Introduction

While metals are naturally present in the environment, high concentrations, like those found in areas surrounding mining and smelting operations, have been shown to disrupt many critical biogeochemical processes within soils, due in part to the negative effects they impose on soil microbial communities. Microbial communities are known to be important for a wide range of soil processes, including the cycling of nutrients, breakdown of organic matter [18], interaction with plant roots, and formation of soil structure [25]. The sensitivity of microbial communities to metal contamination has been studied previously in both field-contaminated and laboratory-spiked soils. High levels of soil metal contamination have been correlated with decreased microbial activity [30, 40], and biomass [18, 22], as well as impaired enzyme function and nutrient cycling [7, 19]. Similarly, metal contamination has been shown to impart selective pressure on soil microbial communities, leading to shifts in community structure [8] and decreases in biodiversity [22, 30]. Despite significant attempts to characterize the negative effects of metals on microbial communities, questions remain regarding the variation in soil microbial community responses to metal stress.

In 1934, Bass-Becking theorized that high microbial abundance and dispersal rates foster ubiquitous distribution in the environment [3]. This idea that “everything is everywhere and the environment selects” suggests that differences in soil microbial community composition is a result of the collection of stresses and/or substrates available in the immediate soil environment [35]. Building off of this idea, this study attempts to address the following question: Will microbial communities, indigenous to soils with unique local environmental conditions, converge to a common community when exposed to the same selective pressure (i.e., metal contamination)? To answer this question, we compared the effects of metal stress on the structure and function of bacterial communities indigenous to three different soils, each with a history of background, moderate, or high levels of long-term metal contamination, respectively. Based on previous work in this area, we hypothesize that metal amendment will lead to shifts in both bacterial community structure and function. More specifically, we hypothesize that metal-associated selection pressure will cause the bacterial communities from each soil to converge to a common community, exhibiting similar structural and functional diversity. Finally, we hypothesize that the shifts in microbial community structure and function will be a direct effect of metal stress, rather than an artifact caused by acidification of soils from metal-salt amendments. Counter-ion and pH controls were included for each treatment to aid in delineation of metal, counter-ion, and pH effects. From this study, a better understanding of bacterial community responses to metal-associated selection pressure will be achieved.

Methods

Site Description and Sampling Procedure

Aerial and fluvial deposition of metals, following a century of mining and smelting activity, has resulted in widespread contamination of the soil surrounding the Anaconda Smelter site in Anaconda, Montana. Contaminants of concern at this site include copper (Cu), zinc (Zn), lead (Pb), cadmium (Cd), and the metalloid arsenic (As). Previous work at this site placed soils within a 24-km radius of the stack in the path of aerial metal deposition [36]. Two sites within 24 km of the smelter stack (a severely impacted smelter site and a moderately impacted smelter site [(N 46.1814, W 112.7633) and (N 46.09752, W 112.09207), respectively]) were selected based on the degree of total soil metal contamination [37]. Bacterial communities within each soil are assumed to have been exposed to either chronically high or moderate levels of metals, respectively. The vegetative type and cover within the two smelter-impacted sites has been characterized previously (percent vegetative cover, percent cover by grasses, forbs, live woody, dead woody, bare ground, gravel, and rocks) in 2000 using 1-m2 Daubenmire quadrants. A third site, located 30 miles from the Anaconda Smelter site (N 45.75610, W 112.7639) was selected, and the bacterial community from this site represents a nonsmelter impacted community. The nonimpacted site was chosen based on the similarity of vegetative cover and the background concentration of metals in the soil.

In August of 2005, soil from the severely and moderately impacted smelter sites and the nonimpacted site were sampled to a depth of 15 cm. The soil from each site was sieved in the field to remove the >2 mm size fraction, homogenized and stored at 4°C for 6 months prior to commencement of the laboratory dosing experiment.

