Extreme Arsenic Bioaccumulation Factor Variability in Lake Titicaca, Bolivia

Latin America, like other areas in the world, is faced with the problem of high arsenic (As) background in surface and groundwater, with impacts on human health. We studied As biogeochemical cycling by periphyton in Lake Titicaca and the mine-impacted Lake Uru Uru. As concentration was measured in water, sediment, totora plants (Schoenoplectus californicus) and periphyton growing on stems, and As speciation was determined by X-ray absorption spectroscopy in bulk and EDTA-extracted periphyton. Dissolved arsenic was between 5.0 and 15 μg L−1 in Lake Titicaca and reached 78.5 μg L−1 in Lake Uru Uru. As accumulation in periphyton was highly variable. We report the highest As bioaccumulation factors ever measured (BAFsperiphyton up to 245,000) in one zone of Lake Titicaca, with As present as As(V) and monomethyl-As (MMA(V)). Non-accumulating periphyton found in the other sites presented BAFsperiphyton between 1281 and 11,962, with As present as As(III), As(V) and arsenosugars. DNA analysis evidenced several taxa possibly related to this phenomenon. Further screening of bacterial and algal isolates would be necessary to identify the organism(s) responsible for As hyperaccumulation. Impacts on the ecosystem and human health appear limited, but such organisms or consortia would be of great interest for the treatment of As contaminated water.


Sites description
. Dates of the five sampling campaigns and corresponding level of the Lake Tititcaca at Huatarata (source: http://www.senamhi.gob.bo/).
Concentrations of As, Mn, Fe were measured by ICP-MS (Agilent 7500 CX) at ENS Lyon for the filtered waters and for the plant and sediment digests. The solutions run on ICP-MS were prepared in HNO3 0.5 N, spiked with 2 µg L -1 indium, which is used as internal standard. Arsenic was measured on 75 As with H2 reaction mode and He collision mode, 56 Fe with He (collision mode), 55 Mn with He (collision mode) and Ar gas carrier. 75 As measurements with H2 or He in general agreed well; there is also in general a very good consistency between 55 Mn measurements made with He (collision cell) and Ar. The detection limit of As, Fe, Mn in the solutions run on the ICP-MS was ca. 0.1 µg L -1 . Filtered water samples were diluted 1:2 as well as 1:5 in HNO3 0.5 N medium; thus the final detection limit was 0.2 µg L -1 . Measurements were made in the range 1-40 µg L -1 for As, 1-300 µg L -1 for Fe and Mn. Results obtained for both dilutions (1:2 and 1:5) agree well. For plant, sediment and periphyton samples, the evaporated digest was first taken in 2 mL; an aliquot of the digest (50-500 µL depending on the digest concentration) was diluted in 10 mL HNO3 0.5 N (dilution 1:200-1-:20) so that the final concentration of the solution run on ICP-MS falls within the ranges given above (if this is not the case, further appropriate dilution was made). Repeated analysis of filtered water and solid samples within and between analytical sessions gives a precision of ± 5%.
Major elements were measured in the filtered solutions (Ca, K, P, S, Al) and in the plant and sediment digests (Fe, Mn, Ca, K, P, S, Al) by inductively coupled plasma spectrometry -atomic emission spectrometry (ICP-AES) using an Agilent 720 ES at ISTerre. Arsenic was also measured by ICP-AES when the concentration of the solutions run on ICP-AES was above detection limit of 50 µg L -1 . For the filtered solutions, there was no dilution before analysis. For the solid samples, the evaporated digest was diluted in 50 mL with 2% HNO3. Calibration was performed by dilution of standard solutions at 1000 µg L -1 . A set of water and solid samples has been measured independently by ICP-AES and ICP-MS for As, Mn, Fe and there was in general a good agreement (within a few %) between both measurement techniques. The detection limit for each element in the solutions run on the ICP-AES and the corresponding concentrations in mg kg -1 of dry solid material before digestion are given below: Limit of quantification (LQ) for the ICP-AES measurements For both analytical techniques, solid reference materials (JSd1, MESS3, BRC679) digested using the protocol described above and solution standards (Roth) were measured at each session and at different times during the session. Measured values agreed with certified values within less than 5 %. To further test the quality and consistency of the analyses, sediment and periphyton samples were digested in parallel in the two laboratories, and analyzed for As, Fe and Mn both by ICP-AES at ISTerre and by ICP-MS at ENS Lyon. Results were in very good agreement, with a few % difference.

