The use of temporary pond microcosms for aquatic toxicity testing: direct and indirect effects of endosulfan on community structure

The use of temporary pond microcosms for aquatic toxicity testing: direct and indirect effects of endosulfan on community structure

Aquatic Toxicology 41 (1998) 101 – 124 The use of temporary pond microcosms for aquatic toxicity testing: direct and indirect effects of endosulfan o...

258KB Sizes 0 Downloads 51 Views

Aquatic Toxicology 41 (1998) 101 – 124

The use of temporary pond microcosms for aquatic toxicity testing: direct and indirect effects of endosulfan on community structure Michael J. Barry *, Dean C. Logan Key Centre for Applied and Nutritional Toxicology, RMIT-Uni6ersity, City Campus, GPO Box 2476V, Melbourne 3001, Australia Received 4 October 1996; received in revised form 21 July 1997; accepted 28 July 1997

Abstract Replicate aquatic microcosms can be established using sediment from the basin of dried temporary ponds. When the sediment is flooded with water, the resting stages or eggs of zooplankton, phytoplankton, macrophytes and filamentous algae are activated and a complex community develops in a few weeks. The aim of this study was to test the utility of using temporary pond microcosms for aquatic toxicity testing, by dosing a set of microcosms with the organochlorine pesticide, endosulfan and measuring physical, chemical and biological responses over a period of 10 weeks. A total of 20 3-l microcosms were established and after 6 weeks dosed with endosulfan at 0, 1, 10 or 50 mg l − 1. The application was repeated after 3 weeks. The pH, conductivity, dawn and dusk oxygen levels, ammonia, nitrite, orthophosphate, chlorophyll-a, zooplankton and phytoplankton were measured at weekly intervals. After 10 weeks the total composition of each microcosm was determined and endosulfan residues in the macrophytes and sediments measured. The half-life of endosulfan in the water column was approximately 24 h. After 10 weeks between 6 and 12% of the initial dose could be recovered from the sediment in a toxic form (a or b endosulfan or endosulfan sulfate). Total mean endosulfan residues were 0.57, 27 and 61 ng g − 1 dry weight in macrophytes and 1.49, 7.0 and 36 ng g − 1 dry weight in the sediments for the 1, 10 and 50 mg l − 1 treatments, respectively. At 10 and 50 mg l − 1 endosulfan, populations of ostracods and calanoid copepods were eliminated and the cladoceran Ceriodaphnia sp. significantly reduced. There were significant increases in Simocephalus sp., tardigrades, filamentous algae, chlorophyll-a and macrophytes. Orthophosphate concentrations were correlated with macrophyte biomass and were lower in the treated microcosms. Nitrite levels were also lower in the treated microcosms. There was quantal shift in composition of the animal community between 1 and * Corresponding author. Current address: Department of Biological and Food Sciences, Victoria University of Technology, St. Albans, 3021, Australia. 0166-445X/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved. PII S 0 1 6 6 - 4 4 5 X ( 9 7 ) 0 0 0 6 3 - 5

102

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

10 mg l − 1 endosulfan which could be visualized using the ordination technique multi-dimensional scaling. The mechanisms underlying this quantal shift in community structure are analyzed and the advantages and limitations of using temporary pond microcosms for toxicity testing are discussed. © 1998 Elsevier Science B.V. All rights reserved. Keywords: Replicate aquatic microcosm; Endosulfan; Sediment; Ceriodaphnia; Simocephalus

1. Introduction The aim of ecotoxicology is to measure and predict the effects of chemicals on the natural environment. Ecosystem level toxicity testing has been widely advocated as an important adjunct to single species bioassays for this assessment process (Kimball and Levin, 1985, Cairns, 1983, 1986; Meyer and Ellersieck, 1986; Kenaga, 1987; Perry and Troelstrup, 1988). In recent years several standardised protocols have been developed to measure the impact of chemicals at the ecosystem level and numerous papers have appeared using ad hoc approaches. Major limitations of ecosystem level bioassays include the inherent variability of multi-species biological systems and the complexity of resulting data. Another major constraint with ecosystem bioassays is the time and cost involved in the establishment and monitoring of test systems and low replication levels normally employed. Significant perturbations of ecological processes can be difficult to detect against background variation and the discriminatory power of such systems can be low. As a consequence of these problems, the US Environmental Protection Agency no longer routinely requests aquatic mesocosm studies as a prerequisite for pesticide registration (Shaw and Manning, 1996). Lentic ecosystem level bioassay tests can be classified on the basis of test volume (B 1 l to \ 600 m3) and on the basis of composition (from completely natural to entirely synthetic). Small volume or microcosm studies provide a viable alternative to large scale mesocosm testing and a number of protocols have been proposed. Microcosms can be entirely ‘synthetic’ or constructed entirely from defined media and laboratory raised populations (Taub et al., 1983). Alternatively, they can be ‘natural’ systems, constructed from sediment and water collected in the field (Van Donk et al., 1995). Synthetic microcosms have the advantage of simplicity and a high degree of reproducibility, however, they represent a relatively abstract system with tenuous links to the real world. Another problem with synthetic microcosms is that they usually rely on a limited number of ‘laboratory weed species’, taxa which are easy to culture in the laboratory (Taub et al., 1983). In contrast, systems based on natural communities often provide a closer approximation to natural ecosystems, but are harder to replicate. The composition, structure and function of such systems may approximate the organisation of larger systems, indicating that interactions observed in the laboratory should reflect at least a subset of the biotic and chemical processes which occur in the wild (Odum, 1969; Neill, 1975). Such systems may therefore be of great utility for answering particular questions about the effects of chemicals on ecosystem function, but are of less value to regulators who require consistent comparisons across time and space.

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

103

Temporary ponds occur over most of Australia, even in the coastal mesic well-watered zone (Lake et al., 1989). They have high species diversity and there is a large degree of overlap with the biota of more permanent waters (Perry, 1981), suggesting that they may make a good model for toxicity testing. Sediment from dried temporary ponds contains the resting stages of a wide range of phytoplankton, zooplankton, filamentous algae and macrophytes. When this sediment is flooded with water a complete community will develop. The aim of this study was to test the suitability of temporary pond microcosms for aquatic toxicity studies and to demonstrate the utility of the system, by measuring the direct and indirect effects of an agricultural chemical on biotic and abiotic parameters in the microcosms. The cyclodiene pesticide, endosulfan, was selected as the toxicant for this study. Endosulfan is widely used in Australia as an insecticide on cotton and vegetable crops and is regarded as a serious potential contaminate of Australian waterways (Peterson and Batley, 1993; Sunderam et al., 1992). It is a central nervous system stimulant (Eldefrawi and Eldefrawi, 1989) and the major molecular target for endosulfan is believed to be the GABAA receptor (Lawrence and Casida, 1984; Bloomquist and Soderlund, 1985).