Experimental Design

A plastic-lined cement mixer was used to homogenize 24 kg each of nonimpacted, moderately impacted and severely impacted soil, respectively. Metal-salt solutions [H2NaAs04, Pb(NO3)2, Cu(SO4), Cd(SO4), and Zn(SO4) dissolved in ultrapure H20(>18 MΩ cm)] were added to approximately 2,500 g of the homogenized soil from each site, to create a total of six metal-salt amended treatments (RL, RH, R2H, LH, L2H, and H2H, Table 1). Metal-salt solutions were mixed into soil treatments for 30 min with an electric five-speed food processor to ensure even distribution and uniform concentration. The stainless-steel mixer attachments and the plastic-lined cement mixer were rinsed for 30 min with 10% HNO3 between treatments to prevent cross-contamination of soil metals. Unamended control treatments for nonimpacted, moderately impacted, and severely impacted soil (Treatments R, L, and H, respectively, Table 1) received equivalent amounts of ultrapure H20 and were mixed for 30 min with an electric five-speed food mixer.

Table 1 Nominal metal concentrations and corresponding day 7 soil pH (mean (SE)) of the nonimpacted reference site (Reference), moderately impacted (Low), and severely impacted (High) smelter site treatments, following amendment of soils
Full size table

Results from a pilot study demonstrated that the addition of metal-salt solutions significantly decreased the pH of the amended soils. Soil pH in treatments receiving the highest concentrations of soil metals decreased 2 pH units and remained decreased relative to the unamended control treatments over time (data not shown). Because the effects of pH on soil microbial community structure and function were found to be confounding factors in the analysis and interpretation of results in previous studies [12, 16, 33, 43], six additional treatments (RLC, RHC, R2HC, LHC, L2HC, and H2HC, Table 1) were designed to control for the effects of increased concentrations of counter-ion or changes in soil pH. Counter-ions (added as NaNO3 and NaSO4) were mixed into 2,500 g of soil from each site, as described above. Additionally, 1 M HNO3 was mixed into control soils until the control treatment soil pH was equal to that of its corresponding metal-amended treatment (where treatments RLC, RHC, R2HC, LHC, L2HC, and H2HC correspond to metal-amended treatments (RL, RH, R2H, LH, L2H, and H2H, respectively).

Once each of the 15 metal-salt amended and control treatments were thoroughly mixed, 140 g of soil from each treatment were used to seed individual, sacrificial “conetainers” (Stuewe & Son, Corvalis, OR, USA). Each conetainer was modified to include a filter paper plug at the base and a polyfilm seal (perforated to allow air exchange) at the top, to prevent loss of soil through the base and excessive loss of soil moisture from evaporation. A total of 18 identical conetainers were created for each of the respective treatments and were stored at room temperature. Soil moisture was monitored in conetainers in each treatment over time, and ultrapure H2O was added each week to maintain the soils between 12% and 20% moisture (based on dry weight).

Sampling and Analysis

On sampling days 7, 21, and 63 (after amendment of soils with metal salts), five replicate conetainers from each treatment were sacrificed. A 40-g subsample of soil from each conetainer was removed and stored at 4°C for up to 1 week prior to microbial analysis. Remaining soil from replicate containers in each treatment was composited and used to characterize the soil physiochemical, nutrient, and metal profiles. Additionally, soil physiochemical, nutrient, and metal profiles were preformed on day 0, using composited soil from each treatment.

A subset of composited soil from each treatment was oven-dried at 105°C for 48 h to determine soil oven-dry weight equivalent (ODE) and the percent moisture. Soil nutrient [nitrate (NO3), phosphorus (P), and percent base saturation of the elements (potassium (K), magnesium (Mg), calcium (Ca)], and physical [pH, % organic matter, cation exchange capacity (CEC), particle size distribution] profiles were analyzed by Waters Agricultural Laboratories, Owensboro, KY, USA. Total extractable metals were digested following EPA method 3050B. Soluble soil metals were extracted by shaking soil in a 1:10 (w/v) 0.01 M CaCl2 solution for 16 h, as described previously [17]. Total extractable and soluble metals (Zn, Pb, As, Cd, and Cu) were analyzed by inductively coupled plasma atomic emission spectroscopy (Texas Tech University, GeoAnalytical Laboratory).

