Figure 2 shows the chromatographic separations carried out with a Dionex Ion AS18 column of glyphosate, AMPA, glufosinate, and internal standards using the developed chromatographic method. The analytical procedure was validated through the linear range, instrumental limit of detection (LOD), instrumental limit of quantification (LOQ), procedural blank, method detection limit (MDL), method quantification limit (MQL), repeatability, trueness, and extraction yield. The internal standard method was used to quantify the pesticides and labelled glyphosate and AMPA were chosen as internal standards due to the similar instrumental and preanalytical behaviour to target pesticides. The linearity of the calibration curves was evaluated using standard solutions prepared in ultrapure water with a constant concentration of labelled glyphosate and AMPA of 1.26 and 1.00 μg L−1, respectively. Linearity was evaluated at 0.025, 0.05, 0.1, 0.2, and 0.5 μg L−1 for LW; 0.025, 0.05, 0.1, 0.2, 0.5, and 1 μg L−1 for SPM; and 0.2, 0.5, 1, 5, and 10 μg L−1 for SED using PM3 samples.
The matrix effect (ME) is obtained by separating the signal response of a standard present in the sample extract with the response of a standard prepared in the ultrapure water, expressing the result as a percentage. A value of 100% is obtained when no there is a matrix effect while a value < 100% is indicative of an ionization suppression (Matuszewski et al. 2003). Figure S1 shows the ME at three glyphosate concentrations (0.05, 0.1, and 0.5 µg L−1) in LW, SPM, and SED. To the best of our knowledge, this is the first study where a full quantitative analysis of matrix effects affecting glyphosate analysis in any matrix has been carried out. This matrix effect evaluation has highlighted that an external calibration curve in pure solvents is not a reasonable calibration method as instrumental signal suppression is not properly considered.
By considering the ratio between the concentration of pesticides and internal standards and the ratio between the relative peak areas, linearity was evaluated. LOD and LOQ values are calculated as three and ten times the signal-to-noise ratio, respectively, of the known absolute amounts of the analysed target compound in a standard solution. The MDL and the MQL were evaluated as 3 and 10 times the standard deviation of these field blanks, respectively (Bliesner 2006). Linearity, LOD, LOQ, MDL, and MQL for glyphosate, AMPA, and glufosinate are reported in Table 1. The LODs achieved for glyphosate in this study were at least one order of magnitude lower than other methods developed with ion chromatography reported in recent studies (Zhu et al. 1999; Popp et al. 2008; Chiesa et al. 2019; Dovidauskas et al. 2020). This huge step forward improving the sensitivity in terms of LODs allowed us to assess the presence of glyphosate at an ultra-trace level. The instrumental precision was evaluated and CV% value (reported as a percentage and calculated from the average (A) and standard deviation (SD), calculated as (SD/A) × 100) was always below 10%.
Another important parameter of the method validation is trueness. It is expressed as a per cent error (Table 1), calculated as (Q − T)/T × 100 where Q is the determined value and T is the “true” value. The error for glyphosate was calculated performing the same pre-analytical procedure achieved with the samples from the Lagoon of Venice. For the evaluation of the extraction yield to estimate the procedural extraction efficiency for SPM and SED, the isotopically labelled glyphosate-2-13C,15 N was added after the PTFE filtration. Extraction yield and error values for glyphosate in SPM and SED are reported in Table 1.
The recovery of the analytical procedure for SPM and SED for glyphosate ranged between 93 and 94%, comparable to the values found by Peruzzo et al. (2008) and Imfeld et al. (2013) that obtained a recovery of 86% in soybean cultivation and wetland sediments, respectively. The method allows the analysis of several samples, which is particularly interesting when a routinely monitor the seawater quality of a region is needed.
Occurrence of glyphosate in the Venice Lagoon
Glyphosate was determined in the LW, SPM, and SED collected in the Lagoon of Venice in spring 2019, autumn 2019, summer 2020, and winter 2021 (Table S2) to assess the occurrence of glyphosate depending on the seasons and to understand how the rivers and/or the local inputs can influence its concentration in the Venice Lagoon. Yearly, the Regional Agency for the Environmental Prevention and the Protection of Veneto (ARPA Veneto) presents a report of the water quality of the Veneto Region, studying plenty of environmental parameters and pollutants. From 2014 to 2019, ARPA Veneto has monitored the presence of glyphosate in some rivers that flows into the Lagoon of Venice: driven by the growing diversity of uses and dramatic increases in volumes applied, the rivers Dese, Zero, Osellino, and Naviglio Brenta (Fig. 1) showed a non-negligible presence of glyphosate in 2018 and 2019, with a constant exceeding of the concentration limits fixed by the Italian Legislative Decree n. 152/2006 (0.1 µg L−1).
