ARTICLE IN PRESS Ecotoxicology and Environmental Safety 73 (2010) 360–369
Contents lists available at ScienceDirect
Ecotoxicology and Environmental Safety journal homepage: www.elsevier.com/locate/ecoenv
Toxicity of selected pesticides to freshwater shrimp, Paratya australiensis (Decapoda: Atyidae): Use of time series acute toxicity data to predict chronic lethality Kumar A.a,n, Correll R.a,b, Grocke S.a, Bajet C.c a
CSIRO Land and Water, PMB 2, Urrbrae 5064, Australia Rho Environmetrics, PO Box 366, Highgate 5063, Australia c ˜os College, Laguna 4031, Philippines National Crop Protection Center, University of the Philippines at Los Ban b
a r t i c l e in f o
a b s t r a c t
Article history: Received 15 July 2008 Received in revised form 1 September 2009 Accepted 1 September 2009 Available online 14 November 2009
Toxicity of six pesticides (carbaryl, chlorpyrifos, cypermethrin, dimethoate, diuron and fenarimol) to the freshwater shrimp, Paratya australiensis was assessed after 96 h exposures. Of the six pesticides tested, alpha cypermethrin was the most toxic to the shrimp followed by chlorpyrifos, carbaryl, dimethoate, fenarimol and diuron. Regression methods for predicting chronic toxicity (lethality) from acute lethality data with shrimp were developed and compared, and it was found that the log–log model gives the most reliable predictions of the probability of death as a function of extended exposure times. Based on this model, chronic toxicity (21 days) to P. australiensis was estimated as 0.0058 mg/L for alpha cypermethrin, 4.9 mg/L for carbaryl, 0.004 mg/L for chlorpyrifos, 89 mg/L for dimethoate, 240 mg/L for diuron and 1500 mg/L for fenarimol. Acute LC10 values were also useful predictors of the chronic lethality. The log–log model was used to derive extrapolated chronic values that were compared to measured experimental chronic values for two fish species. The predictions of chronic toxicity based on acute toxicity data were found to give credible results for both fish species. These predictions of chronic toxicity can therefore be used in ecological risk assessments to fill in gaps with reasonable confidence where no measured estimates of chronic toxicity are available. Crown Copyright & 2009 Published by Elsevier Inc. All rights reserved.
Keywords: Pesticide Shrimp Time series Acute toxicity Prediction Chronic toxicity Extrapolation Log–log model
1. Introduction Hazards posed by indiscriminate and widespread use of a variety of pesticides have attracted global attention. Studies on the toxicity of pesticides to both aquatic and terrestrial non-target organisms have been undertaken in an effort to meet the demand for data pertaining to the environmental safety of these toxicants. In recent years, acute toxicity tests involving aquatic invertebrates have been used for ecotoxicological evaluation (Goss and Sabourin, 1985; Vittozi and De Angelis, 1991; Abdullah et al., 1994; Calleja et al., 1994; Phyu et al., 2004, 2005). Considering the varying degrees of sensitivity among the numerous test criteria and test species available, it is important to detect adverse effects of contaminants to organisms belonging to different trophic levels to obtain a measure of the potential impact on an ecosystem. The majority of insecticides currently in use are organophosphate (OPs), carbamate and synthetic pyrethroid compounds.
n
Corresponding author. Fax: + 61 8 8303 8565. E-mail address:
[email protected] (A. Kumar).
These compounds are widely used because of their relatively nonpersistent characteristics in the environment. Although these compounds offer the advantage of rapid degradation in the environment, they generally lack target specificity and have high acute toxicity toward many non-target invertebrate and vertebrate species. Aquatic invertebrates have been shown to exhibit a wide range of sensitivities to OPs, carbamate and pyrethroid pesticides and have been used as biomonitors for pesticide contamination. Data from various field studies have indicated that crustaceans are among the first taxa to either diminish in abundance or disappear from benthic communities when exposed to pollution (Long et al., 2001). Due to their sensitivity to xenobiotic compounds, the freshwater shrimp, P. australiensis (Decapoda: Atyidae), has been used a bioindicator to assess effects of contaminants in the Australian environments (Abdullah et al., 1993, 1994; Olima et al., 1997; Phyu et al., 2005). In Australia, diuron has around 2200 registered uses, including the application to croplands, orchards, plantations, nurseries and commercial areas. It is also used to control weeds in flood mitigation channels. This pesticide has been detected frequently in surface waters of the Murray–Darling Basin, usually between
0147-6513/$ - see front matter Crown Copyright & 2009 Published by Elsevier Inc. All rights reserved. doi:10.1016/j.ecoenv.2009.09.001
ARTICLE IN PRESS KumarA et al. / Ecotoxicology and Environmental Safety 73 (2010) 360–369
0.2 and 3.0 mg/L (Boey and Cooper, 1996). Toxicity of diuron to native invertebrate species such as the freshwater shrimp has not previously been researched. Chlorpyrifos has over 1200 registered uses on around 100 crops (NRA 1997), including cotton, wheat, vegetables and fruit. Chlorpyrifos is also used as a pesticide on pasture lands, machinery, stored grain and hides and also has domestic uses in Australia such as pesticide applications to lawns, gardens and pets. Abdullah et al. (1993) investigated inhibition and recovery of acetylcholinesterase activity (AChE) in P. australiensis during chlorpyrifos exposures. Severe toxicity was observed in P. australiensis causing 495% AChE inhibition within 30 min exposure at concentrations Z10 mg/L. Dimethoate is a phosphorodithioate OP pesticide used in Australia for control of a variety of pests including aphids and mites in a range of food crops, cotton and tobacco (Tomlin, 1994). Currently, there is lack of Australian ecotoxicological studies on dimethoate. However, a water quality guideline for its use in Australia has been derived from the overseas data (ANZECC and ARMCANZ, 2000). There is very limited or no Australian ecotoxicological data on carbaryl, alpha cypermethrin and fenarimol. Of these three pesticides, fenarimol is used as a fungicide. Carbaryl is a broadspectrum carbamate pesticide used in Australia against a range of insects, mites, lice, millipedes and other pests (APVMA, 2006). It is used in a diverse range of situations encompassing a wide range of agricultural crops (ornamentals, fruit and vegetables). Alpha cypermethrin is a pyrethroid insecticide registered in Australia for application to fruit, grains and vegetables. Water quality guidelines have not been developed for these three pesticides (ANZECC and ARMCANZ, 2000). Therefore there is a need for toxicity data on local species to develop water quality guidelines for these pesticides. While there is a general lack of acute toxicity data, especially for Australian species, there are even less data on chronic toxicity of these pesticides, required for sound management of the environment. Because of the large number of chemicals, high cost and long duration of chronic tests, resources are insufficient to obtain experimental information about long-term impacts for all these chemicals over a range of native species. Acute data is considered inadequate for developing protective water quality guidelines. The Australian guidelines use experimentally derived chronic toxicity data to derive their high reliability guidelines, and extrapolated chronic toxicity data to derive their medium and low reliability guidelines, and do not use acute toxicity data in its raw form to derive guidelines at all. Thus it is of great interest to relate acute and chronic toxicity of chemicals and develop statistical techniques that can predict chronic toxicity based on data from acute toxicity experiments. The main objectives of this study were twofold: (i) to determine the acute toxicity of pesticides such as carbaryl, chlorpyrifos, alpha cypermethrin, dimethoate, diuron, and fenarimol to Australian freshwater shrimp, P. australiensis, and (ii) to predict the chronic lethality of these pesticides to P. australiensis based on the regression analyses of the acute toxicity data and compare against the measured values, where possible.
