Remobilisation of uranium from contaminated freshwater sediments by bioturbation

Previous studies have demonstrated that benthic macro-invertebrate bioturbation can inﬂuence the remobilization of uranium initially associated with freshwater sediments resulting in a high release of this pollutant through the overlying water column. Giving the potential negative e ﬀ ects on aquatic biocenosis and the global ecological risk, it ap- 5 peared crucial to improve our current knowledge concerning the uranium biogeochemical behaviour in sediments. The present study aimed to assess the biogeochemical modiﬁcations induced by Tubifex tubifex (Annelida, Clitellata, Tubiﬁcidae) bioturbation within the sediment permitting to explain such a release of uranium. To reach this goal, uranium distribution between solid and solute phases of a reconstructed benthic sys- tem (i.e. in mesocosms) inhabited or not by T. tubifex worms was assessed in a 12 day laboratory experiment. Thanks notably to ﬁne resolution (mm-scale) measurements (e.g. DET gels probes for porewater, bioaccumulation in worms) of uranium and main chemical species (iron, sulfate, nitrate, nitrite), this work permitted (i) to conﬁrm that the removal of bottom sediment particles to the surface through the digestive tract of worms greatly favours the oxidative loss of uranium in the water column, and (ii) to demonstrate that both uranium contamination and bioturbation of T. tubifex substantially inﬂuence major microbial-driven biogeochemical reactions in sediments (e.g. stimulation of denitriﬁcation, sulfate-reduction and iron dissolutive reduction). This study provides the ﬁrst demonstration of biogeochemical modiﬁcations induced by bioturbation in freshwater uranium-contaminated sediments.

the total concentration in uranium, the presence of organic and mineral ligands (Ragnarsdottir and Charlet, 2000;Davis et al., 2002Davis et al., , 2004Davis et al., , 2006Denison, 2004;Curtis et al., 2006) and microbial activity (Renshaw et al., 2007). In most surface freshwaters (i.e. under oxic conditions, pH 5-9), uranium occurs under a free and soluble form, the uranyl ion UO 2+ 2 , which is at the (+VI) oxidation state. Depending on the pH and the 20 ionic composition of the water, the uranyl ion can form complexes with other ions, principally hydroxyles or carbonates, phosphates, fluorides, chlorides, but also with some organic compounds such as humic acids (Ragnarsdottir and Charlet, 2000;Marang, 2007). Adsorption on mineral (e.g. oxy/hydroxydes of iron or manganese and clays) or organic particular phases also plays an important role by reducing the mobility of 25 uranyl ions in water . Thereby, uranium coming in contact with the sediment is either on a soluble form, free or complexed, and will diffuse towards the porewater, or sorbed to suspended matter and will be incorporated by sedimentation. In anoxic sediments, uranium is reduced to U(+IV) which is an insoluble form and tends Introduction to be immobilised and thus to accumulate in the deeper sediment layers by formation of insoluble nanoparticulates or large aggregated oxides like uraninite or schoepite (Liger et al., 1999;Phrommavanh, 2008). Additionally, the reduction of uranium can also occur biotically through metal-reducing bacteria metabolism in sediment (Lovley et al., 1991). This process has notably been used in recent bioremediation programs of con-5 taminated sites where immobilisation of uranium in sediments was favoured by organic amendment at the sediment surface (Renshaw et al., 2007;Wall and Krumholz, 2006;Wilkins et al., 2006;Barlett et al., 2012). Inversely, mechanical disturbances of top sediment can hamper the incorporation of uranium in bottom anoxic layers and its immobilization. Among them, the modifications of sediment properties induced by benthic macro-invertebrate activities (i.e. bioturbation) are likely to be of first importance. As demonstrated for diverse trace metals, bioturbation can increase their remobilization from the sediment through the overlying water by favouring the oxidative loss of previously accumulated solid phases (Motelica-Heino et al., 2003;Naylor et al., 2004Naylor et al., , 2006Naylor et al., , 2012. Three major processes 15 are involved: (i) oxygen penetration in bottom sediment due to active water pumping in burrows (i.e. bioirrigation), (ii) removal of particles from the bottom sediment to its surface, (iii) and indirect effects due to induced heterogeneity (e.g. redox conditions, organic matter availability, fluxes of solutes) and stimulation/inhibition of microbial communities. However, little work has been undertaken so far to evaluate these processes 20 in sediments accumulating uranium in freshwater systems (Komlos et al., 2008;Phrommavanh, 2008) and the few previous studies on this topic concerned marine ecosystems (Zheng et al., 2002;Morford et al., 2009). In freshwaters, some organisms able to survive in contaminated environments, such as tubificid worms (Annelida, Clitellata), are known to induce a strong sediment reworking that could impact the remobilisation 25 of metals (Zheng et al., 2002;Alfaro-De-la-Torre and Tessier, 2002;Ciutat and Boudou, 2003;Ciutat et al., 2007;De Haas et al., 2005;Petersen et al., 1995;Soster et al., 1992;Zoumis et al., 2001). Tubificid worms represent a dominant group of freshwater benthic macro-invertebrates. Their populations can reach very high densities notably Introduction in organic-rich sediments. Despite no active bioirrigation of their gallery network, the effects induced by tubificid worms on the sediment matrix are crucial for the global biogeochemical functioning and the metal distribution at the benthic interface. Due to their particular mode of feeding which consist on ingestion of sediment particles in bottom layers and rejection at the sediment-water interface (i.e. upward-bioconveying, 5 sensu Gérino et al., 2003), they create a removal of associated compounds including metals initially immobilized under reduced forms. This phenomenon has been recently observed in uranium-contaminated sediment inhabited by Tubifex tubifex worms (Lagauzère et al., 2009a-c). In these mesocosm experiments, it was demonstrated that despite important ecotoxic effects of uranium on the worms (e.g. malformations, autotomy, mortality), their bioturbation activity remained sufficiently important to stimulate diagenetic processes (e.g. increase of oxygen uptake) and to induce a 2-to 10-fold higher release of uranium through the overlying water. However, underlying biogeochemical processes occurring under these conditions still need to be assessed to explain the remobilization of uranium. 15 The main goal of this study was thus to assess the influence of the bioturbation of T. tubifex in a benthic ecosystem for which sediment was initially contaminated with uranium. We conducted a laboratory experiment using mesocosms with natural sediment artificially contaminated in uranium and inhabited or not by T. tubifex. The distribution of uranium between solid and solute phases, including bioaccumulation in T. tubifex 20 worms, was estimated to calculate fluxes between the sediment and the water column, and a mass budget. In parallel, high-resolution profiles of dissolved uranium, iron, manganese, sulfates, nitrates and nitrites were measured to give an overview of the main biogeochemical reactions occurring in sediment and to propose hypothesis of their interactions with uranium in presence or absence of T. tubifex worms. Introduction

