4.1 Overview of Pesticide Dissipation Processes in Soil and Water

The transport and distribution of pesticides in the environment are determined by numerous factors, including the physicochemical properties of the active compounds, the mechanisms of formulation, the rate of application, as well as local conditions such as topography, the soil and vegetation characteristics, and the hydro-climatic conditions. Pesticide behaviour in soil and water is determined by chemical, physical, and biological processes, such as sorption–desorption, volatilization, degradation, uptake by plants, and leaching, all of which contribute to pesticide dissipation, which includes degradative and non-degradative processes. The relative significance and cumulative effect of these processes vary according to both pesticide and matrix properties. Transformation of pesticides involves both biotic processes (e.g., microbial degradation) and abiotic processes (e.g., photochemical reactions). The contribution, rate, and extent of the transformation processes for a given pesticide are determined by its chemical structure and environmental conditions. While evaluating the specific contributions of pesticide transformation pathways is essential for risk assessment and decision-making on the catchment scale, robust approaches to characterize and constrain these processes pesticide transformation are mostly missing.

For instance, physicochemical properties of the soil (pH and SOM content, redox gradients in soils), biological properties (activity and distribution of microorganisms), and environmental conditions controlling soil temperature and moisture generally can affect both biotic or abiotic transformations. However, within soils biotic transformation of pesticides typically dominates over abiotic transformation, e.g., photochemical transformations which are typically restricted to only the top sub-millimetre layers of soil. In contrast, in shallow surface waters, phototransformation (with indirect photolysis being more relevant to pesticide degradation) can significantly contribute to pesticide degradation. For some pesticides, e.g., organophosphate pesticides, hydrolysis can also be a relevant abiotic reaction in the aqueous phase. However, for most pesticides, hydrolysis requires specific conditions such as high or low pH or low-redox environments. Given the multitude of processes influencing pesticide dissipation in the environment, accurate tracking of degradation as the primary process for removing parent compounds necessitates an approach grounded in multiple lines of evidence. This approach should encompass the integration of various complementary methods, which are briefly described below.

4.2 Evaluating Pesticide Transformation in Soil and Water: Current Approaches and Limitations

4.2.1 Pesticide Concentration Analysis

The rapid development of analytical instrumentation has enabled the detection and monitoring of multiple pesticide compounds in soil and water at low environmental concentrations (Alder et al. 2006; He and Aga 2019). Prior to analysis, different methods are commonly used to extract and concentrate pesticides from environmental samples, including solid-phase extraction (SPE) for water and quick, easy, cheap, effective, rugged, and safe procedure (QuEChERS) for soil (Anastassiades et al. 2003; Bonansea et al. 2013). Analyses by gas chromatography-mass spectrometry (GC–MS) or liquid chromatography-tandem mass spectrometry (LC–MS/MS) are the two most common methods for monitoring pesticide transformation in soil and water.

Pesticide concentration analysis is typically used in routine pesticide monitoring (e.g., by environmental agencies) and for examining in situ pesticide degradation. For instance, the concentration-based approach of monitoring 76 pesticide residues in 317 soil samples across the European Union revealed that over 80% of the soil samples contained pesticide residues (Silva et al. 2019). The fungicide boscalid was the most common pesticide molecule detected. Indeed, boscalid is very persistent in the soil matrix, with a dissipation half-life (DT50) ranging from 130 to 337 days (Karlsson et al. 2016). However, while monitoring approaches based on concentration analysis alone enable quantification of pesticide residues in the environment, they remain limited in distinguishing between physical and degradative processes, both of which can contribute to overall pesticide dissipation. Consequently, a comprehensive framework to examine pesticide transformation and dilution in the environment, and especially process-based information, is mostly lacking. Although a substantial volume of data on pesticide degradation has been gathered through regulatory testing, the capacity to anticipate and quantify the extent of degradation and the specific pathways involved in various field conditions still poses a considerable challenge (Fenner et al. 2013).

Incorporating an examination of not only the parent pesticide molecules but also their transformation products (TPs) offers a valuable approach to relate concentration monitoring and the degradation processes taking place in the field, thus enabling a more comprehensive identification and quantification of these in situ processes. Additionally, TPs may alter the ecotoxicological effects of pesticides on ecosystems. Through targeted and non-targeted methods, i.e., used to characterize chemical structures of unknown compounds, this complementary analysis is thus fundamental to evaluate the environmental fate and effects of pesticides. TPs analysis is easier when TPs are known, and more challenging when non-target analyses are used. Nevertheless, recent advances in the interpretation of the non-target analysis allow for the identification of previously unidentified TPs. For instance, suspect screening of 242 pesticide TPs combined with national monitoring data has recently identified previously undetected TPs, even in the absence of the parent compound (Menger et al. 2021). However, this approach introduces additional complexity, as TPs are liable to not only be produced but degraded as well in a natural environment. Hence, even when TP-based analyses are applied, quantifying degradation based on concentration alone is subject to large uncertainty. Concentration monitoring of both parent compounds and their TPs is frequently integrated with numerical modeling approaches to simulate the degradation of parent compounds and the subsequent production and breakdown of resulting TPs within a specific study site.

4.2.2 Modelling Approaches

Computational models can be used to predict variables related to pesticide persistence and mobility in soil and water. As a result, they provide a means to describe and predict the reactive transport of pesticides with various degrees of uncertainty based on the input parameters (Goumenou et al. 2021). Numerous studies have attempted to model the behaviour and distribution of pesticides in the environment, ranging from the micro-environment to the global scale. Different modelling approaches have been evaluated in several review papers (Mottes et al. 2014; Wang et al. 2019). Typically, in most models, degradation is one of the most sensitive model parameters determining the fate of pesticides in soil and losses to surface water and groundwater (Dubus et al. 2003). Accurately capturing the dynamics of pesticide and TPs degradation is therefore particularly critical in the context of risk assessment using fate models since this can introduce considerable uncertainty into model predictions (Lindahl et al. 2008). In addition, scaling up pesticide dissipation processes requires conceptual simplification of localized processes with a minimal loss of information (Imfeld et al. 2021). For instance, understanding the relationship between degradation rates and measured field parameters (e.g., microbial activity, temperature and soil moisture) allows for the application of pesticide fate models at larger scales (Ghafoor et al. 2011). Reasonable constraints on these field parameters can be made by iterating between model development, sampling, and measurement schemes. In recent years, the Soil and Water Assessment Tool (SWAT) (Arnold et al. 1993) has been proposed as a comprehensive modelling technique to relate processes affecting pesticides in sediment–water interfaces to catchments and river basins (Imfeld et al. 2021).