Soil Bacterial Analysis

Biolog

Biolog GN microtiter plates (BIOLOG, Hayward, CA, USA) were used to assess bacterial activity and carbon substrate utilization patterns (SUPs), as described previously [14, 45, 48]. Briefly, Biolog plates were inoculated using 10 g ODE of soil from each sacrificed conetainer (n = five per treatment, per timepoint). Soil was blended with 0.2% water agar using an electric food processor and then serially diluted using sterile water. The 96-well plates were inoculated and incubated at 25°C for 72 h. Absorbance was analyzed every 12 h at a wavelength of 590 nm. Bacterial activity represents the total amounts of carbon substrate utilized on each Biolog plate (i.e., the sum of the total absorbance). Carbon SUPs represent the identities of substrates utilized on each Biolog plate.

Biomass

A modified chloroform-fumigation-extraction method was used to determine the effects of metal contamination of soil microbial biomass, as described previously [26]. Biomass was measured in duplicate, and the average microbial biomass in each sacrificed container was used for statistical comparison (n = 5 per treatment, per timepoint).

DGGE

UltraClean Soil DNA Extraction kits, were used to extract genomic DNA from each sacrificed conetainer (n = 5 per treatment, per timepoint), following manufacturer’s instructions (MO BIO Labs, Carlsbad, CA, USA). DNA extracted from 1 g soil was polymerase chain reaction (PCR)-amplified using the bacterial primer 341 f with a GC-clamp (5′-CGC CCG CCG CGC GCG GCG GGC GGG GCG GGG GCA CGG GGG GCC TAC GGG AGG CAG CAG-3′) and the universal primer 519r (5′-ATT ACC GCG GCT GCT GG-3′) (Integrated DNA Technologies, Coralville, IA, USA). Extracted DNA (1 µl) was PCR-amplified as follows: each 25 µl reaction contained: 20 pmol of each primer, 0.4 mM of each dNTP, 1× buffer (containing 2 mM MgCl2), 1.5 U Taq (Takara Bio, Japan), and sterile water. Thermocycler conditions: initial denaturation at 94°C for 2 min; followed by 35 cycles of denaturation, annealing, and extension (at 98°C for 10 s, 54°C for 40 s, 72°C for 1 min, respectively); followed by a final extension at 72°C for 10 min. PCR products were stored at −20°C prior to DGGE analysis.

The effect of metal contamination on species richness and bacterial community structure was assessed via denaturant gradient gel electrophoresis (DGGE) of 16 S rDNA. Previously established protocols [24, 29] were modified and optimized for use on the BioRad DCode DGGE system as follows: an 8% w/v polyacrylamide gel containing a denaturant gradient ranging from 35% to 36% [where 100% denaturant contains 40% formamide (Sigma) and 7 M Urea (BioRad Laboratories, Richmond, CA, USA) was used to separate PCR-amplified 16 S rDNA. Amplicons were initially pulsed through the wells at 90 V for 15 min and then ran at 60 V for 16 h at a constant temperature of 60°C. A standard marker containing a mixture of Pseudomonas aeruginosa, Shewanella putefaciens, Sphingomonas sp., Ralstonia sp., Desulfovibrio sp., was run in the beginning, middle, and end lanes of each gel. After being stained in ethidium bromide, polyacrylamide gel were digitally photographed under UV light (Kodak Imaging) and analyzed using GelComparII software.

Data Analysis

Differences in the soil physiochemical profile and the soil metal profile between the moderately and severely impacted smelter sites (L and H, respectively) were analyzed by analysis of variance and compared to the nonimpacted site (R) using Dunnett’s multiple comparison test. Bacterial activity, a measure of the amounts of substrates utilized on Biolog plates, was calculated based on raw difference data (i.e., the absorbance registered in the control well subtracted from the absorbance registered in each of the 95 substrate-containing wells [14]). Carbon substrate utilization patterns (SUPs) were generated with respect to the identities of substrates utilized.

Gel CompareII professional software (Applied Maths, Austin, TX, USA) was used to analyze PCR-DGGE banding profiles. All gels were normalized using identical standards run in multiple lanes on each gel. Bands within each lane were manually identified and quantified using a best-fit Gaussian curve. All gel images were compiled, and band lanes were assigned (optimization was adjusted to 0.50 and position tolerance was adjusted to 1.00) [Heather Christensen, Applied Maths, personal communication]. Quantitative band tables, representing both the presence/absence and intensity of bands in each lane, were generated.