Content in anions, dissolved organic carbon, dissolved hydrogen sulfides in water samples
Anions were measured at IGE (OSUG-Grenoble) by ionic chromatography using a 332 Metrohm apparatus. External calibration was done using monomolecular standards at 1000 mg L -1 , with various dilutions to cover the range of concentrations. The accuracy, evaluated on a multimolecular standard (Carl Roth 2668.1), was between 3 and 11%. The drift of the machine during the measurement session, corrected with the repeated analysis of a PO4 3standards, was between 3 and 13%.
Dissolved organic carbon (DOC) was measured with a TOC-VCSN analyzer from Shimadzu. The DOC is transformed into CO2, and detected by infrared. External calibration was done using certified standards at 100 mg L -1 (ChemLab). The accuracy, evaluated on a certified standard, was between 0.8 and 11%. The precision, evaluated by repeated measurements on the same sample, was < 2.5%. The drift of the machine, corrected with the repeated analysis of standards during the measurement session, was between 1 and 11%.
Samples for dissolved hydrogen sulfides (H2S) were collected directly into a degassed vacuum container, previously filled with 0.5 mL of a diamine mixture prepared as recommended (Reese et al., 2011). The method used is a modification of the previously described ones (Small and Hintelmann, 2007;Small and Hintelmann, 2014). It determines the concentration of H2S and HS -(converted into H2S by the reagents). Briefly, 20 µL of the sample with the diamine mixture was injected into an Agilent 12600 HPLC with a Poroshell 120 EC-C18 Agilent column with a mix of 20% acetonitrile, 18% methanol, 20% sodium acetate buffer (pH 5.2, 0.05 mM) at 35°C and 1.1 mL min-1. Concentrations were determined using the Radiello® calibration solution for H2S Code 171.

AVS, SEM and loss on ignition for sediment samples
Acid volatile sulfides (AVS) measurements have been performed using miniaturized and duplicate apparatus developed at ISTerre adapted from (Allen et al., 1993). To avoid sample oxidation, samples were kept in flask filled until the top and were kept under N2 atmosphere once at the laboratory. AVS extraction and quantification consisted on degasing AVS from the wet sediment by acidifying with HCl 6N and bubbling under N2 to generate H2S gaz. Aliquots of wet sediment ranged from 30 to 600 mg, in order to match the calibration range. H2S gaz was then trapped in a sodium hydroxide solution to form a stable molecule finally quantified by a spectrophotometric method with the generation of methylene blue complex referred as Cline's method (Cline, 1969). AVS results are expressed in µmol g -1 DW after freeze drying an aliquot of fresh sediment to determine its water content. Calibration was performed using a solution prepared with Na2S, 9H2O reagent and titrated by an iodometric method (Fishman and Friedman, 1989). Measurement accuracy was determined by analyzing 2 times several samples, and ranged from 8 to 16%. The simultaneously extracted metals (SEM) including Cd, Cu, Ni, Pb, Zn, As and Ag were analyzed by ICP-AES (Agilent 720 ES) at ISTerre following the protocol by Di Toro et al. (Di Toro et al., 2005). Standards were prepared with monometallic ICP standard solutions at 1000 mg L-1 diluted in HCl 6N in order to avoid any matrix effect. The machine drift was corrected based on the regular analysis of a standard during the sequence. It was always <5%. Results were corrected from the blanks measured in the same conditions as the samples. The organic content of the sediments was evaluated by the loss on ignition (LOI), which is the percentage of weight lost after 3 hours at 550°C.

