Introduction

Aquatic invertebrate communities are generally exposed to multiple stressors that potentially include complex mixtures of contaminants. The impacts of these are of ecological concern, particularly during an organism’s sensitive, developmental life stages (Oros and Werner 2005; Geist 2011; Brooks et al. 2012; Connon et al. 2012). Concentrations of pesticides found in waters that receive agricultural and urban runoff often do not occur at levels that result in direct mortality (Scholz et al. 2012). Although these chemicals may be present in aquatic environments at relatively high concentrations during the peak application periods of spring and early summer, water flow and adsorption of pesticides to sediments and surfaces dictate that organisms are rarely exposed to elevated concentrations that cause mortality for continuous periods of time (Phillips et al. 2012; Beketov et al. 2013; Jeon et al. 2013). However, because pesticides and their breakdown products are retained in sediments and may gradually re-dissolve or otherwise remain available at more consistent low levels (e.g., through diet or contact), research on chronic and environmentally typical low-level exposures is needed. Exposure to low-level concentrations for extended periods of time, or to moderate concentrations for multiple brief periods, can potentially result in physiological impairments (Nyman et al. 2013; Scherer et al. 2013). Examples of reported effects in invertebrate organisms include reduction in emergence (Du et al. 2013), case-abandonment (Johnson et al. 2008) and reduced growth (Rakotondravelo et al. 2006). Many insecticides may have effects that can lead to long-term severe health impacts or reproductive impairment, which are often not detectable using traditional toxicity testing methods (Christensen et al. 2005; Connon et al. 2012). Effect-based endpoints, designed to assess sublethal impairments are often more sensitive and better predictors of deleterious effects associated with contaminated water and sediments (Maul et al. 2008; Connon et al. 2012; Deanovic et al. 2013; Rasmussen et al. 2013).

Even though aquatic organisms are generally exposed to contaminant mixtures, data used in ecotoxicological risk assessment are predominantly based on single substance evaluation (Junghans et al. 2006; Backhaus et al. 2013; Gregorio et al. 2013). This regulatory approach, though informative, may underestimate ecological relevance because the effects of contaminant mixtures is known to differ from that predicted based on the sum of individual contaminant effects (Backhaus et al. 2000; Altenburger et al. 2004; Nøargaard and Cedergreen 2010). Two fundamental concepts exist, which are devised to evaluate the general relationships between the effects of single substances and their corresponding mixtures: concentration addition (CA), in which the effect of each contaminant can be expressed as if it were a dilution of the other and is based on the assumption of a similar action, and independent action (IA), which is based on the assumption of probabilistic independence of the effects of dissimilarly acting agents (Faust et al. 2000; Backhaus et al. 2004; Belden et al. 2007; Cedergreen 2014). In brief, if chemical effects do not interact in a mixture, the influences of each compound are effectively additive, and the actual effect of the mixture is adequately described by either CA or IA as a reference model. However, interactions may occur and chemicals in a mixture may produce effects that are synergistic (more severe) or antagonistic (less severe), than predicted by either reference model (Lydy et al. 2004; de Zwart and Posthuma 2005; Jonker et al. 2005; Belden et al. 2007). Additive, synergistic, and antagonistic responses have been documented for mixtures of various classes of pesticides, and the majority of studies that have been conducted, have focused on binary mixtures. Our study expands this approach to include tertiary mixtures of contaminants that are often detected in surface waters worldwide, and makes use of sublethal endpoints that are likely to have substantial ecological relevance.

We investigated the combined effects of three commonly used insecticides: two pyrethroids; permethrin (type I pyrethroid) and lambda-cyhalothrin (type II pyrethroid), and one organophosphate; chlorpyrifos, on survival and the sublethal endpoints of immobility and growth of Chironomus dilutus following a 10-days exposure. C. dilutus is used as a standard invertebrate species used in toxicity testing, and is among the numerous non-target species that are potentially affected by pesticide runoff. It has been shown to be highly sensitive to pyrethroids and organophosphates in field and laboratory studies (Weston et al. 2004; Anderson et al. 2006; Du et al. 2013; Li et al. 2013). The life cycle of C. dilutus is comprised of three aquatic stages (egg, four larval instars and pupae) and a terrestrial adult stage. The larval stage is representative of organisms living in the benthic zone. During this stage organisms burrow in the upper sediments, and utilize organic matter and sediment particles to build their protective cases. Like many benthic organisms, they feed on detrital particles, making them ideal organisms for testing the sediment–water interphase.

