Correspondence: Catarina M. Magalhães, Centre of Marine and Environmental Research (CIIMAR), University of Porto, Rua dos Bragas, no. 289, 4050-123 Porto, Portugal. Tel.: +351 22 206 2285; fax: +351 22 206 2284; e-mail: email@example.com
In this study, the effects of increasing copper (Cu) concentrations on the denitrification pathway and on the diversity of a denitrifier community and gene expression were evaluated in intertidal sandy sediments of the Douro River estuary (north-west Portugal). The results demonstrated that sediment denitrification rates were highly affected by Cu; almost complete inhibition (85%) of the process was observed in sediments amended with 60 μg Cu g−1 wet wt sediment. Moreover, the addition of Cu stimulated the accumulation of both N2O and NO2− and inhibited the rate of NO3− uptake. Further, the amendment with even the lowest Cu concentration (4 μg Cu g−1 wet wt sediment) yielded a drastic decrease in the abundance of nirK, nirS and nosZ (between 79% and 81%) assessed by means of real-time PCR. In agreement, reverse transcription-PCR–denaturing gradient gel electrophoresis analysis of nirK, nirS and nosZ transcripts showed a progressive decrease in the diversity of the transcription products of these genes with an increase of the Cu concentration.
Estuaries are also sinks of metals of different origins (Williams et al., 1994; Almeida et al., 2004). Once introduced into the aquatic environment, contaminants can accumulate in sediments (Williams et al., 1994), and the bioavailability and toxicity to the biota are determined by metal speciation (e.g. Burton & Scott, 1992). Trace metals can negatively affect aerobic and anaerobic microbial respiration, biomass, N-mineralization, nitrification and the microbial community structure of soils, sediments and aquatic habitats (Giller et al., 1998; Holtan-Hartwig et al., 2002; Granger & Ward, 2003). In addition, some investigators have demonstrated inhibitory effects of metals on denitrification (Sakadevan et al., 1999; Holtan-Hartwig et al., 2002; Magalhães et al., 2007a). However, the question of whether metal contaminants can directly affect essential biogeochemical processes at the genomic and transcriptional levels, i.e. at the ‘preprotein’ level, has rarely been considered (Kourtev et al., 2009). In the present study, we examine how one major metal pollutant, copper (Cu), can affect denitrification activity with regard to the diversity of genotypes and transcripts of nitrite (nirS and nirK) and nitrous oxide reductase (nosZ) genes in an estuarine microbial community. The Douro River estuarine sediments show a clear signature of anthropogenic contamination of Cu, indicating possible toxicity for sediment macrobiota (Mucha et al., 2003, 2004, 2005). In addition, the watershed drains the vineyards of the Port Wine-producing region, which are regularly sprayed with Cu-based fungicides that represent a major source of Cu to the Douro River estuarine system.
Materials and methods
Site description and sample collection
The Douro River estuary is located in the second most populated coastal zone of Portugal. Three main cities (Porto, Vila-Nova-de-Gaia and Gondomar; 700 000 inhabitants) are located within the estuary, exerting an important anthropogenic pressure in terms of urban runoff, sewage discharge, land reclamation and onshore constructions. A total of eight wastewater treatment plants drain into the estuary. In addition, herbicides and pesticides used in the vast port wine watershed are transported to the Douro River estuary. For this study, sediment samples were collected from the dominant intertidal sandy flat, within the lower Douro River estuary (for site location, see Magalhães et al., 2003). These sediments are mainly composed of highly permeable, coarse sand and gravel (>0.5 mm) with total recoverable Cu <4.5 μg Cu g−1 dry wt sediment (Magalhães et al., 2007a). However, a range of Cu concentrations between 1.2 and 94 μg Cu g−1 dry wt sediment have been measured in the Douro River estuarine sediments (Mucha et al., 2003). High positive redox potentials were measured in situ within the 0.5 cm depth (Eh=144.9±20.1 mV) (Magalhães et al., 2007a). This site has been characterized physically, chemically and biologically in previous studies (e.g. Mucha et al., 2003; Magalhães et al., 2005a, b). The sampling survey was performed during March 2007, at low tide and at a single location. A total of 20 cores (3 cm diameter and 8 cm long) were collected within 50 cm distance between each other, covering an area of approximately 3 m2. In addition, 5 L of estuarine water was collected nearby the sample site and stored in acid-cleaned, polyethylene bottles. The 20 cores collected were homogenized, to be used as a composite sample, stored in sterile plastic bags and transported to the laboratory in the dark, in refrigerated ice chests for processing.
