PII:
S0269-7491
(97)00
i 04-8
Environmental Pollution, Vol. 98, No. 1, pp. 7~80, 1997 © 1998 Elsevier Science Ltd. All fights reserved Printed in Great Britain 0269-7491/97 $17.00 + 0.00
ELSEVIER
DEVELOPMENT OF ZINC BIOAVAILABILITY A N D TOXICITY FOR THE SPRINGTAIL FOLSOMIA CANDIDA IN A N EXPERIMENTALLY C O N T A M I N A T E D FIELD PLOT
C. E. Smit, a* P. v a n Beelen b a n d C. A. M . V a n G e s t e l a
aDepartment of Ecology and Ecotoxicology, Vrije Universiteit, De Boelelaan 1087, 1081 HV Amsterdam, The Netherlands bLaboratoryfor Ecotoxicoiogy, National Institute of Public Health and the Environment, PO Box 1, 3720 BA Bilthoven, The Netherlands (Received 11 April 1997; accepted 21 July 1997)
Abstract The influence of outdoor exposure conditions and ageing of contamination on the toxicity of zinc was investigated for the springtaU Folsomia candida to evaluate the validity of a standardised soil toxicity test. In three successive years, animals were incubated in an experimentally contaminated field plot. During the first months after construction of the test field, total zinc concentrations of the soil decreased rapidly due to leaching of excess zinc with rainwater, while increased sorption of the remaining residues resulted in a reduced bioavailability of the metal. Although variation between replicates was substantial, the ECsos for the effect of zinc on reproduction of F. candida determined in the field experiments differed by less than a factor of two from effect concentrations obtained in laboratory tests in which the same soil was used. Expression of the ECsos on the basis of water soluble zinc allowed for a comparison with effect concentrations estimated for other soil types. ECsos were comparable with literature data, which indicates that bioavailability of zinc is the main factor determining toxicity for F. candida. It is concluded that laboratory based toxicity data are suitable to predict effects of zinc for F. candida under outdoor conditions, provided that the bioavailability of zinc is determined accurately using water soluble concentrations. © 1998 Elsevier Science Ltd. All rights reserved
1997). Laboratory tests are suited to investigations of cause effect relationships, and results of these studies improve insight into mechanisms of toxicity. Since essential ecosystem characteristics are not incorporated in single species toxicity tests, an extrapolation of laboratory results to the field situation seems hard to make (Cairns, 1984; Van Straalen et al., 1994). Some characteristics of laboratory tests, for instance the relatively short incubation time of the test substances, can result in an overestimation of toxicity. Others, such as the absence of secondary stress, may lead to decreased toxicity compared with field situations. This implies that no definite answer can be given regarding whether a laboratory to field extrapolation factor should be greater or smaller than one (Van Straalen and Denneman, 1989). To improve the ecological relevance of terrestrial toxicity tests, a number of model ecosystems have been used to evaluate the effects of chemicals on soil invertebrates under more natural conditions (Verhoef and Van Gestel, 1995; Sheppard, 1997). In addition, field tests can be used to validate the outcomes of such tests. Terrestrial field studies are restricted mainly to the application of pesticides to experimental plots, after which the effects on endemic soil fauna are recorded (Eijsackers and Van de Bund, 1980; Edwards and Bohlen, 1992). Other types of field tests involve the in-situ exposure of test organisms to contaminated soil or sediment using caged indicator species (Callahan et al., 1991; Menzie et al., 1992; Burton et al., 1996). The results of these tests are variable because of differences in soil type and experimental methods. Furthermore, field soils are often polluted with various or unknown contaminants and the absence of an uncontaminated reference substrate with the same soil characteristics hampers the interpretation of test results. The present study was conducted within the framework of a research project which was started to evaluate the validity of laboratory tests to predict effects of zinc in the field situation. As an intermediate between laboratory and field, an experimental field plot was constructed to investigate the toxicity of zinc for the springtail Folsomia candida in artificially contaminated soil under outdoor conditions. Experiments
Keywords: Toxicity, zinc, field test, Collembola.
INTRODUCTION During the past years methods have been developed to predict the potential hazard of chemicals for terrestrial ecosystems. Most risk assessment procedures are based on laboratory toxicity tests which, for reasons of cost effectiveness and international standardisation, are straightforward and relatively simple (Van Gestel et al., *To whom correspondence should be addressed. Centre for Substances and Risk Assessment, National Institute of Public Health and the Environment, PO Box 1, 3720 BA Bilthoven, The Netherlands. Fax: +31-30-274 4401; e-mail: ce.smit@ rivm.nl 73
74
C . E . Smit et al.
were repeated in three successive years during which the zinc concentrations in the soil were monitored. The results were used to evaluate the validity of the standardised F. candida soil toxicity test (ISO, 1994) under field conditions and to gain insight into the development of contamination through time.