We selected the three insecticides because they are among the most frequently detected insecticides in aquatic habitats worldwide (Amweg et al. 2006; Sprague and Nowell 2008; Hintzen et al. 2009; Trimble et al. 2009; Bereswill et al. 2013; Li et al. 2013), and are known to be highly toxic to aquatic invertebrates and fish species (Werner et al. 2010; Phillips et al. 2012). All three pesticides are used for similar pest treatments such as the cultivation of vegetables, fruits, grains, and for landscape maintenance, and were repeatedly detected in the same water or sediment samples in recent studies (Weston et al. 2008; Budd et al. 2009; Weston et al. 2013). The selected pesticides are all neurotoxins with different neurological target sites and/or modes of action. Although modulations of the voltage-gated sodium channels are similar between the two types of pyrethroids, the degree of modification of sodium currents is different. Single sodium channel currents are prolonged to a greater extent with type II than type I pyrethroids (Wouters and van den Bercken 1978; Clark and Matsumura 1982; Nasuti et al. 2003). Organophosphates (chlorpyrifos) on the other hand, inhibit acetylcholine esterase activity (Malison et al. 2010; Hua et al. 2013) directly impacting the synaptic signal. Varying modes of action could drive potential interactive exposure effect differences between single chemicals and complex mixtures.

Given that pesticides predominantly occur at concentrations below those that cause mortality, sublethal endpoints are more applicable at evaluating environmental relevance. By assessing the effects of the three pesticides in 10-days toxicity tests, we evaluate the use and effectiveness of sublethal endpoints in mixture toxicity testing for regulatory applications and monitoring studies.

Materials and methods

Test organisms and acclimation

Chironomus dilutus were obtained from Aquatic Biosystems, Fort Collins, CO, USA. Under typical laboratory conditions, C. dilutus begin to pupate, and emerge as adults 21 days after hatching, thus 2nd instar larvae (8–10 days old) were used as to avoid emergence during the 10-days exposures. Upon arrival, dissolved oxygen (DO >2.5 mg/L), and temperature (23 ± 2 °C) of transport water sub-samples were measured, and were within acceptable ranges stipulated by U.S. EPA standard test protocols (U.S. EPA 2000). Healthy animals were moved to aerated 7-L aquaria, fed, and acclimated to laboratory test conditions for 48 h. During the acclimation period, approximately 50 % of the transport water was changed twice daily and refilled with test control water, i.e., deionized water, modified to attain U.S. Environmental Protection Agency (U.S. EPA) moderately hard specifications (hardness 90–100 mg/L CaCO3, alkalinity 50–70 mg/L as CaCO3, SC 330–360 µS/cm and pH 7.8–8.2) (Eide and Johansson 1994; U.S. EPA 2000, 2002). C. dilutus were fed 10 ml of 4 g/L Tetramin® slurry daily.

Chemicals and chemical analysis

Chlorpyrifos (purity >98 %, CAS number 2921-88-2), lambda-cyhalothrin (purity >98 %, CAS number 91465-08-6), and permethrin (purity >98 %, CAS number 52645-53-1) were purchased from Chem Service (West Chester, PA, USA). Pesticide stock solutions were prepared in methanol and spiked into laboratory control water to achieve exposure concentrations illustrated in Tables 1 and 2, with a final methanol concentration of 0.05 %. Before adding the pesticide solutions into the test beakers, three 1-L water samples for each single-chemical exposure and the tertiary mixture exposure were collected, and stored at 4 °C for subsequent chemical analyses. Within 24 h, the samples were extracted by solid phase extraction (Supelclean™ ENVI™-18 SPE Tubes, 500 mg, Sigma-Aldrich, St. Louis, MO, USA), and analyzed using an HP-6890 gas chromatograph (Agilent Technologies, Palo Alto, CA, USA) coupled to an HP-5973 N quadrupole mass spectrometer detector, operated in electron capture negative ionization mode (GC-ECNI-MS) with methane as the reagent gas, equipped with a split-splitless injector (280 °C, splitless, 1.5-min purge time) and a Supelco DB-5MS column (30 m × 0.25 mm with a 0.25 µm film thickness). Instrumental calibration was performed using nine sets of calibration standard solutions containing all three pesticides (each purchased as 100 µg/mL solution in acetonitrile, Chem Service, West Chester, PA), the surrogate trans-permethrin D6 (EQ Laboratories, Atlanta, GA), and an internal standard; dibromooctafluorobiphenyl (Chem Service, West Chester, PA) in hexane. Quantity was calculated based on peak area and comparing them to the standard curves. Quality-assurance/quality-control was conducted by analyzing a method blank of deionized water (Milli-Q) to ensure that no contamination occurred during sampling extraction and analysis. The surrogate trans-permethrin D6 was added to each sample, including the blank, to monitor matrix effects and overall method performance. Surrogate recoveries were on average 103 % with a range between 79 and 112 % indicating high extraction efficiency. Reported values were not corrected for surrogate recovery. Dibromooctafluorobiphenyl was added to sample extracts before analysis in order to correct quantitative differences in extract volume as well as to monitor instrument conditions. No pesticides were detected in the control or the method blank. Measured concentrations by GC-ECNI-MS were lower than nominal concentrations; average recoveries for lambda-cyhalothrin were 65 % (range 62–86 %), for permethrin 46 % (range 34–58 %), and for chlorpyrifos 78 % (range 71–86 %). The average proportion of the measured concentrations recovered for each pesticide was used as a factor to estimate the realized concentrations for each exposure (Tables 1, 2), and are presented as measured concentrations hereon.

