Use of Ra isotopes to deduce rapid transfer of sediment-derived inputs off Kerguelen

The Southern Ocean is known to be the largest high-nutrient, low-chlorophyll (HNLC) region of the global ocean due to iron limitation. However, a large phytoplankton bloom develops annually downstream of the Kerguelen Islands, a bloom which is sustained partly by iron released from the sediments deposited onto the shelves. In the framework of the KEOPS-2 project, we used radium isotopes (Ra, T1/2 = 3.66 d; Ra, T1/2 = 11.4 d; Ra, T1/2 = 5.75 yr) to provide information on the origin of iron fertilization and on the timescales of the transfer of sedimentderived inputs (including iron and other micronutrients) towards offshore waters. Significant Ra and Ra activities were found in the near vicinity of the Kerguelen Islands, in agreement with the short half-lives of these isotopes. Significant Ra and Ra activities were also detected up to 200 km downstream of the islands and more unexpectedly in offshore waters south of the polar front. These observations thus clearly indicate (i) that the sediment-derived inputs are rapidly transferred towards offshore waters (on timescales on the order of several days up to several weeks) and (ii) that the polar front is not a physical barrier for the chemical elements released from the sediments of the Kerguelen Plateau. The Ra data set suggests that iron and other micronutrients released by the shelves of the Kerguelen Islands may contribute to fueling the phytoplankton bloom downstream of the islands, despite the presence of the polar front. However, the heterogeneous distribution of the Ra and Ra activities in surface waters suggests that this supply across the front is not a continuous process but rather a process that is highly variable in space and time.


Introduction
The Southern Ocean is recognized as the major high-nutrient, low-chlorophyll (HNLC) region of the global ocean.Despite high nutrient concentrations, the phytoplankton growth was shown to be limited by the very low iron concentrations in surface waters of the Southern Ocean (De Baar et al., 1995;Martin et al., 1990).Dissolved iron is, however, supplied to the surface waters in several locations of the Southern Ocean where iron is released by the shelf sediments, but this natural iron fertilization remains spatially limited (Tagliabue et al., 2014).Consequently, high phytoplankton biomass can be found offshore the Antarctic continental shelf (Arrigo et al., 2008;Moore and Abbott, 2002) or in the vicinity of subantarctic islands (Blain et al., 2001;Korb et al., 2004;Pollard et al., 2007).
One of the largest phytoplankton blooms is observed offshore of the Kerguelen Islands, in the Indian sector of the Published by Copernicus Publications on behalf of the European Geosciences Union.
V. Sanial et al.: Use of Ra isotopes to deduce rapid transfer of sediment-derived inputs Southern Ocean (Blain et al., 2001).This phytoplankton bloom extends more than 1000 km downstream of the Kerguelen Islands and shows two main features: (i) a plume that extends northeast from the islands and north of the polar front (PF) that shows high mesoscale and temporal variability, and (ii) a larger bloom to the southeast of the islands and south of the PF (Blain et al., 2001(Blain et al., , 2007)).The two areas are separated by a narrow band of relatively low chlorophyll concentration associated with the PF that follows the inner shelf edge between 200 and 500 m isobaths (Park and Gamberoni, 1997;Park et al., 1998b).While Park et al. (2008a) suggest that the northward geostrophic flow associated with the PF may possibly block any southward penetration of lithogenic inputs released by the Kerguelen Islands, the numerous eddies and meanders formed along the PF may contribute to transport of chemical elements between the northern Kerguelen Plateau and offshore waters.
The KEOPS-2 (KErguelen Ocean and Plateau compared Study) project aimed at understanding the circulation patterns off the Kerguelen Islands and the mechanisms of iron fertilization in that area.The KEOPS-2 cruise was conducted during austral spring 2011 to the east of the Kerguelen Islands.Natural radio-tracers such as radium isotopes ( 223 Ra, T 1/2 = 11.4 d; 224 Ra, T 1/2 = 3.66 d; 228 Ra, T 1/2 = 5.75 yr) have already been proved to be powerful tools to track the origin and fate of chemical elements -including iron and other micronutrients -which are released by the sediments deposited on the shelves (Annett et al., 2013;van Beek et al., 2008;Charette et al., 2007;Dulaiova et al., 2009;Sanial et al., 2014).In this work, we refer to these latter inputs as "sediment-derived inputs".Radium isotopes are produced by the decay of particle-bound thorium isotopes in the sediments and are delivered to the open ocean by diffusion and advection processes.Thus, a water mass that interacts with shelf sediments is potentially enriched in radium and in other elements that also diffuse out of the sediments (e.g., iron and other micronutrients).While iron may then be removed from the water column by biotic or abiotic processes, radium behaves as a conservative tracer.Radium is only affected by radioactive decay and mixing in such a way that the water body keeps the signature of its contact with the sediments.The radium signature of a given water mass may then be transferred by diffusion and advection towards offshore waters.The presence of significant Ra activities in offshore waters thus indicates that the water body has interacted with shallow sediments.Alternatively, vertical mixing may also transport Ra towards surface waters.Because radium isotopes decay, they can be used as chronometers to estimate the time elapsed since the water body left the shelf, which in turn gives information on how quickly the microelements released by the shallow sediments may be transferred to offshore waters (Moore, 2000).In this work, we examined the distribution of 223 Ra, 224 Ra and 228 Ra in surface waters downstream of the Kerguelen Islands in order (i) to investigate the origin and dispersion of the sediment-derived inputs, including iron, and (ii) to determine the apparent ages of offshore waters that provide information on the timescales of the transfer of water and associated chemical elements between the shelves and offshore waters.In addition to the Ra distribution in surface waters, we report several vertical profiles of 223 Ra, 224 Ra and 228 Ra that provide constraints on the vertical transport of chemical elements associated with vertical mixing.