Preparation of the aquaria
Surface sediment and water were collected in a dead arm of the Esparron Lake, a reservoir-lake upstream from a man-made dam on the river Verdon (Alpes de Haute-Provence, Southern France). This site was chosen for the pristine quality of its surface 5 water and the fine texture of the sediment (median grain size, D 50 = 33.8 µm). Sediments were treated to eliminate the grosser particulate phases (vegetal fragments, stones, wastes) and the maximum of organisms by sieving at 2 mm and freezing at −20 • C for 48 h. The sediments were then homogenised mechanically and stored at 4 • C until the preparation of the aquaria. The water was filtered at 20 µm to eliminate 10 macro-and meiofauna and also stored at 4 • C before use. The principal physical and chemical parameters of the sediment and overlying water were reported in a previous work (Lagauzère et al., 2009c). Briefly, analyses revealed a highly calcareous medium (total calcite in sediments: 70 %, water hardness: 152 Eq mg CaCO 3 L −1 ), with a pH of 8.2-8.6 and a low organic matter content (2.4 %) for a lake sediment. 15 The sediment was artificially contaminated with a solution of uranyl nitrate UO 2 (NO 3 ) 2 · 6H 2 O in a large HDPE container (height = 65 cm, ∅ = 41 cm, 68 L drum from CurTec ® , the Netherlands) in order to obtain a nominal uranium concentration of 600 µg U g −1 (dry weight). The tank was then daily shaken during 2 weeks before the beginning of the experiment to allow adsorption of uranium on the sediment par-20 ticles and a homogenous contamination. The effective measured concentration after 2 weeks was 539 µg U g −1 . This concentration level was chosen as a function of results obtained in previous works with similar experimental conditions. It corresponds to a bioturbation activity though diminished that generates an important remobilisation of uranium from the sediment to the water column (Lagauzère et al., 2009a, c (12 cm in diameter, 20 cm in height). The aquaria were filled with 10 cm of sediment and an overlying 10 cm water column. They were randomly placed in a large tank with controlled temperature (21 • C) and photoperiod (16 h light/8 h dark). Each aquarium received a constant air bubbling in the water column. This setting has been stabulating for 4 weeks before the introduction of the T. tubifex worms. Slight additions of distilled 5 water were done on a daily basis to compensate the water losses due to evaporation. Volumes extracted for sampling were replaced by initial lake water.

Origin, acclimatization and introduction of benthic organisms
The T. tubifex worms were purchased from a commercial breeding (Grebyl and Fils, Arry, France). They were acclimated to the experimental conditions during several weeks and were fed twice a week with Tetramin ® flakes (Tetra Werke, Melle, Germany) put into suspension (3 mg ind −1 from a 10 g L −1 suspension). Before the beginning of the experiment, the worms were starved in artificial sand for 48 h. In each aquarium devoted for worm addition, 28 g of T. tubifex was added at the sediment surface. This mass corresponds to ca. 60 000 individuals per square meter, which is representative 15 of an average natural density for freshwater ecosystems (Budd, 2005). The air bubbling was stopped for three hours to allow the worms to settle to the sediment-water interface.

Experimental procedure
On the 9 aquaria prepared for each condition (contaminated/control), 3 were retrieved 20 on the first day of the experiment (day 0) i.e. after the 4 weeks of stabilization, 3 received T. tubifex worms on day 0 and were retrieved after twelve days (day 12), and the remaining 3 did not receive any organisms and were also retrieved on day 12. The coding of experimental treatments was the following: C-for control, U-for contaminated sediment, T-for presence of T. tubifex, followed by −0 or −12 for the time of sampling Introduction (for instance, UT-12 corresponds to "contaminated sediment/presence of worms" after 12 days).

Measurements in the water column: temporal monitoring
Temperature, pH and dissolved oxygen concentration of the water column were mea-5 sured one day before the introduction of the T. tubifex worms and then every second day until the end of the experiment (day 12). The concentration in total uranium in the water column was monitored from non-filtered and acidified water samples (2 % HNO 3 ) analysed by ICP-AES (Optima 4300 DV, Perkin Elmer, USA). The apparent uranium exchange flux between the sediment and the water column was calculated from the 10 uranium concentrations at day 0 and day 12 (based on the variation of uranium in the water column divided by the area of the water-sediment interface). The concentrations of dissolved chemical species, including uranium, were estimated from the average values measured from the water-exposed part of the DET probe (see below). The differences in concentrations between day 12 and day 0 were also expressed in terms of 15 mean fluxes.