2. Methods Sediment was collected from the basin of a dried temporary pond in south-western Victoria, Australia. The sediment was homogenised by sifting through a 3-mm sieve and stored in a dark sealed container in the laboratory for 6 months prior to the start of the experiment. Of the sifted sediment, 130 g were added to each of 20, 4-l glass aquaria and covered with a thin layer (500 g) of coarse aquarium gravel. The gravel was used to prevent disturbance of the sediment when the microcosms were stirred for zooplankton sampling. Aquaria were placed in a four by five block design under a bank of triphosphor fluorescent lights with a 14 h light/10 h dark photoperiod. Microcosms were aerated by a gentle flow of 0.2 mm filtered air from standard aquarium air-diffusers. Air temperature was maintained at 20°C 9 3. Development of the systems was initiated by adding 3 l of distilled water to mimic natural flooding. Further distilled water was added when required to replace losses due to evaporation and chemical sampling. At 3 weeks post-flooding, the tanks were cross-seeded to increase homogeneity by siphoning 1.5 l from each tank, mixing and redistributing the water. The microcosms were inspected at regular intervals for the presence of concostracans (Limnetis sp.) and shield shrimps (Lepiduras apus 6iridis). These organisms were removed as soon as they were sighted. Preliminary studies indicated that the distribution of these species between tanks was likely to be uneven. Given their large size relative to the microcosm volume it was concluded that their presence in some microcosms may unduly bias the results.

104

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

At 6 weeks post-flooding, 16 microcosms were selected for inclusion in the study. The main selection criterion was similarity in extent of macrophyte development. The 16 tanks were randomly assigned to one of four treatment groups each of which was replicated four times: control, 1, 10 or 50 mg l − 1 endosulfan. Technical endosulfan (70% a isomer, 30% b isomer) was added to each of the aquaria at the above concentrations in 50 ml of acetone, with controls receiving acetone only. The microcosms were immediately stirred with a glass rod to mix the toxicant. The day of the first pesticide application was designated day 0. Dosing of the microcosms was repeated on day 21. The concentration of endosulfan in the water was measured at 2, 4, 8, 24, 48 and 144 h after the initial application of the pesticide and 24 and 48 h after the second application. Endosulfan residues in the sediment and macrophytes were measured at the end of the study. Water-borne endosulfan was measured by extraction of a 10-ml sample of microcosm water with 2 ml of iso-octane. Measurements were made on a Shimadzu® gas-liquid chromatograph (GC) by electron-capture detector (ECD). Samples were quantified against standards for a and b endosulfan and endosulfan sulfate. A standard for endosulfan diol was not available. At the end of the experiment (day 71), the total macrophyte population of each tank, plus any attached filamentous and epiphytic algae which could not be separated from the macrophytes, was rinsed free of sediment, freeze-dried, weighed and stored at −20°C until analysis. Endosulfan residues were extracted from the macrophytes by placing in a soxhelet apparatus with 150 ml of solvent (90% hexane, 10% acetone). Following extraction, endosulfan was concentrated in 1.5 ml of iso-octane by rotary evaporation. Samples were cleaned on a Florisil column with an elution wash mix (50% dichloromethane, 0.35% acetonitrile, 49.65% hexane) and endosulfan residues measured by ECD-GC. Sediment samples were also collected at the end of the experiment for measurement of endosulfan residues. Following removal of the macrophytes, the sediment was thoroughly mixed and a subsample freeze-dried and stored at − 20°C until analysis. Endosulfan residues were extracted from triplicate 20 g samples using the methods noted above for the macrophytes. When initial samples were run on the GC a large peak, identified as elemental sulfur, was found to mask endosulfan peaks on some chromatograms. However, several clean-up procedures were tested, to develop a method which would remove the elemental sulfur without affecting the endosulfan which also contains sulfur. The most satisfactory method involved treating the extracts with copper strips to produce copper sulfite. Tests with spiked samples indicated that degradation of endosulfan was minimal. Aldrin was used as an internal standard for all procedures and endosulfan concentrations were corrected for aldrin recoveries. Pesticide grade solvents were used in all extractions. All physical, chemical and biological measurements were made on the same days: −1, 3, 7 and weekly intervals thereafter until day 70. The pH, temperature, conductivity and oxygen concentration of the water in each microcosm were measured pre-dawn (i.e. immediately before lights-on) on each sampling day. Temperature and oxygen concentration were also measured at dusk (i.e. shortly before lights-off) on the day prior to sampling (i.e. day − 2, 2, 6, 13…).

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

105

The nutrients (ammonia, nitrite, orthophosphate) were sampled within 1 h of lights-on and analysed the same day. The microcosms were stirred with a glass rod and two 5-ml samples of water were withdrawn from each microcosm for each nutrient and immediately sealed teflon vials and kept on ice in the dark until analysis. For chlorophyll analysis a single 50-ml water sample withdrawn from each microcosm, filtered through a 60-mm mesh and stored on ice in the dark until analysed. Analyses were ordered in such a way, so as to minimise degradation of potentially unstable compounds, particularly chlorophyll-a and orthophosphate. Ammonia, reactive nitrite and reactive phosphorus concentrations were measured spectrophotometrically using a modification of the method of Strickland and Parsons (1972). As sample volumes were 5 ml compared to the 100 ml recommended by Strickland and Parsons, all procedures were scaled to the volumes required and compared against standard curves after correction for turbidity in the blanks. The methods were tested and validated for use with low nutrient concentrations before commencement of the study. Two readings were made for each sample and the mean value calculated. A mean for each microcosm was then calculated using the average of these two means. Chlorophyll-a was measured flurometrically (Clesceri et al., 1989) and samples were corrected for phaeophytin by acidification with HCl. Zooplankton were sampled immediately after collection of physico-chemical data. The tanks were gently stirred to evenly distribute the zooplankton prior to sampling. A glass rod with a rubber stopper fixed to the end was placed on the sediment and a glass tube 5 cm in diameter inserted over it. The entire column of water inside the glass tube was removed and poured into a beaker. This procedure was repeated giving a sample volume of approximately 160 ml. The precise volume was measured by pouring the contents of the beaker into a measuring cylinder. The sample was then poured through a 60-mm sieve to remove the metazoans and the water returned to the microcosm. The zooplankton were stored in the refrigerator prior to enumeration which occurred within 3 h of collection. Immediately prior to counting the animals were killed by placing in 70% ethanol. Zooplankton were counted in a Borgov tray under a stereo-binocular microscope. Phytoplankton were collected by stirring the microcosms with a glass rod and then taking a 10-ml sample with an automatic pipette. The samples were immediately transferred to tapered glass test-tubes and Lugol’s iodine added to make a 1% solution. Samples were stored in the dark for 4–6 days to allow phytoplankton time to settle. Approximately 9 ml of supernatant was drawn off and the phytoplankton resuspended in the remaining volume. This liquid was transferred to pre-weighed teflon vial and a new weight calculated. This value was used to determine the volume of the remaining liquid. The teflon vials were sealed and stored in the dark until analysis. The biomass of macrophytes in the microcosms at the start of the experiment was estimated by counting the number of plants in each tank and assigning a value between one and five to each macrophyte depending on its size. The smallest macrophytes were scored as one and the largest as five. A measure of the total biomass of macrophytes in each microcosm was then calculated as the sum of the scores.

