Different decreasing rates of chemical threshold concentrations can be explained by their toxicokinetic and toxicodynamic characteristics

Science of the Total Environment xxx (xxxx) xxx

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Different decreasing rates of chemical threshold concentrations can be explained by their toxicokinetic and toxicodynamic characteristics Fan Wu a, Yongfei Gao a, Zhiling Zuo a, Jianfeng Feng a,⇑, Zhenguang Yan b, Lin Zhu a a Key Laboratory of Pollution Process and Environmental Criteria of Ministry of Education and Tianjin Key Laboratory of Environmental Technology for Complex Trans-Media Pollution, College of Environmental Science and Engineering, Nankai University, Tianjin 300071, China b State Key Laboratory of Environmental Criteria and Risk Assessment, Chinese Research Academy of Environmental Sciences, Beijing 100012, China

h i g h l i g h t s

g r a p h i c a l a b s t r a c t


All 20 chemical’s HC5 in this study

decreased with exposure time.
Chemical’s decreasing rates of HC5 (k)

are different among chemicals with range of from 1.23E to 06 to 5.70E-02 1/h.
The k is significantly negatively related to the bioconcentration factor (BCF) and positively associated with the parameter the damage recovery rate (kR). Higher assessment factor (AF) should be adopted in deriving the acute water quality criteria (WQC) for some chemicals with high BCF and low kR.

a r t i c l e

i n f o

Article history: Received 8 August 2019 Received in revised form 17 October 2019 Accepted 25 October 2019 Available online xxxx Editor: Daniel Wunderlin Keywords: Chemicals Threshold concentrations HC5 Bioconcentration factors Damage recovery rates

a b s t r a c t To protect ecosystems, threshold concentrations (e.g., HC5) for chemicals are often derived using the toxicity data obtained at fixed times. Since the toxicity (e.g., LC50) usually decreases with exposure time, the threshold concentrations are expected to be time-dependent, giving rise to the uncertainty in the chemical environmental criteria. Here, using the LC50 data with at least 3 different exposure durations (24, 48 and 96 h) for compounds, we explored the time evolutions of HC5 across 20 chemicals. Results showed that all chemical’s HC5 decreased with time, but their decreasing rates of HC5 (k) are significantly different: for some chemicals the k are lower than 0.001 (e.g., methoxychlor and dieldrin), while for some chemicals the k are higher than 0.05 (e.g., PCP and aldicarb). Furthermore, we found that k is negatively related to the bioconcentration factors (BCF), and positively related to the damage recovery rates (kR). Our work demonstrated that time is an important source of the ecological threshold uncertainty, and this uncertainty is associated with chemical-specific toxicokinetic and toxicodynamic characteristics. We recommend that to effectively protect the ecological communities, higher assessment factor should be adopted in deriving the acute environmental criteria for these chemicals with high BCF and low kR, fluoranthene and diazinon. Ó 2019 Elsevier B.V. All rights reserved.

1. Introduction

⇑ Corresponding author. E-mail address: [email protected] (J. Feng).

Ecological risk assessment of chemicals plays important roles in ecotoxicology and aims to find the toxicity threshold concentrations (e.g., LC50 or predicted no-effect concentration (PNEC)) in

https://doi.org/10.1016/j.scitotenv.2019.135234 0048-9697/Ó 2019 Elsevier B.V. All rights reserved.

Please cite this article as: F. Wu, Y. Gao, Z. Zuo et al., Different decreasing rates of chemical threshold concentrations can be explained by their toxicokinetic and toxicodynamic characteristics, Science of the Total Environment, https://doi.org/10.1016/j.scitotenv.2019.135234