Pore-water concentration profiles of the dissolved chemical species
In each aquarium, the concentration profiles of the dissolved elements were determined by using two constrained DET probes (Diffusive Equilibrium in Thin-films) purchased from DGT Research Ltd (Lancaster, UK). One of the probes was devoted to 20 the analysis of major cations and uranium by ICP-AES (Optima 4300 DV, Perkin Elmer, USA) whereas the second one was used for the analysis of major anions by ionic liquid chromatography (DX120, column AS11HC 4 mm, eluant KOH, Dionex, Sunnyvale, USA). These peeper-gel probes consisted of plastic holders ( agarose-gel strips (1 mm × 1 mm × 18 mm) were exposed through a 0.2 µm Nylon membrane to the ambient medium for equilibration. Before use, the probes were deoxygenised in a 0.01 M NaCl solution with nitrogen for 48 h. They were afterwards deployed in the sediment of each aquarium 48 h before the desired analysis time (i.e. 2 days before the introduction of the organisms for treatments C-0 and U-0 and at 5 day-10 for treatments C-12, CT-12, U-12 and UT-12). After sampling, the probes were placed in a glove box under nitrogen atmosphere. The gels strips were gently retrieved and directly eluted in 1 mL HNO 3 2 % for samples analysed by ICP-AES (major cations and uranium) and in 1 mL Milli-Q water for samples analysed by ionic chromatography (major anions). All measurements were performed in the 6 days following the sampling. 10 During their storage, the samples were kept at 4 • C and daily hand-shaked. Calculations were made based on the assumption of gel strips dimension homogeneity.

Determination of net accumulation rates in overlying waters
Accumulation rates were calculated from differences between start-end concentrations of dissolved species divided by the duration of the experiment. Positive values indicate a net input to the overlying water. In contrast, negative values suggest a net elimination/output of dissolved species.

Determination of diffusive instantaneous fluxes at the sediment-water interface and sediment uranium consumption rates
Each vertical profile of dissolved chemical species was simulated using the PROFILE 20 software (Berg et al., 1998) in order to estimate uranium consumption rates below the sediment-water interface and instantaneous diffusive fluxes of other solutes. Briefly, the diffusive flux J at the sediment-water interface was estimated from the sediment porewater concentration profile obtained with DET probe using the Fick's first  (Li and Gregory, 1974;Berner, 1980): where φ is the porosity, D s is the diffusion coefficient in sediments (cm 2 s −1 ), C is the concentration (mmol cm −3 ), z is the depth (cm) and ∂C(z)/∂z is the concentration gradient across the sediment-water interface. 5 D s was estimated from values of the diffusion coefficient in water D reported in the literature corrected for temperature and salinity (Li and Gregory, 1974;Boudreau, 1997;Schultz and Zabel, 2000) and the tortuosity θ of the sediment using the following equation: where θ was calculated from the porosity using the empirical relation given by Boudreau (1997): In our experiments the subsurface porosity was estimated at 0.75 in absence of worms and 0.83 in presence of worms, independently of the uranium contamination. 15 Thus the tortuosity values were θ 2 = 1.57 without worms (treatments C-0, C-12, U-0, U-12) and θ 2 = 1.37 with worms (treatments CT-0, CT-12, UT-0, UT-12).
Differential equations were then solved numerically at a steady state to reproduce C(z) and provide the best "reasonable" estimation of R(z) (Berg et al., 1998). The boundary conditions were the concentration in the overlying water at the top and an 20 absence of flux in deeper sediment. All fluxes were integrated over overlying water height in order to be compared to net accumulation rates in overlying waters. 10,2013 Remobilisation of uranium from contaminated freshwater sediments S. Lagauzère et al.

Determination of the uranium concentration in the solid phase of the sediment
After aforementioned analyses, at day 0 for U-0 and at day 12 for U-12 and UT-12, the water was retrieved with a syringe and the sediment column was gently sliced at a resolution of 1 cm. Each slice was dried at 60 • C for 72 h and then homogenised 5 by manual grinding with a mortar. Three sub-samples of 1 g of dry sediment were then mineralised by successive addition of HNO 3 , HCl and H 2 O 2 . After two cycles of mineralization/evaporation (105 • C, 90 min) the solutions were filtered at 0.45 µm (Minisart acetate cellulose filters) and subsequently analysed by ICP-AES to determine the concentration of total uranium in the sediment. These data were used to calculate 10 the global mass budget of uranium over the duration of the experiment.

Determination of the uranium bioaccumulation in T. tubifex worms
Four hours before dismantling the contaminated aquaria containing T. tubifex worms (UT-12) the air bubbling was stopped to allow the worms to come out of the sediment. A sample of worms was collected with a pipette to determine the bioaccumulation of 15 uranium in them. After two hours of integument depuration in non-contaminated water at ambient temperature (22-25 • C), samples of worms were dried at 60 • C for 48 h and then mineralised with the following procedure. Each sample (ca. 630 mg) was mineralized by addition of 5 mL of HNO 3 at 65 % and 5 mL H 2 O 2 at 30 % followed by two cycles of heating at 95 • C for 90 min. After complete evaporation, the solid residues 20 were put in suspension in 10 mL HNO 3 at 2 % at ambient temperature for 24 h. Samples were then filtered at 0.45 µm (Minisart acetate cellulose filters) and analysed by ICP-AES. The quality control sample was prepared by adding a known concentration of uranium to mineralized T. tubifex worms; this preparation was necessary because no certified biological material was available for uranium measurement by ICP-AES in 25 aquatic organisms. To estimate the amount of uranium in the total biomass the measured concentration was related to the mass of worms initially introduced in the aquaria BGD 10,2013 Remobilisation of uranium from contaminated freshwater sediments S. Lagauzère et al. with a minoration of 10 % to take into account the mortality of the species at this concentration of uranium (Lagauzère et al., 2009c).