4.2.3 Biomolecular Analysis

The structural and functional characteristics of microbial communities are also potential indicators for monitoring and evaluating the degradation of a given pollutant in the environment. Following environmental DNA (eDNA) extraction, mainly quantitative polymerase chain reaction (qPCR) and functional gene arrays can be used to examine the response of microbial communities to exposure to pesticides and the potential of microbial population to adapt and degrade specific pesticides (Fenner et al. 2013). However, both techniques rely on knowledge of the biologically driven pesticide degradation pathway and gene sequence coding, which are not typically available (Imfeld and Vuilleumier 2012). In addition, detection and quantification of pesticide-degrading genes from environmental DNA generally feature the genetic potential for pesticide degradation but not its expression. In some cases, RNA transcripts may help follow up target active microorganisms involved in the degradation process (Monard et al. 2013). However, RNAs remain difficult to retrieve and conserve from complex environmental samples such as soil. In future, next-generation approaches may build on approaches involving specific degradation genes and their expression to also include patterns of microbial diversity using metagenomics and meta-transcriptomics (Jeffries et al. 2018; Rodríguez et al. 2020; Malla et al. 2022). This could be applied to both probe the effect of pesticides on soil or water microorganisms, as well as to evaluate the pesticide degradation functions.

Beyond these approaches, compound-specific stable isotope analysis (CSIA) already offers a complementary approach with untapped potential to evaluate the sources and transformations of pesticides in the environment.

4.2.4 Compound-Specific Stable Isotope Analysis (CSIA)

4.2.4.1 Principles

Degradation processes (biotic or abiotic) typically lead to changes in the ratio of stable isotopes (e.g., 13C/12C) in organic molecules. This change is related to the kinetic isotope effect (KIE) and depends on the preferential rate of bond breakage for weaker bonds (those with light isotopes, e.g., 12C for carbon) compared to bonds bearing a heavy isotope (13C for carbon). This may lead to an enrichment in the remaining contaminant in 13C relative to the initial isotope composition (Hunkeler et al. 1999; Meckenstock et al. 1999; Sherwood Lollar et al. 1999). The isotope effects produced during physical processes, such as volatilization, dissolution, dispersion, and sorption are typically smaller, usually insignificant (relative to analytical uncertainty) compared to the large KIE that may be produced by bond cleavage (Hunkeler et al. 2008; Kuntze et al. 2020 and reference therein). Therefore, change of 13C/12C isotope ratios during pesticide degradation can provide direct evidence of its degradation. The CSIA approach has been extensively applied over the past two decades to measure and quantify the degradation of legacy pollutants in industrial-contaminated sites. Industrial sites, unlike agricultural fields, are typically characterized by very high concentrations of the pollutants due to the presence of non-aqueous phase liquids (Hunkeler et al. 2008).

4.2.4.2 Application of CSIA

To quantify the magnitude of the isotope fractionation during the given degradation process, isotope fractionation factors (ε) need to be derived in the laboratory. To do so, change in the isotope signatures, e.g., δ(13C), can be related to the extent of degradation with the Rayleigh equation (Eq. 4.1):

$$\ln \left( {\frac{{\delta (^{13} {\text{C}})_{t} + 1}}{{\delta (^{13} {\text{C}})_{0} + 1}}} \right) = \varepsilon \times {\text{ln}}\left( f \right),$$
(4.1)

where δ(13C)0 and δ(13C)t represent the isotope signatures at time 0 and t of the degradation, respectively, whereas f is the fraction of the remaining pesticides at time t (Fig. 4.1a).

Fig. 4.1
Two scatterplots of the change in delta versus degradation. a. A curved curve denotes a stable carbon-isotope composition. b. A slanting curve in an increasing trend. y = a x + b.

a—Change of stable carbon-isotope compositions during degradation; b—Rayleigh equation for the same dataset

Full size image

Accurate determination of ε values is key to quantifying degradation processes in situ. From different forms of the Rayleigh distillation equation standard linear regression without forcing through zero is preferred (Scott et al. 2004) (Fig. 4.1b).

The extent of biodegradation (%) of an organic compound can be calculated in situ using the determined ε according to Eq. 4.2, without concentration data, eliminating the influence of physical processes.

$$B = \left( {1 - \left( {\frac{{\delta (^{13} {\text{C}})_{t} + 1}}{{\delta (^{13} {\text{C}})_{0} + 1}}} \right)^{{\frac{1}{{{\upvarepsilon }_{{{\text{bulk}}, {\text{C}}}} }}}} } \right) \times 100$$
(4.2)

For GC- and LC-isotope ratio mass spectrometry (IRMS) analysis, organic compounds are converted into a simple gas for which the isotope ratio is determined (e.g., CO2 for CSIA of carbon, N2 for CSIA of nitrogen). Due to the total conversion of the molecule to a measuring gas, position-specific changes in the isotope composition of a target compound cannot be ascertained. Moreover, the isotope fraction is diluted with any additional atom that is not directly involved in the reaction. In the case of large molecules due to the number of carbon atoms per molecule, a large extent of degradation is required to measure significant stable carbon-isotope fractionation (Δ(13C) > 2 ‰) to identify degradation in the field (Hunkeler et al. 2008). The isotope fractionation criterion for positive identification of in situ degradation of Δ(13C) > 2 ‰ is typically established based on total analytical uncertainty (± 0.5 ‰), incorporating accuracy and reproducibility (Sherwood Lollar et al. 2007), which may vary among compounds and elements analysed.

Even fractionation experiments under conditions controlled in the laboratory can be influenced by several phenomena, including additional rate-limiting steps, transport across the cell membrane (Renpenning et al. 2015; Ehrl et al. 2019), low substrate bioavailability (Sherwood Lollar et al. 2010), or substrate-enzyme binding (Mancini et al. 2006). These rate-limiting steps can “mask” the ε normally associated with the particular bond-cleavage involved. In some cases, this can lead to an underestimation of isotope fractionation, making it difficult to determine the appropriate ε to use for quantifying the biodegradation of contaminants.