Multivariate statistics were used to analyze Biolog SUPs and community DGGE banding profiles as follows: (1) To test the hypothesis that metal-amendment would cause shifts in bacterial community structure and function, canonical correspondence analysis (CCA) was used to explain the differences in bacterial community SUPs and DGGE banding profiles, with respect to soil pH and/or total extractable soil metal concentration (i.e., direct gradient analysis). Ellipses, representing the standard error of the multivariate means, were used to characterize the differences among treatments. Eigenvalues, which estimate the percent of variation in the data that is accounted for by environmental variables, is denoted by subscripts. (2) To test the hypothesis that unique bacterial communities would converge on a common community following metal-amendment, the SUPs and DGGE banding profiles supported by bacterial communities in the 2*High treatments (R2H, L2H, and H2H) were compared by correspondence analysis (CA). First, bacterial SUPs and DGGE banding profiles on sample days (7, 21, and 63) were analyzed independently, and differences within treatments on each sample day were compared. Next, SUPs and DGGE banding profiles that were analyzed from each sample day were analyzed jointly (i.e., all three treatments and all three sample days analyzed on the same CA plot) to visualize convergence over time. All statistical tests were conducted using R [32].

Results

The nonimpacted site (R), the moderately impacted smelter site (L), and the severely impacted smelter site (H) are significantly different in terms of their soil physiochemical profiles (Table 2). The soils from each site are classified as sandy loam and share similar soil texture; however, the soil physiochemical profiles among the sites are variable. Notably, soil pH, cation exchange capacity (CEC), and organic matter levels were highest in the severely contaminated soil and lowest in the moderately contaminated soil. The nonimpacted site, moderately impacted smelter site, and the severely impacted smelter site were also significantly different in terms of their total metal profiles (Table 3 and 4). Metal concentrations in severely impacted site were significantly elevated relative to the metal concentrations in the moderately and nonimpacted sites.

Table 2 Soil physiochemical parameters [mean (SE)] of the nonimpacted reference site (R) and the moderately impacted (L), and the severely impacted (H) sites (n = 6 per site)
Full size table
Table 3 Total extractable metal concentrations (mean (SE))
Full size table
Table 4 CaCl2 extractable metal concentrations (mean (SE))
Full size table

Soils from the nonimpacted, moderately impacted, and severely impacted sites were amended with metal salts and counter-ions to create a total of 15 treatments. The total extractable metal concentrations corresponded with the nominal concentrations in each treatment, with few exceptions (Table 3). Compared to total extractable metals, the CaCl2 extractable metal concentrations were significantly lower in all treatments (Table 4). Metal concentrations in pH-control treatments (RLC, RHC, R2HC, LHC, L2HC, and H2HC) were not significantly different in terms of total extractable metal concentrations, relative to their respective unamended treatments (R, L, and H, respectively) (data not shown). On the other hand, several pH-control treatments (LHC, L2HC, and H2HC) had increased levels of CaCl2 extractable metal concentrations relative to their respective unamended treatments (L and H), likely due to soil acidification.

The level of activity supported by bacterial communities indigenous to the nonimpacted (R) and moderately impacted soil (L) was elevated compared to the activity supported by bacterial communities native to the severely impacted soil (H) (Fig. 1). As seen by the decrease in bacterial activity supported by microbial communities following metal-amended, high concentrations of soil metals negatively affected bacterial activity in all three soils. For example, levels of activity were significantly decreased in R2H, LH, L2H, and H2H, relative to their respective unamended control treatments (R, L, and H). Because the level of activity supported within pH control treatments (RLC, RHC, R2HC, LHC, and H2HC) remained similar to the level of activity supported by bacterial communities, the unamended treatments (R, L, and H, respectively), the observed decrease in activity is likely a result of metal-associated selection pressure rather than due to changes in soil pH. The exception to this trend appears to be treatment L2HC, which supported decreased activity as a result of soil acidification. Notably, the level of bacterial activity in LH, L2H, and H2H metal-amended treatments remained depressed throughout the course of the experiment; whereas bacterial activity in the R2H treatment returned to preexposure levels by day 63.