Table 1 Measured and nominal concentrations of lambda-cyhalothrin, permethrin, and chlorpyrifos at test initiation for single-chemical exposures of C. dilutus over 10 days
Full size table
Table 2 Measured and nominal concentrations of lambda-cyhalothrin, permethrin, and chlorpyrifos in the mixture exposure at test initiation for mixture exposures of C. dilutus over 10 days
Full size table

Exposure tests

Mixture toxicology employs a dimensionless ratio, the toxic unit (TU), to generate a normalized scale (McCarty et al. 1992). Each toxicant concentration is considered as a fraction of its individual toxicity, which most commonly is expressed in terms of its LC50. The total TU of a mixture is the sum of the individual fractions. The TU approach assumes dose additivity (Altenburger et al. 2004). Thus, the sum of TU describes the joint chemical concentration of a mixture, given a known common effect concentration, which can be used to assess the toxicity of a tertiary mixture as follows:

$$ TU_{summation} = \frac{{Cw_{1} }}{{LC50_{1} }} + \frac{{Cw_{2} }}{{LC50_{2} }} + \frac{{Cw_{3} }}{{LC50_{3} }} $$
(1)

where Cwi is the concentration of chemical 1 in a mixture and LC50i is the LC5096h for chemical 1 (McCarty et al. 1992). For example, the sum of 1/3 of the LC5096h of each pesticide equals a TU value of 1. Pesticides were combined in an attempt to produce equitoxic mixtures and to evaluate their interactive effects (Belden and Lydy 2006; Symington et al. 2011). The treatments used in the toxicity tests were based on expected LC5096h values determined by other research groups using similar test methods. Targeted nominal LC5096h values for C. dilutus were 189.00 ng/L permethrin (Harwood et al. 2009), 37.90 ng/L lambda-cyhalothrin, and 470.00 ng/L chlorpyrifos (Ankley and Collyard 1995). Based on each reported LC5096h value, we exposed animals to single-pesticide TU values of 0.167, 0.25, 0.33, 0.5, 1, 1.5, and 3 (Table 1), and the following mixture TU values: 0.125, 0.25, 0.5, 0.75, 1, 1.5, and 3 (Table 2). These concentrations were specifically chosen to include environmentally typical concentrations as reported in previous studies, as well as high values that may occur transiently (Anderson et al. 2006; Budd et al. 2009; Werner et al. 2010; Weston and Lydy 2010). Single-pesticide responses were then used to predict the combined mixture toxicity, and compared to the observed mixture response (see Statistical analysis).

Mixture and single-chemical exposures were conducted at the same time over a 10-day period using the same batch of animals (2nd instar larvae). Tests were conducted at 23 ± 2 °C with a 16-h light: 8-h dark photoperiod, and consisted of four replicate 1 L glass beakers, each containing a layer of 10 g clean and autoclaved silica sand as a substrate, 750 mL of treatment water, and 10 organisms. The sand allowed chironomids to build their cases. Once test solutions were added to the test vessels at test initiation, organisms were randomly placed into each beaker.

Mortality was recorded daily, at which time 80 % of each test solution was renewed, and any dead organisms and debris were removed from the test vessels. Water quality parameters (pH, specific conductance, DO, and temperature) of renewal water and wastewater were measured. At test initiation and after each water renewal, organisms were fed 1.5 ml of 4 g/L Tetramin® slurry. Test vessels were randomly redistributed within the exposure chamber at each renewal day.

Mobility of each organism was determined at test termination (day 10) using video analysis. Chironomids are generally sedentary if food and oxygen are sufficient in the immediate area, but are inclined to be mobile when they are not provided substrate. Therefore, surviving organisms were carefully teased from their cases and transferred individually into corresponding filming chambers; a 5.1 cm diameter well in a five-welled white PVC plate containing corresponding treatment water without substrate. To improve lighting quality and contrast of the videos, the white PVC plate was then placed on a light board. The positioning of the plate and video settings were standardized for each recording. Videos were recorded in MPEG-2 format, using a Panasonic® black and white CCTV Camera (12 V DC) filming all five filming chambers from the top. The camera was connected to a portable laptop-computer via a USB Frame grabber (Model WinTV-HVR 950, Hauppauge Computer Works, Hauppauge, NY). Thirty frames per second were collected for each organism, over an 80 s period. Recorded videos were then analyzed using Ethovision XT 6.1 Software (Noldus Information Technology Inc., Leesburg, VA) to determine percentage immobility.

