Hierarchical Responses of Soil Invertebrates (Earthworms) to Toxic

blanks, and certified reference material (calcareous soil sediment BCR CRM ... A comparison of soil concentrations measured by Spurgeon and Hopkin...
0 downloads 0 Views 281KB Size
Download PDF
Environ. Sci. Technol. 2005, 39, 5327-5334

Hierarchical Responses of Soil Invertebrates (Earthworms) to Toxic Metal Stress D A V I D J . S P U R G E O N , * ,† H U W R I C K E T T S , ‡ CLAUS SVENDSEN,† A. JOHN MORGAN,‡ AND PETER KILLE‡ Centre for Ecology and Hydrology, Monks Wood, Abbots Ripton, Huntingdon, Cambridgeshire, PE28 2LS, United Kingdom, and School of Biosciences, University of Cardiff, P.O. BOX 911, Cardiff, CF1 3US, United Kingdom

The concept of a hierarchical cascade of biological responses to stress occurring across different levels of biological organization is an underlying principle of both theoretical and regulatory ecology/ecotoxicology. This study investigates the reality of this cascade for earthworms exposed to toxic metal stress. Gene expression was the most sensitive endpoint (EC50 ) 616 µg Zn g-1) followed by the integrity of coelomocyte lysosomal membrane (EC50 ) 645 µg Zn g-1). This confirms that, in accordance with the cascade concept, suborganism level endpoints respond at lower metal concentrations than higher organization endpoints. The relative sensitivity of the higher organization parameters was not as predicted by the cascade. Organic material removal was more sensitive (EC50 ) 997 µg Zn g-1) than L. rubellus reproduction (EC50 ) 3236 µg Zn g-1), L. rubellus population size (EC50 ) 5000-11500 µg Zn g-1), and earthworm community diversity (EC50 ) 1737 µg Zn g-1). This can be attributed to (1) the relative insensitivity of L. rubellus to metals and (2) general toxic effects of metals on the earthworm energy budget (and thus feeding). On the basis of these results, it can be concluded that predictive assessments of the consequences of environmental stressors needs to include approaches that respect the relative sensitivities of different taxa, while retrospective appraisals should exploit the sensitivity of low organization level responses.

Introduction A central paradigm in (eco)toxicology is that there is a tiered cascade in biological responses to increasing chemical exposure. Represented in Figure 1, this cascade proposes that the first responses to stress will occur at low levels of biological organization (e.g., changes in gene and protein expression). If the stress persists or the severity increases (due either to increased exposure or to environmentally induced changes in organism sensitivity), these molecular level effects will accrue to cause cellular alterations (e.g., effects on organelles) followed by tissue level dysfunction (e.g., carcinoma). Further increases in the stress will eventually lead to changes in the life history parameters of the organism (e.g., altered reproduction rate, reduced growth, * Corresponding author phone: + 44 1487 772 563; fax: + 44 1487 773 467; e-mail: [email protected]. † Centre for Ecology and Hydrology. ‡ University of Cardiff. 10.1021/es050033k CCC: $30.25 Published on Web 06/15/2005

 2005 American Chemical Society

FIGURE 1. Schematic diagram of the hierarchical relationship between ecotoxicological responses measured at different levels of biological organization. altered life-span), which, in accordance with population theory, if sustained will give rise to changes in population structure. If these population changes result in local extinctions, community diversity will be reduced, leading ultimately to changes in the functional integrity of ecosystems. The notion of a hierarchical cascade linking the severity of chemical exposure to the biological level of organization at which effects can be measured is a well-recognized and fundamental principal (1). The cascade is brought to notice in the introductory chapters of standard ecotoxicology publications (e.g., refs 2 and 3), and although the form of the presentation may differ, the underlying principle that “processes at one level take their mechanism from the level below and find their consequence at the level above” remains consistent (4). Despite the central position occupied by the hierarchical response cascade in the partner fields of stress biology and ecotoxicology, to date, no single study has undertaken a systematic comparison of the relative sensitivity to stress of biological responses at different levels of biological organization from gene to ecosystem. So far, only links between responses at two or three biological levels of organization (e.g., population to community and gene/protein expression to cellular integrity) have been investigated (5-8). The absence of studies assessing effects at a range of levels of organization probably results from the divergence of approaches used in stress physiology and ecological effect assessment. Stress physiology has focused on understanding the mechanisms for maintaining homeostasis and the consequences for biochemical pathways, physiological performance, and tissue and cell morphologies of any departures from this state (9). In contrast, ecological effect assessment has focused only on ecologically meaningful phenomena, such as population, community, and ecosystem functional changes, usually under field conditions (10). The divergence of stress physiology and stress ecology has recently been recognized as a barrier to predicting the consequences of contemporary environmental scenarios in industrial societies where typically biological systems, including man, are chronically exposed to complex mixtures of stressors (11, 12). To unify the mechanistic and ecological approaches, a first step would be to better characterize the hierarchical response cascade to confirm the relative sensitivities of traits from different levels of biological organization. This study constructs such an analysis by considering the multiple biological responses of earthworm genes, cells, VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5327

FIGURE 2. Location of field sites in the region around a primary cadmium/lead/zinc smelter located at Avonmouth, southwest UK. Axes give ordnance survey grid reference in kilometers. Shaded area indicates the urban area of Bristol. M48, M4, and M5 indicate position of major roads. individuals, populations, communities, and decomposer functional assemblages to toxic metals along a field contamination gradient.

