL. Fsed is significantly higher at station GeoB8463 (2478 mg m
2 d
1) than at station GeoB84101 (570 mg m
2 d
1), reflecting the role of this part of the Namibian slope as an area of intensive particle deposition. 234Th-based sediment accumulation rates are somewhat higher than corresponding 14Cbased accumulation rates (707 and 260 mg m
2 d
1, respectively), which integrate over much longer time intervals. The particle residence times, calculated as t¼

I SPM . F sed

are 9 and 5 weeks for GeoB84101 and GeoB8463, respectively (Table 4). In 2000, particle settling fluxes were determined in the mid water column (960 m) somewhat further down the slope (1800 m water depth) off Walvis Bay (206 mg m
2 d
1) and Lu¨deritz (316 mg m
2 d
1) with sediment traps (G. Fischer, pers. comm.). The mass fluxes calculated from 234Th exceed the trap data by factors of 2–13, reflecting the dominant role of lateral particle transport on the continental slope off Namibia. Applying Corg values of BNL particles of 12% as determined by Inthorn et al. (submitted for publication), the flux of Corg into the sediment (FCorg) amounts to 68 mg Corg m
2 d
1 at station GeoB84101, and 297 mg Corg m
2 d
1 at station GeoB8463 (Table 4). These values are of the same magnitude as carbon mineralization rates of 117 mg Corg m
2 d
1 at a station at 1000 m depth and about 15 km onshore of station GeoB84101 and 72 mg Corg m
2 d
1 at station GeoB8463 (Aspetsberger, 2006), indicating that, notwithstanding the large fraction of advected particles, Corg is remineralized to a large extent in the surface sediment. 4.3.4. Model comparision The three models represent extreme cases in describing particle transport and exchange processes at the sediment–water interface. All three models

can explain 234Th depletion in the BNL and 234Th excess in the surface sediments. Good correlation of SPM and 234Thdepl supports model 1, which addresses 234Th scavenging in the BNL, resettling of the particles, non-local mixing in the surface sediments and resuspension of lowactivity particles. It has to be kept in mind though that model 1 is incompatible with net sedimentation. Radiocarbon data clearly show net sedimentation, but 14C integrates over much longer time intervals than 234Th so that it does not necessarily reflect the sampling situation. Seasonal changes may play a role. Model 2 includes sedimentation of 234Th-bearing particles through the water column and into the BNL. This reduces the calculated residence time of particles in the BNL. The extreme case model 2a, where the full export from the surface-water is assumed to enter the BNL, is unlikely, because it disregards mineralization and 234Th release in the water column, and it predicts 234Th inventories in the sediment that are in excess of those observed. Model 2b, where only 30% of the 234Th export from surface waters reaches the BNL, is more realistic. Predicted sediment inventory fits the data at station GeoB8463 but is still too high at station GeoB84101 (Table 4). Model 3, focusing on sedimentation of particles after long-range transport in the BNL, better reflects the role of the Namibian continental slope as a particle depocenter. Additionally, the calculated mass fluxes are consistent with carbon mineralization rates in this area. On the other hand, model 3 does not give an explanation for the correlation of SPM and 234Thdepl, leading to the conclusion that reality is presumably best described by a combination of the model approaches. 5. Conclusions In spite of the limited extent of our study, particle exchange and transport at the sediment–water interface and in the BNL are successfully addressed with the natural radiotracer 234Th. Other studies point to the importance of deep lateral particle transport on the continental slope off Namibia (Giraudeau et al., 2000; Inthorn et al., 2006b; Inthorn et al., 2006; Mollenhauer et al., 2003), but here the first quantitative measures of the processes are given. With respect to the observed 234Th distribution, three models could explain depletion in the BNL and excess in the surface sediments, all