106

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

At the completion of the study the water was siphoned from each tank using plastic tubing and passed through a 60-mm sieve to collect the metazoans. The zooplankton were preserved in 70% alcohol and stored at 4°C until they could be enumerated. There was considerable growth of filamentous algae in some of the tanks and it was entangled with the macrophytes. It was not possible to disentangle the filamentous algae from the plant material so a semi-quantitative measure of its biomass was calculated by assigning a value between one and five to each microcosm depending on the quantity of filamentous algae present. A value of one indicated little visible filamentous algae and a value of five indicated abundant growth. The macrophytes together with entangled filamentous algae were carefully removed from the sediment rinsed free of mud, then transferred to 100-ml teflon tubes for freeze drying and weighing. All sediment that was not required for pesticide analysis was preserved in 70% ethanol and stored at 4°C until it could be sorted. Prior to enumeration of benthic fauna, a 25% subsample of the sediment was removed and divided into six size classes by sieving. The total number of animals in the subsample was counted in a Borgov tray under a stereo-binocular microscope. The effects of endosulfan on zooplankton densities, during the 10-week sampling period, were statistically analysed using one-way analyses of variance (ANOVA) for each sampling date. The effects of pesticide treatment on total numbers of each of the major metazoan taxa in the tanks at the end of the study were also measured using one-way ANOVA. Tukey’s test was used to determine which treatments were different when the overall ANOVA was significant. The effects of endosulfan on the total quantity of filamentous algae in the microcosms at the conclusion of the study was analysed using non-parametric ANOVA on the semi-quantitative scores. The effects of endosulfan on the biomass of macrophytes in each microcosm at the conclusion of the study was similarly analysed. As variation in the biomass of macrophytes in each tank at the beginning of this study may have influenced the final result, a second analysis was performed. Microcosms were ranked from 1 to 15 on the basis of their macrophyte score at the start of the experiment with one representing the tank with the least macrophytes. The microcosms were again ranked from 1 to 15 at the conclusion of the study on the basis of total dry weight of plant material. A new score was then calculated for each tank by subtracting the original score from the final one. Thus, a tank which was ranked 4th initially and 7th at the end would have a score of − 3. A tank which was ranked 12th at the start and 2nd at the end would be given a score of +10. The effects of endosulfan on these scores were analysed with non-parametric ANOVA. Changes in physico-chemical parameters were regarded as indirect effects. A priori, it was assumed that there would be more inherent variability in indirect effects compared with direct effects on the fauna. Simple linear regression on the values for each microcosm were used to identify treatment related trends in physico-chemical parameters. Regression was selected for analysis because it is a more appropriate model for identifying dose related trends.

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

107

The use of multiple analyses for the weekly sampling data increased the chances of making a Type 11 error, or finding a significant difference where none really existed. To reduce the probability of making a Type 11 error, biological significance was only ascribed to results which showed significant differences on several dates. All data were checked for conformity with the assumptions required for valid parametric analysis. Data which showed significant heteroscendacity was log-transformed (Zar, 1984). A significance level of PB 0.05 was selected as the critical value in all analyses. Analyses were performed using the program Superanona® v. 1.11. The effects of endosulfan on the total community structure at the conclusion of the study were assessed using the ordination technique non-metric multi-dimensional scaling (MDS). MDS is a procedure for fitting a set of points in space such that the distances between the points correspond as closely as possible to a given set of dissimilarities between objects (Wilkinson, 1989). Microcosms with similar faunal composition should cluster close together in multi-dimensional space and those which are dissimilar, further away. Numerical abundances of the nine most common taxa in the microcosms were standardised and the Pearson correlation matrix calculated. Data were ordinated using the program MDS in the statistical package Systat (Wilkinson, 1989). 3. Results The microcosms developed rapidly following flooding and protozoa were observed in the tanks at the end of the first week. Macrophytes began to emerge from the sediment in the second week. Metazoans, including copepod nauplii, ostracods and cladocerans emerged towards the end of the second week and rapidly proliferated. Some edge effects were apparent, with the front corner tanks showing less primary production than others and those at the rear, closest to the wall, showing the greatest. As there was no relationship between tank position and temperature, it was assumed that these differences were due to variation in light intensity reaching the tanks from the top and reflection from the rear wall. Following the initial sampling on day − 1, it was noted that one of the control tank was significantly different from the other systems. This microcosm had the lowest macrophyte biomass and the ammonia concentration was an order of magnitude higher than other tanks. The microcosm was also relatively depauperate in zooplankton species. The replicate appeared to be an outlier in terms of stage of development and was excluded from further analysis. The logic for this exclusion was the need for a high degree of homogeneity between all tanks at the start of the experiment to maximise the power of study to discriminate endosulfan-induced effects from background ‘noise’ variability. Endosulfan rapidly partitioned out of the water column. The measured concentrations of endosulfan in the water fell by 50% within the first 24 h and a further 50% in the next 24 h. After 7 days no endosulfan was detected in the water of the low and medium treatment microcosms, but 4 mg l − 1 was recorded in the high treatment tanks (Fig. 1(A)). A similar pattern was observed after the second application of endosulfan on day 21.

108

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

Fig. 1. (A) Concentration of endosulfan (a+ b +endosulfan sulfate) in water column following initial appliaction of the pesticide. (B) Average ratio of a and b endosulfan and endosulfan sulfate in sediment and macrophyte samples from microcosms. Key: horizontal stripes, a endosulfan; clear, b endosulfan; solid, endosulfan sulfate.

After 10 weeks a total of 0.70 mg endosulfan (as a+ b+ endosulfan sulfate), or 11.6% of the total amount applied, was recovered from the sediment and macrophytes of the low dose microcosms. For the medium dose microcosms, 3.69 mg of endosulfan or 6.15% of total dose was recovered, and for the high dose microcosms, 17.95 mg of endosulfan or 5.98% of the total dose was recovered (Table 1). Endosulfan residues were detected in both the macrophyte and sediment samples (Table 1). Mean concentrations of endosulfan in the macrophytes were 0.57, 27.4 and 61 ng g − 1 dry weight for low, medium and high dose treatments, respectively. Mean concentrations of endosulfan in the sediments were 1.49, 6.97 and 35.73 ng g − 1 dry weight for low, medium and high dose treatments, respectively. The mean ash free dry weight of the sediment was 1.78% of the initial weight, indicating a low organic content. The actual organic content of the temporary pond sediment was much higher but the organic content was diluted by the sand cover which was placed on top of the sediment at the start of the Table 1 Mean9 S.E. endosulfan residues (a+b endosulfan+endosulfan sulfate) in sediment and macrophytes remaining in the microcosms 10 weeks after the initial application: the total amount of endosulfan remaining and the percentage of the total application this represents are also presented Endosulfan application rate (mg l−1)

1 10 50

Endosulfan residues (ng g−1 dry wt.)

Sediment

Macrophytes

1.49 (0.49) 6.97 (0.72) 35.73 (3.13)

0.57 (0.29) 27.47 (4.92) 61.44 (22.55)

Total endosulfan recovered (mg)

Total dose still in toxic form (%)

0.70 (0.21) 3.69 (0.43) 17.95 (1.55)

11.6 (3.50) 6.15 (0.72) 5.98 (0.52)

