**4.1 Sorption-desorption study**

**4.1.1 Kinetic study** 

Sorption-desorption kinetic of IMI on the selected Croatian soils were studied using the initial IMI concentration of 10 mg/L in order to estimate the time needed to achieve the sorption and desorption equilibrium. The results for the case of Istria II soil are presented in Figure 5. Similar trends were observed in all tested soils. Equilibrium time for sorption process was reached at 48h, while for desorption process equilibration was achieved within 144h.

Fig. 5. Sorption-desorption kinetic of IMI on Istria II soil sample. The initial concentration of IMI was 10 mg/L.

## **4.1.2 Sorption equilibrium study**

Figure 6a shows the sorption isotherms for IMI in the tested soils. All of the sorption isotherms are of L-type (Giles; et al., 1960) showing a convex initial curvature. This isotherm type indicates a decrease in specific sorption sites when the concentration of insecticide increases; however, in the case of IMI the curves did not reach the plateau of saturation.

In Table 2 the sorbed amount of IMI during the sorption processes in the tested soils is presented. The percentage of IMI sorbed on the Istria II soil was 35.96, 36.00, 37.13 and 37.88% at concentrations of 10, 5, 2.5 and 1 mg/L, respectively, whereas in the Istria I, Krk II and Krk I soil, the percentage sorbed ranged from: 30.41-33.01, 18.45-23.68 and 16.50-19.75% at the same respective concentrations. The percentage IMI sorbed was higher in Istria II soil than in the other soils.

Sorption-desorption processes have a significant effect on pesticide persistence, pesticide concentration level in the soil solution and on the transport of pesticides from agricultural field to other environmental compartments (Arias-Estevez et al., 2008). An understanding of the variability of pesticide persistence and sorption-desorption processes within and among regions can improve the accuracy of estimates of the behavior and fate of pesticide in the soil and provide an additional support in design of intervention strategies against

Sorption-desorption kinetic of IMI on the selected Croatian soils were studied using the initial IMI concentration of 10 mg/L in order to estimate the time needed to achieve the sorption and desorption equilibrium. The results for the case of Istria II soil are presented in Figure 5. Similar trends were observed in all tested soils. Equilibrium time for sorption process was reached at 48h, while for desorption process equilibration was achieved within

> **Time (h)** 0 20 40 60 80 100 120 140 160

Fig. 5. Sorption-desorption kinetic of IMI on Istria II soil sample. The initial concentration of

Figure 6a shows the sorption isotherms for IMI in the tested soils. All of the sorption isotherms are of L-type (Giles; et al., 1960) showing a convex initial curvature. This isotherm type indicates a decrease in specific sorption sites when the concentration of insecticide increases; however, in the case of IMI the curves did not reach the plateau of

In Table 2 the sorbed amount of IMI during the sorption processes in the tested soils is presented. The percentage of IMI sorbed on the Istria II soil was 35.96, 36.00, 37.13 and 37.88% at concentrations of 10, 5, 2.5 and 1 mg/L, respectively, whereas in the Istria I, Krk II and Krk I soil, the percentage sorbed ranged from: 30.41-33.01, 18.45-23.68 and 16.50-19.75% at the same respective concentrations. The percentage IMI sorbed was higher in Istria II soil

Sorption process Desorption process

**4. Results and discussion** 

groundwater pollution.

**4.1.1 Kinetic study** 

IMI was 10 mg/L.

saturation.

than in the other soils.

**4.1.2 Sorption equilibrium study** 

144h.

**4.1 Sorption-desorption study** 

**Imidacloprid sorbed (mg/kg)**

0

5

10

15

20

Fig. 6. a) Sorption and b) desorption isotherms of IMI in the tested soils represented by the Freundlich model. Values are means ± standard deviations. Symbols represent the experimental data, while lines represent the theoretical curves described by the Freundlich model.


*<sup>a</sup>* sorbed amount of IMI after 48 h of sorption reaction time; *b* desorbed amount of IMI after 144 h of desorption reaction time.

Table 2. The sorbed and desorbed amount of IMI in the tested soils in relation to the initial concentration.

All sorption data fit the Freundlich equation (*R2* > 0.966) and Table 3 summarizes the sorption capacity (*KFsor*) and intensity (*1/nsor*) values. The *KFsor* values obtained from the Freundlich

Behavior and Fate of Imidacloprid in Croatian Olive Orchard Soils Under Laboratory Conditions 503

 pH clay CEC OC pH 1.00 -0.21 -0.24 -0.15 clay -0.21 1.00 **0.85 0.82**  CEC -0.24 **0.85** 1.00 **0.79**  OC -0.15 **0.82 0.79** 1.00 *K*Fsor -0.24 **0.79 0.82 0.85**  DT50 0.21 **-0.67 -0.76 -0.64** 

OC – organic carbon content; CEC – cation exchange capacity;

0.049), as well as between soils Krk I and Krk II (*p* = 0.050).

Table 4. Kendall-Tau correlation test for soil properties and sorption and

that the OC content predominantly influenced IMI sorption on the tested soils.

degradation parameters of IMI, n = 12 (Bold typeface indicates significant correlations

Nonparametric regression showed that the amount of OC in the soil, the CEC and clay amount affected the sorption processes, but multiple linear regression equations suggested

Given the difference between tested soils in the studied regions, statistically significant differences in soil sorption coefficients, *KFsor* were found among the studied regions, i.e. Krk and Istria region (*p* = 0.004). In addition, results of sorption study within the regions showed a statistically significant diffrence in *KFsor* values between the soils Istria I and Istria II (*p* =

The OC partition coefficient, *KOCsor* (Equation 2) usually illustrate the hydrophobicity of the pesticide and may be used to estimate or predict the migration and behavior of an organic pesticide in the environment (Xue et al., 2006). Thus, defined coefficient, normalized to the proportion of OC, should have a constant value for each pesticide molecule and the same values in the soils with different content of organic matter. However, variability in *KOCsor* values for the soils of different type and characteristics, and even for the soils with the same content of organic matter, indicated that not only organic matter content, but also its structure, aromaticity and polarity, affected the distribution of pesticide molecules in the soil/water system (Schwarzenbach et al., 2002). The values of *KOCsor* coefficient for IMI in the tested soils varied from 172 to 305 L/kg (Table 3), and they are consistent with reported *KOCsor* values (Cox et al., 1998b; Krohn & Hellpointner, 2002), especially for soils with similar textural characteristics. Estimated values of our study prove that, according to the McCall classification for the mobility of pesticides (McCall et al., 1980), IMI can be categorized as having a medium mobility (*KOCsor* 150 - 500 L/kg) (Comfort et al*.*, 1994), showing less tendency to be sorbed by the examined soils. Therefore, these *KOCsor* values, together with reported *KOW* values (3.7) (Krohn & Hellpointner, 2002) and a great water solubility (0.51g/L) (Tomlin, 2001) suggest a potential of IMI to leach to groundwater. However, the results of field studies have showed the lack of leaching for IMI, which could be due to a larger sorption potential at a lower concentration compared to higher concentration range (Kamble & Saran, 2005), or as a result of an increase in the sorption of IMI with time in the

*K*Fsor– Freundlich coefficient of sorption;

with *p < 0.05*).

soil (Oi, 1999).