2

F. Wu et al. / Science of the Total Environment xxx (xxxx) xxx

order to protect the ecosystem from unacceptable harm. The species sensitivity distributions (SSDs) approach has been often used worldwide since the 1990s for deriving environmental quality criteria (Schmitt-Jansen and Altenburger, 2005; Staples et al., 2004; Stephan, 2002). SSDs use statistical distribution models (e.g., lognormal or log-logistic distribution) to construct cumulative distribution plots of effect concentrations (e.g., median lethal concentration, LC50, sub-lethal/chronic effects using EC10 or NECs), from which a hazardous concentration based on the LC50 value is estimated as HC5 (the concentration corresponding to the lower 5th percentile of cumulative probability), in order to protect 95% of the species. To determine the toxicity threshold that can be used to protect the ecosystem structure and function, HC5 is generally divided by an assessment factor (AF) (Amiard et al., 2015; Iino et al., 2005). While the SSDs approach is a common method that is used worldwide for deriving water quality criteria, it also has several shortcomings (Belanger et al., 2017; Forbes and Calow, 2002), such as not assessing multiple exposure profiles and accounting for the exposure time (Fox and Billoir, 2013; King et al., 2015; Ockleford et al., 2018). Previous studies have demonstrated that the toxicants’ LC50 varies with the exposure time and HC5 derived from SSD based on the LC50 value is expected to be time-dependent (Fox and Billoir, 2013; Jager et al., 2006; Zhang and Van Gestel, 2017). For example, Fox and Billoir theoretically presented that HC5 decreased with time (Fox and Billoir, 2013), and King et al. (2015) noted that the HC5 estimated by the LC50 decreased with exposure time, and that the HC5 based on 72-h exposure experiments was under protective (King et al., 2015). However, few studies have examined how the exposure duration affects the variability in the HC5 values and if there are differences in the time evolution of HC5 among different toxicants. The present study seeks to use the LC50 data with at least 3 different exposure durations (24, 48 and 96 h) for every compound, together with mathematical methods and the SSD model to evaluate the variability in the HC5 values associated with exposure time, which involves two tasks: (1) quantification of the possible different decreasing rates of HC5 with increasing exposure time for various pollutants, and (2) exploration of the mechanism of this phenomenon by the analysis of toxicokinetic (TK) and toxicodynamic (TD) characteristics. This work is beneficial for identifying the magnitude of the variability in the HC5 values based on the exposure time and provides guidance for reducing the uncertainty in the time-course HC5 values based on the TK-TD process.

2. Materials and methods 2.1. Toxicity data All the data used to construct the SSDs were obtained from the ECOTOX database in 1915–2019 (http://www.epa.gov/ecotox/). The criteria for the selection of the chemicals were as follows: (1) 4 or more species were available for each toxicant, considering the uncertainty in HC5 can be significantly reduced when number of data points are > 4 (Del Signore et al., 2016); (2) the datasets of each species contained acute toxicity endpoints (LC50) based on survival with exposure durations of 24, 48 and 96 h. Then, SSDs were constructed and HC5 values with exposure durations of 24, 48 and 96 h of all of the chemicals were calculated, with their 95% confidence intervals based on the bootstrap method (Efron, 1979); (3) For each toxicant, the toxicology maybe different for the same species, mainly due to the variation of external factors such as test conditions. The geometric mean, as the normal method of dealing with this variation (Xu et al., 2015), was calculated and used as the final toxicity value. Following the criteria, 20 chemi-

cals, namely, aldrin, DDT, deltamethrin, heptachlor, dieldrin, toxaphene, methoxychlor, endosulfan, pentachlorophenol (PCP), chlorpyrifos, cypermethrin, carbofuran, malathion, carbaryl, aldicarb and 2,4,5-trichlorophenol, propiconazole, fluoranthene, alpha-cypermethrin and diazinon were selected. The corresponding CAS numbers for the 20 chemicals was provided in Table S1. Additional information about the number of species and species group is provided in Table S2. 2.2. Calculation of the decreasing rates of HC5 Several models were developed based on the assumption of a decrease in the LC50 with time, two often used models are as follows: model I (Crommentuijn et al., 1994): LC50 = a/(1  exp(b  t)) and model II (Green, 1965):LC50 = a + 1/(b  t). Considering that the HC5 calculated based on LC50 decreases with increasing exposure time and converges to a constant (HC5 based on the NECs) (King et al., 2015), the relationship between HC5 and time can be described as following 2 models:

HC 5 ðt Þ ¼

HC 51 1  ekt

HC 5 ðt Þ ¼ HC 51 þ

1 kt

ð1Þ ð2Þ

where HC5(t) is the HC5 value after t hours of exposure (mg/L), and t is the exposure time (h). The HC51 parameter is the asymptotic HC5, describing the expected threshold concentration for an infinite exposure time. The k indicates how fast HC5 decreases with exposure time [1/h]. Equations (1) and (2) were used to fit the HC5 values at 24, 48, and 96 h by nlsLM function in R, and the results for all of the chemicals were shown in Table S3 and S4. The best-fitting model was identified using the Akaike Information Criterion (AIC) (Burnham and Anderson, 2004). We selected the parameter k as the indicator describing the rate of decreasing in hazardous concentrations (HC5) corresponding to the chemicals. 2.3. Toxicokinetic and toxicodynamic parameters TK-TD models can be used to simulate the time-course of processes resulting in toxic effects on organisms. The TK illuminates the processes of uptake and elimination of a toxicant, and the TD links the internal concentration to the effect on individual organism (Jager et al., 2011). Here, the generalized unified threshold models of survival (GUTS) is used to simulate the time-course of processes leading to toxic effects on organisms, considering GUTS is a more general model (Ashauer et al., 2016; Ockleford et al., 2018). In the GUTS framework, the chemical’s TK process is determined by both the uptake of a chemical in proportion to an exposure concentration and elimination of the chemical in relation to the internal concentration. Its simplest form, a one-compartment TK model with first-order kinetics, can be used as follows (Jager et al., 2011):

