Mutual interaction between arsenic and biofilm in a mining impacted river

Mutual interaction between arsenic and biofilm in a mining impacted river Laura Barral-Fraga a, ⁎, Diego Martiñá-Prieto b , María Teresa Barral b ,SoizicMorin c , Helena Guasch a a Institute of Aquatic Ecology, Department of Environmental Science, University of Girona, Girona, Spain b Department of Soil Science and Agricultural Chemistry, Faculty of Pharmacy, Campus Vida, University of Santiago de Compostela, Santiago de Compostela, Spain c Irstea, UR EABX, Bordeaux, France HIGHLIGHTS •Epilithic biofilms growing in a goldmining impacted river accumulate high As concentrations. •Arsenic inhibits algal growth and increases bacterial and dead diatom densities. •Methylated As-species found intracellularly suggests As-detoxification by biofilms, even under eutrophic conditions. •Nutrients, DOC, temperature or light availability must be considered when analyzing effects of As in freshwater ecosystems. GRAPHICAL ABSTRACT abstractarticle info Article history: Received 2 February 2018 Received in revised form 17 April 2018 Accepted 21 April 2018 Available online 3 May 2018 Editor: Daniel Wunderlin Gold mining activities in fluvial systems may cause arsenic (As) pollution, as is the case at the Anllóns River (Galicia, NW Spain), where high concentrations of arsenate (As V ) in surface sediments (up to 270 mg kg −1 )were found. A 51 day-long biofilm-translocation experiment was performed in this river, moving some biofilmcolonized substrata from upstream (less As-polluted) to downstream the mine area (more As-polluted site), to explore the effect of As on benthic biofilms, as well as their role on As retention and speciation in the watersediment interface. Eutrophic conditions (range: 0.07–0.38 mg L −1 total phosphorus, TP) were detected in water in both sites, while sediments were not considered P-polluted (below 600 mg kg −1 ). Dimethylarsenate (DMA V ) was found intracellularly and in the river water, suggesting a detoxification process by biofilms. Since most As in sediments and water was As V , the high amount of arsenite (As III ) detected extracellularly may also confirm As V reduction by biofilms. Furthermore, translocated biofilms accumulated more As and showed higher potential toxicity (higher As/P ratio). In concordance, their growth was reduced to half that observed in those non-translocated, became less nutritive (less nitrogen content), and with higher bacterial and dead diatom densities. Besides the high As exposure, other environmental conditions suchas the higher riparian coveratthe more As-polluted site could contribute to those effects. Our study provides new arguments to understand the contribution of microorganisms to the As biogeochemistry in freshwater environments. © 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Keywords: Biospeciation Microalgae Diatoms Bacteria Microbial ecotoxicology Microbial biogeochemistry 1. Introduction Microorganisms constitute the majority of all living matter on Earth, most of them living in the form of multicellular aggregates commonly Science of the Total Environment 636 (2018) 985–998 ⁎Corresponding author. E-mail address: [email protected] (L. Barral-Fraga). https://doi.org/10.1016/j.scitotenv.2018.04.287 0048-9697/© 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Contents lists available at ScienceDirect Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv referred to as biofilms, attached communities consisting especially of bacteria, algae, fungi and protozoa embedded within a polysaccharide matrix and generally located in close association with wet surfaces and interfaces (e.g., Mora-Gómez et al., 2016). Biofilm communities play a role in the environment both in maintaining and improving the ecological health of freshwater ecosystems (Romaní, 2010;Battin et al., 2016). In rivers, biofilms are the first to interact with dissolved substances, such as pollutants, having a high capacity to accumulate heavy metals from the surrounding environment, and being able to actively influence their sorption, desorption and transformation (Behra et al., 2002;Guasch et al., 2010). Furthermore, they are also a site for biotransformation and/or transfer of chemicals to other aquatic organisms (e.g., Guasch et al., 2016). As a major component of benthic biofilms, diatoms (microscopic, unicellular brown algae) are considered good indicators of environmentalconditions due to theirquick response to environmental changes and their cosmopolitan distribution in aquatic systems, making them widely used in ecotoxicological studies (Morinet al., 2012;Luís et al., 2013, 2016).By carryingoutecotoxicological studies in the field, the effects of pollution may be evaluated under real exposure conditions, using a set of biofilm parameters (i.e. endpoints or biomarkers) together with the analysis of water chemistry and the prevailing environmental conditions (Guasch et al., 2010, 2016, 2017). For instance, biofilm translocation experiments in fluvial systems using biofilm developed on artificial substrates are considered an active biomonitoringapproach toassess the effects of metal pollution on these natural communities (Bonet et al., 2014;Morin et al., 2016). Another interesting aspect of biofilms is that they allow the coexistence of microniches of different physiological requirements, allowing the simultaneous, but spatially separated occurrence of opposite redox processes in the same biofilm environment (e.g., Kulp et al., 2004;Huang, 2014). This characteristic contributes to the fact that biofilms play a major role in driving biogeochemical cycles (e.g., Huang, 2014). For instance, autotrophic and heterotrophic communities in biofilms may drive, directly or indirectly, a complex interplay of As mobilization, sequestration and transformation (speciation) processes that determine the fate of this metalloid in the environment (Huang, 2014;Wang et al., 2015). Arsenic may be mobilized during gold mining activities since gold- and As-bearing minerals coexist (e.g., Garelick et al., 2009). In river systems, arsenic is predominantly bound to sediments, which may contain high amounts of this element, especially in mining areas (e.g., Drewniak and Sklodowska, 2013). In these areas, arsenic concentrations can reach up to hundreds or thousands of mg kg −1 in sediment (Smedley and Kinniburgh, 2002;Rubinos et al., 2011). Furthermore, mean As concentrations of 137 μgL −1 , and even up to 7900 μgL −1 ,havebeenfoundin river waters close to mining areas (Smedley and Kinniburgh, 2002), clearly exceeding the concentration limit of chronic As exposure in freshwaters set at 150 μgL −1 (Criteria Continuous Concentration, CCC) by the Aquatic Life Criteria (USEPA, 2014). Sediments may act as a sink for pollutants that can subsequently be released upon changes in environmental conditions (e.g., Magbanua et al., 2013). Particularly, changes in water chemistry, such as the introduction of high