ARTICLE IN PRESS M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

emphasizing the significance of processes at the sediment–water interface. High mass flux into the sediment at a slope depth depocenter, compared to order-of-magnitude lower vertical particle fluxes towards nearby sediment traps, strongly support the findings of the other studies. Several studies have used the inventory of 234Th in surface sediments as a tool to quantify recent sedimentation rates (e.g., DeMaster et al., 1994; Schmidt et al., 2002). The present study, where the depletion of 234Th in the BNL is of the same order of magnitude as the excess inventory in the sediment, shows the importance of scavenging of 234Th in the BNL, a process that has been measured in very few studies. If only BNL fluxes are considered, similar residence/removal times at all three stations indicate generally identical transport/removal processes along the slope, while the extent and particle concentration of the BNL as well as the sediment accumulation rate vary significantly. Together with other new data from the same study area, e.g., regional distribution and particle content of the BNL and organic carbon distribution, mineralization, quality and burial in the sediments, the 234Th results provide the basis for a budget of the particle and organic carbon fluxes at this highly productive continental margin. Additionally, such information is of value for a better understanding of the controls of the formation of the sedimentary record on short time scales and the controls of food supply to deep-sea benthos. However, a greater number of stations and a better vertical resolution of samples would be necessary to differentiate between resuspension and sedimentation processes, to verify the quantities, to get an overview of the fluxes on larger scales in time and space, and to localize sources and sinks. In future work, it would also be desirable to have trap measurements of 234Th fluxes just above the BNL (cf Smoak et al., 1999) to fill in the missing data in our model 2. 234 Th and SPM samples from the BWS represent an in situ resuspension of particles from the sediment surface. The results are the first specific 234 Th activity data of the easily resuspendable fraction of the surface sediment. This will prove useful for future modeling of 234Th exchange at the sediment–water interface. Acknowledgments We would like to thank the captain and crew of R.V. Meteor for their strong support during cruise

1759

M57/2. We are very grateful for the introduction to the 234Th-sampling procedure and the measurements provided in a very friendly way by I. Voege. We highly appreciate the provision of sensors and technical help by M. Bergenthal, and the in situ pump by W. Balzer and Olaf Wilhelm. G. Fischer kindly provided unpublished sediment trap data. T. Ferdelman and three anonymous reviewers contributed many helpful comments on the manuscript. This research was funded by the Deutsche Forschungsgemeinschaft as part of the Research Center ‘‘Ocean Margins’’ (RCOM) of the University of Bremen, Contribution no. 406.

References Aller, R.C., DeMaster, D.J., 1984. Estimates of particle flux and reworking at the deep-sea floor using 234Th/238U disequilibrium. Earth and Planetary Science Letters 67, 308–318. Aspetsberger, F., 2006. Benthic carbon turnover in continental slope and deep sea sediments: importance of organic matter quality at different time scales. Berichte, Fachbereich Geowissenschaften Bremen, 245, Bremen, 136pp. Bacon, M.P., Rutgers van der Loeff, M.M., 1989. Removal of thorium-234 by scavenging in the bottom nepheloid layer of the ocean. Earth and Planetary Science Letters 92, 157–164. Bailey, G.W., 1991. Organic carbon flux and development of oxygen deficiency on the modern Benguela continental shelf south of 221S: spatial and temporal variability. In: Tyson, R., Pearson, T.H. (Eds.), Modern and Ancient Continental Shelf Anoxia, vol. 58. Geological Society Special Publication, pp. 171–183. Barnes, C.E., Cochran, J.K., 1990. Uranium removal in oceanic sediments and the oceanic U balance. Earth and Planetary Science Letters 97, 94–101. Biscaye, P.E., Anderson, R.F., 1994. Fluxes of particulate matter on the slope of the southern Middle Atlantic Bight: SEEP-II. Deep-Sea Research Part II 41 (2/3), 459–509. Bishop, J.K.B., 1999. Transmissometer measurement of POC. Deep-Sea Research Part I 46, 353–369. Boetius, A., Springer, B., Petry, C., 2000. Microbial activity and particulate matter in the benthic nepheloid layer (BNL) of the deep Arabian Sea. Deep-Sea Research Part II 47, 2687–2706. Buesseler, K.O., Benitez-Nelson, C.R., Moran, S.B., Burd, A., Charette, M., Cochran, J.K., Coppola, L., Fisher, N.S., Fowler, S.W., Gardner, W.D., 2006. An assessment of particulate organic carbon to thorium-234 ratios in the ocean and their impact on the application of 234Th as a POC flux proxy. Marine Chemistry 100 (3–4), 213–233. Carr, M.E., 2002. Estimation of potential productivity in Eastern Boundary Currents using remote sensing. Deep-Sea Research Part II 49 (1–3), 59–80. Chen, J.H., Edwards, R.L., Wasserburg, G.J., 1986. 238U, 234U and 232Th in seawater. Earth and Planetary Science Letters 80, 241–251. Coale, K.H., Bruland, K.W., 1985. 234Th/238U disequilibria within the California current. Limnology and Oceanography 30 (1), 22–33.

