Environmental Toxicology and Chemistry, Vol. 33, No. 12, pp. 2724–2732, 2014 # 2014 SETAC Printed in the USA

INVERTEBRATE COMMUNITY RESPONSES TO A PARTICULATE- AND DISSOLVED-COPPER EXPOSURE IN MODEL FRESHWATER ECOSYSTEMS STEPHANIE GARDHAM,*yz ANTHONY A. CHARITON,z and GRANT C. HOSEx yDepartment of Environment and Geography, Macquarie University, New South Wales, Australia zCentre for Environmental Contaminants Research, Oceans and Atmosphere Flagship, CSIRO, New South Wales, Australia xDepartment of Biological Sciences, Macquarie University, New South Wales, Australia (Submitted 5 June 2014; Returned for Revision 9 July 2014; Accepted 18 August 2014) Abstract: Historical contamination has left a legacy of high copper concentrations in the sediments of freshwater ecosystems worldwide. Previous mesocosm studies have focused on dissolved-copper exposures in the overlying waters, which, because of altered exposure pathways, may not accurately predict the effects of copper exposure on invertebrate communities at historically contaminated sites. The present study assessed the effects of copper on the establishment of invertebrate communities within a large outdoor pond mesocosm facility containing environmentally relevant copper-spiked sediments. High particulate copper concentrations (>400 mg/kg dry wt) caused a pronounced effect on the benthic community richness, abundance, and structure in the mesocosms, but particulate copper concentrations below 100 mg/kg dry weight had no effect. Furthermore, there were no effects of copper on the invertebrate communities within the water column, even in the highest copper treatment. The response of the benthic community to copper was influenced by interspecific interactions, the stage of ecological succession, and interspecies variation in sensitivity to copper. The present study demonstrates the importance of using environmentally realistic exposure scenarios that provide both particulate and dissolved exposure pathways. It also emphasizes that risk assessments for aquatic ecosystems should consider the influence of interspecific interactions and interspecies variation in driving the biotic response to contamination. Environ Toxicol Chem 2014;33:2724–2732. # 2014 SETAC Keywords: Metal risk assessment

Mesocosm

Interspecific interactions

Interspecies variation

Benthic macroinvertebrates

Ecological

sediment [18]. Organisms can take up metals, like copper, through exposure to the dissolved phase (for example, facilitated diffusion through permeable surfaces via filter feeding) or the particulate phase (for example, active uptake by deposit feeding) [19,20]. Thus, the dominant exposure route for an individual organism depends on its feeding behavior [19]. In manipulative studies, it is important to establish environmentally relevant partitioning of copper between dissolved and particulate phases to create exposure pathways for organisms that are similar to those present in fieldcontaminated sites. However, the handful of lentic mesocosm studies that have considered the effects of copper on invertebrate communities have not demonstrated copper partitioning representative of historically contaminated sites [14–17,21,22]. These studies exposed invertebrate communities to copper dissolved in the overlying waters at concentrations between 2 mg/L and 420 mg/L. In the context of historically contaminated sites (excluding mine-impacted sites), overlying water copper concentrations at the mid-upper end of this range are unrealistically high. Concentrations at the lower end of the range represent those likely to occur in overlying waters present at historically contaminated sites, so they may be useful for determining effects on invertebrates that are predominantly exposed to copper dissolved in the overlying water. However, in the studies with low overlying water concentrations, sediment and porewater concentrations are likely to have been unrealistically low (although they were generally not measured). For example, Shaw and Manning [14] exposed their mesocosm systems to overlying water copper concentrations between 20 mg/L and 260 mg/L and recorded sediment concentrations of between 4.7 mg/kg and 9.1 mg/kg wet weight at the end of their study. Thus, invertebrates that are predominantly exposed to copper via the particulate phase in sediments would not have been exposed to the concentrations of copper that they would be exposed to in

INTRODUCTION

There has been a sharp increase in the discharge of metals, including copper, into aquatic ecosystems within the last 100 yr [1]. However, with better management of contamination sources, concentrations of dissolved metals are now generally low in surface waters worldwide [2]. In contrast, because sediments are a sink for metals, and the metals cannot be degraded, a legacy of historical contamination remains in the sediments of aquatic ecosystems [3–5]. Concentrations of copper in sediments are often orders of magnitude greater than those in the overlying waters [6]. A review of sediments at freshwater field sites worldwide found a maximum copper concentration of 740 mg/kg dry weight [7,8] (not including mine-impacted sites, where copper partitioning is often very different from other historically contaminated sites [9]). In the field, complex and unknown stressors may exert synergistic or antagonistic effects on biota, making it difficult to elucidate the impact of a specific toxicant [10]. Mesocosms and microcosms avoid many such issues by permitting some control over the experimental conditions and allowing direct tests of causality under realistic environmental conditions. Consequently, they provide a useful tool to assess direct and indirect effects of a contaminant on diverse biological communities [11,12]. From the community-level perspective, an increase in copper concentrations generally leads to shifts in the composition of invertebrate communities and a decline in biomass and total abundance [13–17]. The uptake of metals by benthic organisms is strongly influenced by the partitioning of the metal in the All Supplemental Data may be found in the online version of this article. * Address correspondence to [email protected] Published online 20 August 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/etc.2728 2724

Effects of copper on freshwater invertebrate communities

Environ Toxicol Chem 33, 2014

field-contaminated sites with similar overlying water copper concentrations. The present study describes the effects of copper on the establishment of invertebrate communities within a series of copper-contaminated lentic mesocosms. The mesocosms were manipulated to produce as realistic copper partitioning as possible [8]. The hypothesis was that, with increasing copper concentrations, the structure of invertebrate communities in the water column and benthos would be altered and a decline in abundance, richness, and diversity of invertebrates would be observed relative to the controls. METHODS