Sediment samples were exposed for 6 days to a gradient of CuCl2 concentrations (0, 4, 8, 60 and 154 μg Cu g−1 wet wt sediment) above the natural levels (<4.5 μg Cu g−1 dry wt sediment). For each treatment and controls, nine serum bottles with 6–8 g of wet sediment and 5 mL of filtered (0.2 μm, Schleicher & Schuell membrane filters) estuarine water, amended with 2 mM of glucose and 300 μM of NO3− (KNO3), were incubated in serum bottles open to the air, in the dark, under constant stirring (80 r.p.m.) and temperature (20 °C) with the respective concentration of Cu. After 6 days of Cu exposure (Magalhães et al., 2007a), three of the nine replicates were used to measure denitrification and N2O production rates, three replicates were used to measure NO2− and NO3− net fluxes and the remaining three replicates were used to extract total DNA and RNA.
Denitrification rates were measured using the acetylene inhibition technique according to Sørensen (1978). Briefly, each serum bottle, containing sediment sample and incubation water (amended with 300 μM KNO3 and 2 mM glucose), was hermetically sealed with a butyl stopper and aluminum crimp, and purged for 15 min with helium to remove O2. Each of the three replicates with the different concentrations of Cu was divided into two subsamples to be incubated with and without acetylene (20% v/v). Incubation was performed in the dark for 4 h at a constant temperature (20 °C) and stirring rate (80 r.p.m.). Glucose and NO3− were added so that C and N would not limit denitrification; thus, our measurements reflect the potential rates of the processes under those particular conditions. The linearity of the processes during incubations was confirmed in previous experiments (Magalhães et al., 2005a). At T0 and T4 h, 12 mL of gas was collected (after headspace equilibration via vigorous shaking) from each serum bottle and injected into a Varian gas chromatograph (CP-3800) equipped with an electron-capture detector, two Hay Sep D columns and an automatic back flush system to prevent C2H2 from passing to the detector. The gas samples were collected from each serum bottle by simultaneously adding 12 mL of a 3 M NaCl solution and recovering the gas displaced (Joye et al., 1996). N2O was quantified using daily standard curves generated from certified gas standards, and the detection limit of the method was approximately 20 nM N2O. Nitrous oxide production rates were calculated as the difference between T0 and T4 h N2O concentration in the treatments without C2H2, and the N2 produced via denitrification was calculated as the difference between the N2O produced with and without acetylene (Joye et al., 1996). Nitrite and NO3− production rates were measured in triplicate separate slurries. In this case, after 6 days of metal exposure, samples were processed in the same way as described above for N2 and N2O production rate measurements, but at T0 and T4 h (separate triplicate samples for 0 and 4 h), 2.5 mL of overlying water was collected instead of the gas sample. These samples were centrifuged, 0.2 μm filtered and frozen (−20 °C) for later quantification of NO2− and NO3−. Nitrite and NO3− flux rates were calculated by the difference between concentrations of these N compounds measured at T0 and T4 h. The nitrite concentration was measured as described in Grasshoff et al. (1983) and NO3−+NO2− was determined using the spongy cadmium reduction technique (Jones, 1984). The actual NO3− concentrations were calculated as the difference between NO3−+NO2− and NO2−.