concentrationsof phosphorous (P), may promote the release of As from solid phases through competitive ligand-exchange reactions (e.g., Rubinos et al., 2011). Both the As total concentration and the As chemical form (As species) have to be considered in toxicological analyses, since a key factor in the As mobility and toxicity is its chemical speciation. In natural waters and soils, arsenic is mostly found in inorganic form (iAs) asarsenate (As V , the thermodynamically stable state under oxic conditions)and arsenite (As III , the predominant species in anoxic conditions) (Smedley and Kinniburgh, 2002;Oremland and Stolz, 2003;Sharma and Sohn, 2009). Both Eh (oxidation/reduction potential measurement) and pH are considered the most important abiotic factors controlling As speciation (Smedley and Kinniburgh, 2002). Usually, the form As III is more toxic for the environment thanAs V . In contrast, algae are more sensitive to the As V than to the reduced form As III (Levy et al., 2005;Wang et al., 2015). Microorganisms contribute to the production of these As species and may perform other biotransformations, such as As methylation, resulting in organic products like monomethylarsenate (MMA V )and dimethylarsenate (DMA V ) (e.g., Prieto et al., 2016a). Microorganisms may also produce more complex organic compounds, such as arsenosugars and arsenolipids (e.g., Huang, 2014). According to Hellweger et al. (2003), arsenic biotransformation in microalgae depends on P availability: in P-deficient conditions, As V is uptaken by algae, reduced to As III and further methylated as final products; in contrast, under P-enriched conditions, As V is uptaken and reduced, but not all of it is methylated, resulting in As III excretion into the environment. These transformations have an enormous impact on the environmental behavior of As, since the different chemical forms of As exhibit different toxicity towards higher organisms: usually methyl As III NAs III NAs V Nmethyl As V (e.g., Huang, 2014). In biofilms, arsenate competes with phosphate (PO 4 −3 ) assimilation (essential element for algal growth), due to the similarity of their chemical properties and structure (e.g., Prieto et al., 2013). This similarity leads to cell toxicity due to the replacement of phosphate by As V in, for instance, phosphorylated compounds, which are vital for the cycling of ATP, inhibitingbothphotophosphorylationand oxidative phosphorylation (Levy et al., 2005;Rahman and Hassler, 2014;Wang et al., 2015). Due to the similar properties between As V and PO 4 −3 molecules, the estimation of biofilm As/P ratio allows a better predictability of the As toxicity in microorganisms (e.g., Wang et al., 2013). The implications of As biogeochemistry on the ecology of freshwaters have been poorly addressed, despite having proved the key role of microorganisms, in particular microalgae and bacteria, on the As detoxification and fate in aquatic environments as in mining impacted rivers. With the aim of assessing the fate and effects of the As pollution in a freshwaterecosystem, we carried outa field study (translocation experiment) with benthicbiofilms at the Anllóns River (Galicia, Spain), where high As concentrations in soils and river bed sediments caused by old gold-mining activities were found (Rubinos et al., 2003;Devesa-Rey et al., 2008;Rubinos et al., 2010;Costas et al., 2011). We also intended to study the influence of benthic biofilms on As retention, transformation and mobilization at the water-solid interface of this fluvial polluted-system. Most As in the sediments of the Anllóns River is associated to low-mobility phases (Devesa-Rey et al., 2008;Rubinos et al., 2011). However, previous laboratory experiments show that As biogeochemistry at the Anllóns River is also affected by the biological status of the river sediments, covered by biofilms (Devesa-Rey et al., 2009) mainly constituted by diatoms (Martiñá Prieto et al., 2016). Therefore, we proceed from the assumption that the fluvial biofilm plays a crucial role in the mobilization of As from the sediment to other compartments. We expected to find a causal link between the presence of As in mobile forms in river sediments and their accumulation in the biofilm, causing changes in its structure and function. We also hypothesized that biofilms may transform and excrete into the water the bioadsorbed or bioaccumulated As by transforming it into other inorganic and organic As species (biospeciation), modulating As toxicity in the environment. A set of analyses was carried out to study the ecology of the Anllóns river in two sites, upstream and downstream the mine area, focusing on the relationship between environmental conditions, especially As exposure, and the structure and function of the biofilm community. 2. Material and methods 2.1. Study site This experimentwas carried outat the Anllóns River (Fig. 1), located in Galicia (Northwest Spain). The river is 71 km long, with a mean slope of 7.6%, and an average annual flow of 10.77 m 3 s −1 . The basin covers an area of 516 km 2 mostly dedicated to forestry, agriculture and cattle 986 L. Barral-Fraga et al. / Science of the Total Environment 636 (2018) 985–998 farming. Gold mining activities were active in the area during the Roman Empire and between the years 1895 and 1910. Gold in this area is associated with pyrite and arsenopyrite (Nespereira, 1978). Concentrations of As up to 4000 mg As kg −1 have been detected in superficial soil horizons due to the presence of arsenopyrite mineralization associated with gold ores in hydrothermal quartz veins (Boixet et al., 2007). High total As contents were also detected in the sediments of the Anllóns River, downstream of the gold-mining area (Rubinos et al., 2003), which were attributed to natural geogenic As enrichment exacerbated by miningactivities (Devesa-Rey etal.,2008).Recently, the possibility of exploiting this mineralized zone again was considered, causing social concern among locals. With regard to water quality, there are two important sources of P pollution near the main town of Carballo, located upstream of the gold-mining area (at a distance of 14 km approximately): a wastewater treatment plant and a seafood canning factory (Rubinos et al., 2003). According to EU Directive 75/ 440/CE (Díaz-Fierros, 2003), the Anllóns River has been classified as A2 (potabilization requires physical and chemical treatments) with the exception of a stretch of the river, downstream of Carballo, which is classified as A3 (physical, intense chemical and disinfection treatments). In this research, we carried out a translocation experiment between two sites with similar lithology (alkaline gneiss) and physicochemical conditions, but with different As concentrations in the sediments. We selected one site upstream of the mining area referred to as Upstream site or Up site (coordinates 43.222149, −8.782352); and a second one located just downstream of themining area,with a higher As concentration, referred to as Downstream site or Down site (coordinates 43.230118, −8.831897). 