The KEOPS-2 project
The KEOPS-2 cruise took place east of the Kerguelen Islands (northern Kerguelen Plateau) between 14 October and 23 November 2011 onboard the R/V Marion Dufresne (IPEV: Institut Polaire Français -Paul Emile Victor; TAAF: Terres Australes et Antarctiques Francaises).The KEOPS-2 project was designed to study the mechanisms of natural iron fertilization downstream of the Kerguelen Islands and its impact on ecosystems and biogeochemical cycles.The KEOPS-2 project was labeled as a GEOTRACES process study and followed up a first KEOPS project conducted in 2005 in the area of the southern Kerguelen Plateau (Blain et al., 2007).

Study area
The Kerguelen Plateau, located in the Indian sector of the Southern Ocean, constitutes one of the few physical barriers for the eastward-flowing Antarctic Circumpolar Current (ACC).Various studies provide a detailed description of the general ocean circulation patterns around the Kerguelen Plateau (Charrassin et al., 2004;Park and Gambéroni, 1995;Park et al., 1998Park et al., , 2008bPark et al., , 2009)).An important oceanographic feature of the area is the presence of the PF, which is commonly characterized by the northernmost position of a subsurface temperature minimum bounded by the 2 • C isotherm (Belkin and Gordon, 1996;Park and Gamberoni, 1997;Park et al., 1993).A strong eastward current associated with this front is deflected to the north at 71 • E following the eastern shelf slope of the Kerguelen Plateau between the 200 and 500 m isobaths and forms a cyclonic meander that turns southward at 75 • E (Belkin and Gordon, 1996;Orsi et al., 1995;Park and Gamberoni, 1997;Park et al., 1993;Pollard et al., 2002).Numerous eddies are generated along the PF to the east of the Kerguelen Plateau that can, in some cases, be identified on the satellite composite images of sea surface chlorophyll.The locations of the stations investigated in this study are shown in Fig. 1.

Sample collection
Surface seawater samples were collected at 7 m depth using a clean pump specially designed by IPEV for the KEOPS-2 cruise.Large volumes of surface seawater were collected (250-900 L) and stored in large plastic tanks.We used a CTD (SBE-19plus, Sea-Bird ® ) and a rosette system equipped with 22 × 12 L Niskin bottles to collect seawater samples from various depths throughout the water column.Three samples were also collected directly on two beaches of the Kerguelen Islands (Baie du Morbihan: samples KER-1; Baie des Baleiniers: samples BaieB-1 and BaieB-2).Seawater samples were then passed by gravity through PVC cartridges filled with "Mn fibers" (MnO 2 -impregnated acrylic fiber) following Moore (2008).The flow rate was fixed at ≤ 0.5 L min −1 to provide 100 % extraction efficiency (Moore, 2008;van Beek et al., 2010).The Mn fibers were then rinsed with Milli-Q water and partially dried before analysis.

Sample analysis
The Mn fibers were analyzed using a radium delayed coincidence counter (RaDeCC; Charette et al., 2001;Moore and Arnold, 1996;Moore, 2008).Three counting sessions are necessary to determine both excess 224 Ra and excess 223 Ra activities in the samples.The first counting was performed onboard the research vessel during the cruise and provides the total 224 Ra and 223 Ra activities.The Mn fibers were analyzed again 3 weeks after sampling to determine the 224 Ra activities supported by 228 Th and then after 3 months to determine the 223 Ra activities supported by 227 Ac (Moore, 2000).The 224 Ra activities are corrected for the 224 Ra supported by 228 Th and the 223 Ra activities are corrected for the 223 Ra supported by 227 Ac.The 224 Ra and 223 Ra activities discussed hereafter thus refer to these excess 224 Ra and 223 Ra activities.Uncertainties for both isotopes were calculated following Garcia-Solsona et al. (2008) and were reported with a 1σ confidence interval.
Activities of 228 Ra were then determined using the lowbackground gamma detectors placed at the LAFARA underground laboratory in the French Pyrénées (van Beek et al., 2010(van Beek et al., , 2013)).Mn fibers were either ashed at 820 • C for 16 h (Charette et al., 2001) and analyzed using a well-type germanium detector or compressed and analyzed using a semiplanar detector.Cross-calibrations between the two detectors were made to avoid any bias in the determination of the Ra activities.Each sample was analyzed for ca.120 h to allow for the quantification of the low 228 Ra activities present in Southern Ocean waters (Kaufman et al., 1973). 228Ra activities were determined using 228 Ac peaks (338, 911 and 969 keV).All radium activities are reported in disintegration per minute per 100 L of seawater (dpm 100 L −1 ).The uncertainties reported for gamma counting consist in the error associated with counting statistics (1σ ).