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

109

experiment. When endosulfan concentrations in the sediment were expressed in terms of ash free dry weight concentrations were much higher. The major form of endosulfan detected was endosulfan sulfate. In the macrophytes this accounted for 82% of total endosulfan detected compared with 8% for the a and 9% for the b isomer. This compared with 68, 9 and 22% for endosulfan sulfate, a and b, respectively, in the sediments (Fig. 1(B)). Therefore a greater proportion of the endosulfan had been degraded to the sulfate in the macrophytes as compared to the sediment. The pH of water was normally slightly acidic, fluctuating between pH 6 and pH 7 (Fig. 2(A)). There was a consistent positive relationship between mean pH and endosulfan dose. Regression analysis indicated that this relationship was significant on two dates (days 21 and 49). There was a steady decrease in conductivity from a mean of 230 mS cm2 at the start of the experiment to a mean of 120 mS cm2 at the conclusion (Fig. 2(B)). The decrease was indicative of a dilution effect from repeated sampling for chemical analysis, but was also related to increased macrophyte biomass. Microcosms which were not sampled showed a similar, but lesser decrease in conductivity. There was an effect of endosulfan on conductivity with consistently higher values for control and low pesticide treatments compared with medium and high level tanks. On two dates (days 35 and 49) this relationship was statistically significant. Dawn oxygen concentrations varied between 75 and 95% of saturation. The lowest oxygen concentrations were recorded around day 28 of the study (Fig. 2(C)). There was a steady increase in the mean dusk oxygen concentration over the time course of the study and concentrations were consistently higher in the medium and high dose tanks (Fig. 2(D)). The increase in dusk oxygen concentrations was most likely related to an increasing biomass of macrophytes with time. The concentration of ammonia in the tanks dropped sharply between days − 1 and 3 (Fig. 2(E)). Ammonia levels were generally very low and highly variable and not correlated with endosulfan. Nitrite concentrations also dropped between days 0 and 14 which paralleled the reduction in ammonia levels in the tanks (Fig. 2(F)). Following this initial drop the highest values were consistently recorded in the control microcosms. Nitrite concentrations were negatively correlated with endosulfan. On four dates (days 7, 14, 42, 49) the regression was statistically significant. The behaviour of orthophosphate in the microcosms was complex (Fig. 2(G)). Mean concentrations of orthophosphate at the start of the study were approximately 0.3 mg l − 1. Concentrations increased in all treatments peaking around day 35 followed by a decline. The greatest increases were noted in the control and low pesticide treatments although effects were not significant. The concentration of soluble phosphorus in the water column was related to both the abundance of macrophytes and the biomass of phytoplankton. The data was reanalysed using analysis of covariance (ANCOVA) with the dry weight of macrophytes in each tank at the end of the study as a covariate. The ANCOVA indicated a significant difference on four dates (days 7, 35, 42, 56). Interestingly, a multiple regression using endosulfan dose, dry weight of macrophytes and chlorophyll-a concentration did not increase the significance level of the analysis although r 2 values of around

110

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

Fig. 2. Effect of endosulfan on (A) pH; (B) conductivity; (C) dawn oxygen concentrations; (D) dusk oxygen concentrations; (E) ammonia; (F) nitrite; (G) orthophosphate; (H) chlorophyll-a. Key: () control; (") 1 mg l − 1; () 10 mg l − 1; (“) 50 mg l − 1. * P B0.05.

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

111

0.8 were consistently obtained, indicating that these three factors could account for around 80% of the observed variation in soluble reactive phosphorus. An exception to this rule was day 71, which previously showed no significant differences in any analysis. The multiple regression for orthophosphate on this date was highly significant (P B 0.01) and the r 2 value was 0.91. A total of nine metazoan taxa were commonly found in the microcosms. In addition to these groups a number of other taxa, including the calanoid copepod Boekella major, at least one species of Bosminidae, gastrotrichs and rotifers were collected occasionally in the samples Concostracans and shield shrimps were also observed in some tanks, but as noted in the Section 2, they were removed as soon as they were seen. Ceriodaphnia sp., a small cladoceran crustacean, was the numerically dominant organism in the zooplankton. Mean population numbers oscillated in a cyclic fashion (Fig. 3(A)). Ceriodaphnia was relatively tolerant to endosulfan, but numbers were consistently lower in the high dose tanks. Significant differences could be detected using ANOVA on three dates (days 7, 28 and 42). Population oscillations for this treatment also appeared damped relative to other groups. Two species of Chydoridae were commonly found in the microcosms: Alona cambouei and Chydorus sphaerricus. The chydorids were also tolerant to endosulfan and no significant effects on total population numbers could be detected (Fig. 3(B)). Simocephalus sp. was the largest cladoceran found in the microcosms. This species was only collected in the medium and high dose tanks and was not observed in the control or low endosulfan tanks (Fig. 3(C)). This was not a distributional artefact as ephippial eggs of this species were frequently observed in the sediment of all tanks. Large numbers of Simocephalus did not appear in the microcosms until day 21 and became more abundant towards the end of the study. A total of two species of ostracod were found in the tanks: Cypretta sp. and Eucypris sp. Both species were regularly collected in the zooplankton but Cypretta was the more abundant of the two. This was a reflection of spatial distribution with Eucypris being more commonly associated with the benthos and thus less frequently captured. Both species were strongly affected by endosulfan, being totally eliminated in the high dose tanks (Fig. 3(E)). Low numbers persisted in the medium dose systems but no significant effects could be detected in the 1 mg l − 1 microcosms. Calomoecia was the only calanoid copepod commonly found in the microcosms. It was numerically more abundant early in the study, but numbers declined with time in all microcosms. This species was highly sensitive to endosulfan with complete elimination in the medium and high dose tanks (Fig. 3(D)). Low numbers of calanoids persisted in some of the control and low endosulfan microcosms until the end of the study. One species of planarian (flatworm) was observed in the tanks. These animals were bright green in colour, indicating the presence of symbiotic algae. The flatworms became increasingly common as the study progressed but changes in population density were not correlated with endosulfan. At the conclusion of the study it was possible to gain a clearer picture of the spatial distribution of the major taxa and an accurate measure of their total abundances. Fig. 4 illustrates the total abundances of nine taxa (Ceriodaphnia sp.

112

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

Fig. 3. Effect of endosulfan on (A) Ceriodaphnia sp.; (B) Chydoridae; (C) Simocephalus sp.; (D) Calamoecia sp.; (E) Total ostracods. Key: () control; (") 1 mg l − 1; () 10 mg l − 1; (“) 50 mg l − 1. * P B0.05, ** P B0.01.

chydorids, Simocephalus sp., Cypretta sp., Eucypris sp., Calamoecia sp., tardigrades, oligochaetes and planarians) and their spatial distribution. Those cited as being collected in open water also include animals associated with the macrophytes and filamentous algae. Ceriodaphnia was only found in the open water and the population density was negatively correlated with pesticide concentration (Fig. 4(A)). The chydorids were evenly distributed between the sediment and open water and were not effected by

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

113

the pesticide (Fig. 4(B)). Simocephalus was found only in the high and medium dose microcosms (Fig. 4(C)). The ostracod Cypretta sp. was collected from both open water and the sediment, although numbers were higher in the benthic samples (Fig. 4(D)). There was a significant effect of endosulfan on this species (PB 0.001), with greatest numbers in the control and low treatment tanks, lower numbers in the medium level treatments and very few individuals in the high endosulfan treatments. The second ostracod, Eucypris sp., was more prevalent in the sediment than open water (Fig. 4(E)). There was a significant effect of endosulfan on this species (P B 0.0001) with patterns being similar to Cypretta. Both the initial dose of endosulfan and concentrations of the pesticide in the sediment were good predictors of ostracod population densities. Endosulfan dose could explain 73 and 75% of the

Fig. 4. Effect of endosulfan on total numbers 9S.E. of animals in microcosms 10 weeks after initial application. (A) Ceriodaphnia sp.; (B) Chydoridae; (C) Simocephalus sp.; (D) Eucyprus sp.; (E) Cypris sp.; (F) Tardigrada; (G) Oligochaeta; (H) Planaria. Common alphabetic superscripts indicate no significant difference by Tukey’s test where overall ANOVA was significant.