dC i ðtÞ ¼ kin C w ðt Þ  kout C i ðtÞ dt

ð3Þ

where Ci is the internal concentration (mol/kg), Cw is the exposure concentration (mol/L), kin is the uptake rate constant (L/(kg∙t)) and kout is the elimination rate constant (1/t). The bioconcentration factor (BCF) is an important indicator that describes the accumulation level of toxicants (TK) in organisms which consider both the uptake and elimination processes. The BCF reflects the bioconcentration potential of chemicals and is considered to be an inherent property of the substance that is independent of the actual chemical concentration in the environment (Gobas et al., 2009). BCF can be

Please cite this article as: F. Wu, Y. Gao, Z. Zuo et al., Different decreasing rates of chemical threshold concentrations can be explained by their toxicokinetic and toxicodynamic characteristics, Science of the Total Environment, https://doi.org/10.1016/j.scitotenv.2019.135234

F. Wu et al. / Science of the Total Environment xxx (xxxx) xxx

expressed as the ratio between the uptake from water (kin) and the elimination constant (kout). In practice, the BCF of the chemicals can also be estimated from regression-based method using the octanol– water partitioning coefficient (KOW) (Grisoni et al., 2018), considering that the European Chemical Agencies (ECHA) guidance document and various studies indicates that the log BCF increases linearly with log KOW values < 5 (Arnot and Gobas, 2006; Veith et al., 1979). The BCF of the chemicals were obtained at http:// www.chemspider.com. For the TD process, the scaled damage, denoted D(t) (e.g. in mol/ kg), is connected to the internal concentrations, according to increasing internal concentrations and a possible individual-level repair of the scaled damage with damage repair rate constant kR (1/t) following:

dDðtÞ ¼ kR ðC i ðt Þ  DðtÞÞ dt

ð4Þ

The damage dynamics is then related to an individual hazard state variable, resulting in mortality when an internal damage threshold is exceeded. Two death mechanisms (Ockleford et al., 2018), namely stochastic death (SD) and individual tolerance (IT), are used in this study. The SD model has one value for the threshold of survival and after exceeding it, an organism has an increased probability to die, presented in Eqs. (5)–(6):

dHðtÞ ¼ b  maxð0; Dðt Þ  zÞ dt

ð5Þ

SSD ðtÞ ¼ eHðtÞ  ehb t

ð6Þ

where dH(t)/dt is the hazard rate [1/t], b is the killing rate constant (kg/(mol∙t), z is the threshold for effects (mol/kg), hb is the background hazard rate (1/t), SSD(t) is the survival probability [unitless] in GUTS-SD model. The IT model assumes the threshold for death to be drawn from an individual tolerance distribution, which can be described in Eqs. (7)–(8):

1

F ðt Þ ¼ 1þ

max DðsÞ b ð0stm Þ

SIT ðt Þ ¼ ð1  FðtÞÞ  ehb ðtÞ

ð7Þ

ð8Þ

where F(t) is the cumulative log  logistic distribution of the tolerance threshold over time [unitless], b is shape parameter for the distribution of threshold, m is the median of the distribution of thresholds (mol/kg). SIT(t) is the survival probability [unitless] in GUTS-IT model. More details about the TD parameters for several chemicals used in this study are described in Tables S5-S6. The model used by Schuler et al. (2004) is following damage assessment model (DAM) proposed by Lee et al., 2002. Basically, GUTS integrates all previously published TKTD models (including DAM model) for survival that we are aware of (i.e., models that include aspects of both TK and TD) (Jager et al., 2011).