Color data
High-resolution maps (1/25 • × 1/25 • ) of chlorophyll concentration (mg m −3 ) were constructed by a 10-day weighted mean of MODIS and MERIS measurements.These satellite products were delivered 3 times a week in near-real time during the cruise from Ssalto/Duacs and CLS (Collecte Localisation Satellites, Toulouse, France) with support from CNES (Centre National d'Etudes Spatiales, France).These images were used to define the sampling strategy in the investigated area.

Surface drifters
Drifters provided by the US National Ocean and Atmospheric Administration (NOAA) Global Drifter Program (GDP) were also released.The drogue is centered at 30 m depth.These drifters thus provide information on the mean currents in the surface mixed layer and on the dispersion of water masses due to eddy activities.Successive positions of the drifter were transmitted to the R/V Marion Dufresne four times a day by the NOAA GDP center.The time-irregular positions of the drifter were interpolated into a regular time step of 12 min and a low-pass filter of 48 h was then applied to filter all tidal currents and inertial oscillations.

Lagrangian particle analysis
The Lagrangian particle analysis was based on total surface currents, which are the sum of the absolute geostrophic V. Sanial et al.: Use of Ra isotopes to deduce rapid transfer of sediment-derived inputs currents (deduced from altimeter product) and Ekman currents (daily mean).The Ekman component is deduced from the European Centre for Medium-Range Weather Forecasts (ECMWF) wind stress analysis applying a regional Ekman model, specifically adjusted for the Kerguelen area.The altimeter current products were produced by Ssalto/Duacs and distributed by AVISO, with support from CNES.Total surface currents were delivered every day with a 1/8

Lagrangian model
The altimetry-derived velocities providing the geostrophic mesoscale velocity at the ocean surface were analyzed in near-real time with a Lagrangian model.This model was inspired by Mongin et al. (2009), who reconstructed the extension of the Kerguelen chlorophyll plume with a transport scheme based on altimetry.The model created thousands of virtual surface drifters released on the shelf break of Kerguelen (2000 m isobaths; apparent age= 0).The trajectories were constructed by backward-in-time integration of the altimetric velocity field and were stopped when a hit over the Kerguelen shelf break was detected (indicating a trajectory coming from the shelf) or when a maximum integration time -set to 120 days -was reached (indicating no interaction with the shelf on the past 120 days).This model was applied successfully by Sanial et al. (2014) to highlight the key role played by surface horizontal transport in defining the extension of the springtime chlorophyll plume in the Crozet area.

Hydrological context during the KEOPS-2 cruise
The KEOPS-2 cruise lasted almost 2 months (October-November 2011).During that period, the phytoplankton bloom developed off the Kerguelen Islands (Fig. 2).The satellite composite images of sea surface chlorophyll reveal a complex shape of the phytoplankton bloom that may be associated with the complex hydrography of the area.A high concentration of chlorophyll first appeared close to the Kerguelen Islands (October 2011) before spreading out in offshore waters until covering a large part of the study area at the end of November 2011.East of the Kerguelen Islands, a narrow band of low chlorophyll concentration is associated with the northward branch of the PF that splits the phytoplankton bloom into two parts.
The PF also delimits two surface water masses characterized by a strong contrast in temperature and salinity; the Antarctic Surface Water (AASW) is located south of the PF, and the Subantarctic Surface Water (SASW) is located north of the PF (Emery and Meincke, 1986).The potentialtemperature-salinity diagrams of the water masses investigated in this study are shown in Fig. 3.The SASW is iden-tified only at station F-L, suggesting that this station is located north of the PF.The Winter Water (WW), a typical feature of the Antarctic zone, is the winter remnant of the Antarctic Surface Water characterized by a subsurface temperature minimum layer around 200 m depth (Park et al., 1998a(Park et al., , 2008b(Park et al., , 2014)).The WW is found on all the vertical profiles reported here except for station F-L, thus confirming its location north of the PF.Below the WW, three water masses can be identified: the Upper Circumpolar Deep Water (UCDW), the Lower Circumpolar Deep Water (LCDW) and the Antarctic Bottom Water (AABW) (Park et al., 1993(Park et al., , 2008b)).Note that the AABW is only found on the F-L profile (commonly observed below 2600 m in this area; Park et al., 2008b).