Statistical analysis
All statistical analyses were performed with the Statistica software (StatSoft, Inc., OK, USA). Before each statistical analysis, the normality of the data was tested with 5 a Shapiro-Wilk's test and the homogeneity of the variances by a Levene's test. These tests were repeated after transformation of the data when these hypotheses were not fulfilled. A threshold of significance of 5 % was applied for all statistical analysis.
i. The temporal variation of physico-chemical parameters of the water column were analysed by repeated-measures ANOVAs (RM-ANOVAs) for each treatment. 10 ii. The variation of total uranium concentration in the water was analysed by a Student's t test. To separate the effects of bioturbation and the effects of uranium contamination, the variations of other parameters (iron, sulfates, nitrates and nitrites) were analysed by two-ways ANOVAs ("Tubifex"/"uranium") completed by post-hoc Fisher's LSD tests. 15 iii. The diffusive fluxes at the sediment-water interface in the different treatments were analysed by one-way ANOVAs and compared by post-hoc Fisher's LSD tests. 20 For all the experimental treatments, the water temperature was maintained at 21.2 (±0.1) • C, the dissolved oxygen concentration at 8.1 (±0.3) mg L −1 and the pH at 8.4 BGD 10,2013 Remobilisation of uranium from contaminated freshwater sediments S. Lagauzère et al. (±0.2) with no significant difference between the treatments and times of sampling (RM-ANOVAs: P > 0.05; two-ways ANOVAs: P > 0.05). The concentration of total uranium in the water column of contaminated aquaria, with and without T. tubifex worms, according to time is presented in Fig. 1, with the apparent net accumulation rate of uranium in the water column as an inset. In both 5 cases, the concentrations of uranium increased during the experiment but they reached clearly higher values in presence of worms in the sediment (117 ± 9.6 nmol U cm −3 , i.e. > 3 times the concentration at day 0). It resulted in a net accumulation rate of total uranium significantly higher (i.e. 5 times) in presence of worms (Student's test: t = −14.7, P = 0.000). The concentration of dissolved uranium measured with DET 10 probes (see below) showed the same results ( Fig. 2a-c).

Chemistry of the water column
Although several other compounds were effectively measured during the analyses, only total dissolved concentrations of iron, sulfate, nitrate and nitrite were retained here as certain compounds (e.g. manganese, potassium, phosphates) had concentrations below the detection limits of the measurement devices (ICP-AES: 1 µg L −1 , ionic chro-15 matography: 10 µL m −1 ), or simply provided no relevant information (e.g. chlorides, calcium).

Uranium
The water column and sediment porewater concentrations profiles of uranium for the three contaminated treatments (U-0, U-12 and UT-12) are presented in Fig. 2a-c. In 15 all cases, the average concentrations of uranium determined in the water column by the DET probes was similar to those obtained from non-filtered water samples (abovementioned results), indicating that almost all of uranium in the water column was under a dissolved form. These results confirm an increase of uranium concentrations in the water during the experiment with a much more pronounced effect in presence of 20 worms. Under the sediment-water interface, the shape of uranium profiles is similar in all treatments and shows decreasing concentrations with depth ( Fig. 2a-c), probably due to its diffusion from the overlying water and its reduction within the anoxic sediment. Modelling estimations from PROFILE reveal significant dif- flux more than doubled with time, from −0.09 ± 0.00 × 10 −3 nmol U cm −3 h −1 (U-0) to −2.21 ± 0.01 × 10 −3 nmol U cm −3 h −1 (U-12). In response to increasing concentrations in overlying water and subsequently increasing chemical gradient, uranium consumption rate increased, indicating a strong kinetic dependency to substrate (i.e. uranyl) concentration. On the other hand, the diffusive flux was lower with bioturbation (−0.07 ± 0.01 × 10 −3 nmol U cm −3 h −1 ). Indeed, uranyl production occurred in a fine layer directly under the interface and the reduction started 1 cm deeper than in undisturbed sediment. This production, accounting for 5.75 × 10 −5 nmol U cm −2 s −1 , may be relative to oxidation of upward-diffusing reduced uranium and then limited the total inward flux.

10
Order of magnitudes of instantaneous fluxes (Fig. 2d) and time integrated fluxes (inset graph of Fig. 1) can be compared: they both indicate that the presence of worms clearly increased the uptake flux of uranium as a response to increased transfer to overlying water from dissolution/release of particle-bounded uranium. Biogeochemical pathways responsible for uranium consumption at depth did not seem to be quantita- 15 tively modified by worm activities, at least at the experiment time scale.

