Environ. Sci. Technol. 2009, 43, 7237–7244
In Situ Monitoring of the Diurnal Cycling of Dynamic Metal Species in a Stream under Contrasting Photobenthic Biofilm Activity and Hydrological Conditions M A R Y - L O U T E R C I E R - W A E B E R , * ,† TEDDY HEZARD,† MATTHIEU MASSON,† ¨ RG SCHA ¨ FER‡ AND JO CABE, Department of Inorganic and Analytical Chemistry, University of Geneva, Sciences II, 30 Quai E.-Ansermet, CH-1211 Geneva 4, Switzerland, Universite´ de Bordeaux, UMR 5805 EPOC, Avenue des Faculte´s, 33405 Talence, France
Received January 27, 2009. Revised manuscript received April 8, 2009. Accepted April 14, 2009.
The diurnal evolution of the dynamic fraction, i.e., the potentially bioavailable fraction, of Cd, Cu, and Pb in a small river impacted by mining and smelting waste was studied in situ, under contrasting biofilm activity and hydrological conditions, using an automated voltammetric analyzer. The in situ, near realtime measurements revealed persistent dynamic metal species diurnal cycles. These cycles were affected mainly by the biochemical conditions rather than hydrological conditions. The data obtained from the in situ measurements, coupled with complementary laboratory analyses, revealed that: various processes control the diurnal dynamic metal species cycles in the studied site; the trends of the diurnal cycles of the dynamic metal species can be different from those observed for the dissolved metal species measured in filtered samples. Moreover, the dynamic fraction of a given cationic metal can show diurnal cycles with opposite trends depending on the environmental conditions. All these findings highlight the interest and importance of automated, continuous measurements of specific relevant environmental metal fractions, compared to punctual weekly or monthly traditional sampling strategies of total dissolved metal analysis, to allow more appropriate water quality control and reliable assessment of metal ecotoxicological impact.
Introduction The studies performed during the past decade in close-toneutral pH rivers and streams demonstrated that many trace metals undergo diurnal cycles, with total dissolved metal concentrations often changing 1- to 5-fold during a 24 h period (1-3). The growing data on diurnal metal cycles highlight that metal biogeochemical processes are very dynamic, and that short-term (daily and bihourly) variations in metal concentrations and speciation can be similar in magnitude to those previously thought to occur only at the seasonal time scale (4). This knowledge is important for both * Corresponding author phone: 0041-22-3796048; fax: 0041-223796830; e-mail:
[email protected]. † University of Geneva. ‡ Universite´ de Bordeaux. 10.1021/es900247y CCC: $40.75
Published on Web 05/01/2009
2009 American Chemical Society
interpreting existing data banks and designing more appropriate monitoring procedures for water quality control (5). Various possible biogeochemical and physical processes may cause these diel metal cycles (e.g., refs 3, 6): discharge fluctuations, redox and photochemical reactions involving metal (hydrous)oxides, precipitation/dissolution of carbonates or other minerals, sorption processes, and biological absorption and/or uptake-regeneration cycles. Among these processes, sorption processes induced either by pH diurnal cycle linked to metabolic activity of benthic photosynthetic biofilms (refs 3, 7, 8 and references therein), or temperaturedependent adsorption onto actively precipitating Fe, Mn, and Al oxides (2, 9, 10) were reported to be the main processes controlling metal diurnal cycles in a number of freshwater streams. However, most of these studies were performed in summer under low-flow conditions and thus little is known about the seasonal variability of diurnal metal cycles with respect to contrasting conditions/sites. Moreover, samples were collected hourly by hand, or with commercially available automated water samplers, immediately filtered, and acidified for later laboratory analysis. This intensive and laborious sample collection and processing approach typically limits hourly sampling to 1-3 days at maximum. More frequent analysis is required for further study and to understand the seasonal occurrence and magnitude of diel metal cycles. For this purpose, automated, continuous in situ chemical analyzers offer tremendous advantages over traditional sampling methods (e.g., refs 11, 12). While submersible probes for continuous monitoring of pH, dissolved oxygen, temperature, and conductivity are now routinely deployed for long-term, high resolution in situ monitoring, development of other in situ chemical analyzers has lagged far behind. Recently, however, few systems developed for in situ applications were systematically characterized, optimized, and applied in various environmental studies to monitor trace metals at time