Effects of temperature on the dissipation of total- and water-extractable pesticides in Japanese soils

Abstract

We investigated the dissipation of 27 pesticides in five Japanese soils at three temperatures and the variability of activation energies (E_a). The dissipation of total pesticides extracted sequentially using water and acetone was fitted to a single first-order (SFO) model. The E_a values calculated from the dissipation rate constants of the SFO model showed a normal distribution with a median of 61.1 kJ mol⁻¹. The dissipation of water-extractable pesticides (i.e., phytoavailable pesticides) was fitted to a double first-order in parallel model with two dissipation rate constants: k₁ and k₂. The E_a values calculated from k₁ and k₂ showed normal or lognormal distribution, and the medians of the normal distribution calculated from k₁ and k₂ were 62.8 and 45.2 kJ mol⁻¹, respectively. Furthermore, the method for estimating the biphasic dissipation of phytoavailable pesticides at different temperatures by using the median E_a values of the laboratory experiment was demonstrated in a field experiment.

Introduction

Recently in Japan, some pesticides applied to previous crops and remaining in the soil have been detected in succeeding crops at levels exceeding the uniform residue limit of 0.01 mg kg⁻¹ applied for all combinations of crops and pesticides without maximum residue limits.^1,2) It is prohibited to sell or distribute crops that contain pesticides above the uniform residue limit. To prevent pesticide contamination of succeeding crops, understanding the dissipation behavior of phytoavailable pesticides in soil is crucial. Our previous research³⁾ showed that the uptake concentrations of leafy vegetables cultivated in pesticide-applied soils had a higher positive correlation with the concentrations of water-extractable than total-extractable pesticides (i.e., the concentrations of pesticides extracted sequentially from soils using water and acetone). This result suggests that the phytoavailability of the residual pesticides in soil can be assessed using water-extractable concentrations in soil. In addition, we indicated that the dissipation behavior of water-extractable pesticides was quite different from that of total-extractable pesticides in soils with organic carbon (OC) content of 1% or more.⁴⁾ In other words, the dissipation time for 50% (DT₅₀) of water-extractable pesticides in soil was clearly shorter than that of total-extractable pesticides, and the dissipation of water-extractable pesticides was better fitted by biphasic models, i.e., a double first-order in parallel (DFOP) model as compared with a single first-order (SFO) model applied to the dissipation of total-extractable pesticides in soil.

The dissipation process in soil is one key factor governing the fate and behavior of pesticides in the environment and is affected by soil conditions such as temperature, moisture, pH and microbial activity.^5,6) Japan is located in the mid-latitudes of the Northern Hemisphere, has a long landmass extending north and south and has four distinct seasons; thus, temperatures differ greatly depending on the region and season.⁷⁾ Investigating the effect of temperature on the dissipation of pesticides in soils is very important for estimating the pesticide concentrations in succeeding crops taking temperature fluctuations into account.

The effects of temperature on the dissipation of pesticides expressed by the SFO model have been reported in many previous studies.^8–15) In general, the relationship between the first-order dissipation rate constant (k) of the SFO model and temperature is analyzed using the activation energy (E_a) calculated from the Arrhenius equation.¹²⁾ Using E_a values, it is possible to estimate the k values at different temperatures. The scientific report of the European Food Safety Authority (EFSA) indicated that the statistical distribution of median E_a values for each pesticide, derived from 99 datasets corresponding to 53 pesticides, was lognormal with a median value of 65.4 kJ mol⁻¹.¹²⁾ It is suggested that this value be used as a default in an environmental exposure assessment considering the effect of temperature. However, there is no report on the distribution of E_a values derived from the SFO model in Japanese soils including volcanic ash soil (Andisol). Although Andisols cover only about 0.8% of the earth’s surface,¹⁶⁾ they play a crucial role in Japanese agriculture, considering that they cover approximately half of all upland fields in Japan.¹⁷⁾

In addition, it is necessary to examine the effect of temperature on the biphasic dissipation of water-extractable pesticides, so as to analyze the seasonal and regional difference in pesticide concentrations in succeeding crops. The DFOP model consists of two compartments: the first compartment with a faster dissipation rate governs the earlier part of the dissipation curve; and the second compartment with a slower dissipation rate governs the later part.¹⁸⁾ The dissipation behavior in the first and second compartments follows first-order kinetics, and thus two dissipation rate constants are required: k₁ and k₂. So far, there is no report on the variability of each E_a value calculated for k₁ and k₂ according to the Arrhenius equation.

The aims of this study were to investigate the distribution of E_a values measured using typical Japanese soils, including Andisols, and 27 pesticides with different chemical classes and physico-chemical properties to determine whether the distribution of E_a values in Japanese soils differed from those reported by the EFSA, and to examine the differences in E_a values between k₁ and k₂ so as to estimate the biphasic dissipation of phytoavailable pesticides considering temperature fluctuations. Furthermore, the method used to estimate the first-order and biphasic dissipations of pesticides at different temperatures based on E_a values from laboratory experiments was verified by a field experiment.