361
presented real data examples for predicting chronic toxicity from acute toxicity results based on analyses of acute/chronic ratios (ACRs). The AF is the ratio of the maximum acceptable concentration (MATC) to the acute median lethal concentration (LC50). The MATC is a chronic measure calculated as the geometric mean of the no-observed effect concentration (NOEC) and the lowest observed effect concentration (LOEC) (Rand, 1995). An ACR is obtained by dividing an acute measure for a particular organism for a particular toxicant by a chronic measure for the same toxicant and organism (Rand, 1995; Warne, 1998; Raimondo et al., 2007). Therefore, an ACR can be derived from dividing any acute measure with any chronic measure, and is not restricted to just the MATC and LC50. There are problems with the use of ACRs (Lange et al., 1998; Warne, 1998; Calow and Forbes, 2003). Firstly, the loose definition of an ACR, where an ACR can be derived using an NOEC and a LC50 for example, but then can be applied to MATCs or LOECs to derive PNOECs. Secondly, the numerical values of ACRs are dependant on the biological endpoints measured in the toxicity data used to derive the ACRs. There is no sound theoretical basis for their use (Warne, 1998). Time course distinguishes acute and chronic toxicity and also relates them in view of the whole toxicity process as an integrated and progressive development through time of exposure. A time to response approach gives a better understanding of toxic effect over time and survival time has shown applicability in ecotoxicological investigations (Newman and Aplin, 1992 and Lee et al., 1995). According to Newman and Aplin (1992) a survival time modelling extracts more information from toxicity data than routine end point methods but does not preclude estimation of standard endpoints. They used survival time modelling and advocated using the hazard function of Cox and Oakes (1984). Their concept has been developed by Sun et al. (1995), who provided a mechanism for using data from 24, 48, 72 and 96 h trials to estimate the parameters that relate concentration and exposure time to the probability of survival. Their approach can be summarised in their Eq. (6), where they describe the quantile surviving as Q ðt; xÞ ¼ exp ½axb t c where x is the concentration of the toxicant and t the exposure time and a, b and c are constants that are estimated from the acute exposure data. Their method can be used to produce confidence intervals of the survival rates at any nominated time. Those authors chose to use the 1% death rate at 90 days as their approximation to MATC. Use of the technique of Sun et al. (1995) requires revisiting the raw data used in the establishment of the acute toxicity estimates. While the technique of Sun et al. (1995) is very attractive, implicit is the concept of the product of exposure time and concentration. Their equation implies that at very large times and non-zero concentration there would be zero survival. In other words, all very long-term chronic LC50s would be zero. Lee et al. (1995) used a multifactor probit analysis to derive the relationship between the quantile affected, time and concentration. They considered a range of models but found the model Probit P ¼ a þ b logðCÞ þ d=t
2. Statistical models used for predicting chronic toxicity from acute toxicity data There are several methods available in the toxicological literature for predicting chronic toxicity (typically 496 h) from acute toxicity (typically r96 h) data. For example, Mount and Stephan (1967) introduced a procedure for estimation of chronic toxicity by using an application factor (AF) and Kenaga (1982)
(where C is the concentration and t the exposure time) had the lowest heterogeneity factor. Use of that model must also take into account correlations induced by repeated measures on the same animals during the trial. Other workers have based their extrapolations on the published LC50s for various times (Hemming et al. 1989; Lee et al., 1995), where LC50 data for various exposure times (from acute toxicity tests) are extrapolated to chronic toxicity. Lee et al. (1995)
ARTICLE IN PRESS 362
KumarA et al. / Ecotoxicology and Environmental Safety 73 (2010) 360–369
list 8 formulae that have been used for extrapolation. Ideally the formula should include a logarithmic function of the concentration as this is typically used in ecotoxicological trials. The logarithmic scale has the advantage that when back transformed, the results are guaranteed to be positive, thus avoiding nonsensical negative concentrations. A further useful feature is to have the reciprocal of t in the prediction equation. This term has the natural interpretation of representing chronic toxicity when this term is zero for as t is then interpreted as being very large. Both these features are displayed in Eq. (7) of Lee et al. (1995), namely ln ðYÞ ¼ a þb=t where Y could represent an LC50. It should also be noted that the probit equation of Lee et al. (1995) given above includes both the logarithmic concentration and the reciprocal term in time. The use of extrapolations based on LC50 rather than LC10 or LC05 has been recommended by Slaughter et al. (2007). Slaughter et al. (2007) argued that LC50s had narrower confidence intervals and are model independent as compared to low-effect LCx values where there is more uncertainty.