DT50- time for 50% of the initial residue to degrade.

isotherm model were 1.28, 1.53, 2.60 and 3.28 (mg/kg)/(mg/L)1/n for Krk II, Krk I, Istria I and Istria II soil, respectively. The highest *KFsor* value for IMI, is indicative of the strongest retention by the soil matrix. A primary consequence of strong retention of IMI is its limited mobility in the Istria II soil profile and thus lower risk of ground water contamination.


Table 3. The Freundlich sorption parameters, values of organic carbon/partition coefficient (*KOC*) and Gibbs free energy *(*Δ*G0*) for IMI in the tested soils.

In our study, the behavior of sorption was nonlinear. This is based on the best-fit estimated parameter *1/n* under the value of one (*1/n* < 1) (Table 3). In fact, for the Krk soils, *1/n* values were lower (0.894 and 0.907) than for the Istria soils (0.937 and 0.945). Nonlinear isotherm behavior is a measure of the extent of heterogeneity of retention reactions and the presence of sites having variable affinities for sorption of IMI by the soil matrix surface. Based on the estimated *1/n* values, an increased amount sorbed by soil is anticipated in all soils at low IMI concentration. A higher initial insecticide solution concentration led to the change of the affinity between insecticide molecules and soil, probably due to decreased accessibility to the free sorption sites (Kamble & Saran, 2005).

In the present study, *KFsor* values varied between the tested soils, indicating that the differences between the soils strongly influence the sorption. Several studies have shown that soil properties, particularly the soil organic matter and clay content play a key role in the performance of applied pesticides (Cox et al., 1998a; Fernandez-Bayo et al., 2007; Kamble & Saran, 2005; Liu et al., 2006; ten Hulscher & Cornelissen, 1996). In order to elucidate the factors that affect sorption of IMI on the soils, the *KFsor* values were correlated with the OC and clay content, CEC, and pH using a nonparametric Kendall-Tau correlation test. Correlation analysis between sorption coefficients (*KFsor*) and selected soil properties showed a significant correlation between *KFsor* and the OC content, CEC and clay, but the correlation between *KFsor* and pH was not significant (Table 4). It has been postulated that soil pH has an influence on pesticide sorption only when the *pKa* or *pKb* are within approximately two units of the soil pH (Farenhorst, 2006). As the pH value in examined soils ranged from 4.76 to 7.12, which was significantly below the *pKa* value of IMI (pKa =11.2), the effect of soil pH was not noticeable. Thus, these results suggest that the OC content and CEC had the major influence on the IMI sorption in these soils.

In addition to nonparametric test, multiple linear regression was used, which simultaneously compares various soil properties and sorption coefficients (*KFsor*), and leads to a linear predictive model for *KFsor* value (Golfinopoulos & Arhonditsis, 2002). These models may be useful for identifying areas (homogeneous soil types) where surface water resources could be threatened by pesticide contamination and for identification of pesticides which are more easily leached through the soil profile (Golfinopoulos & Arhonditsis, 2002). Multiple liner regression resulted in the following correlation:

$$\mathrm{K\_F^{\mathrm{sor}}} = 0.4177 \,\mathrm{OC} + 0.0037 \,\mathrm{C.EC} + 0.0398 \,\mathrm{day} - 0.0866 \,\mathrm{pH} + 1.009 \left(\mathrm{R}^2 = 0.989\right) \tag{12}$$


OC – organic carbon content; CEC – cation exchange capacity;

*K*Fsor– Freundlich coefficient of sorption;

502 Pesticides in the Modern World - Risks and Benefits

isotherm model were 1.28, 1.53, 2.60 and 3.28 (mg/kg)/(mg/L)1/n for Krk II, Krk I, Istria I and Istria II soil, respectively. The highest *KFsor* value for IMI, is indicative of the strongest retention by the soil matrix. A primary consequence of strong retention of IMI is its limited mobility in

Krk I 1.53 (± 0.06) 0.894 (± 0.019) 0.996 278.18 -13.72 Krk II (1.28 ± 0.06) 0.907 (± 0.022) 0.998 304.76 -13.94 Istria I (2.60 ± 0.10) 0.945 (± 0.021) 0.998 200.00 -12.91 Istria II (3.28 ± 0.09) 0.937 (± 0.016) 0.997 171.73 -12.54 Table 3. The Freundlich sorption parameters, values of organic carbon/partition coefficient

*G0*) for IMI in the tested soils. In our study, the behavior of sorption was nonlinear. This is based on the best-fit estimated parameter *1/n* under the value of one (*1/n* < 1) (Table 3). In fact, for the Krk soils, *1/n* values were lower (0.894 and 0.907) than for the Istria soils (0.937 and 0.945). Nonlinear isotherm behavior is a measure of the extent of heterogeneity of retention reactions and the presence of sites having variable affinities for sorption of IMI by the soil matrix surface. Based on the estimated *1/n* values, an increased amount sorbed by soil is anticipated in all soils at low IMI concentration. A higher initial insecticide solution concentration led to the change of the affinity between insecticide molecules and soil, probably due to decreased accessibility to

In the present study, *KFsor* values varied between the tested soils, indicating that the differences between the soils strongly influence the sorption. Several studies have shown that soil properties, particularly the soil organic matter and clay content play a key role in the performance of applied pesticides (Cox et al., 1998a; Fernandez-Bayo et al., 2007; Kamble & Saran, 2005; Liu et al., 2006; ten Hulscher & Cornelissen, 1996). In order to elucidate the factors that affect sorption of IMI on the soils, the *KFsor* values were correlated with the OC and clay content, CEC, and pH using a nonparametric Kendall-Tau correlation test. Correlation analysis between sorption coefficients (*KFsor*) and selected soil properties showed a significant correlation between *KFsor* and the OC content, CEC and clay, but the correlation between *KFsor* and pH was not significant (Table 4). It has been postulated that soil pH has an influence on pesticide sorption only when the *pKa* or *pKb* are within approximately two units of the soil pH (Farenhorst, 2006). As the pH value in examined soils ranged from 4.76 to 7.12, which was significantly below the *pKa* value of IMI (pKa =11.2), the effect of soil pH was not noticeable. Thus, these results suggest that the OC content and CEC had the major

In addition to nonparametric test, multiple linear regression was used, which simultaneously compares various soil properties and sorption coefficients (*KFsor*), and leads to a linear predictive model for *KFsor* value (Golfinopoulos & Arhonditsis, 2002). These models may be useful for identifying areas (homogeneous soil types) where surface water resources could be threatened by pesticide contamination and for identification of pesticides which are more easily leached through the soil profile (Golfinopoulos & Arhonditsis, 2002).

( ) *sor <sup>2</sup> K = 0.4177 OC + 0.0037 CEC + 0.0398 clay – 0.0866 F pH + 1.009 R = 0.989* (12)

1/nsor R2 KOCsor

(L/kg)

ΔG (kJ/mol)

the Istria II soil profile and thus lower risk of ground water contamination.