Materials and Methods

1. Laboratory experiments

The test conditions—chemicals, soils, laboratory incubation except for incubation temperature, sequential extraction of soils and pesticide analysis—are described in our previous report,⁴⁾ in which soil samples were incubated at 25±2°C.

In brief, 27 pesticides (chemical purity of >97.0%) including 15 insecticides and 12 fungicides were selected on the basis of hydrophobicity, i.e., the range of their octanol–water partition coefficients (log P_ow=−0.549–4.84) as shown in Table 1. On the basis of the analytical methods, pesticides were divided into three analytical groups: A, B and C. Five typical Japanese agricultural soils, with a wide range of OC contents (0.6–86.5 mg g⁻¹), including two Andisols were used in laboratory experiments (Table 2).

Table 1. Test pesticides^a)

Compound	CAS no.	Pesticide typeb)	Substance group	log Powc)	log Swd)	Analytical groupe)
						Laboratory test	Field test
Dinotefuran	165252-70-0	I	neonicotinoid	−0.549	4.60	A	A
Imidacloprid	138261-41-3	I	neonicotinoid	0.570	2.79	A	A
Dimethoate	60-51-5	I	organophosphate	0.704	4.60	A
Clothianidin	210880-92-5	I	neonicotinoid	0.905	2.53	A	A
Thiacloprid	111988-49-9	I	neonicotinoid	1.26	2.26	A	A
Fosthiazate	98886-44-3	I	organophosphate	1.68	3.95	A	A
Metalaxyl	57837-19-1	F	phenylamide	1.75	3.92	A	A
Ethiprole	181587-01-9	I	phenylpyrazole	1.99	0.964	A
Azoxystrobin	131860-33-8	F	methoxyacrylate	2.50	0.826	C
Methidathion	950-37-8	I	organophosphate	2.57	2.38	A
Fenobucarb	3766-81-2	I	carbamate	2.78	2.62	B	C
Boscalid	188425-85-6	F	pyridinecarboxamide	2.96	0.663	B
Flutolanil	66332-96-5	F	phenylbenzamide	3.17	0.904	B	C
Procymidone	32809-16-8	F	dicarboximide	3.30	0.391	B	C
Fenitrothion	122-14-5	I	organophosphate	3.32	1.28	C
Kresoxim-methyl	143390-89-0	F	oximinoacetate	3.40	0.301	B
Tetraconazole	112281-77-3	F	triazole	3.56	2.19	C
Chloroneb	2675-77-6	F	chlorophenyl	3.58	0.903	B
Diazinon	333-41-5	I	organophosphate	3.69	1.78	C
Propiconazole	60207-90-1	F	triazole	3.72	2.18	C
Fipronil	120068-37-3	I	phenylpyrazole	3.75	0.577	C
Cadusafos	95465-99-9	I	organophosphate	3.85	2.39	B
Diclocymet	139920-32-4	F	carboxamide	3.97	0.805	C
Trifloxystrobin	141517-21-7	F	oximinoacetate	4.50	−0.215	B
Tolclofos-methyl	57018-04-9	F	organophosphate	4.56	−0.150	C	C
Tetradifon	116-29-0	I	bridged diphenyl	4.61	−1.11	C
Fenthion	55-38-9	I	organophosphate	4.84	0.623	B

^a) Reprintcd with permission from ref.4. Copyright (2016) American Chemical Society.^{4) b)} (I) Insecticides; (F) fungicides. ^c) Octanol–water partition coefficient and ^d) water solubility (mg/L) was obtained from the Pesticide Properties Database of IUPAC¹⁹⁾ except for P_ow of diclocymet and S_w of ethiprole, which was obtained from The Pesticide Handbook.^{20) e)} Pesticides were divided into three analytical groups according to the analytical methods. The details of analytical methods are described in our previous report.²¹⁾

Table 2. Test soils.^a)

Soil	Experiment	Classificationb) (USDA)	Classificationc) (Japan)	Textured)	OCe) (mg g−1)	CECf) (cmol(+) kg−1)	Clay (mg g−1)	pH (H2O)
LS1	Laboratory	Entisol	Sand-dune Regosol	Sand	0.6	3.4	24	7.5
LS2	Laboratory	Ultisol	Yellow soil	Light clay	10.2	11.4	390	5.3
LS3	Laboratory	Entisol	Gray lowland soil	Silty clay	14.6	18.2	253	5.8
LS4	Laboratory	Andisol	Andosol	Loam	52.1	33.8	108	5.5
LS5	Laboratory	Andisol	Andosol	Silty loam	86.5	35.4	18	5.8
FS	Field	Andisol	Andosol	Silty loam	49.3	26.0	73	6.3

^a) Data are shown in our previous report.^{4) b)} According to U.S. Department of Agriculture Soil Taxonomy.^{16) c)} According to the criteria adopted by the Cultivated Soil Classification Committee.^{22) d)} According to the International Society of Soil Science.^{23) e)} Organic carbon content. ^f) Cation exchange capacity.