3. Materials and methods 3.1. Pesticide standards and other chemical reagents All pesticides used in this study were technical material ( 497% purity). Stock and working solutions of pesticides were prepared in analytical-grade acetone (99% purity) as a carrier solvent. All stock solutions were stored at 4 1C. All working solutions were made immediately prior to use. 3.2. Shrimp collection and maintenance of cultures Freshwater shrimp (P. australiensis) were collected from the Finniss River area located approximately 70 km to the southeast of Adelaide. This is a pristine site in the natural reserve area. Pesticides were not detected in the sediments and waters collected from the Finnis River. On arrival at the laboratory the shrimp were acclimated in a 60 L aquarium at 237 1 1C and a light intensity of less than 800 lx at the surface of the water with a 16:8 h light-to-dark regime for at least 2 weeks prior to use in the test. The shrimp were fed Seramins tropical flake food twice a day. Shrimp were fed ad libitum during the acclimation period, but were not fed for 24 h prior to and during testing. Shrimp were acclimated in the dilution water and test tanks at the experimental temperature for at least 2 weeks before the experiments were run. 3.3. Shrimp toxicity tests The toxicity tests with the selected pesticides consisted of nine treatments including control, a solvent control. The pesticide concentrations used in the definitive experiments were based on range finding tests. The semi-static tests exposed the shrimp (10–15 mm long) for 96 h with replacement of the test solution every 24 h. The tests were conducted using three replicates of each treatment. Each replicate consisted of a 1 L beaker that contained 800 mL of test solution and then randomly allocated shrimp. The beakers were covered with plastic film to reduce volatilization and then randomly allocated in an incubator set at 237 1 1C with a light intensity of approximately 800 lx at the surface of the water and a 16:8 h light-to-dark regime. The test end point was survival. Daily measurements were made of pH, conductivity and dissolved oxygen. The numbers of dead shrimps were counted and immediately removed every 12 h until the end of the test at 96 h. Mortality was defined as lack of movement or response to gentle prodding (ASTM, 1998). The toxicity tests were considered valid if the mortality in each of the water and solvent controls was not greater than 10% and the DO level in all the test solutions remained greater than 60% saturation. 3.4. Fish fry bioassay for 28 days for validation of chronic lethality predictive model Newly hatched larvae of Murray rainbowfish, Melanotaenia fluviatilis (Atheriniformes: Melanotaenidae) were obtained from Ausfish, Queensland. Fish larvae were acclimatised to laboratory conditions for two weeks in 5-L glass aquaria before they were used for the bioassays. For bioassays, 500 mL glass beakers
containing 400 mL of test solutions were arranged at random in the incubator and fish fry were gently added to each beaker using random numbers. The incubator was set at 247 1 1C with a light intensity of approximately 800 lx at the surface of the water and a 16:8 h light-to-dark regime. Fish fry were exposed to 1, 5, 10, 20, 40, 160 and 120 mg/L chlorpyrifos for 28 days. Negative and positive solvent controls were run concurrently. There were four replicate test beakers for each concentration with five fish fry in each beaker. In total, 20 fish fry were exposed to each pesticide concentration. The tests solutions were renewed every Monday, Wednesday and Friday for 4 weeks. The test beakers were covered with plastic film and were not aerated during experimentation to reduce chlorpyrifos evaporation. Larval fish fry were fed 250 mL of 0.8 g/L ‘Liquid Small Fry Food’ (Wardley’s) per day. Fish fry were fed during acclimation and experimentation period. Each beaker was checked every 24 h under a dissecting microscope for dead organisms. Death was confirmed when no response was recorded when organisms were gently prodded with a stainless steel probe and the heart was no longer beating. Water quality parameters were maintained with dissolved oxygen 460% saturation, conductivity 1000–1500 mS/cm, pH 7–8.5 and temperature 24.370.8 1C. Fish mortality was observed for 28 days.
3.5. Pesticide analyses Water samples were collected from all treatments before and after the renewal of the test solutions to measure the concentration of pesticides during various exposures. Pesticides were extracted from 500 mL of water samples by solid-phase extraction (SPE) using Supelco Supelclean Envi-C18 6 cm3/1 g SPE cartridges. The cartridges were conditioned with 3 mL methanol then 3 mL Milli-QTM water before running the water samples. Elution was with 3 1 mL aliquots of acetonitrile into tapered glass vials. The eluent was evaporated down to 1 mL under a gentle stream of N2 and then made up to a final volume of 2 mL in acetonitrile. Samples to be run on GC–MS were evaporated down to dryness and prepared to a final volume of 1 mL in hexane. All samples were vortex mixed prior to their transfer into vials. Chlorpyrifos extracts were analysed by GC–MS. Carbaryl, diuron, dimethoate and fenarimol were analysed using an Agilent 1100 HPLC-DAD. Alpha cypermethrin levels were not measured during this experiment due to very low concentrations used in the bioassays. All HPLC analyses were run on an Agilent 1100 Series HPLC with a quaternary pump using an Alltech Alltima C18 5 mm 250 4.6 mm ID column with an Alltech Alltima C18 5 mm 7.5 4.6 mm ID guard column. Detection wavelength was set at 220 nm using a UV diode array detector (Agilent). Detection times for the pesticides studied are listed in Table 1. The mobile phase was acetonitrile and water with varying ratios for individual pesticides at a flow rate of 1 mL/min. Chlorpyrifos in lower concentration levels was analysed by an Agilent 6890 Series GC system with an Agilent 5973 Network Mass Selective Detector. GC–MS conditions were as follows: column, a fused silica J&W HP5-MS 30 m length 0.25 mm I.D. 0.25 mm film thickness. The column temperature was programmed at 60 1C at 0 min, held for 30 s, then ramped to 280 1C over 6 min at a ramp rate of 40 1C/min and held for 3 min at 280 1C. The injector temperature was 230 1C with pulsed splitless injection. Carrier gas helium set at a flow rate of 1.1 mL/min. Table 1 summarises the pesticide compounds studied and the mobile-phase ratios and retention times for analysis. The recovery efficiency of pesticide spiked water samples was always greater than 90% for all pesticides analysed under this investigation. The minimum detectable quantity of the pesticides in the water samples (detection limit, DL) was estimated as the lowest standard sample/ concentration factor. In the current investigation, DL for chlorpyrifos was 0.005 and 0.1 mg/L for carbaryl, fenarimol and diuron. For dimethoate, DL was 1 mg/L.
