To the best of our knowledge, the present study and our previous one  are the first to demonstrate pesticide contamination of public sites in agricultural landscapes using grass samples. For soil samples, pesticide contamination of playgrounds was reported in one European  and one US study . The US study  exclusively focused on residues of persistent organochlorine pesticides which have been banned since the 1970s, while the European study performed on playgrounds in Sarajevo  investigated also the variety of contaminants reported for public sites in South Tyrol. The soil samples of these studies were contaminated with up to 42 different pesticide residues with concentrations up to 0.1 mg kg−1 soil for Europe and up to 0.06 mg kg−1 for playgrounds in the USA [17, 47]. We argue that pesticide residues of grass samples might give a more immediate picture of actual pesticide contamination and off-site drift than soil samples due to possible faster degradation of residues on the surface. Our previous study showed that irradiance can be a driving factor to lower pesticide contamination by photodegradation [12, 48, 49].
Concerning pesticide classes and concentrations, the findings of the current study are comparable with an Austrian study sampling field margins near apple orchards  showing that 73% of the detected residues were fungicides, the dominant pesticide class used in intensive apple and wine production. Furthermore, 23 of the detected residues reported in our current study were also detected in a recent monitoring study in Germany using passive air samplers which analyzed > 500 substances in 116 sites across Germany . Because glyphosate was the only pesticide found in all German sites we assume that glyphosate might also be present at our study sites if the samples had been analyzed with the appropriate methods .
Origin of detected pesticides
We assume that the main sources of pesticide residue contamination in our study were drifting during application or secondary drift via volatilization and/or dust from soil or plants after application [53, 54]. Of the 33 substances identified in the grass samples in the present study, 97% were approved for use in fruit orchards and vineyards in Italy at the time of sampling . Propiconazole and diphenylamine are not approved for fruit growing, but are approved for general agriculture and are additionally contained in wood preservatives. The origin of the two detected preservative agents (diphenylamine and 2-phenylphenol) is also most likely agricultural, as both are used for post-harvest treatments to control storage disease in apples and citrus fruits. In Italy, diphenylamine is used as an anti-scald agent to avoid skin browning . Diphenylamine is also used for non-agricultural dyes.
Only 10 of the detected pesticides, or 30%, have approval for non-professional users, nine fungicides, and one insecticide [55,56,57]. Therefore, we assume that the majority of residues in our samples very most likely derived from agricultural applications, though we cannot rule out contributions from other sources.
According to the official risk assessments, all detected residues (except the preservative-agent 2-phenyl-phenol, Table 1) are characterized by low volatility and thus are expected not to move more than a few meters beyond the application site . However, our results suggest that evaporation from plant surfaces or particle drift may be significant sources for off-site contamination. It is important to point out that the European Food Safety Authority (EFSA) builds its risk assessment on expectations and modeling. Our data show that these models seem to underestimate the real-life situation and we, therefore, plead for a stricter application of the precautionary principle.
Potential impact of pesticide residues on human health
We assume that vegetables and fruits from nearby private gardens are as likely to be similarly contaminated by pesticide drift as the grass on public sites. Hence, we argue that the residue levels in grass samples might also be a valid indicator for overall contamination in public sites and private gardens.
Comparing residue levels in the grass with MRLs for lettuce, spinach, and strawberries, the levels for fluazinam and captan detected in this study, exceeded the surrogate MRLs 24-fold and 15-fold, respectively. Moreover, fluazinam is an endocrine active substance of category 2  and captan is classified as a category 2 carcinogen, according to the EU pesticide database .
The multiple detections of fluazinam on three sites on the following sampling dates (spring and summer) at the end of May and the end of July with decreasing concentrations indicate a possible persistence of this residue. The half-life of fluazinam is up to 69 days  (Table 1) and the concentration decline from spring to summer was 80% (Fig. 3). Residues of the fungicides captan and dodine did not decline and their half-lives are only between 3 and 5 days, indicating that both are less persistent and that the higher concentrations detected in summer occurred through repeated pesticide application (Fig. 3).
Generally, residue concentrations in grass samples were low, but the year-round detection and the reported contamination of the last years [12, 27, 28] indicate chronic exposure of humans and the environment. Among the detected pesticides, chlorpyrifos and chlorpyrifos-methyl are the most hazardous. Residues of these chemicals were detected in the grass samples of 2018 and in samples from earlier years [12, 27, 28]. Low doses of chlorpyrifos have been shown to lead to brain anomalies in fetuses and children , and to affect locomotor activity, behavior, and neurotransmitter systems in rats . Recently a study  re-evaluated the low dose effects of chlorpyrifos in developmental neurotoxicity and corrected the misleading results of the original assessment ; meanwhile, these two insecticides are banned in the EU.