intervals of 1 min to 1 h using voltammetric techniques (11). They include gel-integrated microelectrodes (GIME) and Au/Hg microelectrodes incorporated in various types of deployable platforms. The GIME microsensor presents key features for environmental trace metal monitoring (12). In particular, when a microsensor such as the GIME is used with anodic stripping voltammetric technique under appropriate conditions, the voltammetric signal is selectively due to the so-called dynamic metal complexes, i.e., those which are sufficiently labile (large dissociation rate) and mobile (large diffusion rate) (13). In natural waters, the dynamic metal species include mostly the free metal ion and metal complexes with small inorganic (e.g., OH-, CO32-, SO42-) and few small organic ligands (e.g., malonate, citrate, some fulvic acids). These metal species represent the maximum fraction of metal potentially bioavailable (12, 14, 15), and thus their measurements are of prime interest to assess metal ecotoxicological impact. The GIME-based systems allow in situ, autonomous, simultaneous measurements of the dynamic species of Cu(II), Pb(II), Cd(II), and Zn(II) at subnanomolar concentrations at time intervals of typically 30 min to 1 h. Another important feature of the GIME is linked to the capability and efficiency of its gel membrane to minimize the chemical (fouling) and physical (ill-controlled hydrodynamic conditions) interferences (16). This characteristic is essential to obtaining a sensor that is reliable and robust enough to be used for in situ long-term monitoring. The objective of this work was to apply a GIME based analyzer to monitor in situ and simultaneously, at the hourly time scale, the temporal evolution of the dynamic fraction of Cd, Pb, and Cu in a stream affected by former coal mining VOL. 43, NO. 19, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7237
and Zn ore treatment waste. The simultaneous detection of these trace metals is of interest as (i) Cd and Pb are extremely toxic even at low concentrations, whereas Cu may be essential or toxic depending on the conditions; (ii) these metals have different complexation/adsorption properties and thus may present different temporal evolutions. In parallel, master variables (T, pH, O2, conductivity) were monitored in situ, and water samples were collected for complementary analyses (particulate and total dissolved-metal concentrations; DOC; fulvic-humic acids; particulate and total dissolved Mn/Fe (hydrous)oxide concentrations; alkalinity; and major ions). The data obtained from in situ measurements and laboratory analyses during two field work periods were integrated to understand the diurnal behavior of the dynamic metal species under contrasting hydrological and biogeochemical conditions and to identify the processes which control these cycles.
Experimental Section Monitoring/Sampling Site. In situ continuous monitoring of the concentration of the Cd, Cu, and Pb dynamic species (Cddyn, Cudyn, Pbdyn) and sampling for ancillary measurements were performed at the Joanis site near the outlet of the RiouMort watershed (France) (Supporting Information Figure S1) in the periods 26 April to 2 May 2007 and 21-28 April 2008. The Riou-Mort watershed (155 km2) is known as the major source of the historical polymetallic pollution of the LotGaronne-Gironde fluvial-estuarine system, resulting from former open-cast coal mining and Zn ore treatment (17, 18). Two decades after the end of mining and ore treatment activities, and despite important remediation efforts, leaching and mechanical erosion of metallurgic tailings still contribute to important amounts of metals/metalloids from various, partly unidentified, surface and subsurface sources (17) to the downstream river system. The Riou-Mort River (mean annual discharge 1.9 m3s-1; DIREN Midi-Pyre´ne´es; France) shows highly variable discharges, with high-intensity flash floods (8-12 h) (17). In Situ Monitoring of the Diurnal Cycling of the Dynamic Metal Species. Continuous monitoring of the Cddyn, Cudyn, Pbdyn concentrations, using square wave anodic stripping voltammetry (SWASV), was performed with a voltammetric in-line analyzer (VIA-Field) (www.Idronaut.it). This instrument is a portable version of the voltammetric in-situ profiling system (VIP) (11, 19) designed for real-time on-field monitoring and screening of the dynamic metal species in shallow rivers and runoff waters. The system consists of several units incorporated in a watertight plastic box: (i) an electronic housing containing a potentiostat and an internal computer to manage the voltammetric measurements, the data acquisition and storage, the communication interface for data exchange through a standard RS-232 cable with a laptop computer; (ii) a mini-flow through Plexiglas voltammetric cell; (iii) a peristaltic pump; and (iv) an online