Environmental risk assessments are generally performed for either terrestrial or aquatic systems, while these systems sometimes exist in close proximity. The objective of this study is to compare environmental risks along gradients from aquatic to terrestrial conditions. The assessment involved chemical analysis (including bioavailable fractions), as well as bioassays and bioaccumulation experiments using aquatic and terrestrial organisms. The results demonstrate that sediments and soils from neighbouring aquatic and terrestrial systems may render different assessments in terms of environmental risks. Metal availability for oligochaetes appeared to be limited in the aquatic environment as compared to the terrestrial environment, while the reverse was observed for organic contaminants. This paper aims to illustrate the use of various assessment techniques within a framework to compare ecological risks in aquatic and terrestrial environments. The obtained results are useful when considering (a prioritisation of) remedial actions.
In tidal areas and river flood plains, where soil originates from continuous sedimentation and regular flooding still occurs, the sources of contamination in the semi-aquatic and terrestrial sediments and soils are similar to those in adjacent aquatic sediments. Decision-making on possible remedial actions is improved by a good understanding of ecological risks associated with the contamination. Such assessments should cover both aquatic as well as terrestrial aspects and should make a direct comparison between these two parts of the ecosystem as much as possible. Only on the basis of such integrated assessment remedial actions can be directed towards those aspects from which the ecosystem most strongly benefits. However, ecotoxicological and environmental risk assessment studies typically focus on either aquatic or terrestrial systems (Chapman, 1986; Gaudet et al., 1995; Hill et al., 1993; Klerks and Weis, 1987; Posthuma and Van Straalen, 1993; Tarradellas et al., 1997).
A scheme for such an integrated assessment of aquatic sediments, flood plains as well as terrestrial ecosystems is therefore needed, focusing on specific characteristics. For example, indicators and assessment methods for ecological integrity of semi-aquatic terrestrial environments are required (Innis et al., 2000). In addition, challenging aspects refer to the choice of test organisms since none of the available standard test organisms for whole sediment or terrestrial assays can be used on all samples along an aquatic—terrestrial gradient without a severe adjustment of the samples (for example water content). On the other hand choosing more location specific test organisms might result in difficulties in comparing test results. From the chemical point of view, bioavailability is one of the main aspects. There has been an increase in knowledge concerning physical chemical parameters that influence the bioavailability in both aquatic and terrestrial environments. Many authors designed models that predict the toxicity or bioaccumulation of specific contaminants using parameters like pH, acid-volatile sulfide, organic matter content, cation exchange capacity and others (e.g., Ankley et al., 1994; Burkhard, 1998; Lock and Janssen, 2001; Park and Erstfeld, 1999). The importance of these parameters will however most likely vary along aquatic-terrestrial gradients. Besides, not only the organic matter quantity but also the quality will influence bioavailability and also this parameter will vary along the gradient.
To gain insight in the comparability of environmental risks in closely linked aquatic and terrestrial systems, a study was undertaken in a freshwater tidal area, the Sliedrechtse Biesbosch in the Rhine-Meuse Delta in the Netherlands. The delta has collected large amounts of contaminated sediments especially during the 1970s due to industrial emissions in both the Rhine and Meuse rivers. Because of seasonal high river discharges, the area is flooded once or twice each year, having caused widespread contamination of the terrestrial as well as the aquatic environment. Previous studies indicated serious environmental risks in the aquatic environment (Den Besten et al., 1995; Peeters et al., 2001), but little or no information is available concerning the terrestrial environment. With the decrease of industrial and domestic contamination over the past decades (Admiraal et al., 1993), the overall quality of newly deposited sediments in the river Rhine has improved. As the influx of contaminants has been reduced, the authorities are considering remediation of contaminated sediments in order to improve the prospects of this area as a unique wetland system. To this effect, pilot remediations have been performed in both Rhine and Meuse branches that enclose the area (Den Besten et al., 2000). Considering remedial action raises the question as to whether neighbouring soils should be included in remediation activities as well, as contamination of these soils occurred simultaneously. Besides, continued occasional flooding may cause redistribution of contaminants through erosion of terrestrial locations, thus recontamination of remediated aquatic locations may occur.
This study aimed at the development of an approach, suitable for areas where both aquatic and terrestrial systems exist in close proximity. For example, ecological field observations were left out, since it was felt that a proper assessment of this parameter in the intertidal mudflats was not yet possible. On the other hand, greater emphasis was put on bioavailability.