114

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

variance in Eucypris and Cypretta population numbers, respectively. The amount of variance that could be explained by endosulfan concentrations in the sediment was marginally lower, being 67 and 74% for the two species. The tardigrades (water bears) were primarily associated with the sediment and were more abundant in the medium and high level treatments (PB 0.05) (Fig. 4(F)). Oligochaetes were also associated with the sediment but tended to be less abundant in the medium and high level microcosms (Fig. 4(G)). Flatworms were found in both open water and sediment but population densities were not significantly correlated with endosulfan (Fig. 4(H)). A total of three major classes of primary producers were measured; phytoplankton, filamentous algae and macrophytes. The sampling system employed did not provide an adequate measure of quantitative phytoplankton composition, however, a qualitative picture of the phytoplankton could be obtained. The phytoplankton included Vol6ox sp. (which was collected with the zooplankton), Closterium sp., Chlamydomonas sp., Chlorella sp., at least one euglenoid species and at least one unidentified diatom species. There were blooms of both volvox and Closterium in some aquaria and an unidentified jelly like mass was also observed on the walls of some of the tanks. Although individual phytoplankton species could not be quantified, chlorophyll-a provided an indication of overall algal biomass. There was a positive relationship between endosulfan concentration and chlorophyll-a, with the regression being significant on three dates (days 28, 42 and 49) (Fig. 2(H)). The two dominant genera of filamentous algae were Spirogyra sp., (at least three species were present), Oedogonium, Rhizoclonium sp. and Mougeotia sp. were also present. There was little visible filamentous algae in the microcosms at the start of the experiment, however, there were large quantities in the medium and high dose microcosms at the end of the study (Fig. 5(A)). At least several species of macrophytes were collected. Because of their small size the specimens could not be identified with confidence but tentative identifications indicated Myriophylum propinquum, Juncus triarticulatus and Potomogeton pectinatus. All species were present in all tanks but the proportional representation of each varied. A grass like plant was also present in many of the microcosms, but had begun to decay by the end of the experiment. There was significant growth of the macrophytes over the 10-week sampling period but dry weight at the end of the study was not correlated with endosulfan dose (Fig. 5(B)). When change in rank order of the tanks between the start and conclusion of the study was compared, it was apparent that macrophyte growth was greatest in the medium and high level microcosms (Fig. 5(C)). It should be noted that the total macrophyte biomass also some included filamentous algae which was entangled with the macrophytes and this may have contributed to an apparent increased growth rate in the treated microcosms. Multi-dimensional scaling (MDS) was used as an ordination procedure to investigate the effects of endosulfan on the whole invertebrate community at the conclusion of the study (Fig. 6(A)). When the data were scaled in two dimensions, the three control and four low treatment systems formed a relatively tight cluster, indicating a high degree of homogeneity. The medium and high treatment tanks form a diffuse but over-lapping group stretching away from the other points. The

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

115

Fig. 5. Effects of endosulfan on mean9S.E. (A) abundance of filamentous algae; (B) dry weight of macrophytes; (C) relative change in ranking of microcosms based on biomass of macrophytes at start and end of experiment. Common alphabetic superscripts indicate no significant difference by Tukey’s test where overall ANOVA was significant.

scores on axis 1 of the MDS plot were significantly correlated with both pesticide dose and endosulfan residues in the sediment. A plot of log (dose+ 0.1) against score on MDS Axis 1 could explain 77% of the variance in this parameter. Ordination (MDS) of the main biotic and abiotic parameters separated the variables into two groups along Axis 1 (Fig. 6(B)). Endosulfan, tardigrades, Simocephalus, chydorids, tardigrades, planarians, filamentous algae and chlorophyll-a clustered in this first group, all being factors which positively correlated with endosulfan. Volvox and macrophytes were outliers of this group. The other main group was more diffuse, containing the ostracods, Ceriodaphnia and orthophosphate. Calanoids and oligochaetes were outliers on this group. Orthophosphate and Vol6ox received strong positive loadings on Axis 2 and macrophytes received strong negative loadings. Macrophytes may have played an important role in the cycling of phosphorus in the microcosms. In step-wise multiple regression of end-point data the only significant predictor of orthophosphate was macrophyte dry weight (P = 0.004; r 2 =0.48). What factors regulated phytoplankton and filamentous algae? A step-wise multiple regression indicated that Ceriodaphnia, Simocephalus and orthophosphate were significant predictors of chlorophyll-a. However, as Simocephalus was in low

116

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

numbers and was positively correlated with chlorophyll-a, both may have been responding to a third factor and not causally linked. When Simocephalus was removed from the analysis Ceriodaphnia, filamentous algae and orthophosphate were the main predictors of chlorophyll-a. Filamentous algae was also positively correlated with chlorophyll and was removed from the regression leaving Ceriodaphnia and orthophosphate as the only predictors of chlorophyll-a (r 2 = 0.77, PB 0.01).

Fig. 6. (A) Plot of multidimensional scaling scores on two axes for microcosms at end of experiment based on faunal composition. C, control; L, 1.0 mg l − 1; n, 10 mg l − 1; H, 50 mg l − 1 endosulfan. (B) Plot of multidimensional scaling scores on two axes for main biotic and abiotic parameters and endosulfan dose at the end of the study. Key: cal, Calonoida; cer, Ceriodaphnia sp.; chl, chlorophyll-a; chy, Chydoridae; cyp, Cypretta sp.; end, endosulfan; euc, Eucypris sp.; fil, filamentous algae; mac, macrophytes; oli, Oligochaeta; pho, orthophosphate; pla, planarians; sim, Simocephalus sp.; tar, Tardigrada; vol, Vol6ox sp.

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

117

A step-wise multiple regression of end-point data indicated that Cypretta, Ceriodaphnia, chydorids, chlorophyll-a and macrophyte biomass were all significantly correlated with filamentous algae. However, only Cypretta and Ceriodaphnia showed negative correlations with algal growth. When these factors were regressed only Cypretta remained significant (r 2 = 0.76, PB 0.001). This ostracod could account for 76% of the variation in filamentous algae growth indicating that grazing by this species was most possibly a major limiting factor.

4. Discussion Less than 10% of the endosulfan which was added to the microcosms remained as either endosulfan a or b or endosulfan sulfate after 10 weeks. The remainder must have been degraded to a non-toxic product or lost through volatilisation. The main degradation products of endosulfan are endosulfan diol, endosulfan ether, endosulfan hydroyether and endosulfan lactone (Peterson and Batley, 1991). Endosulfan diol is regarded as relatively non-toxic and hydrolysis can occur by chemical or biological means (Goebel et al., 1982). Endosulfan has a weak UV absorbence and does not absorb at all \260 nm. Therefore, photodegradation was unlikely to be an important route of detoxication. Fields studies with endosulfan have demonstrated that volatilisation is an important route of loss for this pesticide under field conditions (Singh et al., 1991). As the microcosms were aerated, this may have been a significant route of loss from the tanks. In aqueous systems endosulfan is believed to bind preferentially to the sediment (Peterson and Batley, 1991). Bacteria and fungi can degrade endosulfan to the sulfate or diol (Martens, 1976) and a significant amount of endosulfan may have been degraded by bacteria. Macrophytes are extremely tolerant to endosulfan and can metabolise it to endosulfan sulfate. Filamentous and blue-green algae can also absorb large quantities of endosulfan and metabolise it to the diol (Goebel et al., 1982). Significant amounts of endosulfan were found in the plant tissues at the end of the study, suggesting that this may have been an important route of degradation. As filamentous algae increased in the treated microcosms this may have dampened the biological effects of endosulfan. Very little a endosulfan remained in the microcosms at the end of the study despite this being the main component of the technical material (70%). Although the a endosulfan was degraded equally in sediments and macrophytes, b endosulfan appeared to degrade more readily in the macrophytes. Dying plant tissues may have also been an important source for the slow release of endosulfan back into the microcosms. In an earlier study, Barry (1996) noted that endosulfan could adsorb to algae and was an important sink for the pesticide in a simple flow-thru culture system containing Daphnia and the alga Selenastrum capricornatum. As the population density of Daphnia increased, the algae level decreased and the measured concentration of endosulfan in the water increased. Conversely, when the Daphnia population densities decreased, the alga population numbers increased and the measured concentration of endosulfan in the water decreased. This finding suggests endosulfan can undergo reversible binding