Radium distribution in surface waters
The radium activities reported in this study are shown in Table 1 and fall in the range of previous radium data reported for surface waters near islands of the Southern Ocean (Annett et al., 2013;Charette et al., 2007;Dulaiova et al., 2009;Hanfland, 2002;Kaufman et al., 1973;Sanial et al., 2014;van Beek et al., 2008).The highest 223 Ra, 224 Ra and 228 Ra activities are found in seawater samples collected at shallow stations near the Kerguelen Islands (bathymetry < 200 m; Fig. 4).The radium activities then gradually decrease offshore.Several samples, however, display significant 224 Ra activities in samples collected offshore (Fig. 4a): stations UW-21-23-34 and TEW-7 located along the PF; stations UW-32, E-1 and TEW-5 south of the PF; and station TNS-1 north of the PF.A greater number of offshore stations exhibit significant 223 Ra activities, which agrees with the longer half-life of the 223 Ra isotope (Fig. 4b).The stations displaying significant 224 Ra activities also display significant 223 Ra activities.The radium activities are especially high at station TNS-2, located north of the PF, and at stations E-1 and G-1, located south of the PF.Station G-2 was visited twice and showed high 223 Ra and 224 Ra activities on both visits.Station A3 located on the southern Kerguelen Plateau was also visited twice.Significant 223 Ra and 224 Ra activities were determined in the water sample collected at station A3-1 during the first visit at station A3 (note, however, that these activities are low) but were both below the detection limit at station A3-2 (second visit at station A3).In contrast, the 228 Ra activities are similar for the two visits to A3 (Table 1).All surface samples display significant 228 Ra activities up to ca. 300 km offshore from the Kerguelen Islands (i.e., station TEW-8).Relatively high values are observed at stations TNS-2 and UW-32, located north and south of the PF, respectively (Fig. 4c).Station R-2, which was chosen as the reference station for typical HNLC waters east of the Kerguelen Islands, shows significant 223 Ra, 224 Ra and 228 Ra activities in surface waters.

Vertical distribution of Ra isotopes
The study of the vertical distribution of Ra isotopes allows us to provide constraints on the vertical transport of Ra associated with vertical mixing.Consequently, these profiles help us to define the origin of the Ra enrichments observed in surface waters off the Kerguelen Islands (lateral versus vertical supply of Ra).The major water masses identified with the potential-temperature-salinity diagrams throughout the water column are reported for each profile.The shallow Ra profiles (stations TEW-3, G-1 and A3-2) are shown in Fig. 5 and the deep profiles (stations F-L, E-4W and E-1) are shown in Fig. 6.
The 223 Ra and 224 Ra activities are usually higher in samples collected near the seafloor and are below the detection limit at intermediate depths (Table 2; Figs. 5 and 6).Significant 223 Ra and 224 Ra activities are observed in surface and/or subsurface waters several kilometers offshore from the islands, in particular at stations G-1 and E-1, located south of the PF, and at station F-L, located north of the PF.The vertical profiles of 223 Ra and 224 Ra are quite unique at station F-L.Although (i) this station is located far from the Kerguelen Islands and (ii) the bottom is at 2670 m depth, the 223 Ra and 224 Ra activities are relatively high throughout the water column (Fig. 6).Significant 228 Ra activities were found in the different water columns investigated in this study.The 228 Ra activities at stations TEW-3 and G-1 are relatively high and uniform throughout the water column.The 228 Ra activities at station A3 are uniform in the upper 250 m and then increase with increasing water depth.The vertical 228 Ra profiles at the deep stations (F-L, E-1 and E-4W) exhibit an in- creasing trend with increasing depth reflecting the diffusion of radium out of the sediments.This latter pattern is also especially marked at station A3 on the southern Kerguelen Plateau (Fig. 5).