Weight was determined using the same organisms used for mobility assessments. Following video recording, the organisms were transferred from the filming chambers onto pre-weighed tin dishes (pooled per treatment replicate), desiccated at 60 °C following methods described by Nahon et al. (2010), and weighed using a Mettler® AE 100 balance (0.1 mg accuracy). To examine 10-day growth, the weights of a separate set of four replicates of ten organisms were measured at test initiation and compared to the weights of surviving individuals at test termination. Weight is presented as mg/surviving individual. Data for both sublethal endpoints is presented up to 1 TU as mortality in higher treatments was greater than 50 %.

Statistical analysis

Median lethal and effective concentrations (LC50 and EC50) and no observed effect concentrations (NOEC) were determined for mortality and the sublethal endpoints of immobility and growth, based on measured exposure concentrations (Tables 1 and 2). We tested for significant differences of the treatments compared to the controls using Analysis of Variance, or where parametric assumptions were not met, a Kruskal–Wallis test, including a Dunnett’s post hoc comparison, using Minitab 16 Statistical Software 2010 (Minitab, Inc., State College, PA, USA). The significance level or α used in all these tests was p ≤ 0.05. All differences discussed below are significant unless otherwise noted. Concentrations that caused a 50 % reduction in survival (LC50) and growth (EC50) were determined by fitting non-linear regression curves to the toxicity data used in the DRC package version 2.3-96 (Ritz and Streibig 2005) R (R Core Team 2014). EC50 concentrations for the effects on immobility and for chlorpyrifos on weight were not calculated because the levels of responses did not amount to 50 % relative to the control. For all data, log-normal, log-logistic and Weibull functions were fitted with the optimal model fit chosen for each dataset by the distribution that had the lowest Akaike’s Information Criterion value. The fit of the optimal model was confirmed by a goodness of fit test. Three-parameter regression models were fitted, assuming a lower limit of 0.

The dose–response data were described with a log-logistic dose–response model with an upper limit of

$$ 1:y = \frac{1}{{1 + ({\raise0.7ex\hbox{$x$} \!\mathord{\left/ {\vphantom {x e}}\right.\kern-0pt} \!\lower0.7ex\hbox{$e$}})^{b} }} $$
(2)

Where e is the effect concentration (LC50), and the parameter b denotes the relative slope around e. Once the dose–response curves were fitted, joint effect predictions in relation to concentration addition (CA) and independent action (IA) were made. For the prediction of mixture toxicities the complete concentration–response range between 1 and 99 % effect was predicted according to both the CA as well as the IA concept. CA assumes that the chemicals act as dilutions of each other (Bliss 1939; Hewlett and Plackett 1952). For the concept of CA the calculation was conducted as outlined in Backhaus et al. (2000). Briefly, the total concentrations of each mixture were calculated in steps of 1 % as:

$$ LCx_{mix} = \left( {\mathop \sum \limits_{i = 1}^{n} \frac{{p_{i} }}{{ECx_{i} }}} \right)^{ - 1} $$
(3)

Where p i is the fraction of compound i present in the mixture. The resulting 99 pairs were connected with straight lines, visualizing the predicted concentration response-curve.

To calculate the mixture effects according to IA, the individual compounds were expressed as fractions pi of the total concentration c mix. The overall effect of any given total mixture concentration can be calculated as

$$ L\left( {c_{mix} } \right) = 1 - \mathop \prod \limits_{i = 1}^{n} \left[ {1 - F_{i} (c_{i} )} \right] $$
(4)

where L(c mix) denotes the predicted effect (scaled from 0 to 1) of an n-compound mixture, c i is the concentration of the ith compound, and F i is the effect of that concentration if the compound is applied singly. To calculate the mixture effects predicted according to IA, the 99 predicted mixture concentrations obtained from the concentration–response range for the CA model described in Eq. (3) were used.

The observed data were considered to be significantly different from the predicted model if the 95 % confidence intervals of the observed toxicity values did not overlap the value predicted by the model.

The toxicity of each of the chemicals in a mixture can differ substantially, therefore each chemical’s relative contribution to toxicity was expressed by using toxicity units rather than their actual concentrations (LC50) to predict the joint toxicity (Lydy and Austin 2004; Jonker et al. 2005). An LC50 resulting in a TU >1 represents a synergistic type of joint action, whereas a TU <1 indicates an antagonistic type of action (Pape-Lindstrom and Lydy 1997; Sørensen et al. 2007).