Experimental Procedures Study Site. All analyses were conducted at a series of rough grassland sites in an area to the northeast of a primary Cd, Pb, and Zn smelter located at Avonmouth, southwest England (Figure 2). Operating until December 2002, this factory was one of the largest smelters of its type in the world and a major source of metal pollution (13). Previous studies in the area have indicated the presence of elevated metal concentrations in soils at least 15 km downwind of the factory (14). The sites chosen for use in this study were located along a gradient of contamination ranging from severe (soil containing up to 5% metal) to approaching background. Ecosystem Functioning, Earthworm Community Structure, and L. rubellus Population Size. The functional activity of the decomposer community at selected Avonmouth sites was measured by Filzek et al. (15) using the bait lamina test. This assay measures the removal of food pellets contained in small holes within plastic strips (16) and, thus, gives a measure of the removal of organic material by soil fauna and microbes as a part of carbon cycling. Bait lamina feeding was measured at six of the gradient sites (12, 9, 7, 4, 2, and 2a) using three replicate plots of 16 strips. These were inserted vertically in a four-by-four grid with 15-20 cm between adjacent strips and the topmost bait positioned just below the soil surface. Earthworm diversity and L. rubellus abundance were measured at all 14 gradient sites by Spurgeon and Hopkin (17). Raw data from three seasonal samples taken in this study (spring, autumn, and winter) were reanalyzed to interpret metal effects. Diversity, as a measurement of community effect, was calculated as the Shannon Wiener statistic (H′) for each site in each of the three seasonal surveys. The L. rubellus population size was calculated as the average number of L. rubellus present per m2 at each site in the three seasonal samples. Although collected some 3 years prior to the bait lamina data and 5 years before the reproduction, NRR-T, and mt-2 expression data, the comparison of soil metal concentrations measured in site soils in each study showed only small differences in measured metal concentrations. Additionally, observations made during the collec5328

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005

tion of the soils used for the mesocosm study and of worms for further biomarker studies in 2002 (18) indicated there has been little change in the distribution of common earthworm species around the smelting works in the intervening years. Lifecycle Effects, Cellular Integrity, and Gene Expression in L. rubellus. In situ measurements of life cycle performance and physiological responses in field populations were not logistically feasible in this study. Instead, physiological and demographic response parameters in naı¨ve worms exposed to selected Avonmouth soils in mesocosms constructed as described by Svendsen and Weeks (19) were measured. Soil from five sites (12, 9, 7, 4, and 2) was dug (from a series of patches) to a maximum depth of 10 cm after removal of the turf layer. Soil was returned, frozen at -20 °C to remove the indigenous earthworm fauna, dried at ambient temperature, and then screened through a 2 mm mesh. Twelve L of each soil was used in each of four replicate mesocosms per site. As a negative control, a clean clay loam soil (Broughton Loam, Kettering, UK) sieved through a 2 mm mesh and amended with the 3% composted bark was used. A representative sample of soil from each mesocosm was also taken at this time for chemical analyses. The soils for use in the test were wetted to 60% of water holding capacity in watertight bags and then left to stabilize for 2 weeks with the bags left open to the air before being transferred to the mesocosms. The use of a stabilization period was important as it was possible that the soil manipulation (freezing-drying) could have caused changes in bioavailability. The length of the stabilization period used has been found to be suitable for the initial equilibration of metal spiked soils (20) and was so considered adequate to allow the field soils prepared for the study to return to equilibrium. To initiate the exposure, 12 adult (fully clitellate) L. rubellus (mean weight 1150 mg) were added to each replicate. The mesocosms were then embedded 30 cm (over drainage gravel) into the outdoor field plot located in southeast UK (Ordnance Survey Grid Reference TL 798202) and left for 70 days. The mesocosms were set up in January 2001, and the experiment was terminated after 70 days exposure. During the exposure, the soil temperature at 2 and 20 cm depth was monitored hourly using automated temperature data loggers. Daily rainfall was also recorded. Each unit was checked twice weekly to ensure the integrity of all seals and surface fed every 14 days with 10 g dry weight of horse manure (from a known source not subject to recent medication) rewetted to 80% moisture content. Temperature in the field plot was automatically recorded at 0, 10, and 20 cm depth, and the daily rainfall was measured. At the end of the exposure, surviving worms were hand-sorted from the soil and placed on moist filter paper for 24 h to allow egestion of gut contents. Soil from the mesocosms was sieved through a 1 mm mesh to collect any cocoons. This number was then compared to the survival data to allow cocoon production rate (cocoons/ worm/week) in each replicate to be calculated. To establish toxic responses at the cellular level, metal effects on lysosomal membrane stability were measured using the neutral red retention time (NRR-T) assay (21, 22). The NRR-T assay measures changes in the fragility of earthworm coelomocyte lysosome membranes following exposure to stress (23). Measurements were conducted using an in vitro technique (22) in four starved worms from each mesocosm. Quantitative reverse transcriptase polymerase chain reaction (Q-RT-PCR) was used to quantify gene expression. RNA samples were collected from amputated caudal tissue from three worms per mesocosm (this was the maximum number of worms per replicate that could be analyzed given the resources available in the project). Total RNA was isolated using the tri-reagent method of Chomczynski and Sacchi (24). After resuspension in nuclease-free water, RNA was