This highlights the relevance of multi-element CSIA (ME-CSIA) to circumvent the limitation of only one element CSIA approach. In theory, all elements in the molecule are affected in a similar way by the masking processes mentioned above, eliminating the bias observed for ε.

4.2.4.3 Multi-element CSIA (ME-CSIA)

ME-CSIA can provide additional support for interpreting in situ results, specifically when complex pollution scenarios with multiple pollution sources or different degradation processes occur. Many pesticides contain nitrogen atoms (e.g., carbamates, acetanilides, thiabendazoles, triazoles), and therefore nitrogen CSIA in addition to carbon CSIA may help tracing pesticides degradation. In such a case, transformation mechanisms are identifiable from dual-isotope plots [e.g., δ(13C) versus, e.g., δ(15N)], reflecting underlying carbon- and nitrogen-isotope effects. For example, C-N dual-CSIA allows the identification of degradation mechanisms of six chloroacetanilide and acylalanine pesticides during abiotic hydrolysis (Masbou et al. 2018a). In ME-CSIA lambda (Λ) is the slope of the dual-isotope plot, and it reflects the changes in the stable isotope composition of one element versus the second element, which can be more specific to a reaction, and thus inform about transformation processes in the laboratory or in the field. The York method, which incorporates uncertainty in both variables (Ojeda et al. 2019), better adapts to the wide set of data conditions observed for dual-isotope data, and the natural logarithmic form is used for large isotope fractionation (Höhener and Imfeld 2021) (Fig. 4.2.).

Fig. 4.2
Two scatterplots. a. Delta S versus delta. b. I n versus in. Both denote a slanting line in an increasing trend. Raw delta values are transformed delta values.

Dual-isotope plot of a—raw Δ values according to (Ojeda et al. 2019) and b—ln-transformed δ values according to (Höhener and Imfeld 2021) for the same dataset

Full size image

Laboratory derived ε and Λ values for various pesticides and different degradation pathways can be found in recent review papers (Elsner and Imfeld 2016; Kuntze et al. 2020; Cui et al. 2021; Won et al. 2021). Nevertheless, CSIA datasets for pesticides during biotic and abiotic degradation processes are still currently limited, mainly due to analytical challenges (see below).

4.3 Compound-Specific Isotope Analysis (CSIA) of Pesticides in Soil and Water

4.3.1 Potential of Pesticide CSIA

Current research on pesticides focuses on toxicity, degradability, degradation pathways, and the formation of transformation products. However, novel management strategies relying on the natural and engineered degradation of pesticides and following the precautionary principle are warranted. Synthetic pesticides generally undergo degradation tests prior to use. The recurrent detection of herbicides in groundwater and soil emphasizes the difficulties in extrapolating laboratory tests to environmental conditions. Beyond the outcome of regulatory testing, current approaches are limited in their ability to (i) predict over relevant long time scales the degradation of micropollutants in the environment, (ii) evaluate, beyond environmental monitoring, the prevailing dissipation zones and periods (“hot/cold spots and moments”) in integrative studies on the catchment scale (Elsner and Imfeld 2016). Pesticides CSIA can provide additional assistance in this regard.

Although the analytical challenges currently limit field studies, including pesticide CSIA, applications in the environment are now within reach and can be foreseen in different scenarios of increasing complexity:

  • Point source scenario. The fate of organochlorine pesticides, such as hexachlorocyclohexane (HCH), has been evaluated in several contaminated field sites with point source pollution using CSIA (Bashir et al. 2015; Chartrand et al. 2015; Liu et al. 2017, 2021; Wu et al. 2018, 2019a; Qian et al. 2019). In analogy to steady-state assessments of legacy contaminants, continuous pesticide release from a single point source related, for example, to pesticide spills during production, can be studied.

  • Event-based scenario. Usually, pesticide pollution in agricultural fields does not conform to single point source scenarios. Instead, it often manifests as pesticide pulses, which can be traced in event-based studies to elucidate the dynamics of “hotspots” and “hot moments” (e.g., Riml et al. 2013). The isotope data of pesticides in surface water may enhance the interpretation of transformation pathways in heterogeneous reactive compartments. The first field studies have recently been reported (Schreglmann et al. 2013; Alvarez-Zaldivar et al. 2018).

  • Large-scale studies. Comprehensive studies on the catchment scale may also integrate multiple pesticide sources and events over multiple seasons (Moschet et al. 2013). Pesticide CSIA may thus also serve comprehensive studies on a larger spatial scale with multiple sources and events over a season but may miss where, when, and how degradation occurs. For example, the stable carbon-isotope signature was used to evidence the migration and degradation of DDTs in a large-scale study of arable soils across China (Niu et al. 2016). In this respect, although multi-element CSIA may also be desirable to further improve characterization, its feasibility is challenged by the sampling window, where quantification limits can be achieved.

Additionally, the combination of passive sampling and pesticides CSIA opens novel opportunities. Passive sampling techniques have been widely used to integrate mean concentrations of organic contaminants over extended sampling periods (< 50 days) and to pre-concentrate micropollutants, allowing their detection in trace amounts in surface waters (10–50 pg/L). A passive sampling approach combined with CSIA has been previously applied for vapour intrusion studies of legacy contaminants, such as benzene (Goli et al. 2017), and has recently been tested for pesticide in wetlands in agricultural fields (Gilevska et al. 2022). δ(13C) and δ(15N) of pesticides in Polar Organic Chemical Integrative Sampler (POCIS) deployed in surface water would reflect their degradation in both water and hydrologically connected soil. Therefore, when compared to the initial isotope signature of the pesticide formulation applied, isotope signatures from POCIS can indicate degradation in both water and soil in the entire agricultural catchment.

Complementarily, pesticide chirality can provide yet another tracer to understand the fate of chiral pesticides in the environment. One-third of marketed pesticides are chiral, and most of them are used as racemates, despite the fact that the desired activity usually depends on one enantiomer, while the other(s) has adverse environmental effects (Maia et al. 2017). As many degradation processes are enantioselective, monitoring changes in the enantiomeric fraction (EF) can be used to follow degradation in situ. Furthermore, combining CSIA and enantioselective analysis techniques for the enantioselective stable isotope analysis (ESIA) may enhance the evaluation of sources and transformation processes of individual pesticide enantiomers in the environment (Milosevic et al. 2013; Elsner and Imfeld 2016; Jin and Rolle 2016; Masbou et al. 2018b, 2023). For example, combined evidence from concentrations, enantiomer ratios, and isotope composition confirmed the degradation of the herbicide dichlorprop and its metabolite in the hotspot at the contaminated site in Denmark (Milosevic et al. 2013).