3

3. Results 3.1. Relationships between HC5 and exposure time for different chemicals Details regarding the log-normal fitted SSD curves and the HC5time plots for the 20 chemicals examined in this study were provided in Figures S1-S2. For each compound, the asymptotic concentrations (HC51s) were extrapolated based on the equation (1), and the corresponding information is presented in Fig. 1. The HC51 of aldicarb (144.91 mg/L) is>100 mg/L, suggesting that the chemical may generate low toxicity for the tested species (Moore et al., 2010). With the exception of PCP (38.12 mg/L), carbaryl (4.77 mg/L) and carbofuran (2.67 mg/L), the HC51 values for the other chemicals are close to zero (Fig. 1). For example, the HC51 of dieldrin is approximately 0.0003 mg/L, which shows that dieldrin may elicit strong toxicity to organisms, leading to a potential relative higher risk to ecosystem. It is advised that these pollutants with lower HC51 should be paid more attentions by the management agencies in the ecological risk assessment. 3.2. Decreasing rates of HC5 for chemicals The parameter k [1/h] was estimated by Eq (1) and was selected as the indicator describing the rate of decreasing in hazardous concentrations (HC5) corresponding to the chemicals. For all of the chemicals, the HC5 values calculated based on the LC50 values decrease with increasing exposure time. However, their decreasing rates of HC5 (k) are different: for some chemicals the k values are lower than 0.001 (e.g., fluoranthene, diazinon, methoxychlor and dieldrin), while for some chemicals the k values are higher than 0.5 (e.g., PCP and aldicarb) (Table 1). 3.3. The effects of TK-TD parameters on the decreasing rates of HC5 Fig. 2 showed that the k values decrease significantly with log (BCF) increasing (p < 0.01), indicating that BCF is an important factor that affects the rate of decrease of HC5 (the parameter k). Fig. 3 showed that the rate of decrease of HC5 (k) is positively related to the parameter kR both in the GUTS-SD and GUTS-IT model (R2 > 0.5, p < 0.01), but k is not related to other TD parameters (Figures S3S4), which suggests that the damage recovery rates might play important role in explaining the decreasing rates of HC5 (i.e. time variability of threshold concentration). As shown in Fig. 3, the faster toxicodynamic recovery (i.e. larger kR values), the slower decreasing rate of HC5 is (i.e. the larger k values). This phenomenon can be attributed to the assumptions about how death affect survival after constant exposure (Ashauer et al., 2015). An extreme scenario assumes that the tested organisms hold instantaneous toxicodynamic recovery during constant exposure duration (kR = infinite), the corresponding survival remains constant, i.e. LC50 is consistent at different exposure time. And subsequently the decreasing rate of HC5 based on LC50s (k value) is timeindependent during exposure duration. In contrast, to organism with slower toxicodynamic recovery (lower kR), the survival declines with increasing exposure time, and the longer time is needed for organisms to reach time-independence HC5 (lower k value).

2.4. Statistical analysis

4. Discussion

Relationships between k and TK-TD parameters were examined by linear regression (correlation coefficient R2 and significant level p < 0.01) through a maximum likelihood method. All the statistical analyses were implemented and solved in R 3.5.0 (http://www.Rproject.org/).

4.1. Relationships between HC5 and exposure time for different chemicals Recent studies demonstrated that the HC5 is not time invariant (Fox and Billoir, 2013; King et al., 2015). The present study

Please cite this article as: F. Wu, Y. Gao, Z. Zuo et al., Different decreasing rates of chemical threshold concentrations can be explained by their toxicokinetic and toxicodynamic characteristics, Science of the Total Environment, https://doi.org/10.1016/j.scitotenv.2019.135234

4

F. Wu et al. / Science of the Total Environment xxx (xxxx) xxx

Fig. 1. Time-course of the hazardous concentrations (HC5s) derived from the log-normal species sensitivity distributions (SSDs) based on 24 h, 48 h, (72 h) and 96 h-LC50s for the 20 chemicals examined in this study. Table 1 Different decreasing rates (k) of the hazardous concentrations (HC5s) calculated by equation (1) for the 20 chemicals examined in this study with their standard error. Chemical

k [1/h]

Standard Error

Fluoranthene Dieldrin Methoxychlor Diazinon Heptachlor Deltamethrin DDT Cypermethrin Aldrin Propiconazole Toxaphene Alpha-cypermethrin Carbaryl Chlorpyrifos Malathion Endosulfan 2,4,5-Trichlorophenol Carbofuran Aldicarb PCP

1.23E-06 4.28E-06 3.95E-04 7.04E-04 1.10E-03 4.29E-03 5.22E-03 8.82E-03 9.16E-03 9.58E-03 1.15E-02 1.23E-02 1.61E-02 1.94E-02 2.01E-02 2.86E-02 3.39E-02 4.95E-02 5.19E-02 5.70E-02

2.69E-03 4.03E-03 2.43E-03 5.25E-03 3.65E-03 1.05E-02 2.28E-03 7.62E-03 2.02E-02 5.67E-03 9.31E-03 1.07E-02 3.60E-03 5.12E-03 4.12E-03 5.99E-03 6.18E-03 2.38E-02 2.02E-02 4.98E-03

extended some common results of the time-dependent of hazardous concentration (HC5) by the analysis of various chemicals.