Nucleic acid extraction
Total DNA was extracted from 1 g wet weight of sediment using a PowerSoil DNA isolation kit (MoBio Laboratories Inc., Solana Beach, CA). The reproducibility of the amount of the DNA extracted was tested in triplicate sandy sediment samples (CV=14%). The efficiency of DNA extraction from sandy sediment samples was tested according to Okano et al. (2004) by adding a known number of Ruegeria pomeroyi cells to the sediment. The number of R. pomeroyi cells in the liquid medium was determined after 4′,6′-diamidino-2-phenylindole (DAPI) staining under an epifluorescence microscope, and the nosZ abundance was calculated on the basis that R. pomeroyi has one nosZ copy per cell (Moran et al., 2004). A total of 4.0 × 106R. pomeroyi cells were added to 1 cm3 of sandy sediment and DNA was extracted in triplicate as described above. The calculated nosZ abundance in samples with R. pomeroyi cells, was on average 8.0±0.3 × 106 cells cm−3 of sediment, and the difference between nosZ abundance in the samples with and without target DNA addition measured by real-time PCR was 1.1±0.3 × 106 cells cm−3 of sediment. Therefore, the extraction efficiency of the PowerSoil DNA isolation kit was 27.5±2.2%, which is in agreement with DNA recovery efficiencies calculated in other studies (e.g. Mumy & Findlay, 2004; Okano et al., 2004).
Total RNA was extracted from 2 g wet weight of sediment using the PowerSoil RNA isolation kit (MoBio Laboratories Inc.). DNA was removed from RNA by treatment with 1 U μL−1 of DNase I (Sigma) using DNA- and RNA-free reagents, followed by PCR using general 16S rRNA gene bacteria primers to check for traces of genomic DNA contamination. Reverse transcription (RT) was performed using the Omniscript RT kit (Qiagen) by adding 13 μL of total RNA to an 18-μL RT mixture according to the manufacturer's instructions. Synthesized cDNA was further used for amplification according to the protocol described below.
Quantitative real-time PCR
Quantitative PCR targeting nirS, nirK, nosZ and bacterial 16S rRNA gene was performed in a real-time PCR system (iQ5, Bio-Rad) using the previously described primer sets (Table 1). PCR reactions were run in triplicate with 4 ng of template DNA in a 25 μL reaction volume containing 12.5 μL of iQ SYBR green supermix (Bio-Rad), 2 μL (nosZ) or 1.25 μL (nirS, nirK) of each primer (10 μM), 1–3 μL of template and nuclease-free water (Promega). All reactions were performed in white 0.2 mL strips with ultraclear optical flat caps (Bio-Rad). For all primer sets, the thermal cycle was programmed to 5 min of precycling at 94 °C, eight cycles of 94 °C denaturation for 30 s, 63 °C annealing for 30 s, 72 °C extension for 30 s and 80 °C for 18 s, followed by 35 cycles where the annealing temperature was changed to 57 °C. PCR reaction mixtures with all reagents, except template DNA, served as the negative control. Standards consisted of cloned DNA fragments from sandy sediment samples of the Douro River estuary containing the target region of each primer set plus 0.5 μL of DNA pool from our environmental samples. The PCR product obtained was gel purified using the QIAquick gel extraction kit (Qiagen) and the amplicons were cloned using the TOPO TA cloning kit (Invitrogen) according to the manufacturer's instructions, with the exception of a slight modification (vector DNA and salt solution were decreased to 0.5 μL and chemically competent cells decreased to 25 μL). Plasmids were isolated using the GeneElute plasmid miniprep kit (Sigma) and the DNA concentrations of purified plasmids were determined fluorometrically using the PicoGreen dsDNA quantitation kit (Molecular Probes, Invitrogen, Eugene, OR). Standard curves were generated in duplicate for each primer set and amplification of standards was linear over six orders of magnitude, using different concentrations of plasmid DNA (from 0.2 to 0.2 × 10−6 ng DNA); PCR reactions with standards and environmental samples were run together in the same PCR block. The R2 values between plasmid DNA copy numbers and the calculated threshold cycle value across the specified concentration ranged between 0.98 and 1.00 and amplification efficiency ranged between 95% and 101% for all the standard curves generated. Target copy numbers for each reaction were calculated from the standard curves, assuming that the average molecular mass of a double-stranded DNA molecule is 618 g mol−1; the data are presented in gene abundance per cm3 of sediment, based on the DNA extraction efficiency calculated above. Melting curves and agarose gels of the qPCR products were run following each qPCR assayed to confirm that the fluorescence signal originated from the specific PCR products. Also, to confirm the specificity of the qPCR assays, a qPCR product generated by each primer sets was cloned as described above; four colonies were selected randomly from the library of each primer set, insert size verified by digestion with EcoRI and sequenced. These sequences were compared with reference sequences from GenBank using the basic local alignment search tool (blast; Altschul et al., 1990) and demonstrated the specificity of the amplicons generated by qPCR with all primer sets.