4.3.2 Challenges of Pesticide CSIA

The occurrence of very low (sub-µg/L) concentrations of pesticides and their polarity are two major analytical challenges to be addressed to expand CSIA approaches to pesticides. First, the enrichment of sufficient analyte (typically a few ng of C or N on column per injection are needed) requires the extraction of large amounts of water, soil, or vegetal material. The detection limits for nitrogen CSIA of pesticides are much higher than for carbon, as typically there are fewer atoms of N than C in the pesticide molecule, making it difficult to apply ME-CSIA in situ (Elsner and Imfeld 2016). Any scaling up of the extraction method should be monitored for potential isotope effects. As shown by Melsbach et al. (2021), increasing the volume of water during SPE above 10 L may alter the δ(13C) of atrazine.

For accurate and precise CSIA, complete chromatographic separation of all compounds is required. Therefore, any chromatographic interferences can strongly limit the applicability of the CSIA. As the majority of pesticides are non-volatile or semi-volatile on-column liquid injection is typically used as a sample introduction technique when pesticides are analysed with GC-IRMS. This technique, in contrast to static or dynamic headspace sampling used for volatile organic compounds, can introduce many matrix interferences from compounds with similar physicochemical properties from soil or water co-enriched during extraction procedures. For example, during large-volume SPE, non-volatile matrix components from environmental samples are thus enriched together with the target compounds, which is defined here as the matrix effect. Extract clean-up procedures are thus often necessary prior to the application of CSIA to environmental samples (see Sect. 4.4.2.3).

The second challenge lies in the polarity of the more polar pesticides and their transformation products. This generally requires the use of a derivatization step prior to GC separation or the use of LC-IRMS, which is currently limited to δ(13C) analyses. LC-IRMS is further constricted due to high detection limits (at least one order of magnitude higher than GC-IRMS). LC-IRMS has also limited application due to the incompatibility of the method with organic solvents, such as methanol, typically used for chromatographic separation (Gilevska et al. 2014; Perini and Bontempo 2021). Therefore, the most common derivatization step is chosen to circumvent the polarity of the parent compound or daughter products. The choice of the derivatization method strongly depends on the chemical structure of the pesticide. Methylation of hydroxyl and amino groups is typically achieved with trimethylsilyldiazomethane (TMSD) and trimethylsulfonium hydroxide (TMSH) (Reinnicke et al. 2010; Mogusu et al. 2015; Melsbach et al. 2019; Torrentó et al. 2019). Any additional step, such as derivatization, requires strict screening for any isotope fractionation during the sample preparation. Additionally, the stable carbon-isotope composition should be corrected if additional carbon atoms are introduced during derivatization.

4.4 Soil and Water Sample Preparation and Measurements with GC-IRMS for Stable Carbon- and Nitrogen-Isotope Composition of Pesticides

4.4.1 Sampling Strategy

The following sampling strategy is suggested to quantify pesticide degradation extent under field conditions in small agricultural catchments (i.e., 10–100 ha). Both sampling and interpretation should be adapted for larger scales to include multiple sources and events over an agricultural season.

First, sufficient knowledge regarding hydrological and hydro-climatic conditions and functioning is mandatory for the application of pesticide CSIA at the catchment scale (Table 4.1). Hydro-climatic data should provide sufficient resolution to evaluate (i) mean daily rainfall, (ii) mean rainfall intensity, (iii) total rainfall, (iv) mean daily reference evapotranspiration, (v) mean daily temperature, (vi) mean daily discharge normalized by the total catchment area, (vii) time of concentration, and (viii) proportion of days in a month when rainfall occurred (% Wet Days). Subsurface travel time should be defined precisely, possibly with preliminary hydrological studies using the stable hydrogen and oxygen isotope composition of water [δ(2H) and δ(18O)], refer to Sect. 6.3.1.

Table 4.1 Typical sampling scheme for water and soil samples for implementing pesticide CSIA on the catchment scale
Full size table

The soil sampling frequency should be adapted to the pesticide application and the expected degradation kinetics. Transects should be selected to account for soil type, heterogeneity, and the variability of moisture conditions, drainage characteristics and to maximize the number of plots where the pesticide is applied. In addition, a digital elevation model (DEM) may be used to determine local slopes and to estimate the topographical wetness index (TWI) [-]. TWI mainly quantifies the impact of topography on soil moisture. Soil crust development should be characterized across the catchment after a precipitation event as a function of mm of cumulative rainfall. This should allow us to evaluate the reduction in the soil infiltration capacity due to crusting, and to interpret the temporal evolution of TWI along with rainfall-runoff data.

During sampling, the water discharge at the catchment outlet should be continuously monitored to evaluate hydrological functioning and to establish water and pesticide mass balances. After sampling, the sub-samples of water may be pooled into composite samples according to hydrograph characteristics and pesticide concentrations, yielding one or more samples weekly with the chosen water volume. This volume should be selected to ensure that the concentrations of each pesticide are sufficient for CSIA analysis. An example of the sample size and the potential for pesticide and isotope analysis is illustrated in Fig. 4.3.

Fig. 4.3
A graph of concentration versus volume. The multi-line depicts carbon C S I A, carbon and nitrogen C S I A, and concentration analysis. The maximum volume typically collected for pesticide C S I A marked as a vertical line at 10.

Relationship between the necessary S-Metolachlor concentrations in water and collected water volume for reliable carbon or nitrogen CSIA and concentration analysis (Alvarez-Zaldivar et al. 2018; Torrentó et al. 2019). The solid black line represents feasibility of carbon CSIA, solid grey line represents feasibility of carbon and nitrogen CSIA, dashed line represents feasibility of concentration analysis. Note a different scale for the concentration analysis. Note that this relationship is site-specific

Full size image

4.4.2 Water and Soil Sample Processing and Conservation

Water samples should be collected using a refrigerated autosampler in the field, stored in the dark at 4 °C during collection (to avoid photolysis and limit further biodegradation), and placed on ice during transportation to the laboratory for immediate filtering (on 0.7 or 0.45 µm glass fibre). Water samples should be kept at 4 °C from the collection up to the extraction and should be preferably further filtered and analysed (using SPE, see Sect. 4.4.3) within 24 h. Preliminary tests on the effect of water sample sampling, transport, and storage should be done prior to the study. Water samples may also be frozen for longer-term conservation, but preliminary tests of the effect of freezing on both extraction yield and stable isotope fraction are required for each targeted compound.