In Fig. 1, the hazardous concentration (HC5) decreases with exposure time among all exanimated chemicals. However, the decreasing rates of HC5 are not consistent between different chemicals, for example, a steady state for HC5 to carbofuran is observed to be reached within 96 h exposure (2.67 mg/L), but for dieldrin the HC5 still decreases after 168 h, which indicates that for these chemicals, only toxicity values with the longest 96 h exposure duration are used to derive acute water quality criteria, as a standard procedure to derived environmental threshold, might be insufficient, it should be conducted longer exposure experiments. Fig. 1 compares the decreasing rate of threshold of each toxicant visually and provides the predicted asymptotic hazardous concentration for 5% of the species (HC51), which examined the important issue of time dependence in the risk assessment. 4.2. Decreasing rates of HC5 for chemicals. The smaller the k value, the slower the HC5 declines and vice versa. For example, the k values for carbofuran and dieldrin are 0.0495 1/h and 0.00000428 1/h, respectively, suggesting that the HC5 of dieldrin decreases slower than the HC5 of carbofuran. Concretely, the HC5 at 96 h and HC51 of carbofuran are 2.69 mg/L and 2.67 mg/L, respectively, whereas for dieldrin, the HC5 at 96 h and HC51 are 0.74 mg/L and 0.0003 mg/L, respectively. Thus, for these chemicals with large k values (k > 0.05), their HC5s at 96 h nearly

Please cite this article as: F. Wu, Y. Gao, Z. Zuo et al., Different decreasing rates of chemical threshold concentrations can be explained by their toxicokinetic and toxicodynamic characteristics, Science of the Total Environment, https://doi.org/10.1016/j.scitotenv.2019.135234

F. Wu et al. / Science of the Total Environment xxx (xxxx) xxx

Fig. 2. Relationship between log BCF and the decreasing rate of HC5 (k), where k is estimated by equation (1). The dotted line corresponds to the 95% confidence interval.

reach their HC51s. In contrast, for these chemicals with small k values (k < 0.001), their HC5s need longer exposure time (>96 h) to reach their HC51s, demonstrating the incipient thresholds (e.g. HC5) based on short-term (96 h) exposure experiments may be under protective. Therefore, it is advised to use larger AFs for chemicals with the lower parameter k values (<0.001), such as dieldrin, methoxychlor, fluoranthene, and diazinon. It should be noted for some chemicals, the HC5s were estimated based on relative few species, which may account for some of the uncertainties in their k values. For example, SSDs of fluoranthene and propiconazole only consist of 4 species, one should be careful when adopting their HC5s and k values. 4.3. The effects of TK-TD parameters on the decreasing rates of HC5 We found that then TK parameter (k) decreases significantly with log (BCF) increasing (p < 0.01). The underlying cause about this

5

relationship between k and BCF can be explained by the fact that chemicals with higher BCF need a longer to reach steady-state (equilibrium) and the time in toxicity tests is too short to reach that (McCarty et al., 2011). The uptake and accumulation of environmental toxicants is the basis of the triggering of acute and chronic toxicity. Therefore, the cumulative toxicity action is usually one of the indicators used for assessing the toxicity of environmental pollutants and is an important reference for the development of environmental quality criteria for a compound. Bioaccumulation is a key determinant of the fate and the effects of toxicants (Grisoni et al., 2018), in particular for chemical pesticides, because it may lead to the death of the organism when the body concentration accumulated by an organism is higher than a certain tolerance threshold (lethal body concentration). Since the BCF value for the chemicals was estimated from the toxicokinetic process, for compounds with higher Kow or BCF values (i.e., highly hydrophobic chemicals), the time to reach equilibrium with the surrounding aqueous phase is longer, leading to that the relatively short exposure duration in most toxicity test may not be sufficient to reach a time independent toxicity value (e.g., LC501). As described by Henk et al, for toxicants with a log Kow (log-transformed octanol  water partition coefficient) of up to 3.8, the internal concentration should to reach an equilibrium value after 4 days (96 h) of exposure in small fish and smaller aquatic organisms (Verhaar et al., 1999). Many studies also documented that the uptake clearance coefficient (kin) generally increased with increasing log Kow of a compound, while the elimination rate constant (kout) was negatively correlated with log Kow (F. Landrum, 1989; Lydy et al., 1992; W. Lohner and J. Collins, 1987), indicating that the chemicals with the higher Kow were accumulated easily and were eliminated less, leading to a relative lower k value. The mode of action (MOA) is related to the chemical structure (Escher and Hermens, 2002) and often conserved across biota because they are triggered by common molecular initiating events (Ankley et al., 2010). The toxicants in our study covered five chemical classes (organophosphates, carbamates, baseline toxicants, uncouplers, sodium channel modulators), each representing a distinct molecular initiating event (MIE) and cellular toxicity pathway (Ashauer and Jager, 2018). The present study found that for the examined chemicals, going from the reversible (uncoupling: 2,4,5-trichlorophenol, PCP; carbamate AChE inhibition: carbaryl, carbofuran, aldicarb) to the irreversible modes of action (MOAs) (organophosphate AChE inhibition: diazi-

