Chemicals and reagents

The pesticide standards with purity > 98% were provided by Sigma–Aldrich (Gillingham, UK) and Dr. Ehrenstorfer (Augsburg, Germany). Selected characteristics of compounds are included in Table 1. A concentration of 500 or 1000 μg mL^-1 solution was initially prepared in methanol for each standard and then solutions with all of the analytes were prepared at lower concentrations in order to prepare the calibration curves and spike the ultra-high quality (UHQ) or uncontaminated river water samples to evaluate the matrix effect.

River water sampling

During 2016 surface water samples were collected at various points in the Cachapoal River basin in Central Chile (located in the region with the greatest pesticide sales in the country), comprising two rivers (Claro and Cachapoal), two creeks (La Cadena and Zamorano) and two irrigation canals (Las Cabras and Tahuilla). This region has an area of 6370 km² and the main crops are grapes, apples, peaches, plums, cherries, nectarines, pears, oranges, walnuts, avocados and lemons (Figure 1). Two samplings were carried out, one at the end of the summer (April) and the other in winter season (July), with a total of thirty water samples collected (eighteen in April and twelve in July). The samples were taken in 2-L amber glass bottles and transported to the laboratory in coolers with ice. In less than 48 hours, they were passed through nitrocellulose filters with a pore size of 0.45 μm and stored at −20°C until their pre-concentration by SPE.

Figure 1 Map of land use of Cachapoal River basin (Central Chile) and surface sampling points. Self-elaboration based on data Fruit Catastro year 2015 [²⁴]. 

Analytical methodology and GC-MS chromatographic conditions

The effects of sorbent type, the solvent used in the elution and pre-concentrated sample volume were previously evaluated to optimize the pre-concentration process by SPE. Five SPE sorbents for extracting the compounds of interest were compared Oasis HLB (60 mg, Waters), LiChrolut EN (200 mg, Merck), Strata X (60 mg, Phenomenex), silica C18 (Sep-Pak Plus 900 mg, Waters) and carbon (SampliQ 500 mg, Agilent)). The procedure was followed as indicated by Herrero-Hernández, et al. [²¹]. In brief, they were conditioned with 10 mL of methanol and 10 mL of UHQ water and a volume of 100 mL of UHQ water spiked with all of the analytes at a concentration of 1 μg L^-1 was concentrated. The elution was carried out with 8 mL of eluent, which was subsequently evaporated to dryness under a nitrogen stream in an EVA-EC2-L evaporator (VLM GmbH, Bielefeld, Germany). The residue was redissolved in 1.0 mL of methanol and the extract was filtered with 0.45 μm GHP Acrodisc filters for subsequent analysis by GC-MS. The effects of five organic solvents (acetone, acetonitrile, methanol, hexane, ethyl acetate and a combination of acetone and acetonitrile) as eluents and five spiked UHQ water volumes (50, 100, 250, 500 and 1000 mL) were also assessed.

The analysis was carried out in an Agilent 7890 gas chromatograph coupled to an Agilent 5975C Triple Axis MSD mass spectrometer and the separation was done using a ZB-5MS capillary column (Phenomenex) (30 m x 0.25 mm id with 0.25 μm film thickness). The chromatographic conditions that allowed good separation and quantification of the analytes were as it is indicated by Herrero-Hernández, et al. [²¹]. The total analysis time was thirty minutes. To identify each pesticide and degradation product, the three most abundant ions were selected by separately injecting each compound and, for quantification, the ion found in the greatest proportion was selected.

Analytical method validation

The different analytic parameters were assessed using samples of UHQ water or uncontaminated natural water spiked with known concentrations of the different compounds. The matrix effect was evaluated in duplicate by comparing the analyte signal obtained when pre-concentrating 500 mL of UHQ water with that obtained when pre-concentrating 500 mL of river water, both spiked with a concentration of 0.1 μg L^-1 of all compounds. The linearity of the calibration curves was evaluated at a concentration range between 0.1 and 1.5 μg L^-1 using calibration standards prepared in uncontaminated river water (matrix-matched calibration standards) to ensure good quantification. Calculations were made using peak areas. The accuracy (average recovery) was determined by evaluating the recovery percentage for five replicates and precision of the method (repeatability and reproducibility) was assessed by determining the concentration, on the same day, of five samples of river water (500 mL) spiked with 0.1 μg L^-1 of each compound, as well as the concentration variation of each analyte, for the same sample, for five days. The limits of detection (LODs) and quantification (LOQs) were estimated as the analyte concentration with a signal-to-noise-ratio of 3 and 10, respectively.