After collection, soil samples should be kept in an ice box during transport to the laboratory and kept frozen at −20 °C until analysis. Soils should be homogenized, quartered before sieving (e.g., according to NF X 31,100 standard), and sieved at 2 mm. Water content, pH, organic content, CEC, and other parameters can be measured to characterize soil samples and help in data interpretation.

4.4.3 Extraction Methods for CSIA from Water and Soil

To expand the use of pesticide CSIA to agricultural catchments, it is crucial to use an appropriate extraction method. Extraction methods for pesticide residues from environment matrices for ME-CSIA should: (i) provide sufficient analyte mass for reliable isotope analysis, (ii) cause no isotope fractionation [Δ(HE)], (iii) be applicable to a wide range of pesticides and matrices, (iv) limit matrix co-enrichment to avoid co-elution during chromatographic separation.

4.4.3.1 Extraction Methods from Water

To date, SPE has been the most common method used to extract water samples for pesticide CSIA although liquid–liquid extraction has been used for small amounts of water (from 0.5 mL) in laboratory experiments (Chevallier et al. 2018; Knossow et al. 2020). The extraction methods should be tested within the expected concentration range and target environmental matrix to determine the feasibility of pesticide CSIA from water samples. SPE has been tested previously in combination with CSIA of atrazine, acetochlor, S-Metolachlor, metalaxyl, butachlor, alachlor, terbutryn, chlordizon, bentazone, dichlorvos, dimethoate, omethoate and several of their metabolites (Schreglmann et al. 2013; Elsayed et al. 2014; Wu et al. 2014; Schürner et al. 2016; Masbou et al. 2018a; Torrentó et al. 2019; Drouin 2021; Droz 2021; Pérez-Rodríguez et al. 2021). To ensure maximum recovery, the type and quantity of sorbents and extraction eluents should be adjusted to the physical properties of the analytes and matrix. As the physicochemical properties of parent compounds and their TPs may differ significantly, different sorbents or sorbent combinations may be used to pre-concentrate the parent compounds and the TPs (Torrentó et al. 2019).

The majority of pesticide CSIAs studied did not show isotope fractionation using SPE (Schreglmann et al. 2013; Elsayed et al. 2014; Wu et al. 2014; Masbou et al. 2018a; Torrentó et al. 2019; Droz 2021; Pérez-Rodríguez et al. 2021). A sample size of up to 10 L is more frequently used for SPE and pesticide CSIA, as increasing the volume > 10 L could change the carbon isotope ratios δ(13C) of pesticides (Melsbach et al. 2021). However, this change may not be correlated with the SPE procedure but rather with the matrix effect on the measurement, decreasing oxidation capacity, and increasing background levels and instrument maintenance issues. Therefore, prior to the pesticide measurements, clean-up strategies (discussed below) are typically applied to minimize the matrix effect. Nevertheless, current SPE methods allowed carbon and nitrogen pesticide CSIA in the ng/L to µg/L of pesticide concentration range (Schreglmann et al. 2013; Alvarez-Zaldivar et al. 2018; Torrentó et al. 2019). This underscores the feasibility of carbon and nitrogen CSIA from water samples in agricultural settings throughout the agricultural season.

4.4.3.2 Extraction Methods from Soil and River Sediment

When choosing an extraction method and solvent, one should take into account the physicochemical properties of pesticides, including their hydrophobicity and acid dissociation constant, as well as soil properties like pH, organic matter content, and moisture levels. For optimal results, extraction tests must be carried out with the studied soil or sediment to ensure high recovery and non-significant isotope fractionation and evaluate the matrix effect due to the co-extraction of soil organic matter. Modifications to existing methods and protocols, such as increasing the sample volume from 5 to 20 g or sequential use of extraction solvents, should also be evaluated.

Pesticide extraction methods from soil and sediment for reliable pesticide CSIA have been already used in both laboratory and field studies (Alvarez-Zaldivar et al. 2018; Masbou et al. 2018b; Wu et al. 2019b; Pérez-Rodríguez et al. 2021). Ivdra et al. (2014) proposed a modified ultrasonic-assisted extraction (MUSE) without carbon-isotope fractionation associated with extraction [Δ(13C) ≤ 0.4) for hexachlorocyclohexanes (HCHs). A modified MUSE method (Ivdra et al. 2014) was also tested with ethylacetate (Alvarez-Zaldivar et al. 2018; Masbou et al. 2018b) or dichloromethane:pentane (Droz et al. 2021; Pérez-Rodríguez et al. 2021) as the extraction solvent. Another study used accelerated solvent extraction (ASE) for the extraction of HCHs from soil and plants, which enabled stable C, H, and Cl isotope analysis (Wu et al. 2019b; Liu et al. 2020, 2021). The application of QuEChERS for the extraction of metolachlor from two agricultural soils led to an insignificant (< 1 ‰) isotope fractionation for carbon CSIA (Torrentó et al. 2021).

Current methods allow for carbon and nitrogen CSIA in soil samples in the range of ng/g to µg/g range for carbon and several µg/g for nitrogen CSIA (Alvarez-Zaldivar et al. 2018; Masbou et al. 2018b; Droz 2021). This currently restricts the application of carbon and nitrogen CSIA of pesticides to source areas and laboratory studies, respectively. In order to reduce concentration ranges for ME-CSIA, there is a need for simple and fractionation-free purification techniques that can be applied to a wide spectrum of environmental soils and pesticides.