Soil KFsor

(*KOC*) and Gibbs free energy *(*

(mg/kg)/(mg/L)1/n

the free sorption sites (Kamble & Saran, 2005).

influence on the IMI sorption in these soils.

Multiple liner regression resulted in the following correlation:

Δ

DT50- time for 50% of the initial residue to degrade.

Table 4. Kendall-Tau correlation test for soil properties and sorption and degradation parameters of IMI, n = 12 (Bold typeface indicates significant correlations with *p < 0.05*).

Nonparametric regression showed that the amount of OC in the soil, the CEC and clay amount affected the sorption processes, but multiple linear regression equations suggested that the OC content predominantly influenced IMI sorption on the tested soils.

Given the difference between tested soils in the studied regions, statistically significant differences in soil sorption coefficients, *KFsor* were found among the studied regions, i.e. Krk and Istria region (*p* = 0.004). In addition, results of sorption study within the regions showed a statistically significant diffrence in *KFsor* values between the soils Istria I and Istria II (*p* = 0.049), as well as between soils Krk I and Krk II (*p* = 0.050).

The OC partition coefficient, *KOCsor* (Equation 2) usually illustrate the hydrophobicity of the pesticide and may be used to estimate or predict the migration and behavior of an organic pesticide in the environment (Xue et al., 2006). Thus, defined coefficient, normalized to the proportion of OC, should have a constant value for each pesticide molecule and the same values in the soils with different content of organic matter. However, variability in *KOCsor* values for the soils of different type and characteristics, and even for the soils with the same content of organic matter, indicated that not only organic matter content, but also its structure, aromaticity and polarity, affected the distribution of pesticide molecules in the soil/water system (Schwarzenbach et al., 2002). The values of *KOCsor* coefficient for IMI in the tested soils varied from 172 to 305 L/kg (Table 3), and they are consistent with reported *KOCsor* values (Cox et al., 1998b; Krohn & Hellpointner, 2002), especially for soils with similar textural characteristics. Estimated values of our study prove that, according to the McCall classification for the mobility of pesticides (McCall et al., 1980), IMI can be categorized as having a medium mobility (*KOCsor* 150 - 500 L/kg) (Comfort et al*.*, 1994), showing less tendency to be sorbed by the examined soils. Therefore, these *KOCsor* values, together with reported *KOW* values (3.7) (Krohn & Hellpointner, 2002) and a great water solubility (0.51g/L) (Tomlin, 2001) suggest a potential of IMI to leach to groundwater. However, the results of field studies have showed the lack of leaching for IMI, which could be due to a larger sorption potential at a lower concentration compared to higher concentration range (Kamble & Saran, 2005), or as a result of an increase in the sorption of IMI with time in the soil (Oi, 1999).

Behavior and Fate of Imidacloprid in Croatian Olive Orchard Soils Under Laboratory Conditions 505

affinity for the IMI. In our study, *1/n* constants ranged from 0.654 to 0.836 with deviations from the linear function ranging from 16.4% (Krk II soil) to 34.6% (Istria II soil). Fernandez-Bayo et al. (2007) made similar observations for IMI in Spain soils. This could be explained by a possible hysteresis effect taking place during desorption, involving various forces that caused the amount of IMI retained to be higher after desorption than after sorption at the unit equilibrium concentration. Hysteresis is manifested by an increase in the difference between the sorption and desorption isotherm slopes. Conceptually, the lack of similarity between sorption and desorption due to the hysteresis is likely a results of binding to organic matter and clay particles. Clay fraction is of great importance because it can enter into interactions with natural organic matter in the soil and it can control its structural configuration (Gunasekara & Xing, 2003). Particularly, in the interaction of organic matter with the clay fraction crystal-amorphous complexes are formed which can increase the nonlinearity of the sorption isotherm. Several studies have illustrated hysteretic behavior of IMI (Fernandez-Bayo et al., 2007;

Krk I 4.66 (± 0.16) 0.786 (± 0.049) 0.988 847.27 0.879 Krk II 3.65 (± 0.05) 0.836 (± 0.023) 0.982 869.05 0.922 Istria I 12.41 (±0.39) 0.749 (± 0.041) 0.997 954.61 0.793 Istria II 16.05 (±0.54) 0.654 (± 0.036) 0.989 840.31 0.698

Table 5. The Freundlich desorption parameters and hysteresis index (H) for IMI in the tested

For the Krk II soil, the Freundlich *KFsor* was 1.28 (mg/kg)/(mg/L)1/n and the *KFdes* was 3.65 (mg/kg)/(mg/L)1/n, while the *KOCsor* and *KOCdes* were 304.76 and 869.05 L/kg. These findings indicated that IMI was weakly sorbed, but slightly held by the soil. In contrast, the *KFsor* and *KFdes* values for the Istria II soil were 3.28 and 16.05 (mg/kg)/(mg/L)1/n, while the *KOCsor* and *KOCdes* were 171.73 and 840.31 L/kg. The IMI was sorbed firmer by the Istria II soil and retained, though surprisingly the low *KOCsor* values of Istria II soil suggested that OC played less of role in sorption than with the lower OC content in Krk II soil. The difference may lie in the different pH values for the soils. The pH for the Krk II soil was 6.88 (nearly neutral) while the Istria II soil pH was 4.76 (acidic). Ping et al. (2010) found that IMI was sorbed more strongly at pH 4.5 than at pH 7.5. The effect of pH is probably due to the increased polarity of the humic material and the electrostatic interaction of the pesticide with soil particles at

In order to estimate the discrepancies between sorption and desorption isotherms hysteresis coefficient *H* was calculated, and its values for the tested soils are presented in Table 5. When the value of *H* is lower, sorption-desorption hysteresis is more pronounced. We can see that the highest hystersis effect (the lowest *H*) was observed in the Istria II soil.

1/ndes R2 KOCdes

(L/kg)

H

Papiernik et al., 2006).

soils.

higher pH.

Soil KFdes

(mg/kg)/ (mg/L)1/n

**4.1.4 Sorption-desorption hysteresis** 

At equilibrium, the pesticide distribution between the solid and aqueous phases is ultimately governed by the sorption Gibbs free energy (Δ*G°*). The change in the Δ*G°* as a result of sorption process, was calculated from the thermodynamic relationship:

$$
\Delta G^{\text{O}} = -RT\ln K\_{\text{OC}} \tag{13}
$$

where, Δ*G°* is the free energy change (kJ/mol), *T* is the absolute temperature (K), *R* is the universal gas constant (8.314 J/molK). Δ*G°* values of sorption processes ranged from -13.94 to -12.54 kJ/mol and are listed in Table 3. The Δ*G°* values obtained in the present study indicate that the sorption capacity of the soils would be in the order of Istria II soil > Istria I soil > Krk I soil > Krk II soil. The greater the absolute magnitude of Δ*G°* value, the greater is the extent to which the sorption reaction may take place. A small negative value of Δ*G°*  indicated the exothermic nature of the reaction and a spontaneous process. In such cases, it can be inferred that the sorption of IMI takes place via physical processes involving weak attractive forces (ten Hulscher & Cornelissen, 1996), primarily by dissolution-like partition of IMI into soil organic matter (Sheng et al., 2001).