quantified spectroscopically, and first-strand synthesis was conducted using equal quantities (4 ng) of total RNA in a 40 µL reaction catalyzed by AMV reverse transcriptase (Sigma Chemicals, Poole, UK) and primed by 15-T and random hexamer oligonucleotides. For each cDNA sample, the transcripts encoding the metal binding protein metallothionein-2 (mt-2) were quantified using the primers, probe, standards, and reaction conditions detailed in Galay Burgos et al. (25). Measurements were made in triplicate for a 1 µL sample of each cDNA. Measurements of a housekeeping gene β-actin were also made. Normalization of the mt-2 data using these values, however, failed to reduce variance between individual worms or replicates. Thus, since a similar volume of total RNA was used for reverse transcription and all reactions were undertaken at the same time with reagents taken from single batches, the transcript levels are expressed simply as molecules/ng of total RNA (26). Soil Metal Measurements. Percentage loss on ignition, soil pH, and water holding capacity were measured using established procedures (27). Nitric acid extractable Cd, Cu, Pb, and Zn were measured in six replicate soil samples for all 14 gradient sites by Spurgeon and Hopkin (17). Concentrations of these metals were also analyzed in nitric acid extracts of the soil used in each mesocosm. For each mesocosm, duplicate single 1 g soil samples were digested according to Hopkin (14) and analyzed by flame atomic adsorption spectrophotometry (Varian Spectra 30, GTA 95). Analyses included appropriate spikes, blanks, and certified reference material (calcareous soil sediment BCR CRM 141R and Tort-1 Lobster hepatopancreas) to provide quality assurance for the analytical results. Statistics. Site effects on each response were investigated using the general linear model (GLM). Prior to analysis, data sets were checked for homogeneity of variance using Bartlett and Levene’s tests. Residuals resulting from the GLM were also checked for normality and independence, and, if necessary, the data were transformed until it showed a normal distribution. Where significant differences (p < 0.05) were found in the GLM, Tukey’s test was used to ascertain between which sites any differences occurred. The sensitivity of each parameter was assessed by fitting a logistic model of response at each site against the log soil Zn concentration either measured by Spurgeon and Hopkin (17) (bait lamina, community diversity, and L. rubellus population size) or in the mesocosm soil (reproduction, NRRT, and mt-2 expression). Responses were fitted against log Zn concentration, as previous work at Avonmouth has suggested that this metal is most likely to cause any toxic effects (28, 29). The approach of considering only the most toxic component of the metal mixture present is pragmatic and recognizes the difficulty in predicting the combined biological effects of metals (30). Because, however, the major contaminant metals along the transect from the smelter (Cd, Cu, Pb, and Zn) are strongly cocorrelated (13), the relationships for Zn would be consistent with those for other metals and also total soil metal load. Bait lamina feeding, community diversity, L. rubellus population size, and reproduction were all fitted against soil Zn as a proportion of the control response. For NRR-T, as times did not reach zero in any treatment, the lowest mean recorded value (for site 2 soil exposed worms) was taken as a minimum, and NRR-T was then expressed as a proportion of control retention after removal of this minimum time from both values. Log10 transformed mt-2 expression was compared as the proportional increase in expression between the baseline (mean in the control) and maximum (mean in site 2 exposed worms). Bait lamina data were fitted using measurements from each of the three replicate plots used by Filzek et al. (15). Community diversity and L. rubellus population size were fitted using the site mean value for each of the three seasonal

TABLE 1. Metal Concentrations (µg g-1 Dry Weight Soil) in Soils Collected from Three Sites under Deposition Plume of Primary Cd, Pb, and Zn Smelter Located at Avonmouth, Southwest Englanda Cd (µg g-1)

Cu (µg g-1)

Pb (µg g-1)

Zn (µg g-1)

site mean ( SD mean ( SD mean ( SD mean ( σ∂ control 0.158 ( 0.03 22.3 ( 1.9 45.7 ( 1.9 121 ( 23 12 7.56 ( 1.09 47.4 ( 6.8 271 ( 83 542 ( 80 9 6.79 ( 1.78 45.7 ( 8.7 176 ( 94 489 ( 118 7 17.1 ( 7.3 113 ( 17 1460 ( 180 734 ( 378 4 65.4 ( 5.2 204 ( 90 2280 ( 510 2669 ( 390 2 31.5 ( 5.1 221 ( 41 2120 ( 250 9769 ( 330 a All values were measured in samples taken after preparation for use in the mesocosm test and are means ((SD) of six replicate analyses.

measurements of Spurgeon and Hopkin (17) as separate replicates. Cocoon production rate, NRR-T, and mt-2 were fitted using the value determined in each replicate mesocosm.

Results and Discussion Soil Metal Concentrations. Metal measurements in the soils collected for the mesocosm study indicated that concentrations were highest at the sites closest to the factory and showed a trend of decrease with distance from the source (Table 1). A comparison of soil concentrations measured by Spurgeon and Hopkin (17) and in the mesocosm soil indicated a high correlation (Cd r2 ) 0.77; Cu r2 ) 0.98; Pb r2 ) 0.93; Zn r2 ) 0.94). Slope parameters for these regressions were less than 1 (0.67-0.76), indicating that concentrations were lower in the mesocosm soil than in the samples taken by Spurgeon and Hopkin (17). This is unlikely to be due to changes in site soil metal concentrations between the sample dates. Instead, it can be explained by the fact that the site has been contaminated by aerial deposition giving higher average metal concentrations in the surface soil (2 cm depth) analyzed by Spurgeon and Hopkin (17) than the 10 cm depth of soil collected for the mesocosm study. Organic Material Removal Feeding, Earthworm Diversity, and L. rubellus Population Size. A chi-squared test of the arcsine transformed bait lamina feeding data obtained by Filzek et al. (15) indicated significantly lower feeding at sites 7, 4, 2, and 2a (but not site 9) when compared to site 12. As these sites are the most contaminated, this indicated a probable effect of the metals present on the rate of organic material removal. Spurgeon and Hopkin (17) collected no worms at sites 1 and 2, and a lower H′ value was also recorded at sites 3-8. This reduction in diversity was due to the absence of species such as Aporrectodea caliginosa and Aporrectodea rosea that were always collected at more distant sites. L. rubellus population counts suggested it to be one of the least sensitive species, as it was present, although in highly variable numbers, at all sites except 1 and 2. Reproduction, Cellular Integrity, and mt-2 Expression. Soil temperatures in the mesocosms were low at the start of the exposure in January (regularly below 0 °C at surface), thereafter rising over the 70 day exposure period (Figure 3a). Heavy rainfall occurred during the exposure, with a total of 58 rain days (Figure 3b). The range of weather conditions encountered in the mesocosm exposure confirms the relevance of the exposure conditions to those prevailing at Avonmouth (which is located approximately 300 km directly west of the experimental site) during the months in which earthworms are most active (autumn to spring). This means that effect measurements made in the outdoor exposure can be related to those occurring in worms along the field gradient. VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5329

FIGURE 3. Soil temperature at the soil surface (a) and daily rainfall (b) measured over the 70 day duration of the mesocosm study.