Origin of the radium enrichments in surface waters
The relatively high radium activities ( 223 Ra, 224 Ra and 228 Ra) observed in surface waters east of the Kerguelen Islands may be explained either by the vertical transport or diffusion that supplies radium to surface waters or by the lateral advection of waters that have recently interacted with shallow sediments (Blain et al., 2001;Park et al., 2008a;van Beek et al., 2008).When considering solely the 228 Ra vertical profileswhich show in most cases an increase in 228 Ra activities with increasing depth -it cannot be excluded that the vertical mixing contributes to an increase in radium activities in surface waters.However, the 224 Ra and 223 Ra vertical profileswhich show higher Ra activities in the upper and in the deep water column but Ra activities below the detection limit in the mid-water column -clearly indicate that the higher 224 Ra and 223 Ra activities in surface waters cannot be explained by vertical mixing.The 224 Ra and 223 Ra enrichments in surface waters are thus more likely explained by the lateral advection of waters that have recently interacted with shallow sediments.
The northward advection of a water mass that has interacted with the shallow sediments deposited on the shelves of Heard Island has been identified as a pathway for the micronutrients that sustain the phytoplankton bloom on the southern Kerguelen Plateau (Chever et al., 2010;van Beek et al., 2008;Zhang et al., 2008).The presence of a chlorophyll plume that expands northward from the southern Kerguelen Plateau may also support the existence of this northward advection (Fig. 2).The observation of significant 224 Ra and 223 Ra activities in surface waters at station A3-1 confirms this circulation pattern and suggests that the transit time of the waters that interacted with the shelves of Heard Island may be < 1 month between Heard Island and station A-3.This is in agreement with the Ra data obtained in 2005 during the KEOPS-1 project, where significant 224 Ra and 223 Ra activities were also found in surface waters at station A3 (van Beek et al., unpublished data).When the waters move further north towards the area investigated in this study, the 224 Ra and 223 Ra activities will then continue to decay.Two drifters released during the KEOPS-2 cruise at station A3 allow us to provide constraints on the transit time between the southern Kerguelen Plateau and the studied area (east of Kerguelen at around 49 • S).A first drifter recirculated around station A3 nearly 20 days before it moved slowly northward.It took approximately 60-75 days for the drifter to reach the investigated area located to the east of the Kerguelen Islands (Fig. 7).It took approximately 53-65 days for the second drifter to reach the area to the east of Kerguelen.Such transit times agree with the estimate of Park et al. (2008b) during the KEOPS-1 project (i.e., several months between Heard Island and the eastern flank of the Kerguelen Islands).With such a transit time, a water body that interacted with the shelves of Heard Island should not contain any remaining short-lived radium isotopes when reaching the eastern flank of the Kerguelen Islands.As a consequence, the 224 Ra and 223 Ra activities found in offshore waters east of the Kerguelen Islands, south of the PF, are best explained by diffusion or advection of Ra via waters that recently interacted with the shallow sediments of the northern Kerguelen Plateau.This scenario, however, implies that the Ra isotopes (and potentially other chemical elements such as iron) were transferred offshore across or via the PF.High dissolved and particulate trace element concen- trations (Fe, Ni and Co) were also found east of the PF, confirming that chemical elements may be transported offshore across or via the PF (Quéroué et al. 2014;van der Merwe et al., 2014).Among the potential mechanisms allowing surface waters to be transported eastward across the PF, one can invoke either (i) the wind stress (eastward winds are especially strong in that region) or (ii) eddies that form along the PF and that could promote the passage of chemical elements across the front.
However, a contribution of surface waters originating from the southern Kerguelen Plateau may not be completely excluded.In contrast to 224 Ra and 223 Ra, which both disappear due to radioactive decay along the northward transport (> 2 months), 228 Ra with a longer half-life would remain in these waters.The 228 Ra activities observed to the east of the Kerguelen Islands may thus be partly explained by an advective transport of waters originating from the south.It cannot be excluded, therefore, that the northward advection origi- nating from the southern plateau contributes to the natural fertilization of the investigated area, in addition to the input of chemical elements across the PF that was shown by the short-lived isotopes.
South of the Kerguelen Islands (i.e., along the PF at stations UW-23 and UW-24 or south of the PF, e.g., at stations UW-15, UW-16, R-2; Fig. 4), it cannot be completely excluded that the observed radium enrichments are partly ex-plained by an input of radium associated with the Leclaire Rise, located west of the Kerguelen Islands at ca. 350 m depth (Weis and Frey, 2002).Station R-2, which is located east of the Leclaire Rise south of the PF, shows significant 223 Ra and 224 Ra activities in surface waters.Although these activities are relatively low (0.016 and 0.057 dpm 100 L −1 , respectively), they suggest that the waters downstream of the Leclaire Rise may be impacted by this topographic feature.However, sample UW-14 collected in surface waters lying above this rise does not show significant 223 Ra and 224 Ra activities and only low 228 Ra activity, which suggest that vertical mixing may not efficiently transport radium released by the shallow sediments towards surface waters above this topographic feature.Note that the influence of the Leclaire Rise on the chemical element concentrations downstream of the rise is also observed in Fe and other trace metal (REEs, Mn, Al) concentrations, but only in waters lying in the 200-500 m depth interval (Bowie et al., 2014;van der Merwe et al., 2014;Grenier et al., 2015).