#### **4.1.3 Desorption equilibrium study**

Plot of the desorption isotherms for IMI are shown in Figure 6b. It can be seen that the slopes of the desorption isotherms are clearly different from those of the sorption isotherms. The characteristic steep slopes of all isotherms are observed at the low equilibrium concentrations of IMI corresponding to the low initial IMI content in solutions. With increase of the IMI concentrations in solutions the curve slopes become less steep.

The desorption data are also shown in Table 2, where desorbed amount was expressed as a percentage of the total amount sorbed. In all the tested soils significant differences of the amount desorbed between different concentrations and between the tested soils were observed. As the initial IMI concentration increased from 1 to 10 mg/L, the desorbed amount, as a percentage of the total sorbed, increased from 12.70 to 27.71% for the Istrian soils and from 41.05 to 58.75% for the Krk soils. The highest percentage of desorption was achieved for the Krk II soil, where actual amounts of recovery ranged from 43.98 to 58.75%of that sorbed by soil. This suggests that half of the amount sorbed was retained by the Krk II soil regardless of initial concentration. The lowest measured recovery of the amount of the desorbed IMI was observed for the Istria II soil. The higher release for Krk II than Istria II soil is likely due to difference in OC content. In a desorption study of IMI on a Hungarian soil, (Nemeth-Konda et al., 2002) found that the amount of IMI desorbed following three desorption steps (average the initial concentrations) was 62±15%. In their study, a similar background solution (CaCl2) was used except that sorption was limited 24h.

The Freundlich desorption coefficient values (*KFdes*) for the tested soils were higher than sorption values (*KFsor*), while desorption *1/n* values were lower than the Freundlich sorption equilibrium values (Table 5). *KFdes* value was highest for the Istria II soil (clay soil with 1.91% OC) followed by the Istria I, Krk I and the Krk II soil (sandy loam soil with 0.42 % OC) which exhibited the lowest *KFdes*. A higher *KFdes* value indicated a stronger affinity for the IMI. In our study, *1/n* constants ranged from 0.654 to 0.836 with deviations from the linear function ranging from 16.4% (Krk II soil) to 34.6% (Istria II soil). Fernandez-Bayo et al. (2007) made similar observations for IMI in Spain soils. This could be explained by a possible hysteresis effect taking place during desorption, involving various forces that caused the amount of IMI retained to be higher after desorption than after sorption at the unit equilibrium concentration. Hysteresis is manifested by an increase in the difference between the sorption and desorption isotherm slopes. Conceptually, the lack of similarity between sorption and desorption due to the hysteresis is likely a results of binding to organic matter and clay particles. Clay fraction is of great importance because it can enter into interactions with natural organic matter in the soil and it can control its structural configuration (Gunasekara & Xing, 2003). Particularly, in the interaction of organic matter with the clay fraction crystal-amorphous complexes are formed which can increase the nonlinearity of the sorption isotherm. Several studies have illustrated hysteretic behavior of IMI (Fernandez-Bayo et al., 2007; Papiernik et al., 2006).


Table 5. The Freundlich desorption parameters and hysteresis index (H) for IMI in the tested soils.

For the Krk II soil, the Freundlich *KFsor* was 1.28 (mg/kg)/(mg/L)1/n and the *KFdes* was 3.65 (mg/kg)/(mg/L)1/n, while the *KOCsor* and *KOCdes* were 304.76 and 869.05 L/kg. These findings indicated that IMI was weakly sorbed, but slightly held by the soil. In contrast, the *KFsor* and *KFdes* values for the Istria II soil were 3.28 and 16.05 (mg/kg)/(mg/L)1/n, while the *KOCsor* and *KOCdes* were 171.73 and 840.31 L/kg. The IMI was sorbed firmer by the Istria II soil and retained, though surprisingly the low *KOCsor* values of Istria II soil suggested that OC played less of role in sorption than with the lower OC content in Krk II soil. The difference may lie in the different pH values for the soils. The pH for the Krk II soil was 6.88 (nearly neutral) while the Istria II soil pH was 4.76 (acidic). Ping et al. (2010) found that IMI was sorbed more strongly at pH 4.5 than at pH 7.5. The effect of pH is probably due to the increased polarity of the humic material and the electrostatic interaction of the pesticide with soil particles at higher pH.

#### **4.1.4 Sorption-desorption hysteresis**

504 Pesticides in the Modern World - Risks and Benefits

At equilibrium, the pesticide distribution between the solid and aqueous phases is ultimately governed by the sorption Gibbs free energy (Δ*G°*). The change in the Δ*G°* as a

where, Δ*G°* is the free energy change (kJ/mol), *T* is the absolute temperature (K), *R* is the universal gas constant (8.314 J/molK). Δ*G°* values of sorption processes ranged from -13.94 to -12.54 kJ/mol and are listed in Table 3. The Δ*G°* values obtained in the present study indicate that the sorption capacity of the soils would be in the order of Istria II soil > Istria I soil > Krk I soil > Krk II soil. The greater the absolute magnitude of Δ*G°* value, the greater is the extent to which the sorption reaction may take place. A small negative value of Δ*G°*  indicated the exothermic nature of the reaction and a spontaneous process. In such cases, it can be inferred that the sorption of IMI takes place via physical processes involving weak attractive forces (ten Hulscher & Cornelissen, 1996), primarily by dissolution-like partition

Plot of the desorption isotherms for IMI are shown in Figure 6b. It can be seen that the slopes of the desorption isotherms are clearly different from those of the sorption isotherms. The characteristic steep slopes of all isotherms are observed at the low equilibrium concentrations of IMI corresponding to the low initial IMI content in solutions. With increase of the IMI concentrations in solutions the curve slopes become

The desorption data are also shown in Table 2, where desorbed amount was expressed as a percentage of the total amount sorbed. In all the tested soils significant differences of the amount desorbed between different concentrations and between the tested soils were observed. As the initial IMI concentration increased from 1 to 10 mg/L, the desorbed amount, as a percentage of the total sorbed, increased from 12.70 to 27.71% for the Istrian soils and from 41.05 to 58.75% for the Krk soils. The highest percentage of desorption was achieved for the Krk II soil, where actual amounts of recovery ranged from 43.98 to 58.75%of that sorbed by soil. This suggests that half of the amount sorbed was retained by the Krk II soil regardless of initial concentration. The lowest measured recovery of the amount of the desorbed IMI was observed for the Istria II soil. The higher release for Krk II than Istria II soil is likely due to difference in OC content. In a desorption study of IMI on a Hungarian soil, (Nemeth-Konda et al., 2002) found that the amount of IMI desorbed following three desorption steps (average the initial concentrations) was 62±15%. In their study, a similar background solution (CaCl2) was used except that sorption was limited

The Freundlich desorption coefficient values (*KFdes*) for the tested soils were higher than sorption values (*KFsor*), while desorption *1/n* values were lower than the Freundlich sorption equilibrium values (Table 5). *KFdes* value was highest for the Istria II soil (clay soil with 1.91% OC) followed by the Istria I, Krk I and the Krk II soil (sandy loam soil with 0.42 % OC) which exhibited the lowest *KFdes*. A higher *KFdes* value indicated a stronger

*<sup>0</sup> <sup>Δ</sup>G = - RTlnKOC* (13)

result of sorption process, was calculated from the thermodynamic relationship:

of IMI into soil organic matter (Sheng et al., 2001).