Table 1. Oligonucleotide probes used in this study
The DGGE technique (Myers et al., 1985) was performed using a CBS Scientific DGGE system (Del Mar, CA) essentially following Magalhães et al. (2007b). PCR and RT-PCR targeting nirS, nirK, nosZ and PCR targeting bacterial 16S rRNA gene were performed in a regular PCR system (Eppendorf, mastercycler gradient), using the same protocols and primer sets described for qPCR (Table 1), although a 40-bp GC clamp was added to the 5′-end of each forward primer (Table 1). PCR product (700 ng) from each replicate sample was loaded on a 6.5% polyacrylamide gel containing a gradient of denaturant (urea and formamide) from 45% to 65% for the bacterial 16S rRNA gene and 40–80% for DNA and cDNA nirS, nirK and nosZ. PCR reactions containing genomic DNA from Clostridium perfringens and Bacillus thuringiensis (Sigma) were used as DGGE standards. Gels were run for 14 h at a constant voltage of 100 V (3.75 V cm−1) in 1 × TAE buffer at 60 °C. Gels were stained for 30 min with 1 × SYBR Green. The DGGE gel images were converted to a densitometry scan and aligned using image analysis software (gelcompar II version 5.1, Applied Maths). The presence or absence of DGGE bands in each sample provided the input variables to evaluate the differences in the diversity of nirS, nirK, nosZ and bacterial 16S rRNA genes and transcripts by hierarchical cluster analysis based on Bray–Curtis similarities (Clarke & Warwick, 1994).
Total cell counts
For the total count of microbial cells, 1 g of homogenized sediment sample was added to 2.5 mL of saline solution (0.2 μm filtered, 9 g L−1 NaCl, 200 μL of 0.2 μm filtered, 12.5% v/v Tween 80) and fixed with 100 μL of formaldehyde (0.2 μm filtered, 4% v/v). The slurries were stirred at 150 r.p.m. for 15 min, followed by sonication for 20–30 s at a low intensity (0.5 cycle, 20% amplitude). Subsamples of the slurries were then stained with DAPI and incubated in the dark for 12 min (Porter & Feig, 1980). Samples were filtered onto black Nuclepore polycarbonate filters (0.2-μm pore size, 25 mm diameter, Whatman, UK) under gentle vacuum and washed with autoclaved 0.2-μm-filtered distilled water. Membranes were set up in glass slides and cells counted at 1875X on an epifluorescence microscope (Labophot, Nikon, Japan).