3.6. Calculations and statistical analyses The 24, 48, 72 and 96 h LC50 for pesticide exposures to shrimp was computed ¨ by trimmed Spearman Karber method of Hamilton et al. (1977, 1978). 14 and 28 day LC50 values for Murray rainbowfish during 28 days exposure to chlorpyrifos ¨ were also computed using the trimmed Spearman Karber method. Chronic toxicity was predicted from the acute toxicity data by three methods. The first was by a linear regression of the log LC50 values against the inverse of time of exposure (log-inverse method). As discussed by Mayer et al. (1994), the intercepts of these regressions represented zero reciprocal of time, which is equivalent to the inverse of infinite time. This was taken as being the value for chronic toxicity. That formula was then used to estimate chronic toxicity based on the acute (less than or equal to 96 h or 4 days) LC50 data. The all important assumption of linearity in the first model was tested by fitting a quadratic term to the regression. The second method was based on a regression of log LC50 against log of exposure time (log–log method). The slope of the regression b was then used to obtain acute (96 h= 4 d) to chronic ratio (ACR) use the formula ACRT ¼ exp ½bðln4 lnTÞ þ e
ARTICLE IN PRESS KumarA et al. / Ecotoxicology and Environmental Safety 73 (2010) 360–369
363
Table 1 Properties of pesticides studied in this investigation. Pesticide
Pesticide class
Mobile phase ratio (CH3CN:H2O)
Retention time (min) at 1 mL/min flow rate
Minimum detection limit (mg/L)
Alpha cypermethrin Carbaryl Chlorpyrifos Dimethoate Diuron Fenarimol
Pyrethroid Carbamate Organophosphate Organophosphate Phenylurea herbicide Pyrimidine fungicide
Not analysed 40:60 Not applicable 40:60 60:40 60:40
Not analysed 6.87 6.20 (GC–MS) 5.38 6.27 9
– 0.1 0.005 1.0 0.1 0.1
Table 2 Measured concentrations of pesticides before and after 24 h renewal of the test solutions in 96 h shrimp bioassays. Chlorpyrifos Nominal concentration (mg/L)
Carbaryl Measured concentration Nominal concentration (mg/L) (mg/L)
0.001 0.005 0.01 0.05 0.1 0.4 0.8
o DL o DL 0.0087 0.005 0.037 0.008 0.077 0.01 0.36 70.17 0.81 70.43
0.1 0.5 1.5 4.5 13.5 20
Dimethoate
Measured concentration (mg/L) oDL 0.22 7 0.03 0.90 70.18 2.67 7 0.56 8.52 7 0.87 15.5 7 1.4
Fenarimol Nominal concentration (mg/L)
Measured concentration (mg/L)
0.05 0.1 0.5 2 5 10 20
0.047 0.18 0.077 0.04 0.337 0.17 1.387 0.25 3.87 1.6 8.267 0.9 16.07 2.8
Diuron
Nominal concentration (mg/L)
Measured concentration (mg/L)
Nominal concentration (mg/L)
Measured concentration after 24 h (mg/L)
0.05 0.10 0.50 2 5 10 20
0.057 0.01 0.107 0.09 0.407 0.16 1.73 7 0.79 4.17 7 1.10 7.12 7 1.45 19.73 7 0.18
0.25 1 5 10 20
0.22 7 0.15 0.80 70.33 4.39 7 1.76 8.5 7 2.9 16.4 7 3.6
Values reported as measured concentrations; mean 7standard deviation, n= 8 replicates/pesticide concentration; o DL =below detection limit.
where T is the chronic time point measured in days and the random error e is assumed to have constant variance in each case. A variation on this method was used by Barata et al. (1999), where they modelled probits of mortality (P) using the equation P ¼ a þ b ln ðCÞþ
g ln ðTÞ
diuron and fenarimol after 24 h exposures were closer to the nominal concentrations (70–100%). 4.2. Toxicity data
þe
They restricted the data to those observations where the probits corresponded to mortalities between 10% and 90% and fitted the parameters by least squares. The third method used 10 as an ACR based on NOEC levels (ACR method).
4. Results During all 96 h bioassays, the dissolved oxygen measured was always greater than 80% saturation and pH ranged from 7.7 to 7.9. Conductivity of test solutions varied from 390 to 420 mS/cm in all tests and temperature remained quite consistent during various exposures (23.470.8 1C).
4.1. Pesticide concentrations Test solutions were renewed every 24 h and were analysed at regular intervals to compare the nominal versus measured concentrations. Measured concentrations of carbaryl at the end of 24 h were between 40% and 60% of the nominal concentrations (Table 2). Chlorpyrifos test concentrations also showed marked variation after 24 h exposures and measured concentrations varied between 60% and 100% of the nominal concentrations (Table 2). However, measured concentrations of dimethoate,
P. australiensis was highly sensitive to alpha cypermethrin with the lowest 96 h LC50 value of 0.019 mg/L. The 96 h LC50 values of chlorpyrifos and dimethoate to P. australiensis were 0.06 and 800 mg/L, respectively (Table 3). Fenarimol and diuron were the least toxic pesticides with 96 h LC50 values of 3400 and 8800 mg/L, respectively (Table 3). 4.3. Predicting chronic lethality Fig. 1 provides simple linear regressions of the log LC50 value regressed against the inverse of time of exposure. Various parameters from the regression analyses are given in Table 4. Because the regression used the log of the LC50 values, the intercept must be exponentiated to give the back-transformed estimate of chronic lethality. For chlorpyrifos the exponential of 3.59 is 0.028, which is then an estimate of chronic lethality. Estimates of lethal toxicity of the other pesticides are also given in Table 4. The assumption of linearity of the regression was tested by the inclusion of a quadratic term. The quadratic term was found to be significant for all six pesticides indicating the assumption of linearity was unacceptable. A variation on the regression model is ln ðLC50t Þ ¼ a þ b=t k þ e
ARTICLE IN PRESS 364
KumarA et al. / Ecotoxicology and Environmental Safety 73 (2010) 360–369
Table 3 96 h LC50 values for six pesticides during acute exposures to shrimp, Paratya australiensis. Pesticide
96 h LC50(mg/L)
Lower confidence interval (mg/L)
Upper confidence interval (mg/L)
Alpha cypermethrin Carbaryl Chlorpyrifos Dimethoate Diuron Fenarimol
0.019 12 0.063 800 8800 3400
0.017 7 0.052 670 7100 2800
0.025 15 0.072 980 10,000 4100
0.05 Cypermethrin
Chlorpyrifos
0.4 0.2 0.1
0.00
0.01
0.02 0.03 1/hours
0.03 0.02
0.04
0.00
0.01
0.02 0.03 1/hours
0.04
0.00
0.01
0.02 0.03 1/hours
0.04
0.00
0.01
0.02 0.03 1/hours
0.04
20
Dimethoate
Carbaryl
24
16 12
3000
1000
10 0.01
0.02 0.03 1/hours
0.04
6000 Diuron
Fenarimol
0.00
4000
5e+04
5e+03 3000 0.00
0.01
0.02 0.03 1/hours
0.04
Fig. 1. Chronic lethality prediction of selected pesticides based on acute toxicity data using log-inverse time extrapolation.
Table 4 Use of log-inverse time extrapolation method for estimation of chronic toxicity. Pesticide
Estimate of intercept
Predicted chronic toxicity (mg/L)
Modified estimate (mg/L)
Alpha cypermethrin Carbaryl Chlorpyrifos Dimethoate Diuron Fenarimol
4.11 2.30 3.59 6.04 8.55 8.01
0.016 10 0.028 421 5180 3020
0.0081 5.93 0.0054 115 620 1836
The modified estimate is based on a log-inverse of square root of time (k= 0.5).
where LC50t is the LC50 at time t and e is a random error. That model also has the property that the reciprocal of its intercept is a measure of chronic toxicity. Various values of k were tested and the best estimate of linearity was obtained using a very small value of k. The estimates of chronic toxicity were sensitive to the choice of k as illustrated by using a value for k= 0.5 (i.e. Ot) as shown in Table 4 and Fig. 2.