MATERIALS AND METHODS Construction of the field plots and contamination of the test soil Six plots were constructed at the university campus of the Vrije Universiteit, based on Faber and Verhoef (1991) (Fig. 1). The plots, indicated A to F, were divided in 10 equal compartments of l x 0 . 5 m (lxw) using stainless steel plates of l x0.5m (wxh) which were driven about 10cm deep in the sand. A layer of sharp sand on top of a waterproof foil was placed under the plots. A drainage system was installed to control soil moisture content. The soil was collected at Panheel, Limburg, The Netherlands in February 1994. About 8m 3 soil was sieved (5ram), homogenised and dried to a moisture content of 6% on a dry weight basis. Soil was contaminated by subsequently mixing 30 kg lots of soil with 9% sharp sand and 21itres of the appropriate ZnC12 (Merck, rein, > 95% pure) solution to achieve a desired concentration range of 0, 32, 56, 100, 180, 320, 560, 1000, 1800 and 3200/zg Zn g-i dry soil at a soil moisture content of 13% on a dry weight basis. Zinc concentrations were assigned to each of the compartments following a completely randomised block design (Sokal and Rohlf, 1995), leading to six replicated series of the complete concentration range; 130 kg of contaminated soil was added to every test compartment, which was equal to a 30-crn deep soil layer. Monitoring of zinc concentrations Soil samples were collected at regular time intervals to determine soil pH-KCI and soil zinc concentrations. The first sampling of plots A - F took place directly after construction of the field plot in July 1994. Thereafter, soil samples were taken from plots B-F in February and October 1995 and May 1996. The total zinc fraction was obtained by digestion of 500 mg dry soil in 5 ml concentrated HNO3. CaC12 exchangeable and water soluble zinc were obtained by shaking 10g of moist soil with 100ml of a 0.01 M CaCI2 solution or 100ml of double distilled water. The extracts were filtered through a
SP~
~ .....
0 l .
,
S
m
]
~
'
Fig. 1. Enclosure plot (5x 1m), divided into 10 compartments, built of stainless steel plates (SP). S = sand; CS = contaminated soil.
0.45-/zm filter and acidified by addition of 0.7% HNO3. Zinc concentrations in the extracts were determined by ICP-AES and expressed on the basis of dry soil. During the first rainfall after the construction of the test field, samples were taken from the drainage system to determine the time that zinc leaching occurred. Experimental design Toxicity tests were performed in the test field plots during the late summers of 1994, 1995 and 1996, respectively. For the first experiment, which was started at the end of August 1994, soil cores (10cm) were sampled from the compartments and put into 14cm high plastic tubes (13 6 cm) with a gauze bottom (0.6 mm). Six replicate test containers were placed into the respective compartments of five plots (B-F), leading to 300 test containers in total. An extra series of test containers was incubated in the control soils to determine the time offspring were produced and the test could be terminated. The test containers were covered by a gauze lid. The experiment was started with juvenile F. candida of the same age (10--12 days old) and fresh weight (8.7 + 3.6/zg; ± SD) which were obtained by synchronising the egg laying of a laboratory breeding stock (Smit and Van Gestel, 1996). At the start of the experiment, 10 individuals were transferred to each of the test containers and about 10mg dried baker's yeast (Oetker B.V., Veenendaal, The Netherlands) was added for food. Food was supplied weekly when needed. After eight weeks of incubation, the test containers were sacrificed to determine survival and reproduction according to Smit and Van Gestel (1996). A second experiment was started in the first week of September 1995. Based on the results of the first experiment, smaller test containers (7 cm high, 13 6 cm) were used which were filled with a 5-cm deep soil layer. Four replicate test containers were placed in the compartments of four plots (C-F). Juvenile F. candida (1012 days old; 17.1 ±4.1 tzg) were transferred to the test containers of three plots (C-E) as described above. Dried baker's yeast was added for food when needed. After five weeks of exposure, offspring were observed in the controls and survival and reproduction were determined. In the fourth plot (F), adult animals with an average fresh weight of 111.5± 18.7/zg (+SD; n= 10) were placed into the test containers. Incubation of this test series was terminated when offspring were observed after four weeks of exposure. The last experiment was started in the second week of September 1996. For this experiment, test containers with a gauze bottom of 0.45 mm were used. Moist soil was collected from the respective test compartments of four plots (B-E) and 4-mm sieved. Test containers were filled in duplicate with a 5-cm deep layer of 4-ram sieved moist soil for every test compartment, and heated in a microwave oven for 2m in at 600W (65°C) to kill unwanted animals (Kammenga et aL, 1996). At the start of the experiment, 10 adult F. candida (76.1 4-15.4#g ( ± S D ; n = 10) were transferred into the test containers and about 10mg dried baker's yeast was added for
Development of zinc bioavailability and toxicity food. After five weeks of incubation, test containers of plot C, D and E were sacrificed to determine survival and reproduction, the last series (plot B) was incubated for six weeks. Ambient temperature was monitored continuously during all the toxicity tests using a thermohygrograph placed at 25 cm above soil level. The moisture content of the test field soil was kept constant at field capacity by the drainage system.
