Identification of pesticides transformation products in a vineyard watershed by coupled targeted, enhanced suspect, and non-target screening

Overview & Synopsis

The past decade has brought a more holistic view of watershed contamination by organic chemicals via the deployment of powerful analytical strategies such as suspect and non-target analysis. Aided by Liquid Chromatography hyphenated to High Resolution Mass Spectrometry (LC-HRMS), these analyses allow the search of a larger number of contaminants as well as their transformation products (TPs) in environmental matrices. In France, vines are among the crops that use the most pesticides in relation to the cultivated agricultural areas. These regular inputs lead to long-term contamination of vineyard watersheds by pesticides and their TPs, thus jeopardizing local aquatic ecosystems. However, knowledge about TP pesticides regarding their identity and presence in watershed is still insufficient, in contrast to the extensive information available for most parent pesticides. There is a need to combine powerful analytical strategies based on LC-HRMS, which are seldom used simultaneously. We present here the workflows developed and applied in order to characterize a vineyard watershed contamination by pesticides and theirs TPs. Surface waters and soils samples were collected in a vineyard watershed located in the Beaujolais region of France (near the city of Lyon). River surface waters were sampled during rainy periods, favoring runoff of chemical compounds from fields. Soils samples were collected from several parcels with different agricultural practices, such as organic or conventional farming. To be as non-selective as possible, sample preparation of surface waters was minimal and consisted in a simple filtration step. Pesticides and their TPs in soils samples were extracted by Pressurised Liquid Extraction. Both filtered water samples and soils extracts were analyzed by LC-HRMS. The chromatographic separation was performed on reverse phase LC using a water/acetonitrile gradient and was completed by a detection with a Quadrupole Time of Flight operating in positive ionisation mode. Three data treatments were applied for each set of HRMS data. First, targeted analysis was performed with the UNIFI software (Waters) in order to quantify more than 30 known pesticides. Second, a novel enhanced suspected screening workflow was applied, with a database creation assisted by several in silico tools. This suspect screening analysis was conducted with UNIFI, in order to identify pesticides transformation products. The third data treatment consisted of a non-target screening analysis using the MZmine3 software on converted mzXML data, followed by a multivariate analysis carried out with R, in order to highlight pesticide TPs potentially missed during the suspect screening. The combined application of these three HRMS data treatments allowed the identification of pesticides and theirs TPs in soil samples, with variations observed among parcels from the same vineyard catchment. Differences were observed for legacy pesticides such as diuron or norflurazon, as well as their desmethyl TPs, beween two parcels that differ in their farming practices. Results of identification of TPs within soils and surface waters by suspect and non-target screening were cross-compared, allowing a better understanding of the presence and fate of detected pesticides and TPs within the vineyard watershed.