234Th deficit and excess were examined in the upper 500 m of the Southern Ocean from Sub-Antarctic to Seasonal Ice Zones (Australian sector) during austral spring 2001. 234Th fluxes at 100 m indicate that particle export was low in the North (46.9–51.0°S), minimal in the Polar and Inter-Polar Frontal Zones and high in the South (≥61°S). These results are in tight agreement with new production estimates from the same cruise. Our data indicate that Polar and Inter-Polar Frontal Zones were not zones of intense export in the Australian sector at this time of year, in contrast with other sectors of the Southern Ocean. Also, we highlight the usefulness of 234Th excess below the mixed layer as a tool to study mesopelagic remineralization.
 The radioisotope 234Th is produced in ocean water from 238U decay. Because of 234Th's short half-life (24.1d) and its strong adsorption on particles, its deficit with respect to 238U has been used as a powerful tool to study biogenic particle export from surface ocean [Coale and Bruland, 1985]. In contrast, 234Th excess with respect to 238U has been only occasionally reported and explained as the result of remineralization process [e.g., Usbeck et al., 2002].
 The Southern Ocean plays a key role in the global climate. Indeed, global modeling output indicates that it acts as a net sink for atmospheric CO2, mainly due to phytoplankton uptake [Takahashi et al., 2002]. Thus, studying biogenic particle export and remineralization in the upper few hundred meters of the Southern Ocean water column is important to improve our understanding of atmospheric CO2 sequestration by the global ocean.
 Here we present a study of 234Th:238U disequilibria performed in the upper 500 m of the Southern Ocean during spring 2001 (WOCE SR3 transect, Australian sector). We report 234Th fluxes along the transect, highlight 234Th excess as a tool for studying particle remineralization in subsurface water, and discuss the particle export variability in the Southern Ocean.
2. Experimental and Methods
234Th:238U disequilibrium was studied during the CLIVAR SR3 cruise (aa0103 30 October–14 December 2001). Seven stations were sampled between 46.9°S and 64.9°S; stations at 63.9 and 60.9°S were re-visited after a delay of 11 and 17 days, respectively. The location of the stations as well as the major fronts and zones and their acronyms are indicated on Figure 1. The station at 63.9°S was located within sea ice (first and second visits), while the station at 64.9°S was ice-free.
 The upper 125 to 500 m were sampled for total (i.e., dissolved + particulate) 234Th using Niskin bottles, typically from 12 depths. Total 234Th was processed following Pike et al. . Briefly, 230Th was added as a yield monitor to a 4L-seawater sample. After equilibration, Mn precipitate was formed by addition of KMnO4 and MnCl2 to scavenge Th and filtered on quartz fiber filter. Beta radioactivity was measured on board generally over a 12–24 hour counting period using low level beta counters (Risø, Denmark). Filters were re-counted after 6 months for beta background, after which 230Th was analyzed by ICP-MS and Th yield calculated. Average Th recovery was 91%. The final relative uncertainty on total decay corrected 234Th activity was 2%. 238U activity was calculated from salinity following Chen et al. . Total accuracy of the method was evaluated by comparison to deep water samples where 234Th = 238U.
P: net loss of 234Th on sinking particles (i.e., vertical 234Th flux); i: layer of water column; n: number of layers of interest; λ: decay constant for 234Th (λ = 0.0288d−1); z: layer thickness; 238U: activity of 238U; 234Th: activity of total 234Th; t1, t2: time of the first and second visit of a station, respectively. Uncertainties on fluxes at 100 m and 300 m were 50 and 120 dpm/m2/d, respectively. We assume in these models that supply and loss of 234Th via horizontal and vertical mixing are small, relative to the rates of 234Th production from 238U, its decay, and loss on sinking particles. Physical processes have only been found to impact 234Th activity balances significantly in sites of intense upwelling or coastal regions where tidal pumping and activity gradients are a larger component of the 234Th activity balance [Buesseler, 1998]. Using the ratio of particulate C to 234Th, one can also derive POC flux from the 234Th flux but we leave this discussion to a forthcoming article where this approach will be more fully evaluated and compared to other estimates of C export.