75
4000
"6 3000
"~2000 0 C
N 1000
Caleulatiom and Jcmfisties Sorption of zinc to the test field soil was described using a Freundlich isotherm:
i
0
i
1000
i
i
2000
I
i
3000
4000
nominal zinc (gg/g dry soil)
Cs = Kf* --w Cl/n
(1)
where Cs is the concentration of zinc adsorbed to the soil (/zgg -1 dry soil), Cw is the zinc concentration in the water extract (/~gml-l), Kf is the Freundlich sorption constant (mlg -~) and 1In is a shape parameter of the Freundlich isotherm. Estimates for Kf and 1/n were obtained by linear regression in a double logarithmic plot of Cw versus Cs. A nested ANOVA (Sokal and Rohif, 1995) was used to test for significant differences in survival and reproduction between corresponding compartments of the enclosure plots. If no significant 'plot' effect was observed, data were pooled. SiLmificant differences in reproduction between concentrations were determined using the Tukey's test for multiple comparison of means. Data transformation was included when data were not normally distributed or when variances where not homogeneous. The non-parametric Kruskal-Wallis test was used when data transformation was not possible. The tests mentioned are incorporated in the TOXSTAT ® 3.3 software package (Gulley et al., 1990). Measured total, CaCI: exchangeable and water soluble zinc concentrations were used to estimate the 50% effect concentrations using the logistic model of Haanstra et aL (1985).
Fig. 2. Total zinc concentrations of the test field soil as a function of nominal concentrations added. Symbols indicate the average of plots A-F in July 1994 (O), and of plots B-F in February 1995 (I-1), October 1995 (A) and May 1996 (O). Error bars indicate standard deviations.
amount of zinc was recovered in October 1995 and May 1996. Figure 3 shows the pH-KCl as a function of the measured total zinc concentration. The initial pH-KCI of the Panheel soil was 5.2. At the time of first sampling in July 1994, pH in the control soil compartments was slightly raised to a maximum of 6.0 4-0.2 (average 4- SD, n=6). In February 1995, average pH (±SD) was raised to values between 6.44-0.04 and 7.1 4.0.1 (n = 5) and was negatively correlated with zinc concentration. From February 1995, only minor pH changes were detected. Figure 4 shows the CaCl2 exchangeable zinc fraction measured in the field enclosure plots as a function of the total zinc concentration. In July 1994, an average (±SD) of 40.4+ 10% of the total zinc was found in the CaC12exchangeable fraction at the highest nominal zinc concentration. Variation between the compartments 7.5
RESULTS
l ~ v ~ of tim test i~kl ~)il wire mae In July 1994, measured and nominal zinc concentrations were broadly similar (Fig. 2). The first rainfall after the construction of the test field o~urred during the first week of August 1994. Concentrations in the drainage water started to increase almost immediately and reached a maximum in September and decreased to normal levels in December 1994, indicating a substantial leaching of zinc from the soil. As a result, total zinc concentrations in soil decTeased between July 1994 and February 1995, after which they stabilised over the following two years. Concentration decrease was most pronounced at nominal concentrations higher than 560/zg Zng -I dry soil. At the highest concentration of 3200/~g Zng -I dry soil, 71 and 68% of the initial
ov. -1i
6.0 5.5 5.0
i
0
I
i
i
1000 2000 total zinc (gg/g dry soil)
i
i
3000
Fig. 3. Average pH-KCI as a function of the total zinc concentration of the test field soil measured in plots A-F in July 1994 (O), and of plots B-F in February 1995 (D), October 1995 (A) and May 1996 (O). Error bars indicate standard deviations.
76
C . E . Smit et al.
2000
25 t.)
0
"R
1500
20
%/•
~O ~
¢3.