2.2. Experimental set-up Small (8.5 × 2 cm) and large (12 × 9 cm) sandblasted glass tiles, fixed to cement slabs with neutral silicone sealant (Quilosa, Madrid, Spain), were placed horizontally on the riverbed at the Upstream site for natural biofilm colonization and growth (Fig. 2). The use of artificial glass substrates reduces the heterogeneity that occurs on natural substrates (Cattaneo et al., 1998), making quantitative sampling easier, and allows tocontrol the maturity degree of the biofilm for the analyses. Translocation was carried out after 5 weeks of biofilm colonization: half of the artificial substrates weremoved from the Upsite to the Downsite. The experiment started at that moment and several samplings were carried out in both sites to simultaneously assess the fate and effects of As on biofilms. The experiment lasted 51 days, from July 8th until August 28th 2014. Fig. 1. Study site at the Anllóns River (Galicia, NW Spain). “U”(in gray circle) represents the Upstream sampling site (located upstream of the mining area); “D”(in grey circle) represents the Downstream sampling site (located downstream of the mining area). The grey arrow indicates the biofilm translocation from the Upstream to the Downstream sampling site. 987L. Barral-Fraga et al. / Science of the Total Environment 636 (2018) 985–998 2.3. Sampling Biofilm and river water were sampled 5 times during the experiment: on days 2, 4, 7, 22 and 51 after translocation. Two samplings were carried out for dissolved As concentration: after 22 days of translocation at the Downstream site, and at the end of the experiment (day 51 after translocation) at both sampling sites. Sediments were sampled twice: on days 7 and 51 after translocation. 2.3.1. River water and sediment sampling and preparation Triplicate water samples (1 L) were collected and immediately filtered using GF/F glass microfiber filters (Whatman, ~0.7 μmofpore size) for nutrients analyses, including total dissolved phosphorus (TP), total dissolved nitrogen (TN) and total dissolved organic carbon (DOC). In the laboratory, some samples were filtered through 0.45-μm cellulose nitrate membrane filters NCS 045 47 BC (Albet LabScience, Dassel, Germany) for soluble reactive phosphorus (SRP); and some were filtered with sterile 0.45-μm Whatman Puradisc 25AS™syringe filters (GE Healthcare Europe GmbH, Barcelona) for DOC determination. For the analysis of dissolved As, water was filtrated through 0.45 μm nylon membrane filters (Whatman). All river water samples were frozen until analysis. The GF/F filters were also kept for dry weight (DW) determination of river water suspended solids (SS). Complex sediment samples were collected using a small plastic shovel from the top 5 cm of the riverbed sediment at various points from each site, and mixed in situ in hermetic plastic containers, topped up to prevent oxidation during the transport to the laboratory. There, sediment samples were centrifuged (3000 rpm, 15 min) and then solid sediment samples were freeze-dried and sieved (b2 mm). 2.3.2. Biofilm sampling Colonized artificial substrates were sampled at random and in triplicate in the field, and the biofilm was scraped into Falcon® tubes or glass vials (Anorsa, Spain). Forbacterial densityand diatom analyses (relative abundance and quantification of live diatom community), samples were preserved in glass vials with 5 mL of filtered river water with 37% formaldehyde (Panreac, Spain) (dilution 0.5:4.5), with 3 replicates for each one. All samples were kept in dark conditions in the fridge until they were analyzed in the laboratory (1 month for diatom analyses; 3 months for bacterial density). In turn, biofilm from large substrata was scraped into Falcon® 50 mL tubes and then some fresh samples were used to determine the concentration of Total bioaccumulated As and its species distribution in the different biofilm compartments, while other samples were freeze-dried to determine total biofilm dry weight biomass (DW) and elemental composition. Non-scraped samples from small artificial substrates were transported in triplicate to the laboratory, in dark conditions and in an oxygenated box with river water, for in vivo fluorescence analyses. 2.4. Sample analysis Most chemical analyses were doneusing certified reference material and following international and national standard methods (APHA, 1995, 2005;AENOR, 1994, 2012). 2.4.1. Environmental endpoints 2.4.1.1. Light measurements. Environmental light intensity (μmol photons s −1 m −2 ) was measured in situ with a light sensor (LICOR Inc., Lincoln, Nebraska, USA) at each sampling time at both sites. Light measurements were carried out on submerged biofilm in order to measure light reaching benthic biofilm, as well as below the riparian zone and below a nearby zone without vegetation to calculate the cover (%) of the sampled river sections. 2.4.1.2. River water analyses. Water temperature, dissolved oxygen, pH and electrical conductivity were measured in the field at each sampling time using multiparametric probes (WTW METERS, Weilheim, Germany). Redox potential was determined with a HANNA HI 9025 portable pH-Eh meter equipped with a Pt combination redox electrode (Hanna Instruments, Eibar, Spain). Eh values obtained with the Pt-Ag/ AgCl electrode were corrected to refer them to the standard hydrogen electrode by adding 245 mV. Suspended solids (SS) were analyzed in river water during the experiment (on days 2, 7, 22 and 51 after the translocation) in both sampling sites, according to APHA (1995). Dissolved phosphorus was determined by means of an acid digestion of the filtered samples (~0.7 μm) with 1 mL of H 2 SO 4 31% and 0.4 g of (NH 4 ) 2 S 2 O 8 at 121 °C during 30 min (APHA, 2005). Phosphorus in the extracts was determined using the phosphomolybdenum blue method described by Murphy and Riley (1962). Soluble reactive phosphorus was determined in 0.45 μmfiltered samples following the same colorimetric method. Total dissolved nitrogen was determined by digestion of filtered samples (~0.7 μm) with H 2 SO 4 and 3% H 3 PO 4 using the Kjeldhal method (UNE-EN 25663: AENOR, 1994). Dissolved organic carbon was determined in 0.45 μmfiltered samples using a Total Organic Carbon Analyzer Model TOC-5000 (Shimadzu, Kyoto, Japan), by subtracting the inorganic carbon (IC) concentration from the total carbon (TC) concentration. Arsenic concentration in river water was analyzed in samples taken at translocation day 22 and 51 following the method described by Prieto et al. (2016a). Total As concentration was determined in 0.45-μmfil- tered samples by Inductively Coupled Plasma Spectrometry (ICP-MS, Varian 820MS), equipped with collision reaction interface (CRI) Fig. 2. Experimental setup at the Anllóns River. Artificial substrates before the biofilm colonization (a) and after 5 weeks of colonization (b), when the biofilm was translocated from the Upstream to the Downstream site. 988 L. Barral-Fraga et al. / Science of the Total Environment 636 (2018) 985–998 technology to reduce polyatomic interferences. For As speciation (As V , As III ,DMA V , MMA V and Arsenobetaine, As-Bet), High-Performance Liquid Chromatography coupled with Inductively Coupled Plasma Spectrometry (HPLC-ICP-MS) was used. The detection limits under the experimental conditions were 2.8, 4.1, 2.9, 4.6 and 2.5 ng L −1 for As V , As III , MMA V ,DMA V and As-Bet, respectively. The certified reference material EnvironMAT-Drinking Water HIGH EP-H-1 (Catog. number: 140- 025-032, SCP Science) was used for quality control. 