Timescales of the offshore transport of surface waters
Once released into the water column, radium isotopes are subject to dilution, mixing and radioactive decay.The decay of short-lived radium isotopes in offshore waters provides information of how quickly chemical elements (including micronutrients) also released by the sediments are diluted and dispersed into the ocean (Moore, 2000).The presence of 224 Ra and 223 Ra in offshore waters thus indicates that the waters have recently been in interaction with the sediments.In contrast, when both 224 Ra and 223 Ra activities are below the detection limit, this suggests that the water bodies have not been in contact with the sediments over the past 2 months (this is represented in light gray in Fig. 8).The water samples that display significant 223 Ra activity but no 224 Ra (represented in dark gray in Fig. 8) suggest that the interaction between the water body and the sediment occurred between 1 month ( 224 Ra activities < DL) and 2 months ago (significant 223 Ra activities).When both the 224 Ra and 223 Ra activities were significant, apparent ages could be calculated following Moore (2000): In this study, we only reported the apparent ages deduced from the 224 Ra / 223 Ra ratios because we showed that both the 224 Ra and 223 Ra determined east of Kerguelen originate from the shallow sediments of the Kerguelen Islands (see Sect. 4.1).Apparent ages were thus calculated using an initial 224 Ra / 223 Ra ratio that was obtained by averaging the ratios found at stations located on the northern Kerguelen Plateau (< 200 m water depth).In contrast, we cannot exclude that 228 Ra has various origins (Kerguelen Islands and/or Heard Island).The use of the 224 Ra / 228 Ra or 223 Ra / 228 Ra ratios to derive apparent ages is thus compromised because it is not possible to determine a single initial ratio in this case.
Several offshore samples display a young apparent age (4-8 days), suggesting a rapid transport of radium between the shallow waters of the northern Kerguelen Plateau and offshore.Station TNS1, located north of the PF, is reached after 5 days.This observation agrees with the circulation pattern in this area, with waters flowing eastward and that may interact with the shallow northern Kerguelen Plateau (Park et al., 2014).This is also in agreement with the drifters launched during the KEOPS 2 project that also highlighted such advection along the PF (Fig. 7) (Zhou et al., 2014).Station UW-21, located ca.50 km offshore, and station E1 and station UW-32, located ca. 200 km offshore, also show relatively young apparent ages (4, 5 and 6 days, respectively).Because all these stations are located south of the PF, this suggests that the sediment-derived inputs may be rapidly transferred towards offshore waters across the PF.Station R-2, located south of the PF, also displays a young apparent age.At station Kerfix located close to station R-2, Jeandel et al. (1998) reported westward currents associated with a recirculation pattern that may transport chemical elements originating from the Kerguelen Plateau.The Ra signal may then be transported eastward, as suggested by the significant 223 Ra activities also observed east of station R-2, south of the PF (Figs. 4  and 8).Alternatively, the Leclaire Rise, located west of R-2, may impact the surface waters, thus leading to a young age for this water sample.Because the 224 Ra and 223 Ra activities found at station A3-1 were attributed to the northward advection on the southern Plateau, the apparent age at station A3-1 was calculated assuming that the initial 224 Ra / 223 Ra ratio off Heard Island is similar to that off Kerguelen Islands (Fig. 8).This hypothesis may be correct since the geological contexts of the two islands are similar.The apparent age thus calculated provides an estimate for the transit time of surface waters above the southern Kerguelen Plateau between Heard Island and station A3.However, during the second visit to station A3 (A3-2), the 224 Ra and 223 Ra activities were below the detection limit.This may highlight the temporal variability in the circulation patterns in this area: the transit time of surface waters between Heard Island and station A3 may thus vary with time, ranging from 1 week to 1-2 months.On such timescales, the 228 Ra activities do not significantly decay, which would explain why similar 228 Ra activities were found during the two visits to A3.Finally, the spatial variability in the distribution of the apparent ages in offshore waters suggests that the passage of chemical elements across the PF is a sporadic process, which may contribute to explaining the mosaic structure of the phytoplankton bloom.Future studies in the area could aim to track more precisely the sedimentary sources of radium (and other chemical elements) and to quantify the radium fluxes out of the sediments using, for example, the method developed by Cai et al. (2012).