Other dissolved species
Here again, for clarity and interest, only dissolved concentration profiles of total iron, sulfate, nitrate and nitrite are presented (Figs. 3-6).
Total iron -without bioturbation (C-0, C-12), the iron profiles show increasing con-20 centrations from the sediment-water interface to a certain depth and then a decrease (Fig. 3a). They are characteristics of an upward diffusion of Fe 2+ in top sediment, the remobilisation of iron through dissolutive reduction of iron oxi/hydroxides in deeper sediment, and a removal by precipitation onto mineral phases within the bottom sediment.
With bioturbation (CT-12), the concentrations of iron in the sediment were globally lower 25 and decreased gradually with depth but no peak appeared on the profiles (Fig. 3b) same trends were observed in contaminated aquaria (U-0, U-12, UT-12) with an additional increase of concentrations in the deepest part of the profiles (Fig. 3b). All profiles indicated an upward diffusive flux at the top of the sediment (Fig. 3c) though no dissolved iron was detected in the water column. This could be explained by a direct precipitation of released iron since the water was well oxygenated. 5 Sulfate -in the uncontaminated aquaria (C-0, C-12), the sulfate concentration profiles showed a net production under the sediment-water interface and consumption in depth (Fig. 4a), resulting in a net outward flux towards the water (Fig. 4c). With bioturbation (CT-12), the highest concentrations were detected in the water column and sulfates diffuse towards the sediment where they were directly consumed under the interface 10 ( Fig. 4d). The presence of uranium also modified these profiles (Fig. 4b): the production of sulfate under the sediment-water interface was not yet observed in undisturbed aquaria (U-0, U-12) but occurred in a fine layer in bioturbated ones (UT-12). Finally, the consumption of sulfate in bottom sediment probably due to bacterial sulfate-reduction resulted in an inward diffusive flux in all contaminated aquaria (Fig. 4c). 15 Nitrate -in control aquaria without worms (C-0, C-12), the concentrations of nitrate were low and the profiles were poorly marked (i.e. no significant flux at the sedimentwater interface), only indicating a slight consumption within sediment ( Fig. 5a and d).
On the other hand, the concentrations were higher in presence of worms (CT-12) and the profile showed a slight outward diffusive flux ( Fig. 5a and d). In contaminated 20 aquaria, the profiles appeared clearly different. Without bioturbation (U-0, U-12), despite some irregularities in the shape of the profiles, the concentrations of nitrate gradually increased in the sediment, indicating a clear production of this compound (Fig. 5b). The resulting diffusive fluxes were directed towards the sediment (Fig. 5c and d). Here again, in presence of worms (UT-12), the concentrations were higher and the profile 25 showed a negative peak across the sediment-water interface reflecting consumption of nitrate (Fig. 5b). However, production also occurred in this case but deeper in the sediment. The net diffusive flux was directed towards the overlying water (Fig. 5d) Nitrite -compared to nitrate, the concentrations of nitrite were very lower. Except for the U-0 condition for which PROFILE modelling estimated a clear production of nitrite within sediment, the other profiles indicated very slight diffusive fluxes related rather to consumption in undisturbed sediment (C-0, C-12, U-12) and to production in bioturbated sediment (CT-12, UT-12) (Fig. 6a-d).

Bioaccumulation in T. tubifex worms
The average concentration in uranium determined in the tissues of the T. tubifex worms after 12 days of exposure was 37.4 (±8.3) µg U g −1 dry weight. As the concentration of uranium in the surrounding sediment was around ten times higher, it can be concluded that these organisms did not bioconcentrate uranium (Bioconcentration factor 10 BCF ≪ 1). However, it can be noted a low bioconcentration by comparing the value of bioaccumulation to the concentration in the water closed to the sediment-water interface (BCF = 1.3 ± 0.1) or to the porewater concentration in the sediment surrounding worms (BCF = 2.3 ± 0.1). Table 1 presents the mass balance of uranium between the different compartments (water column, sediment, pore water, organisms) for the initial conditions (day 0) and after 12 days with or without bioturbation. Although equilibrium was not reached after 12 days of experiment, since between treatments U-0 and U-12 the concentration in water continued to rise (∼ 1 %), the results confirmed that uranium did not remain within 20 the sediment and was released through the water column. Bioturbation has a stronger effect on this remobilisation with more than 5 % of uranium being removed from the sediment towards the water phases at the end of the experiment.