4.4.3.3 Clean-Up Procedures

A number of clean-up procedures can be applied to address the issue of co-enrichment to maximize the analytical performance of pesticide CSIA extraction from environmental matrices without altering the isotope ratio of the target compounds. These include: (i) the addition of a sorbent, such as primary secondary amine or graphitized carbon black, to remove pigments, such as chlorophyll (Anastassiades et al. 2003; Wilkowska and Biziuk 2011), (ii) chromatography HPLC separation or column chromatography (Schreglmann et al. 2013; Mogusu 2016), and (iii) the use of molecularly imprinted polymers (MIP) (Bakkour et al. 2018). MIP is likely the most effective clean-up method for CSIA. However, MIP is not commercially available for all classes of compounds and therefore must be specifically synthesized prior to clean-up. Furthermore, samples should only be processed in glass when using organic solvents, especially dichloromethane, to reduce matrix interferences from the extraction procedure.

Carbon and nitrogen CSIA in complex matrices can also benefit from two-dimensional gas chromatography (GC × GC)-IRMS. This approach has been applied to polychlorinated biphenyls, chloronaphthalenes, and chlorofluorocarbons (Horii et al. 2005; Horst et al. 2015). In GC × GC-IRMS, the system needs to be equipped with a column-switching device, such as moving capillary stream switching (MCSS), or with a 6-port valve. By column-switching parts of the effluent from the first column are cut and transferred to a second column, where separation of compounds of interest can be enhanced. In addition, the use of (GC × GC)-IRMS can improve sensitivity by eightfold (Horst et al. 2015) by enhancing chromatographic separation. Therefore, this approach has potential for pesticide CSIA application.

Whenever possible, one or several clean-up strategies, depending on the complexity of the matrix, should be applied during pesticide CSIA. Along with improving the background of pesticide CSIA, it will reduce the need for increased maintenance of the GC-IRMS instrument, including oven and column replacement as well as blockage of capillaries. It is also possible to cut the first 10–20 cm of the GC column if some small chromatographic interferences are observed in blanks preventing retention of the entire column (be aware of the change in retention time and the measurement windows). Complementarily with clean-up procedures, it may be possible to extend the temperature programme to improve the level of background and chromatographic separation when the background compromises the GC-IRMS measurements. However, this will increase the measurement times.

It is worth mentioning that pesticide CSIA must be preceded by quantitative evaluation of the pesticides in the extracts to optimize injections of the target analytes. Prior CSIA, rigorous quality assurance practices and referencing strategies must be established to ensure that isotope measurements are accurate and reproducible.

4.4.4 Optimal GC-IRMS Conditions for Pesticide CSIA

4.4.4.1 Referencing Strategies

An important component of quality assurance is the use of analytical standards of pesticides calibrated on international isotope scales. The referencing strategy of the “identical treatment principle” (Werner and Brand 2001) should be implemented: (i) to measure external standards before and after the sample to correct for an offset, and (ii) to use a reference material that is identical to the target substance. To obtain in-house isotopic standards, a mass spectrometer equipped with an elemental analyzer and isotope ratio mass spectrometer (e.g., Flash EA IsoLinkTM CN IRMS, Thermo Fisher Scientific, Bremen, Germany) or offline conversion followed by dual-inlet-IRMS are typically used. The δ(13C) and δ(15N) values should be normalized to VPDB and Air scales, respectively, using a minimum of two international reference materials with the range of isotope δ values, that would encompass all the measured samples, are chosen as an anchor point for the regression line, e.g., IAEA600 [δ(13C) = – 27.77 ‰] and USGS41 [δ(13C) =  + 37.63 ‰] (Coplen et al. 2006). Then, the third standard, which has the value between the chosen anchors is treated as an unknown sample for quality control (QC), and is used to evaluate combined analytical uncertainty (Coplen 1988). As a part of inter-laboratory comparison, it is recommended to measure the same standards at other lab facilities using another EA-IRMS.

Unless otherwise specified, the uncertainties for pesticide CSIA are reported as standard deviation (1σ) calculated from replicate measurements. The combined analytical uncertainty of δ(13C) and δ(15N) values should not differ (≤ 0.5 ‰) from values obtained by EA-IRMS.

The standard injection frequency depends on the pesticides, the element measured, the matrix, and the number of pesticides analysed. For relatively simple environmental matrices (e.g., groundwater), it is recommended to measure an in-house pesticide mix standard with known isotope composition at least every six samples to control the retention time of the target compounds and assess the instrument performance (e.g., conversion efficiency). The standard mixtures are injected every three samples, however, for simultaneous δ(15N) analysis of several pesticides for complex environmental matrix (e.g., soil). Peak amplitudes of injected standards should closely match those of analysed samples (Sherwood Lollar et al. 2007). To conserve the combustion capacity of the oven, the measurement window (effluent mode) should be kept to a minimum: opened before the targeted peak elution and closed shortly thereafter.

As part of CSIA method validation, the linearity range should be determined for each pesticide. The linearity range specifies the range of measurements with sufficient precision and accuracy, indicating, within an acceptable range [e.g., ± 0.5 ‰ for δ(13C)], that the stable isotope composition is independent of the amount of compound injected (Jochmann et al. 2006). The method detection limit (MDL) is the point with the lowest (or highest) concentration within a ± 0.5 ‰ linear interval of the mean value for the standard measured with GC-IRMS and with good reproducibility in triplicate measurements (< 0.5 ‰). In Fig. 4.4a shows the linearity for δ(13C) for pesticide tebuconazole, with three data points (filled circles) outside ± 0.5 ‰ linear interval from the mean value or with low reproducibility of triplicate measurements (> 0.5 ‰). Such measurements are not taken into consideration and are considered to be outside the linear range. Figure 4.4b shows the linearity for δ(15N) for terbutryn, with MDL corresponding to a significantly higher concentration of the compound in the sample.

Fig. 4.4
Two scatterplots of delta versus concentration. The title reads linearity for tebuconazole and linearity for terbutryn. It depicts M D L 3.6 milligrams per liter 200 millivolts and M D L 100 milligrams per 314 millivolts.