The log–log model (Eq. (8) of Lee et al., 1995), which was previously suggested by Heming et al. (1989), was found to give a near linear relationship (Fig. 3). The slopes of the regression lines, together with the ACRs are shown in Table 5. Based on this model, 21 days chronic lethality of P. australiensis were estimated as 0.0058 mg/L for alpha cypermethrin, 4.9 mg/L for carbaryl, 0.004 mg/L for chlorpyrifos, 89 mg/L for dimethoate, 240 mg/L for diuron and 1500 mg/L for fenarimol (Table 6).
ARTICLE IN PRESS KumarA et al. / Ecotoxicology and Environmental Safety 73 (2010) 360–369
Cypermethrin
Chlorpyrifos
0.500
0.050
0.005 0.00
0.05
0.10
0.15
0.04 0.02 0.01
0.20
0.00
0.05
Dimethoate
Carbaryl
20 15 10
0.05
0.10
0.15
0.00
Diuron
Fenarimol
0.15
0.20
0.15
0.20
100 0.05
0.10 1/hours
3000 2000 0.10
0.20
500
0.20
5000
0.05
0.15
2000
1/hours
0.00
0.10 1/hours
1/hours
0.00
365
0.15
5e+04 5e+03 5e+02
0.20
0.00
0.05
1/hours
0.10 1/hours
Fig. 2. Chronic lethality prediction of selected pesticides based on acute toxicity data using log-inverse square root time extrapolation.
Carbaryl
Chlorpyrifos
10
LC50
LC50
20
5
2e-02
5e-04 1
2
5 10 20 50 Exposure time (Days)
100
1
2
Cypermethrin
100
Dimethoate 2000 LC50
0.03 LC50
5 10 20 50 Exposure time (Days)
0.01
200 50 10
1
2
5 10 20 50 Exposure time (Days)
100
1
2
Diuron
100
Fenarimol 7000
LC50
1e+05
LC50
5 10 20 50 Exposure time (Days)
1e+03
3000 1000
1e+01 1
2
5 10 20 50 Exposure time (Days)
100
1
2
5 10 20 50 Exposure time (Days)
Fig. 3. Relationship between LC50 and time using a log–log scale.
100
ARTICLE IN PRESS 366
KumarA et al. / Ecotoxicology and Environmental Safety 73 (2010) 360–369
Table 5 Estimation of chronic toxicity using the log–log method, with extrapolation from 96 h LC50 to 21 and 90 days. Pesticide
Slope
Alpha cypermethrin Carbaryl Chlorpyrifos Dimethoate Diuron Fenarimol
21 days
0.72 0.53 1.67 1.32 2.17 0.51
90 days
ACR (21 days)
Estimate (mg/L)
ACR (96 days)
Estimate (mg/L)
3.3 2.4 16 8.9 37 2.3
0.0058 4.9 0.0039 89 240 1500
9.5 5.3 180 62 870 4.9
0.0020 2.2 0.00035 13 10 690
The slope is the slope of the log LC50 log time regression, the factor is the divisor required to convert the 96 h to the chronic toxicity and the estimate is the estimate of chronic toxicity. The extrapolations assume linearity of the regression.
Table 6 Summary of estimates of LC50, LC10, NOEC (based on 96 h exposure) and estimates of chronic toxicity based on the safety factor method, log-linear extrapolation and log– log extrapolation to 21 days exposure. Pesticide
96 h LC50 (mg/L)
LC10 (mg/L)
NOEC (mg/L)
Safety factor (mg/L)
Log-linear (mg/L)
Log–log (mg/L)
Alpha cypermethrin Carbaryl Chlorpyrifos Dimethoate Diuron Fenarimol
0.019 12 0.063 800 8800 3400
0.01 4.4 0.02 550 4700 1700
0.01 1.5 0.01 500 5000 2000
0.001 0.15 0.001 50 500 200
0.016 10 0.028 421 5180 3020
0.0058 4.9 0.0040 89 240 1500
Table 7 LC50 values obtained from three replicates of an acute toxicity experiment conducted on shrimp, Paratya australiensis to assess 96-h toxicity and the predicted chronic (21 days) toxicity to fenarimol. Exposure time
24 h 48 h 72 h 96 h Predicted 21 days
LC50 values (mg/L)
Table 8 Estimate of LC50 of rainbow fish exposed for different durations to chlorpyrifos. Time
Standard error
Replicate A
Replicate B
Replicate C
7.065 5.578 4.409 3.722 1.803
5.524 5.639 3.754 2.623 1.499
7.065 5.213 4.082 3.460 1.308
0.036 0.011 0.020 0.046 0.144
The standard errors are based on the variation among the replicates.
Chronic values for each pesticide were estimated by safety factor approach (dividing 96 h NOEC value by 10). These data, together with LC10 values are presented in Table 6. 4.4. Estimation of the standard error of the prediction The estimation of the standard error of the prediction from the log–log model is made difficult because typically the same animals are observed over time (24, 48, 72 and 96 h), so the observations are not independent. An estimate based only on a small number of degrees of freedom can be obtained by considering differences among the estimates of chronic toxicity based on each replicate. Data from 3 replicates of LC50 of P. australiensis to fenarimol are shown in Table 7. The predicted chronic toxicity values show good consistency with a mean of 1.330 and a standard error of 0.129.
24 h 48 h 72 h 96 h 14 days 28 days
LC50 (mg/L)
Prediction
Estimate
Lower
Upper
407 244 169 122 71 45
334 184 121 86 47 31
496 325 237 174 106 66
NA NA NA NA 44 24
Also given are predicted values for LC50 measured at 14 and 28 days as predicted from the acute toxicity data using the log–log method.
and 96 h data using the log–log method. The agreement of the prediction and observed values for the 14 and 28 days data is fair. The confidence interval of the observed LC50 at 14 and 28 days were available and are shown, but the prediction also has a standard error as discussed in Section 4.4. When both these sources of variation are taken into account, the predicted estimates of chronic toxicity are very credible. Where raw data are available, it would be possible to estimate the confidence interval by bootstrapping, where the bootstraps are based on sampling without replacement of the animals that were used in the trial.