Statistical data analysis

Intra-group comparisons were carried out to verify differences in the total concentration of pesticides between the collected samples at the end of summer and during the winter season. The statistical t student was used for this purpose and the p values less than 0.05 were considered significant. The statistical analyses were performed using R software. It is important to mention that for logistical problems, Las Cabras Canal and Cachapoal River were not sampled at both seasons; therefore the statistical analysis was performed for Tahuilla Canal, Cadena Creek, Zamorano Creek and Claro River.

Optimization of GC-MS chromatographic conditions

To optimize the GC-MS chromatographic conditions, the individual standards were injected in methanol at a concentration of 1 μg mL^-1. Each compound was analyzed in scan mode to identify the most abundant ions and retention times. The conditions described by Herrero-Hernández, et al. [²¹] were initially used, maintaining the carrier gas flow rate at 1.5 mL min^-1 and using splitless injection mode. However, it was necessary to modify the temperature ramp in order to achieve good peak resolution and separation. The initial conditions were the following: from 100 to 200°C at 20°C/min (holding time 1 min), to 210°C at 1°C/min, to 230°C at 5°C/min, and to 315°C at 50°C/min (2.3 min), total chromatographic time of 24 min. In this research, the chromatographic conditions that allowed good separation and quantification of the analytes were the following: electron impact ionization mode, an ionization energy of 70 eV, an electron multiplier voltage of 1700 V, a source temperature of 230°C, a transfer line temperature of 150°C, helium carrier gas, a flow rate of 1.0 mL min^-1 and an injection volume of 5 μl at 220°C in splitless mode. The following temperature gradient was used: from 80 to 180°C at 20°C min^-1 (2 minutes), from 180 to 190°C at 1°C min^-1 and from 190 to 310°C at 80°C min^-1 (11.5 minutes). The three most abundant ions of each compound were selected (one to quantify and two to confirm) by comparing the spectra from the mass library (NIST 08) with those obtained when injecting the standards in scan mode. Finally, three different windows were created as a function of retention times in which the selected ions for the compounds included in each window were determined (Table 2).

Optimization of the pre-concentration stage and analytical method validation

To optimize the SPE method, the recovery percentage of each of the compounds was evaluated by modifying the sorbent type, the solvent used for elution and the sample volume. Results obtained for five different types of sorbent tested are shown in Figure 2a and 2b. In general, the recovery percentages obtained with C₁₈ and LiChrolut cartridges were similar for most of the compounds, with values that fluctuated between 51 and 91%. The lowest recovery percentages were obtained using the carbon-based sorbent, with the fungicides cyprodinil and pyrimethanil and the insecticide chlorpyrifos presenting the lowest recovery percentage (< 20%). The best results were obtained using Oasis HLB and Strata X cartridges, with 88 and 69% of the studied compounds, respectively, presenting a recovery percentage over 60%.

Figure 2 Recoveries of pesticides after solid phase extraction with different sorbents (a and b). In all cases, 100 mL of UHQ water are preconcentrated with all compounds at 1.0 μg L-1 and eluted with 8 mL of methanol. 

According to these results the Oasis HLB sorbent was selected to evaluate the efficacy of different organic solvents tested and the preconcentrated sample volume. The use of hexane resulted in the worst recovery percentages, which were < 50% for all of the compounds. When ethyl acetate was used, only eleven compounds presented a recovery percentage over 60%. Meanwhile, when using methanol, acetone or acetonitrile, more than 70% of the compounds presented good recoveries, with the compound type varying as a function of polarity (Figure 3a and 3b). Therefore, it was decided to use a sequential elution with acetone and acetonitrile, resulting in recovery values > 70% for 73% of the studied compounds.

Figure 3 Recoveries of pesticides after solid phase extraction with different solvents using Oasis HLB catridges (a and b). In all cases, 100 mL of UHQ water are preconcentrated with all compounds at 1.0 μg L-1 and eluted with 8 mL of methanol. 

Another of the studied parameters was pre-concentrated sample volume, for which 50, 100, 250, 500 and 1000 mL of UHQ water spiked with 0.2 μg of each compound were passed through Oasis HLB cartridges. The results showed that the recovery percentage remains constant when pre-concentrating between 50 and 500 mL of water, and decreases considerably when pre-concentrating a greater sample volume. Considering this variation and the time required to pre-concentrate the various volumes of water, it was decided to work with a volume of water of 500 mL since it allows a greater recovery percentage to be obtained.