3. 234Th Activity
 The lowest and highest 234Th activities were measured in surface and subsurface water of southern stations, respectively (Figure 2). With respect to 238U, there was a deficit of 234Th in the upper water column; it increased and deepened between the first and second visits at 60.9 and 63.9°S. As the 234Th deficit is due to exported particles which remove adsorbed 234Th [Coale and Bruland, 1985], it defines an “export layer” where 234Th < 238U activity. The export layer depth varied from 75 m at 63.9°S (SIZ)—where sea ice was melting—to 200 m in the PFZ.
 Below the export layer a zone of 234Th excess, i.e., 234Th > 238U, was often observed. In some cases (e.g., AZ-S and SIZ, first visits), 234Th excess was observed at depths as deep as 400–500 m. 234Th excess at similar depths has been reported in previous studies but only few authors have interpreted these data since (1) particle export from the surface layer is usually of primary interest and (2) previous methods did not allow for high vertical resolution needed to resolve excess 234Th. Rutgers van der Loeff et al.  reported a large 234Th excess in surface water near the retreating ice edge and under the ice. These authors argued that this was related to 234Th accumulation in the ice (adsorption on phytoplankton and/or ice) and release during melting. Usbeck et al.  reported 234Th excess between 100 and 400 m within the Weddell Gyre and interpreted it as being induced by remineralization of particles exported from the surface layer. Buesseler et al.  report a persistent 234Th excess at 80–120 m in Fe-depleted water during SOFeX. These authors attribute this to release of 234Th to the dissolved phase from sinking particles or disaggregation of sinking into suspended material.
 During the CLIVAR SR3 cruise, the transition from the surface export layer to deeper waters is defined by the gradient in 234Th activity from deficit to excess (Figure 2). This gradient was either small and stretched over a quite broad depth interval (e.g., at 56.9°S) or sharp and stretched over a narrow depth interval (e.g., at 63.9°S). 234Th gradients superimposed nicely onto density gradients (Figure 2) and were usually very pronounced in the south and smoother in the north. The 234Th increase below the export layer indicates that this density gradient defines the depth at which net particle export on sinking particles exceeds its supply.
 Biogenic particulate barium (Baxs) in mesopelagic waters is used as a proxy of remineralization in mesopelagic waters [Dehairs et al., 1997]. Baxs was studied during the CLIVAR SR3 cruise (D. Cardinal et al., Particulate Ba distributions and fluxes suggest latitudinal variations of carbon remineralization in the Southern Ocean, submitted to Deep-Sea Res. I, 2003, hereinafter referred to as Cardinal et al., submitted manuscript, 2003) (Figure 2). At most of the stations, Baxs increased from 50–100 m to a maximum at 200–300 m. Baxs and 234Th gradients superpose (e.g., at 60.9°S, second visit). This feature was not observed in the SAZ, however 234Th was sampled only in the first upper 250 m, i.e., above the Baxs increase. In addition, O2 concentration decrease coincides with Baxs and 234Th increase in and south of PFZ (not shown), i.e., where mesopelagic remineralization seems to be higher.
 From the above it appears that (1) 234Th excess starts within or just below the density gradient at the bottom of the upper mixed layer, (2) relative to particle export, particle break-up and remineralization were intense within the first 300–500 m along the transect, especially from IPFZ to SIZ, and (3) 234Th excess is a powerful indicator of remineralization processes in the mesopelagic zone. Estimates of remineralization flux give values ranging between 60 ± 90 dpm/m2/d at 46.9°S and 3800 ± 200 dpm/m2/d at 60.9°S with highest values in the AZ-S and SIZ. However, except for the two northern stations, these values may be underestimates since the deepest samples of the profiles were not always at equilibrium, indicating that remineralization process may still occur at deeper depths. Thus, for future studies, we recommend sampling the upper 500–600 m with a high vertical resolution in order to test the use of 234Th excess as a proxy of mesopelagic particle break-up and remineralization.
4. 234Th Fluxes
4.1. Steady Versus Non-Steady State Models
 The steady state model (Equation 1) indicates that the lowest 234Th fluxes at 100 m were observed in the central part of the study area (PFZ and IPFZ) and during the first visit of the AZ-S (Figure 3). The highest values correspond to the second visit of the AZ-S and to the station at 63.9°S. After ca two weeks the 100 m-flux increased slightly at 63.9°S, and increased by a factor of ca 4 at 60.9°S. Except at 46.6 and 53.7°S, fluxes at 300 m (Figure 3) are much smaller than at 100 m and even negative, due to particle break-up and remineralization.