ARTICLE IN PRESS 1760

M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761

Cochran, J.K., 1992. The oceanic chemistry of the Uranium and Thorium-series nuclides. In: Ivanovich, M., Harmon, R.S. (Eds.), Uranium-series Disequilibrium: Applications to Earth, Marine, and environmental Sciences, second ed. Clarendon Press, Oxford, pp. 334–395. DeMaster, D.J., Brewster, D.C., McKee, B.A., Nittrouer, C.A., 1991. Rates of particle scavenging, sediment reworking and longitudinal ripple formation at the HEBBLE site based on measurements of 234Th and 210Pb. Marine Geology 99, 423–444. DeMaster, D.J., Pope, R.H., Levin, L.A., Blair, N.E., 1994. Biological mixing intensity and rates of organic carbon accumulation in North Carolina slope sediments. Deep Sea Research Part II: Topical Studies in Oceanography 41 (4–6), 735–753. Ferdelman, T.G., Fossing, H., Neumann, K., 1999. Sulfate reduction in surface sediments of the southeast Atlantic continental margin between 151380 S and 271570 S (Angola and Namibia). Limnology and Oceanography 44 (3), 650–661. Gardner, W.D., Walsh, I.D., Richardson, M.J., 1993. Biophysical forcing of particle production and distribution during a spring bloom in the North Atlantic. Deep-Sea Research Part II 40 (1/2), 171–195. Gardner, W.D., Richardson, M.J., Carlson, C.A., Hansell, D., Mishonov, A.V., 2003. Determining true particulate organic carbon: bottles, pumps and methodologies. Deep-Sea Research Part II 50 (3–4), 655–674. Giraudeau, J., Bailey, G.W., Pujol, C., 2000. A high-resolution time-series analyses of particle fluxes in the Northern Benguela coastal upwelling system: carbonate record of changes in biogenic production and particle transfer processes. Deep-Sea Research Part II 47 (9–11), 1999–2028. Glud, R.N., Gundersen, J.K., Jo¨rgensen, B.B., Revsbech, N.P., Schulz, H.D., 1994. Diffusive and total oxygen uptake of deep-sea sediments in the eastern South Atlantic Ocean: in situ and laboratory measurements. Deep Sea Research Part I 41 (11/12), 1767–1788. Inthorn, M., Mohrholz, V., Zabel, M. 2006a. Nepheloid layer distribution in the Benguela upwelling area offshore Namibia. Deep-Sea Research Part I, this issue [doi:10.1016/ j.dsr.2006.06.004]. Inthorn, M., Wagner, T., Scheeder, G., Zabel, M., 2006b. Lateral transport controls distribution, quality and burial of organic matter along continental slopes in high-productivity areas. Geology 34 (3), 205–208. Jahnke, R.A., Reimers, C.E., Craven, D.B., 1990. Intensification of recycling of organic matter at the sea floor near ocean margins. Nature 348, 50–54. Liu, Z., Stewart, G., Kirk Cochran, J., Lee, C., Armstrong, R.A., Hirschberg, D.J., Gasser, B., Miquel, J.-C., 2005. Why do POC concentrations measured using Niskin bottle collections sometimes differ from those using in-situ pumps? Deep Sea Research Part I: Oceanographic Research Papers 52 (7), 1324–1344. Mau, S., Sahling, H., Rehder, G., Suess, E., Linke, P., Soeding, E., 2006. Estimates of methane output from mud extrusions at the erosive convergent margin off Costa Rica. Marine Geology 225, 129–144. McCave, I.N., 1986. Local and global aspects of the bottom nepheloid layers in the world ocean. Netherlands Journal of Sea Research 20 (2/3), 167–181.