2.4.1.3. Sediment sample analysis. The pH and Eh were analyzed in situ with a HANNA HI 9025 portable pH-Eh meter equipped with a Pt combination redox electrode (Hanna Instruments, USA). Particle size distribution was determined by dry sieving (2, 1, 0.5, 0.25, 0.1 and 0.05 mm). A representative aliquot of the b2 mm fraction was milled and sieved (b50 μm) for specific analysis. Total phosphate was determined in previously ignited (450 °C for 1 h) samples (b50 μm), using the molybdenum blue method (Murphy and Riley, 1962) after acid digestion (concentrated HF + H 2 SO 4 + HCl 10:1:10, 220 °C). Total Kjeldhal nitrogen was determined by wet digestion of samples (b50 μm) with H 2 SO 4 , following the Kjeldahl method as described in Guitián and Carballas (1976). The determination of % OM was carried out following the UNE-EN 13039 standard (AENOR, 2012), by calcination at 450 °C for 2 h of 5 g dried samples. Total carbon was calculated using the equation %C = %OM/1.724. TotalAs concentration in the sediment (b50 μm) was determined by X-ray fluorescence (XRF) spectrometry (custom built, equipped with a Philips high-voltage generator and a Mo anode of 2.2 Kw as X-ray source), following the considerations described by Devesa-Rey et al. (2008). The accuracy of the XRF measurement of total As was checked by using the certified reference material BCR CRM-277b. For this reference material, the As concentration (mg kg −1 ) measured was 45.4 ± 4.1 (certified value 47.3 ± 1.6). The easily-extractable As concentration (total As and As speciation) was obtained from freeze-dried sediment samples (b2 mm fraction) using a phosphate buffer, following the method of Gleyzes et al. (2002).The total As concentration as well as the As speciation (As V , As III ,DMA V , MMA V and As-Bet) were determined by ICP-MS (Varian 820MS) and by HPLC-ICP-MS (Varian Prostar 230 HPLC-Varian 820MS), respectively. The detection limits were similar to those determined for water analysis. Bioavailable concentrations of As at the Down site were measured twice (days 7 and 51 after translocation) using diffusive gradients in thin films (DGT) devices (DGT Research Ltd., Lancaster, UK) placed at the interface water-sediment, and following the procedures indicated by Prieto et al. (2016a). The mass of As in the resin gel (M), the timeaveraged DGT concentrations (C DGT ) and the flux (F) of As measured by DGT were calculated according to Zhang and Davison (1995), and DGT® technical documentation. 2.4.2. Biological endpoints 2.4.2.1. In vivo fluorescence measurements. Several small biofilmcolonized artificial substrates (8.5 × 2 cm 2 )wereusedtoanalyze in vivo fluorescence in the laboratory, using a MINI-PAM fluorometer (HeinzWalz, Effeltrich, Germany), at a constant temperature (20 °C) and from a constant distance between the light-emitting diode and the samples (5 mm). First, the minimum fluorescence yield (F 0 )was given by the fluorometer in dark adapted samples and then a saturation pulse was applied to obtain the maximum PSII quantum yield (Ymax). After 15 min of light adaptation, a saturation pulse of actinic light was applied to the samples to obtain the effective PSII quantum yield (Yeff). According to Corcoll et al. (2012a),theF 0 parameter can be used as an estimation of algal biomass. The Ymax is defined as a measurement of the photosynthetic capacity of the community, whereas the Yeff is a measurement of the community photosynthetic efficiency. 2.4.2.2. Biofilm elemental composition. Total DW biomass was obtained by weighing freeze-dried samples of biofilm, from which nutrient stoichiometry (C, N, P content and molar ratios) was analyzed. Carbon and N biogenic elements were determined using an elemental analyzer (PerkinElmer 2400). In turn, TP biogenic element was determined after biofilm digestion in 4 mL of concentrated HNO 3 (65% Suprapure, Merck, Germany) and 1 mL of H 2 O 2 (33% Suprapure Merck). Digestion transformed all organic P forms into inorganic forms and, then, total dissolved P content was obtained according to the method detailed in the SRP protocol. Quantification of the C, N and P elements of the benthic biofilm were performed on a dry weight basis, and C/N and N/P biofilm molar ratios werethen calculated (Sternerand Elser,2002;Muñoz et al., 2009;Scharler et al., 2015). 2.4.2.3. Bacterial density. Bacterial density was determined by flow cytometry, adapted from Amalfitano et al. (2009) and Perujo et al. (2016) for biofilm samples. Samples (1 mL) were placed in a glass vial for sonication with 4 mL of detaching solution (1/5 sample dilution) added tohelp separate cells and avoid their aggregation. After sedimentation of larger biofilm cells (and possibly some sediment particles), a sample purification process was carried out with Nycodenz® (Sigma- Aldrich, USA) to ensure the elimination of those larger particles and cells in the supernatant (1/10 diluted sample). Then, 400 μLofsample (1/10 diluted) were stained with Syto13 (4 μL Fisher, 5 μM solution) and incubated in dark conditions for 15–30 min (1/100 final sample dilution). To normalize fluorescence data, bead solution (10 μLof10- 6 beads mL −1 ,Fisher1.0μm) was added to the samples in a known concentration. The fluorescence color of the beads was yellow-green (505/515), with maximum absorption and emission wavelengths at 495 nm and 505 nm, respectively. Bacterial density was measured using flow cytometry (FACSCalibur, Becton–Dickinson) with a selected size of ~0.1 μm (corresponding to most mean heterotrophic bacterial size). Results are referred to as bacteria cm −2 . 2.4.2.4. Quantitative estimates of live diatom community. The quantitative estimates of live diatoms were carried out according to Morin et al. (2010), using a Nageotte counting chamber (Microgravure Precis, France) and a light microscopy at a 10× magnification (photomicroscope Nikon Eclipse 80i, Nikon Co., Tokyo, Japan). Counting wasseparated into 2 types: empty cells that were considered ‘dead’and cells occupied by chloroplasts were considered ‘alive’.Datawererecorded as cells per unit area of sampled substrate (cells cm −2 ). 