Air-dried soil with an 8 g of dry weight (DW) equivalent was placed in a 50 mL glass centrifuge tube. The water content of the soils was adjusted to 50–60% of the water-holding capacity during the test period. After a pre-incubation period of 10 days, 80 µL of an acetone stock solution (100 µg mL⁻¹) of each pesticide group was added to a soil sample in a test tube; the initial pesticide concentration was 1 µg g⁻¹ DW. Duplicate soil samples were incubated in darkness for 0, 2, 7, 14, 30, 60 and 120 days at 10±2°C and at 35±2°C. After each incubation period, the soil samples were analyzed using a sequential extraction method using distilled water and acetone.

Forty mL of distilled water was added to the soil samples in the tubes. The tubes were agitated on a thermostat shaker (TAITEC, Saitama, Japan) for 24 hr at 120 rpm and 25±2°C in darkness. After being shaken, mixtures were centrifuged at 1200×g for 30 min. Fifteen mL of the supernatant was withdrawn and used to quantify the water-extractable pesticides, and then 20 mL of supernatant was discarded. Subsequently, 30 mL of acetone was added to the remaining sample; the tubes were shaken in a thermostat shaker at 120 rpm and 25±2°C for 20 min, centrifuged at 1200×g for 10 min, and the supernatant was then carefully decanted. This extraction procedure was repeated twice more. The collected supernatant was evaporated in a rotary evaporator to reduce the volume to <15 mL, and then used to quantify the acetone-extractable pesticides in soil.

The aliquots (15 mL) of water extracts and concentrates (<15 mL) of acetone extracts from soils were analyzed using three different methods for each analytical group, as previously described in detail.²¹⁾ The aliquots and concentrates were cleaned with a diatomite column (InertSep K-solute 20 mL; GL Sciences, Tokyo, Japan), and the following solid phase extraction cartridges were used: a PSA column (500 mg; Supelco, Bellefonte, PA, USA), an Accell CM column (500 mg; Waters, Milford, MA, USA) and an ENVI-Carb II/PSA column (500 mg/500 mg; Supelco) for analytical groups A, B and C, respectively. The cleaned samples were analyzed by liquid chromatography-tandem mass spectrometry for group A (Supplemental Tables S1 and S2) and gas chromatography-mass spectrometry for groups B and C (Supplemental Tables S3 and S4).

The concentration of water-extractable pesticides in soil, the C_WE (µg g⁻¹ DW), and the concentration of total-extractable (water- and acetone-extractable) pesticides in soil, the C_TE (µg g⁻¹ DW), were calculated as described in the previous study.⁴⁾

2. Data analysis

Two kinetic models, SFO (Eq. 1) and DFOP (Eq. 2), were used to describe the dissipation of the C_WE and the C_TE over time.¹⁸⁾ The dissipation parameters were obtained using the least-squares method with Solver, the Microsoft Excel add-inn.

  

(1)

  

(2)

where C(t) is the C_WE or the C_TE value after t days, C₀ is the C_WE or the C_TE value at time t=0, k is the dissipation rate constant (day⁻¹), f is the fraction of C₀ applied to the first compartment, and k₁ and k₂ are the dissipation rate constants (day⁻¹) in the first and second compartments, respectively. The goodness of fit for each model was assessed using the correlation coefficient (r) and model error (χ² error, %), which was calculated as follows¹⁸⁾:

  

(3)

where C is the calculated value, O is the observed value, Ō is the mean of all observed values and χ²_m,α is the tabulated chi-square value where m is the degrees of freedom—the number of measurements minus the number of model parameters—and α is the probability (5%).

For the SFO model, the DT₅₀ (days), which is the time required for a 50% decrease in the C_TE, was calculated as follows¹⁸⁾:

  

(4)

The relationship between the dissipation rate constant (k, day⁻¹) and the temperature is represented by the Arrhenius equation¹²⁾:

  

(5)

where A is an empirical constant, E_a is the activation energy (kJ mol⁻¹), R is the gas constant (8.3145 JK⁻¹ mol⁻¹) and T is the absolute temperature (K).

According to Eq. 5, the E_a values were calculated for k of the SFO model and for k₁ and k₂ of the DFOP model using laboratory incubation data at 10, 25 and 35°C. All data at 25°C used for data analysis were reported in our previous study.⁴⁾ To ensure the quality of the E_a calculation, the following conditions for data selection had to be satisfied:

• – The measuring points in laboratory incubation exceeded 4.
• – The r values of the SFO and DFOP models exceeded 0.7.
• – The χ² error values of the SFO and DFOP models were below 15%.¹⁸⁾
• – The f values of DFOP models were not zero or one.
• – When Arrhenius fits were done with three temperature points, the r values between ln k and 1/T for Eq. 5 were below −0.7.
• – When Arrhenius fits were done with two temperature points, the calculated E_a was positive.