1000
E
/"
I
I
•
15
U
~
500
998
'10 ,
i
0
I
i
I
i
2000 1000 total zinc (p.g/g dry soil)
I
3000
I
was relatively large. Average maximum CaCI2 exchangeability decreased to 23.74- 1.5 and 19.2+2.5% in February and October 1995, respectively. In May 1996, a maximum of 19.1% of the total zinc was found in the CaC12 extracts and variability between the compartments was reduced in comparison with the first sampling date. Freundlich adsorption coefficients (Kf) and corresponding 1In values are shown in Table 1 for the respective sampling dates. Sorption increased with time and sorption isotherms deviated from linearity except for the second sampling date. At the first sampling date, maximum water solubility was 1134/~g Zn g-i dry soil, corresponding with 36.8% of the total zinc concentration at the highest nominal concentration of 3200#g Zng -l dry soil• In May 1996, only 1.5% of the total amount of zinc was water soluble at this concentration. Temperature during the experiments
The average temperature measured over 24h (Fig. 5) was between 8.0 and 18.0°C during the experiment in 1994 and ranged from 11.3 to 17.7 °C in 1995. In 1996, average daily temperatures between 10.1 and 14.4°C were recorded during the experiment. During the 1994 experiment, minimum temperatures were between 6.0 Table 1. Freundlich adsorption coefficients Kfand corresponding l/n values estimated for zinc contaminated soil from experimental field plots, r2=coefficient of determination of the regression of eqn 1
October 1995 May 1996
',.
94
August
February 1995
¢ V
O) > ¢6
Fig. 4. Average CaC12 exchangeable zinc concentration as a function of the total zinc concentration of the test field soil measured in plots A-F in July 1994 (O), and of plots B-F in February 1995 (Fq), October 1995 (A) and May 1996 (<>). Error bars indicate standard deviations.
July 1994
< •
10
0
Sampling date
,.
Kf
l /n
r2
214 (179-255) 586 (521-661) 557 (442-701) 1107 (986-1245)
0.47 (0.39-0.55) 0.96 (0.87-1.05) 0.53 (0.42-0.65) 0.68 (0.63-0.73)
0.757 0.904 0.677 0.952
!
September
October
Fig. 5. Average daily temperature at 25 cm above soil level during the fieldexperimentsperformed in 1994 (--), 1995 ( - - - ) and 1996 (.-.).
and 16.5°C, whereas maximum temperatures ranged from 8.5 to 20.5°C. In 1995, temperature was generally higher with minimum temperatures ranging from 9.0 to 15.5°C and maximum temperatures between 14.0 and 22.5°C. In 1996, minimum temperatures between 7.5 and 13.5°C were recorded, whereas the maximum temperatures were between 12.0 and 18.5°C. The average temperatures calculated over the whole test period were 13.7 (1994), 15.4 (1995) and 12.4°C (1996). Effects of zinc on survival and reproduction
In 1994 and 1995, no significant plot effect on survival of F. candida was observed (nested ANOVA, p = 0.709 and 0.138) and data of all plots were pooled. In 1996, survival was significantly different between plots (p < 0.05). Control survival was low in all experiments, and variation between replicates was large, especially during the first experiment in 1994 when survival in the control replicates was between 0 and 70%. The highest average survival (+ SD) of 53.7 + 22.4% was observed at an exposure concentration of 560/zgg -1 dry soil. In 1995, average control survival was somewhat higher (33%). Incubation of adult animals in one of the plots resulted in a slightly higher survival (46%), but differences were not significant. Average survival was highest at the exposure concentration of 1800/~g Zn g-~ dry soil (58.8 4-29.3%). In 1996, when adult animals were incubated, control survival was higher and reached values between 40 and 65%. No relationship between zinc concentration and mortality was observed in any of the experiments. This was most pronounced in 1996, when average survival at the highest concentration was between 60 and 70%. In some compartments, average survival was even higher; for instance, in plot D a 95% survival was observed at exposure concentration of 1800/~g Zn g-t dry soil. No significant plot effect was observed for reproduction in 1994 (nested ANOVA, p=0.356). In 1995, nested ANOVA for plots C, D and E resulted in a p-value
Development of zinc bioavailability and toxicity of 0.372. When plot F, incubated with adults animals, was included in the analysis, the plot effect was also not significant (p = 0.513). In 1996, one out of the four plots tested showed a significant plot effect. No separate concentration response relationship could be estimated for this plot (E), and data were excluded from further calculations. The number of juveniles is shown as a function of the nominal zinc concentration in Fig. 6. During the first experiment in 1994, a low and variable control reproduction of 57 + 43 juveniles per test container (average +SD; n=30) was observed. In the next two years, juvenile numbers (average + SD) in the control soil were 189+85 (1995; n = 16) and 132+46 (1996; n=6). Apart from F. candida, other soil arthropods were observed in some of the test containers at the termination of the tests, even in 1996 when the soil was microwave heated prior to the experiment. In 1994 and 1995, a significant reduction of the number of juveniles was observed at nominal concentrations of 1800 and 3200/~g Zn g-I dry soil (Kruskal-WaUis and Tukey's test, p<0.05), whereas in 1996 reproduction was decreased significantly at 3200/zg Zng -1 dry soil (Tukey's test, p < 0.05). Since no significant plot effect was observed and 95% confidence intervals of the ECs0 values estimated for the respective enclosure plots overlapped, reproduction data were pooled to estimate a combined effect concentration. ECso values for the effect of zinc on reproduction estimated on the basis of total, CaCI: exchangeable and water soluble zinc concentrations are given in Table 2, together with results of laboratory experiments performed with soil from the same field plot. The first toxicity experiment in 1994 was carried out at the time zinc concentrations of the test field soil were still decreasing due to leaching of zinc from the soil with rainwater. Therefore, zinc concentrations which were 300
250 f 200
1
ii!