2.4.2.5. Relative abundances of the diatom species. Samples for diatom community identification were prepared in permanent slides as recommended by Leira and Sabater (2005), after digestion with 35% HCl and 30% H 2 O 2 .Diatomidentification to the lowest taxonomic level possible was carried out following standard references and recent nomenclature updates: i.e., Krammer and Lange-Bertalot (1986–1991) and Coste and Rosebery (2011). About 400 frustules were counted per slide, as far as possible, using a light microscope (Nikon E600, Tokyo, Japan) with Nomarski differential interference contrast optics at a magnification of ×1000. Additionally, diatom species richness (S), Shannon-Wiener index of diversity (H), and species evenness (J) were calculated (Shannon and Weaver, 1949;Pielou, 1975). 2.4.2.6. Biofilm arsenic content. Extracellular and intracellular As extraction was carried out in translocated biofilms of the Down site on days 7 and 51 after translocation. For the extractions, 2 g of biofilm sample were used. As a first step, a rinse solution was obtained by washing thesamples with river water to extract the soluble As slightly associated to the matrix of extracellular polymeric substances (EPS) and cell surfaces. Subsequently, a sequential extraction was performed following the procedure of Levy et al. (2005) for extracellular As, consisting in a phosphate extraction, and that of Miyashita et al. (2009) for intracellular As, consisting in its extraction from lysed cells with a methanol- 989L. Barral-Fraga et al. / Science of the Total Environment 636 (2018) 985–998 water solution. The details of the extraction procedure are indicated in Prieto et al. (2016a). The total As concentration, as well as the As speciation (As V ,As III ,DMA V ,MMA V and As-Bet) were determined by ICP-MS (Varian 820MS) and by HPLC-ICP-MS (Varian Prostar 230 HPLC-Varian 820 MS), respectively. The detection limits were similar to those determined for water analysis. Total amount of As in the biofilm was also analyzed, in freeze-dried biofilm samples, after digestion in 4 mL of concentrated HNO 3 (65% Suprapure, Merck, Germany) and 1 mL of H 2 O 2 (33% Suprapure Merck), using a high performance microwave digestion unit (Milestone, Ethos Sel, Sorisole (BG), Italy). Then, samples were diluted to 15 mL with milli-Q water and dissolved total As concentration was measured using ICP-MS (7500c Agilent Technologies, Inc. Wilmington, Denmark). 2.5. Data analysis Homogeneity of variances and normality of data were checked prior to statistical analyses, using SPSS v19.0 software. Before being included in the analysis, variables (except pH and Eh of the river water and the sediment were transformed: most of them were ln(x + 1)- transformed (sample sizen ≥3), includingmost ofbiofilmmetrics(photosynthetic parameters, that is Fo,Ymax and Yeff, live diatom quantification, diatom diversity indices, diatom relative abundance, bacterial density, biofilm DW, biofilm TP and TN content, total As and Asspecies accumulated in biofilms, and the light irradiance reaching biofilm communities); but proportions and percentages were “arcsine square root”transformed (%C, %N, %P, C/N, N/P and As/P all in biofilms, and % riparian cover). In the cases where sample size were too small (n = 2) only mean values were calculated (that is, in river waters: DOC, TP, TN, SRP, total As and As species; in sediments:TP, TN, TC,OM, total As by XRF and the easily extractable-As). The pH and Eh of water and sediment, as well as the water velocity, conductivity, temperature, and its content on oxygen and nutrients (TP, TN, SRP) were not transformed (n = 1). Changes in biofilm biomass (Fo) during both colonization (5 weeks) and experiment period under As exposure (51 days) were fittoa3- parameter log-normal curve using SigmaPlot v.11.0. We tested for differences between the estimated parameters using the following equation: y¼ae −0:5x−x0 b ðÞ 2 where y represents the periphyton biomass, ais associated with the peak of the curve, bis the rate of inhibition after the peak, x 0 is the time required to reach the maximum or peak value, and xis time in days. For other biofilm metrics and light measurements, a Two-Way Repeated Measures ANOVA was carried out with all the variables that were analyzed in triplicate every sampling day. That is, the riparian cover percentage, the light irradiance reaching biofilm communities, and the biofilm parameters, such as photosynthesis (Fo,Ymax and Yeff), live diatom quantification, biogenic elements (%C, %N and %P) and their ratios (C/N and N/P), as well as the total As concentration accumulated and the As/P ratio in the biofilm. The time variable (expressed as translocation days) was the within-subject continuous variable, while the treatment (the Downstream and Upstream sites) was the between-subject variable. The As in different biofilm compartments (total As and As species) and bacterial density were checked by Two-Way ANOVA, since samples had been taken twice during the experiment, while One-Way ANOVA was applied to diatom diversity indices (S,H,J). Finally, post hoc Bonferroni's tests were applied to locate significant differences. All these analyses were carried out using SPSS v19.0 software. Diatom relative abundance (≥1%) of the last day (translocation day 51) was represented in a Non-Metric Multidimensional Scaling plot (NMDS), to show possible variations of community composition between sites. Multi-response permutation procedures (MRPP) were used to test for inter-site versus intra-site heterogeneity in diatom community structure (Zimmerman et al., 1985) based on Bray Curtis distance (same as for the NMDS). PC-ORD software (version 6.0; McCune and Mefford, 2011) was used to perform the analyses. The effect of the environmental factors on the biological responses was studied using redundancy data analysis (RDA), using variables taken at both sites (Up, Down) and every sampling day (2, 4, 7, 22 and 51). Previously, variables with a strong inter-correlation were eliminated to avoid collinearity, and then, two principal component analyses (PCA) were carried out to select the variables with the highest explained variance between sampling sites and time: one PCA using the explanatory variables (mainly physico-chemical) and the other one using the response variables (mainly biological variables). Finally, the RDA was carried out on the explanatory variables that best explained the variation of the response matrix (Borcard et al., 2011), corresponding to the light cover percentage, river water velocity and conductivity, river water nutrients (TP, TN, SRP), pH and Eh of sediment, and the biofilm As/P ratio. Biofilm As/P ratio was considered an explanatory variable due to the fact that it may determine the activity and metabolism of the biofilm, since it is the best proxy to estimate As toxicity in cells. The response variables comprised some biofilm metrics as DW, TP and TN content and C/N ratio, as well as photosynthetic parameters (Fo and Yeff) and live diatom density. Correlation analyses were carried out using SPSS v19.0 software. The “vegan”package of R statistical software (version 3.2.2 for Windows; www.r-project.org) was used to carry out both PCAs and the final RDA. The dynamics and distribution of As species (mean %) in the different environmental and biofilm compartments of the Downstream site (sediment, river water, biofilm rinse solution, extra- and intracellular biofilm) were represented in cumulative bar charts. 