Lagrangian particle analysis
To provide additional constraints on the origin of the Ra signal in offshore waters, Lagrangian analyses derived from total surface currents were conducted at several stations (Fig. 9).A 2-month backward analysis -to account for the life time of 223 Ra -was performed starting from the sampling date for targets that were centered on the station locations.
The Lagrangian analyses for the southern stations A3-1, A3-2, UW-35 and G-1 are reported in Fig. 9.The backward trajectories provide a similar pattern and indicate a southern origin for the surface waters found at these stations.This pattern agrees with the trajectories of the two drifters released in situ at station A3 (Fig. 7).Waters that have interacted with the shallow shelves of the southern Kerguelen Plateau (Heard Island) may thus reach the investigated area.In particular, this northward advection may explain the significant Ra activities determined at stations G-1 and A3-1.Note that the distance covered by the backward trajectories over the 2 months is short, thus reflecting the relatively slow currents in this area.
Lagrangian analyses were also performed for several northern stations: E-1, UW-31, UW-32, TEW-7 and F-L.The trajectories are represented in shades of red in Fig. 9. Stations F-L, TEW-7 and UW-31 are located relatively close to each other, east of the Kerguelen Islands in the area of the southern branch of the cyclonic meander formed by the PF.Their backward trajectories display a similar feature and all point to the same origin, which is the northern Kerguelen Plateau.This suggests that chemical elements originating from the Kerguelen Plateau may be transported offshore via the PF.The transit time given by the Lagrangian analysis is approximately 1 month between the coast of the Kerguelen Islands and the investigated stations.With a transit time of approximately 1 month, the 224 Ra activities should have disappeared due to radioactive decay -or should be close to the detection limit -while the 223 Ra activities should have significantly decayed.As a comparison, the 224 Ra and 223 Ra activities are below the detection limit at stations F-L and UW-31, whereas significant 223 Ra and 224 Ra activities were found at station TEW-7.Such a discrepancy between the investigated stations may highlight the spatial variability in the circulation patterns in this area or that the Ra activities are close to the detection limits.Both 224 Ra and 223 Ra activities are also significant at station E-1 located in the center of the cyclonic meander formed by the PF.The Lagrangian analysis suggests that the surface waters at station E-1 originate from the southwest.These waters flow northwards before reaching the PF area and then follow the eastern shelf of the northern Kerguelen Plateau.When passing close to the PF, these waters may receive significant Ra inputs (and potentially other sediment-derived inputs) that could be transported via or across the front in this area.This Ra signal may then be transferred to station E-1, as suggested by the backward trajectories.This hypothesis is also supported by the study conducted by Zhou et al. (2014), who identified a northeastward drift of surface waters originating from the Kerguelen Plateau.Finally, the backward trajectories at station UW-32 -which also displayed significant 223 Ra and 224 Ra -highlight the spatial variability in that area: while several trajectories originate from the south, several other trajectories follow the PF and the shelves of the northern Ker- guelen Plateau, where these waters could also potentially receive sediment-derived inputs.

Comparison of the apparent radium ages with an altimetry-based Lagrangian model
The timescale of the offshore transport of surface waters was also investigated using an altimetry-based Lagrangian model (Fig. 10).The color bar indicates the time (number of days) elapsed since the water body left the 2000 m isobath.A color code similar to that of Fig. 8 was used.A color palette from red to yellow highlights the rapid offshore transport of the surface waters (surface waters < 6 days).The dark-gray coding illustrates surface waters that left the 2000 m isobath less than 1 month ago.Finally, surface waters that left the 2000 m isobath more than 1 month ago are represented in light gray.As a comparison, the radium apparent ages are reported on the map using the same color code.Young ages can be found close to the 2000 m isobath, along the PF.Surface waters < 1 month follow the cyclonic meander formed by the PF, while waters older than 2 months are found in the center of the meander.Note that the altimetry-derived Lagrangian analysis may misplace structures with errors of ∼ 10 km (e.g., d 'Ovidio et al., 2010), which is comparable to the width of the structures visible in the map.It may thus be difficult to compare quantitatively the altimetry-derived ages with the ages determined in situ.Nevertheless, two important considerations can be made: (i) the order of magnitude of the satellite-derived and in situ ages are consistent in the region, and (ii) considering a west-east transect from Kerguelen, both estimations indicate a transition from young to old and then once again young ages,

Conclusions
The observation of short-lived Ra isotopes ( 223 Ra and 224 Ra) in surface waters east of the Kerguelen Islands, south of the PF, clearly indicates that these waters have recently interacted with shallow sediments.Neither the shelves of Heard Island -located hundreds kilometers south of the study area -nor the vertical mixing of deep waters that interacted with bottom sediments can account for the short-lived radium enrichments found in surface waters.The 223 Ra and 224 Ra activities south of the PF are thus best explained by waters that interacted with the shelves of the Kerguelen Islands.This finding implies that chemical elements can be transported across or via the PF.Among the potential mechanisms allowing surface waters to be transported eastward across the PF, one can invoke either (i) the wind stress (eastward winds are especially strong in that region) or (ii) eddies that form along the PF and that may promote the transport of surface waters and associated chemical elements.The spatial variability observed in the 223 Ra and 224 Ra distribution in surface waters south of the PF suggests that the input of waters and associated chemical elements across the PF -potentially driven by wind stress or eddies -act as sporadic pulses that may highly vary in both space and time.This pathway may thus constitute a mechanism that contributes to fertilizing the phytoplankton bloom with iron and other micronutrients east of the Kerguelen Islands, south of the PF.This finding shows that the PF may not act as a strong barrier for surface waters and associated chemical elements, a finding that may also apply to other frontal systems of the world's ocean.