i
i::? 0
32
100 180 320 560 100018003200 nominal total zinc (t~g/g dry soil) 56
Fig. 6. Average number of juveniles produced per test container as a function of the nominal zinc concentration after incubation of F. cand/da in artificially contaminated soil during the field experiments performed in 1994 (ll), 1995 (r~) and 1996 (D). Error bars indicate standard error of the mean. Asterisks indicate significant differences from control (p < 0.05).
77
determined in soil samples taken from one of the plots in August 1994 (results not shown) and the soil analysis data of February 1995 were both used to estimate effect concentrations. It can be argued that the actual ECs0 values are lying in between the two estimates. Soil zinc concentrations determined in October 1995 and May 1996 were used to estimate effect concentrations for the experiments carried out in 1995 and 1996. Effect concentrations determined in the field experiments increased with time when based on total and CaC12 exchangeable zinc concentrations. The ECs0 based on water soluble concentrations was highest in 1994 and decreased the next year, but the confidence intervals were large and differences between experiments were not significant.
DISCUSSION Soil analysis Literature data on the adsorption of zinc on soils show a wide range of adsorption coefficients. The adsorption coefficients found in this study (Table 1) are in agreement with data of Buchter et al. (1989), who determined Freundlich adsorption coefficients of zinc in 11 natural soils and found Kf values between 2 and 774 ml g-l. In a previous study, a Kf value of 143 ml g-l with a l/n value of 0.40 was obtained for Panheel soil which was freshly contaminated with ZnC12 (Smit, 1997). The adsorption coefficient and corresponding 1/n determined in July 1994 compares very well to this value, and is in agreement with Kfs of 220 and 238mlg -1 (1/n=0.47 and 0.42) reported by Smit and Van Gestel (1996) for artificially contaminated natural soils. Except for the second sampling date, the 1/n values found in the present study differed from unity, indicating that adsorption was concentration dependent. This effect decreased with time, and the 1In value found in May 1996 was close to the average value of 0.739 reported by Buchter et al. (1989) for 11 natural soils. The Kf value obtained in May 1996 is comparable with the adsorption coefficient of 1474ml g-i reported by Smit and Van Gestel (1996) for zinc polluted smelter soils having similar soil characteristics as the soil used in this study. In our study, the most pronounced changes in zinc concentrations occurred during a relatively short period of time after the first rainfall in the autumn of 1994 when leaching of excess, readily soluble ZnC12 resulted in decreased total zinc concentrations. At the same time, an increase in pH was observed, probably due to the gradual release of calcium from the sharp sand which was used to improve soil structure. In studies with natural soil, Barrow (1986a) has shown that a one unit increase of pH by the addition of CaCO3 increased the zinc retention capacity of the soil by a factor of about 10. The combination of decreased total zinc concentrations and increased sorption resulted in a rapid reduction of the CaCI2 exchangeable and water soluble zinc concentrations during the first month after construction of the enclosure plots. Thereafter, bioavailability of zinc
78
C . E . Smit et al.