Results

Water quality parameters remained stable throughout all exposures and mean control survival of C. dilutus was 75 % (SE = ± 0.13) meeting the U.S. Environmental Protection Agency minimum acceptance criteria for this species (U.S. EPA 2000). Control mortality was only observed within 24 h after test initiation, and was likely caused by random handling stress when transferring the chironomids into treatment beakers.

The tertiary mixture exposure resulted in a less severe response than the pesticides applied singly. In single-chemical tests, lambda-cyhalothrin was the most toxic pesticide to C. dilutus resulting in an LC50 value of 32.99 ng/L (±2.56 SE), followed by permethrin (159.41 ng/L, ±16.36 SE), and chlorpyrifos (571.49 ng/L, ±88.68 SE) (Table 3). These values were 1.1–1.5 times higher than the corresponding single-pesticide NOEC values for lambda-cyhalothrin, permethrin, and chlorpyrifos (21.98, 144.59, and 361.90 ng/L, respectively). The LC50 value for the observed mixture exposure was 1.90 TU (±0.28 SE), indicating an antagonistic response (>1 TU). Survival of C. dilutus was reduced in concentrations ≥1.5 TU of lambda-cyhalothrin (32.97 ng/L) and chlorpyrifos (542.85 ng/L), whereas the mixture exposure caused no significant reduction in survival ≤1.5 TU (10.25 ng/L lambda-cyhalothrin, 31.94 ng/L permethrin, 193.40 ng/L chlorpyrifos) indicating an IA response (Fig. 1). The CA and IA concept were applied to predict mixture toxicity based on the single-pesticide exposures, and compared to the observed mixture response (Fig. 2). The observed mixture response was most consistent with the IA concept, whereas the CA concept overestimated the combined effect.

Table 3 Effect concentrations calculated for 10-days exposures of C. dilutus to lambda-cyhalothrin (LC), permethrin (Perm), and chlorpyrifos (CLF) applied singly and in mixture (Mixture) of C. dilutus. NOEC no observed effective concentration, LC50 lethal concentration resulting in 50 % mortality of the population, EC50 effect concentration resulting in 50 % reduction in growth, SE standard error. ‘—‘ indicates value was not calculable
Full size table
Fig. 1
figure 1

Percent survival of C. dilutus following 10-days exposures to lambda-cyhalothrin, permethrin and chlorpyrifos, and corresponding mixtures up to 3 TU. Non-linear regression curves were fitted to the toxicity data using the DRC package version 2.3-96 (Ritz and Streibig 2005) in R (R Core Team 2014). Average control survival = 75 %

Full size image
Fig. 2
figure 2

Observed and predicted toxicities of tertiary mixtures of lambda-cyhalothrin, permethrin and chlorpyrifos on C. dilutus. Mixture concentrations derived from LC50 values of the individual components. Mixture = observed toxicity of the tertiary mixture; CA prediction = Prediction according to concentration addition model; IA prediction = Prediction according to independent action model

Full size image

Average control immobility was 77.89 % (±6.69 SE) over the recording time of 80 s. Single-pesticide exposure caused a decrease in mobility across all concentrations tested with an average immobility of 93.68 % (±0.44 SE) at 0.25 TU for all three pesticides (Fig. 3). The exposure to the tertiary mixtures decreased mobility at 1 TU with an average immobility of 94.23 %. Exposure to chlorpyrifos individually caused the greatest inhibition of mobility at 1 TU (361.90 ng/L), with an average immobility of 93.15 % (±0.71 SE), followed by permethrin (90.42 %, ±0.59 SE), mixture (88.58 %, ±0.99 SE), and lambda-cyhalothrin (87.48 %, ±1.79 SE). Comparing NOEC values for immobility and survival, permethrin exposures caused the greatest difference (NOEC for immobility 8 times smaller than for survival), followed by chlorpyrifos (6.5 times), lambda-cyhalothrin (5 times), and the mixtures (3 times).

Fig. 3
figure 3

Percentage immobility over 80 s of recording time of C. dilutus following 10-days exposures to lambda-cyhalothrin, permethrin, chlorpyrifos, and corresponding mixtures up to 1 TU (survival ≥50 %). NS non-significant to controls, all others p < 0.05. Data points are connected to aid visualization. Average control Immobility = 77.88 %