It is desirable that Arrhenius fits were conducted using k values for at least three temperatures. However, the number of E_a values calculated from three temperature points was small in this experiment, so the E_a values were also calculated using two temperature points.

3. Field experiments

To verify the method for correcting the dissipation rate constants (k, k₁ and k₂) at different temperatures by using E_a values measured in laboratory experiments, field experiments were conducted in a field located in Tsukuba City, Japan, during spring–summer and autumn–winter periods. Details of the spring–summer experiment were described in a previous report,⁴⁾ and the autumn–winter experiment was conducted in the same way as the spring–summer test except for crop cultivation. The pesticides (dinotefuran, imidacloprid, clothianidin, thiacloprid, fosthiazate, metalaxyl, fenobucarb, flutolanil, procymidone and tolclofos-methyl) and the characteristics of the field soil (FS) are shown in Tables 1 and 2, respectively. The spring–summer and autumn–winter experiments were started on May 12 and November 11, 2015, respectively, and the corresponding daily mean temperature of Tsukuba City during these periods ranged from 16–30°C (mean 23°C) and 1–17°C (mean 8°C), respectively. One liter of a mixed solution of the 10 pesticides (200 mg L⁻¹ each), prepared by dilution of commercial formulations (emulsifiable concentrates, water-soluble powders and water-dispersible powders) with water, was evenly applied to triplicate plots (1 m×1 m) using a watering can. Subsequently, the soil surface (approximately 0–20 cm depth) was tilled with a walking-type tilling apparatus. Furthermore, the seeds of komatsuna (Brassica rapa var. perviridis ‘Yokattana’) were sown and cultivated in each plot only for the autumn–winter experiment. Four cores were taken from each plot at soil depth of 0–10 cm at 0 (immediately after tilling), 2, 7, 14, 21, 28, 35, 42, 49, 56, 63, 70, 77 and 84 days and at 0, 28 and 64 days after the application of pesticides for the spring–summer and autumn–winter experiments, respectively. The cores for each plot were combined and mixed well. The sampled soil equivalent to 5 g DW was analyzed using a sequential extraction method in the same way as in the laboratory experiment.

The temperature correction of the dissipation rate constants using E_a values from laboratory experiments was performed using the following equation¹²⁾:

  

(6)

where k_T1 and k_T2 are the dissipation rate constants at temperatures T₁ (23°C, i.e., 296 K of the mean temperature during the spring–summer experiment) and T₂ (8°C, i.e., 281 K of the mean temperature during the autumn–winter experiment). The k_T1 values were calculated from the SFO (Eq. 1) and DFOP (Eq. 2) models using the dissipation data of the spring–summer experiment, and then the dissipation of pesticides in the autumn–winter experiment was estimated using k_T2 calculated by Eq. 6. We compared the estimated values with the measured values on 28 and 63 days after the application of pesticides for the autumn–winter experiment.

4. Statistical analysis

The statistical outliers of the calculated E_a values were found with the Smirnov–Grubbs test at the 5% level of significance using Microsoft Excel 2010, and then the normality or lognormality (log indicates the natural logarithm) of the E_a distribution was checked by Shapiro–Wilk’s test using SPSS Statistics 25 (IBM Corporation, Armonk, NY, USA). Normality or lognormality assumptions were rejected if the p-value for fit was below 0.05. A compound-specific E_a value was reported to exist; for example, the E_a values of phenylureas were statistically lower than those of non-phenylureas.¹²⁾ Therefore, considering inter-compound variability, the tests for normality or lognormality of E_a distribution were performed not only with individual E_a values across pesticides (all data) but also with the medians of E_a values among test soils (n≥2) for each pesticide.¹²⁾

Results and Discussion

1. Dissipation behavior of pesticides in laboratory experiments

The details of fitting the SFO and DFOP models to the dissipation data for all pesticides and all soils are summarized in Supplemental Tables S5 and S6. The dissipation of pesticides in soils showed different behaviors depending on the temperature and combinations of pesticides and soils, and there were differences between total-extractable and water-extractable concentrations (C_TE and C_WE). Figure 1 shows an example of the fungicide propiconazole in soils LS1 and LS5. The dissipation in soils was influenced by temperature, and the dissipation rate constants in both soils increased with increasing temperature.

Fig. 1. Dissipation of total- and water-extractable propiconazole in soils. SFO and DFOP indicate a single first-order model and a double first-order in parallel model, respectively.

Overall, the dissipation of the C_TE was basically fitted by the SFO model as shown in our previous study.⁴⁾ The χ² error value of the SFO model was mostly below 15% (Fig. 2), which is the threshold value suggested by the Forum for the Co-ordination of Pesticide Fate Models and their Use.¹⁸⁾ However, focusing on the dissipation of the C_TE at a high temperature (35°C), the χ² error values of the SFO model tended to be higher than those of the DFOP model (Fig. 2). In such cases, it was preferable to use the DFOP model to describe the dissipation of the C_TE.