Materials and methods
Reinhold-Dudok van Heel and Den Besten (1999) and Peeters et al. (2001) provide an extended description of the area selected for this study. Sediments and soils were sampled in 1999 along two sites comprising gradients of aquatic to terrestrial locations: 1) shallow water, 2) shallow water with reeds, 3) an intertidal mud flat, 4) reed vegetation, and 5) a marsh forest. These five locations were each sampled at two transects, “Sneepkil” (S) and “Gat van de Hengst” (G), both located in the Sliedrechtse Biesbosch and used as replicates. Since the research intended to focus on possible differences in bioavailability along the gradients, sample locations were carefully selected, trying to keep the total concentrations of the contaminants constant along the gradient as well as between the two sites. Possible differences might therefore be more directly linked to differences in bioavailability.
In an attempt to preserve the in situ conditions in the sediments and soils as much as possible, sub-sampling for the different assays was done in the field. For the same reason, samples were not further treated (e.g. homogenisation, sieving etc.) upon arrival in the laboratory. Sample codes, descriptions of the sampling locations and some general sediment and soil characteristics are listed in Table 1. All bioassays and bioavailability measurements were started within one week after sampling.
Elementary organic carbon was determined after Houba et al. (2000). Particle size distributions were determined using the pipet method (Dutch guideline NEN 5753).
Total concentrations of Cd, Cu, Pb, Ni, Zn, Cr and As were determined in sediment/soil and biota using acid destruction followed by ICP-MS. Hg was determined with flameless atomic absorption spectrometry. Calcium chloride (CaCl2) extraction was performed after Houba et al. (2000). In short, 10 g of sediment/soil was extracted for 2 hours with a 0.01 M CaCl2 solution at a solid:liquid ratio of 1:10 w/v. After settling, an aliquot of the suspension was centrifuged, and the supernatant was filtered (0.45 μ m) and acidified prior to analysis with ICP-MS. To account for expected differences in relation to the water content of the samples, calcium chloride extractions were performed with both untreated and dried sediment/soil samples. Measurements in the untreated sediment were used to evaluate the aquatic bioassays, while measurements in the dried samples were used to evaluate the terrestrial tests.
Analyses were aimed at: PAHs (fluorene, phenanthrene, anthracene, fluoranthene, pyrene, benzo[a]anthracene, chrysene, benzo[b]fluoranthene, benzo[k] fluoranthene, benzo[a]pyrene, benzo[ghi] perylene, and indeno[1,2,3-cd]pyrene), PCBs (28, 52, 101, 138, 153 and 180) and some other chlorinated compounds (pentachlorobenzene, hexachlorobenzene, p,p′-DDD and p,p′-DDE). Ten Hulscher et al. (2003) provide a full description of the extraction of sediments, soils and biota, the Tenax extraction (Cornelissen et al., 2001) and subsequent analysis of the organic contaminants.
Bioassays and bioaccumulation tests
Selected toxicity tests were performed according to the origin of the sample, aquatic tests on sediments, and terrestrial tests on soils. Samples from the intertidal mud flats were tested with all organisms to facilitate the comparison of ecological risks. Furthermore, the Microtox assay with Vibrio fischeri and a bioaccumulation assay were performed on all sediments. Due to heavy rains in the period preceding the sampling, the water content of the soils was too high for terrestrial worms. Therefore, soils as well as sediments from the intertidal mudflats were air dried at room temperature for a maximum of four consecutive days. The soils and sediments were visually checked and when deemed sufficiently dry, collected and stored ad 4°C until the start of the experiment.
Vibrio fischeri (Microtox®)
The bacterium test was performed according to ISO 11348-3 (1998). Inhibition of bioluminescence was measured after 5, 15 and 30 minutes. Measurements were conducted in duplicate against a control of dilution medium. For each exposure time an EC20 (vol% pore water) was determined.
The water flea test is based on OECD 202 (1995). Pore water was diluted to obtain a dilution series of 100%, 56%, 32% and 10%. For each concentration, 10 test vessels were prepared, and in each one daphnid (age <24 hours) was placed. The experiment was terminated after 16 days and the intrinsic population growth rate was calculated according to Van Leeuwen et al. (1985). No Observed Effect Concentrations are reported, based on mortality and reproduction.
The chironomid larvae test was performed after Van de Guchte et al. (1993). In each vessel, 25 Chironomus riparius larvae (1st larval stage, age <24 hours) were placed. The experiment was terminated after 10 days. Larvae were counted and their biomass (dry weight) was determined per experimental unit. Average mortality, average larval development and average biomass were determined and compared with results from a reference sediment.
A collembolan reproduction test (ISO 11267, 1999) was performed with Folsomia candida. Soil (20 g wet weight) was placed in 100 ml vessels with 10 springtails (age 10–12 days) in four-fold. After four weeks the experiment was terminated by flushing out the springtails, and surviving adults and juveniles were counted. Average mortality and average number of juveniles per adult were determined and compared with results from a reference soil tested simultaneously.