Figure 1
figure 1

Total microbial activity, relative to the amounts of substrates utilized on Biolog plates, supported by bacterial communities within each treatment, following amendment. (Refer to Table 1 for description of treatment amendments). The black, dark grey, and light grey bars represent the total activity on sample days 7, 21, and 63, respectively. Error bars correspond to the mean ± SE for each treatment (n = 6)

Full size image

Microbial biomass was decreased in the smelter-impacted sites (L and H), relative to the nonimpacted site (R) (Fig. 2). As seen by the decrease in biomass in treatments following metal-amendment (RL, RH, R2H, LH, L2H, and R2H), relative to the unamended treatments (R, L and H, respectively), high concentrations of soil metals negatively affected biomass in all three soils. Biomass was also significantly affected by soil pH, as was seen by the decreased biomass in treatments RLC, RHC, R2HC, and H2HC, relative to their unamended soil treatments (R and H, respectively). Bacteria in treatments LHC and L2HC did not support decreased biomass as a result of soil pH. In all soils, the decreased biomass observed as a result of metal-amendment was more pronounced than the decrease in biomass as a result of soil pH (i.e., compare RL to RLC, RH to RHC, R2H to R2HC, etc).

Figure 2
figure 2

Total microbial biomass supported by bacterial communities within each treatment, following amendment (Refer to Table 1 for description of treatment amendments). The black, dark grey, and light grey bars represent the total activity on sample days 7, 21, and 63, respectively. Error bars correspond to the mean ± standard error for each treatment (n = 6)

Full size image

The bacterial communities indigenous to the nonimpacted (R) and smelter-impacted (L and H) sites were unique in terms of their structural diversity, estimated by canonical correspondence analysis (CCA) of community DGGE-banding patterns (Fig. 3a) and functional diversity, estimated by CCA of SUPs (Fig. 3b). In these soils, bacterial community structure and function were highly correlated with soil metal concentration, where metal concentration accounted for 51% and 37.6% of the total variation in the data sets, respectively. Amendment of the soils with increasing concentrations of metals caused significant shifts in both bacterial community structure and function (Fig. 4). Acidification of the soils also was observed to cause shifts in both structural and functional diversity (Fig. 4). For instance, as seen by their proximity in two-dimensional space, on day 63, the structural diversity of the microbial community within the metal-amended treatments (RL, RH, and R2H) and the pH control treatments (RLC, RHC, and R2HC) were significantly shifted relative to the unamended control treatment (R) (Fig. 4, Panel 1). Soil pH and soil metal concentration accounted for 45.5% of the total variation observed, indicating that the structural diversity of microbial communities within the nonimpacted smelter site is sensitive to both metal stress and changes in soil pH. However, the structural diversity of the microbial communities within the metal-amended treatments (RL, RH, and R2H) were also significantly shifted relative to their respective pH-control treatments (RLC, RHC, and R2HC). Soil metal concentrations alone accounted for 37.5% of the total variation, indicating that metal contamination had a direct effect on soil microbial community structure, above and beyond that of pH only. The carbon use profile of the microbial community in the nonimpacted reference site was also observed to be sensitive to changes in soil pH and metal contamination, where metal contamination and soil pH accounted for 25.1% and 9.7% of the total variation, respectively (Fig. 4, Panel 2). These effects were most pronounced in the most contaminated and acidified treatments (R H, R 2H, and R 2HC). Treatment RLC was not significantly different than the control treatment, as seen by their overlapping ellipses.