 The main assumption of the steady state model (Equation 1) is that there is no change over time in 234Th activity (and consequently on 234Th fluxes) at least at the time scale of 234Th (t1/2 = 24d). This is clearly not the case at 60.9°S and to a lesser extent at 63.9°S which show considerable change in the 234Th flux over ca two weeks. Furthermore, under steady state conditions mesopelagic remineralization should at most be as large as surface export, i.e., export fluxes at depth should be nil or positive. The 300 m-flux is highly negative at 60.9°S and 64.9°S, indicating that surface export is greater than remineralization, which suggests that the steady state model is not valid for predicting export fluxes at these stations. In contrast, for the northern stations, the steady state approach seems valid since export and remineralization are in balance. This situation is consistent with our earlier findings [Savoye et al., 2004] suggesting that the southern part of this transect acts as a pulsed production system contrasting with a more stable system in the north.
 The repeat visits at 60.9 and 63.9°S after a delay of 11 and 17 days, respectively, allow calculation of the 234Th flux using a non-steady state model (Equation 2). The 100 m-fluxes were ca 3000 dpm m−2 d−1, greatly exceeding those obtained with the steady state model. In the following discussion and in Table 1, we use the steady state flux values from SAF to IPFZ and the non-steady state flux values in AZ-S and SIZ.
Table 1. 234Th Flux (dpm/m2/d) at 100 m (#: 200 m) in the Southern Ocean From Steady-State and Non Steady-State Models
PFZ, IPFZ, PF
SAZ, SAF, PFZ, IPFZ, AZ-S: see Figure 1; STZ: Sub-Tropical Zone; PF: Polar Front. ‘/’ indicates a range of values.
 During spring 2001 234Th fluxes were low in the north (ca 630 dpm/m2/d), minimal in the Polar and Inter Polar Frontal Zones (ca 300 dpm/m2/d) and high in the south (ca 3000 dpm/m2/d). The latter were similar to fluxes reported by Buesseler et al.  at 170°W but much higher than fluxes reported by Rutgers van der Loeff et al.  in the Atlantic sector for early spring conditions (Table 1). In contrast, from the SAZ to the IPFZ, 234Th fluxes were much smaller in our study than generally reported in the literature, being similar only with early spring values [Rutgers van der Loeff et al., 1997]. Our 234Th fluxes indicate low particle export in the SAZ and SAF, very low particle export in the PFZ and IPFZ and high export in the AZ-S and SIZ during spring 2001. Interestingly, the phytoplankton assemblage was dominated by small cells (picoplankton and flagellates) from SAZ to IPFZ and by large cells (diatoms and dinoflagellates) in AZ-S and SIZ [Savoye et al., 2004], suggesting that large phytoplankton contributed more efficiently to particle export.
 Latitudinal variations of 234Th fluxes were in tight agreement with variations in new production (Figure 3)—estimated for the same cruise from f-ratio and primary production—which is considered as ‘exportable’ production. 234Th data and new production values both indicate that PFZ and IPFZ were the regions of lowest export, contrasting with the widely accepted idea that the Polar Front favours export [e.g., Usbeck et al., 2002]. This could be specific to the Australian sector because of hydrodynamic conditions: in the Australian sector (140°E), Polar Front divides into two branches 4–5° apart from each other [Trull et al., 2001]. Elsewhere, the Polar Front branches are located closer to each other [Pollard et al., 2002].
 Authors thank the crew of the R/V Aurora Australis for assistance on board, Steve Pike and John Andrews for assistance with 234Th analyses and Steve Rintoul as chief scientist. This work was supported by the OSTC PODOII Programme on Global Change, Ecosystems and Biodiversity, Brussels, Belgium (Contracts EV/03/7A and EV/37/7C). K.O. Buesseler was supported by WHOI Ocean Life Institute Fellowship and US National Science Foundation. Shiptime for this work was provided via Australian Antarctic Science grants #1343.