Study design

Twenty pond mesocosms were established on the Macquarie University campus (33.769468S, 151.114968E), in northern Sydney, New South Wales, Australia. The infrastructure and experimental design of the mesocosms are described in detail by Gardham et al. [8] and briefly below. Each mesocosm had a volume of 1500 L, was sunk into the ground, shaded by 70% shade cloth, and aerated. Inputs from rainwater were sufficient to maintain water depth. Sediments were spiked in situ with copper and allowed to equilibrate for 2 mo to create environmentally relevant copper partitioning [8]. The mesocosms were opened to allow biotic colonization on 1 November 2010 (0 d). The mesocosms were allowed to colonize naturally via mobile (flying or crawling) biota, and some biota colonized the mesocosms following the introduction of a plant specimen of Vallisneria spiralis to each mesocosm after the first 6 mo. The experimental design consisted of a control (C) and 4 sediment copper concentrations (very low [VL], low [L], high [H], and very high [VH]), each with 4 replicates (Table 1). The water quality parameters and invertebrate communities (sections Copper concentrations and water quality and Invertebrate community sampling) were monitored every month in the first 6 mo (first spring/summer season; at 31 d, 72 d, 100 d, 135 d, and 161 d) and then at 12 mo (365 d), 13 mo (407 d), and 16 mo (497 d) (second spring/summer season). The final sample was taken on 12 March 2012 (497 d). The concentrations of copper in the overlying waters, porewaters, and sediments are described and discussed in detail by Gardham et al. [8]. The change in partitioning over time between treatments is also discussed within the Results section of the present study, where additional statistical analyses are included. Copper concentrations and water quality

Copper concentrations in the sediments, porewaters, and overlying waters were measured as per Gardham et al. [8]. Conductivity, pH, turbidity, dissolved oxygen, and the redox Table 1. Copper concentrations in the sediment, porewaters, and overlying waters by treatment after pseudo-equilibration had been reached (mean  standard error of the mean; n ¼ 7)a Treatment

Sediment copper (mg/kg dry wt)

Porewater copper (mg/L)

Overlying water copper (mg/L)

Control Very low Low High Very high

4.6  0.7 71  4.6 99  7.1 410  41 711  46

1.5  0.2 2.8  0.2 4.6  1.2 18  3.1 30  2.5

1.3  0.3 2.2  0.2 3.8  0.5 7.8  0.6 11  0.9

a

See Gardham et al. [8].

2725

potential (vs the standard hydrogen electrode) of the overlying water were recorded using a multimeter (HI 9828; Hanna Instruments). Invertebrate community sampling

For the benthos, 3 cores (6-cm diameter) were collected from the surficial layer of the sediment (top 1–2 cm) and were mixed together. Then, 100 mL of that pooled sediment was retained and preserved in 100% ethanol for later processing. Samples were processed by a flotation procedure adapted from that described by Anderson [23]. In this procedure, each sample was drained of ethanol through a 106-mm mesh sieve, placed in a container, and flooded with saturated glucose solution. The sample was stirred to assist organisms to float to the surface, and then the glucose solution was decanted through the 106-mm mesh sieve to collect the floating organisms. Each sample was processed twice and soaked in water for 5 min between processing to remove any residue of the glucose solution. The animals collected from both extractions were pooled and identified. Water column invertebrates were collected using a plastic tube (5-cm diameter) that was inserted vertically from the water surface to the sediment at a random location within the mesocosm. A stopper was placed at the bottom of the tube, the tube was removed from the mesocosm, and the water collected was retained. This process was repeated 4 times, and the samples were pooled. The pooled sample was mixed, then a subsample (1 L) was filtered through a 106-mm sieve, and the animals retained were preserved in 70% ethanol. At 16 mo, standardized sweep net samples were also collected using a kick net (0.12 m2 opening) with 500-mm mesh. The nets were swept across the surface of the sediment and macrophytes in one-half of each mesocosm. This process was repeated 3 times. Large macroinvertebrates (i.e., those obvious to the eye without magnification) were live-picked by 2 people until all visible animals were removed. Data analysis

Water quality data (including copper concentrations) were analyzed to identify key differences in water quality among treatments over time. A repeated-measures analysis of variance (RM-ANOVA) was performed on the data of each parameter (between-subjects factor: treatment; within-subjects factor: time) followed by least significant difference pairwise multiple comparison tests to assess the differences between individual treatments and time points using PASW Statistics [24]. Where a significant interaction between treatment and time was found, a one-way ANOVA followed by least significant difference pairwise multiple comparison was performed on the data from within each time point to identify where a significant difference among treatments occurred, and a separate RM-ANOVA was performed for the data of each treatment to identify changes in the parameter within each treatment over time. Data were transformed to meet assumptions of homogeneity of variance and normality as described in Supplemental Data, Table S1. Where transformed data failed Levene’s test, the significance level (a) was 0.01; for all other comparisons, a was 0.05 [25]. If the assumption of sphericity (assessed by Mauchly’s test) was not met during the RM-ANOVA analyses, the Greenhouse–Geisser correction was used. The effects of treatment and time on the invertebrate assemblages identified in the benthic, water column, and sweep net sample data were analyzed separately. For each data set, community indices (total abundance, taxonomic richness, and diversity [H0 ]) were each analyzed using a RM-ANOVA with post hoc tests (as described for the water quality data). The

2726

Environ Toxicol Chem 33, 2014

benthic and water column biota data sets were then analyzed using principal response curves, a multivariate method of analysis [26]. Recommended software settings in Canoco [27] were used on appropriately transformed data (Supplemental Data, Table S1). The principal response curve identifies differences in community composition between treatments and controls over time. The accompanying species weights indicate the effect of the treatment regime at the species level. Taxa with strong positive species weights respond to the treatment regime similarly to the community response shown in the principal response curve. Taxa with strong negative weights respond to the treatment regime in the opposite pattern to the principal response curve. Those taxa with weights close to 0 either do not respond to the treatment regime or respond in a way unlike that shown in the principal response curve. The significance of the principal response curve was tested using Monte Carlo permutation tests (using a ¼ 0.05). To further interrogate the pattern observed in the principal response curve, the variation in community composition between treatments at each time point was analyzed by using permutational multivariate analysis of variance (PERMANOVA) [28] followed by post hoc pairwise comparisons. The PERMANOVA analyses were based on Bray–Curtis similarities of transformed data (Supplemental Data, Table S1) and were performed in Primer 6þ [29]. To understand the responses of individual taxa to copper treatment, an RM-ANOVA was performed (as described for the water quality data) on appropriately transformed (Supplemental Data, Table S1) abundance data of sensitive taxa, identified by the principal response curve, within each invertebrate data set. Where present, even if not identified in the principal response curve analysis, the response of Physa acuta (Gastropoda: Physidae) to copper treatment was characterized because field observations indicated that the gastropod was abundant in the C/VL/L treatments and rare in the H/VH treatment. For the sweep net data set, a one-way ANOVA (as there was only 1 time point) with least significant difference post hoc analysis was performed on the transformed (Supplemental Data, Table S1) abundance data of P. acuta. RESULTS