TABLE 2. Percentage Survival, Juvenile Production Rate,a NRR-T,b and mt-2 Transcriptsc in L. rubellus Exposed to a Control and Five Soilsd Collected along a Transect from Primary Zinc/Lead/Cadmium Smelting Works Located at Avonmouth, Southwest Englande

site

% survival mean ( SD

reproduction (cocoon/worm/week) mean ( SD

NRR-T (min) mean ( SD

mt-2 (molecules/ng of total RNA) mean ( SD

control 12 9 7 4 2

68.8 ( 20.8a 64.5 ( 14.2a 62.5 ( 16a 52 ( 17.1a 62.5 ( 17.3a 20.8 ( 14.4b

0.211 ( 0.041a 0.298 ( 0.059a 0.333 ( 0.017a 0.277 ( 0.135a 0.197 ( 0.074a 0.004 ( 0.004b

48.7 ( 15a 50.5 ( 11a 41.3 ( 9.1ab 21.6 ( 16.4b,c 14.2 ( 9.7c 11.8 ( 6.6c

2.38 × 103 ( 8.4 × 102a 4.36 × 103 ( 3.26 × 103a 6.27 × 103 ( 3.78 × 103a 7.78 × 104 ( 9.3 × 104b 7.33 × 104 ( 6.6 × 104b 7.33 × 104 ( 6.15 × 104b

a Juveniles/worm/week. b Mean of four replicates, each itself being the mean of three worms. c Mean of four replicates, each itself being the mean of three worms. d Sites 12, 9, 7, 4, and 2. e All values are the mean ( SD. Values not sharing the same letter are significantly different at p < 0.05.

There was a significant effect of site soil on earthworm mortality (df 5, F ) 4.44, GLM, p < 0.001), with survival significantly reduced as compared to all other treatments in the site 5 soil (Tukey’s test p < 0.05). Reproduction showed a hormesis like response (31), with the cocoon production rate higher in site 9 soil than in either site 12 or the control (Table 2). GLM indicated a significant site effect on cocoon production (df 5, F ) 11.44, p < 0.001), with reproduction significantly lower at site 5 than in the control and all other site soils (Tukey’s test p < 0.05). NRR-T in coelomocytes of worms exposed to the reference and site 12 soils approached the maximum value of 60 min for the assay (Table 2). GLM indicated a highly significant site effect on NRR-T (df 5, F ) 8.85, p < 0.001). Retention was significantly lower in worms exposed to soil collected from sites 7, 4, and 2 when compared to times in the control and site 12 soils (Tukey’s test p < 0.05). Mean mt-2 expression was higher in worms exposed to all of the smelter contaminated soils (ratio at least 2.56) when compared to reference worms. GLM indicated a significant site effect on log transformed mt-2 expression (df 5, F ) 7.9, p < 0.002), with significantly higher transcript levels in worms exposed to soil from sites 3-5 than in controls (Tukey’s test p 0.59) except L. rubellus population size (r2 ) 0.07). The poor fit for this later parameter was due to the highly heterogeneous nature of the population counts (Figure 4c), and as a result, the EC50 for Zn was assumed to lie in the range between the highest concentration at which L. rubellus was collected (5000 µg Zn g-1 at site 4) and the lowest Zn concentration at which this species was not represented within the fauna (11500 µg Zn g-1 at site 2). Comparing parameters, organic material removal was found to be inhibited at lower soil Zn concentrations (EC50 (95% CIs) ) 979 (467-2041) µg Zn g-1) than earthworm community diversity (EC50 (95% CIs) ) 1737 (1318-2291) µg Zn g-1), L. rubellus population size (EC50 ) 5000-11500 µg Zn g-1), or L. rubellus reproduction (EC50 (95% CIs) ) 3236 (1175-8912) µg Zn g-1). NRR-T (EC50 (95% CIs) ) 645 (548-758) µg Zn g-1) and mt-2 expression (EC50 (95% CIs) ) 616 (549-691) µg Zn g-1) were both more sensitive

FIGURE 4. Fitted dose-response relationships for organic material removal, earthworm diversity, L. rubellus population size, L. rubellus reproduction, coelomocyte NRR-T, and mt-2 gene expression. Responses are calculated as a percentage of the control response, except NRR-T, which is expressed as a proportion of control retention after removal of the baseline minimum retention time and the log mt-2 expression, which is plotted as the proportion increase in expression between the baseline and maximum value. Organic material removal, community diversity, and L. rubellus population size were fitted against log Zn concentrations measured by Spurgeon and Hopkin (17); L. rubellus reproduction, NRR-T, and mt-2 expression were plotted against log Zn concentrations measured in the mesocosm soils (Table 1). than any higher organization endpoint. Worm mt-2 expression showed the highest sensitivity, with the EC50 for this response being lower than for any high organization level parameter by a factor of 1.6. Significance of the Results. An underlying principle in biology is that there is a hierarchical cascade of responses to stress (3, 4). In toxicology, the response cascade concept underpins the widespread use of toxicity data for single lifecycle traits (e.g., reproduction and survival) in chemical risk assessment. Thus, it is assumed that an observation of no effect at the individual level will mean there will be no effect of the same concentration for populations, communities, and ecosystems. This is thought because (i) the tradeoff mechanisms provide compensation between changes in individual lifecycle traits and effect on population growth rate (7); (ii) the loss of sensitive species can be compensated by replacement by more tolerant species, thereby conserving diversity (37); and (iii) current evidence points to the presence of some functional redundancy within ecological communities (38). On the basis of this appropriation of the hierarchical response cascade, the current state-of-the-art method in predictive ecological risk assessment is to model no-observedeffect-concentration (NOEC) values for lifecycle responses in diverse species as a cumulative sensitivity distribution (39) and from this to calculate the concentration of a chemical predicted to affect only a small percentage of species (usually 5%) in a theoretical community (40). This estimated value (the hazardous concentration for 5% of species or HC5) can then be used as an environmental quality standard (protection threshold) in environmental policy. Despite well-known concerns regarding the scientific and ethical validity of the species sensitivity approach when based on individual traits (41, 42), the method has become widely used in ecotoxicology (43-45) and currently represents a cornerstone of chemicals policy within the European Union (46).