Full size image

Initial weight (T0) of four subsamples of C. dilutus was 0.345 mg/individual (±0.04 SE; Fig. 4). Final weight of control organisms was 1.932 mg/individual (±0.07 SE) equaling a 5.6-fold growth over the 10-days test period. Both pyrethroids and the mixture led to an inhibition in C. dilutus growth, while chlorpyrifos had no detectable effect on this endpoint. Compared to the final control weight, reductions were recorded at concentrations ≥0.25 TU for lambda-cyhalothrin (from 32.6 % up to 65.8 % in the highest concentration) and the mixture (25.2–63.1 %), and ≥0.167 TU for permethrin (33.0–77.8 %). In detail, exposure to lambda-cyhalothrin resulted in a growth inhibition up to a final weight of 0.661 mg/surviving individual (±0.08 SE) at 1 TU (21.98 ng/L lambda-cyhalothrin). Mixture exposures resulted in a final weight of 0.712 mg/surviving individual (±0.03 SE) at 1 TU (7.45 ng/L lambda-cyhalothrin, 24.90 ng/L permethrin, 129.70 ng/L chlorpyrifos). Exposure to permethrin resulted in a final weight of 0.429 mg/surviving individual (±0.03 SE) at 1 TU (96.39 ng/L permethrin). NOEC values for the growth endpoint, for both mixture and single-pyrethroid concentrations were 5.0–12.0 times lower than the respective NOEC for survival (Table 3), with the mixture exposure (NOEC = 0.125 TU) causing the greatest difference (12 times lower than for survival), and permethrin (NOEC < 18.21 ng/L, 8 times lower) and lambda-cyhalothrin (NOEC = 4.39 ng/L, 5 times lower) the smallest differences. EC50 values for growth were on average 1.8–7.0 times lower than the respective LC50 for survival, with permethrin (EC50 = 22.51 ng/L, ±7.75 SE) representing the largest difference and lambda-cyhalothrin (EC50 = 18.13 ng/L, ±5.20 SE) the smallest. The EC50 value for the mixture (0.49 TU) was 3.9 times lower than the respective LC50 (1.90 TU).

Fig. 4
figure 4

Final weight in mg/surviving individual of C. dilutus following 10-days exposures to lambda-cyhalothrin, permethrin, chlorpyrifos, and corresponding mixtures up to 1 TU (survival ≥50 %). NS non-significant to controls, all others p < 0.05. Non-linear regression curves were fitted to the toxicity data using the DRC package version 2.3-96 (Ritz and Streibig 2005) R (R Core Team 2014). Horizontal bars represent average initial weight at test initiation (T0) (0.345 mg/individual) and average weight of the control at test termination (T10) (1.932 mg/surviving individual), and are presented to facilitate the comparison with initial and final weights along increasing toxic units

Full size image

Discussion

The assessment of three insecticides, lambda-cyhalothrin, permethrin, and chlorpyrifos, demonstrates that mixtures can affect survival in an antagonistic manner (1.90 TU) and result in effects on sublethal endpoints that are up to 12.0 times lower than the corresponding NOEC for survival. The sublethal effect concentrations for growth inhibition caused by lambda-cyhalothrin (EC50 = 18.13 ng/L), permethrin (EC50 = 22.51 ng/L) and the tertiary mixture (0.25 TU = 2.09 ng/L lambda-cyhalothrin, 6.30 ng/L permethrin, and 30.94 ng/L chlorpyrifos) in this study are within the range of environmentally relevant concentrations reported in previous monitoring studies in different states of the USA (Anderson et al. 2006; Smith and Lizotte 2007; Werner et al. 2010). For example, studies in Californian streams by Budd et al.(2009) and Weston et al.(2014) detected lambda-cyhalothrin at concentrations of 1.4–27.0 ng/L, and permethrin between 4 and 470 ng/L. Chlorpyrifos was detected at concentrations between 1.2 and 226.0 ng/L (Weston and Lydy 2010). Even though concentrations for lambda-cyhalothrin in these monitoring studies were on average lower than for the other two pesticides, the lower effective concentrations of 18.13 ng/L (EC50 Growth) and <3.69 ng/L (NOEC Immobility) determined in this study suggest that lambda-cyhalothrin is the most toxic, and of the three, the pesticide of greatest concern in terms of potential ecotoxicological effects on invertebrate populations and aquatic communities.

The EC50 of the mixture of 0.49 TU indicates that 0.163 TU (LC8) of each chemical in a tertiary mixture results in a 50 % growth inhibition relative to controls. The levels of 0.167 TU as measured in our single-chemical exposures were 3.67 ng/L lambda-cyhalothrin, 16.07 ng/L permethrin, and 60.31 ng/L chlorpyrifos; these too are lower than the reported environmentally relevant concentrations. These results highlight the pressing need to adequately assess the sublethal effects of contaminant mixtures as they co-occur in the environment.

