Pesticide analytical standards (≥ 96%) were obtained from Dr. Ehrenstorfer GmbH (Augsburg, Germany). The main physical-chemical properties of the active ingredients are shown in Table 1. Experimental values of the octanol/water partition coefficient (K_OW), soil/organic partition coefficient (K_OC), aqueous solubility (S_W), and groundwater ubiquity score (GUS) index were taken from The Pesticide Properties DataBase (AERU, 2015) and KOCWIN™ included in the EPI Suite v4.11 program provides by USEPA (2015).

Stocks solutions (1000 µg/mL) of each pesticide standard were prepared in acetonitrile, protected from light and stored at 5ºC. Several standard solutions, with concentrations of 0.5-200 µg/L, were injected to obtain the linearity of detector response and the detection limits (LOD) of the compounds studied. Pesticide grade acetonitrile and dichloromethane were supplied by Scharlab (Barcelona, Spain).

Three different agricultural soils were chosen for this study. Soil A (Hipercalcic calcisol), soil B (Calcaric regosol) and soil C (Haplic calcisol) were taken from the Campo de Cartagena (Murcia, south-eastern Spain). Soil samples were collected from the surface (top 20 cm), air-dried, and passed through a 2 mm sieve. The experimental studies began within four days after sampling and storage under cool conditions. The composition of the soils is shown in Table 2.

The main characteristics of the soil (total organic carbon, pH, alkalinity, texture) were determined by standard soil analysis techniques (Reeuwijk, 2002) The clay content was determined by means of a powder X-ray diffractometer (XRD) on a Philips PW 1700 using Cu-Kα radiation at a scanning speed of 1º (2θ) after washing, disaggregation and dispersion of the sample.

The experiment was conducted according to the OECD guidelines (OECD, 2007). Downward movement of the pesticides was studied in polyvinyl chloride (PVC) columns of 40 cm (length) × 4 cm (i.d). There were three sets of columns, each filled with 200 g of a different kind of soil: A, B or C. The top 3 cm of the columns were filled with sea sand and the bottom 3 cm with sea sand plus nylon mesh with an effective pore diameter of 60 µm to minimizing the dead-end volume and prevent losses of soil during the experiment. Experiments were performed in triplicate, at room temperature and protected from direct light. Before the application of the pesticides, columns were conditioned with 0.01 M CaCl₂ in distilled water to their maximal water holding capacity and then allowed to drain for 24 h. The pore volume (PV) of the packed columns was estimated by the weight difference of water-saturated columns versus dry columns. The calculated PVs (mL) of the soil columns after saturation were 70.1±1.9 (soil A), 73.6±2.2 (soil B), and 79.3±1.6 (soil C). After this, 1.0 mL of methanol/water solution (10+90, v/v) containing 20 µg of every compound were added to the top of each column. Twenty four hours after pesticide application, the compounds were leached by adding 800 mL of 0.01 M CaCl₂ during 16 days with a peristaltic pump. CaCl₂ instead of water was used in order to minimize soil mineral balance disruption. The leachates (50 mL/day) were quantitatively collected at the bottom of the columns, and then filtered through nylon membrane filter (0.45 µm). After this time, the columns were opened and the soil separated in two segments of approximately 10 cm each.

The soil degradation experiments were conducted according to the OECD procedure (OECD, 2002). To ascertain pesticide persistence, soil samples (50 g, on a dry weight basis) were placed in incubation flasks (n=3). Two milliliters of a methanol/water solution (10+90, v/v) containing all the studied pesticides was applied in order to achieve an initial concentration of 100 µg/kg of each. All the flasks were incubated in the dark for 120 days at 21±2ºC with the soil moisture ranging from 40-60% of the water holding capacity. Throughout the incubation period water losses exceeding 10% of the initial values were compensated by the addition of fresh distilled water every two weeks. Relative humidity was maintained at 70%. After 7, 15, 30, 60, and 120 days of incubation, three flasks from each treatment were taken and kept at 4°C for 2-3 days until analysis.

The dissipation of all pesticides could be described by pseudo first-order kinetics: lnC_t = lnC₀−kt, where C_t is the concentration of pesticide in soil; C₀ and k denote y-intercept values and the slope of the dissipation lines, respectively; and t is the post-application time in days. Hence, their half-lives (t_½) were calculated from the equation t_½ = ln2/k. The curve fitting and statistical data were obtained using SigmaPlot v. 12.0 statistical software (Systat, Software Inc., San Jose, CA, USA).