Table 2. ECso values and 95% confidence intervals for the effect of zinc on the reproduction of Foisomia eandida after incubation in zinc contaminated test field soil in laboratory and field experiments during three successive years
Date of experiment
August 1994
Lab or field experiment
Field
December 1994
Laboratory
September 1995
Field
February 1996
Laboratory
September 1996
Field
ECso (95% confidence interval)
Reference
Total
Exchangeable zinc (/zgg-I dry soil)
Water soluble
979a (892-1074) 90b (670-1213) 1705 (1246-2329) 1491 (1143-1947) 2178 ( 1527-3106) 1749 (886-3450)
238a (I 77-322) 179b (144-222) 268 (204-350) 255 ( 177-367) 332 (269-368) 359 (125-1033)
70.2a (48.3--102) 13.1b (--) 19.3 (15.1-24.2) 30.1 ( 17.8-51.0) 22.1 ( 14.5-33.8) 20.2 (1.81-225)
This study This study 1 This study 2 This study
Effect concentrations are based on total, CaCI2 exchangeable and water soluble zinc concentrations. aEffect concentrations based on soil analysis data of August 1994. bEffect concentrations based on soil analysis data of February 1995. References: 1: Smit and Van Gestel (1995); 2: Smit (1997). further decreased with time. From July 1994 to May 1996, the maximum CaCI2 exchangeability decreased by 25%, and the water solubility of zinc at the highest nominal concentration declined from 36.8 to 1.5% of the total zinc concentration. The results of the present study clearly demonstrate the effect of ageing on the sorption of zinc. In several studies it was shown that an increase in contact time increased the retention of heavy metals by soils (Barrow, 1986b; Mann and Ritchie, 1994). It is assumed that zinc sorption onto soil involves an initial, rapid adsorption reaction followed by diffusive penetration of the surface (Barrow, 1986b). This sorption process can be simulated artificially, as shown by Briimmer et ai. (1983), who have used an eight weeks period of alternate drying and rewetting to equilibrate zinc with soil after contamination with ZnC12. The levels of water solubility observed from February 1995 are in good agreement with data on field soils. Bogacz (1993) has reported an average water solubility of zinc of 1.1% of the total zinc concentration after analysis of 49 soils. Maximum water soluble levels of 3.73 and 2% have been determined for soil samples from zinc polluted smelter areas (Spurgeon and Hopkin, 1995, 1996a). Although sorption still increased after February 1995, the results indicate that within a few months the test field soil had developed to a situation more or less comparable to a field situation. Develo0ment of zinc toxicity with time; comparison with laboratory tests
In 1995 and 1996, control reproduction met the requirement of a minimum of 100 juveniles per test container proposed in the Collembola test guideline (ISO, 1994). The relatively low numbers of juveniles observed in the experiments can be partly explained by the low average temperatures during the tests. The development time of F. candida is temperature depen-
dent (Snider, 1973; Johnson and Wellington, 1980), and the optimum temperature for reproduction of F. candida is between 15.5 and 21°C (Snider, 1973). At lower temperatures, animals will reproduce later and eggs will need longer time to hatch (Snider, 1973). The longest development time was observed in 1994, and control survival and reproduction were low compared to the other years. Large soil columns were used to incubate F. candida during the 1994 experiment, and animals might have experienced problems reaching the food which was placed on top of the soil. In general, performance was better when experiments were started with adult animals. In studies with the collembolan Lobella maxillaris, Choudhuri et al. (1979) have shown that the first juvenile stages are sensitive to low humidity. It can be assumed that juvenile animals, which have to complete several moulting cycles before reaching the reproductive age, are more susceptible to drought and other adverse conditions than adults. A relationship between mortality and zinc concentration was not observed in any of the experiments. Results from previous studies indicated that mortality is a variable and insensitive parameter to estimate effects of contaminants on F. candida (Crommentuijn, 1994; Van Gestel and Hensbergen, 1997). Instead, production of juveniles appeared to be a sensitive and reproducible parameter to assess zinc toxicity for F. candida (Smit, 1997). Exposure temperature and incubation time determine the number of offspring produced, but at incubation temperatures between 13 and 20°C, the influence of these test conditions on the toxicity of zinc is small when reproduction is considered (Smit and Van Gestel, 1997; Van Gestel and Hensbergen, 1997). Therefore, reproduction represents a suitable parameter to compare results of different experiments. ECs0s based on total zinc concentrations increased with time, indicating that increased sorption and
Development of zinc bioavailability and toxicity
reduced bioavailability caused a reduction in toxicity. Expression of the ECs0 on the basis of CaC12 exchangeable zinc concentrations reduced the differences in effect concentrations observed between the respective experiments. ECsos obtained from the last two experiments ( 1995 and 1996) are in good agreement with data obtained from studies in which soil from the test field was used in laboratory toxicity tests (Table 2). Assuming that the actual ECso based on water soluble zinc for the 1994 experiment is indeed lower than the estimated value of 70.2/~g Zng -I dry soil, the differences between laboratory data and ECs0s obtained in the present field study are all less than a factor of 2. Given the great differences in test conditions, the difference between field ECsos and laboratory data is rather small. ECsos based on total zinc concentrations in the field experiments were three to ten times higher than effect concentrations reported from studies with freshly contaminated soil (Smit and Van Gestel, 1996; Van Gestel and Hensbergen, 1997). Spurgeon and Hopkin (1995, 1996b) studied population density and species diversity of earthworms in the vicinity of a zinc smelter, and conclude that differences in bioavailability are the main reason that earthworms could survive in the field at soil zinc concentrations which exceeded by far the laboratory based effect concentrations. Smit (1997) concludes that water soluble zinc concentrations are most appropriate to account for differences in bioavailability when different soil types are used, and estimated an average ECso based on water soluble zinc concentrations of 15.4#gg -! dry soil for 10 tests in which different soils were used. Water soluble ECs0s obtained in the present study are in the same order of magnitude, which indicates that bioavailability of zinc is the main factor determining toxicity for F. candida. This supports the conclusion of Van Gestel and Van Straalen (1994) that field effects can be predicted from laboratory studies provided that the test species is the same, and the exact exposure concentration is known.