Fig. 3. Relationship between log-transformed kR and the rate of decrease of HC5 (k) for chemicals. The kR is estimated by equation (4) with the death mechanism of SD (A) and IT (B) and k is estimated by equation (1). Color indicates different mode of action (MOA) of chemicals. Black: AChE inhibition (carbamate), Blue: uncoupling of oxidative phosphorylation, Red: AChE inhibition (organophosphates), Green: baseline toxicity, Cyan: neurotoxicity. The dotted line corresponds to the 95% confidence interval. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Please cite this article as: F. Wu, Y. Gao, Z. Zuo et al., Different decreasing rates of chemical threshold concentrations can be explained by their toxicokinetic and toxicodynamic characteristics, Science of the Total Environment, https://doi.org/10.1016/j.scitotenv.2019.135234

6

F. Wu et al. / Science of the Total Environment xxx (xxxx) xxx

non, chlorpyrifos, malathion), then to baseline toxicity (propiconazole, fluoranthene), the kR value decreased, which could provide some reference for the damage dynamics depending on the mechanisms of action of the compound. Generally, our results supported the hypothesis that TD parameters (especially for the kR) clustered according to mode of action (Ashauer and Jager, 2018; Ashauer et al., 2015). Thus, the TD parameters could be applied in ecological risk assessment of time-variable exposures, although there is some variability (e.g. kR values provided in GUTS-SD model are for two species only, GUTS-IT model for four species). Thus, further research is necessary to determine the effects of the TD parameters on the rates of decrease of HC5 with more species data in the future. Baseline toxicity is also a reversible mechanism, in which the response is directly related to the concentration in the membrane, the LC50 of baseline toxicants is quantitatively related to the bioaccumulation kinetics, i.e., the time it takes for the organism to reach equilibrium with the surrounding aqueous phase. Because the time to reach equilibrium is longer for propiconazole and fluoranthene with higher BCF (hydrophobicity), leading to relatively low k values (Fig. 1). The present study showed that the chemical’s HC5 decreased with time. The decreasing rate of HC5 for chemicals is associated with the TK parameters (BCF, p < 0.01) and TD parameters (kR, p < 0.01). However, we noticed that there are several chemicals lies outside of the confidence interval. For example, the k values of the diazinon and propiconable are relative lower (<0.001), while for PCP and carbofuran, the k values are relative higher. These deviations can be explained by the fact that the damage recovery rate (kR) of diazinon and propiconable are relatively smaller, while for the PCP and carbofuran, their kR values are the largest among the 10 chemicals (Fig. 3). Thus, the decreasing rates of HC5 depend not only on the toxicokinetics (BCF) but also on the toxicodynamics (kR). Environmental Implications. Currently, the quantification of the relationship of ecological thresholds to TK-TD model parameters remains challenging for chemicals risk regulatory framework. The present study demonstrated that exposure time is an important determinant of ecological thresholds based on the SSD approach. Furthermore, the decreasing rates (k) of chemicals’ ecological thresholds are negatively associated with BCF and positively related to the kR. Therefore, we recommend that for chemicals with high BCF or low kR, higher AF should be adopted in deriving the environmental acute criteria and ecological risk assessment, such as fluoranthene and diazinon. Recently, TK-TD model has been proposed to be used in the Tier-2 effect assessment procedure to expand the risk assessment of pesticides for aquatic organisms (Ockleford et al., 2018). Our work also highlights the importance of the TK and TD parameters in explaining the time varying of the chemical threshold concentrations, which can be viewed as efforts to extend the applicability of the TK-TD to ecological risk assessments. As the underlying mechanism of the TK parameters on the time evolution of the chemical threshold concentrations is generally well understood, the TD parameters effect is more complex, due to the fact that the toxicodynamic recovery is a result of cellular and physiological compensating mechanisms and repair in an organism’s stress response to the toxicant (Ashauer et al., 2015). Thus, more studies are necessary to determine the effects of the TD parameters on the rates of decrease of HC5, given that the TD parameters (e.g. damage recovery rate) across different chemicals and species are investigated in the future. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments This study was supported by the National Key Research and Development Program of China (2018YFC1406403), National Natural Science Foundation of China (41877498) and Major Science and Technology Program for Water Pollution Control and Treatment (2017ZX07301).

Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.scitotenv.2019.135234. References Amiard, J.-C., Amiard-Triquet, C., Mouneyrac, C., 2015. Chapter 2 - Conventional Risk Assessment of Environmental Contaminants, Aquatic Ecotoxicology. Academic Press, pp. 25–49. Ankley, G.T., Bennett, R.S., Erickson, R.J., Hoff, D.J., Hornung, M.W., Johnson, R.D., et al., 2010. Adverse Outcome Pathways: A Conceptual Framework To Support Ecotoxicology Research And Risk Assessment. Environ. Toxicol. Chem. 29, 730– 741. Arnot, J.A., Gobas, F., 2006. A review of bioconcentration factor (BCF) and bioaccumulation factor (BAF) assessments for organic chemicals in aquatic organisms. Environ. Rev. 14, 257–297. Ashauer, R., Albert, C., Augustine, S., Cedergreen, N., Charles, S., Ducrot, V., Focks, A., Gabsi, F., Gergs, A., Goussen, B., Jager, T., Kramer, N.I., Nyman, A.M., Poulsen, V., Reichenberger, S., Schafer, R.B., Van den Brink, P.J., Veltman, K., Vogel, S., Zimmer, E.I., Preuss, T.G., 2016. Modelling survival: exposure pattern, species sensitivity and uncertainty. Sci. Rep. 6, 11. Ashauer, R., Jager, T., 2018. Physiological modes of action across species and toxicants: the key to predictive ecotoxicology. Environ. Sci-Proc. Imp. 20, 48–57. Ashauer, R., O’Connor, I., Hintermeister, A., Escher, B.I., 2015. Death dilemma and organism recovery in ecotoxicology. Environ. Sci. Technol. 49, 10136–10146. Belanger, S., Barron, M., Craig, P., Dyer, S., Galay-Burgos, M., Hamer, M., Marshall, S., Posthuma, L., Raimondo, S., Whitehouse, P., 2017. Future needs and recommendations in the development of species sensitivity distributions: Estimating toxicity thresholds for aquatic ecological communities and assessing impacts of chemical exposures. Integr. Environ. Assess. Manage. 13, 664–674. Burnham, K.P., Anderson, D.R., 2004. Multimodel inference: understanding AIC and BIC in model selection. Sociological Methods & Research 33, 261–304. Crommentuijn, T., Doodeman, C.J.A.M., Doornekamp, A., Van Der Pol, J.J.C., Gestel, C. A.M., Bedaux, J.J.M., 1994. Lethal body concentrations and accumulation patterns determine time- dependent toxicity of cadmium in soil arthropods. Veith, D., DeFoe, D.L., Bergstedt, B.V, 1979. measuring and estimating the bioconcentration factor of chemicals in fish. Del Signore, A., Hendriks, A.J., Lenders, H.J.R., Leuven, R., Breure, A.M., 2016. Development and application of the ssd approach in scientific case studies for ecological risk assessment. Environ. Toxicol. Chem. 35, 2149–2161. Efron, B., 1979. Bootstrap methods: another look at the Jackknife. Ann. Statist. 7, 1– 26. Escher, B.I., Hermens, J.L.M., 2002. Modes of action in ecotoxicology: Their role in body burdens, species sensitivity, QSARs, and mixture effects. Environ. Sci. Technol. 36, 4201–4217. F. Landrum P., 1989. Bioavailability and toxicokinetics of polycyclic aromatic hydrocarbons sorbed to sediments for the amphipod Pontoporeia hoyi. Forbes, V.E., Calow, P., 2002. Species sensitivity distributions revisited: a critical appraisal. Hum. Ecol. Risk Assess. 8, 473–492. Fox, D.R., Billoir, E., 2013. Time-dependent species sensitivity distributions. Environ. Toxicol. Chem. 32, 378–383. Gobas, F.A.P.C., de Wolf, W., Burkhard, L.P., Verbruggen, E., Plotzke, K., 2009. Revisiting bioaccumulation criteria for POPs and PBT assessments. Integr. Environ. Assess. Manage. 5, 624–637. Green, R.H., 1965. Estimation of tolerance over an indefinite time period. Ecology 46, 887. Grisoni, F., Consonni, V., Vighi, M., 2018. Detecting the bioaccumulation patterns of chemicals through data-driven approaches. Chemosphere 208, 273–284. Iino, F., Takasuga, T., Senthilkumar, K., Nakamura, N., Nakanishi, J., 2005. Risk assessment of short-chain chlorinated paraffins in Japan based on the first market basket study and species sensitivity distributions. Environ. Sci. Technol. 39, 859–866. Jager, T., Albert, C., Preuss, T.G., Ashauer, R., 2011. General unified threshold model of survival - a toxicokinetic-toxicodynamic framework for ecotoxicology. Environ. Sci. Technol. 45, 2529–2540. Jager, T., Heugens, E.H.W., Kooijman, S.A.L.M., 2006. Making sense of ecotoxicological test results: towards application of process-based models. Ecotoxicology 15, 305–314. King, G.K.K., Delignette-Muller, M.L., Kefford, B.J., Piscart, C., Charles, S., 2015. Constructing time-resolved species sensitivity distributions using a hierarchical toxico-dynamic model. Environ. Sci. Technol. 49, 12465–12473.