4.5.2. Effect of carbaryl on Channelfish, Nuria danrica Data for acute and chronic mortality data are available for the effect of carbaryl on Channelfish, N. danrica (Abbasi and Soni, 1991). A regression of ln(LC50) against ln(t) of their data yielded the equation
4.5. Comparison of prediction with measured observation using log– log method
lnðLC50t Þ ¼ 11:05 0:4679lnðtÞ
4.5.1. Effect of chlorpyrifos on rainbowfish, M. fluviatilis A comparison of the measured and predicted and LC50 values are shown in Table 8. The predictions were based on the 24, 48, 72
The equation predicts 18,900 mg/L as the LC50 at 21 days. The lower bound (assuming independence of observations) for the predicted value estimated by this method is 14,500 mg/L. That estimates is slightly higher than the observed value of 12,590 mg/
ARTICLE IN PRESS KumarA et al. / Ecotoxicology and Environmental Safety 73 (2010) 360–369
L. Given that no estimate of the error of the observed value is available, the two measures of chronic toxicity are comparable.
5. Discussion 5.1. Comparison of the acute toxicity data P. australiensis, has been found to be a well-suited organism in ecotoxicological studies by Abdullah et al. (1993, 1994), Olima et al. (1997) and Phyu et al. (2004, 2005). In the present study, toxicity of six pesticides (carbaryl, chlorpyrifos, cypermethrin, dimethoate, diuron and fenarimol) to the freshwater shrimp was assessed after 96 h exposures. Of the six pesticides tested, alpha cypermethrin was the most toxic to the shrimp followed by chlorpyrifos, carbaryl, dimethoate, fenarimol and diuron. Accord´ ing to Sa nchez-Fortun and Barahona (2005), among aquatic invertebrates, the organisms that are most sensitive to pyrethroids are surface-dwelling insects, mayfly nymphs and some of the larger crustaceans and zooplanktons. Several other studies have also shown that aquatic invertebrates and fish are extremely sensitive to pyrethroids (Anderson, 1989; Mian and Mulla, 1992; Friberg-Jensen et al., 2003) with LC50 values for most pyrethorids found to be less than 1 mg/L. In the present investigation, 96 h LC50 and LC10 values for P. australiensis during alpha cypermethrin exposures were 0.02 and 0.01 mg/L. The effect of the pyrethroid insecticide alpha cypermethrin on a natural freshwater community was investigated by Friberg-Jensen et al. (2003) in small in situ enclosures over an 11-day period. According to the current investigation, alpha cypermethrin proved to be acutely toxic to crustaceans in enclosures receiving nominal alpha cypermethrin concentrations of 0.13 mg/L. No-ObservedEffect-Concentration (NOEC) and Median Effect-Concentration (EC50) for the total crustacean community and cladoceran and copepod subgroups ranged between 0.02–0.07 and 0.04–0.17 mg/ L, respectively, with copepods being less sensitive than cladocerans. Smith and Stratton (1986) also found lobsters and shrimps to be highly susceptible to pyrethroids. Even at non-lethal concentrations, there could be significant behavioural changes in aquatic invertebrates, such as in their ability to respond to tactile stimuli, which may affect their survival. In the present study, chronic lethality of alpha cypermethrin to P. australiensis was predicted to be observable at 0.016 mg/L. Such low concentrations of alpha cypermethrin could be expected in the aquatic environments receiving this insecticide as run-off or spray drift, although it is hard to quantify alpha cypermethrin concentrations at such low levels. In this study, chlorpyrifos exhibited severe toxicity to P. australiensis with a 96 h LC50 of 0.06 mg/L and a LC10 of 0.02 mg/L and with 100% mortality at concentrations above 1 mg/L within 24 h. According to Abdullah et al. (1993) chlorpyrifos exposure to P. australiensis at concentrations of 10 and 100 mg/L resulted in potent toxicity with exposed shrimp surviving less than 30 min of exposure. They also reported that exposure of shrimp at 0.05 mg/L of chlorpyrifos for 7 days resulted in significant mortalities in shrimp with 100% mortality of shrimp on the 19th day of exposure. These results are in accordance with the current investigation. Among the two OPs tested, dimethoate was found to be less toxic (96 h LC50 800 mg/L) to P. australiensis than chlorpyrifos. Roast et al. (1999) also observed greater toxicity of chlorpyrifos (96 h LC50 0.13 mg/L) to the Western European hyperbenthic leach, Neomysis integer than in comparison to dimethoate (96 h LC50 540 mg/L). There is very limited data on fenarimol toxicity to aquatic invertebrates. In the current study, fenarimol was moderately toxic to shrimp during 96 h acute exposures with 96 h LC50 and
367
LC10 values of 3.4 and 1.7 mg/L, respectively. Based on regression analyses of the time series acute toxicity data, chronic lethality could be observed in shrimp at 3 mg/L fenarimol. Recently, there have been studies reporting anti-ecdysteroidal activity of fenarimol to the crustaceans such as daphnids by lowering endogenous ecdysome levels and by delaying moulting in a concentrationdependent fashion (Mu and LeBlanc, 2002, 2004). Such physiological effects of fenarimol on other invertebrates such as shrimp should be further investigated as fenarimol could be impacting only invertebrates while not affecting other aquatic vertebrates. Diuron was found to be the least toxic herbicide to P. australiensis. Nebeker and Schuytema (1998) investigated acute and chronic effects of the herbicide diuron on several aquatic organisms. Survival and reproduction of Daphnia pulex was impacted during diuron exposure with 96 h and 7-day LC50 values of 17.9 and 7.1 mg/L, respectively. In the present study, P. australiensis exhibited greater toxicity to diuron exposure with a 96 h LC50 of 8.8 mg/L. Based on this study, toxicological endpoints such as 96 h LC50 and LC10 values for six commonly used pesticides in Australian agriculture could be derived. Currently there are no water quality guidelines available in Australia for pesticides such as carbaryl, dimethoate, fenarimol and alpha cypermethin. Such toxicity data on local species is highly beneficial in developing site-specific water quality guidelines. In addition, value of such data can be enhanced by using time-to death data following techniques such as survival time modelling as recommended by Newman and Aplin (1992). This approach can assist in predicting ecological risk of such contaminants at very low concentrations and over an extended period of exposure.