CONCLUSION From the results of this study it is concluded that large changes in bioavailability occur when an artificially contaminated soil is incubated under outdoor conditions. When based on accurately determined water soluble concentrations, the toxicity of zinc for Folsomia candida under outdoor conditions is in good agreement with laboratory data.
ACKNOWLEDGEMENTS The authors are indebted to: G. P. Green of the Panheel Groep for supplying the test field soil; R. Baerselman, A. J. Folkerts, R. Henzen, A. Kalif, K. Verhoef, R. Verweij, and colleagues of the Department of Ecology
79
and Ecotoxicology for their help in the construction of the test field plot; staff members of the Laboratory of Ecotoxicology and the Laboratory of Analytical Chemistry of the National Institute of Public Health and the Environment for the analysis of soil samples; E. Stam for his help with statistics. Thanks are due to N. M. Van Straalen and two anonymous referees for comments on the manuscript. M. Aldham-Breary improved the English style. This study was financially supported by the Dutch Ministry of Housing, Spatial Planning and Environment.
REFERENCES
Barrow, N. J. (1986a) Testing a mechanistic model. IV. Describing the effects of pH on zinc retention by soils. Journal of Soil Science 37, 295-302. Barrow, N. J. (1986b) Testing a mechanistic model, lI. The effects of time and temperature on the reaction of zinc with a soil. Journal of Soil Science 37, 277-286. Bogacz, W. (1993) Influence of soil properties on the distribution of zinc in polish soils. Geomicrobiology Journal 11,299308. Brfimmer, G., Tiller, K. G., Herms, U. and Clayton, P. M. (1983) Adsorption-desorption and/or precipitation-dissolution processes of zinc in soils. Geoderma 31,337-354. Buchter, B., Davidoff, B., Amacher, M. C., Hinz, C., Iskandar, I. K. and Selim, H. M. (1989) Correlation of Freundlich Kd and n retention parameters with soils and elements. Soil Science 148, 370-379. Burton, G. A., Hickey, C. W., DeWitt, T. H., Roper, D. S., Morrisey, D. J. and Nipper, M. G. (1996) In situ toxicity testing: teasing out the environmental stressors. SETAC Europe News 7, 6-9. Cairns, J. (1984) Are single species toxicity tests alone adequate for estimating environmental hazard? Environmental Monitoring and Assessment 4, 259-273. Callahan, C. A., Menzie, C. A., Burmaster, D. E., Wilborn, D. C. and Ernst, T. (1991) On-site methods for assessing chemical impact on the soil environment using earthworms: a case study at the Baird and McGuire superfund site, Holbrook, Massachusetts. Environmental Toxicology and Chemistry 10, 817-826. Choudhuri, D. K., Bhattacharyya, B., Mukherjee, P. and Pande, T. (1979) The effects of various physical factors on the biology of Lobella (Lobella) maxillaris Yosii, 1966 (Collembola: Insecta). Revue d'Ecologie et Biologie du Sol 16, 241-257. Crommentuijn, G. H. (1994) Sensitivity of soil arthropods to toxicants. Ph.D. thesis, Vrije Universiteit Amsterdam. Edwards, C. A. and Bohlen, P. J. (1992) The effects of toxic chemicals on earthworms. Reviews of Environmental Contamination and Toxicology 125, 23-99. Eijsackers, H. and Van de Bund, C. F. (1980) Effects on soil fauna. In Interactions Between Herbicides and the Soil, ed. R. J. Hance, pp. 255-305. Academic Press, London. Faber, J. H. and Verhoef, H. A. (1991) Functional differences between closely related soil arthropods with respect to decomposition processes in the presence or absence of pine-tree roots. Soil Biology and Biochemistry 23, 15-23. Gulley, D. D., Boelter, A. M. and Bcrgrnan, H. L. (1990) TOXSTAT ~ Release 3.3. Fish Physiology and Toxicology Laboratory, Department of Zoology and Physiology, University of Wyoming. Haanstra, L., Doelman, P. and Oude Voshaar, J. H. (1985) The use of sigmoidal dose response curves in soil ecotoxicological research. Plant and Soil 84, 293-297.
80
C.E. Smit et al.