118

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

dynamics with the plant material. It has important biological implications because endosulfan adsorbed to algae has low toxicity to Daphnia (Barry et al., 1995). Thus, when zooplankton densities are low and large amounts of phytoplankton are present in the water, relatively little of the pesticide may be bioavailable to the organisms. But when zooplankton densities are high and phytoplankton levels are low, more endosulfan may be free to exert a toxic impact on the biota. Several species were eliminated or were severely reduced by endosulfan in the medium and high treatments. The main species to suffer the direct effects of endosulfan were the ostracods, calanoids and to a lesser degree Ceriodaphnia. The present study supports the finding of Lee (1979)-cited in Peterson and Batley (1991) that ostracods are more sensitive to endosulfan than cladocerans, but contrasts with his conclusion that copepods are less sensitive than either species. The calanoid copepods in the tanks were eliminated at 10 mg l − 1 endosulfan indicating that they were more sensitive than the cladocerans and at least as sensitive as the ostracods. Cyclopoid copepods are regarded as being tolerant of endosulfan; Lu¨demann and Neumann (1960) cited in NRCC (1975) reported no mortality at 100 mg l − 1 after 96 h for Cyclops strenus. It is possible that the copepods referred to by Lee were also cyclopoids which would indicate that there are significant differences in tolerance to endosulfan within the class copepoda with calanoids being more sensitive than cyclopoids. There was a significant effect of endosulfan on the cladoceran Ceriodaphnia sp. at the highest pesticide concentration. The mean population density of Ceriodaphnia in the high treatments were lower than other treatments on several occasions and population fluctuations appeared damped in the second half of the study. The 48 h EC50 for endosulfan to cladocerans range from 0.3 mg l − 1 for Daphnia longispina (Magdza, 1983) to 62 mg l − 1 (Schoettger, 1970) and 740 mg l − 1 (Lemke, 1980) for Daphnia magna. Sunderam et al. (1992) reported 48 h EC50’s of 215 and 490 mg l − 1 for Moinadaphnia macleayi and Ceriodaphnia cf dubai, respectively, and chronic no observable effect concentrations of 20 and 10 mg l − 1. A concentration of 120 mg l − 1 caused reproductive impairment of D. magna Fernandezcasalderrey et al. (1993). The concentrations to which Ceriodaphnia were exposed in this experiment therefore fell in the range noted to cause sub-lethal responses in Cladocera. It has been suggested that large cladocerans may be more sensitive to pesticides than smaller ones (Gliwicz and Sieniawaska, 1986). Hanazato (1991) found a positive correlation between cladoceran size and sensitivity to the pesticide carbaryl in a microcosm study. This relationship clearly did not apply to endosulfan in the present experiment. Simocephalus is a large species, adults may reach 4 mm in length. They were clearly less sensitive to endosulfan than Ceriodaphnia which had a maximum size of B1 mm. The very small chydorids, however, were also quite tolerant to endosulfan. Size may have an important modifying effect on toxicity, but other factors such as routes of exposure, bioavailability of the toxicant to sensitive organ systems and the extent of detoxication pathways will be the major determinants of impact. Multi-dimensional scaling indicated a quantal shift in community composition, occurring between 1 and 10 mg l − 1 endosulfan. Ostracod and ceriodaphnid popula-

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

119

tions declined following addition of endosulfan suggesting that the pesticide was directly toxic to these groups. Simocephalus, tardigardes, phytoplankton, filamentous algae and possibly macrophyte biomass all increased following the addition of endosulfan, suggesting that these events were indirect effects of the toxicant. Correlational data suggested that the decline in ostracods was responsible for the growth of filamentous algae. The growth of filamentous algae and macrophytes, in turn, may have been responsible for lower nutrient levels in the treated microcosms. Orthophosphate and nitrite concentrations were also lower in the treated microcosms compared with controls. As Brock et al. (1995) found that vascular plants showed the greatest response to nutrient loading in aquatic microcosms and Van Donk et al. (1993) postulated that the aquatic plant Elodea played a key role in the cycling of nutrients in shallow ponds, the endosulfan-related changes in nutrient concentrations may have been a response to macrophyte and filamentous algae growth. The presence of Simocephalus, only in the medium and high endosulfan microcosms paralleled the development of filamentous algae. It is possible that filamentous algae is an important food source or substrate for the simocephalids and their appearance only in the 10 and 50 mg l − 1 endosulfan microcosms does suggest that they were being excluded by food limitation in the control tanks. The other animal taxa to show significant increase population growth in the pesticide treated microcosms were the water bears. These sediment-dwelling animals may have benefitted from increased food levels or reduced predation following the demise of the ostracods. There is little information available on the ecological effects of ostracods, however, Diner et al. (1986), found that ostracods can have a significant effect on production and community metabolism in simple laboratory microcosms. The results of the present study indicate that ostracods may have an important role in determining the structure of temporary pond systems. In this context ostracods may fit the keystone species concept of Paine (1969). Hallmarks of a keystone species are, firstly, that their presence is crucial to maintaining the organisation and diversity of their ecological communities. Secondly, it is implicit that these species are exceptional relative to the rest of the community in their impact (Mills et al., 1993). At least five classes of keystone species have been recognised: predator, prey, plant, link and modifier. Ostracods would clearly be classified as a keystone predator as grazing on filamentous algae had a major effect on the structure and diversity of the microcosm community. This study showed a significant impact of endosulfan at 10 mg l − 1. Concentrations of 0.44 mg l − 1 endosulfan have been reported from lagoons in cotton growing regions of New South Wales, Australia (Chapman et al., 1993), however, as the pesticide does not persist in the water column, peak concentrations were probably higher. A concentration of 1530 mg l − 1 endosulfan was found in ditch water in British Columbia, Canada, shortly after the application of the pesticide, indicating the potential for very high pulses of this pesticide to enter the environment (Wan, 1989). Although endosulfan is generally regarded as being less toxic to invertebrates than to fish (Chapman et al., 1993), the results of the present study demonstrate