Figure 3
figure 3

Direct gradient analysis of bacterial communities native to the nonimpacted reference site (R), moderately impacted smelter site (L), and severely impacted smelter site (H) on sampling day 63. Variation in community DGGE banding profiles (a) and substrate utilization profiles (b) are projected onto two axes via canonical correspondence analysis (CCA). Ellipses represent the SE of the means of each treatment. Arrows represent significant correlations of total extractable metal concentration (Cd, Cu, As, Pb, and Zn) with ordination axes and is denoted (M). Eigenvalues, which estimate the percent of variation in the data that is accounted for by metal concentration, is denoted by subscripts

Full size image
Figure 4
figure 4

Direct gradient analysis of the shifts in structural and functional diversity of bacterial communities in each soil, following amendment (Refer to Table 1). Variation in community DGGE banding profiles (Panel 1) and variation in substrate utilization profiles (Panel 2) are projected onto two axes via canonical correspondence analysis (CCA). Ellipses represent the standard error of the means of each treatment. Arrows represent significant correlations of soil pH and total extractable metal concentration (Cd, Cu, As, Pb, and Zn) with ordination axes and is denoted (pH or M, respectively). Eigenvalues, which estimate the percent of variation in the data that is accounted for by the environmental variables, is denoted by subscripts

Full size image

Similarly, the structural and functional diversity of the bacterial communities indigenous to the smelter-impacted soils followed the trends previously observed in the nonimpacted soil (Fig. 4, panels 1 and 2, respectively). Bacterial communities from these soils were sensitive to both decreased soil pH and increased metal stress; however, as shown by the percent of the total variation accounted for by each environmental variable, the effects of soil metal concentrations superseded the effects of soil pH in both soils. For example, metal concentration accounted for 40.6% and 19.0% of the total variation observed in DGGE profiles and SUPs of microbial communities from the moderately impacted soil (LH and L2H), respectively, whereas soil pH only accounted for 8.5% and 6.1% of the total variation in this soil. Similarly, metal concentration accounted for 59.4% and 36.8% of the total variation observed in DGGE profiles and SUPs of microbial communities from the severely impacted soil (H2H), respectively, whereas soil pH only accounted for 18.6% and 12.6% of the total variation in this soil.

To test the hypothesis that the extreme levels of soil metal contamination would select for structurally and functionally similar microbial communities in each soil, DGGE banding profiles and SUPs of microbial communities within the most highly metal-contaminated treatments (R2H, L2H, and H2H) were compared using CA. The SUPs and DGGE banding profiles within each community were generated following 7, 21, and 63 days of incubation. SUPs and DGGE banding profiles from each sample day were analyzed independently and jointly to determine if the three communities converged during the course of the incubation. As seen by their proximity in two-dimensional space, when the three sample days are analyzed independently, convergence of treatments R2H, L2H, and H2H is not observed (Fig. 5). Bacterial communities from the three treatments were structurally distinct on day 7 and did not converge to a structural community by day 63 (Fig. 5, panel 1), and, bacterial communities from each site failed to converge on a functionally similar community by day 63, despite evidence of overlapping carbon use profiles on day 7 (Fig. 5, panel 2). Likewise, when all three treatments and all three sample days were analyzed jointly on the same CA plot, convergence of community DGGE and SUP profile was not observed (data not shown).

Figure 5
figure 5

Direct gradient analysis of the shifts in structural and functional diversity of bacterial communities native to each site following amendment with the highest nominal metal concentration (2*High). Variation in community DGGE banding profiles (Panel 1) and variation in substrate utilization profiles (Panel 2) are projected onto two axes via correspondence analysis (CA). Ellipses represent the standard error of the means of each treatment

Full size image

Discussion

Through the selection of tolerant species and selection against sensitive species, soil metals have the capacity to shape the structural and functional diversity of microbial communities in soils [18]. The strong selective pressure of metal contamination is well documented, and soils exposed to long-term metal contamination have been shown to support metal-tolerant communities. For instance, microbial communities chronically exposed to a gradient of Zn concentrations in field-contaminated soil, were found to have increased tolerance to Zn, as measured via the pollution-induced community tolerance method (PICT) [6, 38]. Likewise, microbial communities exposed to long-term metal-contaminated sewage sludge exhibited an increase in community tolerance to metals [46], and forest soils exposed for centuries to naturally elevated soil lead concentrations were observed to support an abundance of lead-tolerant species [1]. Due to the diversity, ubiquity, and heterogeneity of bacterial species in soils [34], it is likely that microbial communities native to pristine soil environments support species inherently tolerant to metal insults. For example, lead-tolerant species were isolated from previously uncontaminated forest soils [1], and metal-tolerant communities were established following the artificial contamination of pristine soils with Zn, Cd, Cu, and Ni [9].