3. Results 3.1. Site conditions 3.1.1. Site physicochemical data Physical and chemical conditions during the experiment are summarized in Tables 1a, 1b and 2. Riparian cover was high at both sites, but especially at the Down site (Table 1a), causing slightly differences in light conditions between both sampling sites. Low light irradiance in biofilms was detected on days 7 and 22 (b50 μmol photons s −1 m −2 ), especially at the Down site on day 22 (4.90 ± 0.24 μmol photons s −1 m −2 ). In contrast, high irradiances (N1000 μmol photons s −1 m −2 ) enough to cause photo-inhibition were measured on day 2, proving high and significantly temporal light variability. River water (Table 1a) showed a well oxygenated status, neutral pH and low mineralization. High concentrations of TN and TP were detected, especially at the end of the experiment, when high SRP concentrations were also observed. Conversely, higher concentrations of suspended solids were found at the beginning of the experiment. Low As concentrations were detected in all river water samples (ranging from 0.56 at the Up site to 1.83 μgL −1 at the Down site). The sediments had neutral pH (ranging from 6.80 to 7.59) and Eh values corresponding to a suboxic state (100 mV bEh b400 mV) at this pH. A higher percentage of OM was observed in the sediments of the Up site and especially at the beginning of the experiment (in July), coinciding also with high TN concentrations. These values decreased during the experiment, resulting in high values of the C/N ratio, especially at the Up site at the end (Table 1b). Regarding the total amount of As in sediments, very high concentrations (ranging from 31 to 100 mg kg −1 ) were detected at both sites, and especially at the Down site (Table 1b), where the highest concentrations of the easilyextractable As (ranging from 1.77 to 2.23 mg kg −1 ) were also found (Table 1b). Moreover, the As time-averaged DGT concentration measured at the Down site is indicative of the average As concentration 990 L. Barral-Fraga et al. / Science of the Total Environment 636 (2018) 985–998 released from the sediment, was of 0.36 μgL −1 and, although the total As retained by DGT devices increased with time (0.05 μgonday7and 0.25 μg on day 51), the As time-averaged concentration decreased over the experiment (0.44 ± 0.00 μgL −1 after 7 days of translocation; 0.29 ± 0.00 μgL −1 after 51 days, at the end of the experiment). Finally, sediment samples showed grain size differences between sites (Table 2), with a higher percentage of fine sand at the Down than at the Up site. 3.1.2. Benthic biofilm analyses The evolution of the biofilm after the translocation of artificial substrates and until the end of the experiment is reflected in Table 3,showing significant differences between the translocated and nontranslocated biofilms for the following biofilm metrics: the minimum fluorescence yield (Fo), the density of live diatoms and total bacteria, theamountof accumulated As and the As/P ratio, as well as theelemental composition (% C, % N and % P) and the C/N and N/P ratios. In the non-translocated biofilms (Upstream site), Fo was higher (mean value 345.95 ± 145.90) than in the translocated ones (195.14 ± 45.70). Furthermore, the Fo parameter was significantly fitted to a curve (Fig. 3) showing how the growth of the translocated biofilms located at the Downstream site reached half of those non-translocated (Up site). The percentages of C, N and P were generally higher in the non-translocated biofilms (Up site, with mean values of 23.27 ± 4.35 for % C, 3.97 ± 0.83 for % N and 0.19 ± 0.03 for % P), than in those translocated at the Down site (showing values of 18.27 ± 5.99 for % C, 2.68 ± 0.89 for % N and 0.12 ± 0.03 for % P). However, C/N and N/P molar ratios were higher in the translocated biofilms (7.08 ± 0.84 of C/N and 22.09 ± 5.41 of N/P) than in those non-translocated (5.97 ± 0.17 of C/N and 20.50 ± 2.64 of N/P). Live diatom density was, in accordance with the Fo parameter, higher in the non-translocated biofilms (492.99 × 10 3 ± 193.35 × 10- 3 cells cm −2 ) than in those translocated (4.92 × 10 3 ±2.75×10- 3 cells cm −2 ). In contrast, a higher density of bacteria was found in the translocated biofilms (Down site), especially at the end of the experiment. During the experiment, the As concentration was generally higher (p = 0.017) in the translocated biofilms (mean values of 15.20 ± 6.85 μgAsg −1 biofilm at the Down site) than in the non-translocated ones (mean values 8.82 ± 2.24 μgAsg −1 biofilm at the Up site). Differences were more apparentin the As/P ratio (p b0.001), reaching mean values of 8.75 × 10 −3 ± 5.10 × 10 −3 in the translocated biofilms, while 2.27 ×10 −3 ±1.17×10 −3 were found in the non-translocated biofilms. Summarizing, the biofilms at the Up site (non-translocated biofilms) contained higher nutrient content (lower C/N and C/P ratios), as well as a higher algal biomass (Fo) and had more live diatoms than at the Down Table 1a Physico-chemical properties: environmental light measurements (riparian cover percentage and light reaching biofilms) and physico-chemical properties of river water in the Upstream (Up)and Downstream (Down) sampling sites of the Anllóns River, on days 2, 4,7, 22 and 51 (corresponding to the last day of the experiment) after the translocation. Single values or mean values are shown (n = 3 for riparian cover, light irradiance reaching biofilm and total As on 22Down; n = 2 for DOC, nutrients and total As; n = 1 for pH, Eh, conductivity, oxygen and temperature). nd = no data. Statistical results of riparian cover for effects on time (F = 151.741, p b0.001; degrees of freedom, df = 4) and sites (F = 1593.701, p b0.001; df = 1) were achieved by Two-Way Repeated Measures ANOVA andBonferroni's test, as well as results of light irradiancereaching biofilm for effects on time (F = 1192.900; p b0.001; degrees of freedom, df = 4) and sites (F = 7.093; p = 0.056; df = 1). *Significant differences (p ≤0.05) between sampling sites at a specific date (from Bonferroni's test). ªValues of pH and Eh of sediments (n = 1) are indicated in brackets with the river water pH and Eh values, respectively. Translocation date & Site Environmental light River water a Riparian cover (%) Light reaching biofilm (μmol photons s −1 m −2 ) pH Eh (mV) Electrical cond. (μScm −1 ) O 2 (mg L −1 ) T (°C) Water velocity (m s −1 ) SS (mg L −1 ) DOC (mg L −1 ) TP (mg L −1 ) SRP (mg L −1 ) TN (mg L −1 ) Total As (μg L −1 ) 2Up 90.09 ± 0.81* 1023.00 ± 113.32* 7.07 (7.59) 415 (215) 141 8.88 17.60 0.98 3.60 nd 0.07 0.03 1.07 nd 2Down 98.62 ± 0.14* 1454.67 ± 32.96* 7.16 (7.16) 493 (232) 142 9.28 17.70 0.38 12.60 nd 0.08 0.06 0.81 nd 4Up 