Mollenhauer, G., Schneider, R.R., Mu¨ller, P.J., SpieX, V., Wefer, G., 2002. Glacial/interglacial variability in the Benguela upwelling system: Spatial distribution and budgets of organic carbon accumulation. Global Biogeochemical Cycles 16 (4). Mollenhauer, G., Eglinton, T.I., Ohkouchi, N., Schneider, R.R., Mu¨ller, P.J., Grootes, P.M., Rullko¨tter, J., 2003. Asynchronous alkenone and foraminifera records from the Benguela Upwelling System. Geochimica et Cosmochimica Acta 67 (12), 2157–2171. Moran, S.B., Charette, M.A., Pike, S.M., Wicklund, C.A., 1999. Differences in seawater particulate organic carbon concentration in samples collected using small- and large-volume methods: the importance of DOC adsorption to the filter blank. Marine Chemistry 67, 33–42. Muir, G.K.P., Pates, J.M., Karageorgis, A.P., Kaberi, H., 2005. 234Th:238U disequilibrium as an indicator of sediment resuspension in Thermaikos Gulf, northwestern Aegean Sea. Continental Shelf Research 25 (19–20), 2476–2490. Rutgers van der Loeff, M.M., Boudreau, B.P., 1997. The effect of resuspension on chemical exchanges at the sediment–water interface in the deep sea—A modelling and natural radiotracer approach. Journal of Marine Systems 11 (3–4), 305–342. Rutgers van der Loeff, M.M., Moore, W.S., 1999. Determination of natural radioactive tracers. In: Grasshoff, K., Ehrhardt, M., Kremling, K. (Eds.), Methods of Seawater Analysis. Verlag Chemie, Weinheim, pp. 365–397. Rutgers van der Loeff, M.M., Meyer, R., Rudels, B., Rachor, E., 2002. Resuspension and particle transport in the Benthic Nepheloid Layer in and near Fram Strait in relation to faunal abundances and 234Th depletion. Deep-Sea Research Part I 49, 1941–1958. Santschi, P.H., Guo, L., Walsh, I.D., Quigley, M.S., Baskaran, M., 1999. Boundary exchange and scavenging of radionuclides in continental margin waters of the Middle Atlantic Bight: implications for organic carbon fluxes. Continental Shelf Research 19 (5), 609–636. Sauter, E.J., Schlu¨ter, M., Wegner, J., Labahn, E., 2005. A routine device for high resolution bottom water sampling. Journal of Sea Research 54 (3), 204–210. Savoye, N., Buesseler, K.O., Cardinal, D., Dehairs, F., 2004. 234 Th deficit and excess in the Southern Ocean during spring 2001: particle export and remineralization. Geophysical Research Letters 31 (12). Schmidt, S., van Weering, T.C.E., Reyss, J.L., van Beek, P., 2002. Seasonal deposition and reworking at the sediment–water interface on the northwestern Iberian margin. Progress in Oceanography 52 (2–4), 331–348. Seiter, K., Hensen, C., Zabel, M., 2005. Benthic carbon mineralization on a global scale. Global Biogeochemical Cycles 19. Smoak, J.M., Moore, W.S., Thunell, R.C., Shaw, T.J., 1999. Comparison of 234Th, 228Th, and 210Pb fluxes with fluxes of major sediment components in the Guaymas Basin, Gulf of California. Marine Chemistry 65 (3–4), 177–194. Thomsen, L., McCave, I.N., 2000. Aggregation processes in the benthic boundary layer at the Celtic Sea continental margin. Deep-Sea Research Part I 47 (8), 1389–1404. Thomsen, L., van Weering, T.C.E., 1998. Spatial and temporal variability of particulate matter in the benthic boundary layer

ARTICLE IN PRESS M. Inthorn et al. / Deep-Sea Research I 53 (2006) 1742–1761 at the N.W. European Continental Margin (Goban Spur). Progress in Oceanography 42, 61–76. Turnewitsch, R., Springer, B., 2001. Do bottom mixed layers influence 234Th dynamics in the abyssal near-bottom water column? Deep-Sea Research Part II 48, 1279–1307. Usbeck, R., Rutgers van der Loeff, M.M., Hoppema, M., Schiltzer, R., 2002. Shallow mineralization in the Weddell Gyre. Geochemistry. Geophysics, Geosystems 3(1), 1008, doi:10.1029/2001GC000182. Usbeck, R., Schlitzer, R., Fischer, G., Wefer, G., 2003. Particle fluxes in the ocean: comparison of sediment trap data with

1761

results from inverse modeling. Journal of Marine Systems 39 (3–4), 167–183. Waples, J.T., Benitez-Nelson, C., Savoye, N., Rutgers van der Loeff, M., Baskaran, M., Gustafsson, O¨., 2006. An introduction to the application and future use of 234Th in aquatic systems. Marine Chemistry 100 (3/4), 166–189. Wefer, G., Fischer, G., 1993. Seasonal patterns of vertical particle flux in equatorial and coastal upwelling areas of the eastern Atlantic. Deep-Sea Research Part I 40 (8), 1613–1645.