The Osellino and Dese rivers had the annual highest concentration in 2019, reaching 0.4 and 0.3 µg L−1, respectively. No data are present about the seasonal variability of pesticides in these rivers. The occurrence of polar pesticides as glyphosate in the Lagoon of Venice has not been established: considering the strong influence of river discharges in the lagoon, the possible impact on the microbial community (Ker et al. 2006; Gaupp-berghausen et al. 2015) and the effects of herbicides on non-target aquatic species (Kittle and McDermid 2016) are crucial to identify the amount of glyphosate in the Venice Lagoon environment.
The glyphosate concentrations for each season in LW, SPM, and SED, in the sites St. Erasmo and Mouth of Dese River, are provided in Fig. 3. The concentrations of glyphosate during the whole sampling period are reported in Table S3. Our results showed the presence of glyphosate mainly in St. Erasmo (SE) and Mouth of Dese River (DE) suggesting that the local and river intake of glyphosate was critical, either from agricultural or non-agricultural activities. Considering the ARPA Veneto values, the presence of glyphosate in the DE could be attributed to agricultural activities, but also to point sources, such as the cleaning of machinery on the riverbank, discarding of recipients with herbicide wastes. Glyphosate can be exported mainly by runoff and by underground leaching from agricultural soils towards surface water, especially when rainfall occurs close to its application (Yang et al. 2015; Grandcoin et al. 2017). The occurrence of glyphosate in the St. Erasmo is likely due to the agricultural activities and vineyards present throughout the island. Apart from DE and SE, the glyphosate concentrations in the other sampling sites were below the MQL. Domestic and urban usage and tidal circulation could explain the sporadic presence of glyphosate in SPM and SED in Rio Marin and Petta di Bò sites (Table S3).
The glyphosate loads in the SPM and SED were always far greater in DE, compared to SE. On the other hand, glyphosate in St. Erasmo LW was < BDL in spring 2019 and showed a similar concentration and trend compared to Dese for the rest of the sampling (Fig. 3). This is might due to the glyphosate adsorbing properties to soil particles (Giesy et al. 2000): during transport over long distances from inland cultivation by Dese River, glyphosate is able to interact with water-soluble organic matter, clay particles, and iron oxides which also belong to the colloidal fraction (Vereecken 2005). This leads to a water-to-colloid transfer during the glyphosate transport. This phenomenon in SE seems to have a much smaller effect compared with DE. This could be explained by the constant tidal removal process of dissolved glyphosate by Adriatic seawater exchange that lacks the possibility to undergo the colloid transfer. The absence of glyphosate in sediments of SE site is may due to the grave seabed that does not favour the adsorption of glyphosate.
The reported seasonal loads represent the budget of glyphosate received by the Venice Lagoon from the drainage basin and are therefore fundamental in planning for the control of water quality. In the framework of system management, an interesting result is related to the temporal variability of the delivered glyphosate and to the importance of regional rainfall events on the overall pollutant transfer (Collavini et al. 2005). Consequently, the seasonal variability is particularly visible for DE in all the studied matrixes (Fig. 3). The risk of offsite transport glyphosate should be particularly evaluated due to the features of the Dese River. The river is a resurgence stream and flows in a predominantly agrarian territory, with intensive farming activities (89%). A more limited portion of the watercourse (11%) runs through urbanized areas. The results shown in Fig. 3 revealed that the concentration trend of glyphosate in the Mouth of Dese River site is strongly dependent on the matrix. LW showed the highest concentration in summer 2020 for both sites, following seasonal-related behaviour connected to the spraying of herbicides during seasonal agriculture. While the load of glyphosate on SED was quite constant through the sampling campaign in the DE site, SPM show an enrichment in spring 2019 and winter 2020/2021. The mechanism behind any seasonality in the Venice Lagoon of glyphosate at the DE can only be hypothesized. As described above and in other studies (Ronco et al. 2016; Mac Loughlin et al. 2020), the presence of glyphosate is very often associated with the SPM and its trend can be associated to local rainfall and water-to-colloid transfer during transport. Furthermore, it is demonstrated that pH, salinity, and temperature variability significantly influence the adsorption behaviours of glyphosate (Zhang and Huang 2011) and therefore its concentration in the LW. A decrease in either the seawater pH or the temperature enhances the adsorption of these compounds onto marine sediments, while the negative correlation between salinity and adsorption suggests the greater mobility of these compounds in marine than in freshwater systems (Skeff et al. 2018). Further investigation is required to link the evolution of dissolved and particulate pesticides, in order to better understand the relationship between them and the effect of their presence in the Venice Lagoon.