Bioturbation effects in uncontaminated sediment
Although the main goal of this study was to investigate the distribution and transfers of uranium in bioturbated sediment, the comparison of parameters in uncontaminated aquaria (C-0, C-12, CT-12) have permitted to improve substantially our global knowl-5 edge of biogeochemical processes influenced by the bioturbation of T. tubifex worms. It is thus important to rapidly discuss these results. The most visible consequence of bioturbation was the changes of the water quality with increasing concentrations of sulfate and nitrate. Since there was not influx of water during the experiment (excepted additions made to compensate evaporation and sampling), these results can be only 10 related to changes occurring within the sediment.
In the absence of uranium contamination, the worms reached the bottom of the aquaria (10 cm) and induced a high upward advection of sediment particles towards the water column (Lagauzère et al., 2009a, c). A fine layer of mucus-bounded faecal pellets appeared at the top of the sediment. The sediment-water interface area mod-15 erately increased, the oxygen penetration depth was reduced from 3 to 2 mm, and the diffusive oxygen uptake (DOU) of sediment was enhanced by 14 % (Lagauzère et al., 2009b). In accordance with previous works reported in the literature, it has been suggested that these observations were relative to a global stimulation of microbial activity through notably a supply of labile organic matter, the re-fractioning of sediment parti-20 cles, and the providing of new microniches for micro-organisms involved in diagenetic processes (Mermillod-Blondin et al., 2013). The upward-bioconveying of sediment particles also induces oxidation of reduced compounds removed from the bottom sediment (i.e. iron sulfide Fe-S), and enhances fluxes of nutrients due to higher diffusion and advection processes (Matisoff, 1995;Mc Call and Fisher, 1980;Mermillod-Blondin et al., 25 2005;Nogaro et al., 2007;Svensson et al., 2001). In the case of high densities of benthic worms (20 000-70 000 ind m −2 ), as used in our study, high removal of reduced compounds coupled with intensive aerobic microbial activity in the faecal pellet layer 17018 Introduction hampers the oxygen penetration in surface sediments (Pelegri and Blackburn, 1995). These authors suggested that denitrification and sulfate-reduction would be favoured by these suboxic/anoxic conditions whereas nitrification would be limited to a very fine layer under the sediment-water interface. In the present work, the profiles obtained from DET-gel probes partially confirm these assumptions. For instance, the higher sul-5 fate concentrations measured in bioturbated aquaria (CT-12) can be explained by removal of reduced sulfur and its subsequent oxidation in the water. Sulfate ions diffuse therefore towards the sediment where they are rapidly consumed by sulfate-reducers. Inversely, in undisturbed sediment (C-0 and C-12), the concentration profiles show a peak under the sediment-water interface that reflects production of sulfate through 10 iron sulphide oxidation since top sediment was more oxygenated than in bioturbated sediment. This apparent production in sediment can also be due to gypsum dissolution as gypsum particles are commonly present in the Verdon River, the main tributary of the Esparron Lake (E. Viollier, personal communication, 2012). Without upward advection of sediment particles, as induced by worms (see below), the concentrations of 15 sulfate remained low in the water column.
On the other hand, nitrate and nitrite concentration profiles do not directly fit the assumption of an enhanced denitrification and a limited nitrification. In all cases, the concentrations were rather homogeneous with sediment depth that indicates a lower nitrogen turn-over. Nevertheless, the only remarkable profiles correspond to aquaria 20 with bioturbation (CT-12). In this case, concentrations were also more or less constant on the entire profile but much higher than in undisturbed aquaria. As suggested by Stief and De Beer (2002), this result can be explained by a high coupling of denitrification and nitrification processes resulting in a fast turn-over of nitrogen in sediment. Through enhanced organic matter mineralization and their own excretion, the worms lead to 25 the production of ammonium which is in return re-oxidized into nitrite and nitrate. The nitrate concentration profile shows however a low diffusion towards the sediment which can be related to consumption by denitrification. Finally, it is important to note the downward shift of iron profiles with bioturbation, which indicated a higher dissolutive reduction of Fe(III)-oxi/hydroxides compared to the undisturbed sediment. Here again, the particle reworking of the sediment column was likely to enhance the removal of iron from bottom sediment towards a more reactive zone and then to increase its global turn-over.