5.2. Comparison of chronic toxicity estimates Typically, acute toxicity tests with aquatic organisms provide lethality estimates for a series of toxicant concentrations at 24, 48, 72, and 96 h of exposure. In the current study, a statistical model based on regression analyses was developed that utilized acute toxicity data to establish the relation of lethality to toxicant concentration and exposure time for predicting chronic lethality. These predictions of chronic toxicity could be used in ecological risk assessments to fill in gaps with reasonable confidence where no estimates of chronic toxicity are available. The safety factor method is typically based on the NOEC concentration. Use of NOECs has been criticised by various workers (Crane and Newman, 2000) as they are dependent not only on the toxicity but also on the amount of replication and the choice of concentrations. It would be more satisfactory to base the safety factor method on an estimate that is less dependent on the experimental design, for example LC50 or LC10. Table 5 indicates that for some compounds (especially diuron) the ACR is much larger than 10, which is the usual safety factor used to convert acute to chronic toxicities. For these compounds a larger safety factor would be required. The methods outlined in this paper indicate how an estimate could be obtained. In the present study, a different approach from the ACR concept is to model the effect of time on survival or on LC50. The linear extrapolation based on the regression of log LC50 against inverse of time is very attractive, as it does not require a further definition of chronic toxicity. Unfortunately as seen in Fig. 1 the critical assumption of linearity is not tenable. This is in contrast to the findings of Mayer et al. (1994), who found the relationship between log LC50 and inverse of time to be linear. Based on our data, the log-inverse method cannot be recommended. The variation where the regression is against a reciprocal of a power of exposure time requires an arbitrary choice of the power. The
ARTICLE IN PRESS 368
KumarA et al. / Ecotoxicology and Environmental Safety 73 (2010) 360–369
appeal of 1/t being zero for chronic toxicity is also shared by 1/tk, where k is any positive power of t. Our experience is that setting k= 1 results in non-linear relationships. Choosing a value of k= 0.5 gave a better approximation but there was still some evidence of non-linearity (Fig. 2). Unfortunately the estimate of the chronic LC50 is sensitive to the choice of k, so some other approach is required. The result of the extrapolation from acute to chronic toxicity is sensitive to the choice of k. The best approximation to linearity was found with very low values of k, suggesting the appropriate choice would be k= 0, which would be interpreted as the equivalent of taking the log of t (Fig. 3). When k was chosen as 0.5 (as shown in Table 5), the results were more similar to those obtained from the log–log model based on a 21-day exposure (Table 6) but there was still evidence of non-linearity. Because of the non-linearity in all the relationships based on 1/tk, models of this form cannot be used for extrapolation. The definition of chronic toxicity therefore has to be accompanied by a choice of exposure time. The log–log model (Eq. (8) of Lee et al., 1995) was found to give a near linear relationship. Use of this relationship, in common with the approach of Sun et al. (1995), requires nomination of a time for the assessment of chronic toxicity. The choice of exposure time in the estimation of chronic toxicity can be guided by the lifecycle length of the test organism. For P. australiensis a choice of 21 days would be appropriate. For other species such as fish, a 21-day time-frame has been also used for fish chronic tests; this is a compromise between the lifecycle length and feasibility. The use of extrapolations based on LC50 rather than LC10 or LC05 has been recommended by Slaughter et al. (2007) because they found extrapolations based on LC50 values are more stable. Of all measures like LC50 and LC10, LC50 has the smallest standard error. Basing the extrapolations on LC50 is therefore good practice. In the present study, LC10 values for the six pesticides tested were closer to the predicted chronic LC50 values. Both the log–log extrapolation and the LC10 approaches are recommended for estimating chronic toxicity. Other types of extrapolation such as those suggested by Slaughter et al. (2007) could potentially be combined with these methods to give a better coverage of chronic toxicity estimates. Their suggestion of using the same safety factor for all times and all toxicants requires further consideration.
6. Conclusions In the most agricultural areas, aquatic organisms inhabiting water bodies have potential to be exposed to a very low concentration of pesticides due to spray drift and run-off. To assess the significant biological outcome of such exposures, it is necessary to study the potential impairments of essential functions of an organism at sub-lethal concentrations. Due to high costs and long duration of chronic tests, there is a need to make long-term predictions about the likelihood of adverse effects caused by low-moderate contaminant levels. Uses of time series acute toxicity data proved to be highly beneficial in making such predictions under this study and were found to be credible when compared with the available observed value for the two pesticides.
Acknowldgments This work was funded by the CSIRO and ACIAR. All bioassays were conducted with shrimp and fish larvae in accordance with national and institutional guidelines for the protection of animal welfare.
References Abbasi, S.A., Soni, R., 1991. Studies on the environmental impact of three common pesticides with respect to toxicity towards a larvivore (Channelfish N. denricus). J. Inst. Public Health Eng. 2, 8–12. Abdullah, A., Kumar, A., Chapman, J., 1994. Inhibition of acetylcholinesterase in the Australian freshwater shrimp (Paratya australiensis) by profenofos. Environ. Toxicol. Chem. 13, 1861–1866. Abdullah, A.R., Lim, R.P., Chapman, J.C., 1993. Inhibition and recovery of acetylcholinesterase in Paratya australiensis exposed to the organophosphate insecticide chlorpyrifos. Fresenius. Environ. Bull. 2, 752–757. Anderson, R.L., 1989. Toxicity of synthetic pyrethroids to freshwater invertebrates. Environ. Toxicol. Chem. 8, 403–410. ANZECC and ARMCANZ (Australian and New Zealand Environment and Conservation Council, and Agriculture and Resource Management Council of Australia and New Zealand), 2000. Australian and New Zealand Guidelines for Fresh and Marine Water Quality. ANZECC and ARMCANZ, Canberra, Australia. APVMA, 2006. The reconsideration of registrations of products containing carbaryl and their associated labels Part 2. Uses of Carbaryl in Agricultural Situations. Vol. 1, Review Summary, July 2006. Australian Pesticides & Veterinary Medicines Authority, Canberra, Australia. ASTM, 1998. Standard guide for conducting acute toxicity tests on aqueous effluents with fishes, macroinvertebrates, and amphibians. Designation E1192, American Society for Testing and Materials. Barata, C., Baird, D.J, Markich, S.J., 1999. Comparing metal toxicity among Daphnia magna clones: an approach using concentration–time-response surfaces. Arch. Environ. Toxicol. 