Still, questions remain regarding the capacity for metals to shape microbial communities in soils and the uniformity of microbial community responses to metal stress across soil types. In this study, three soils in the vicinity of a copper smelter, previously exposed to background, low or high concentrations of metals, were artificially contaminated with metal-salts, and the shifts in the structural and functional diversity of each community as a result of metal stress was compared. Based on the assumption that microbial communities are shaped by the selective forces in their local environments, we question whether metal stress will cause initially distinct microbial communities to shift toward structurally and functionally similar communities. The idea of microbial communities with unique structural profiles converging to a similar community profile, as a result of strong selection pressure, has been addressed previously. While in one study, indigenous groundwater microbial communities were seen to converge on a common community in bioreactors treating aromatic hydrocarbon-contaminated groundwater [23], microbial communities from different soils did not converge on a similar hydrocarbon-degrading community when exposed to diesel contamination [5], and unique soil communities did not converge on a similar community profile following fumigation with chloroform [10]. To our knowledge, this is the first study to characterize the convergence following metal stress.

In this study, the nonimpacted reference site supported a bacterial community with inherently higher levels of activity and biomass, relative to the smelter-impacted sites. These results are consistent with previously reported results, in which microbial communities from metal-contaminated sites were shown to support decreased levels of biomass relative to uncontaminated sites [12, 18]. Following amendment with metal-salts, bacterial communities from each site supported decreased levels of activity, and biomass, indicating that the communities were similarly sensitive to metal stress. However, while the levels of activity remained depressed in soil communities native to the smelter-impacted sites, the levels of activity supported by the bacterial community from the nonimpacted smelter site (R2H) returned to preexposure levels throughout the course of the study. Previously reported findings of pollution-induced community tolerance (PICT), suggest that the previously exposed communities (i.e., communities from the smelter-impacted sites) would be more tolerant to additional metal stress than previously unexposed communities [6]. The observed “resiliency” of the community in treatment R2H is likely related to PICT, where the initial high concentrations of metals killed off the majority of metal-sensitive species, which then served as substrate for the surviving community [9, 13].

In addition to differences in activity and biomass, the bacterial communities native to each site were significantly different in terms of their structural and functional diversity. Following amendment with the same nominal concentrations of metals-salts, the bacterial communities native to the nonimpacted, moderately impacted, and severely impacted smelter sites failed to converge on a functionally or structurally similar community. The structural and functional differences in the microbial communities from each site following similar levels of metal stress indicate that the cumulative effects of all environmental variables (i.e., soil pH, organic matter, metal bioavailability, and/or the initial composition of the community) may outweigh the effects of a single environmental stressor in governing bacterial community succession.

The Biolog technique has become widely used in literature, as a measure of bacterial community function, with respect to carbon substrate utilization [14, 21, 22, 45, 48]. As a culture-based technique, however, Biolog is inherently biased due to selective enrichment [2], which may ultimately serve to misrepresent the indigenous populations [6]. These selection biases have been addressed previously [20]. Additional bias in the Biolog technique, which may be relevant to this study, is related to pH. Biolog plates are buffered at a pH of 6.5, and highly acidic soils have been shown to affect Biolog results [47]. The soil pH in our treatments range from 8.11 (H) to 4.53 (L2H). While we made every effort to control for the confounding effects of soil pH on the bacterial communities in this study, we did not control for the potential bias associated with acidic soil pH in the Biolog technique. Therefore, results should be interpreted with caution.