Bioturbation effects in uranium-spiked sediment
As already mentioned, previous experiments conducted in strictly same conditions have permitted to obtain a solid basis to understand the influence of T. tubifex bioturbation in uranium-spiked sediments and to choose the most suitable experimental design for the present study (Lagauzère et al., 2009a-c). At the tested concentration of 10 ca. 600 µg U g −1 dry sediment (i.e. > 100 times the background level for freshwater sediments), the population of worms was reduced by 10-20 % after 12 days of exposure. Uranium induced several negative effects on T. tubifex (e.g. malformations, autotomy, loss of biomass) that became significant only for these higher concentrations. Nevertheless, their bioturbation activity was significantly affected from this level of contamina-15 tion. The most easily observable effects were a 2-fold reduction of the total length of the gallery network, a net concentration in upper layers of sediments (< 2 cm), a lower maximal penetration depth (6 cm) and asynchronous, disordered and slower movements of sampled individuals. By simulating vertical profiles of particle tracers (microspheres and luminophores) with diagenetic models, it was possible to estimate a decrease of 20 advective (60-70 %) and diffusive transports (25-30 %) induced by bioturbation in contaminated sediments (Lagauzère et al., 2009a). Likewise, in presence of worms the porosity was only enhanced of 10 % in the first cm of uranium-spiked sediments while it reached more than 20 % within 2 cm in uncontaminated sediments. Finally, it is important to note that the lower penetration and dispersal of worms within the sediment 25 column led to a 2-fold higher ingestion rate compared to control sediment, with a peak in the nearest cm under the sediment surface. As a main hypothesis, it was suggested that the behavior of worms resulted in a trade-off between avoiding contaminated sed-17020 Introduction iment and yet remaining within it to meet their physiological needs and to be protected from predators. Despite the negative effects of uranium on the activity of T. tubifex worms, the concentration of uranium in the water column was multiplied by 5 in the course of the three aforementioned studies, a result again confirmed in the present case. As demonstrated 5 for other metals, such a release of uranium can be explained by the removal of particles from anoxic layers to the surface of sediment through the digestive tract of worms. Nevertheless, underlying biogeochemical processes needed to be investigated furthermore to support this hypothesis and to provide additional information on the influence of tubificid worms in contaminated sediments. Previous results relative to DOU (Diffusive oxygen uptake) of sediments, which is a representative parameter of the global functioning of sediment, have shown a net increase due to both uranium contamination (+24 %), T. tubifex bioturbation (+14 %), and the combination of these two factors (+53 %). The present study has permitted to confirm and quantify the direct and indirect oxidative loss of uranium initially associated with sediment and to demonstrate 15 a clear impact of uranium and/or bioturbation on microbial-driven diagenetic reactions.
First, it is important to consider the information provided from experimental units without T. tubifex worms. As confirmed by DET profiles, uranium released in the water was actually produced in the oxic sediment (< 3 mm). Initially, uranium was added to the sediment in close beakers kept without oxygenation for 4 weeks. In this state, most 20 of uranium would be under insoluble and reduced form U(+IV). After assembling the aquaria, the sediment became in contact with aerated water and uranium could gradually desorbed and be oxidized into its soluble form U(+VI). This result is supported by previous experimental work demonstrating that the oxidation of reduced uranium solid phases is a fast process (Anderson et al., 1989;Cochran et al., 1986) Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | tion in the overlying water. Although the direct oxidative loss of uranium could explain the enhanced DOU of sediment (+24 %), it was assumed that uranium could influence several microbial-driven diagenetic processes. First of all, sulfate concentration profiles show the disappearance of the sulfur oxidation layer in top sediment and only reflect a low diffusion of sulfate from the water to 5 the sediment and then a lower sulfate-reduction rate. In the present case, the microorganisms involved in the sulfur cycle were not stimulated and even rather inhibited by uranium contamination.
In the same manner, profiles of nitrate/nitrite concentrations in the contaminated aquaria show clear differences with the control aquaria. The concentrations in the sed-10 iment were higher that could reflect intensive nitrification process. However, these concentrations might be more probably explained by the way of sediment contamination itself since uranyl nitrate was used. Nevertheless, this nutrient supply seemed to not be consumed by denitrification. Here again, the uranium contamination did probably not influence micro-organisms involved in the nitrogen cycle, all the more than they were 15 not already very active in the uncontaminated sediment.
In contrast, uranium has induced slight modifications of the iron cycle as indicated by higher concentrations of dissolved iron in bottom sediment due to potentially enhanced dissolutive reduction of Fe(III) oxi/hydroxides.
Although no strong stimulation of microbial activity linked to sulfur, nitrogen or iron cy-20 cles by uranium was put in evidence in this work, some examples have been reported in the literature. Most of these studies deal with the immobilization of uranium in sediment by favoring its reduction in the context of bioremediation of contaminated water. Indeed, uranium can be a potential substrate for anaerobic respiration (Lovley et al., 1991). Most of dissimilatory metal-reducing bacteria capable of conserve energy by coupling 25 organic matter and H 2 oxidation with reduction of metallic ions can also reduce uranyl. As well, some sulfate-reducing bacteria can catalyze the reduction of ferrous ions or uranyl ions without keeping energy or grow up with these ions as sole electron acceptors . In contaminated sites, it was demonstrated that anaerobic prokaryotes were easily cultivable on nuclear wastes, with a clear dominance of nitratereducing organisms (Akob et al., 2007). Besides these reduction processes, oxidation of uranium (uraninite) can also be biotically favored in presence of nitrate or Fe(III) oxides but it remains currently unknown if bacteria turn a gain of energy from this reaction (Borch et al., 2010). Comparatively to these studies, the present experiment represents the first assessment of uranium contamination on microbial communities not previously exposed to pollution. More generally, toxicity of uranium to micro-organisms has been so far poorly investigated and appears to be lower than for other heavy metals (Nies, 1999). A case of resistance has been also reported in an aerobic bacterium capable for detoxification through formation and rejection of intra-cytoplasmic granules (Suzuki and Banfield, 2004). Here, we indirectly demonstrated an inhibition of the activity of micro-organisms involved in sulfur and nitrogen cycles after some weeks of exposure to contaminated sediment. However, from these results, it is not possible to conclude to long-term effects of such a contamination and potential adaptations processes of micro-organisms could not be excluded (Hoostal et al., 2008). Finally, in absence of 15 available estimations of aerobic microbial activity alone, the higher oxygen consumption of sediment in presence of uranium should be mainly attributed to its oxidation that leads to increase its concentration in the water by solubilisation. Without more evident stimulation of metal reducers, the oxidation of uranium may be primarily abiotic. The presence of T. tubifex worms introduced another transfer pathway of uranium 20 previously associated with sediment. By conveying particles directly from the depth of maximal ingestion rate (2 cm), which is within the reducing zone, to the sediment surface, oxidative dissolution of uranium was greatly enhanced. In addition, the contact of ingested particles with digestive tract solutions is likely to have contributed to uranium dissolution. Already observed in marine environments accumulating authigenic 25 uranium (Zheng et al., 2002), these results confirm what was previously reported for other metals in freshwater sediments (Ciutat and Boudou, 2003;Ciutat et al., 2007;Krantzberg, 1985;Matisoff, 1995). The flux modelling corresponding to UT-12 profile indicates that this process was all the more so efficient that bioturbation also limited the diffusive flux from the water to the sediment due to the presence of an oxidation layer in top sediment. As well, a low bioaccumulation of sediment-associated uranium during the digestive tract of particles into the worms has favoured its release. Only the bioconcentration factors relative to water were higher than 1, indicating that diffusion of dissolved uranium through the tegument of worms may be the dominant way 5 of exposure and accumulation. However, this effect was weak, probably because of the water hardness (152 Eq mg CaCO 3 L −1 ) and the pH (8.6) that considerably limit the bioavailability of uranium (Markich, 2002;Sheppard et al., 2006). Nevertheless, longterm effects need to be assessed furthermore to determine potential adaptation of worm populations or on the contrary increased bioaccumulation and toxic effects due 10 to increasing uranium concentration in the water. Although the upward particle advection induced by the worm feeding mode was apparently the most influent factor in the present case, the diagenetic behaviour of metallic pollutants can also be indirectly modified by bioturbation-driven changes of sediment properties (e.g. redox boundaries). In presence of T. tubifex worms, the DOU of sedi-15 ment increased of 18 % in comparison with undisturbed aquaria (U-0 and U-12) and of 53 % when sediment was initially not contaminated (C-0 and C-12) (Lagauzère et al., 2009b). From these results it was suggested that bioturbation stimulated aerobic reactions already favoured by uranium. However, the present experiment did not confirm totally this hypothesis since anaerobic reactions were mainly stimulated. Like in control 20 aquaria, sulfate and nitrate concentrations increased in the water due to modifications in the sediment. The profile of sulfate seems to result of combined effect of uranium (i.e. inhibition of micro-organisms involved in sulfur cycle) and bioturbation (i.e. upwardbioconveying of reduced sulfur from bottom sediment, subsequent oxidation in the water and diffusion towards the sediment where it was consumed by sulfate-reduction), 25 the latter apparently the dominant. For nitrate, the same trend was observed as the profiles were marked by a negative peak around the interface that can be explained by the use of the nitrate supply due to contamination by uranyl nitrate. Although nitrate were not reduced in absence of worms (U-0 and U-12), the bioturbation seemed to have stimulated denitrification in UT-12 aquaria, a reaction which was already slightly enhanced in control conditions . It is however difficult to explain nitrate concentration increase at depth since the bioturbation activity was damped from 2 cm. The occurrence of transient anaerobic nitrification based on ammonium oxidation by manganese oxides, as proposed in the case of newly deposited sediment (i.e. flood 5 sediments or turbidites) in recent studies (Anschutz et al., 2002;Chaillou et al., 2008;Clément et al., 2005) could be considered for instance. The nitrate profile was particularly complex in contaminated sediment where the interplay with denitrification seems to play an important role. In contrast, the dissolutive reduction of iron oxi/hydroxides did not seem to be modified by the presence of worms. 10 The higher oxygen consumption of sediment was therefore mainly attributed to oxidation of reduced materials removed from bottom anoxic sediment. Nevertheless, aerobic microbial respiration could have been stimulated by the higher supply of labile organic matter due to toxic effect on worms themselves (i.e. mortality of 10-20 % of the worm population, enhanced mucous production by resistant individuals and lost of the 15 posterior part of their bodies through a detoxification process).
Finally, after only 12 days of bioturbation, around 5 % of uranium initially associated with sediment was removed towards the aqueous phases. Based on estimation of diffusive fluxes in the different experimental treatments and uranium accumulation in water, a conceptual outline is proposed in Fig. 7 to visualize the transfers of ura-20 nium in presence of worms in the sediment. Considering a Predicted No-Effect Concentration (PNEC) of 23 mg U L −1 , as estimated from our experimental conditions (pH 8.6, water hardness 152 Eq mg CaCO 3 L −1 ), the concentration reached in the water (∼ 30 mg U L −1 ) due to bioturbation represents a serious threat for the aquatic biota (Sheppard et al., 2006). However, the benthic response to increasing concentration 25 in the water corresponds to enhanced downward diffusive fluxes. Until a steady state would be reached, it would account for increasing bioaccumulation of uranium and possible toxic effects slowing down bioturbation activity. In the other hand, long-term adaptations are likely to occur for a tolerant species such as T. tubifex if some populations are gradually or punctually exposed to uranium contamination in the environment.