As well as using single species lifecycle toxicity data for predictive risk assessment, the same endpoints (survival, reproduction, and growth) can also be used to retrospectively monitor the effects of deliberate or accidental chemical release (47). Such biomonitoring studies also embrace the concept of the hierarchical response cascade. Thus, if a study to measure the effects of a chemical in water or soil indicates no effect on the lifecycle of an individual, then it is usually assumed that exposed populations and communities will also not be affected. As environmental regulators become better at identifying historic and new hot-spots of contamination, biomonitoring approaches will increasingly be used to steer environmental management and cleanup decisions (48). Because the hierarchical cascade concept has such an underpinning role for predictive and retrospective risk assessment, it is a remarkable omission that no previous study has quantified the minimum degree of protection (if any) conferred on a higher level of organization by the observation of a no-toxic-effect at the preceding level. This study is thus, to our knowledge, the first to attempt to validate this protective assumption. Assessment of the effects of toxic metal stress, known to be derived primarily from the effects of Zn exposure (28, 29), on earthworm responses from gene expression to ecological function indicated that at lower organization levels (subcellular), the predictions of the cascade are supported. Expression of the gene encoding the metal binding protein mt-2 was the most sensitive measurement, with a 50% increase in induction from baseline to maximum found at 645 µg Zn g-1. The induction of changes in gene expression was followed at slightly higher soil Zn concentrations by changes in the stability of the lysosomal membrane caused probably by the effect of increasing cellular metal (Zn, but also Cd, Cu, and Pb) concentrations. Indeed, Stu ¨ rzenbaum et al. (5, 49) have established a likely mechanistic link between mt-2 expression and the lysosomal VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5331

FIGURE 5. Revised schematic diagram of the sensitivity relationship of responses at different levels of biological organization based on the comparative analysis conducted in this study. compartment since the expressed mt-2 protein is entrapped into cadmium rich granules (cadmosomes) probably formed from modified lysosomes (50). At higher levels of organization, the relative sensitivity of the measured parameters did not accord with the predictions of the hierarchical cascade. Thus, while as expected L. rubellus reproduction was more sensitive than L. rubellus population size, both were less sensitive than earthworm diversity, and all three of these are less sensitive than organic material removal. That earthworm population size was less sensitive than reproduction points to the fact that trade off mechanisms and density dependence regulation both permit population persistence even when vital rates are impaired (7, 51). That L. rubellus population size was less sensitive than earthworm diversity can be explained by two previous findings. First, L. rubellus has been shown to be less sensitive to Zn than other soil dwelling worms (e.g., A. caliginosa and A. rosea) (52). Second, Spurgeon and Hopkin (17) found that as sensitive species were eliminated from the fauna due to the presence of high metal concentrations in the soil, they were not replaced by more tolerant species. This meant that along the contamination gradient, H′ was effectively determined by the sensitivity of species, and as one of the more tolerant worms, the population of L. rubellus was maintained at sites where species richness was reduced. Clearly, the apparent low sensitivity of reproduction and population size could have been negated if a more sensitive species than L. rubellus had been chosen for the bioassay and population analysis. The absence of a reliable supply of other field species and the molecular genetic tools needed to undertake the required analyses prevented such work in this study. For the scenario investigated, however, the comparative sensitivity of the different measured traits can be summarized by redrawing the hierarchical response as depicted in Figure 5. During the assessment, the high sensitivity of organic material removal was something of a surprise. The presence of functional redundancy within the decomposer community in soil has been indicated in previous work (53, 54), and it may have been expected that this would compensate for the loss of sensitive species. The bait lamina results, however, suggest that this is unlikely to be the case. One possible cause of the relatively high sensitivity of comminution could be that among earthworms, which have been shown to be one of the most important taxa involved in bait removal (55), the dominant species at the less contaminated sites (A. caliginosa) is also one of the most sensitive. Reduction in the abundance of this species (e.g., at sites 7 and 4) may, thus, have a strong effect on bait removal rates. A second possible mechanism for the sensitivity of organic removal could be through metal stress induced anorexia. This is a pollutant-induced depres5332

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005

sion of feeding rate that has been found in invertebrates, such as Daphnia magna, exposed either to single metals or complex effluents (56, 57). The observed effect may, thus, be due to a toxic effect of the metals on energy assimilation in soil invertebrates, meaning that changes in the rates of organic material removal may reflect a subindividual level physiological response rather than a community level functional endpoint. Implications for Risk Assessment. In this study, the effects of metal exposure (primarily Zn) on biological responses have been investigated, thereby allowing the sensitivity of the different traits to be compared. While recognizing that this comparison describes responses only for a single exposure and species combination, meaning it is not clear how generally applicable the results may be, this dissection of the hierarchical response cascade conducted can help outline initial approaches for both predictive and retrospective assessments of stressor effects. For both types of appraisal, an obvious approach would be to use only sensitive species to generate data to be used to protect populations, communities, and ecosystems. This strategy is, however, widely accepted as flawed (58) since the sensitivities of different (even closely related) species can vary for different stressors, and it will seldom be clear which species and which life stage should be tested and when. For predictive ecological risk assessment, these current results lend support to the use of the sensitivity distribution approach (39). Given that a range of species (both sensitive and tolerant) are included in the analysis, any environmental quality standard calculated from toxicity data for lifecycle traits would be expected to protect the populations, given the apparent lower sensitivity of population abundance as compared to individual traits (see Figure 4c,d). Community diversity was more sensitive than the lifecycle parameter (reproduction). This reflected the comparative insensitivity of the tested species (L. rubellus) in comparison to other field species. This demonstrates that in the species sensitivity approach, threshold values derived from single species data and likely to be protective only if toxicity data for a representative range of species are included in the assessment. Decomposition also proved more sensitive than reproduction, due possibly to metal effects on the physiology of invertebrate feeding and acquisition of energy. There is, thus, a risk that this endpoint would not be protected by environmental quality standards set using data for individual traits. An alternative approach for protecting comminution would be to base environmental quality standards on toxicity data for the feeding rate (59, 60) or, alternatively, through data collected concerning effects on digestive enzyme activity and metabolism (61). The paucity of data for such parameters, however, makes it unlikely that such analyses could be done soon. For retrospective risk assessment and monitoring, two approaches are commonly recommended. Ecological survey can be used to confirm impacts at the community level. While relevant, such appraisals have the fundamental flaw that they record changes that have already occurred and, as such, are not protective. An alternative to ecological survey is to use individual based bioassays. Again, however, for retrospective analyses, these are problematic since it will usually not be clear in each circumstance which is the most suitable (sensitive) species to use in the assessment. To counter this problem, a suite of bioassays using diverse taxa might be beneficial; however, this is both time-consuming and still provides no guarantee of long-term protection for the most sensitive species. The potential pitfalls concerning the use of ecological survey and individual bioassays for retrospective risk assessments highlighted previously confirm a fundamental limitation with these approaches; namely, that the measure-