Copper concentrations and water quality

Concentrations of copper in the sediments, porewaters, and overlying waters generally increased from the C < VL < L < H < VH treatments (Supplemental Data, Figure S1). There was a significant interaction between treatment and time on copper concentrations in the sediments (F17,64 ¼ 2.6, p ¼ 0.003), porewaters (F14,53 ¼ 3.2, p < 0.001), and overlying waters (F13,48 ¼ 4.2, p < 0.001) (Supplemental Data, Table S2). In general, there was a significant difference between the copper concentrations in the sediments among treatments; however, the VL and L treatments were only significantly different (p < 0.05) on 4 occasions (Supplemental Data, Figure S1A). In addition, at 161 d, there was no significant difference between sediment copper concentrations in the H and VH treatments (p ¼ 0.12). Within the C, L, H, and VH treatments, copper concentrations changed significantly over time (p < 0.05); however, post hoc analysis generally showed no difference between individual time points. Porewater copper concentrations were similar among C, VL, and L treatments, and they were usually significantly higher in the H and VH treatments (Supplemental Data, Figure S1B) (Supplemental Data, Table S2). Within the C, L, H, and VH

S. Gardham et al.

treatments, porewater copper concentrations changed significantly over time (p < 0.05); they decreased (up to 42 d) initially, but then a plateau was reached (Supplemental Data, Figure S1B). In the overlying waters, the concentrations of copper in the C, VL, and L treatments were similar, although they were often significantly lower in the C than the VL treatment (Supplemental Data, Figure S1C and Table S2). The H and VH treatments were also generally significantly different from each other (70% of the time; p < 0.01). Within each treatment, overlying water copper concentrations changed significantly over time, but there was no underlying directional trend (Supplemental Data, Figure S1C). There was an interaction between treatment and time on pH (F16,61 ¼ 4.6, p ¼ < 0.001), which varied between 6.8  0.1 (C, 247 d) and 8.8  0.1 (C, 310 d) (Supplemental Data, Figure S2A; Supplemental Data, Table S2). Within each sample point there was generally (60% of the time) no effect of treatment (Supplemental Data, Figure S2A). Where a significant effect of treatment was found, the pH of the C treatment was lower than that of the other treatments. This generally occurred in the first spring/summer season. The maximum difference in pH between treatments at an individual time point was 1.3 units (161 d). Although there were significant fluctuations in pH over time (p < 0.001; Supplemental Data, Table S2), there was no consistent directional trend (Supplemental Data, Figure S2A). The redox potential varied between 250  4.0 mV (VH, 407 d) and 480  5.9 mV (VH, 365 d; Supplemental Data, Figure S2B); although there was no consistent directional trend in redox potential over time, it did fluctuate significantly (F2.4,37 ¼ 293, p ¼ < 0.001; Supplemental Data, Table S2). The redox potential was not affected by treatment (F4,15 ¼ 1.1, p ¼ 0.39), and there was no interaction between treatment and time (F9.8,37 ¼ 2.5, p ¼ 0.021; a ¼ 0.01; Supplemental Data, Table S2). There was no directional trend in conductivity over time, but it fluctuated significantly (F1.8,27 ¼ 110, p < 0.001); it varied between 100  24 mS/cm (VL, 15 d) and 250  14 mS/cm (C, 100 d), except for at 135 d, when a higher conductivity was recorded across all treatments (390  11 mS/cm, n ¼ 20; Supplemental Data, Figure S2C). There was no significant effect of treatment on conductivity (F4,15 ¼ 0.60, p ¼ 0.67), nor was there an interaction between treatment and time (F7.3,27 ¼ 1, p ¼ 0.44; Supplemental Data, Table S2). At the start of the colonization period, dissolved oxygen concentrations were low (e.g., 28  0.4%, 4 d [n ¼ 20]). After 31 d, however, the overlying waters in the mesocosms were generally well oxygenated (Supplemental Data, Figure S2D). This change in dissolved oxygen over time was significant (F3.7,55 ¼ 190, p < 0.001). Treatment did not affect dissolved oxygen concentrations (F4,15 ¼ 0.59, p ¼ 0.68), and there was no interaction between treatment and time (F15,55 ¼ 0.75, p ¼ 0.72) (Supplemental Data, Table S2). Turbidity changed over time (F4.2,64 ¼ 60, p < 0.001), being highest when the mesocosms were first opened up to colonization, and then gradually decreasing. Turbidity remained between 0.6  0.1 and 5.5  2.3 nephelometric turbidity units (NTUs) after 135 d (Supplemental Data, Figure S2E and Table S2). There was no effect of treatment on turbidity (F4,15 ¼ 0.13, p ¼ 0.97), nor was there a significant interaction between treatment and time (F17,64 ¼ 1.2, p ¼ 0.28) (Supplemental Data, Table S2). In summary, the primary abiotic difference between the treatments was copper. Statistically, treatments could be grouped into C, VL/L, H, and VH by sediment copper concentrations, C/VL/L and H/VH by porewater copper