On the other hand, the K_d (distribution coefficient) value was considered a measure of pesticide adsorption capacity by the soils. The sorption of pesticides was determined using a batch equilibrium method (OECD, 2000). Triplicates 5 g of each soil were equilibrated with 20 mL of an aqueous solution of pesticides at an initial concentration (C_i) of 25 ng/mL for 24 h at 22±1ºC. The amount adsorbed (C_S) was considered to be the difference between C_i and that remaining after equilibration (C_e). K_d values were obtained from the relationship between C_s and C_e(K_d=C_s/C_e). The organic carbon normalized adsorption coefficient (K_OC) was calculated as (K_d∙100)/%OC.

Water samples (10 mL) were extracted via a liquid-liquid extraction with acetonitrile (10 mL) by sonication, followed by a salting-out step with 2 g NaCl. The samples were shaken and centrifuged for 10 min at 3000xg. Dried soil samples (10 g) were extracted with 20 mL of acetonitrile/water (2/1) by sonication. Then, 20 mL of dichloromethane were added and the mixture was centrifuged for 10 min at 1900xg. Finally, water and soil extracts were filtered quantitatively through a glass funnel with a DP302 filter separation phase paper, 150 mm diameter (Albet, Barcelona). The organic phase was concentrated to dryness by rotary vacuum evaporation. The residue was redissolved in 1 mL of acetonitrile, filtered through 0.45 µm filter and analyzed by liquid chromatography in tandem with mass spectrometry (LC-MS²).

The separation, identification and quantification of pesticides studied were carried out by an HPLC system (Agilent Series 1100, Agilent Technologies, Santa Clara, CA, USA) equipped with a reversed phase C8 analytical column of 150 mm × 4.6 mm and 5 µm particle size (Zorbax Eclipse XDB-C8). The mobile phases A and B were acetonitrile and 0.1% formic acid, respectively. The gradient program started with 10% A, constant for 5 min, followed by a linear gradient to 100% A to 35 min. After this run time, 10 min of post-run time followed using initial 10% of A. The flow-rate was constant (0.6 mL/min) during the whole process. In all cases 5 µL of sample were injected. For the mass spectrometric analysis, an Agilent G6410A triple quadrupole mass spectrometer equipped with an ESI interface operating in positive ion mode, using the following operation parameters: capillary voltage, 3000 V; nebulizer pressure, 40 psi; drying gas, 9 L/min; drying gas temperature, 350ºC. Mass spectra were recorded across the range 50–1000 m/z.

The calibration curves were prepared in soil and water. The calibration samples were analyzed by spiking pesticides at 0.5-200 µg/L levels into water and soil samples in five replicates. In all cases the correlation coefficients were >0.99. Limit of detection (LOD, defined as the lowest amount of pesticide in a sample which can be detected but not necessarily quantitated as an exact value) and limit of quantitation (LOQ, as the lowest amount of pesticide in a sample which can be quantitatively determined with suitable precision and accuracy) were evaluated by injecting standard solution into blank matrix at different concentration levels. LOQs were calculated from the signal-to-noise (S/N) ratio of 10. LOQs varied from 0.02-0.3 µg/L and 0.07-0.7 µg/kg in water and soil, respectively.

The shape of the curves for soils B and C (Fig. 2) indicates a certain interaction with the organic and inorganic soil colloids unlike soil A. Clay-humus interactions contribute to the high organic matter content of clay soils because the organic matter entrapped in the micropores (<1 µm) formed by clay particles is physically inactive to decomposing organisms. Dissolved pesticides may be carried downward rapidly by water through large macropores. This type of non-uniform movement, referred as preferential flow, greatly increases the chances of groundwater pollution, while organic matter improve soil water retention since it increases both infiltration rate and water-holding capacity (Brady & Weil, 2010).

It is generally accepted that the leaching of a compound depends mainly on its degradation and sorption to the soil (Moorman et al., 2001). Adsorption is the binding of pesticides to mineral or organic matter. Sorption retards movement, and may also increase persistence because the pesticide is protected from degradation. Among various soil properties, OC content is the single largest factor that has maximum influence on pesticide degradation, adsorption and mobility in soil (Wauchope et al., 2002). In our case, the three soils have similar clay content (29-33%). Moreover, the percentage of weight of montmorillonite, an expanding 2:1-type silicate clay with a very high total specific surface area (800 m²/g) is also similar (37-40%) in the three studied soils. Therefore, the differences observed in the leachability can be attributed to the different organic carbon content in the soils (0.7-1.8%), mainly for the soil C.