**4.1.3 Desorption equilibrium study** 

less steep.

24h.

In order to estimate the discrepancies between sorption and desorption isotherms hysteresis coefficient *H* was calculated, and its values for the tested soils are presented in Table 5. When the value of *H* is lower, sorption-desorption hysteresis is more pronounced. We can see that the highest hystersis effect (the lowest *H*) was observed in the Istria II soil.

Behavior and Fate of Imidacloprid in Croatian Olive Orchard Soils Under Laboratory Conditions 507

**midacloprid residue (mg/kg)**

**I**

**Imidacloprid residue (mg/kg)**

0.0

Fig. 7. Degradation of IMI in the tested soils at 0.5 mg/kg concentration level. Values are means ± standard deviations. Symbols represent the experimental data, while lines

represent the theoretical curves fitted by the first-order kinetics model or two-compartment

The results from the curve fitting analysis are shown in Figure 7 and 8. In each figure, the measured data are shown together with the curves simulated by the first-order kinetic model or by two-compartment model. IMI degradation in all tested soils at the low concentration level appears to be adequately described by two-compartment model. In contrast, at the high concentration level the experimental data were better described using the first-order kinetic model rather than the two-compartment model, except for Istria II soil, where biphasic kinetic was observed (Figure 8d). In fitting two-compartment model, we assumed that the fast degradation phase occurred from 0 to 15 days after application and that the slow degradation phase occurred thereafter. This was visually determined based on the changes of the slopes of the degradation curves. Capri et al. (2001), who studied the degradation of IMI in Italian soils, found that IMI concentration decreased rapidly in the first 10 days followed by a slower decrease in the total amount recovered. The derived rate constants as well as DT50 and DT90 for each model are shown in Table 6 together with the correlation coefficient for each curve (*R2*) and with root mean square error (RMSE). Both statistical indices (*R2* and RMSE) indicated that the first-order kinetic model better described IMI degradation at the high concentration level in all tested soils, except in Istria II soil, than at the low concentration level (Table 6). At the low concentration level, the more complex two-compartment model generated smaller RMSE.

0.1

0.2

0.3

0.4

0.5

0.6

0.0

0.1

0.2

0.3

0.4

0.5

0.6

**b) Istria I soil**

**d) Istria II soil**

**Time (days after application)** 0 50 100 150 200

> Experimental (mean ± stdv ) Fitted, Two-compartment (*R<sup>2</sup>* = 0.9978 )

**Time (days after application)** 0 50 100 150 200

Experimental (mean ± stdv ) Fitted, Two-compartment (*R<sup>2</sup>* = 0.9907 )

**a) Krk I soil**

**c) Krk II soil**

**Imidacloprid residue (mg/kg)**

0.0

**Imidacloprid residue (mg/kg)**

0.0

model.

0.1

0.2

0.3

0.4

0.5

0.6

0.1

0.2

0.3

0.4

0.5

0.6

**Time (days after application)** 0 50 100 150 200

> Experimental (mean ± stdv ) Fitted, First-order (*R<sup>2</sup>*

= 0.9917)

**Time (days after application)** 0 50 100 150 200

Experimental (mean ± stdv ) Fitted, Two-compartment (*R<sup>2</sup>* = 0.9939 )

In Krk soils, no higher differences between sorption-desorption isotherm slopes were found, and therefore, hysteresis coefficients near to unit (*1/ndes* ≈ *1/nsor*) indicates the high reversibility of IMI sorption by these two soils. Coefficient *H* was lower for Istrian soils than for Krk soils (*1/ndes*<*1/nsor*). This indicates that a significant amount of the sorbed IMI is very difficult to desorb from Istrian soils which may have been caused by a higher amount of OC and clay content in Istria soils, leading to a higher sorption capacity than for Krk soils.

According to the dual model for the sorption of organic pesticides on the soil organic matter, sorption takes place through two sorption mechanisms: the partition and the sorption (Pignatello & Xing, 1996). Soil organic matter has not a uniform continuous phase and is rather represented as a three dimensional matrix, in which the condensed and amorphous phases form separated microenvironment. According to the proposed mechanism, at low concentrations of IMI, the sorption sites in the condensed aromatic area are occupied first, while at higher concentrations of IMI the sorption sites in the amorphous and aliphatic regions start to fill. This effect caused a pronounced hysteresis in the range of lower concentrations, which is consistent with the results obtained for the sorption of IMI. Since sorption area contains a limited number of high-energy sorption sites, molecules of sorbate occupy first these places at low concentrations, meaning that at low concentration the sorption mechanism dominates over the partition (Gunasekara et al., 2003). In addition, de Jonge & Mittelmejer-Hazeleger, 1996) showed that natural organic matter, has high microporosity, with a radius of pores <20 Å, and, therefore, the authors assumed that the observed sorption-desorption hysteresis can be the result of irreversible "trapping" of IMI molecules in the pores of natural organic matter. If we assume that the pore radius is 10 Å, than the calculated pore volume is about 4200 Å3. Since the volume of one IMI molecule is 275 Å3, it is possible that "irreversible entrapment" caused the observed sorption-desorption hysteresis.

#### **4.2 Degradation study**

#### **4.2.1 Persistence of IMI**

Results of degradation studies for IMI at two concentration levels (0.5 and 5 mg/kg) in four Croatian soils at various times are plotted in Figures 7 and 8. A visual examination of degradation pattern for the IMI in all tested soils suggests significant deviation from the first-order kinetic (*R2* range from 0.95 to 0.98). Consequently, alternative twocompartment model was used to describe the observed two-phase kinetic and to derive DT50 and DT90 (50 and 90 % degradation time) values. In our study, therefore, we presented the experimental data as a concentration of IMI degraded from the initial application on soil, and the corresponding values for DT50 and DT90, were graphically estimated from Figures 7 and 8.

Following the treatment of soils at 0.5 and 5 mg/kg concentration level, the average initial concentration varied from 0.43 to 0.49 and 4.71 to 5.43 mg/kg (Figure 7 and 8). In all the tested soils, the residues persisted beyond 180 days at both levels and 4.7-13.8 % loss was recorded on day 7, 12.0-39.9 % on day 30, 32.7-65.2 % on day 90, and 55.0-82.6 % on day 180. The greatest loss of IMI was found in a clay soil with a higher OC content (Istria II soil, 75.4 -82.61 %) and the lowest in a sandy loam soil with a lower OC content (Krk II soil, 55.0- 58.1 %).