Effects of copper on freshwater invertebrate communities

Benthic community response

The total abundance of benthic invertebrates increased up to 161 d, as the benthic communities within the mesocosms became established (F7,105 ¼ 58, p < 0.001). After 161 d the total abundance was quite stable among treatments, but on occasion a spike in mean total abundance was recorded in the VL, L, or H treatments (Figure 1A), because of an unusually high abundance in a particular replicate. There were differences in the total abundance of benthic invertebrates among treatments (F4,15 ¼ 6.9, p ¼ 0.002); in the VH treatment the benthic invertebrate abundance was significantly lower compared with all other treatments (p < 0.01; Supplemental Data, Table S4). There was no significant interaction between time and treatment on total abundance of benthic invertebrates (F28,105 ¼ 1.6, p ¼ 0.058; a ¼ 0.01). There was a significant interaction between treatment and time on taxonomic richness of the benthic invertebrates (F28,105 ¼ 1.6, p ¼ 0.047). Within each treatment, taxonomic richness increased as the benthic communities in the mesocosms became established (p < 0.001; Figure 1B and Supplemental Data, Table S4). During the establishment of the benthic communities, the taxonomic richness in the H/VH treatments was consistently lower than the taxonomic richness in the C/VL/L treatments and, corresponding to this, there was a significant effect of treatment on each sampling occasion from 72 d to 365 d (p < 0.05; Figure 1B and Supplemental Data, Table S4). Taxonomic richness was reasonably constant from 407 d (Figure 1B), and there was no significant effect of treatment on taxonomic richness at 407 d and 497 d (p > 0.05; Supplemental Data, Table S4). There was a significant interaction between treatment and time on benthic diversity (F28,105 ¼ 1.9, p ¼ 0.012). Within each treatment, benthic diversity changed over time (p < 0.05), increasing up to 135 d (during the community’s establishment) and then leveling out (Figure 1C). Within each time point there was generally no effect of treatment on benthic diversity; diversity was only significantly different among treatments at 100 d (F4,15 ¼ 3.3, p ¼ 0.04). When the mesocosms were first opened for biotic colonization, the benthic communities were similar across all treatments (Pseudo-F4,15 ¼ 0.33, p[perm] ¼ 0.94). Over the first spring/ summer season (up to 161 d), as the benthic communities became established, the H/VH treatments deviated from the C/VL/L treatments (evidenced by the H/VH treatment lines

ns

A

C

C

B

B

ns

A AB B AB

*

A A A

*

C

A AB AB BC

AB ABC B

5

* A A AB

*

BC

*

AC

0 ns

ns

2.0

ns

*

ns

ns

ns

ns

1.5

135 161 Time (d)

365

407

497

B

Total abundance (N)

C

ns 10

A A AB B

In total, 35 taxa were recorded across all data sets (Supplemental Data, Table S3). Meiofauna comprised 88% of the total abundance, and were dominated by 3 Ostracoda taxa (collectively 36%) and Nematoda (26%). The most abundant macrofauna were Diptera (10%), which were dominated by the Chironomidae subfamilies Chironominae and Tanypodinae. Other relatively abundant macroinvertebrates were Libellulidae (Odonata), P. acuta (Gastropoda: Physidae), and Mesovelia sp. (Hemiptera: Veliidae).

800 700 600 500 400 300 200 100 0

B Taxonomic richness

The aquatic community

2727

A

Diversity (H')

concentrations, and C/VL/L, H, and VH by overlying water copper concentrations. Conductivity, dissolved oxygen, redox potential, and turbidity were stable among treatments. Although the overlying water pH was significantly affected by treatment at some time points, this was rare and did not reflect pronounced differences in the environment; for example, the overlying water was still neutral/slightly alkaline where the pH was significantly different among treatments.

Environ Toxicol Chem 33, 2014

1.0 0.5 0.0 31

72

100

Figure 1. Benthic community (A) total abundance, (B) taxonomic richness, and (C) diversity over time. The concentrations included a control (white), very low (dashed horizontal), low (diagonal cross-hatching), high (dashed vertical), and very high (black). The main results of the one-way analyses of variance performed at each time point on the taxa richness and diversity data (because of a significant interaction between time and treatment for those data) are displayed above each respective time point. An asterisk indicates significance at p < 0.05. Letters above each bar indicate the outcomes of least significant difference post hoc analyses; treatments that do not share a letter within a time point are significantly different. ns ¼ not significant.

moving further from C/VL/L treatments in Figure 2). At the start of the second spring/summer period (365 d) the apparent difference between H/VH and C/VL/L treatments was still present, but all of the treatments gradually became more similar; and at 497 d, there was little difference among treatments (Figure 2). Treatment accounted for 25% of the total variance in the benthic community across all time points. The first axis of the principal response curve for the benthic community was significant (p ¼ 0.002) and accounted for 44% of the variance explained by the treatment regime (Figure 2). The second axis accounted for 18% of the variance explained by the treatment regime and was not significant (p ¼ 0.13); thus, only the first axis is presented and discussed. There was a significant interaction between time and treatment on the benthic community assemblages (Pseudo-F28,120 ¼ 1.4, p ¼ 0.002). Within each time point, there was no significant difference between the benthic communities in the C/VL/L treatments. The apparent difference between C/VL/L treatments and the VH treatment was generally significant (p < 0.05), or near significant (p < 0.091), from 72 d onward (including the second spring/summer season through to 497 d; Supplemental Data, Table S5). The apparent differences between the H treatment and C/VL/L treatments were significant (p < 0.05), or nearly significant (p < 0.063), in all 3 of the C/VL/L treatments on only 2 occasions (74 d and 409 d; Supplemental Data, Table S5). The taxa most sensitive to copper treatment (i.e., with species weights >2) were Chironominae, Cladocera, and Ostracod Sp. 2 (Figure 2). There was a significant interaction between time and treatment on the abundance of Chironominae (F28,105 ¼ 2.9, p < 0.001). The effect of time on Chironominae abundance in the C/VL/L/H treatments was significant (p < 0.01; Supplemental Data, Table S4). The taxa increased in abundance over the first spring/summer season (up to 161 d), as the benthic

2728

Environ Toxicol Chem 33, 2014

Figure 2. Principal response curves with species weights showing the effect of copper on establishment of the benthic invertebrate community. The concentrations included a control (open square), very low (open triangle), low (open circle), high (solid diamond), and very high (solid square). For species weights, tick marks on the species weights axis indicate other taxa with calculated species weights (labels have been removed for clarity). [Color figure can be viewed in the online issue which is available at wileyonlinelibrary.com]

communities in the mesocosms became established. However, by the second spring/summer season (365 d), overall Chironominae abundance was lower than at the end of the first spring/ summer season (161 d) in the C/VL/L/H treatments. In the VH treatment, where the abundance of Chironominae was consistently low, there was no effect of time (F7,21 ¼ 1.4, p ¼ 0.28). Within most sampling points (75%), treatment affected