83.82 ± 0.19* 156.20 ± 3.40* 7.05 (7.24) 431 (240) 145 9.41 16.50 0.93 nd nd 0.08 0.03 1.03 nd 4Down 95.00 ± 0.64* 40.90 ± 1.44* 7.19 (7.24) 470 (241) 145 9.02 17.80 0.35 nd nd 0.09 0.05 1.09 nd 7Up 80.89 ± 2.22* 31.01 ± 0.49* 7.00 (7.38) 431 (191) 149 8.77 17.60 0.87 0.60 nd 0.08 0.04 0.50 nd 7Down 97.66 ± 0.24* 42.69 ± 0.58* 7.22 (7.30) 461 (329) 146 8.98 18.70 0.34 3.10 nd 0.11 0.04 0.36 nd 22Up 93.30 ± 0.95* 45.02 ± 0.34* 6.81 (7.55) 420 (224) 153 8.53 19.10 0.67 1.90 4.45 0.09 0.01 3.65 nd 22Down 91.53 ± 0.26* 4.90 ± 0.24* 7.06 (7.13) 459 (250) 152 8.92 19.20 0.55 1.90 3.39 0.09 0.05 1.94 1.83 51Up 71.20 ± 2.51* 155.43 ± 7.21* 6.83 (6.80) 519 (253) 158 8.00 18.90 0.49 1.20 2.92 0.38 0.07 2.15 0.56 51Down 88.38 ± 1.12* 52.51 ± 2.13* 7.13 (6.86) 464 (281) 161 8.86 18.70 0.20 0.50 2.64 0.19 0.06 3.15 0.79 Table 1b Physico-chemical properties of sediments in the Upstream (Up) and Downstream (Down) sampling sites of the Anllóns River, on day 7 after the translocation and day 51 (the end of the experiment). Mean values are shown (n = 2). Values of pH and Eh are shown in Table 1a. Translocation date & Site Sediment OM (%) TC (mg kg −1 ) TN (mg kg −1 ) C/N TP (mg kg −1 ) Total As (XRF) (mg kg −1 ) Easily-extractable As (mg kg −1 ) 7Up 4.91 28.50 × 10 3 1169.73 24 363.25 31 1.32 7Down 2.71 15.70 × 10 3 604.21 25 381.75 54 2.23 51Up 2.33 13.50 × 10 3 327.22 41 402.80 51 0.66 51Down 2.13 12.40 × 10 3 379.68 32 253.10 110 1.77 991L. Barral-Fraga et al. / Science of the Total Environment 636 (2018) 985–998 site, where the biofilms had a higher bacterial density. The As content of the biofilms and the As/P ratio were also higher at the Down site, evidencing higher potential As toxicity. 3.1.3. Diatom community composition At the end of the experiment, 34 species of diatoms were identified, and 4 more only at genus rank (see the Supplementary table in the online version). The most abundant species identified were Cocconeis placentula Ehrenberg var. placentula (55.37%) and Mayamaea atomus (Kützing) var.atomus(18.34%). Regarding sampling sites, C. placentula var. placentula (67.60%) and Cocconeis pediculus Ehrenberg (14.16%) were the most abundant species at the Up site (non-translocated biofilms), whereas M. atomus (33.48%) dominated in the samples of the Down site (translocated biofilms) together with C. placentula var. placentula (43.21%). No significant differences were found between the translocated and non-translocated biofilms in the diatom diversity indices at the end of the experiment (S, H and J). The NMDS ordination showed slight differences in the diatom assemblages between the samples at this moment (Fig. 4), suggesting natural variability between communities ofthesampledsites(inter–replicates).However,lessvariability of species composition was found within than between sites (MRPP value: A = 0.14, p= 0.12). 3.1.4. The influence of environmental parameters on the biofilm responses Redundancy Analysis (RDA) plot (Fig. 5)showstheinfluence of environmental parameters (in grey) on the responses attributed to the biofilm (parameters in italics and black), in every sampling day (2, 4, 7, 22 and 51) and sampled site (D for Down site, U for Up site). Based on this analysis, 72.17% of the total variance attributed to all physicochemical parameters is by the riparian cover, the SRP, TP and TN of the river water, the conductivity and velocity of the river water, the pH and Eh of sediments, and the biofilm As/P ratio asa parameter of potential toxicity. Almost half of the total variance (49.16%) is expressed by the first axis, which arranges the sampling sites according to the N and P contents, live diatoms and algal biomass (Fo), all of which are higher in the biofilms located at the Up site, where environmental conditions are described by higher water TN, TP and electrical conductivity. In turn, the C/N ratio was higher in the translocated biofilms at the Down site, where a higher As/P biogenic ratio was also detected, as well as the highest riparian cover. Therefore, sampling site distribution on axis 1 followed a gradient related to canopy cover, water chemistry and As pollution, separating most Downstream (on the left) from Upstream samples (on the right of the plot). The second axis explains 23.01% of the total variance and shows an event that happened on day 7 in both sampling sites, related to an increase in water velocity and water turbidity, causing an increase in Table 2 Grain size distribution (%) of the sediments of the Anllóns River, in the Upstream and Downstream sampling sites, on 7 and 51 days after the translocation. Grain size fraction Upstream Downstream 7 days 51 days 7 days 51 days b50 μm 11.9 8.1 6.9 4.5 50–100 μm 2.1 2.5 3.4 2.9 100–250 μm 22.4 18.4 46.6 42.4 250–500 μm 15.0 13.6 37.0 41.7 500–1000 μm 16.1 18.6 5.0 5.7 1–2 mm 32.5 38.8 1.1 2.8 Table 3 Biofilm metrics of the non-translocatedand translocated biofilms located Upstream (Up) and Downstream (Down) of the mine area, respectively, sampled at several times after the translocation: on days 2, 4, 7, 22 and 51 (corresponding to the last day of the experiment). Value units may refer to the surface of the colonized artificial-glass substrates (cm 2 )orthebiofilm freeze-dried mass (g). Mean values±standard deviations (n = 3) are shown. nd = no data. Except for Bacteria density, all statistical results (F and p) for effects on time (degrees of freedom, df = 4) and treatment (df = 1) were achieved by Two-Way Repeated Measures ANOVA and Bonferroni's test. For Bacteria density, statistical results (F and p) for effects on time (df = 1) and treatment (df = 1) were achieved by Two-Way ANOVA and Bonferroni's test. Significant differences in time or site effects are set at p ≤0.05. *Significant differences (p ≤0.05) between sampling sites at a specific date (from Bonferroni's test). (*) Marginal significance (0.05 bp≤0.1) between sampling sites at a specific date (from Bonferroni's test). Translocation date & Site Biofilm Fo Ymax Yeff Live Diatoms (×10 3 cells cm −2 ) Bacteria (×10 6 cells cm −2 ) DW (mg cm −2 ) C (%) N (%) P (%) C/N N/P As (μg g −1 ) As/P 2Up 271 ± 112.46 0.46 ± 0.14 0.28* ± 0.05 4.96 ± 1.27 0.12 ± 0.04 0.40 ± 0.31 13.96 ± 4.26 2.30 ± 0.79 0.13 ± 0.02 6.13 ± 0.24 17.39 ± 3.23 10.42 ± 2.27 0.004* ± 0.002 2Down 179.33 ± 55.89 0.48 ± 0.28 0.35* ± 0.18 9.97 ± 8.21 0.08 ± 0.04 1.24 ± 1.25 12.84 ± 5.98 1.48 ± 0.38 0.10 ± 0.03 8.36 ± 1.96 16.04 ± 7.07 17.73 ± 6.43 0.013* ± 0.006 4Up 147.67* ± 15.31 0.54 ± 0.06 0.40 ± 0.02 7.66 ± 10.96 nd 0.66 ± 0.29 27.23 ± 8.32 4.70 ( * ) ± 1.60 0.21 ± 0.08 5.85* ± 0.25 23.11 ± 1.77 7.76 ± 2.16 0.002 ± 0.001 4Down 228.00* ± 18.73 0.58 ± 0.04 0.34 ± 0.03 2.06 ± 1.24 nd 1.23 ± 0.61 14.87 ± 5.52 2.21 ( * ) ± 0.93 0.09 ± 0.03 6.80* ± 0.30 23.92 ± 2.93 24.51 ± 18.55 0.014 ± 0.010 7Up 354.33 ± 206.36 0.28 ± 0.14 0.23 ± 0.13 0.95* ± 0.28 nd 3.30 ± 1.51 16.22 ± 4.41 2.57 ± 0.74 0.14 ± 0.04 6.34 ± 0.17 18.66 ± 3.37 7.89 ± 3.42 0.002 ± 0.002 7Down 306.67 ± 34.65 0.25 ± 0.19 0.22 ± 0.17 