Fig. 2. Comparison of model errors (χ² errors) between a single first-order (SFO) model and a double first-order in parallel (DFOP) model. The dashed line shows the 1 : 1 line.

The dissipation behavior of the C_WE in soil LS1 was similar to that of the C_TE (Fig. 1). Among the tested soils, LS1, which had extremely low OC content, showed the lowest apparent sorption coefficient (K_d,app)⁴⁾ and the highest water extractability (C_WE divided by C_TE). That is, most of the pesticide residues in LS1 were extracted by water. Overall, the χ² error values of the SFO model in LS1 for the C_TE and the C_WE were almost the same as those of the DFOP model (Fig. 2); thus, the dissipation of pesticides in LS1 could be expressed by the SFO model.

However, the dissipation of the C_WE in soil LS5, which had the highest OC content, was faster than that of the C_TE (Fig. 1). As previously reported,⁴⁾ the K_d,app values of pesticides in LS5 increased with time, i.e., the water extractability of pesticides decreased with time, so that differences in the dissipation rate constants of the C_WE and the C_TE arose. The dissipation of the C_WE in LS5 was not fitted by the SFO but by the DFOP model regardless of temperature (Fig. 1). For the dissipation of the C_WE at 35°C, the χ² error value of the SFO model was 29.9%, and the visual fit was not acceptable (Fig. 1). In this case, the DFOP model with a χ² error value of 6.4% was preferable for expressing the dissipation of the C_WE (Fig. 1). A good fitting of the DFOP model to the dissipation data of the C_WE and differences in the χ² errors of the SFO and DFOP models were also observed in soils LS2, LS3 and LS4, and the difference in the χ² error between the SFO and DFOP models tended to increase in the two Andisols (LS4 and LS5) with higher OC content (Fig. 2). The reason for this was that the dissipation of the C_WE was strongly affected by a time-dependent increase in K_d,app,⁴⁾ the rate of which was high for soil with high OC content and showed a biphasic pattern, as shown in our previous study.⁴⁾ Thus, the dissipation of the C_WE in soils, except for LS1 (OC=0.6 mg g⁻¹), should be described by the DFOP model regardless of the temperature.

2. Distribution of E_a values derived from the dissipation of total-extractable pesticides in laboratory experiments

In order to investigate the effect of temperature on the dissipation of the C_TE, the E_a values were derived from Eq. 5. The statistical distributions of the E_a values calculated using the SFO and DFOP models are shown in Supplemental Table S7. The E_a values for the DFOP model of soil LS1 were not obtained because the dissipation of both the C_TE and the C_WE in LS1 could be expressed by the SFO model as described in Section 1 above. The number of individual E_a values calculated for k₁ and k₂ of the DFOP model was low compared to that of the SFO model (Supplemental Table S7). This was because the dissipation of the C_TE for some combinations of pesticides and soils tended to be fitted not by the DFOP model but by the SFO model (in some cases, the f values of DFOP models for the C_TE were zero or one, Supplemental Table S5), the Arrhenius fits were not good, and the calculated E_a values were negative in many cases. Furthermore, the k₂ values of the DFOP model at low temperatures were often calculated as zero (Supplemental Table S1); thus, the calculated number of E_a values for k₂ was particularly low. When the k₂ values are zero, it means that the pesticide is not dissipated in the second soil compartment. However, further experiments with a longer incubation period are needed to demonstrate that k₂ values really are zero.

For the SFO model, the individual calculated E_a values using the k values at three temperatures were statistically described by a normal distribution, that is, the p-values of the Shapiro–Wilk’s test exceeded 0.05, and the median E_a value was 62.9 kJ mol⁻¹ (Supplemental Table S2). Additionally, the median E_a values of each pesticide were normally distributed and the median value of the distribution was 61.1 kJ mol⁻¹ (Table 3). Thus, the median E_a of the distribution derived from the median E_a values of each pesticide was almost the same as that derived from individual E_a values. Focusing on the differences in E_a values among soils (Table 4), the average E_a value of soil LS4, an Andisol, was statistically equivalent to those of other soils. Likewise, the average E_a value for soil LS5, also an Andisol, did not significantly differ from that of other soils. Therefore, no specificity of E_a distribution was observed for Andisols, which are the major upland soils in Japan. The EFSA reported that the distribution of median E_a values of the SFO model for each pesticide measured mainly using European soils had a median value of 65.4 kJ mol⁻¹ and this value would be available as a default value for the environmental exposure assessment.¹²⁾ The median value reported by the EFSA was nearly the same as the corresponding value in our study—of 61.1 kJ mol⁻¹—measured using Japanese soils. These results imply that a median E_a value can be used for environmental exposure assessments in Japanese soils when considering the effects of temperature. However, the E_a values varied depending on the combinations of pesticides and soils. Therefore, if soil- and compound-specific E_a values have been measured, the measured values should be used to assess environmental exposure.