Effect of Cu on the denitrification pathway
The results revealed that denitrification within the intertidal sandy sediments of the Douro River estuary was highly sensitive to the progressive addition of Cu (Fig. 1a). Dinitrogen effluxes (measured as a difference between N2O production with and without C2H2) were found to be progressively lower (anova, P<0.05) with an increase of the Cu concentration (Fig. 1a). The addition of the lowest Cu concentration (4 μg Cu g−1 wet wt sediment) inhibited N2 production by 14.7% (Fig. 1a) compared with the control (treatment without the addition of Cu). Further Cu additions showed progressively higher percentages of inhibition of the production of the end product of the denitrification process (38% and 77%, respectively, for 8 and 60 μg g−1 wet wt sediment), with the complete inhibition of N2 release at the highest concentration tested (159 μg Cu g−1 wet wt sediment) (Fig. 1a). In nature, nitrous oxide generally accumulated in lower amounts as it was observed in the control (Fig. 1b). However, a significantly higher N2O accumulation (anova, P<0.01) was observed at the lowest Cu-amended experiment (4 μg Cu g−1 wet wt sediment; Fig. 1b). Our results showed that the enzymes that catalyze the first two steps of the denitrification pathway (reduction of NO3− to NO2− and the reduction of NO2− and N2O) had different sensitivities to Cu (Fig. 2a and b). Nitrate uptake rates were stimulated in the lowest Cu treatment (4 μg Cu g−1 wet wt sediment), because significantly higher rates of NO3− uptake were registered relative to the control (anova, P<0.05). On the other hand, the net NO3− uptake clearly decreased (anova, P<0.05) at the highest concentration of Cu tested (159 μg Cu g−1 wet wt sediment), suggesting a reduction of nitrate reductase activity, probably caused by the general Cu toxicity of bacterial community (Fig. 3). However, the activity of the enzyme NO2− reductase decreased at the 8 μg Cu g−1 wet wt sediment experiment, because the results showed significantly lower NO2− consumption for this concentration compared with the controls (anova, P<0.05). In addition, NO2− reductase was strongly inhibited at Cu concentrations between 8 and 60 μg Cu g−1 wet wt sediment, while NO3− reductase was still in operation. The modest NO2− accumulation for the highest Cu concentration tested coincided with the highest inhibition of NO3− reductase. Also, the significantly lower N2O production rates (anova, P<0.05) registered for 8, 60 and 159 μg Cu g−1 wet wt sediment when compared with the 4 μg Cu g−1 wet wt sediment treatment (Fig. 1b) are presumably due to the inhibition of the previous reduction steps in the denitrification pathway (NO3− or NO2− reduction; Fig. 2a and b).
Effect of Cu on general bacterial abundance and diversity
Bacterial quantification by real-time PCR based on 16S rRNA gene abundance agreed well with the total counts of DAPI-stained cells (Fig. 3). Although higher values were registered for the 16S rRNA gene abundance, probably due to the presence of multiple copies of the 16S rRNA genes that characterize several bacterial species (Fig. 3a). A significant, positive correlation was observed between bacterial 16S rRNA gene abundance and DAPI cell counts (R2=0.99, P<0.001, n=15). The results also revealed no significant change in the total bacteria counts among the control and the first two Cu concentration treatments (4 and 8 μg Cu g−1 wet wt sediment). However, bacterial numbers (anova, P<0.05) increased five times in the 60 μg Cu g−1 wet wt sediment treatment, compared with the control and lower Cu concentrations (Fig. 3). On the other hand, the higher Cu concentration tested (159 μg Cu g−1 wet wt sediment) induced a decrease in bacterial numbers compared with the control (anova, P<0.05), suggestive of a toxicity effect (Fig. 3). In agreement, DGGE analysis of total bacteria showed considerable similarity between the bacterial community structure in the first three treatments (0, 4 and 8 μg Cu g−1 wet wt sediment), but a clear shift in the bacterial community structure in experiments amended with 60 and 159 μg Cu g−1 wet wt sediment (Fig. 4). Cluster patterns of the hierarchical cluster analysis based on general bacterial DGGE profiles showed that samples amended with 0, 4 and 8 μg Cu g−1 wet wt sediment formed a closer cluster than communities in the last two treatments (anosim, R=0.95, P<0.01; Fig. 4).
Effect of Cu on nirS, nirK and nosZ abundance and transcription diversity
PCR–DGGE analyses of nirS, nirK and nosZ were performed in order to evaluate shifts in the diversity of the genes that catalyze different steps of the denitrification process in experiments amended with the progressive addition of Cu. Except for slight shifts, the DGGE profiles revealed no clear decrease in the number of bands as the Cu concentrations increased (see the example for nirK PCR–DGGE profiles, Fig. 5). Interestingly, quantitative analysis of nirS, nirK and nosZ showed pronounced decreases of these genes for all experiments amended with different concentrations of Cu (Fig. 6). Thus, while no evident pattern of denitrifier community diversity was found from DGGE analysis, quantitative data clearly showed a strong effect of Cu on the denitrifying bacteria, even at the lower concentration of 4 μg Cu g−1 wet wt sediment.