S. Gardham et al.

Chironominae abundance (Figure 3A and Supplemental Data, Table S4); generally, the abundance of Chironominae in the VH treatment was significantly lower than in the C/VL/L treatments. The abundance of Chironominae in the H treatment was also lower than the C/VL/L treatments, apart from the last 2 sampling time points (407 d and 497 d), where the abundance of Chironominae in the H treatment exceeded all other treatments. The abundance of Ostracoda Sp. 2 increased in the C/VL/L treatments during the first spring/summer season (up to 161 d) but remained low in the H/VH treatments (Figure 3B). By the second spring/summer season (365 d), the abundance of Ostracod Sp. 2 was lower in the C/VL/L treatments (compared with 161 d) and more similar to the H/VH treatments. There was a significant interaction between time and treatment on the abundance of Ostracoda Sp. 2 (F11,40 ¼ 2.6, p ¼ 0.014). The change in abundance of Ostracoda Sp. 2 over time was significant only in the C treatment (F7,21 ¼ 4.8, p ¼ 0.002), and the effect of treatment within each sampling time was significant only 40% of the time (at 100 d, 161 d, and 407 d; Supplemental Data, Table S4). During the first spring/summer season (up to 161 d) the abundance of Cladocera increased in the C/VL/L treatments but remained low in the H/VH treatments (Figure 3C). By the second spring/summer season (365 d), the abundance of Cladocera in the H treatment had increased to be similar to the C/VL/L treatments; then at 497 d, the abundance of Cladocera was low

Figure 3. Population responses of individual benthic taxa over time: (A) Chironominae; (B) Ostracoda Sp. 2; (C) Cladocera; (D) Physa acuta (which appeared in the mesocosms during the second spring/summer season). The concentrations included a control (open square), very low (open triangle), low (open circle), high (solid diamond), and very high (solid square). Values are mean  standard error of the mean (n ¼ 4). The main results of the one-way analyses of variance performed at each time point on the abundances of Chironominae, Ostracoda Sp. 2., and Cladocera (because of a significant interaction between time and treatment for those data) are displayed above each respective time point:  p < 0.05,  p < 0.01,  p < 0.001, ^ ¼ homogeneity of variance, p < 0.05. ns ¼ not significant. [Color figure can be viewed in the online issue which is available at wileyonlinelibrary.com]

Effects of copper on freshwater invertebrate communities

across all treatments. There was a significant interaction between time and treatment on the abundance of Cladocera (F14,53 ¼ 2.1, p ¼ 0.029). The fluctuations in Cladocera abundance over time were significant (or nearly significant) in the C, VL, L, and H treatments (p < 0.055; Supplemental Data, Table S4). Although there was no statistically significant effect of treatment on Cladocera abundance on any sampling occasion, a clear biologically significant difference was apparent between treatments in the first spring/summer season (74–161 d). Physa acuta colonized the mesocosms between the first and second spring/summer seasons (between 161 d and 365 d; Figure 3D). Corresponding to this, the effect of time was significant (F7,105 ¼ 5.6, p < 0.001). In the benthic samples, the abundance of P. acuta was not affected by treatment (F4,15 ¼ 1.3, p ¼ 0.3). There was no interaction between treatment and time (F28,105 ¼ 0.72, p ¼ 0.84). Although there was no significant effect of treatment on abundance of P. acuta in the benthic samples, there was a clear biological effect, in which abundance was greater in the C/VL/L treatments compared with the H/VH treatments (Figure 3D); this was also evident from field observations, where the gastropod proliferated in the C/VL/L treatments and was rare or absent in the H/VH treatments (personal observation), and is supported by the sweep net data (see the Sweep net samples section). In summary, there was a clear difference in the composition of the benthic communities between the H/VH treatments and the C/VL/L treatments. During the establishment of the benthic communities in the first spring/summer season, the different abundances of Chironominae, Ostracod Sp. 2, and Cladocera among treatments were the main drivers of the community response. Although, according to the principal response curve, the benthic communities appeared to become more similar among treatments during the second spring/summer season, differences in community composition remained. Water column community response

The total abundance of invertebrates in the water column increased up to 135 d (Figure 4A). During this time, the abundance of invertebrates in the VH treatment was lower than in the C/VL/L/H treatments. After 135 d, the total abundance of invertebrates decreased and then remained low during the second spring/summer season (365–497 d). The effect of time on the total abundance of invertebrates in the water column was significant (F8,120 ¼ 59, p < 0.001). However, there was no significant effect of treatment (F4,15 ¼ 1, p ¼ 0.44) on total abundance, nor was there a significant interaction between time and treatment (F32,120 ¼ 1.2, p ¼ 0.26). The taxonomic richness of the invertebrates in the water column increased up to 72 d and then reached a plateau (Figure 4B). This change over time was significant (F8,120 ¼ 9.5, p < 0.001). There was no effect of treatment on taxonomic richness (F4,15 ¼ 2.4, p ¼ 0.098), nor was there a significant interaction between treatment and time (F32,120 ¼ 1.2, p ¼ 0.22). The diversity of the invertebrates in the water column was low throughout the experimental period, but generally it was higher in the second spring/summer season (365–497 d) compared with the first (up to 161 d), and there was a significant effect of time (F8,120 ¼ 3.6, p < 0.001; Figure 4C). There was no significant difference in diversity of the invertebrates in the water column among treatments (F4,15 ¼ 1.3, p ¼ 0.32), nor was there an interaction between treatment and time (F32,120 ¼ 0.82, p ¼ 0.74). There was no significant difference in the water column invertebrate assemblages between treatments at the start of colonization (Pseudo-F4,15 ¼ 0.86, p[perm] ¼ 0.60). There was

Environ Toxicol Chem 33, 2014

2729

Figure 4. Water-column community (A) total abundance, (B) taxonomic richness, and (C) diversity over time. The concentrations included a control (white), very low (dashed horizontal), low (diagonal cross-hatching), high (dashed vertical), and very high (black).