120

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

that low concentrations of endosulfan can have a significant impact on invertebrate communities. Peterson and Batley (1991) studied the impact and fate of endosulfan on aquatic communities in 25-l microcosms. These systems were constructed using water and sediment collected from a recreational reservoir and contained a range of zooplankton and phytoplankton. Aquarium plants were added to the systems. Despite repeated dosing with endosulfan, Peterson and Batley found noticeable impacts only at 500 and 5000 mg l − 1. They concluded ‘‘This is quite promising from an environmental impact point of view as these levels are considerably higher than would be found even in cases of gross contamination.’’ (Peterson and Batley, 1991). The microcosms of Peterson and Batley were at least an order of magnitude less sensitive than the systems used in the present study. A major flaw with their microcosm experiment was very high between replicate variability at the start of the study. This high variability has greatly reduced the power of their experiment to discriminate significant environmental effects of endosulfan and has resulted in a significant under-estimation of the potential ecological impact of this pesticide. A volume of 3 l was selected for the present study as a compromise between the need to capture a sufficiently large subsample of temporary pond biota and the need for a small, easily reproducible experimental unit. The volume recommended for the ASTM standard microcosms is also 3 l (American Society for Testing Materials, 1992). Preliminary studies indicated that 1-l microcosms were two small because of high variability in plant biomass and patterns of community development in larger microcosms were similar to results from the 3-l tanks (Davies, 1996). One unforseen side effect of the small tank size used in the present study was a dilution of nutrients in the microcosms. Weekly sampling removed 80 ml or 2.7% of total microcosm volume, which resulted in the loss of 21.8% of the original water after 10 weeks. This dilution effect could be detected in the conductivity measurements but visual comparisons of controls with the four microcosms that were initially excluded from the study and therefore, not sampled, indicated similar degrees of macrophyte development and zooplankton communities. The microcosms contained representatives of all the major classes of microcrustacea noted by Lake et al. (1989) at the original sediment collection site. In total, however, only 50% of the microcrustacea species identified by Lake over their 12 month sampling period were found in the present study. Many of these species may not have been present in the subsample of sediment used for this study, while others may not have emerged or prospered because conditions were unfavourable to them. The ephippia of Daphnia carinata were frequently encountered in the microcosm sediment, but this species was never collected in the tanks. As Lake et al. (1989) only collected Daphnia at the field site between August and October, winter and spring months in the southern hemisphere, the microcosm water temperature of 20°C may have been too warm for them to develop. Shield shrimps and concostracans are common temporary pond inhabitants but were actively removed from the microcosms. Both species are grazers and possibly detritivores and may have influenced community development. As this study relied on the organisms which produced resting eggs in the sediment, the insect fauna which dominate many

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

121

temporary ponds in late successional stages were completely absent. Subsequent studies have investigated the impact of adding insect predators of family Notonectidae to the microcosms to mimic the natural invasion of a higher trophic level (Davies, 1996). The development and eventual dominance of the plant community did, however, mirror the pattern observed at the field site and as noted by Yule (1982) and Lake et al. (1989) the flora consisted of mainly native species. This study demonstrated that temporary pond microcosms are a useful system for ecosystem level evaluation of toxicants. They combine a relatively high degree of biological complexity with low levels of between replicate variability and have the advantage of being easily constructed and maintained. Although microcosm studies have many advantages, the significance of any conclusions drawn from studies should be kept firmly in perspective. ‘‘There is cognitive danger that the microcosm (rather than the ecological system) will become the object of study, leading to needless confusion as results are overinterpreted and overextended’’ (Carpenter, 1986). The value of microcosm studies has been hotly debated in recent scientific literature. Several authors have argued that microcosm studies offer the ability to test ecosystem level hypotheses that would otherwise not be practicable to test on a larger scale (Lawton, 1996). Microcosms have played a central role in the development of contemporary ecological concepts (Gause, 1934; Huffaker, 1958; Tilman, 1977), however, others have emphasised the serious limitations of microcosm studies (Carpenter, 1986). When microcosms are used in conjunction which single-species and mechanistic studies, together with environmental monitoring and risk assessment they can be a powerful tool for estimating nature and intensity of any potential risks of chemicals to the environment. Acknowledgements This study was inspired by the work of Mark Perry on temporary ponds. The authors wish to thank Professor Jorma Ahokas and Assoc. Professor Doug Holdway for support and encouragement for this project from its inception. We also wish to thank Victoria Haritos for technical advice on analytical methods for endosulfan. Arthur Paul identified the algae and macrophyte species and Dr Chris Wilson provided identifications for the cladocerans. Finally, we wish to thank Rae Hall for carefully proof reading the manuscript. The study was supported by a small Australian Research Council grant. References American Society for Testing Materials, 1992. Standard Practice for Standardized Aquatic Microcosms: Freshwater. ASTM Annual Book of Standards. ASTM, Philadelphia, PA, pp. 1048 – 1082. Barry, M.J., 1996. The effects of an organochlorine pesticide on different levels of biological organisation in Daphnia. Ecotoxicol. Environ. Saf. 34, 234 – 251.

122

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

Barry, M.J., Logan, D.C., Ahokas, J.T., Holdway, D.A., 1995. The effects of algal food concentration on toxicity of two agricultural pesticides to Daphnia carinata. Ecotoxicol. Environ. Saf. 32, 273 – 279. Bloomquist, J.R., Soderlund, D.M., 1985. Neurotoxic insecticides inhibit GABA-dependent chloride uptake by mouse brain vesicles. Biochem. Biophys. Res. Comm. 133, 37. Brock, T.C.M., Roijackers, R.M.M., Rollon, R., Bransen, F., Van der Heyden, L., 1995. Effects of nutrient loading and insecticide application on the ecology of Elodea-dominated freshwater microcosm. 11. Responses of macrophytes, periphyton and macroinvertebrate grazers. Arch. Hydrobiol. 134, 53–74. Cairns, J. Jr., 1983. Are single species toxicity tests alone adequate for estimating hazard? Hydrobiologia 100, 47–57. Cairns, J. Jr., 1986. The myth of the most sensitive species. Bioscience 36, 670 – 672. Carpenter, S.R., 1986. Microcosm experiments have limited relevance for community and ecosystem ecology. Ecology 77, 677–680. Chapman, J.C., Napier, G.M., Sunderam, R.I.M., Wilson, S.P., 1993. The contribution of ecotoxicological research to environmental protection. Aust. Biol. 6, 72 – 81. Clesceri, L.C., Greenberg, A.E., Trussel, R.R., 1989. Standard methods for the examination of water and wastewater. American Public Health Association, Washington DC. Davies, W.R., 1996. The effect of Notonecid predation and endosulfan on temporary period microcosms used for aquatic toxicity testing. B. Appl. Sci. Honours Thesis. RMIT-University, Australia, 59 pp. Diner, M.P., Odum, E.P., Hendrix, P.F., 1986. Comparison of the role of ostracods and cladocerans in regulating community structure and metabolism in freshwater microcosms. Hydrobiologia 133, 59 –63. Eldefrawi, M.E., Eldefrawi, A.T., 1989. Insecticide actions on GABA receptors and voltage-dependent chloride channels. In: Narahashi, T., Chambers, J.E. (Eds.), Insecticide Action. From Molecule to Organism. Plenum, New York, pp. 1 – 11. Fernandezcasalderrey, A., Ferrando, M.D., Andreumoler, E., 1993. Effects of endosulfan on survival, growth and reproduction of Daphnia magna. Comp. Biochem. Physiol. C. 106, 2437 – 2441. Gause, G.E., 1934. The struggle for existence. Williams and Wilkins, Baltimore. Gliwicz, M.Z., Sieniawaska, A., 1986. Filtering activity of Daphnia in low concentrations of a pesticide. Limnol. Oceanogr. 31, 1132–1138. Goebel, H., Gorbach, S., Knauf, W., Rimpau, R.H., Huttenbach, H., 1982. Properties, effects, residues and analytics of the insecticide Endosulfan. Res. Rev. 83, 1 – 174. Hanazato, T., 1991. Effects of repeated application of carbaryl on zooplankton communities in experimental ponds with or without the predator Chaoborus. Environ. Pollut. 74, 309 – 324. Huffaker, C.B., 1958. Experimental studies on predation, dispersion factors and predator – prey oscillations. Hilgardia 27, 343–383. Kenaga, E.E., 1987. Methods for assessing the effects of mixtures of chemicals on non-human biota as applied to specific taxonomic representatives of individuals or groups of species. In: Vouk, V., Butler, G., Upton, A., Park, D., Asher, S. (Eds.), Methods for Assessing the Effects of Mixtures of Chemicals. Wiley, New York, pp. 395 – 408. Kimball, K.D., Levin, S.A., 1985. Limitations of laboratory bioassays: the need for ecosystem-level testing. Bioscience 35, 165–171. Lake, P.S., Bayly, I.A.E., Morton, D.W., 1989. The phenology of a temporary pond in western Victoria, Australia, with special reference to invertebrate succession. Arch. Hydrobiol. 115, 171 – 202. Lawrence, L.J., Casida, J.E., 1984. Interactions of lindame, toxaphene and cyclodienes with brain specific t-butylbicyclophosphorothionate receptor. Life Sci. 35, 171 – 17X. Lawton, J.H., 1996. The ecotron facility at Silkwood Park: the value of ‘big bottle’ experiments. Ecology 77, 665–669. Lee, K.W., 1979. The effects of thiodan, paraquat, 2,4-D and sevin 85 on the planton community and fish in a ricefield system. M.Sc Thesis. University of Malaysia, Kuala Lumpar, Malaysia, 272 pp.