Please cite this article as: F. Wu, Y. Gao, Z. Zuo et al., Different decreasing rates of chemical threshold concentrations can be explained by their toxicokinetic and toxicodynamic characteristics, Science of the Total Environment, https://doi.org/10.1016/j.scitotenv.2019.135234

F. Wu et al. / Science of the Total Environment xxx (xxxx) xxx Lee, J.H., Landrum, P.F., Koh, C.H., 2002. Prediction of time-dependent PAH toxicity in Hyalella azteca using a damage assessment model. Environ. Sci. Technol. 36, 3131–3138. Lydy, M., Baumann, C.P., Oris, J., Fisher, S., 1992. Effects of sediment organic carbon content on the elimination rates of neutral lipophilic compounds in the midge (Chironomus. Riparius). McCarty, L.S., Landrum, P.F., Luoma, S.N., Meador, J.P., Merten, A.A., Shephard, B.K., van Wezel, A.P., 2011. Advancing environmental toxicology through chemical dosimetry: external exposures versus tissue residues. Integr. Environ. Assess. Manage. 7, 7–27. Moore, D.R.J., Thompson, R.P., Rodney, S.I., Fischer, D., Ramanarayanan, T., Hall, T., 2010. Refined aquatic risk assessment for aldicarb in the United States. Integr. Environ. Assess. Manage. 6, 102–118. Ockleford, C., Adriaanse, P., Berny, P., Brock, T., Duquesne, S., Grilli, S., HernandezJerez, A.F., Bennekou, S.H., Klein, M., Kuhl, T., Laskowski, R., Machera, K., Pelkonen, O., Pieper, S., Smith, R.H., Stemmer, M., Sundh, I., Tiktak, A., Topping, C.J., Wolterink, G., Cedergreen, N., Charles, S., Focks, A., Reed, M., Arena, M., Ippolito, A., Byers, H., Teodorovic, I., 2018. Scientific Opinion on the state of the art of Toxicokinetic/Toxicodynamic (TKTD) effect models for regulatory risk assessment of pesticides for aquatic organisms. EFSA J. 16, e05377. Schmitt-Jansen, M., Altenburger, R., 2005. Predicting and observing responses of algal communities to photosystem II-herbicide exposure using pollutioninduced community tolerance and species-sensitivity distributions. Environ. Toxicol. Chem. 24, 304–312.

7

Schuler, L.J., Landrum, P.F., Lydy, M.J., 2004. Time-dependent toxicity of fluoranthene to freshwater invertebrates and the role of biotransformation on lethal body residues. Environ. Sci. Technol. 38, 6247–6255. Staples, C., Mihaich, E., Carbone, J., Woodburn, K., Klecka, G., 2004. A weight of evidence analysis of the chronic ecotoxicity of nonylphenol ethoxylates, nonylphenol ether carboxylates, and nonylphenol. Hum. Ecol. Risk Assess. 10, 999–1017. Stephan, C.E., 2002. Use of species sensitivity distributions in the derivation of water quality criteria for aquatic life by the U.S. Environmental Protection Agency. Verhaar, H.J.M., De Wolf, W., Dyer, S., Legierse, K., Seinen, W., Hermens, J.L.M., 1999. An LC50 vs time model for the aquatic toxicity of reactive and receptormediated compounds. Consequences for bioconcentration kinetics and risk assessment. Environ. Sci. Technol. 33, 758–763. Lohner, T.W., Collins, W.J., 1987. Determination of uptake rate constants for six organochlorines in midge larvae. Environ. Toxicol. Chem. Xu, F.L., Li, Y.L., Wang, Y., He, W., Kong, X.Z., Qin, N., Liu, W.X., Wu, W.J., Jorgensen, S. E., 2015. Key issues for the development and application of the species sensitivity distribution (SSD) model for ecological risk assessment. Ecol. Ind. 54, 227–237. Zhang, L., Van Gestel, C.A.M., 2017. Toxicokinetics and toxicodynamics of lead in the soil invertebrate Enchytraeus crypticus. Environ. Pollut. 225, 534–541.

Please cite this article as: F. Wu, Y. Gao, Z. Zuo et al., Different decreasing rates of chemical threshold concentrations can be explained by their toxicokinetic and toxicodynamic characteristics, Science of the Total Environment, https://doi.org/10.1016/j.scitotenv.2019.135234