Table 3. Distributions of median values of activation energies (E_a) for each pesticide in laboratory experiments

	Kinetic model	Dissipation rate constant	Distribution check
			Distribution	N	Ea (kJ mol−1)h)			Shapiro–Wilk’s test (p)
					Avg.	Stdi)	Median
TEa)	SFOc)	k	Normality	23	58.5	24.4	61.1	0.231
			Lognormality	23	50.4	0.56	61.1	0.001
	DFOPd)	k1e)	Normality	20g)	104.9	58.5	97.4	0.124
			Lognormality	22	86.5	0.74	94.2	0.479
		k2f)	Normality	15g)	48.5	17.7	44.4	0.091
			Lognormality	16	48.3	0.45	45.0	0.599
WEb)	DFOPd)	k1e)	Normality	26	64.2	17.2	62.8	0.434
			Lognormality	24g)	63.6	0.26	63.2	0.079
		k2f)	Normality	18g)	45.1	12.3	45.2	0.560
			Lognormality	21	48.0	0.48	48.3	0.449

^a) Total-extractable pesticides in soil. ^b) Water-extractable pesticides in soil. ^c) Single first-order model. ^d) Double first-order in parallel model. ^e) Dissipation rate constant of the first compartment. ^f) Dissipation rate constant of the second compartment. ^g) Number of data points except statistical outliers found with Smirnov–Grubbs test at the 5% level of significance. ^h) E_a values were calculated from two data points or more. ⁱ⁾ Standard deviation. In case of lognormality, the values mean Std (ln E_a).

Table 4. Comparison of activation energies (E_a, kJ mol⁻¹) among test soils

Soils	TRa)									WEb)
	SFOc)			DFOPd)						DFOPd)
	k			k1e)			k2f)			k1e)			k2f)
	N	Avg.g)	Median	N	Avg.g)	Median	N	Avg.g)	Median	N	Avg.g)	Median	N	Avg.g)	Median
LS1	16	62.8 A	68.3	NDh)			NDh)			NDh)			NDh)
LS2	26	61.5 A	62.7	17	113.6 A	101.3	18	44.5 A	47.8	20	93.8 B	91.6	16	81.9 A	53.0
LS3	20	62.4 A	61.3	15	151.9 A	158.5	7	51.9 A	43.8	24	52.9 A	51.9	18	41.4 A	42.9
LS4	18	35.2 A	28.5	11	112.3 A	53.7	11	56.0 A	46.3	25	72.7 AB	68.8	24	57.3 A	46.7
LS5	18	55.4 A	51.3	14	109.2 A	71.6	10	75.7 A	54.9	17	68.0 A	62.0	17	72.3 A	71.3

^a) Total-extractable pesticides in soil. ^b) Water-extractable pesticides in soil. ^c) Single first-order model. ^d) Double first-order in parallel model. ^e) Dissipation rate of the first compartment. ^f) Dissipation rate of the second compartment. ^g) Values with the same letter are not significantly different at p<0.05, according to ANOVA followed by Tukey’s multiple comparison test. ^h) No data.

For the E_a values derived from the DFOP model, the individual E_a values calculated from the k₁ and k₂ values at two or more temperatures were statistically described by a lognormal distribution if statistical outliers were excluded from the data set, and the median E_a values of the distributions for k₁ and k₂ were 99.7 and 46.8 kJ mol⁻¹, respectively (Supplemental Table S7). Additionally, the distributions of the median E_a values of each pesticide were also lognormal and the median E_a values of the distributions for k₁ and k₂ were 94.2 and 45.0 kJ mol⁻¹, respectively (Table 3).

3. Distribution of E_a values derived from the dissipation of water-extractable pesticides in laboratory experiments

The statistical distributions of E_a values calculated from k₁ and k₂ of the DFOP model for the C_WE were investigated in the same way as for the C_TE (Table 3 and Supplemental Table S7). The number of individual E_a values calculated from k₁ and k₂ for the C_WE was larger than that for the C_TE (Supplemental Table S7). This is likely because the dissipation of the C_WE was better fitted by the DFOP model than was the dissipation of the C_TE. As described in Section 2 above, the dissipation of the C_TE tended to be fitted not by the DFOP model but by the SFO model. Therefore, the number of E_a values calculated using the DFOP model for the C_TE was smaller than that for the C_WE.

On the whole, the distributions of E_a values except for statistical outliers were normal or lognormal, and the median E_a values of the distributions were very similar regardless of the methods of calculating E_a (Table 3 and Supplemental Table S7). The normal distributions, whose p-values for Shapiro–Wilk’s test were higher than those of lognormal distributions, of the median E_a values of each pesticide for k₁ and k₂ had median values of 62.8 and 45.2 kJ mol⁻¹, respectively (Table 3).