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

123

Lemke, A.E., 1980. Comprehensive report. Interlaboratory comparison of acute testing set. In: US EPA Environmental Research Laboratory, Duluth, MN. Ambient water quality criteria for endosulfan. US Environmental Protection Agency, Washington, DC. van Lu¨demann, D., Neumann, H., 1960. Versache u¨ber die ahde toxische Wirlzong neuzeitlicher Kontakinsektizide auf Su¨sswassertiene. 2 Beitrag. Z. Angew. Zool. 47 (3), 303 – 321. Magdza, C.H.D., 1983. Toxicity of endosulfan to some aquatic organisms of Southern Africa. Zimb. J. Agric. Res. 21, 159–165. Martens, R., 1976. Degredation of Endosulfan-8-9-14C by soil microorganisms. Appl. Environ. Microbiol. 31, 853–858. Meyer Jr., F.L., Ellersieck, M.R., 1986. Manual of acute toxicity: interpretation and data base for 410 chemicals and 66 species of freshwater animals. Resource Publication 160. US Department of Interior, US Fish and Wildlife Service, Washington, DC. Mills, L.S., Soule´, M.E., Doak, D.F., 1993. The keystone species concept in ecology and conservation. Bioscience 43, 219–224. National Research Council of Canada, 1975. Endosulfan: its effects on environmental quality. Report No. 11. NRCC/CNRC Ottowa. Neill, W.E., 1975. Experimental studies on microcrustacean competition, community composition and efficiency of resource utilisation. Ecology 56, 809 – 826. Odum, E.P., 1969. The strategy of ecosystem development. Science 164, 262 – 270. Paine, R.T., 1969. A note on trophic complexity and community stability. Am. Nat. 103, 91 – 93. Perry, M.R., 1981. Some laboratory and field studies of the dynamics of small aquatic ecosystems, with particular reference to the role of predation. B.Sc. Honours Thesis. Monash University, Clayton, Victoria. Perry, J.A., Troelstrup, N.H. Jr., 1988. Whole ecosystem manipulation: A productive avenue for test research? Environ. Toxicol. Chem. 7, 941 – 951. Peterson, S.M., Batley, G.E., 1991. Fate and transport of endosulfan and diuron in aquatic ecosystems. Investigation report CET/LH/IR013. CSIRO Division of Coal and Energy Technology. Peterson, S.M., Batley, G.E., 1993. The fate of pesticides in Australian rivers. Chem. Aust.-Aquat. Environ. Chem. Suppl. 8, 395–397. Schoettger, R.A., 1970. Toxicology of thioden in several fish and aquatic invertebrates. U.S.D.I. Bureau of Sports and Fisheries. Wildl. Invest. Fish Control. 35, 1 – 31. Shaw, J.L., Manning, J.P., 1996. Evaluating macroinverebrate population and community level effects in outdoor microcosms: use of in situ bioassays and multivariate analysis. Environ. Toxicol. Chem. 15, 608–617. Singh, N.C., Dasgupta, T.P., Roberts, E.V., Mansingh, A., 1991. Dynamics of pesticides in tropical conditions. 1. Kinetic studies of volatilization, hydrolysis and photolysis of dieldrin and a and b endosulfan. J. Agric. Food Chem. 39, 575– 579. Strickland, J.D., Parsons, T.R., 1972. A practical handbook of seawater analysis, 2nd ed. Bull. Fish Res. Bd. Can. 167, 1–203. Sunderam, R.I.M., Thompson, G.B., Chapman, J.C., Cheng, D.M.H., 1992. Acute and chronic toxicity of endosulfan to two Australian cladocerans and their applicability in deriving water quality criteria. Arch. Environ. Contam. Toxicol. 27, 541 – 545. Taub, F.B., Read, P.L., Kindig, A.C., Harrass, M.C., Hartmann, H.J., Conquest, L.L., Hardy, F.J., Munro, P.T., 1983. Demonstration of the ecological effects of streptomycin and malathion on synthetic aquatic microcosms. In: Bishop, W.E., Cardwell, R.D., Heidolph, B.B. (Eds.), Aquatic Toxicology and Hazard Assessment, 6th Symp. ASTM STP 802. ASTM, Philadelphia, pp. 5 – 25. Tilman, D., 1977. Resource competition between planktonic algae: an experimental and theoretical approach. Ecology 58, 338–348. Van Donk, E., Gulati, R.D., Iedema, A., Meulemans, J.T., 1993. Macrophyte-related shifts in the nitrogen and phosphorus content of the different trophic levels in a biomanipulated shallow lake. Hydrobiologia 251, 19–26.

124

M.J. Barry, D.C. Logan / Aquatic Toxicology 41 (1998) 101–124

Van Donk, E., Prins, H., Voogd, H.M., Crums, S.J.H., Brock, T.C.M., 1995. Effects of nutrient loading and insecticide application on the ecology of Elodea-dominated freshwater microcosms. 1. Response of plankton and zooplanktivorous insects. Arch. Hydrobiol. 133, 417 – 439. Wan, M.T., 1989. Levels of selected pesticides in farm ditches leading to rivers in the lower mainland of British Columbia. J. Environ. Sci. B24, 183– 203. Wilkinson, L., 1989. SYSTAT: The System for Statistics. SYSTAT, Evanston, IL. Yule, C.M., 1982. Studies on the biology of Dinotoptera species (Plecoptera: Gripopterygidae). In: Victoria, M. Sc. Thesis, Department of Zoology, Monash University, Clayton, Victoria. Zar, J.H., 1984. Biostatistical Analysis. 2nd ed. Prentice-Hall, Englewood Cliffs, NJ.

.