no clear pattern as a result of treatment on the development of the invertebrate community in the water column over time (Figure 5); the first axis of the principal response curve for the water column community accounted for 36% of the variance explained by the treatment regime, but was not significant (p ¼ 0.66). Overall, there was no significant interaction between treatment and time (Pseudo-F32,135 ¼ 1, p ¼ 0.41); however, there were significant differences in the invertebrate assemblages among treatments as they developed (Pseudo-F4,135 ¼ 1.8, p ¼ 0.017). Pairwise comparisons identified significant differences in assemblages among the VL/L and H/VH treatments on several occasions (p < 0.05; Supplemental Data, Table S6). Sweep net samples

There was no significant effect of treatment on total abundance (F4,15 ¼ 0.33, p ¼ 0.85), taxonomic richness (F4,15 ¼ 0.78, p ¼ 0.56), or diversity (F4,15 ¼ 2.5, p ¼ 0.086) of the large-macroinvertebrate assemblages collected with

Figure 5. Principal response curves with species weights showing the effect of copper on establishment of the invertebrate community in the water column. The concentrations included a control (open square), very low (open triangle), low (open circle), high (solid diamond), and very high (solid square). For species weights, tick marks indicate other taxa with calculated species weights (labels have been removed for clarity). [Color figure can be viewed in the online issue which is available at wileyonlinelibrary.com]

2730

Environ Toxicol Chem 33, 2014

sweep nets. The abundance of P. acuta collected in the sweep net samples was analyzed because field observations indicated that the gastropod was abundant in the C/VL/L treatments and rare in the H/VH treatments. Although there was no significant difference among treatments (F4,15 ¼ 1.9, p ¼ 0.17), P. acuta abundance was clearly higher in the C treatment than all other treatments (Figure 6). DISCUSSION

There was a clear and significant difference in the invertebrate assemblages that became established within the highly copper-contaminated mesocosms compared with the lower copper-contaminated and control mesocosms. The richness, abundance, and structure of the benthic invertebrate assemblages, in particular, were strongly affected by copper contamination. Benthic Chironominae, a species of Ostracoda, Cladocera, and P. acuta were particularly sensitive to copper. Contrasting the benthos, invertebrate assemblages in the water column showed little response to copper contamination. This general response of impacted benthic invertebrate assemblages and apparently unaffected assemblages of invertebrates in the water column is as expected in contaminated field sites in which an equilibrium between sediment (with high copper concentrations) and water (with low copper concentrations) has been reached. Our results in terms of biological effects contrast with previous studies and accordingly highlight the importance of establishing environmentally relevant partitioning of copper in experimental systems. Although copper partitioning between the sediment, porewaters, and overlying waters was representative of historically contaminated sites (with high particulate- and low dissolvedcopper concentrations), on occasions the overlying water pH was higher in the copper-contaminated mesocosms than in the control mesocosms. This is important because higher pH levels lower the solubility of metal hydroxides and increase adsorption of metals to the particulate phase [30]. This means that if the pH in the copper-contaminated mesocosms was similar to that in the control mesocosms, the copper may have been more bioavailable. Therefore, the effects of copper on the invertebrate communities observed in the copper-contaminated—compared with the control—mesocosms may be conservative. The importance of an environmentally realistic exposure scenario

In contrast to previous lentic mesocosm studies that considered the effect of dissolved copper in the overlying water on benthic communities [14,16,17], the present study used an environmentally realistic exposure scenario, with high concentrations of copper in the sediment and low dissolved concentrations of copper in the porewaters and overlying waters [8]. As a result, the concentrations of copper in sediments in the present study were much higher than those reported by Hedtke [17] and

Figure 6. Total abundance of Physa acuta among treatments in the sweep net samples. C ¼ control; VL ¼ very low; L ¼ low; H ¼ high; VH ¼ very high.

S. Gardham et al.

Shaw and Manning [14] (711  4.6 mg/kg dry wt in the VH treatment compared with 37 mg/kg dry wt [17] and 5.3 mg/kg wet wt [14]), even though there were similar concentrations of copper in the overlying water (11  0.9 mg/L in the VH treatment compared with 9.3 mg/L [17] and 20 mg/L [14]). It follows that the concentrations of copper in the porewaters were also higher in the present study compared with those previous studies, but this cannot be tested because porewater concentrations were not measured in previous studies. With higher sediment (and likely porewater) copper concentrations, effects were observed on the benthic community when overlying water concentrations of copper were lower than reported in previous studies. Changes in the benthic community were evident in the H and VH treatments, in which average overlying copper concentrations were 7.8  0.6 mg/L and 11  0.9 mg/L, respectively. In comparison, Hedtke [17] and Shaw and Manning [14] observed effects on the benthic community at overlying water copper concentrations of 30 mg/L (not 9.3 mg/L) and >20 mg/L, respectively. This demonstrates that, in agreement with laboratory studies [18,31], the major route of exposure to biota in the benthos is via the porewaters and sediments. In terms of risk assessment, this is extremely important because, based on previous studies [14,17], it may be assumed from overlying water copper concentrations that there would be no effect of copper on the benthic community, when in fact concentrations in the porewaters and sediments could be causing an effect. Because the copper partitioning in the present study is representative of historically contaminated sites [8], it is concluded that chronic copper exposure may affect the invertebrates present in the benthos at such sites. However, because effects on the water column invertebrate community were not evident, the invertebrates in the water column at historically contaminated sites may not be affected. This lack of response in the water column invertebrate community is in line with previous research. Two studies considered the effects of copper on water column invertebrate communities during chronic (5-wk and 32-wk) exposures with overlying water copper concentrations in the same range as those used in the present study. Winner et al. [16] observed changes in the zooplankton community at concentrations of 20 mg Cu/L in the overlying waters, whereas Hedtke [17] observed changes occurring between 9.3 mg Cu/L and 30 mg Cu/L. In the present study, therefore, the concentrations of copper in the highest treatment (VH: average of 11  0.9 mg/L; maximum of 15  0.5 mg/L at 135 d) were between the no-effect and effect concentrations reported previously [16,17]. In addition to considering copper partitioning during exposure experiments, the present study also emphasizes the importance of carrying out truly chronic exposures to understand the long-term effects of a persistent contaminant in the field. The strong response of the benthic community to copper in the present study, carried out over 71 wk, is similar to other longterm studies. Hedtke [17] also recorded severe effects of copper on snails (Viviparus, Physa sp., and Helisoma campanulata) during a 32-wk overlying water copper exposure, and Shaw and Manning [14] observed a pronounced response of the benthic community to copper during 19-wk exposures. During their 5-wk lentic exposures, in contrast, Winner et al. [16] observed a weak response of the benthic community to copper; the densities of small chironomids responded differently depending on season, and the only consistent response in the benthic community to copper was a reduction in the densities of small caenid mayflies at 40 mg/L (compared with 0 mg/L and 20 mg/L