Conclusions
This work confirms the major role of T. tubifex bioturbation in the biogeochemistry of freshwater sediments in general (e.g. stimulation of diagenetic processes, remobilisa-5 tion of reduced materials, redistribution of solutes), and in the remobilization of uranium from the sediment. Despite a lower bioturbation activity in uranium-contaminated sediment, these biogeochemical processes are maintained and some microbial communities are even stimulated (denitrifying and sulfate-reducing bacterial communities) compared to non-bioturbated sediments (Table 2). It remains unclear if these changes indirectly influence the uranium distribution but T. tubifex play a key-role in the remobilisation of this metal from the sediment to the water through upward bioconveying of sediment particles. Long-term effects still need to be assessed, microbial diversity and activity have to be investigated more precisely through molecular approach, and experiments should be extended to real contaminated sites. However this study pro- 15 vides the first demonstration of biogeochemical modifications induced by bioturbation in freshwater uranium-contaminated sediment. As regard to the high tolerance of tubificid worms to sediment contamination and the potential risk for aquatic biocenosis exposed to dissolved uranium, it appears crucial to consider bioturbation as an important factor in further studies, notably in the development of bioremediation strategies 20 where the potential key-role of bioturbation is largely overlooked. Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Hoostal, M. J., Bidart-Bouzat, M. G., and Bouzat, J. L.: Local adaptation of microbial communities to heavy metal stress in polluted sediments of Lake Erie, FEMS Microbiol. Ecol., 65, 156-168, 2008. Komlos, J., Peacock, A., Kukkadapu, R. K., and Jaffé, P. R.: Long-term dynamics of uranium reduction/reoxidation under low sulfate conditions, Geochim. Cosmochim. Ac., 72, 3603-5 3615, 2008. Krantzberg, G.: The influence of bioturbation on physical, chemical and biological parameters in aquatic environments: a review, Environ. Pollut., 39, 99-122, 1985. Kurnaz, A., Küçükömeroglu, B., Keser, R., Okumusoglu, N. T., Korkmaz, F., Karahan, G., and Cevik, U.: Determination of radioactivity levels and hazards of soil and sediment samples in BGD BGD 10,2013 Remobilisation of uranium from contaminated freshwater sediments S. Lagauzère et al. Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Naylor, C., Davison, W., Motelica-Heino, M., Heijdt, L. M., and den Berg, G. A.: Transient release of Ni, Mn and Fe from mixed metal sulphides under oxidising and reducing conditions, Environ. Earth Sci., 65, 2139-2146, 2012. Neame, P. A., Dean, J. R., and Zytaruk, B. G.: Distribution and concentrations of naturallyoccurring radionuclides in sediments in a uranium mining area of northern Saskatchewan, BGD 10,2013 Remobilisation of uranium from contaminated freshwater sediments S. Lagauzère et al.