Uranium Biogeochemistry in the Rhizosphere of a Contaminated Wetland

The objective of this study was to determine if U sediment concentrations in a U-contaminated wetland located within the Savannah River Site, South Carolina, were greater in the rhizosphere than in the nonrhizosphere. U concentrations were as much as 1100% greater in the rhizosphere than in the nonrhizosphere fractions; however and importantly, not all paired samples followed this trend. Iron (but not C, N, or S) concentrations were significantly enriched in the rhizosphere. XAS analyses showed that in both sediment fractions, U existed as UO22+ coordinated with iron(III)-oxides and organic matter. A key difference between the two sediment fractions was that a larger proportion of U was adsorbed to Fe(III)-oxides, not organic matter, in the rhizosphere, where significantly greater total Fe concentrations and greater proportions of ferrihydrite and goethite existed. Based on 16S rRNA analyses, most bacterial sequences in both paired samples were heterotrophs, and population differences were consistent with the generally more oxidizing conditions in the rhizosphere. Finally, U was very strongly bound to the whole (unfractionated) sediments, with an average desorption Kd value (Usediment/Uaqueous) of 3972 ± 1370 (mg-U/kg)/(mg-U/L). Together, these results indicate that the rhizosphere can greatly enrich U especially in wetland areas, where roots promote the formation of reactive Fe(III)-oxides.

Numerical parameters for the fit components are shown next to the scaled and vertically-offset spectra and are also listed in Table S5.

Figure S7
. Linear combination fits of the U L III -edge data from the non-rhizosphere (NR) and the rhizosphere (R) samples with the 3 components discussed in the text: U-carb=U(VI) adsorbed to carboxyl-functionalized beads, U-DFOB=U(VI) associated with the siderophore DFOB, U-Gt=U(VI) adsorbed to goethite.Numerical parameters for the fit components are shown next to the scaled and vertically-offset spectra and are also listed in Table S5.

Figure S8
. Uranium fractionation (%-wt) of contaminated sediments based on sequential extractions operationally defined in Table S6.

List of Tables in Supporting Information
Table S1.Rhizosphere vs. non-rhizosphere chemical properties from sampled collected in the contaminated and uncontaminated portion of Tims Branch (mean ± standard deviation).Table S2.Pearson correlation coefficients for sediment properties in only rhizosphere sediments.Table S3.Pearson correlation coefficients for sediment properties in only non-rhizosphere sediments.
Table S5.Uranyl speciation in the non-rhizosphere (NR) and the rhizosphere (R) samples based on linear combination modeling of U L III -edge EXAFS spectra.Models were either conducted using two components, representing OM-bound U(VI) (a combination of siderophore-bound U(VI) and carboxyl-adsorbed U(VI)), or three components representing OM-bound U(VI) and goethite-adsorbed U(VI).Fit quality illustrated in Figures S6 and S7.Table S6.Sequential extraction method.Numerical parameters for the fit components are shown next to the scaled and vertically-offset spectra and are also listed in Table S4.

S-3
Table S4.Chemical analyses and Fe(III) linear combination fits in three paired non-rhizosphere (NR) and rhizosphere (R) sediment samples.(Uncertainty is ~10 -15% (±5 -7%)).63 ± 29 a Linear Combination Fit: Reference materials used to estimate the percentages of Fe(III) in the different sediment compartments were a SWy-2 montmorillonite standard for Fe(III) in clay; a Fe(III)citrate standard for Fe(III)-OM, and a ferrihydrite and goethite standards for Fe(III) in oxides.b R-factor is a measure of the mean square sum of the misfit at each data point.The lower the value, the better the fit.c NA = not available S-11 Numerical parameters for the fit components are shown next to the scaled and vertically-offset spectra and are also listed in Table S5.S5.

S-13
Table S5.Uranyl speciation in the non-rhizosphere (NR) and the rhizosphere (R) samples based on linear combination modeling of U L III -edge EXAFS spectra.Models were either conducted using two components, representing OM-bound U(VI) (a combination of siderophore-bound U(VI) and carboxyl-adsorbed U(VI)), or three components representing OM-bound U(VI) and goethite-adsorbed U(VI).Fit quality illustrated in Figures S6 and S7.Uncertainties reported by the LC analysis routine were in the range ± 3-7%.a R-factor is a measure of the mean square sum of the misfit at each data point.The lower the value, the better the fit.b Siderophore bound -Reference spectrum from U(VI) complexed with DFOB.This component represents U(VI) binding to a hydroxamate siderophore or carboxyl organics in the solids.c Carboxyl-adsorbed -Reference spectrum from U(VI) associated with carboxyl-functionalized beads.
This component represents U(VI) adsorbed to the carboxyl groups on bacteria or OM.d Goethite-adsorbed -Reference spectrum of U(VI) adsorbed to goethite.