Lumbricus rubellus (toxicity and bioaccumulation)
A test with the terrestrial earthworm Lumbricus rubellus was performed after ISO 11268-2 (1998). In four replicate jars, five adults (individual weight ranging from 1.1 to 1.7 g FW) were placed. After four weeks (at 15°C), surviving worms were collected and transferred into Petri dishes with wetted filter paper to defecate for 48 hours. These animals were divided in two, pooled and analysed for metals and organic contaminants. Cocoons were collected by sieving, transferred into Petri dishes with wetted filter paper and checked for hatching. Mortality, total and average individual biomass and reproduction were determined per glass jar.
Bioaccumulation with aquatic oligochaetes
For each sample two sediment water systems were started in a glass aquarium, each containing 1 Litre sediment and 4 Litres artificial freshwater medium DSW. To each jar 40 g (FW) of oligochaetes (Limnodrilus species) were added. The animals were fed with 2 ml of a 10% (w/v) Aquarian solution. Throughout the experiment, the aquaria were kept in the dark, at 20 ± 2°C. The test was terminated after 28 days. The collected oligochaetes were subdivided for further analysis. Retrieval of the oligochaetes was difficult since non-sieved sediment was used, resulting in relatively low recoveries (ca. 35%). A separate sample of uncontaminated oligochaetes, used to start the experiments, was analysed as a control.
Physical sediment characteristics
The water content (Table 1) of all soil and sediment samples was relatively similar due to rainfall preceding the day of sampling. Carbon content in the sediments increased by a factor of two towards the terrestrial locations along the gradient. In addition, particle size analyses differed between aquatic and terrestrial samples. The particle fraction >50 μ m was larger in the aquatic samples, which may be related to the result of lower net sedimentation rates of smaller particles, an increased erosion or both.
The total concentrations of the different metals are given on a dry weight basis (Table 2), while results of the CaCl2-extraction are given as a fraction (%0) of the total concentration (Figure 1). As desired, total metal concentrations were fairly constant along the aquatic-terrestrial gradients, although especially in two terrestrial samples (G4 and G5) higher concentrations were found (factor 2). The possible influence of the total metal concentration on the available fraction will consequently be small. In most cases CaCl2-extractable fractions obtained from the original field collected sediment were lower than fractions in the dried samples. Lead and chromium were exceptions, showing no significant relationships between the two extraction procedures, whilst these relations were only weak for arsenic. In addition, the CaCl2-extractable fractions of these three metals did not vary along the gradients from shallow water to marsh forest (data not shown). The CaCl2-extractable fractions of cadmium, zinc, copper, and nickel on the other hand clearly increased along the gradients. Furthermore, the dried samples yielded higher extractable fractions compared to the field collected samples (illustrated in Figure 1 for the mud flat samples).
The concentrations of organic contaminants in the sediment and soil samples are given on a dry weight basis (Table 2), whilst results of the Tenax-extraction are given as a fraction (%) of the total concentration (Figure 1). As desired, no significant differences were found in the total concentrations of PAHs, PCBs, DDE and DDD along the gradients from aquatic to terrestrial locations. Significant differences were however found after correction for the organic carbon content, since the terrestrial locations were characterized by significant higher organic carbon contents (OC normalized data not shown, but organic carbon contents are presented in table 1). Pentachloro-and hexachlorobenzene concentrations (dry wt) increased somewhat at terrestrial locations, but the concentrations were low and differences were not statistical significant. In contrast to the total concentrations, statistically significant differences were found in the bioavailable fraction as measured by the Tenax fraction. The Tenax extractable fraction of PAHs, PCBs, DDE and DDD decreased at terrestrial locations as carbon contents increased (Figure 1). Pentachlorobenzene was not detected in the Tenax extracts. The Tenax extractable fraction of hexachlorobenzene did not show any clear trend and varied somewhat between the two sites as well as the different sampling locations (data not shown).
The results of the bioassays are shown in Table 3. With three exceptions, no toxic effects were demonstrated. Results are therefore only presented using average values and standard deviations or 95% c.l. are left out to shorten the overview. Toxic effects were demonstrated in two tests, Vibrio fischeri and Chironomus riparius. With Vibrio fischeri, effects were shown in two aquatic sediments (S1 and S2) from the Sneepkil site, while the Chironomus riparius test revealed an effect with Sneepkil sediment from the intertidal mud flat (S3). The biomass of L3 and L4 larvae was significantly reduced compared to those in the control sediment. These effects were classified as ‘moderate’.