1.88* ± 0.45 nd 1.10 ± 1.29 20.12 ± 2.71 3.20 ± 0.48 0.14 ± 0.02 6.3 ± 0.19 23.21 ± 4.18 12.60 ± 3.16 0.005 ± 0.002 22Up 434.33 ± 84.05 0.47 ± 0.06 0.30 ± 0.08 12.20* ± 3.62 nd 1.10 ± 0.46 26.49 ± 0.40 4.51 ± 0.15 0.25 ± 0.01 5.87 ± 0.11 17.91 ± 0.85 8.78 ± 1.56 0.002 ± 0.000 22Down 231.67 ± 103.16 0.51 ± 0.06 0.40 ± 0.06 5.79* ± 1.12 nd 1.43 ± 0.56 25.26 ± 9.74 3.81 ± 1.78 0.15 ± 0.03 6.88 ± 0.94 25.18 ± 7.46 8.80 ± 1.07 0.004 ± 0.001 51Up 522.43* ± 311.31 0.48 ± 0.08 0.40 ± 0.04 2439.22* ± 950.64 0.11* ± 0.16 1.47 ± 0.54 32.44 ± 4.36 5.75 ( * ) ± 0.87 0.23 ± 0.00 5.65* ± 0.09 25.43 ± 3.97 9.26 ± 1.78 0.001* ± 0.001 51Down 179.20* ± 16.04 0.55 ± 0.02 0.47 ± 0.01 27.41* ± 6.35 1.97* ± 1.24 1.15 ± 1.08 23.45 ± 9.11 3.77 ( * ) ± 1.61 0.15 ± 0.04 6.34* ± 0.41 23.91 ± 4.88 12.38 ± 5.05 0.008* ± 0.006 Time effect F = 1.449 F = 2.984 F= 2.735 F = 35.045 F = 2.942 F = 1.630 F= 4.669 F = 5.711 F = 5.170 F= 3.547 F= 2.547 F = 0.888 F = 1.826 p = 0.264 p = 0.051 p= 0.066 pb0.001 p = 0.125 p = 0.215 p= 0.011 p= 0.005 p= 0.007 p= 0.030 p= 0.080 p= 0.494 p = 0.173 Site effect F = 10.936 F = 0.063 F= 0.312 F = 8.323 F = 6.312 F = 2628.636 F= 7.357 F= 17.822 F= 25.761 F= 8.129 F= 17.413 F= 15.417 F= 145.397 p = 0.030 p = 0.814 p= 0.606 p = 0.045 p = 0.036 p = 0.127 p= 0.053 p= 0.013 p= 0.007 p= 0.046 p= 0.014 p= 0.017 pb0.001 992 L. Barral-Fraga et al. / Science of the Total Environment 636 (2018) 985–998 biofilm DW and a decrease in the Yeff, and illustrating the effects of a previous rain event. 3.2. Arsenic speciation In river water (Table 4), arsenate was the dominant species (99% at the Down site on day 22 and 72.5% on day 51; and almost 89% at the Up site on day 51). At the end of the experiment, As III , DMA V and As-Bet percentages were higher at the Down site, where these species represented all together N27% of the total As, but only 11% at the Up site. In contrast, only inorganic As species were foundin thesediment samples: As V was the predominant As species in both sites, and especially at the Down site, where it represented N86% of the total As (Table 4). As was previously mentioned, the translocated biofilms accumulated more total As than the non-translocated biofilms. For this reason, arsenic speciation was focused on the translocated biofilms (Down site), in the extracellular and intracellular biofilm compartments (Table 5). The total As in the rinse solution, which contains soluble As slightly associated to cell surfaces and EPS matrix, showed the same proportion of As V and As III on day 7, but As V dominated at the end of the experiment. A low amount of DMA V was also detected in this rinse solution (around 2% on day 7 and 0.6% on day 51). Arsenic in the extracellular compartment, where it is adsorbed to cell surfaces and EPS matrix, showed a species distribution very similar to that of the rinse solution, but with higher DMA V concentrations at the end of the experiment. Finally, the total As concentration in the intracellular compartment at the end of the experiment was dominated by As V species, followed by an important amount of DMA V (around 25% on day 7 and 18% on day 51) and As III . Low percentages of As-Bet (ranging from 0.13 to 1.35%) appeared in all the compartments atthe end of the experiment. No MMA V was detected in any sample. In summary, while in river water and sediment As V was the predominant species, other As species were relevant in the biofilm, namely, As III in the rinse solution and extracellular compartment (particularly on day 7), and also DMA V in the intracellular fraction. 4. Discussion 4.1. Site characterization Several environmental factors measured in the sediments and river water were similar at both sampling sites, attributed in particular to the similar lithology and intense farming activities in the basin. Thus, eutrophic conditions were detected in the water in both sites during the whole experiment (experimental ranges: 0.07–0.38 mg TP L −1 and 0.36–3.15 mg TN L −1 ), especially at the end, as most values correspond to eutrophic conditions set at 0.075 mg TP L −1 and 1.5 mg TN L −1 by Dodds et al. (1998). Furthermore, high concentrations of SRP, commonly analyzed asa measurement of the immediately available orthophosphate(PO 4 −3 ) in water and effectively used for predicting algal production (Dodds, 2006;Allan and Castillo, 2007), confirmed these eutrophic conditions (experimental range: 0.01–0.07 mg SRP L −1 , usually 0.01–0.03 mg SRP L −1 ,accordingtoMainstone and Parr, 2002). Fig. 3. Changes in biofilm biomass during the “biofilm colonization days”versus the “experiment days”at the Upstream site (plot on the left) and the Dowstream site (plot on the right). Fitting of Fo parameter over time (days) follows a 3-parameter log-normal curve, assuming growth inhibition at the end of the experiment, and where the aparameter is associated with thepeak of thecurve;the bparameter is the rate of inhibition after the peak;the x 0 parameter isthetime requiredto reach the maximum orpeakvalueand xistime in days. Vertical dotted lines indicate the translocation day (before that time, all samples were located exclusively in the Upstream site, corresponding to the colonization period). Differences between the parameters for each sampling site were tested with two-way ANOVA in order to check if changes in the growth of the communities were statistically significant. Significance was set at p ≤0.05. Red lines indicate 95% confidence band. Fig. 4. Nonmetric dimensional scaling (NMDS) plot showing sampling sites ordination (Up, in green; Down, in red) according to their diatom species composition at the end of the experiment (day 51 after biofilm translocation to the Downstream site). Species abbreviation: ADMI, Achnanthidium minutissimum (Kützing) Czarnecki; AMII, Achnanthidium minutissimum (Kützing) Czarnecki f. inconspicuum (Østrup) Compère & Riaux-Gobin; ADSU, Achnanthidium subatomus (Hustedt) Lange-Bertalot; CPED, Cocconeis pediculus Ehrenberg; CPLA, Cocconeis placentula Ehrenberg var. placentula; ENMI, Encyonema minutum (Hilse in Rabhenhorst) D.G. Mann in Round, Crawford & Mann; EOMI, Eolimna minima (Grunow) Lange-Bertalot; GGRA, Gomphonema gracile Ehrenberg; GPAR, Gomphonema parvulum (Kützing) Kützing var. parvulum f. parvulum; MAAT, Mayamaea atomus (Kützing) Lange-Bertalot var. atomus;NCPR,Navicula capitatoradiata Germain; NGRE, Navicula gregaria Donkin; NFON, Nitzschia fonticola Grunow in Van Heurck; NPAL, Nitzschia palea (Kützing) W. Smith var. palea;PLFR, Planothidium frequentissimum (Lange-Bertalot) Lange-Bertalot; PTLA, Planothidium lanceolatum (Brebisson ex Kützing) Lange-Bertalot; PPDT, Planothidium pseudotanense (Cleve-Euler) Lange-Bertalot; PRBU, Planothidium robustius (Hustedt) Lange-Bertalot; POBG, Psammothidium oblongellum (Østrup) Van de Vijver; UULN, Ulnaria ulna (Nitzsch) Compère; FIST, Fistulifera sp. Species representing ≥1% in at least one sample. 993L. Barral-Fraga et al. / Science of the Total Environment 636 (2018) 985–998