Although differences in the median E_a values for k₂ of the C_TE and the C_WE were not observed, the median E_a value for k₁ of the C_TE was greater than that of the C_WE (Table 3). One reason for this was that the dissipation of the C_WE could be described by the DFOP model regardless of temperature, whereas the DFOP model provided a better fit for the dissipation data of the C_TE at a high rather than a low temperature (Figs. 1 and 2 ). That is to say, the slope on the earlier part (the first compartment) of the dissipation curve of the C_TE at a high temperature was considerably greater than that at a low temperature; thus, the slope of Arrhenius fits (the E_a value) was large. Actually, the ratio of k₁ at 35°C to that at 10°C for the C_TE was higher than that for the C_WE (Fig. 1). In addition, the median E_a values for k₂ were lower than those for k₁ for both the C_TE and the C_WE (Table 3). This result implies that the effect of temperature on the dissipation of pesticide in the second compartment with a slower dissipation rate was small. In other words, the pesticides in the second compartment, which sorb to soil particles and have low bioavailability and biodegradability, were insensitive to temperature as compared with those in the first compartment.

There were no remarkable differences in the E_a values for both k₁ and k₂ among test soils for the C_WE, and the specificity of the E_a distribution for Andisols was not observed in the same way as of the E_a values derived from k of the SFO model for the C_TE (Table 4). Focusing on the difference in the E_a values among test pesticides, the median E_a values of each pesticide for both k₁ and k₂ exhibited high variability. However, the reason for this could not be explained by the differences in the chemical classes and physico-chemical properties of pesticides (Supplemental Table S8).

4. Field experiments

The dissipation of pesticides during the autumn–winter period was estimated using the measured concentration on day 0 of the autumn–winter experiment, the dissipation rate constants for spring–summer experiment, and the E_a values from the laboratory experiment. The median of the distribution, which had a relatively high Shapiro–Wilk’s test p-value, of the median E_a values for each pesticide was used for the estimation using Eq. 6. The corresponding medians for k, k₁ and k₂ of the C_TE were 61.1, 94.2 and 45.0 kJ mol⁻¹, respectively; those for the k₁ and k₂ of the C_WE were 62.8 and 45.2 kJ mol⁻¹, respectively (Table 3).

When the dissipation of pesticides was estimated using the DFOP model, not only k₁ and k₂ values but also f values were also necessary. Therefore, the variability of f values at three temperatures was investigated. Overall, although the f values tended to increase with increasing temperature (Supplemental Tables S5 and S6), the geometric means of the coefficients of variation calculated from the f values at three temperatures for the C_TE and the C_WE were 40.0 and 26.0%, respectively; thus, f values exhibited relatively low variability. The variability of f values differed depending on the combinations of pesticides and soils. However, the variability of f values at three temperatures could not be explained by the difference in physical and chemical properties of pesticides or soils. Therefore, temperature correction of f values was not conducted for estimating the dissipation of pesticides by the DFOP model.

The comparison between the measured and estimated values for nine pesticides, except fenobucarb, whose measured values were below the limit of quantification, in the autumn–winter field experiment, is shown in Fig. 3. When temperature correction of the dissipation rate constants was not performed (Case 1), the estimated values tended to underestimate the measured values. The ratios of the measured values to the values estimated by the SFO model for the C_TE ranged from 0.7–7.4. In particular, the measured values of thiacloprid, fosthiazate, procymidone and tolclofos-methyl, which had relatively low DT₅₀ values (Supplemental Table S9), exceeded the estimated values by more than threefold. However, values estimated using temperature-corrected dissipation rate constants according to Eq. 6 (Case 2) were roughly the same as the measured values (Fig. 3), and the differences between measured and estimated values were smaller than those for Case 1. The ratios of the measured values to the values estimated using the SFO models for the C_TE ranged from 0.4–1.0.

Fig. 3. Comparison between the measured and estimated concentrations in field soil on days 28 and 64. Cases 1 and 2 indicate the estimated results without and with temperature correction, respectively. SFO and DFOP indicate a single first-order model and a double first-order in parallel model, respectively. The solid line shows the 1 : 1 line, and the dashed lines show the 1 : 2 and 2 : 1 lines.

The estimation accuracy of the SFO model for the C_TE was almost the same as that of the DFOP model for both the C_TE and the C_WE (Fig. 3). The ratios of the measured values to the values estimated using the DFOP models for the C_TE and the C_WE in Case 1 ranged from 0.7–4.4 and 0.7–10.4, respectively, whereas the corresponding ratios in Case 2 were 0.4–1.0 and 0.4–1.2. Thus, the estimated values in Case 2 showed good agreement with the measured values. This result suggests that temperature-dependent biphasic dissipation can be estimated using each different E_a value corresponding to k₁ and k₂.