ment endpoint (e.g., mortality, reproduction, and growth) can be only marginally more sensitive than the protection goal (conservation of the diversity and functional integrity of habitats). For retrospective assessment, a new approach is needed, and it is here that mechanistically linked low organization-level effects offer an advantage. Measuring such effects (generically termed biomarkers) in environmental policy has been widely recommended over at least the last two decades (62, 63), despite acknowledged concerns over the robustness and relevance of the approach (64). The work conducted in this study highlights, however, the potential merits (inherent sensitivity) of such approaches, even when measured in a species (L. rubellus) whose individuals and populations are comparatively insensitive to the particular stress. It also confirms that despite the perceived problems, the challenge of developing reliable biomarkers in environmentally relevant species needs to be met to provide the tools needed to monitor the cascade effects of known and emerging environmental stressors. This will allow regulatory agencies to act before and not after environmental damage has occurred.

Acknowledgments This paper was inspired by a presentation given by Jan Kammenga. Jason Weeks contributed to the initial concept, and Steve Hopkin provided detailed background information on the Avonmouth site. The experimental work was supported by a NERC Advanced Fellowship GT5-98-24-DAEC and NERC Grant NER/T/S/2002/00021.

Literature Cited (1) Depledge, M. H. Ecotoxicology: a science or a management tool? Ambio 1993, 22, 51-52. (2) Newman, M. C.; Unger, M. A. Fundamentals of Ecotoxicology, 2nd ed.; Lewis Publishers: Boca Raton, 2003. (3) Walker, C. H.; Hopkin, S. P.; Sibly, R. M.; Peakall, D. B. Principles of Ecotoxicology; Taylor & Francis: London, 1996. (4) Caswell, H. Demography meets ecotoxicology: untangling the population level effects of toxic substances. In Ecotoxicology: A Hierarchical treatment; Newman, M. C., Jagoe, C. H., Eds.; Lewis Publishers: Boca Raton, FL, 1996; pp 255-292. (5) Stu ¨ rzenbaum, S. R.; Winters, C.; Galay, M.; Morgan, A. J.; Kille, P. Metal ion trafficking in earthwormssIdentification of a cadmium-specific metallothionein. J. Biol. Chem. 2001, 276, 34013-34018. (6) DeCoen, W. M.; Janssen, C. R. A multivariate biomarker-based model predicting population-level responses of Daphnia magna. Environ. Toxicol. Chem. 2003, 22, 2195-2201. (7) Forbes, V. E.; Calow, P. Population growth rate as a basis for ecological risk assessment of toxic chemicals. Philos. Trans. R. Soc. London, Ser. B 2002, 357, 1299-1306. (8) Symstad, A. J.; Tilman, D. Diversity loss, recruitment limitation, and ecosystem functioning: lessons learned from a removal experiment. Oikos 2001, 92, 424-435. (9) Boelsterli, U. A. Mechanistic Toxicology: The Molecular Basis of How Chemicals Disrupt Biological Targets; Taylor and Francis: London, 2003. (10) Chapman, P. M.; Loehr, R. C. Relevant environmental science. Environ. Toxicol. Chem. 2003, 22, 2217-2218. (11) Van Straalen, N. M. Ecotoxicology becomes stress ecology. Environ. Sci. Technol. 2003, 37, 324A-330A. (12) Eggen, R. I. L.; Behra, R.; Burkhardt-Holm, P.; Escher, B. I.; Schwegert, N. Challenges in ecotoxicology. Environ. Sci. Technol. 2004, 38, 59A-64A. (13) Martin, M. H.; Bullock, R. J. The impact and fate of heavy metals in an oak woodland ecosystem. In Toxic Metals in Soil-Plant Systems; Ross, S. M., Ed.; John Wiley: Chichester, 1994; pp 327365. (14) Hopkin, S. P. Ecophysiology of Metals in Terrestrial Invertebrates; Elsevier Applied Science: London, 1989. (15) Filzek, P. D. B.; Spurgeon, D. J.; Broll, G.; Svendsen, C.; Hankard, P. K.; Parekh, N.; Stubberub, H.; Week, J. M. Metal effects on soil invertebrate feeding: measurements using the bait lamina method. Ecotoxicology 2004, 13, 807-817.