The coefficient K_OC describes the tendency of a pesticide to bind to soil. At higher values of the K_OC, the greater the sorption potential. Results showed in Table 3 are consistent with the organic carbon content in the soil. Recently, Navarro et al. (2012) pointed the strong decrease of the leachability of 12 substituted phenylurea herbicides through an Endoleptic phaeozen (OC=6.1%) as compared with a Hypercalcic calcisol (OC=0.7%). As a consequence, a strategy to reduce the movement of the pesticides through the soil profile is the addition of organic wastes (Fenoll et al., 2014a,b; 2015).

Chemical reactions between unaltered pesticides and/or their metabolites often lead to the formation of strong bonds (chemisorption), resulting in an increase in the persistence of the residues in the soil, while causing it to lose its chemical identity. From a toxicological point of view, binding of pesticides to humus lead to a decrease of material available to interact with biota, a reduction in the toxicity of the compound, and immobilising the pesticide, thereby reducing its leaching and transport properties ­(Navarro et al., 2007).

The calculated half-lives (as mean value for the three soils) showed in Table 4 are consistent with those included as typical DT50 in the Pesticide Properties Database (AERU, 2015). According to GUS index (Gustafson, 1989), carbofuran and ethoprofos behave as leacher compounds while propyzamide has transient properties (i.e., borderline leacher). As a consequence, their permanence in the soil is lower. In addition to pencycuron (phenylurea fungicide) and cadusafos (organophosphate insecticide) the triazine compounds with the exception of prometon (methoxy-triazine) show moderate persistence in the order triazinones < methylthio-triazines < chloro-triazines. Typical values of t_½ or DT50 for triazine compounds range from 14 to 102 days, with a mean t_½ or DT50 of 36±25 days (Koskinen & Banks, 2008). Triazine retention in soil is influence primarily by OC content, soil clay content ant type, and pH. Other important factors influencing retention include the amount applied, the amount of dissolved OC in soil solution, soil water content and soil contact time (aging). Isoxaben and prometon were the compounds with higher persistence (t_½>100 days).

Pesticide residues can be found in the soil as free and bound residues. Such terms were coined to indicate that the former can be readily extracted from soil without altering their chemical structures, whereas the latter are resistant to such extraction. However, the distinction between these two fractions is not always clear, because while they are in the soil, even the extractable residues are not entirely free from any form of binding because they may be sorbed to the soil solid phases and, therefore, show reduced bioavailability and degradation. The environmental significance of a bound residue depends not on its nonextractability under laboratory conditions, but on its bioavailability. Depending on the nature of binding, immobilised pesticides may be released back to the soil solution or mineralized as a result of changing environmental conditions in the soil (Khan, 1982).

To conclude, the results obtained in this work on texturally similar agricultural soils, soil A (Hipercalcic calcisol), soil B (Calcaric regosol), and soil C (Haplic calcisol) show that all the studied pesticides behave as leacher compounds although in different proportions. Cadusafos, isoxaben, pencycuron, prometryn, propyzamide, simetryn, terbuthylazine, and terbutryn (G1) show medium leachability while atrazine carbofuran, ethoprophos, metamitron, metribuzin, prometon, propazine, and simazine (G2) show high leachability through the soil columns. As a general rule, K_OC (the tendency of a pesticide to bind to soil) values are higher in the order: soil C > soil B > soil A, which agrees with the OC content in each soil. As a consequence, at higher values of the K_OC, the greater the sorption potential and lower the leaching.

On the basis of the half-lives obtained, carbofuran, ethoprophos and propyzamide can be categorized as non-persistent pesticides (t_½<30 days) while isoxaben and prometon can be classified as persistent pesticides (t_½>100 days) showing the rest of the studied pesticides moderate persistence (t_½=30-100 days). As it can be observed, when higher OC content in the soil the greater persistence of the pesticide was observed. Disappearance of the pesticides follows first order degradation kinetics although for the most persistent compounds (isoxaben and prometon) biphasic models could be more appropriate because there is an initial rapid degradation and a subsequent phase in which the compounds disappear more slowly.

Bearing in mind that these residues can reach water bodies by leaching and runoff, and as they are inherently toxic molecules, they can adversely affect aquatic environment and non-targeted organism. Moreover, presence of these pesticides in groundwater is extremely hazardous as groundwater is major source of drinking water. Therefore, the results obtained point to the interest in the use of organic wastes such as agro-industrial, composted manure, sewage sludge, etc., by-products available at low- or no-cost in reducing the pollution of groundwater by pesticide drainage, mainly in arid and semiarid regions, as Mediterranean area where soils generally have low organic matter contents.