Concentration and A—δ(13C) measurements and amplitude of the mass 44 for tebuconazole B—δ(15N) measurements and amplitude of the mass 28 for terbutryn. Circles represent stable carbon- and nitrogen-isotope compositions. Triangles indicate the amplitude of mass 44 and 28 peaks. The solid line represents the calculated mean δ(13C) value (– 29.9 ± 0.2 ‰, n = 51) and δ(15N) value (– 2.3 ± 0.2 ‰, n = 30); dashed lines indicate the ± 0.5 ‰ interval. Measurements were performed in triplicate; the standard deviation of each point is indicated by error bars. The dotted line represents the δ(13C) value of tebuconazole (– 29.7 ± 0.1 ‰, n = 3) and δ(15N) value of terbutryn (– 2.8 ± 0.1 ‰, n = 3) measured by EA-IRMS. Values outside the linear range-filled circles are excluded from the mean δ(13C) and δ(15N) value calculation due to either being outside ± 0.5 ‰ interval from the mean value or due to low reproducibility of triplicate measurement (> 0.5 ‰). MDL—method detection limit. The major principles illustrated in this figure are described in (Jochmann et al. 2006)

Full size image

4.4.4.2 Conversion to gas

Organic compounds are converted to a simple measuring gas CO2 for carbon CSIA and N2 for nitrogen CSIA. For both carbon and nitrogen CSIA, the conversion can be achieved in a combustion reactor comprised of, for example, NiO tube, CuO, NiO, and Pt wires and operated at 1000 °C (P/N 1255321, GC IsoLink II IRMS System, Thermo Fisher Scientific). Within the combustion oven, two reactions occur for nitrogen (some setups use two separate units for oxidation and reduction) (Eq. 4.3).

$${\text{R}} - {\text{NH}}_{2} \to ^{{{\text{Oxidation}}}} {\text{N}}_{2} \mathop \leftarrow \limits^{{{\text{Reduction}}}} {\text{N}}_{x} {\text{O}}_{y}$$
(4.3)

During nitrogen CSIA, the produced CO2 is trapped downstream using a liquid nitrogen trap to reduce measurement interference. Pesticides containing heteroatoms, such as N and Cl, and more complex chemical structures pose challenges to the combustion process, resulting in incomplete conversion and lower sensitivity. CSIA of nitrogen is particularly challenging, since the heavier isotope 15N has a lower natural abundance (0.04%) than carbon heavier isotope 13C (1.1%), and fewer heavy nitrogen atoms are present in organic substances. Moreover, two N atoms are required to form N2, and N2 has a lower ionization efficiency. To obtain a similar precision to that of δ(13C) analysis, theoretically a 50 times higher sample amount is required for δ(15N) analysis, leading to high substance loads that can affect conversion efficiency (Reinnicke et al. 2012). To enhance combustion during pesticide CSIA, different modifications were proposed, combining changes in reactor design, reactor temperature, and oxidation pattern. The variation of these parameters can be optimized if a large offset or low reproducibility is observed during carbon or nitrogen pesticide CSIA. A self-made Ni/NiO reactor operated at 1150 °C (Meyer et al. 2008), a NiO tube/CuO–NiO reactor operated at 940 °C (Reinnicke et al. 2012), a Ni/Ni/Pt reactor operated at 1000 °C (Spahr et al. 2013) have been used to enhance conversion of different pesticides during carbon and nitrogen CSIA.

During the oxidation procedure, oxygen is pumped into the reactor in backflash mode, replenishing the oxygen required for conversion. For nitrogen, excess oxygen might lead to incomplete reduction of NxOy species into N2. The oxidation time should be adjusted to achieve a balance between sufficient oxygen for the quantitate conversion but not too much, so that reduction is not hindered.

4.5 Towards Field Studies Using Pesticide CSIA and Interpretation of Stable Isotope Signatures

Integrative strategies relying on isotope data have been conducted for more than three decades in the case of nitrate and have significantly contributed to the development of water management policies (Nestler et al. 2011; Lutz et al. 2013; Moschet et al. 2013). A similar strategy has been developed for legacy pollutants (Hunkeler et al. 2008) and may be now developed for pesticides in agricultural catchments, as recently described (Alvarez-Zaldivar et al. 2018).

Prior to the sampling campaign, which would include pesticide CSIA information about applied pesticides, application period, published ε and Λ values for pesticides used, the isotope signature of the pesticide formulation should be gathered.

4.5.1 Insights from Pesticide CSIA

4.5.1.1 Occurrence of in Situ Degradation and Pathway Identification

When Δ(13C) or Δ(15N) > 2 ‰ (assuming combined analytical uncertainty of ± 0.5 ‰) of pesticide molecules is observed across space, time, or source (i.e., compared to the isotope signatures of pesticides in applied commercial formulations), pesticide degradation is likely occurring in situ. Further, the degradation pathways may be identified based on the determined ε and Λ values. For example, in the biotic hydrolysis of atrazine by Arthrobacter aurescens TC1, an unusual trend towards more negative δ(15N) values is observed. Protonation of nitrogen atoms during the reaction makes 14N react more rapidly while 15N accumulates in the remaining atrazine (inverse isotope effect) (Meyer et al. 2009). In contrast, oxidative dealkylation by Rhodococcus sp. strain NI86/21 would result in more positive δ(15N) values. As mentioned in Sect. 4.3, the use of dual element isotope plots is preferred, especially when multiple sources are present.

4.5.1.2 Quantitative Assessment of in Situ Degradation

When the degradation pathway is identified in the field, then Eq. 4.2 can be used to calculate the extent of biodegradation. For pesticides, these calculations typically require a large extent of degradation to fulfil Δ(13C) > 2 ‰ requirement. For example, for S-Metolachlor (using εC = − 1.5 ± 0.5 ‰ for biodegradation in soil (Alvarez-Zaldivar et al. 2018; Droz et al. 2021), the extent of degradation of the applied pesticide should be higher than 74% to apply carbon CSIA to identify and quantify degradation in situ. However, owing to the multitude of processes influencing pesticide residues in agricultural fields, such as contamination pulses during runoff events, these estimates of the extent of biodegradation are likely to be conservative. Therefore, a subsurface-surface reactive transport model incorporating CSIA isotope source ratios, ε values for different elements can help examine sources and dissipation of pulses of diffuse contaminants (Van Breukelen 2007; Lutz et al. 2017).

4.5.2 Pesticide Database for Source Identification and Apportionment

The initial stable isotope composition of organic chemicals depends on the conditions and the pathways used to synthesize the compound and thus depends on the manufacturers and the time frame of production. This variation may serve to identify chemical sources or trace the time of contaminant release in the environment. The range of the initial isotopic ratios of these compounds from various manufacturers is thus needed to trace pesticide origin and dissipation.