In conclusion, the median of the distribution of E_a values for first-order dissipation derived from Japanese soils including Andisols was almost the same as that derived from European soils as reported by the EFSA.¹²⁾ The variability of E_a values for k₁ and k₂ calculated from the biphasic DFOP model could also be expressed by normal or lognormal distributions, and the median E_a of the distribution for k₁ was higher than that for k₂. Furthermore, the biphasic dissipation of water-extractable pesticides in a field soil, which was important for evaluating pesticide residue in succeeding crops, could be estimated considering temperature fluctuations, i.e., seasonal differences, by using each median E_a value corresponding to k₁ and k₂. Further verification experiments using other pesticides and fields are required to establish such an estimation method for biphasic dissipation.

Acknowledgments

We thank Shintaro Kanbayashi (Waseda University, Tokyo, Japan) for his help with the laboratory experiments. The test soils were kindly supplied by Aichi Agricultural Research Center and Tochigi Prefectural Agricultural Experiment Station. This research was supported by the Environmental Research and Technology Development Fund (5-1302) of the Ministry of the Environment, Japan.

Electronic supplementary materials

The online version of this article contains supplementary materials (Supplemental Tables S1–S9), which are available at http://www.jstage.jst.go.jp/browse/jpestics/.

References

• 1) S. Hori: Foods Food Ingredients J. Jpn. 212, 496–508 (2007) (in Japanese).
• 2) M. Tanigawa: J. Pestic. Sci. 35, 537–541 (2010) (in Japanese).
• 3) Y. Motoki, T. Iwafune, N. Seike, T. Otani and Y. Akiyama: J. Pestic. Sci. 40, 175–183 (2015).
• 4) Y. Motoki, T. Iwafune, N. Seike, K. Inao and T. Otani: J. Agric. Food Chem. 64, 4478–4486 (2016).
• 5) M. Kah, S. Beulke and C. D. Brown: J. Agric. Food Chem. 55, 4487–4492 (2007).
• 6) A. Ghafoor, J. Moeys, J. Stenström, G. Tranter and N. J. Jarvis: Environ. Sci. Technol. 45, 6411–6419 (2011).
• 7) J. Kobayashi and K. Ikeda: J. Climatol. Weather Forecasting 5, 1000206 (2017).
• 8) A. Walker, Y. H. Moon and S. J. Welch: Pestic. Sci. 35, 109–116 (1992).
• 9) A. M. Cupples, G. K. Sims, R. P. Hultgren and S. E. Hart: J. Environ. Qual. 29, 786–794 (2000).
• 10) S. Taylor-Lovell, G. K. Sims and L. M. Wax: J. Agric. Food Chem. 50, 5626–5633 (2002).
• 11) L. Alletto, Y. Coquet, P. Benoit and V. Bergheaud: Chemosphere 64, 1053–1061 (2006).
• 12) Scientific Opinion of the Panel on Plant Protection Products and their Residues (PPR-Panel) : EFSA J. 622, 1–32 (2007).
• 13) I. Purnama, F. Malhat, P. Jaikaew, H. Watanabe, S. Noegrohati, B. Rusdiarso and M. T. Ahmed: Toxicol. Environ. Chem. 96, 1141–1152 (2014).
• 14) A. J. Cessna, J. D. Knight, D. Ngombe and T. M. Wolf: Can. J. Soil Sci. 97, 717–731 (2017).
• 15) J. M. Marín-Benito, M. J. Carpio, M. J. Sánchez-Martín and M. S. Rodríguez-Cruz: Sci. Total Environ. 653, 1301–1310 (2019).
• 16) U.S. Department of Agriculture: “Soil Taxonomy, A Basic System of Soil Classification for Making and Interpreting Soil Surveys,” 2nd Ed., Agricultural Handbook 436, U.S. Government Printing Office, Washington, DC, 1999.
• 17) H. Obara: Pedologist 44, 134–142 (2000) (in Japanese).
• 18) FOCUS: “Guidance Document on Estimating Persistence and Degradation Kinetics from Environmental Fate Studies on Pesticides in EU Registration,” Report of the FOCUS Work Group on Degradation Kinetics, EC Document Reference Sanco/10058/2005 version 2.0, 1–434 (2006).
• 19) http://sitem.herts.ac.uk/aeru/iupac/ (Accessed 10 June, 2019)
• 20) The 2011 Pesticide Handbook Editorial Committee: “The 2011 Pesticide Handbook,” Japan Plant Protection Association, Tokyo, 2011 (in Japanese).
• 21) Y. Motoki, T. Iwafune, N. Seike, T. Otani and M. Asano: J. Pestic. Sci. 39, 105–114 (2014).
• 22) Cultivated Soil Classification Committee: Misc. Publ. Natl. Inst. Agro-Environ. Sci. 17, 31–55 (1995) (in Japanese).
• 23) H. Murano, Y. Takata and T. Isoi: Soil Sci. Plant Nutr. 61, 688–697 (2015).