(16) To¨rne, E. V. Assessing feeding activities of soil living animals. Pedobiologia 1990, 34, 89-101. (17) Spurgeon, D. J.; Hopkin, S. P. Seasonal variation in the abundance, biomass, and biodiversity of earthworms in soils contaminated with metal emissions from a primary smelting works. J. Appl. Ecol. 1999, 36, 173-183. (18) Spurgeon, D. J.; Svendsen, C.; Hankard, P. K.; Toal, M. E.; McLennan, D.; Wright, J.; Walker, L.; Ainsworth, G.; Wienberg, C. L.; Fishwick, S. K. Application of Sublethal Ecotoxicological Tests for Measuring Harm in Terrestrial Ecosystems; Environment Agency: Bristol, 2004. (19) Svendsen, C.; Weeks, J. M. A simple low-cost field mesocosm for ecotoxicological studies on earthworms. Comp. Biochem. Physiol. 1997, 117C, 31-40. (20) Kula, H.; Heimbach, U. Reproducibility in soil toxicity testing. In Handbook of Soil Invertebrate Toxicity Tests; Løkke, H., Van Gestel, C. A. M., Eds.; John Wiley: Weinheim, Germany, 1998; pp 31-40. (21) Lowe, D. M.; Pipe, R. K. Contaminant induced lysosomal membrane damage in marine mussel digestive cells: An in vitro study. Aquat. Toxicol. 1994, 30, 357-365. (22) Weeks, J. M.; Svendsen, C. Neutral red retention by lysosomes from earthworm (Lumbricus rubellus) coelomocytes: A simple biomarker of exposure to soil copper. Environ. Toxicol. Chem. 1996, 15, 1801-1805. (23) Moore, M. N. Cellular and histopathological effects of a pollutant gradient introduction. Mar. Ecol. Prog. Ser. 1988, 46, 79. (24) Chomczynski, P.; Sacchi, N. Single step method of RNA isolation by acid guanidinium thiocyanate phenol chloroform extraction. Anal. Biochem. 1987, 162, 156-159. (25) Galay-Burgos, M.; Spurgeon, D. J.; Weeks, J. M.; Sturzenbaum, S. R.; Morgan, A. J.; Kille, P. Developing a new method for soil pollution monitoring using molecular genetic biomarkers. Biomarkers 2003, 8, 229-239. (26) Bustin, S. A. Quantification of mRNA using real-time reverse transcription PCR (RT-PCR): trends and problems. J. Mol. Endocrinol. 2002, 29, 23-39. (27) Kalra, Y. P.; Maynard, D. G. Methods manual for forest soil and plant analysis; Forestry Canada: Northwest Region, Northern Forestry Centre, 1991. (28) Spurgeon, D. J.; Hopkin, S. P. Extrapolation of the laboratory based OECD earthworm toxicity test to metal contaminated field sites. Ecotoxicology 1995, 4, 190-205. (29) Hopkin, S. P.; Hames, C. A. C. Zinc, among a cocktail of metal pollutants, is responsible for the absence of the terrestrial isopod Porcellio scaber from the vicinity of a primary smelting works. Ecotoxicology 1994, 2, 68-78. (30) Jonker, M. J.; Sweijen, R. A. J. C.; Kammenga, J. E. Toxicity of simple mixtures to the nematode Caenorhabditis elegans in relation to soil sorption. Environ. Toxicol. Chem. 2004, 23, 480488. (31) Calabrese, E. J.; Baldwin, L. A. Hormesis: U-shaped dose responses and their centrality in toxicology. Trends Pharmacol. Sci. 2001, 22, 285-291. (32) Sterenborg, I.; Vork, N. A.; Verkade, S. K.; van Gestel, C. A. M.; van Straalen, N. M. Dietary zinc reduces uptake but not metallothionein binding and elimination of cadmium in the springtail, Orchesella cincta. Environ. Toxicol. Chem. 2003, 22, 1167-1171. (33) Dallinger, R.; Chabicovsky, M.; Lagg, B.; Schipflinger, R.; Weirich, H. G.; Berger, B. Isoform-specific quantification of metallothionein in the terrestrial gastropod Helix pomatia. II. A differential biomarker approach under laboratory and field conditions. Environ. Toxicol. Chem. 2004, 23, 902-910. (34) Dallinger, R.; Chabicovsky, M.; Berger, B. Isoform-specific quantification of metallothionein in the terrestrial gastropod Helix pomatia I. Molecular, biochemical, and methodical background. Environ. Toxicol. Chem. 2004, 23, 890-901. (35) Shimada, H.; Tominaga, N.; Kohra, S.; Ishibashi, H.; Mitsui, Y.; Ura, K.; Arizono, K. Metallothionein gene expression in the larvae of Caenorhabditis elegans is a potential biomarker for cadmium and mercury. Trace Elem. Electrol. 2003, 20, 240-243. (36) Spurgeon, D. J.; Stu ¨ rzenbaum, S. R.; Svendsen, C.; Hankard, P. K.; Morgan, A. J.; Weeks, J. M.; Kille, P. Toxicological, cellular, and gene expression responses in earthworms exposed to copper and cadmium. Comp. Biochem. Physiol. C 2004, 138, 11-21. (37) Hågvar, S.; Abrahamsen, G. Microarthropoda and Encytraeidae (Oligochaeta) in naturally lead contaminated soil: a gradient study. Environ. Entomol. 1990, 19, 1263-1277. (38) Hector, A.; Schmid, B.; Beierkuhnlein, C.; Caldeira, M. C.; Diemer, M.; Dimitrakopoulos, P. G.; Finn, J. A.; Freitas, H.; Giller, P. S.; Good, J.; Harris, R.; Hogberg, P.; Huss-Danell, K.; Joshi, J.; VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5333

(39) (40) (41) (42) (43)

(44)

(45) (46) (47) (48) (49) (50)

(51)

(52)