In order to evaluate the effect of progressive Cu concentrations on the diversity of nosZ, nirS and nirK transcripts, all RT-PCR products were subsequently resolved by DGGE. The DGGE profiles showed a clear progressive decrease in the diversity of the expressions of these genes, because the number of bands decreased with the progressive increase of Cu for all genes (Fig. 7). Hierarchical cluster analysis of RT-PCR–DGGE profiles based on the presence/absence of bands showed greater similarity between the control and the treatment with 4 μg Cu g−1 wet wt sediment than samples amended with the two highest concentrations (Fig. 7). These results highlight the pronounced inhibition by Cu on the diversity of the transcription of the nitrite and nitrous oxide reductase genes, which is in agreement with the results obtained for the activity measurements carried out concomitantly (Figs 1 and 2).
The activity of denitrifying microorganisms in aquatic ecosystem sediments can considerably improve water quality through a significant net removal of fixed N from the environment (e.g. Seitzinger et al., 2006). In particular, estuarine denitrifiers play an important buffering role in the transport of nitrogen from agricultural and other terrestrial sources into the oceans. However, estuarine sediments also sequester trace metals among other pollutants (Williams et al., 1994; Almeida et al., 2004). Among the most important trace metals, Cu in very small amounts (around 6 pmol L−1, Granger & Ward, 2003) is a necessary micronutrient metal to support life. A Cu limitation can decrease both the growth rates of denitrifying bacteria and the denitrification activity (Granger & Ward, 2003). Indeed, this metal has an extensive involvement in the denitrification pathway because one type of nitrite reductase and the nitrous oxide reductase contains Cu at its reaction center (Zumft, 1997). However, our results demonstrate that the denitrification rates can be inhibited by Cu concentrations between 4 and 159 μg Cu g−1 wet wt sediment above the environmental levels (<4.5 μg Cu g−1 wet wt sediment) (Fig. 1a). The decline in the denitrification rate with increasing concentrations of Cu has also been observed previously (Bardgett et al., 1994; Sakadevan et al., 1999; Holtan-Hartwig et al., 2002), although the Cu concentrations tested were generally higher. For example, Sakadevan et al. (1999) found that denitrification inhibition was achieved only when wetland sediments were amended with 500–1000 μg Cu g−1 wt sediment and additions of 100 μg Cu g−1 wt sediment of sediment stimulated denitrification. Bardgett et al. (1994) found denitrification inhibition along a gradient between 50 and 1300 μg Cu g−1 of dry pasture soil, and according to Holtan-Hartwig et al. (2002), 100 μg Cu g−1 of dry soil amended with ground straw was sufficient to inhibit the nitrous oxide reductase step. In the present study, much lower Cu concentrations inhibited the N2O reduction step in denitrification (4 μg Cu g−1 wet wt sediment; Fig. 1b). This suggests that N2O reductase was either sensitive to the lowest Cu concentrations tested or that Cu caused a shift in the community structure by selecting denitrifiers lacking nosZ. In addition, concentrations of 8–60 μg Cu g−1 wet wt sediment were sufficient to inhibit NO3− and NO2− reduction (Fig. 2). Development of tolerance to heavy metals by microbial communities is commonly observed (Bååth et al., 1998; Holtan-Hartwig et al., 2002); depending on the type of soil/sediment, greater or lesser levels of metal immobilization can occur (Sundelin & Eriksson, 2001; van Griethuysen et al., 2003). Thus, the differences observed in the denitrification tolerance to Cu between different studies could be related to the different levels of microbial tolerance to Cu and/or to differences in metal availability due to the distinct soil matrix interferences.