37, 326–331. Boey, A., Cooper, B., 1996. In: Atrazine and its Ecological Significance. Central and North West Regions Water Quality program, TS96.084. Department of Land and Water Conservation, Paramatta, NSW, Australia. Calleja, M.C., Persoone, G., Geladi, P., 1994. Comparative acute toxicity of the first 50 multicentre evaluation of in vitro cytotoxicity chemicals to aquatic nonvertebrates. Arch. Environ. Contam. Toxicol. 26, 69–78. Calow, P., Forbes, V.E., 2003. Does ecotoxicolgy inform ecological risk assessment?. Environ. Sci. Technol. 37 (7), 146A–151A. Cox, D.R., Oakes, D., 1984. In: Analysis of Survival Data. Chapman & Hall, New York 201pp. Crane, M., Newman, M.C., 2000. What level of effect is a no observed effect?. Environ. Toxicol. Chem. 19, 516–519. Friberg-Jensen, U., Wendt-Rasch, L., Woin, P., Christoffersen, K., 2003. Effects of the pyrethroid insecticide, cypermethrin, on a freshwater community studied under field conditions. I. Direct and indirect effects on abundance measures of organisms at different trophic levels. Aquat. Toxicol. 63, 357–371. Goss, L.B., Sabourin, T.D., 1985. Utilization of alternative species for toxicity testing: an overview. J. Appl. Toxicol. 5, 193–219. ¨ Hamilton, M.A., Russo, R.C., Thurston, R.V., 1977. Trimmed Spearman–Karber method for estimating median lethal concentrations in toxicity bioassays. Environ. Sci. Technol. 11, 714–719. Hamilton, M.A, Russo, R.C., Thurston, R.V., 1978. Correction to trimmed Spearman– ¨ Karber method for estimating median lethal concentrations in toxicity bioassays. Environ. Sci. Technol. 12, 417. Hemming, T.A., Arvind, S., Yogesh, K., 1989. Time–toxicity relationships in fish exposed to the organochlorine pesticide thoxychlor. Environ. Toxicol. Chem. 8, 923–932. Kenaga, E.E., 1982. Predictability of chronic toxicity from acute toxicity of chemicals in fish and aquatic invertebrates. Environ. Toxicol. Chem. 1, 347–358. Lange, R., Hutchinson, T.H., Scholz, N., Solbe, J., 1998. Analysis of ecetox aquatic toxicity (EAT) database II—Comparison of acute to chronic ratios for various aquatic organisms and chemical substances. Chemosphere 36, 115–127. Lee, G., Ellersieck, M.R., Mayer, F.L., Krause, G.F., 1995. Predicting chronic lethality of chemicals to fishes from acute toxicity test data: multi factor probit analysis. Environ. Toxicol. Chem. 14, 345–349. Long, E.R., Hong, C.B., Severn, C.G., 2001. Relationships between acute sediment toxicity in laboratory tests and abundance and diversity of benthic infauna in marine sediments: a review. Environ. Toxicol. Chem. 20, 46–60. Mayer, F.L., Krause, G.F., Buckler, D.R., Ellersiek, M.R., Lee, G., 1994. Predicting chronic lethality of chemicals to fishes from acute toxicity data: concepts and linear regression. Environ. Toxicol. Chem. 13, 671–678. Mian, L.S., Mulla, M.S., 1992. Effects of pyrethroid insecticides on nontarget invertebrates in aquatic ecosystems. J. Agric. Entomol. 9, 73–98. Mount, D.I., Stephan, C.E., 1967. A method for establishing acceptable limits for fish-malathion and butoxyethanol ester of 2,4-D. Trans. Am. Fish. Soc. 96, 185–193. Mu, X.Y., LeBlanc, G.A., 2002. Environmental antiecdysteroids alter embryo development in crustacean Daphnia magna. J. Exp. Zool. 292, 287–292. MU, X.Y., LeBlanc, G.A., 2004. Synergistic interaction of endocrine-disrupting chemicals: model development using an ecdysome receptor antagonist and a hormone synthesis inhibitor. Environ. Toxicol. Chem. 23, 1085–1091. NRA, 1997. In: Guidelines for Registering Agricultural Chemicals – Agricultural Requirements. National Registration Authority for Agricultural and Veterinary Chemicals, Canberra, Australia. Nebeker, A.V., Schuytema, G.S., 1998. Chronic effects of the herbicide diuron on freshwater cladocerans, amphipods, midges, minnows, worms, and snails. Arch. Environ. Contam. Toxicol. 35, 441–446.
ARTICLE IN PRESS KumarA et al. / Ecotoxicology and Environmental Safety 73 (2010) 360–369
Newman, M.C., Aplin, M.S., 1992. Enhancing toxicity data interpretation and prediction of ecological risk with survivial time modelling: an illustration using sodium chloride toxicity to mosquitofish. Aquat. Toxicol. 23, 85–96. Olima, C., Pablo, F., Lim, R.P., 1997. Comparative tolerance of three populations of the freshwater shrimp (Paratya australiensis) to the organophosphate pesticide, chlorpyrifos. Bull. Environ. Contam. Toxicol. 59, 321–328. Phyu, Y.L., Warne, M, St, J., Lim, R.P., 2004. Toxicity of atrazine and molinate to the cladoceran (Daphnia carinata) and the effect of river water and bottom sediment on their bioavailability. Arch. Environ. Toxicol. Chem. 46, 308–315. Phyu, Y.L., Warne, M, St, J., Lim, R.P., 2005. Toxicity and bioavailability of atrazine and molinate to the freshwater shrimp (Paratya australiensis) under laboratory and simulated field conditions. Ecotoxicol. Environ. Saf. 60, 113–122. Rand, G.M., 1995. In: Fundamentals of Aquatic Toxicology. Hemisphere, Washington, DC, USA. Raimondo, S., Montague, B.J., Barron, M.G., 2007. Determinants of variability in acute to chronic toxicity ratios in aquatic invertebrates and fish. Environ. Toxicol. Chem. 26, 2019–2023. Roast, S.D., Thompson, R.S., Donkin, P., Widdows, J., Jones, M.B., 1999. Toxicity of the organophosphate pesticides chlorpyrifos and dimethoate to Neomysis integer (Crustacea: Mysidacea). Water Res. 33, 319–326.
369
´ S., Barahona, M.V., 2005. Comparative study on the environmental Sa nchez-Fortun, risk induced by several pyrethroids in estuarine and freshwater invertebrate organisms. Chemosphere 59, 553–559. Slaughter, A.R., Palmer, C.G., Muller, W.J., 2007. An assessment of two-step linear regression and multifactor probit analysis as alternatives to acute to chronic ratios in the estimation of chronic responses from acute toxicity data to derive water quality guidelines. Integr. Environ. Assess. Manage. 3, 193–202. Smith, T.M., Stratton, G.W., 1986. Effects of synthetic pyrethroid insecticides on nontarget organisms. Resid. Rev. 97, 93–120. Sun, K., Krause, G.F., Mayer Jr., F.L., Ellersieck, M.R., Basu, A.P., 1995. Predicting chronic lethal toxicit of chemicals to fishes from acute toxicity test data: theory of accelerated life testing. Environ. Toxicol. Chem. 14, 1745–1752. Tomlin, C.D.S., 1994. In: The Pesticide Manual: A World Compendium 10th ed. British Crop Protection Council, UK. Vittozi, L., De Angelis, G., 1991. A critical review of comparative toxicity data on freshwater fish. Aquat. Toxicol. 19, 167–204. Warne, M.J., 1998. Critical review of methods to derive water quality guidelines for toxicants and a proposal for a new framework. Supervising Scientist Report 135, Supervising Scientist, Canberra.