Treatment concentrations for each pesticide were initially chosen from literature values to achieve equitoxic concentrations in each mixture, the measured concentrations were, as anticipated, below nominal concentrations, and toxicity levels are known to vary among studies (Wheelock et al. 2005). Differences between nominal and measured concentrations are not unusual, especially when target concentrations are near the limit of detection (Farmer et al. 1995; Amweg et al. 2005). Wheelock et al. (2005) showed that up to 50 % of the pyrethroid can adsorb to the sampling container in 24 h which may be one explanation for the lower measured concentrations in our study.

The realized toxicity levels were therefore not strictly equitoxic. An equitoxic concentration approach was originally chosen to evaluate the joint action of the tertiary mixture by using CA and IA models, and to determine the degree of interaction, similarly to other studies (Denton et al. 2003; Belden and Lydy 2006; Symington et al. 2011; Larras et al. 2013; Norwood et al. 2013; Villa et al. 2014). The CA and IA concepts are usually evaluated using a “fixed ratio design”, where the constituents ratio is kept constant throughout the studies (Barata et al. 2006), however, strictly equitoxic ratios are not essential for mixture toxicity assessments, and non-equitoxic ratios are more reflective of pesticide levels in the environment (Altenburger et al. 2004; Brodeur et al. 2014).

Ecotoxicological effects are known to differ significantly depending on routes of uptake, e.g., aqueous versus dietary exposures, as demonstrated by Werner et al. (2002a). C. dilutus are detritivores, therefore dietary exposure to contaminants is likely (Laskowski 2002; Liu et al. 2004; Yang et al. 2006). Pyrethroids and organophosphates are known to adsorb to particulate matter. C. dilutus spends most of its larval stage in the substrate, therefore aqueous exposure may be reduced for this species, but cuticular exposure from the substrate may be greater. Pyrethroid exposure resulted in reduced growth of C. dilutus, which could have been caused by food avoidance due to insecticide-bound organic material. Alternatively, feeding rates may have been maintained, in which case reduced growth could be a direct effect of these insecticides; e.g., energetic reserves are allocated toward detoxification. Sublethal, behavioral effects such as case abandonment in the caddis fly Brachycentrus americanus, has been reported on exposures to esfenvalerate (type II pyrethroid), and associated with energetically costly activities such as case-rebuilding (Johnson et al. 2008). Induced case-rebuilding has further been associated with reduced growth in the caddis fly Odontocerumalbicorne (Stevens et al. 1999). Growth- and/or behaviorally-related effect concentrations will likely impact population dynamics.

This could occur through several mechanisms, e.g., chironomids are short-lived as adults, and since fecundity is largely determined by size at emergence, contaminant exposure may lead to reduced number of offspring in subsequent generations (Xue and Ali 1994). Smaller individuals may also be more susceptible to predators, may have reduced resistance to other environmental stressors as homeostatic energy demands are increased to contend with contaminant stress (Liber et al. 1996; Sibley et al. 1997; McKenney et al. 1998).

The observed mixture effects on survival (Fig. 2) suggest that the IA model is the most suitable concept to predict the combined toxicity effects of the three chemicals tested in this study, likely due to the differences in modes of action of the assessed pesticides. Other studies have also found that the IA concept provides a reasonable prediction of toxic effect for mixtures containing compounds with different modes of action (including pyrethroids and organophosphates, Backhaus et al. 2000; Faust et al. 2003; de Zwart and Posthuma 2005; Barata et al. 2012). However, the CA model has been highlighted by others (Könemann 1981; Hermens et al. 1984; Cedergreen 2014; Chen et al. 2014) to better predict mixture toxicity (e.g., chlorpyrifos and esfenvalerate; Belden and Lydy (2006)), indicating that the IA under-predicts toxicity. Variable results between studies, which evaluated different types of mixtures containing organophosphates and pyrethroids, may have been due to the toxicity of the most toxic constituent, rather than their modes of action. In this study, lambda-cyhalothrin was 5–12 times more potent than permethrin and chlorpyrifos, respectively, and thus is likely the driver of toxicity in the mixtures. This supports previous postulations that CA and IA models can be driven by one chemical within mixtures if its potency is substantially higher than the other members of the mixture (Heindel et al. 1995; Olmstead and LeBlanc 2005).

Effects on sublethal endpoints of C. dilutus were observed at environmentally relevant concentrations. This was also found in a study by Maul et al. (2008) where the individual exposure to lambda-cyhalothrin or permethrin significantly affected immobilization and growth rate of C. dilutus. Due to the complex life cycle of C. dilutus (involving pupation and emergence events), larval growth is frequently demonstrated to be a more sensitive endpoint than survival during the larval period, because growth may predict survival to adulthood and fecundity (Ankley et al. 1993, 1994; Burton et al. 1996; Maul et al. 2008). Growth inhibition could lead to failure to mature and reproduce. In previous studies, reduced larval growth in C. dilutus negatively affected pupation, emergence (86–100 % reduction), adult female size, number of eggs per female, and fecundity (Liber et al. 1996; Sibley et al. 1997; Ristola et al. 1999).