Concentrations of metals and organic contaminants in Limnodrilus sp. and Lumbricus rubellus are shown in Table 4. Metal concentrations are calculated on a dry weight basis, organic contaminant on a lipid weight basis.
Bioaccumulation of metals
In Limnodrilus sp. very little metal accumulation occurred, probably partly due to relatively high metal concentration in the oligochaetes at the start of the exposure. The exception to this was mercury, which at least doubled in concentration after exposure to Sneepkil sediment and increased by a factor of 5 to 14 in oligochaetes exposed to samples from Gat van de Hengst. In Lumbricus rubellus on the other hand, a distinct accumulation was observed for all metals, except zinc. The zinc concentration appeared to be regulated as it ranged within a narrow bandwidth of 1100 to 1200 mg kg−1 (97–106% of control), despite the variation in zinc concentrations in the soil. Besides these differences between the two species, no significant differences in metal accumulation were observed within either Limnodrilus sp. exposed to sediment or Lumbricus rubellus in soil.
Bioaccumulation of organic contaminants
In Limnodrilus sp. bioaccumulation of organic contaminants clearly occurred. Only the concentrations of o,p′-DDD seemed to remain the same, regardless of the location. The Σ PAH concentrations were between 1.5 and 2.7 times higher compared to the control samples at t = 0. Bioaccumulation was the highest in the aquatic sediments (S1 and G1), and the lowest at the intertidal mud flats (S3 and G3). The Σ PCB concentrations had also increased compared to the control, but no statistical significant differences were found along the gradient. For hexachlorobenzene, the concentrations were generally increased (except at aquatic sediment, S1) and, in contrast to the Σ PAH, hexachlorobenzene bioaccumulation increased, as the samples were less aquatic (but differences were not statistical different). For p,p′-DDD the concentrations increased by a factor of about seven, regardless of the origin of the samples along the gradients.
For Lumbricus rubellus a different pattern emerged, since both Σ PAH and Σ PCB concentrations were highest in the control samples at t = 0. The low background levels in Lumbricus rubellus (as compared to Limnodrilus sp) remained the same or even decreased when exposed to samples from the intertidal mud flats. The concentrations of hexachlorobenzene increased, though no statistical differences were found. Dieldrin was also detected in the exposed animals but like hexachlorobenzene, the concentrations were low and no significant differences were detected.
The results from the chemical analyses and the bioassays were in line with a previous study in this area (Den Besten et al., 2000), concluding that indirect effects caused by secondary poisoning were more important than direct effects posed upon sediment and soil fauna. Only few negative effects were observed in the bioassays. As compared to direct effects, more distinct differences in relation to the gradients were observed in the bioavailability and the accumulation of the contaminants between the aquatic and terrestrial samples. The assessment of the ecological risks was therefore narrowed down to these findings.
Location specific assessment
A clear difference was observed in the bioaccumulation of metals between aquatic and terrestrial assays. With Limnodrilus sp. very little accumulation of metals occurred, whereas Lumbricus rubellus accumulated significant amounts of metals. This difference was especially clear in the intertidal mud flats, which were tested using both organisms. Even though metal natural background levels were higher in aquatic Limnodrilus sp., the terrestrial Lumbricus rubellus had accumulated metals at the end of the exposure period to levels above those found in Limnodrilus sp. This might partly be due to species-specific differences towards metal accumulation (Hopkin, 1989; Miesbauer et al., 2001) but bioaccumulation differences can nonetheless also be expected due to differences in metal availability between aquatic and terrestrial sediments. For instance, the higher CaCl2-fraction of Cd in the terrestrial samples corresponds with an increased bioaccumulation of Cd in Lumbricus rubellus. An improvement in the correlation between cadmium concentrations in soil samples versus internal cadmium concentrations in the organisms is noted if CaCl2-extractable fractions are used instead off total cadmium concentrations (Figure 2).