Jumpponen, A.; Korner, C.; Leadley, P. W.; Loreau, M.; Minns, A.; Mulder, C. P. H.; O’Donovan, G.; Otway, S. J.; Pereira, J. S.; Prinz, A.; Read, D. J.; Scherer-Lorenzen, M.; Schulze, E. D.; Siamantziouras, A. S. D.; Spehn, E. M.; Terry, A. C.; Troumbis, A. Y.; Woodward, F. I.; Yachi, S.; Lawton, J. H. Plant diversity and productivity experiments in European grasslands. Science 1999, 286, 1123-1127. Posthuma, L.; Traas, T. P.; Suter, G. W. The Use of Species Sensitivity Distributions in Ecotoxicology; Lewis Publishers: Boca Raton, FL, 2001. Van Straalen, N. M.; Denneman, C. A. J. Ecotoxicological evaluation of soil quality criteria. Ecotoxicol. Environ. Saf. 1989, 18, 241-251. Forbes, V. E.; Calow, P. Species sensitivity distributions revisited: A critical appraisal. Hum. Ecol. Risk Assess. 2002, 8, 473492. Hopkin, S. P. Ecological implications of 95% protection levels for metals in soils. Oikos 1993, 66, 137-141. Fisher, D. J.; Burton, D. T. Comparison of two U.S. environmental protection agency species sensitivity distribution methods for calculating ecological risk criteria. Hum. Ecol. Risk Assess. 2003, 9, 675-690. Altenburger, R.; Schmitt Jansen, M. Predicting toxic effects of contaminants in ecosystems using single species investigations. In Bioindicators & Biomonitors: Principles, Concepts, and Applications; Markert, B. A., Breure, A. M., Zechmeister, H. G., Eds.; Elsevier Science: Amsterdam, 2003; pp 153-198. Lofts, S.; Spurgeon, D. J.; Svendsen, C.; Tipping, E. Deriving soil critical limits for Cu, Zn, Cd, and Pb: a method based on free ion concentrations. Environ. Sci. Technol. 2004, 38, 3623-3621. European Commission Guidance Document on Terrestrial Ecotoxicology under Council Directive 91/414/EEC; European Commission: Brussels, 2002. Sunahara, G. I.; Renoux, A. Y.; Thellen, C.; Gaudet, C. L.; Pilon, A. Environmental Analysis of Contaminated Sites, 465th ed.; John Wiley & Sons: Chichester, 2002. Whitfield, J. Vital signs. Nature 2001, 411, 989-990. Sturzenbaum, S. R.; Georgiev, O.; Morgan, A. J.; Kille, P. Cadmium detoxification in earthworms: From genes to cells. Environ. Sci. Technol. 2004, 38, 6283-6289. Morgan, A. J.; Sturzenbaum, S. R.; Winters, C.; Grime, G. W.; Abd Aziz, N. A.; Kille, P. Differential metallothionein expression in earthworm (Lumbricus rubellus) tissues. Ecotoxicol. Environ. Saf. 2004, 57, 11-19. Van Straalen, N. M.; Kammenga, J. E. Assessment of ecotoxicity at the population level using demographic parameters. In Ecotoxicology; Schu ¨u ¨ rmann, G., Markert, B., Eds.; John Wiley & Spektrum Akademischer Verlag: London, 1998; pp 621-644. Spurgeon, D. J.; Svendsen, C.; Rimmer, V. R.; Hopkin, S. P.; Weeks, J. M. Relative sensitivity of lifecycle and biomarker responses in four earthworm species exposed to zinc. Environ. Toxicol. Chem. 2000, 19, 1800-1808.

5334

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005

(53) Setala, H.; McLean, M. A. Decomposition rate of organic substrates in relation to the species diversity of soil saprophytic fungi. Oecologia 2004, 139, 98-107. (54) Beare, M. H.; Coleman, D. C.; Crossley, D. A.; Hendrix, P. F.; Odum, E. P. A hierarchical approach to evaluating the significance of soil biodiversity to biogeochemical cycling. Plant Soil 1995, 170, 5-22. (55) Van Gestel, C. A. M.; Kruidenier, M.; Berg, M. P. Suitability of wheat straw decomposition, cotton strip degradation, and bait lamina feeding tests to determine soil invertebrate activity. Biol. Fertil. Soils 2003, 37, 115-123. (56) Barata, C.; Baird, D. J.; Markich, S. J. Comparing metal toxicity among Daphnia magna clones: An approach using concentration-time-response surfaces. Arch. Environ. Contam. Toxicol. 1999, 37, 326-331. (57) McWilliam, R. A.; Baird, D. J. Application of postexposure feeding depression bioassays with Daphnia magna for assessment of toxic effluents in rivers. Environ. Toxicol. Chem. 2002, 21, 14621468. (58) Cairns, J. J. The myth of the most sensitive species. Bioscience 1986, 36, 670-672. (59) Maltby, L.; Naylor, C. Preliminary observations of the ecological relevance of the Gammarus scope for growth assay: effect of zinc on reproduction. Funct. Ecol. 1990, 4, 393-397. (60) Slijkerman, D. M. E.; Baird, D. J.; Conrad, A.; Jak, R. G.; van Straalen, N. M. Assessing structural and functional plankton responses to carbendazim toxicity. Environ. Toxicol. Chem. 2004, 23, 455-462. (61) De Coen, W. M.; Janssen, C. R. The missing biomarker link: Relationships between effects on the cellular energy allocation biomarker of toxicant-stressed Daphnia magna and corresponding population characteristics. Environ. Toxicol. Chem. 2003, 22, 1632-1641. (62) Peakall, D. Animal Biomarkers and Pollution Indicators; Chapman and Hall: London, 1992. (63) Triebskorn, R.; Adam, S.; Behrens, A.; Beier, S.; Bohmer, J.; Braunbeck, T.; Casper, H.; Dietze, U.; Gernhofer, M.; Honnen, W.; Kohler, H. R.; Korner, W.; Konradt, J.; Lehmann, R.; Luckenbach, T.; Oberemm, A.; Schwaiger, J.; Segner, H.; Strmac, M.; Schuurmann, G.; Siligato, S.; Traunspurger, W. Establishing causality between pollution and effects at different levels of biological organization: The VALIMAR project. Hum. Ecol. Risk Assess. 2003, 9, 171-194. (64) Lam, P. K. S.; Gray, J. S. The use of biomarkers in environmental monitoring programs. Mar. Pollut. Bull. 2003, 46, 182-186.

Received for review January 6, 2005. Revised manuscript received April 29, 2005. Accepted April 29, 2005. ES050033K