Effects of copper on freshwater invertebrate communities

treatments). Specifically, Winner et al. [16] also noted that snails were unaffected by copper. Even based on overlying water exposures, Winner et al. [16] should have seen a response (as noted earlier, Hedtke [17] and Shaw and Manning [14] observed strong effects on the benthic community at overlying water copper concentrations of 30 mg/L [not 9.3 mg/L] and 20 mg/L, respectively). Winner et al. [16] supposed that the weak response they observed in the benthos to copper could have occurred because the 5-wk experiments did not incorporate effects on reproduction and recruitment, a theory supported by the present study and the long-exposure studies previously reported [14,17]. The importance of interspecies variation in sensitivity

Within the benthic response, some taxa responded differently compared with previous studies. For example, Tanypodinae were negatively affected by copper [14]; in the present study, however, no effect of copper was identified. Other studies that have considered the effects of copper on Tanypodinae in the field have also found species within the subfamily to be copper tolerant, which supports the lack of response in the present study [32,33]. There was also interspecies variation in copper susceptibility of Ostracoda within the present study, as 1 species of benthic Ostracoda was not affected by copper but another declined in abundance. In contrast, Shaw and Manning [14] showed an increase in benthic Ostracoda. This demonstrates that organisms with similar life histories may respond differently to contaminants and confirms that it is important to identify organisms to an appropriate taxonomic level to allow contaminant specific responses to be identified. The importance of interspecific interactions

Interspecific interactions (interactions between individuals of different species) within the mesocosms influenced the response of the benthic community to copper. In the first spring/summer season (up to 161 d), the abundances of Chironominae, the Ostracoda species, and Cladocera were lower in the H/VH treatments compared with the control. In the second spring/ summer season (from 365 d), however, the abundances of each taxa generally declined in the C/VL/L treatments to become more similar to the H/VH treatments. It is possible that the decrease in abundance of these taxa was attributable to interspecific competition in the low copper treatments between these taxa and the gastropod P. acuta (which colonized the mesocosms and proliferated in those treatments between 161 d and 365 d). Interspecific competition between these taxa has been documented elsewhere [34–37]. For Ostracoda and Cladocera, it may be the result of competition for food; but for Chironominae, there is evidence to suggest that reproductive interference may occur. Devereaux and Mokany [35] showed that Chironomus oppositus avoided P. acuta during site selection for oviposition. Clearly, the interspecific interactions between taxa were important in driving the biotic response to copper, demonstrating that understanding these interactions is necessary when characterizing the potential effects a contaminant has, or will have, on a community. Stage of ecological succession

The present study indicates that the effect of copper on taxonomic richness is most prominent when benthic communities are establishing in new, or recently disturbed, freshwater habitats. An effect of higher taxonomic richness in C/VL/L treatments compared with H/VH treatments was observed only during the establishment of the benthic communities within the

Environ Toxicol Chem 33, 2014

2731

mesocosms. Consistent with this, Shaw and Manning [14] observed no effect on taxonomic richness on already established benthic communities. CONCLUSIONS

There was a strong difference in the benthic communities that developed within the highly copper-contaminated sediments compared with those in the controls. However, there was little effect of copper on the invertebrates in the water column. The present study has demonstrated the importance of using environmentally realistic exposure scenarios, including both realistic partitioning of contaminants between phases in the environment and environmentally relevant exposure times, the latter being particularly important with regard to persistent contaminants in the field. Furthermore, there was evidence to suggest that communities may be particularly sensitive during colonization, with implications for environmental management when sites are disturbed. The present study has demonstrated the importance of considering interspecific interactions and interspecies variation in sensitivity to contaminants in driving the community response. Ultimately, the present study has shown that different communities will respond to contaminants in different ways through both direct and indirect effects. Thus, although guidelines are useful, site-specific assessments on the health of aquatic ecosystems are paramount in understanding the effects of contaminants at those sites. SUPPLEMENTAL DATA

Tables S1–S6. Figures S1–S2. (406 KB PDF). Acknowledgment—Construction of the mesocosms was funded by a Macquarie University Infrastructure Grant. S. Gardham was supported by a Macquarie University Research Excellence Scholarship. The authors thank A. Michie, M. Nagel, and L. Oulton, for their help in running the experiment, and 2 anonymous reviewers for their useful comments. We also thank the CSIRO’s Water for a Healthy Country flagship (partial funding and specialist time). The authors have no conflicts of interest to declare.

REFERENCES 1. Eisler R. 1998. Copper hazards to fish, wildlife, and invertebrates: A synoptic review. Biological Science Report USGS/BRD/BSR—19970002. US Geological Survey Washington, DC. 2. De Deckere E, de Cooman W, Leloup V, Meire P, Schmitt C, Ohe PC. 2011. Development of sediment quality guidelines for freshwater ecosystems. J Soils Sediments 11:504–517. 3. Burton GA. 2002. Sediment quality criteria in use around the world. Limnology 3:65–76. 4. Brils J. 2004. Sediment monitoring under the EU Water Framework Directive. J Soils Sediments 4:72–73. 5. Chon H-S, Ohandja D-G, Voulvoulis N. 2012. The role of sediments as a source of metals in river catchments. Chemosphere 88:1250–1256. 6. Harrahy EA, Clements WH. 1997. Toxicity and bioaccumulation of a mixture of heavy metals in Chironomus tentans (Diptera: Chironomidae) in synthetic sediment. Environ Toxicol Chem 16:317–327. 7. Vinot I, Pihan JC. 2005. Circulation of copper in the biotic compartments of a freshwater dammed reservoir. Environ Pollut 133:169–182. 8. Gardham S, Hose GC, Simpson SL, Jarolimek C, Chariton AA. 2014. Long-term copper partitioning of metal-spiked sediments used in outdoor mesocosms. Environ Sci Pollut Res 21:7130–7139. 9. Ríos CA, Williams CD, Roberts CL. 2008. Removal of heavy metals from acid mine drainage (AMD) using coal fly ash, natural clinker and synthetic zeolites. J Hazard Mater 156:23–35. 10. Hutchins CM, Teasdale PR, Lee SY, Simpson SL. 2008. Cu and Zn concentration gradients created by dilution of pH neutral metal-spiked marine sediment: A comparison of sediment geochemistry with direct methods of metal addition. Environ Sci Technol 42:2912–2918.