oxygen removal system (20). The voltammetric cell is based on a threeelectrode system: a mini- Ag/AgCl/KCl saturated 3% LGL agarose gel (19) and Pt reference and counter electrodes, respectively, and a GIME working electrode. A detailed description of the GIME microsensor production, preparation and analytical merits was given elsewhere (11, 21). Briefly, the GIME array consists of 5 × 20 interconnected iridium microdiscs, each of 5 µm diameter and a center to center spacing of 150 µm, surrounded by a 300 µm thick Epon SU-8 containment ring for the gel. Before use, the Epon SU-8 containment ring is filled with a 1.5% LGL agarose gel and Hg hemispheres are plated on the Ir microdiscs by electrochemical reduction at -400 mV of an acidic 5 mM Hg(CH3COO)2 solution. During the field studies, GIME Hg-film deposition and calibration of the VIA-Field, before and after field monitoring, 7238
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 19, 2009
were performed in the field laboratory. Calibrations were performed in suprapur 0.1 M NaNO3 solutions spiked with Cd(II), Pb(II), and Cu(II) in ranges of 5-50 nM, 2-8 nM and 4-15 nM, respectively. After calibration, the VIA-Field, powered by a 12 V battery, was installed on the riverbank of the Joanis site for autonomous monitoring, under preprogrammed fluidic and SWASV optimized conditions (19, 20), of the target analytes over a 5-6 days period at intervals of 1.5 h and 1 h, respectively, in 2007 and 2008. Water was sampled in the middle of the stream (∼0.3 m above the gravel sediment) using an acid precleaned Tygon tubing coupled to the peristaltic pump of the VIA-Field. Data were collected daily, to check proper functioning of the system. The same Hg hemispheres were used for the calibrations performed before and after the VIA-Field field deployment and during the whole field measurement periods. Standard deviations of the calibration slopes obtained before and after field deployment were 2 weeks); total dissolved metal (Mediss) (in 0.45 µm pore size membrane filtered samples acidified at pH 1 with HNO3 suprapur); and total acid extractable particulate metal (Meac-part) obtained by the subtraction of Mediss to Metot. Water samples were collected using an in-house 12 V battery powered peristaltic pump with acid precleaned Tygon tubing. Filtrations were performed online on the site using either 0.45 µm (and in few cases 0.02 µm) pore size nitrocellulose membranes (Whatman) or, for organics, precombusted (450 °C; 6 h) 0.7 µm pore size GF/F membranes (Whatman) in in-house flow-trough filtration devices. Sampling polypropylene and glass bottles were precleaned using the following procedures: 24 h in 0.1 M HNO3 suprapur, 2 times 24 h in 10-2 M HNO3 suprapur, followed by dipping in Milli-Q water for 12 h after each acid washing step; and 5% DeconexR detergent, 12 h in 10-2 M suprapure HNO3, 12 h in ultrapure water, followed by precombustion 6 h at 500 °C, respectively. Samples for DOC and FA-HA measurements were collected in aluminum wrapped and precleaned glass bottles. Samples for measurements of Metot, Mediss, alkalinity, major ions were collected in precleaned polypropylene bottles, and stored in double polyethylene bags. The samples were acidified on site using HNO3 suprapur (Metot, Mediss) or HCl suprapur (major cations, DOC and FA-HA) acids. Samples were immediately stored in cold boxes, then back to the field laboratory, at 4 °C, in the dark, excepted samples for DOC which were frozen and samples for alkalinity
FIGURE 1. Hydrological data (hourly discharge and water level), organic (DOC values, humic and fulvic acid concentrations, and ratios between HA-FA and DOC) and Mn/Fe species (total dissolved and acid extractable particulate Mn and Fe) concentrations measured in the Riou-Mort River (Joanis site) during the two sampling campaigns. and PO43- measurements, which were analyzed every evening in the field laboratory. Alkalinity was measured by titration with 0.02 M HCl using bromocresol green as indicator and PO43- was analyzed using a Hach spectrometer and the Hach standard procedure. The Metot and Mediss fractions were analyzed by ICPMS (HP 4500) using Rhodium as internal standard. The quality of the analyses was checked by the measurements of standards, international certified reference water (SLRS-4) and blank solutions measured after every batch of 10 samples. Major cations as well as total and total dissolved Mn and Fe were measured by AAS or FAAS (PYE-Unicam Philips); major anions by ionic chromatography (Metrohm model 761). DOC was analyzed after purging with CO2-free air by high temperature combustion using a Shimadzu Total Organic Carbon Analyzer (Shimadzu, TOC 5000). Concentrations of FA-HA were determined by adsorptive voltammetry in presence of Mo(VI) (22) using a µ-Autolab potentiostat (EcoChemie) coupled to a Metrohm VA663 Stand.