Overall, 25 of the 33 identified compounds are EDAs [43, 64], which were present in 96% of the investigated sites, while in 53% of them, EA substances were detected all year round. Endocrine disruptive compounds frequently exhibit non-monotonic dose–response relationships and, therefore, can be effective in concentrations several orders of magnitude below common residue thresholds . Relative to their body weight, children inhale more air, drink more, and eat more food than adults, and some activities of children on playgrounds are likely to bring them in contact with contaminants present in the grass or sand [24, 66]. Endocrine active pesticides interfere with oestrogen or androgen receptors  and have been linked to an increased risk of thyroid, breast, and prostate cancer [1, 67]. Simultaneous exposure to a variety of pesticides might also trigger synergistic effects , which are extremely difficult to examine in risk assessments . Due to the low-dose effect, the non-monotonic dose–response relationship, and the interaction of EDCs with endogenous hormones or other EDCs, it is questionable whether a low dose can ever be considered safe [45, 65]. Moreover, EDCs can cause multigenerational effects [70, 71] even in grandchildren of exposed pregnant women .
Exposure of pregnant women and children to endocrine active pesticides—as were detected in our grass samples—and/or their metabolites have been described in several publications. For example, di-ethyl phosphate and 3,5,6-trichloro-2-pyridinol, both metabolites of chlorpyrifos and chlorpyrifos-methyl, were detected in urine samples from children  and hair samples from pregnant women [74, 75], as well as cypermethrin, cyprodinil, difenoconazole, imidacloprid, oxadiazon, penconazole, propiconazole, pyraclostrobin, pyrimethanil, tebuconazole, tetraconazole, thiacloprid and zoxamid [74, 76]. Many of the EDCs are fungicides—in this study 75%— which is also the most prevalent group of detected residues in the current sites. In the European Union, fungicide sales represent more than 40% of total pesticide sales, and in areas dominated by fruit growing, such as vineyard regions, fungicides account for more than 90% of all pesticide applications .
Fungicides such as propiconazole and tebuconazole have also been identified as potential human health risks by exposure through drinking water . Indeed, contamination of surface water within the study area had been reported for the same pesticides and at similar concentrations as in grass samples from the public sites investigated .
Multiple pesticide contamination occurred on 19 sites at least once during the entire year. The effect of mixtures of active compounds and their specific formulations/additives is still not clear. Exposure due to pesticide drift raises the issue of multiple exposures from various sources  and the still unknown extent thereof. It is obvious that people living in the study area of this survey are simultaneously exposed to the same pesticides in commercial food, vegetables, fruits, and herbs from private gardens, as well as through inhalation of contaminated air.
Plant protection products consist, besides the active compounds, of specific formulations to enhance the effect of the ingredients. It is likely, for instance, that formulations on tween-basis can have an influence on the mobilization, bioavailability, and bioaccumulation of organochlorine contaminants and its metabolites [81, 82]. However, the concealment of the full list of ingredients in formulations makes a detailed investigation impossible.
Impact on the environment
Besides affecting human health, many of the detected pesticides have a proven impact on various non-target organisms and are in general a driving factor in biodiversity decline . Non-target effects of pesticides affect soil biota [84,85,86], bees [87, 88], and other organisms important for the functioning of the agroecosystems . A recent biodiversity assessment conducted in the same region as the presented study showed that butterflies are endangered through pesticide exposure .
We intended to estimate the association between environmental factors and pesticide residues in public sites, as we did in our previous study . Due to a nearly three-fold higher number of different fungicides detected in spring 2018 compared to our previous study  and a higher proportion of public sites contaminated with fungicides, we hypothesized that such findings could relate to the higher rainfall in 2018. However, the relationship between precipitation and detected fungicide residues was not significant, pointing to multifactorial influences.
One public site was without any pesticide residue throughout the year, even though the site is located only 40 m from the closest agricultural area. However, we know from the previous study  that, besides chemical drift, the characteristics of the pesticides applied, as well as the prevailing wind direction and wind speed, may have affected the contamination of these public sites.
Our results show that even a distance of more than 100 m to agricultural sites does not lower the contamination level significantly (Additional file 1: Figure S1) and the relationship between residue load and distance is not linear (Additional file 1: Figure S2). Therefore, an evaluation of the extent of exposure requires further studies and a subsequent risk assessment.
There are already sufficient arguments to ban or at least drastically reduce pesticide application. The most important arguments therefor concerning human and environmental health, sustainability, and costs. The impact of chronic pesticide exposure on human health and the environment is proven [1, 89]. All in all, the health costs to the EU caused by endocrine active compounds, mainly pesticides, have been estimated at more than € 150 billion annually [91, 92]. However, sustainable and ecological agriculture is possible and economically. It generates farm incomes exceeding those from conventional and industrial farms, providing more employment and thus supporting regional economies by using less fossil fuel and contributing to preserving biodiversity .