Uranium Sequential Extraction Method
In an attempt to quantify the fraction of contaminant associated with the mineral matrix, some researchers have turned to selective extraction techniques. 1 Importantly, these selective or sequential extraction techniques have experimental limitations. 1,2 roblems with sequential extractions include incomplete extraction of trace elements or mineral surfaces, non-selectivity of extraction reagents for given soil phases, or re-adsorption of extracted contaminants onto other surfaces.Thus, individual sequential extraction phases may not adequately represent the discrete soil phase to which the contaminant is bound.The actual concentration extracted may only be assumed to be associated with a given operationally defined phase.The sequential extraction method used in this study was taken from Miller et al. 3 and is a modification of ASTM method D 5074-90 4 and ASTM D-3974-81 5 (Table S6).Five extractions were made from each of the five whole sediments.All tests were conducted in duplicate.These sediments were not separated into rhizosphere and non-rhizosphere because there was insufficient amount of rhizosphere sample to complete the analyses.Table S6 The first extraction step was the saturated paste.The purpose of this extract was to provide a solution that would approximate sediment pore water.Conducted in duplicate, 0.5 kg of sediment was brought to saturation with D.I. water using the method of Rhoades 6 This method calls for stirring water into the sediments over a course of two to three days, being careful not to leave standing water on the sediment.The saturated paste was allowed to equilibrate for one week.The sediment pore water was then recovered by centrifugation and filtration.
The remaining four extraction steps (Table S6) were conducted using a platform shaker and in duplicate following the procedure of Miller et al. 3 The first three treatments involved selective extractions, whereas the last treatment was total digestion of the remaining sample.The dilute S-16 acid extractant targeted recovery of the exchangeable fraction.The organically bound extractant targeted recovery of U associated with the more readily oxidizable organic matter fraction, leaving organic matter that is harder to oxidize (i.e., organic matter high in lignin content).The amorphous Fe-oxide extractant targeted recovery of U associated with the amorphous Fe-oxides by reductive dissolution of this sediment fraction and leaves intact the more crystalline fraction that requires a stronger reducing agent for dissolution.Finally, after the sediment had undergone the first four extractions, the remaining sediment was totally digested in aqua regia.The U in the aqua regia digest were presumed to be precipitated or in the mineral structure.This fraction is the least likely to contribute to the sorption/desorption process or to enter the mobile aqueous phase.It is important to note that hydrofluoric acid was not used to dissolve the remaining solids, thus the U associated with the alumino-silicates were not included in this fraction.As such, this extraction step may have slightly underestimated the actual remaining U in the residue faction.The selection of this dissolution method was necessary to address safety concerns associated with working with radioactive materials.
A 0.5 g sediment aliquot was treated sequentially with 30 ml of four extractants.Each extract was recovered by centrifugation after the appropriate contact time and then analyzed for 238 U by ICP-MS.Between each extraction, the sediment was washed three times for 2 hours with 0.01M CaCl 2 .After the final wash, the mass of entrained 0.01M CaCl 2 solution was gravimetrically determined to account for dilution in the following extraction step.
The U mass extracted in the first three extraction steps were used to calculate the desorption K d values (K d-desorb ) as follows: (1)   - =  S6 and are the concentration in the acid extractable, organic, amorphic iron oxide and saturated paste extract, respectively.A bar graph of the data is presented in Figure S8.S6.The C Sat.Paste for #4100 was 1.75 mg/L, for #4103 was 0.033 mg/L, for #4105 was 0.63 mg/L, for #4106 was 1.22 mg/L, and for #4112 was 0.53 mg/L.

Figure S6 .
Figure S6.Linear combination fits of the U L III -edge data from the non-rhizosphere (NR) and the rhizosphere (R) samples with the 2 components discussed in the text: U-carb=U(VI) adsorbed to carboxyl-functionalized beads and U-DFOB=U(VI) associated with the siderophore DFOB.

Figure S2 .Figure. S3 .
Figure S2.Mean relative abundances of bacterial and archaeal sequences at phylum level.(A) Bacterial sequences from non-rhizosphere sediment samples.(B) Bacterial sequences from rhizosphere samples.(C) Archaeal sequences from non-rhizosphere sediment samples.(D) Archaeal sequences from rhizosphere samples.

Figure S5 .Figure S6 .
Figure S5.U L III -edge EXAFS spectra obtained from the non-rhizo (NR, red) and the rhizo (R, blue) samples, going from 4100-top to 4106-bottom.(left) k 2 χ(k) data.(right) FT of the k 2 χ(k) data between k = 2.5 and 11 Å -1 in the left figure.Uncertainties reported by the LC analysis routine were in the range ± 3-7%.

Figure S7 .
Figure S7.Linear combination fits of the U L III -edge data from the non-rhizosphere (NR) and the rhizosphere (R) samples with the 3 components discussed in the text: U-carb=U(VI) adsorbed to carboxyl-functionalized beads, U-DFOB=U(VI) associated with the siderophore DFOB, U-Gt=U(VI) adsorbed to goethite.Numerical parameters for the fit components are shown next to the scaled and vertically-offset spectra and are also listed in TableS5.
+   +    .whereC solid and C aq are the solid and aqueous concentrations respectively.A acid, C Org , and C AmFeOx , and C Sat.Paste are defined in Table

Table S1 .
Rhizosphere vs. non-rhizosphere chemical properties from sampled collected in the contaminated and uncontaminated portion of Tims Branch (mean ± standard deviation).

.
Sequential extraction method.Each extract was recovered by centrifugation after the appropriate contact time and then analyzed for 238 U by ICP-MS.Between each extraction, the sediment was washed three times for 2 hours with 0.01M CaCl 2 .After the final wash, the mass of entrained 0.01M CaCl 2 solution was gravimetrically determined to account for dilution in the following extraction step.