Figure 1 .
Figure 1.Location of stations investigated for Ra analysis.Solid circles represent surface seawater samples.Circles show the locations where vertical profiles were made.KP is the abbreviation for Kerguelen Plateau.Gray shading represents depth in meters.

Figure 2 .
Figure 2. Satellite composite images of sea surface chlorophyll a (mg m −3 ) at successive dates between the beginning of the KEOPS-2 cruise (19 October 2011) and the end of the cruise (23 November 2011).The locations of the water samples collected for radium analysis within the different time intervals are also reported (solid circles).

Figure 4 .
Figure 4. 224 Ra (a), 223 Ra (b) and 228 Ra (c) distributions in surface waters off the Kerguelen Islands.Radium activities are expressed in dpm 100 L −1 .White circles indicate samples with Ra activity below the detection limit.A schematic view of the polar front (PF) is shown.

Figure 5 .
Figure 5. Vertical profiles of 223 Ra, 224 Ra and 228 Ra activities (dpm 100 L −1 ) at the shallow stations.The main water masses are indicated: Winter Water (WW) and Upper Circumpolar Deep Water (UCDW).The bottom depth is denoted by the horizontal lines.

Figure 6 .
Figure 6. 223Ra, 224 Ra and 228 Ra activities (dpm 100 L −1 ) at the deep stations.The major water masses are indicated: Winter Water (WW), Upper Circumpolar Deep Water (UCDW), Lower Circumpolar Deep Water (LCDW) and Antarctic Bottom Water (AABW).The bottom depth is denoted by the horizontal lines.

Figure 7 .
Figure 7. Trajectories of the drifters launched to the east of the Kerguelen Islands during the KEOPS-2 project.The trajectories of the two drifters released at station A3 are reported in color.The equivalent transit time of the two drifters is reported in days along their trajectory.The other drifter trajectories are represented in light gray.

Figure 8 .
Figure 8. Apparent ages of surface waters determined using the 224 Ra / 223 Ra ratios.The offshore apparent ages were estimated using an initial 224 Ra / 223 Ra ratio that was obtained by averaging the ratios found at stations located on the northern Kerguelen Plateau (< 200 m water depth).When both 224 Ra and 223 Ra were significant, apparent ages could be determined (colored symbols).The samples displaying an apparent age between 1 and 2 months are shown in dark gray ( 224 Ra < DL but significant 223 Ra activities).Water samples displaying an apparent age > 2 months are shown in light gray ( 224 Ra < DL and 223 Ra < DL).A schematic view of the polar front (PF) is represented.

Figure 9 .
Figure 9. Lagrangian particle analysis derived from total surface currents (considering absolute geostrophic plus Ekman currents).Solid circles represent the location of the stations.The targets for Lagrangian analysis were centered on and around the station locations (to account for spatial variability).The sampling date is indicated in parentheses.Two-month backward trajectories are shown.The first month of the backward trajectory is represented by a thick line.The second month of the backward trajectory is represented by a thinner line.

Figure 10 .
Figure 10.Ages of surface waters (in days) derived from an altimetry Lagrangian-based model.In situ ages derived from radium isotopes are represented by circles.The color bar indicates the time elapsed since the water body left the 2000 m isobath.
et al.: Use of Ra isotopes to deduce rapid transfer of sediment-derived inputs which is consistent with the existence of a retentive recirculation region centered at about 73 • W, 49 • S.

Table 1 .
Dissolved 223 Ra,224Ra and 228 Ra activities determined in surface samples collected off the Kerguelen Islands.The 223 Ra and 224 Ra activities are excess radium activities (see Methods for details).Activities are expressed in disintegration per minute per 100 L (dpm 100 L −1 ).< DL denotes below the detection limit.

Table 2 .
Dissolved 223 Ra,224Ra and228Ra activities determined in seawater samples collected in the water column using Niskin bottles.The 223 Ra and 224 Ra activities are excess radium activities (see Methods for details).Activities are expressed in disintegration per minute per 100 L (dpm 100 L −1 ).The number of counts detected is also reported in the Table (cnts).< DL denotes below the detection limit.
www.biogeosciences.net/12/1415/2015/Biogeosciences, 12, 1415-1430, 2015 Moore (2000)Ra / 223 Ra) i is the initial ratio in source waters, ( 224 Ra / 223 Ra) obs is the ratio for a given water sample, and λ 224 and λ 223 are the decay constants of 224 Ra and 223 Ra, respectively.The assumptions inherent to this equation can be found inMoore (2000)and are that (1) the223Ra and 224 Ra activities are constant in the source region (i.e., a constant initial 224 Ra / 223 Ra ratio is assumed), (2) the 224 Ra / 223 Ra ratio changes are only due to radioactive decay, and (3) open ocean waters contain no excess 223 Ra and 224 Ra.