2732

Environ Toxicol Chem 33, 2014

11. Brinke M, Ristau K, Bergtold M, H€ oss S, Claus E, Heininger P, Traunspurger W. 2011. Using meiofauna to assess pollutants in freshwater sediments: A microcosm study with cadmium. Environ Toxicol Chem 30:427–438. 12. Relyea R, Hoverman J. 2006. Assessing the ecology in ecotoxicology: A review and synthesis in freshwater systems. Ecol Lett 9:1157–1171. 13. Girling AE, Pascoe D, Janssen CR, Peither A, Wenzel A, Schäfer H, Neumeier B, Mitchell GC, Taylor EJ, Maund SJ, Lay JP, Jüttner I, Crossland NO, Stephenson RR, Persoone G. 2000. Development of methods for evaluating toxicity to freshwater ecosystems. Ecotoxicol Environ Saf 45:148–176. 14. Shaw JL, Manning JP. 1996. Evaluating macroinvertebrate population and community level effects in outdoor microcosms: Use of in situ bioassays and multivariate analysis. Environ Toxicol Chem 15:608–617. 15. Havens KE. 1994. Structural and functional responses of a freshwater plankton community to acute copper stress. Environ Pollut 86:259–266. 16. Winner RW, Owen HA, Moore MV. 1990. Seasonal variability in the sensitivity of freshwater lentic communities to a chronic copper stress. Aquat Toxicol 17:75–92. 17. Hedtke SF. 1984. Structure and function of copper-stressed aquatic microcosms. Aquat Toxicol 5:227–244. 18. Campana O, Simpson SL, Spadaro DA, Blasco J. 2012. Sub-lethal effects of copper to benthic invertebrates explained by sediment properties and dietary exposure. Environ Sci Technol 46:6835–6842. 19. Fukunaga A, Anderson MJ. 2011. Bioaccumulation of copper, lead and zinc by the bivalves Macomona liliana and Austrovenus stutchburyi. J Exp Mar Biol Ecol 396:244–252. 20. Simpson SL, Batley GE. 2007. Predicting metal toxicity in sediments: A critique of current approaches. Integr Environ Assess Manag 3:18–31. 21. Havens KE. 1994. An experimental comparison of the effects of two chemical stressors on a freshwater zooplankton assemblage. Environ Pollut 84:245–251. 22. Moore MV, Winner RW. 1989. Relative sensitivity of Cerodaphnia dubia laboratory tests and pond communities of zooplankton and benthos to chronic copper stress. Aquat Toxicol 15:311–330. 23. Anderson RO. 1959. A modified flotation technique for sorting bottom fauna samples. Limnol Oceanogr 4:223–225. 24. Statistical Package for the Social Sciences. 2009. PASW Statistics for Macintosh, Ver 18.0.3. Chicago, IL, USA.

S. Gardham et al. 25. Underwood AJ. 1997. Experiments in Ecology: Their Logical Design and Interpretation Using Analysis of Variance. Cambridge University Press, Melbourne, Australia. 26. Van den Brink PJ, Ter Braak CJF. 1999. Principal response curves: Analysis of time-dependent multivariate responses of biological community to stress. Environ Toxicol Chem 18:138–148. 27. Ter Braak CJF, Smilauer P. 2002. CANOCO reference manual and CanoDraw for Windows user’s guide: Software for canonical community ordination (version 4.5). Microcomputer Power, Ithaca, NY, USA. 28. Anderson MJ. 2001. A new method for non-parametric multivariate analysis of variance. Austral Ecol 26:32–46. 29. Clarke KR, Gorley RN. 2006. Primer v6: User Manual/Tutorial. PRIMER-E, Plymouth, UK. 30. Atkinson CA, Jolley DF, Simpson SL. 2007. Effect of overlying water pH, dissolved oxygen, salinity and sediment disturbances on metal release and sequestration from metal contaminated marine sediments. Chemosphere 69:1428–1437. 31. Strom D, Simpson SL, Batley GE, Jolley DF. 2011. The influence of sediment particle size and organic carbon on toxicity of copper to benthic invertebrates in oxic/suboxic surface sediments. Environ Toxicol Chem 30:1599–1610. 32. Montz GR, Hirsch J, Rezanka R, Staples DF. 2010. Impacts of copper on a lotic benthic invertebrate community: Response and recovery. J Freshw Ecol 25:575–587. 33. Clements WH, Cherry DS, Cairns JJ. 1988. Structural alterations in aquatic insect communities exposed to copper in laboratory streams. Environ Toxicol Chem 7:715–722. 34. Cuker BE. 1983. Competition and coexistence among the grazing snail Lymnaea, Chironomidae, and microcrustacea in an arctic epilithic lacustrine community. Ecology 64:10–15. 35. Devereaux JS-L, Mokany A. 2006. Visual and chemical cues from aquatic snails reduce chironomid oviposition. Aust J Zool 54: 79–86. 36. Gresens SE. 1995. Grazer diversity, competition and the response of the periphyton community. Oikos 73:336–346. 37. Harvey BC, Hill WR. 1991. Effects of snails and fish on benthic invertebrate assemblages in a headwater stream. J North Am Benthol Soc 10:263–270.

Copyright of Environmental Toxicology & Chemistry is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.