Agra and Soares (2009) also report that environmentally relevant concentrations of chlorpyrifos did not affect growth in C. dilutus supporting the lack of detectable effect on growth in this study. While daily food provision was equal throughout the tests, increased mortality resulting from exposure may have resulted in increased food availability per capita, thus concealing potential growth effects; as reported in other studies (Sibley et al. 1996; Martinez-Jeronimo et al. 2000; Hooper et al. 2003; Rakotondravelo et al. 2006). Furthermore, density can have negative effects on growth, development, and reproduction, as reported by Hooper et al. (2003). But regardless of food per capita and density, growth was also affected by the pyrethroid pesticides, thus the effects observed following pyrethroid exposure, but not following exposure to chlorpyrifos, are potentially due to differences in mode of action of the two insecticide groups. Pyrethroids work by preventing closure of the sodium channels in neuronal membranes affecting both the peripheral and central nervous system which may directly lead to decrease in feeding activity (Landrum et al. 2002), assimilation efficiency (Jager et al. 2006), protein synthesis, and rates of biotransformation and damage repair (Kooijman and Metz 1984). Chlorpyrifos can block the function of acetylcholinesterases (AChE), an important enzyme involved in neurotransmission. A number of studies found that despite a significant inhibition of AChE chironomid larvae are able to survive (Rakotondravelo et al. 2006; Rebechi et al. 2014), while behavioral effects such as limited mobility are present in organisms (Azevedo-Pereira et al. 2011) potentially having a lesser effect on feeding ability.

Mobility was the most sensitive endpoint in this study. A variety of neurotoxic contaminants in aquatic systems may affect mobility at sublethal exposure levels (Christensen et al. 2005; Werner and Moran 2008; Jin et al. 2009), thus suggesting to be a highly environmentally-relevant endpoint. It is especially useful for estimating effects on individual level in fish (Little and Finger 1990; Heath et al. 1993; Geist et al. 2007; Floyd et al. 2008; Beggel et al. 2010), and has also been applied in experiments involving C. dilutus (Hatch and Burton 1999). The assessment of swimming performance incorporates biochemical and physiological effects and directly evaluates the impacts of neurotoxic contaminants on nerve cell transmissions and resulting muscle activity (Heath et al. 1993; Jin et al. 2009). Inability to swim normally after an exposure to insecticides will therefore negatively affect individual fitness and survival, with potential consequences at the population level (Little and Finger 1990; Floyd et al. 2008). In the current study, significant effects on mobility were detected at exposures that are within the range of reported environmental concentrations (Anderson et al. 2006; Budd et al. 2009; Werner et al. 2010; Weston and Lydy 2010).

In ecological risk assessment, safety factors are applied to account for the uncertainty of extrapolating from laboratory toxicity tests to the real environment (EC 2003; U.S. EPA 2004; Jager et al. 2006; Hanson and Stark 2012). Despite these regulation efforts, pesticides exceed sublethal effective concentrations (NOEC and EC50) determined for growth and immobility in this study (Amweg et al. 2006; Weston and Lydy 2010; Bereswill et al. 2013; Li et al. 2013), and have been shown to impact aquatic organisms worldwide at concentrations that current legislation considers environmentally protective (Werner et al. 2002b; Schulz 2004; Weston and Lydy 2012; Beketov et al. 2013). Thus, it is evident that the toxicity of contaminants, and complex mixtures, may be significantly underestimated, by solely utilizing mortality as an endpoint for monitoring ambient water quality, as well as in ecological risk assessments, indicating the need for more sensitive endpoints to adequately assess ecological risk.

Conclusion

Determining the effect of water pollution remains a great challenge for environmental policy and management. This study reinforces the pressing need of integrating sublethal endpoints into regulatory toxicity assessments and monitoring studies to adequately assess the effects of contaminants. Mortality alone, as typically used for ecological assessment and management of industrial chemicals and pesticides, provides limited, to no information on organism fitness caused by pesticide exposure, especially since environmentally relevant concentrations do not generally occur at concentrations that result in direct mortality to aquatic communities. The use of sublethal endpoints, such as growth or mobility, in toxicity tests can indicate the presence of low levels of contaminants in water or sediment samples, sometimes at concentrations below the limit of detection of current-use analytical methods. Therefore it is essential to incorporate such tests into ambient water monitoring efforts and ecological risk assessment. In order to be able to monitor the effects of contaminant mixtures and to safeguard human health and the environment, a more holistic approach is required that includes assessing the combined effects of cumulative exposures to multiple stressors utilizing sublethal endpoints on toxicity exposures.