XAS spectroscopy on periphyton samples
As K-edge XANES measurements on the periphyton were performed at the beamline FAME (BM30B) at the European Synchrotron Radiation Facility (ESRF) in Grenoble, France, operating in 7/8 filling mode, with a current between 160 and 200 mA. The monochromator was a Si(220) double crystal with sagittal focusing. Spectra were recorded in fluorescence mode using a 30element Canberra Ge detector, at 10°K using a He cryostat. As reference spectra included arsenopyrite FeAs III S, arsenic trisuflide (As III 2S3), As III oxide (As III 2O3), sodium arsenite (NaAs III O2), As III sorbed on ferrihydrite, sodium arsenate, As V sorbed on goethite, As V sorbed on ferrihydrite and As III -glutathione (As III -GSH) (last four provided by Raoul Marie-Couture), arsenosugars (glycerol sugars extracted from brown algae Fucus vesiculosus), dimethylarsenate (DMA(V)) and monomethylarsenate (MMA(V)) provided by Iris Koch, and As V sorbed on calcite, and mono-, di-and tetra-thioAs provided by Andreas Scheinost and Britta Planer-Friedrich.
It was not possible to ship frozen periphyton samples from Bolivia to France, so spectra were recorded on freeze-dried samples.
To ensure that this treatment did not alter As speciation, a test experiment was conducted on fresh periphyton collected in France. Periphyton samples were collected in Lake la Batie (le Versoud, France, 45°13'46.646''N 5°51'2.487'' E). They were incubated for 3 h in 500 mL bottles containing the lake water spiked with 1 mg L -1 As III (NaAsO2) or As V (HAsNa2O4) at pH 6.9 (pH of the lake). The periphyton was then collected and pressed to remove the water, half was frozen and half was freeze-dried. As K-edge XANES spectra were recorded on the four samples. The spectra recorded in frozen hydrated and freeze-dried state were very similar ( Figure S2A), and linear combination fits provided similar results, with 3 to 5% difference in the percentages ( Figure S2 B-C). So it was concluded that freeze drying does not alter the speciation of As present in the periphyton.   (Wang and Qian, 2009) and 5ng of DNA. Thermal cycling was carried out in an AmpGene 9700 (ABI) as follows: 10 min at 95°C, 30 cycles at 95°C for 30 s, 60°C for 30 s and 72°C for 40 s and a final extension for 7 min at 72°C. Amplicons were sequenced using MiSeq 250-paired technology (Illumina), with V3 kit, in Getplage sequencing platform (INRA, Toulouse). Data were analysed using FROGS (Find, Rapidly, OTUs with Galaxy Solution) tool (Escudié et al., 2018).
Before statistical analysis, random sampling of filtered data was performed to obtain the same number of reads per sample. Taxonomic biomarkers of As hyperaccumulator periphyton were detected using the LEfSe algorithm (Segata et al. 2012). Briefly, a non-parametric Kruskal-Wallis (p-value <0.05) sum-rank test was performed to detect taxa with significant differential abundances, followed by a pairwise Wilcoxon test (p-value <0.05) in order to detect biological consistency of biomarkers. Finally, linear discriminant analysis (LDA, threshold of 2) leads to estimate the effect size of each differentially abundant taxon. Raw sequences were submitted to the National Center for Biotechnology Information Sequence Read Archive under the number PRJNA508881. Table S2. Concentrations in arsenic in filtered lake waters. As SC : arsenic concentration after filtration on As speciation cartridge, which removes As(V). Data for Fe, Mn, Cl -, S, SO4 2-, H2S, NO3 -, PO4 3-, DOC in filtered waters and pH, Eh, salinity are also presented. Numbers in italics and brackets give the standard deviation measured on triplicate samples. As SC: As content after As speciation cartridge, corresponding to As(III).   Table S4. Concentrations in arsenic and major elements (Fe, Mn, Ca, K, P, S, Al) in totora samples (in g g -1 dry weight) and bioaccumulation factor (BAF) in shoot for As.  There was no periphyton growing in UUH and RH sampling sites. n = number of samples. All periphyton samples from Huatarata (HU) and some of the samples from Cohana bay (BC2, BC3 and BC4)) were enriched in Ca. This enrichment is likely due to the presence of Characeae and other Ca-rich organisms or shells. Si was not analyzed by ICP-MS, but µXRF showed that this species was present as well. It may arise from the presence of diatoms and of detritic particles.    Figure S3. Examples of micro X-ray fluorescence spectra recorded on the periphyton PB4-BC2, after background subtraction. The freeze dried material was pressed into 5 mm pellets without grinding, and µXRF spectra were recorded on various spots of the pellets. Experimental conditions: 20 kV, 200 mA, 60 s acquisition time, beam diameter 300 µm. A variability in peak intensities was observed, indicating a high heterogeneity in the composition of the periphyton. As was detected in all spots, whereas it was not detected in periphyton samples from other sites. Figure S4. As K-edge XANES spectra for some As reference compounds (a) and for the periphyton samples (b, plain lines: experimental, dashed lines: linear combination fits). Figure S5. Structure of microbial communities of two non-hyperaccumulator periphytons (PB4-HU and PB5-BC3) and two As hyperaccumulators (PB4-TBC2-1 and PB4TBC2-2). Replicates are indicated by a lower vowel. Data represents the structure of the communities at the phylum level. Highlighted in dashed lines the class of Betaproteobacteria, belonging to Proteobacteria phylum and the Chloroplasts, related to (eucaryote) microalgae.