ISO (1994) Soil quality-Effects of soil pollutants on Collembola (Folsomia candida): method for the determination of effects on reproduction. ISO/DRAFT. Johnson, D. L. and Wellington, W. G. (1980) Post-embryonic growth of the collembolans Folsomia candida and Xenylla grisea at three temperatures. Canadian Entomologist 112, 687-695. Kammenga, J. E., Van Koert, P. H. G., Riksen, J. A. G., Korthals, G. W. and Bakker, J. (1996) A toxicity test in artificial soil based on the life-history strategy of the nematode Plectus acuminatus. Environmental Toxicology and Chemistry 15, 722-727. Mann, S. S. and Ritchie, G. S. P. (1994) Changes in the forms of cadmium with time in some western Australian soils. Australian Journal of Soil Research 32, 241-250. Menzie, C. A., Burmaster, D. E., Freshman, J. S. and Callahan, C. A. (1992) Assessment of methods for estimating ecological risk in the terrestrial component: a case study at the Baird & McGuire superfund site in Holbrook, Massachussetts. Environmental Toxicology and Chemistry 11,245-260. Sheppard, S. C. (1997) Toxicity testing using microcosms. In Soil Ecotaxicology, eds J. Tarradellas, G. Bitton and D. Rossel, pp. 345-373. Lewis Publishers, Boca Raton. Smit, C. E. (1997) Field relevance of the Folsomia candida soil toxicity test. Ph.D. thesis, Vrije Universiteit Amsterdam. Smit, C. E. and Van Gestel, C. A. M. (1995) Influence of temperature and humidity fluctuations on the sensitivity of Folsomia candida for zinc. In Contaminated Soil '95, eds W. J. Van den Brink, R. Bosman and F. Arendt, pp. 645646. Kluwer Academic Publishers, Dordrecht. Smit, C. E. and Van Gestel, C. A. M. (1997) Comparison of the toxicity of zinc for the springtail Folsomia candida in artificially contaminated and polluted field soils. Applied Soil Ecology 3, 127-136. Smit, C. E. and Van Gestel, C. A. M. (1997) Influence of temperature on the regulation and toxicity of zinc in Folsomia candida. Ecotoxicology and Environmental Safety, 37, 213-222. Snider, R. (1973) Laboratory observations on the biology of Folsomia candida (Willem) (Collembola:Isotomidae). Revue d'Ecologie et Biologie du Sol 10, 103-124.
Sokal, R. R. and Rohlf, F. J. (1995) Biometry. W. H. Freeman, New York. Spurgeon, D. J. and Hopkin, S. P. (1995) Extrapolation of the laboratory-based OECD earthworm toxicity test to metalcontaminated field sites. Ecotoxicology 4, 190-205. Spurgeon, D. J. and Hopkin, S. P. (1996a)Effects of variations of the organic matter content and pH of soils on the availability and toxicity of zinc to the earthworm Eisenia fetida. Pedobiologia 40, 80-96. Spurgeon, D. J. and Hopkin, S. P. (1996b) The effects of metal contamination on earthworm populations around a smelting works: quantifying species effects. Applied Soil Ecology 4, 147-160. Van Gestel, C. A. M. and Van Straalen, N. M. (1994) Ecotoxicological test systems for terrestrial invertebrates. In Ecotoxicology of Soil Organisms, eds M. H. Donker, H. Eijsackers and F. Heimbach, pp. 205-227. Lewis publishers, Boca Raton. Van Gestel, C. A. M. and Hensbergen, P. J. (1997) Interaction of Cd and Zn toxicity for Folsomia candida Willem (Collembola: Isotomidae) in relation to bioavailability in soil. Environmental Toxicology and Chemistry 16, 1177-1186. Van Gestel, C. A. M., Lton, C. D. and Van Straalen, N. M. (1997) Evaluation of soil fauna ecotoxicity tests regarding their use in risk assessment. In Soil Ecotoxicology, eds J. Tarradellas, G. Bitton and D. Rossel, pp. 291-317. Lewis Publishers, Boca Raton. Van Straalen, N. M. and Denneman, C. A. J. (1989) Ecotoxicological evaluation of soil quality criteria. Ecotoxicology and Environmental Safety 18, 241-251. Van Straalen, N. M., Leeuwangh, P. and Stortelder, M. P. (1994) Progressing limits for soil ecotoxicological risk assessment. In Ecotoxicology of Soil Organisms, eds M. H. Donker, H. Eijsackers and F. Heimbach, pp. 397-409. Lewis publishers, Boca Raton. Verhoef, H. A. and Van Gestel, C. A. M. (1995) Methods to assess the effects of chemicals on soils. In Methods to Assess the Effects of Chemicals on Ecosystems, eds R. A. Linthurst, P. Bourdaeu and R. G. Tardiff, pp. 223-257. J. Wiley and Sons Ltd, New York.