The bioaccumulation patterns of organic contaminants (table 4) were reversed to those observed for metals. Concentrations of PAHs and PCBs in Limnodrilus sp. increased during the exposure period, while remaining constant in Lumbricus rubellus. A further illustration of this pattern was found in a direct comparison between the two species exposed to samples from the intertidal mud flats. The background levels of PAHs and PCBs in Lumbricus rubellus remained low, or even decreased further, while these levels in Limnodrilus sp. increased during exposure to the same sediments. The levels of hexachlorobenzene and dieldrin increased also, but the concentrations were low, making it difficult to describe accumulation trends accurately. As illustrated in Figure 2, a large portion of the variation in internal PCB concentrations could not be explained by standardizing concentrations in sediment and soil on organic carbon content. Besides total carbon content, the quality and constitution of the organic matter will also influence the availability of organic contaminants (Garbarini and Lion, 1986; Gunnarson et al., 1999; Rutherford et al., 1992). Especially in the case of aquatic–terrestrial gradients, this may play an important role, since for example the source of the organic matter in sediment strongly differs from that in the soil. Bioavailability measurements, like Tenax extractions, might therefore show an improved correlation with internal PCB concentrations (Figure 2). From this figure, it is clear that a good correlation existed between accumulated PCB and Tenax extractable fractions. This relationship based on Tenax extractable fractions is further described and discussed by Ten Hulscher et al. (2003). In view of these findings, Tenax extraction appeared to be a proficient tool for the assessment of bioavailability of organic contaminants in sediments and soils. Still, a species-specific difference seemed to be present since the relation between the Tenax extractable PCB fraction and the internal concentration in the organisms (mg kg−1 fat) differed between Limnodrilus sp. and Lumbricus rubellus. Limnodrilus sp. seemed to accumulate PCB to a higher extent as compared to Lumbricus rubellus. As indicated above, species-specific differences might be responsible for this difference (e.g. Goerke and Weber, 2001; Schuler et al., 2003). Another explanation may also be possible. In contrast to CaCl2 extractions, Tenax extractable fractions were only analyzed in field collected samples. The terrestrial samples, which were additionally dried prior to the start of the experiments, were not analyzed after drying. As the availability of organic contaminants in terrestrial samples was already low compared to aquatic samples, additional drying might have further decreased availability. Tenax extractable PCB-fractions in terrestrial soil samples (Figure 1) were on average 50% lower compared to the fractions in aquatic sediments. A preliminary correction for this percentage rendered a relationship between Tenax extractable fractions and internal concentrations, which was quite comparable between the two species.
In contrast to the differences between aquatic and terrestrial locations as discussed above, differences among the results obtained by the aquatic bioassays and bioaccumulation experiments on the three different locations (shallow water—intertidal mud flat) were of minor relevance. The same applies to the results obtained by the terrestrial experiments on the three different locations (intertidal mud flat—marsh forest). It is therefore concluded that it does not seem necessary to make distinction between all locations along a gradient. A distinction between either aquatic or terrestrial locations may be sufficient for ecological risks assessments. Intertidal mudflats, being alternately aquatic and terrestrial, resembled more closely aquatic locations based on chemical measurements immediately after sampling (CaCl2 or Tenax). This was probably correlated to the high water content at the time of sampling. However, the same locations seemed to adapt quite fast to changing circumstances. Cadmium, copper and zinc concentrations sharply increased in CaCl2 extractions following the drying of samples (Den Besten, 2003; Zhang et al., 2001). Furthermore, when samples from the intertidal mudflats were tested in bioassays with aquatic oligochaetes internal concentrations did not significantly differ from the shallow water and shallow water with reed locations, while the same was concluded when tested with earthworms for the two terrestrial locations (reed vegetation and marsh forest).
This study aimed at testing a scheme for an integrated assessment of aquatic sediments, flood plains as well as terrestrial ecosystems. Considering the results, it is clear that sediments and soils from neighbouring aquatic and terrestrial systems may render different assessments in terms of environmental risks. Direct toxic effects could only be demonstrated in aquatic samples, and appeared to be moderate. The differences in environmental risks of aquatic sediments and terrestrial soil became more apparent when comparing bioaccumulation patterns. In aquatic environments, metal availabilities for oligochaetes appeared to be limited, whereas organic contaminants (PAHs and PCBs) were bioavailable. In the terrestrial environment the opposite was observed: metals were more bioavailable and organic contaminants were less bioavailable. The use of CaCl2-and Tenax-extraction to describe availability have led to a better understanding of these differences along the gradient. Consequently, such a comparison of environmental risks associated with contaminated sediments and soils along a gradient of aquatic and terrestrial conditions provides useful information when considering (prioritisation of) remedial actions in areas where both aquatic and terrestrial systems exist in close proximity.
This work was financed by the Directorate of Zuid-Holland, Rijkswaterstaat, The Ministry of Transport, Public works and Water management, The Netherlands. The authors would like to thank all persons involved in the sampling expedition. Jos Bodt and Joost van der Pol (Alterra) are kindly acknowledged for assistance on the earthworm assays; Gerda Hopman-Ubbels and Martin van Velzen (Inst. Environmental Studies, The Netherlands) for performing the organic analyses, and Jan Willem Wegener (Inst. Environmental Studies, The Netherlands) for technical advice.