To evaluate the matrix effect, the analyte signals were compared in duplicate at a concentration of 0.1 μg L^-1 by pre-concentrating 500 mL of UHQ water and river water. Most of the compounds presented a signal suppression or enhancement between 5 and 19%, with the exception of the degradation products CGA-92370, DEA, DET and diazoxon, which presented a variation > 25%. Therefore, to minimize the matrix effect and avoid a decrease in or overestimation of the pesticide concentration, matrix-matched standards were used.

The linearity parameters, LODs and LOQs, were determined by preparing a calibration curve between 0.1 – 1.5 μg L^-1 for each compound in 500 mL of river water treated in the same way as the samples. Good linearity was obtained for all the samples, with correlation coefficients > 0.99 and limits of quantification lower than the maximum established by the EC. The LOD fluctuated between 0.011 and 0.030 μg L^-1 for diazoxon and acetochlor, respectively, and the LOQ between 0.036 and 0.099 μg L^-1 for the same compounds. Both parameters were estimated as 3 and 10 times the signal/noise relationship for each analyte, respectively (Table 3).

The precision of the method (repeatability and reproducibility) was evaluated by determining the relative standard deviation of the peak area in five water samples fortified with 0.1 μg L^-1 of each compound. In every case, the pesticide and degradation product signal presented a coefficient of variation < 19%. The accuracy of the method was determined by evaluating the recovery percentage by modifying the type of sorbent and elution solvent and the sample volume that passed through the cartridge. The recovery percentages varied between 61 and 107 %, with a coefficient of variation between 3 and 18%.

Pesticide determination in real surface water samples

The method was applied to the determination of the studied compounds in different samples of surface water from the Cachapoal River in Central Chile. A percentage >87% of pesticides included in this work were detected in one or more of the sampled sites. Of the sixteen studied analytes, atrazine, DET, pyrimethanil, cyprodinil and diazinon were detected in most surface water samples, while fluometuron, acetochlor and CGA 92370 were not detected in any of the analyzed samples. At the end of the summer, pyrimethanil and cyprodinil were the analytes detected in the greatest concentrations, while in winter season, simazine and DET presented the highest concentrations (Table 4). Figure 4a shows that at the end of the summer, DET, pyrimethanil, cyprodinil and chlorpyrifos were detected in more than 15% of the samples in concentrations above the maximum individual limit established by the EC (0.1 μg L^-1). Regarding the concentrations of compounds detected in samples at winter period terbuthylazine, atrazine, DIA, DET and diazinon, were detected in more than 40% of the analyzed samples at concentrations over 0.1 μg L^-1 (Figure 4b). Of all of the sampled sites at the end of the summer, Las Cabras Canal presented the greatest level of pesticide contamination and of the samples collected in winter season, Tahuilla Canal presented the greatest total pesticide concentration. Finally, statistical analysis using the t student showed significant differences in the total concentration of pesticides for the sampled sites at the end of the summer and winter season (Table 5).

Figure 4 Distribution of samples collected at the end of the summer season (a) and at the winter season (b) according to the percentages of samples with not detected compounds, or with concentrations higher than LODs and above 0.1 μg L^-1. 

It is important to stress that in this work the monitored irrigation canals are located in the lower part of the basin (Figure 1), draining a large part of the fields located in the Puemo and Las Cabras area. In addition, the study area presents marked spatio-temporal variations in precipitation levels, with the most intense in the central valley of the basin. Finally, the tributaries of the Cachapoal River present mixed (La Cadena Creek and Claro River) and pluvial (Zamorano Creek) hydrological regimes, with highly variable daily streamflows and monthly averages (for example, 6.81 m³s^-1 in the Claro River and 18.16 m³s^-1 in Zamorano Creek); therefore, all of these characteristics could influence the quantity of pesticides and degradation products that enter the bodies of water in the winter and/or summer period.

Occurrence of herbicides and degradation products derived from triazines in surface water of Cachapoal river basin

Simazine and atrazine were the most ubiquitous herbicides in the sampling performed at the end of the summer period, and were detected in 83 and 100% of the samples, respectively. However, the concentrations did not surpass the limits of quantification of the method. In winter, there was an increase in the concentration of both compounds, with terbuthylazine also detected in 75% of the analyzed samples.