Therefore, prior to applying pesticide CSIA in the field, the isotopic composition of commonly or formerly used pesticides produced by different manufacturers from different regions should be determined for different elements. Relevant pesticide commercial formulations (1–5 mL in clean glass vials stored at 4 °C or at − 20 °C until analysis) can be collected directly at the farms or following official requests to regional and national pesticide providers and manufacturers. The active compounds of the different pesticide formulations can be isolated by either liquid–liquid extraction or measured directly by GC-IRMS. For example, Table 4.2. shows the narrow range of δ(13C) of S-Metolachlor for four different pesticide formulations (from − 31.7 ± 0.2 ‰ to − 32.6 ± 0.5 ‰) and slightly different composition of an analytical standard (− 31.0 ± 0.5 ‰). Based on that range, if the δ(13C) of S-Metolachlor measure in situ is around − 29.0 ‰, it may indicate degradation.

Table 4.2 Range of δ(13C) of S-Metolachlor for a standard (PESTANAL®) and four different pesticide formulation measured by GC-IRMS
Full size table

The ISOTOPEST database (https://ites.unistra.fr/isotopest) was established to address the need for a comprehensive repository of stable isotope signatures of pesticides in commercial formulations. Its primary purpose is to enable seamless inter-laboratory comparisons and foster the utilization of stable isotope data in forthcoming research endeavors concerning pesticide transformation and source tracking in the environment.

4.5.3 Identifying Isotope Fractionation During Degradation in Microcosm Experiments

Degradation laboratory experiments under controlled conditions must clearly demonstrate the relationship between concentrations decrease and changes of stable isotope composition, and identify the prevailing degradation pathway in specific environmental compartments (Hunkeler et al. 2008). Several studies have determined the range of ε and Λ values for various pesticides (Masbou et al. 2018a; Cui et al. 2021; Drouin et al. 2021; Torrentó et al. 2021; Won et al. 2021). However, with the increasing number of commercial formulations, it is likely that specific information concerning isotope fractionation during degradation may not be available for most pesticides. Additionally, field-specific information is preferred for the most accurate identification and quantification of the degradation process. Reference hydrolysis, photolytic, or biodegradation experiments under laboratory-controlled conditions can be carried out to derive field-specific ε and Λ values for pesticides from water, sediment, or soil samples. It is possible that photodegradation may be a relevant process in the first soil mm or water surface (Cui et al. 2021; Drouin et al. 2021). Reference degradation experiments may ideally be carried out with an isolated strain degrading metabolically the targeted pesticide, together with the identification of TPs and targeted microbial assessment (i.e., specific functional genes and/or taxa associated with degradation) to characterize degradation in the field using the multi-line-of-evidence approach.

4.5.4 Outlook

The requirements, bottlenecks, and possible solutions at various scales and levels of complexity to apply pesticide CSIA to field-based studies are summarized in Table 4.3. In most cases, field applications have focused on point source scenarios of persistent pesticides, which typically have higher concentrations and therefore require smaller sample sizes and present fewer measurement challenges (Hunkeler et al. 2008). Degradation is typically assessed by comparing the source isotope signature over time and space. Such applications can be combined with ME-CSIA, ESIA, and microbial analysis to identify the prevailing degradation processes (Bashir et al. 2015; Chartrand et al. 2015; Liu et al. 2017, 2021; Wu et al. 2018, 2019a; Qian et al. 2019). Additionally, the implementation of high-resolution CSIA approaches can provide further constraints on characterizing the dynamics of active degradation zones and compartments (e.g., sediment–water interfaces or soil surface) (Gilevska et al. 2021).

Table 4.3 Requirements, bottlenecks, and possible solution of pesticide CSIA application for different scenarios
Full size table

For pesticide CSIA application to event-based scenarios, a detailed sampling strategy must be developed based on possible sources, hydrology, and land use (see Sect. 4.4.1). Additionally, isotope data can be interpreted within a framework that includes physicochemical and hydrological tracer data to evaluate and quantify pollution sources and the processes that the pesticide has undergone during its retention and transport in a catchment (Elsner and Imfeld 2016). In order to monitor the degradation of pesticides effectively, it is imperative to conduct sampling prior to the pesticide application period. This initial sampling serves the crucial purpose of quantifying background concentrations and isotopic signatures of residual pesticides. Establishing this accurate baseline is essential as it provides a reference point for the assessment of both short-term and long-term impacts following pesticide application. Given that diffuse pesticide pollution typically leads to low environmental concentrations, the collection of substantial and representative volumes of water and soil becomes imperative for the reliable implementation of pesticide CSIA. This need for comprehensive sampling is particularly significant in cases where nitrogen isotopic analysis is involved. In addition, a monitoring strategy incorporating this complementary analysis requires narrower sampling time frames to ensure sufficient sampling resolution to capture short-duration high concentration episodes, such as large runoff events. In future, integrative sampling with passive samplers (e.g., POCIS) can be implemented to ensure that such events will be sampled at sufficiently high resolution. It can be anticipated that the continued improvement of analytical methods for clean-up and extraction for pesticide CSIA will prove invaluable in this application.

The application of CSIA to extensive case studies presents a significant challenge, primarily due to the presence of multiple sources of pesticides, diverse application events, and the unique characteristics of individual sites. To confront this challenge effectively, the catchment areas under investigation must be subdivided in a manner that aligns with their specific land use and water flow directions. This approach enables the utilization of a series of mixing models tailored to the respective subareas, providing a more nuanced and accurate understanding of pesticide dynamics within these complex systems (see Chap. 2). However, in the case of persistent and well-known pesticides, CSIA may offer a unique opportunity to study their long-term degradation, e.g., DDTs (Niu et al. 2016) or chlordecone. Similarly, CSIA holds substantial promise in the investigation of persistent pesticides, like atrazine, which continue to be detected in environmental field sites even decades after their prohibition in Europe (Elsner and Imfeld 2016).

Pesticide application and subsequent pollution remain a major and long-lasting environmental issue that will continue throughout the twenty-first century. The application of pesticide CSIA presents a unique opportunity to identify and quantify pesticide degradation in situ in agricultural catchments, although it requires a careful analysis of its feasibility, a robust sampling design, and a joint interpretation with hydro-climatic, geochemical, and microbial conditions and pesticide concentrations.