Results and Discussion Hydrological Conditions. Although the 2007 and 2008 field campaigns were conducted both at the end of April, large differences in water level and discharge at the monitoring site occurred (Figure 1a) due to very contrasting weather conditions. Spring 2007 was unusually dry with discharges clearly below typical seasonal values and below the annual average (1.9 m3s-1) except for 17 April, when daily discharge reached 3.8 m3s-1. During the in situ monitoring period (26 April to 2 May), water level was ∼0.4 m and average discharge was 0.54 m3s-1 (Figure 1a). In contrast to 2007, heavy rainfalls occurred in April 2008 inducing four short floods, including two major flood events, with daily average discharges representing 5-fold (April 10: Qd ) 10 m3s-1; April 18: Qd ) 10.2 m3s-1) and 10-fold (April 11: Qd ) 19.5 m3s-1; April 20/21: 29.6 m3s-1) the average annual value. Hydrological conditions during the monitoring period (21-28 April) were strongly influenced by the major flood event of April 20 (instantaneous water discharges of up to 73m3s-1 (Figure 1a)), i.e., water discharges and levels VOL. 43, NO. 19, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7239
TABLE 1. Minimum and Maximum Values Measured during Each Sampling Campaign (2007 and 2008) for Alkalinity; Major Cation and Anion Concentrations; Total and Total Dissolved Mn and Fe Concentrations 2007 studied period 26 April to 01 May -1
2008 21 April to 24 April (12 h)
24 April (12 h) to 27 April
alkalinity (meq L /HCO3)
2.0-2.2
1.3-1.5
1.6-1.8
Ca2+ (mM) K+ (mM) Mg2+ (mM) Na+ (mM) Si (mM)
1.3-1.8 0.20-0.24 1.1-1.2 0.61-0.85 0.05-0.07
0.83-1.1 0.08-0.10 0.43-0.80 0.27-0.38 0.17-0.18
1.1-1.3 0.10-0.14 0.83-1.2 0.42-0.49 0.17-0.19
Cl- (mM) SO42- (mM) NO32- (mM) F- (µM) NO2- (µM) PO43- (µM)
0.41-0.48 2.0-2.4 0.13-0.62 13.4-13.8 8.2-26 1.7-2.9
0.16-0.22 0.31-0.73 0.06-0.12 10-11 1.6 0.84-1.7
0.18-0.30 0.75-1.5 0.08-0.17 10-12 1.4-4.2 1.1-1.5
Mntot (µM) Mndiss (µM) Fetot (µM) Fediss (µM)
2.9-6.2 2.8-6.0 1.6-4.6 0.13-0.78
3.3 - 11 1.0-3.7 7.6-51 0.53-1.3
3.5-5.7 1.4-6.0 4.4-7.0 0.22-0.61
decreased first rapidly from 52 m3s-1 and 2.3 m respectively to level off at about 4.3 m3s-1 (10-fold 2007 water discharge) and 0.75 m respectively from April 24 noon to April 27 midnight (SI Table S1), then increased again when rainfall restarted (Figure 1a). Water Composition, Organic and Mn/Fe Species Content. The water composition at the Joanis site during the 2007 and 2008 campaigns is summarized in Table 1 and SI Table S1. The temporal evolution of organic contents and Mn/Fe species is given in Figures 1b-d. During the 2007 campaign, concentrations of FA/HA were 2-2.8 mg L-1 and, by assuming similar C content in the Suwannee River FA (52%), represented typically 25-50% of the DOC (Figure 1b). Mntot and Fetot concentrations were rather similar (1.5-6 µM; Table 1), but showed opposite distributions between acid extractable particulate and total dissolved forms (Figures 1c-d); i.e. g93% of Mn in Mndiss (0.45 µm). The concentrations of major ions were usually relatively constant (Table 1). During the 2008 monitoring period, starting one day after a major flood event, the water composition, organic matter and Mn/Fe species contents showed contrasting variability. The concentrations of FA-HA, DOC, Feac-part, Mnac-part, Siac-part (not shown), and Fediss varied as a function of discharge, i.e., they were higher at the beginning of the study, then decreased rapidly during the first two days and leveled off during April 24 noon to April 27 midnight, to finally increase again when a new raining event started (Figures 1b-d). The Mndiss concentration (Figure 1c) as well as the concentrations of the major cations and SO42- (not shown) showed inverse trends. During the period April 24 noon to April 27 midnight 2008, concentrations of organic compounds, Mn/Fe species and major ions were similar to those obtained during low discharge in 2007 (Figure 1; Table 1; and SI Table S1). This suggests that, except for discharge peaks, the water composition (major ions, organic, and redox species) at the Joanis site is not fundamentally different for contrasting hydrological conditions as those observed in April 2007 and 2008 (Table 1 and SI Table S1). Dissolved Mn and Fe concentrations in the