In Krk soils, no higher differences between sorption-desorption isotherm slopes were found, and therefore, hysteresis coefficients near to unit (*1/ndes* ≈ *1/nsor*) indicates the high reversibility of IMI sorption by these two soils. Coefficient *H* was lower for Istrian soils than for Krk soils (*1/ndes*<*1/nsor*). This indicates that a significant amount of the sorbed IMI is very difficult to desorb from Istrian soils which may have been caused by a higher amount of OC and clay content in Istria soils, leading to a higher sorption capacity than

According to the dual model for the sorption of organic pesticides on the soil organic matter, sorption takes place through two sorption mechanisms: the partition and the sorption (Pignatello & Xing, 1996). Soil organic matter has not a uniform continuous phase and is rather represented as a three dimensional matrix, in which the condensed and amorphous phases form separated microenvironment. According to the proposed mechanism, at low concentrations of IMI, the sorption sites in the condensed aromatic area are occupied first, while at higher concentrations of IMI the sorption sites in the amorphous and aliphatic regions start to fill. This effect caused a pronounced hysteresis in the range of lower concentrations, which is consistent with the results obtained for the sorption of IMI. Since sorption area contains a limited number of high-energy sorption sites, molecules of sorbate occupy first these places at low concentrations, meaning that at low concentration the sorption mechanism dominates over the partition (Gunasekara et al., 2003). In addition, de Jonge & Mittelmejer-Hazeleger, 1996) showed that natural organic matter, has high microporosity, with a radius of pores <20 Å, and, therefore, the authors assumed that the observed sorption-desorption hysteresis can be the result of irreversible "trapping" of IMI molecules in the pores of natural organic matter. If we assume that the pore radius is 10 Å, than the calculated pore volume is about 4200 Å3. Since the volume of one IMI molecule is 275 Å3, it is possible that "irreversible entrapment" caused the observed sorption-desorption

Results of degradation studies for IMI at two concentration levels (0.5 and 5 mg/kg) in four Croatian soils at various times are plotted in Figures 7 and 8. A visual examination of degradation pattern for the IMI in all tested soils suggests significant deviation from the first-order kinetic (*R2* range from 0.95 to 0.98). Consequently, alternative twocompartment model was used to describe the observed two-phase kinetic and to derive DT50 and DT90 (50 and 90 % degradation time) values. In our study, therefore, we presented the experimental data as a concentration of IMI degraded from the initial application on soil, and the corresponding values for DT50 and DT90, were graphically

Following the treatment of soils at 0.5 and 5 mg/kg concentration level, the average initial concentration varied from 0.43 to 0.49 and 4.71 to 5.43 mg/kg (Figure 7 and 8). In all the tested soils, the residues persisted beyond 180 days at both levels and 4.7-13.8 % loss was recorded on day 7, 12.0-39.9 % on day 30, 32.7-65.2 % on day 90, and 55.0-82.6 % on day 180. The greatest loss of IMI was found in a clay soil with a higher OC content (Istria II soil, 75.4 -82.61 %) and the lowest in a sandy loam soil with a lower OC content (Krk II soil, 55.0-

for Krk soils.

hysteresis.

58.1 %).

**4.2 Degradation study 4.2.1 Persistence of IMI** 

estimated from Figures 7 and 8.

Fig. 7. Degradation of IMI in the tested soils at 0.5 mg/kg concentration level. Values are means ± standard deviations. Symbols represent the experimental data, while lines represent the theoretical curves fitted by the first-order kinetics model or two-compartment model.

The results from the curve fitting analysis are shown in Figure 7 and 8. In each figure, the measured data are shown together with the curves simulated by the first-order kinetic model or by two-compartment model. IMI degradation in all tested soils at the low concentration level appears to be adequately described by two-compartment model. In contrast, at the high concentration level the experimental data were better described using the first-order kinetic model rather than the two-compartment model, except for Istria II soil, where biphasic kinetic was observed (Figure 8d). In fitting two-compartment model, we assumed that the fast degradation phase occurred from 0 to 15 days after application and that the slow degradation phase occurred thereafter. This was visually determined based on the changes of the slopes of the degradation curves. Capri et al. (2001), who studied the degradation of IMI in Italian soils, found that IMI concentration decreased rapidly in the first 10 days followed by a slower decrease in the total amount recovered. The derived rate constants as well as DT50 and DT90 for each model are shown in Table 6 together with the correlation coefficient for each curve (*R2*) and with root mean square error (RMSE). Both statistical indices (*R2* and RMSE) indicated that the first-order kinetic model better described IMI degradation at the high concentration level in all tested soils, except in Istria II soil, than at the low concentration level (Table 6). At the low concentration level, the more complex two-compartment model generated smaller RMSE.

Behavior and Fate of Imidacloprid in Croatian Olive Orchard Soils Under Laboratory Conditions 509

the range from 40 to 229 days (first-order kinetic) (Sarkar et al., 2001; Schad, 2001). However, high values of 156 days (Krohn & Hellpointner, 2002) and a greater than a year (Baskaran et al., 1999) have been measured. The DT50 values for IMI degradation in the present study are comparable with those reported under field conditions (Schad, 2001; 96 days) and degradation was slower than in other study at laboratory conditions (Sarkar et al., 2001; 40 days). If we compare the DT50 values derived from the used kinetic models between the two examined initial concentration levels, we can see that these values differed significantly. Higher persistence of IMI was observed at higher initial concentration level (mean DT50 = 118.46 days) compared to lower concentration (mean DT50 = 90.62 days), which was statistically significant (p = 0.020). This seems to lead to the conclusion that concentration

Fitted Model I Model II

or index\* Krk I Istria I Krk II Istria II Krk I Istria I Krk II Istria II

*k*1 (1/d) 0.0076 0.0090 0.0048 0.0114 0.0187 0.0210 0.0055 0.0353 *k*2 (1/d) 0.0020 0.0031 2.21\*10-11 0.0075 *a* (mg/kg) 0.4613 0.4507 0.4297 0.4407 0.2798 0.2928 0.2209 0.2965 *b* (mg/kg) 0.2060 0.1814 0.2106 0.1675 DT50 (d) 91.20 77.02 144.41 60.80 82.39 63.60 145.97 50.18 DT90 (d) 302.97 255.84 479.71 201.98 722.10 466.64 914.97 247.38 *R*<sup>2</sup> 0.9679 0.9695 0.9917 0.9844 0.9939 0.9907 0.9823 0.9978 RMSE 0.0212 0.0219 0.0077 0.0167 0.0103 0.0135 0.0084 0.0071