In this study, we also evaluated whether exposure to a gradient of Cu concentrations could modify denitrifier community structure/diversity. In our first approach, the microbial DNA was targeted; the results for the different genes evaluated (nirS, nirK and nosZ) revealed no clear differentiation of DGGE profiles for the different treatments (Fig. 5). Unfortunately, this method provided no insight into the viability of the organisms, but only into the diversity of the genes preserved in the different treatments, a limitation for this approach. On the other hand, the quantitative analysis performed targeting nirS, nirK and nosZ reflected an acute effect of Cu on the abundance of the microbial communities that mediate the nitrite and nitrous oxide reduction steps (Fig. 6). These findings showed that a high percentage (between 79% and 81%) of denitrifying bacteria in our sediments were particularly sensitive to low Cu concentrations (4 μg Cu g−1 wet wt sediment, plus the natural Cu concentration <4.5 μg Cu g−1 dry wt sediment); the surviving cells represented only around 20% of the total microbial community containing nirS, nirK and nosZ.
The analysis of nirS, nirK and nosZ transcription diversity explained the decrease in the different reduction steps in denitrification activity (Fig. 7); the detrimental effect of Cu occurred through a drastic reduction of denitrifier cells and the transcript diversity of the different genes involved. This subtractive effect by metals on microbial community composition has been identified in other studies (Müller et al., 2001; Holtan-Hartwig et al., 2002; Kourtev et al., 2009). In our study, because most of the denitrifier community were sensitive to Cu, their contribution to the community was lost, and the ability to denitrify was affected even at the 4 μg Cu g−1 wet wt sediment concentration and was completely disrupted after the addition of 60 μg Cu g−1 wet wt sediment. Interestingly, this deleterious effect was not evident when general microbial parameters were evaluated (total counts of prokaryotic cells and 16S rRNA gene abundance). Actually, treatments with Cu concentrations of 60 μg g−1 wet wt sediment of sediments highly stimulated total bacterial growth (Fig. 3), which might be explained by the function of Cu as a micronutrient (Alloway, 1995). However, not all of the bacterial community was affected in the same way. Specifically, in our study, we demonstrate that denitrifier communities were negatively affected in their activity, abundance and transcription of the genes involved on the denitrification pathway. These findings emphasize that the use of general bacterial indicators (e.g. biomass, total bacterial counts, respiration rates) to study the toxic effects of metals or other elements does not allow us to make conclusions about the entire bacterial community, because specific groups of bacteria populations can be affected in very different ways, as demonstrated here. While denitrifiers usually do not represent a high percentage bacterial population (Seitzinger et al., 2006), the biochemical process they perform has significant environmental importance.
Numerical sediment quality guidelines have been developed based on the biological effects associated with the metals described in several studies (NOAA, Long & Morgan, 1990; Long et al., 1990; O'Connor et al., 1998). The effects range-low (ERL) are indicative of concentrations below which adverse effects rarely occur and effects range-median (ERM) are representative of concentrations above which effects frequently occur (Long et al., 1990). These values are used worldwide as guidelines to evaluate the possible toxicological significance of metals or other chemicals in sediments. In the case of Cu, ERL and ERM values were reported to be 34 and 270 mg g−1 sediment (Long et al., 1990). It is, however, important to recognize that while these ERLs are not thresholds below which sediment toxicity is impossible (O'Connor et al., 1998), reduced rates of the denitrification pathway occurred in this study at Cu concentrations much lower than the ERL value (addition of 4 μg Cu g−1 wet wt sediment to samples with <4.5 μg Cu g−1 dry wt sediment). Thus, we believe that this study provides a valuable contribution towards understanding the real impact of metals in an essential biogeochemical process, with profound implications for the management of aquatic ecosystems to promote fixed nitrogen removal.
We thank W.J. Wiebe for editorial assistance. This study was funded by the Portuguese Science and Technology Foundation (FCT) through a PhD fellowship to A.M. (SFRH/BD/46146/2008), a PosDoc fellowship to C.M.M. (SFRH/BPD/14929/2004) and by FCT and COMPETE through a grant to C.M.M. (PTDC/AAC-AMB/113973/2009-FCOMP-01-0124-FEDER-013958).