The degradation products DEA and DET were detected in all of the samples analyzed at the end of summer, although also in concentrations that did not surpass the limits of quantification. In the winter period, 75% of the samples presented a triazine-derived degradation product, with DET exhibiting the greatest concentration, with a maximum of 21.897 μg L^-1 in the Tahuilla Canal. In general, the sites sampled in winter presented a higher concentration of herbicides and degradation products derived from triazine than the sites sampled in summer, surpassing the individual limits (0.1 μg L^-1) for human consumption established by EC. This increase can be explained by the fact that some pesticides whose formulation contains simazine, terbuthylazine and/or atrazine are applied during the autumn or winter, coinciding with the date of the winter period sampling. These results are consistent with those reported by Dores, et al. [²⁵], which showed that in the Primavera do Leste region, Mato Grosso (central-western Brazil), there is an increase in the percentage of pesticides detected in bodies of surface water during the rainy seasons, with atrazine, simazine and DEA detected at the greatest concentrations. Likewise, studies carried out in Europe [²²] show that atrazine, simazine, terbuthylazine and their degradation products are the most frequently detected compounds in surface water in areas with intense agricultural activity.

Occurrence of fungicides in surface water of Cachapoal river basin

Unlike with pesticides, there is little research that accounts for the presence of fungicides in bodies of surface water in agricultural zones. However, studies carried out in Portugal, France, Germany, Australia and the United States have detected residues of pyrimethanil, carbendazim, metalaxyl, propiconazole, myclobutanil and tebuconazole in rivers and groundwater of basins with intense agricultural activity [²⁶]. In Spain, these substances have been detected in the soil and sediments in wine-growing areas [²⁷], as well as in bodies of surface water in various times of the year [²⁸].

In the present work, the most ubiquitous fungicides were pyrimethanil and cyprodinil, which were detected in all of the samples, reaching maximum concentrations of 0.937 μg L^-1 and 0.644 μg L^-1, respectively. These results are much less than those reported by Gregoire, et al. [²⁹], who evaluated the presence of seventeen pesticides over four years in a wine-growing catchment in France, detecting residues of pyrimethanil in all of the surface water samples collected in 2005, with a maximum concentration of 1.8 μg L^-1. Similarly, Wightwick, et al. [³⁰] reported the presence of fungicides in surface water of a catchment with intense horticultural activity in southeastern Australia, finding pyrimethanil in 16% of the analyzed samples; however, in no case was cyprodinil detected. Herrero-Hernández, et al. [²⁸] detected both compounds in three sites in La Rioja (Spain) over a year of sampling. The results show that, while they were not the most ubiquitous compounds, both were detected at a concentration above 0.1 μg L^-1, with pyrimethanil reaching a maximum of 0.590 μg L^-1 in September 2010 and cyprodinil a maximum of 0.981 μg L^-1 in September 2011.

Of all the studied sites in this work, Las Cabras Canal presented the greatest level of fungicide contamination at the end of the summer period and La Cadena Creek the greatest concentration in winter. In general, the total fungicide concentration was higher in the sites sampled at the end of summer, with ranges between 0.090 and 1.580 μg L^-1. This could be because spring and summer make up the main vegetable growing season, which is the period of greatest fungicide use. In addition, in vineyards it is common to apply fungicides as aerosol as a preventive measure; therefore, these substances can reach bodies of water through drift during the crop growing season, coinciding with the sampling period at the end of summer.

Occurrence of insecticides in surface water of Cachapoal River basin

Of the insecticides assessed in this work, diazinon, which was detected in all of the analyzed samples, was the most ubiquitous. This compound presented a greater concentration in the winter period, while in the summer period concentrations did not surpass the limit of quantification of the method (except Tahuilla Canal). Chlorpyrifos was detected in only 30% of the analyzed samples, with a maximum concentration of 0.183 μg L^-1 at the end of the summer. In general, the total insecticide concentration was similar in both sampling periods, which could be due to the heterogeneity of the crops in the basin, variation in the local topography and the physiochemical characteristics of the studied compounds. For example, the pesticides with elevated water solubility and a low tendency to be adsorbed in soil particles (300 < Koc <500) have a greater potential for movement to bodies of water through leaching or surface runoff than those that are less soluble or highly sorbed to soil (Koc > 500) [⁵]. In this sense, although diazinon presents a Koc similar to chlorpyrifos, its water solubility is sixty times greater, allowing this insecticide to enter bodies of water mostly in the dissolved phase in the winter period.

Unlike herbicides and fungicides, insecticides are usually present for a short time period in agriculture-affected surface water, making it crucial to perform monitoring with greater frequency in order to detect these substances. In spite of this limitation, studies have documented the presence of organochlorine, carbamate and organophosphate insecticides in Mediterranean bodies of water, noting that these compounds are generally found in low concentrations (ng L^-1) or detected infrequently [⁸, ³¹]. Finally, it is important to stress that the presence of insecticides in bodies of water constitutes a high risk to the biodiversity of these ecosystems, given that it has been demonstrated that elevated toxicity can have adverse effects on aquatic invertebrates in a short time period [³²–³³].