*k*1 (1/d) 0.0053 0.0063 0.0042 0.0093 0.0053 0.0063 0.0042 0.0356 *k*2 (1/d) 0.0053 0.0063 0.0042 0.0051 *a* (mg/kg) 5.3848 4.5780 4.6514 4.3303 3.1555 2.9662 2.5149 3.4221 *b* (mg/kg) 2.2293 1.6118 2.1365 1.4170 DT50 (d) 130.78 110.02 165.04 74.53 130.78 110.02 165.04 54.86 DT90 (d) 434.45 365.49 548.23 247.59 434.45 365.49 548.23 345.76 *R*2 0.9923 0.9934 0.9919 0.9483 0.9923 0.9934 0.9919 0.9959 RMSE 0.1010 0.0889 0.0769 0.2779 0.1130 0.0984 0.0859 0.0727 \* DT50 and DT90 are times for 50 and 90% of the initial residues to degrade; *k*1 and *k*2 are first-order rate constants in the rapid and slow degradation pools; *a* and *b* are initial concentrations in the rapid and slow degradation pools; *R*2 is the coefficient of determination; RMSE is the root mean square error

Table 6. Fitted parameters for the first-order kinetics model and two-compartment model for

These results suggested that the persistence of IMI was significantly influenced by soil properties. Kendal-Tau correlation analysis between DT50 and selected soil properties demonstrated that IMI persistence in the tested soils was inversely connected to CEC, clay and OC content, with a strongest relationship between DT50 and CEC (Table 4). The analysis showed the positive, but very weak correlation between DT50 with soil pH. Other studies have found reasonable correlation between DT50 and pH. For example, Sarkar et al. ( 2001) showed that the persistence of IMI tended to increase as soil pH increased. In addition,

Concentration level 0.5 mg/kg

Concentration level 5 mg/kg

level significantly affected IMI degradation.

describing IMI degradation in the tested soils.

parameter

Fig. 8. Degradation of IMI in the tested soils at 5 mg/kg concentration level. Values are means ± standard deviations. Symbols represent the experimental data, while lines represent the theoretical curves fitted by the first-order kinetics model or two-compartment model.

The adequacy of the first-order kinetic model description at the high concentration level can be seen by comparing the fitted rate constants (Table 6). The fitted *k1* for the first-order kinetic model was equal (except the Istria II soil) to that in the rapid degradation pool of the two-compartment model. Moreover, the two-compartment model had the same rate constant in the rapid and slow degradation pools, which is also equal to *k1* of the first-order kinetic model. The rate of degradation was highest in the Istria II soil and the lowest in the Krk II soil. The higher degradation rate in Istria II soil than in Krk II soil could be due to the higher OC content of Istria II soil vs Krk II soil.

The calculated DT50 and DT90 values for the tested soils ranged from 50.18-145.97 and 247.38-914.97 days at the low concentration level respectively, and from 54.86-165.04 and 345.76-548.23 days at the high concentration level, respectively (Table 6). The DT50 and DT90 values were highest (for both concentration levels) for the Krk II soil, which were significantly higher than that in the other soils. The lowest DT50 and DT90 were observed in the Istria II soil. Estimated values for DT50 in our study prove that IMI can be categorized as moderately persistent pesticide (DT50 from 30 - 100 days) (Gavrilescu, 2005). Previous studies of IMI degradation at laboratory and at field conditions have reported DT50 values in

**Imidacloprid residue (mg/kg)**

0

**Imidacloprid residue (mg/kg)**

0

Fig. 8. Degradation of IMI in the tested soils at 5 mg/kg concentration level. Values are means ± standard deviations. Symbols represent the experimental data, while lines

represent the theoretical curves fitted by the first-order kinetics model or two-compartment

The adequacy of the first-order kinetic model description at the high concentration level can be seen by comparing the fitted rate constants (Table 6). The fitted *k1* for the first-order kinetic model was equal (except the Istria II soil) to that in the rapid degradation pool of the two-compartment model. Moreover, the two-compartment model had the same rate constant in the rapid and slow degradation pools, which is also equal to *k1* of the first-order kinetic model. The rate of degradation was highest in the Istria II soil and the lowest in the Krk II soil. The higher degradation rate in Istria II soil than in Krk II soil could be due to the

The calculated DT50 and DT90 values for the tested soils ranged from 50.18-145.97 and 247.38-914.97 days at the low concentration level respectively, and from 54.86-165.04 and 345.76-548.23 days at the high concentration level, respectively (Table 6). The DT50 and DT90 values were highest (for both concentration levels) for the Krk II soil, which were significantly higher than that in the other soils. The lowest DT50 and DT90 were observed in the Istria II soil. Estimated values for DT50 in our study prove that IMI can be categorized as moderately persistent pesticide (DT50 from 30 - 100 days) (Gavrilescu, 2005). Previous studies of IMI degradation at laboratory and at field conditions have reported DT50 values in

1

2

3

4

5

6

1

2

3

4

5

6

**b) Istria I soil**

**d) Istria II soil**

**Time (days after application)** 0 50 100 150 200

> Experimental (mean ± stdv ) Fitted, Two-compartment (*R<sup>2</sup>* = 0.9959 )

**Time (days after application)** 0 50 100 150 200

Experimental (mean ± stdv ) Fitted, First-order (*R2* = 0.9934)

**a) Krk I soil**

**c) Krk II soil**

**Imidacloprid residue(mg/kg)**

**Imidacloprid residue (mg/kg)**

0

model.

1

2

3

4

5

6

**Time (days after application)** 0 50 100 150 200

**Time (days after application)** 0 50 100 150 200

higher OC content of Istria II soil vs Krk II soil.

Experimental (mean ± stdv) Fitted, First-order (*R2* = 0.9923)

Experimental (mean ± stdv ) Fitted, First-order (*R2* = 0.9919) the range from 40 to 229 days (first-order kinetic) (Sarkar et al., 2001; Schad, 2001). However, high values of 156 days (Krohn & Hellpointner, 2002) and a greater than a year (Baskaran et al., 1999) have been measured. The DT50 values for IMI degradation in the present study are comparable with those reported under field conditions (Schad, 2001; 96 days) and degradation was slower than in other study at laboratory conditions (Sarkar et al., 2001; 40 days). If we compare the DT50 values derived from the used kinetic models between the two examined initial concentration levels, we can see that these values differed significantly. Higher persistence of IMI was observed at higher initial concentration level (mean DT50 = 118.46 days) compared to lower concentration (mean DT50 = 90.62 days), which was statistically significant (p = 0.020). This seems to lead to the conclusion that concentration level significantly affected IMI degradation.


\* DT50 and DT90 are times for 50 and 90% of the initial residues to degrade; *k*1 and *k*2 are first-order rate constants in the rapid and slow degradation pools; *a* and *b* are initial concentrations in the rapid and slow degradation pools; *R*2 is the coefficient of determination; RMSE is the root mean square error

Table 6. Fitted parameters for the first-order kinetics model and two-compartment model for describing IMI degradation in the tested soils.