Potential Biases Associated with Metal-Salt Amended Mesocosms

Soils artificially contaminated with metal salts (nitrate, sulfate, or chloride salts) have been used extensively in the past to characterize the dose–response relationships of various microbial endpoints, including activity [8, 9, 12, 33, 40], respiration [27, 33, 39, 40], composition (using PFLA) [12, 33], nitrification [27, 39, 40, 43], N-mineralization [39], biomass [12, 21, 44], and tolerance [8, 9] to metal stress. While artificially contaminated microcosm studies have notable advantages over field studies (i.e., they allow for control over the confounding soil parameters (e.g., soil organic matter, soil type, soil cation exchange capacity, soil moisture) and associated stresses (e.g., temperature fluctuations, precipitation, mixed pollution) that can vary between sites in natural systems [27], they introduce their own set of confounding variables [43]. Amendment of soils with metal salts is often accompanied by significant decreases in soil pH [27, 42, 43], which cannot only affect bacterial community structure, activity, biomass, and substrate utilization [25], but is also a driving factor controlling metal bioavailability [7], mobility [28, 40], speciation [16], and toxicity. Because the overall change in soil pH as a result of metal-salt hydrolysis is largely dependent on the concentration of metal-salt amendments, the nature of the soil (e.g., percent organic matter, cation exchange capacity, clay content) and the inherent buffering capacity of the soil [41], the degree to which metal-salt amendments lead to changes in soil pH may vary between studies. Several studies have noted the confounding effects of pH on the microbial endpoints tested in artificially spiked soils [12, 16, 33, 43]; however, a few studies have included pH controls, and the effects of soil acidification on microbial communities is rarely considered when interpreting results [41].

In this study, a series of pH-control treatments were used to account for the changes in soil pH as a result of metal-salt amendment. Decreases in soil pH were observed to decrease bacterial biomass and activity and lead to shifts in both microbial structural and functional diversity. As a general trend, bacterial community function, with respect to the identities of substrates utilized on Biolog plates, seemed to be less sensitive to decreases in soil pH, than community structure in the nonimpacted and moderately impacted site communities (as seen by the overlapping functional profiles of R and RLC and L and LHC, respectively), which may indicate that bacterial community functional diversity endpoints have a higher resilience to changes in pH than bacterial community structure endpoints in these soils. Additionally, the bacterial community native to the nonimpacted reference site appeared to be relatively robust in terms of its ability to tolerate changes in soil pH, whereas the moderately impacted and severely impacted soil communities appeared to be more sensitive.

In addition to artifacts caused by changes in soil pH, differences in the soil physiochemical profiles between the sites add further sources of bias, and the direct comparison of bacterial community responses between the three soil types may be hindered due to the differences in soil metal bioavailability. The CaCl2 extraction method has been shown to provide a good estimate of the bioavailable fraction of metals in soils [11, 17, 31]. In this study, metals in the severely impacted site appear to be less bioavailable than metals in the nonimpacted site or the moderately impacted site, as seen by the decreased CaCl2 extractable metal concentration in H2H compared to R2H or L2H. The reasons for this decreased bioavailability may be twofold: (1) Because soils were dosed with metal-salt solutions to bring their total metal concentration to a nominal level, the soil from the severely impacted site, which had significantly higher levels of aerially deposited metals, received less metal salt, relative to the moderately impacted site. Metals in contaminated field soils are less bioavailable than metals in artificially spiked due to weathering, aging of the soils, and complexation of metals within the soil matrix [4, 27, 39]. (2) The severely impacted site had significantly elevated soil organic matter, likely due to decreased mineralization of organic matter by soil microbes, as was seen previously in soils surrounding metal-contaminated smelter sites [15]. The moderately impacted site, which supported the lowest levels of organic matter, appears to have the greatest bioavailable fraction of metals. Bioavailability of metals may be controlled by levels of organic matter in the soils [12, 18].

Conclusions

While the use of artificially spiked soil mesocosms for characterizing effects of soil metals on bacterial communities has noted advantages over field studies, the potential biases associated with these studies need to be considered when interpreting metal effects. Controlling for changes in soil pH was useful in this study to delineate pH-associated effects from metal effects; however, differences in soil metal bioavailability and toxicity between sites remains a potential confounding factor. Nevertheless, this study confirms the variation in soil bacterial community responses to selection pressure across different soil types and highlights the complexity of bacterial community responses to metal stress. Additionally, although the bioavailable concentrations of soil metals varied among treatments from different soil types due to differences in bioavailability, the general trend suggests that bacterial communities did not converge on a common community. Extension of these studies using soils with similar levels of bioavailability may help determine the potential for bacterial communities to converge as a result of metal stress.