These results suggested that the persistence of IMI was significantly influenced by soil properties. Kendal-Tau correlation analysis between DT50 and selected soil properties demonstrated that IMI persistence in the tested soils was inversely connected to CEC, clay and OC content, with a strongest relationship between DT50 and CEC (Table 4). The analysis showed the positive, but very weak correlation between DT50 with soil pH. Other studies have found reasonable correlation between DT50 and pH. For example, Sarkar et al. ( 2001) showed that the persistence of IMI tended to increase as soil pH increased. In addition,

Behavior and Fate of Imidacloprid in Croatian Olive Orchard Soils Under Laboratory Conditions 511

concentration level during the period of 150-180 days after application. Formation of the 6- CNA from IMI in soil has been reported earlier (Scholz, 1992). 6-CNA accounted for a maximum of about 15 and 10% of the initial concentration of IMI for the 5 and 0.5 mg/kg, respectively, in the Istria II soil. The corresponding minimum values for 6-CNA were 6 and

**6-CNA (mg/kg)**

Fig. 9. Formation of 6-CNA in the tested soils at concentration level of: a) 5 and b) 0.5 mg/kg. Values are means ± standard deviations. Symbols represent the experimental data,

The sorption-desorption and degradation of IMI was examined to understand the influence of concentration and soil properties on its behavior and fate in soils of Croatian coastal regions. The experimental data revealed that the sorption and desorption isotherms of IMI in the tested soils were nonlinear over the concentration range used, which can be best described by the Freundlich equation. Soil sorption capacity of IMI depended significantly on the soil properties. Especially, the sorption behavior of IMI was largely dependent on the soil OC content, where the soils with higher OC content (Istria soils) showed higher sorption capacity and less potential mobility of IMI. Given the spatial difference between tested soils, statistically significant differences in soil sorption capacity were found among and within soils of Istrian and Krk region. According to calculated *KOC* values, IMI can be categorized as a medium mobility pesticide indicating that rational use of IMI entails little danger of the ground-water contamination. In all soils, a higher sorption capacity was observed at lower IMI concentrations, indicating that the percentage of desorbed amount of pesticide increased with increasing initial solution concentration. Desorption experimental data deviated significantly from the sorption data, indicating that these processes were distinctly different in tested soils. It can be assumed, that the desorption process appeared to be the result of a complex, time dependent interplay of several chemical and physical processes and irreversible binding of IMI to soil surfaces, leading to hysteresis. The negative and low values of the Gibbs free energy of the IMI sorption indicated exotermic characteristics of sorption reaction and corresponded to the physical process, suggesting that partitioning into soil organic matter was the main mechanism of IMI sorption in the soils used. IMI kinetic behavior in all tested soils at the high concentration level can be described by the first-order

while vertical bars represent the standard deviation in the triplicate samples.

**b) IMI (0.5 mg/kg)**

Krk I soil Istria I soil Krk II soil Istria II soil

**Time (day)** 0 20 40 60 80 100 120 140 160 180 200

9%.

**6- CNA (mg/kg)**

0

**5. Conclusions** 

200

400

600

800

**a) IMI (5 mg/kg)**

 Krk I soil Istria I soil Krk II soil Istria II soil

**Time (day)** 0 20 40 60 80 100 120 140 160 180 200

multiple liner regression confirmed that IMI persistence was primarily correlated with OC content, with a regression equation of:

$$\text{DT}\_{50} = 72.7581 \,\text{OC} \cdot \text{-4.9850} \,\text{CEC} \cdot \text{-0.4689} \,\text{clay} + 11.8127 \,\text{pH} + 116.50 \left(\text{R}^2 = 0.825\right) \tag{14}$$

The DT50 values were further tested to determine the effects of soil type on the DT50 of IMI. Statistically significant differences in soil persistence, were found among the Krk (mean DT50 = 132.09 days) and Istria (mean DT50 = 77.00 days) region (*p* = 0.000002). In addition, results of degradation study within the regions showed a statistically significant difference in DT50 values between the soils Istria I (mean DT50 = 92.53 days) and Istria II (mean DT50 = 61.46 days) (*p* = 0.002), as well as between soils Krk I (mean DT50 = 108.16 days) and Krk II (mean DT50 = 156.02 days) (*p* = 0.00001).

Examining the soil properties reveals that the most contrasting difference between tested soils, with respect to IMI degradation, is soil OC content, which was in the range from 1.30 to 1.91% for Istrian soils and from 0.42 to 0.55% for Krk soils. Higher OC content in Istrian soils would cause more IMI sorption by the soil based on the concept proposed by (Park et al., 2003). An equilibrium sorption study, conducted to verify this hypothesis showed that IMI equilibrium sorption constants were higher for the Istrian soils than for the Krk soils (2.60 and 3.28 for the Istria I soil and Istria II soil; 1.28 and 1.53 for the Krk II soil and Krk I soil). Higher OC content in Istrian soils might have been accompanied by higher microbial population and activities that promoted biodegradation processes of IMI (Cox et al., 1997; Getenga et al., 2004; Park et al., 2003). In describing degradation of 2,4-D (2,4 dichlorpphenoxyacetic acid) Picton & Farenhorst (2004) hypothesized a mechanism according to which initially readily available chemical resulted in apparent rapid degradation, while subsequent increased binding to soil caused noticeable reduction in degradation rate. When comparing IMI degradation in all the tested soils, we observed that IMI degraded faster in Istrian soils than in Krk soils, although more IMI was sorbed in Istrian soils. Thus, it appears that sorption did not significantly inhibit IMI degradation in the soil. Otherwise, IMI should have degraded faster in the Krk soils than in the Istrian soils. Fitting two-compartment model to the measured data showed that the degradation rate constants in the rapid degradation pool in the Istrian soils were greater than those in the Krk soils at both concentration level (Table 6). Calculations using two compartment model based on the data in Table 6 revealed that readily available IMI amount in the rapid degradation pool initially represented 51-58 and 54-59% of the applied IMI in Krk soils and 62-64 and 65- 71% in Istria soils at the low and high concentration level, respectively. These results suggested that IMI in the rapid degradation pool is not equivalent to the dissolved IMI molecule, as Wolt, (1997) proposed. Although pesticide molecule in soil solution is generally thought to be readily available to microorganisms for biodegradation, there is evidence that sorption can accelerate pesticide degradation (Park et al., 2003).

#### **4.2.2 6-CNA formation**

Metabolism of IMI was also studied in four Croatian soils at both concentration levels. The amount of 6-CNA, which was detected in all the tested soils as a metabolic product, varied irregularly with the time (Figure 9). The maximum concentration of 6-CNA in the tested soils was in the range from 280 to 720 µg/kg for the 5 mg/kg concentration level, while the corresponding concentration of 6-CNA was from 36.2 to 54.9 µg/kg for the 0.5 mg/kg concentration level during the period of 150-180 days after application. Formation of the 6- CNA from IMI in soil has been reported earlier (Scholz, 1992). 6-CNA accounted for a maximum of about 15 and 10% of the initial concentration of IMI for the 5 and 0.5 mg/kg, respectively, in the Istria II soil. The corresponding minimum values for 6-CNA were 6 and 9%.

Fig. 9. Formation of 6-CNA in the tested soils at concentration level of: a) 5 and b) 0.5 mg/kg. Values are means ± standard deviations. Symbols represent the experimental data, while vertical bars represent the standard deviation in the triplicate samples.
