The main challenge of studies under field realistic conditions is to overcome the diverse confounding factors. Ideally, identical environmental conditions prevail at all investigated study sites to relate any differences found exclusively to the treatment and increase the validity of the results. However, in a large-scale monitoring study like the one presented in this issue, equal conditions cannot be ensured as variability is part of the natural system (Liess et al. 2005). Nevertheless, where variable conditions cannot be avoided, the parameters can be measured and included as covariates in the statistical analyses, thus, providing a better understanding of complex interactions under realistic field conditions.
In the following, the measures applied to account for common uncertainties in field studies are highlighted. These measures include (i) comparable conditions at the reference and test sites in terms of land use, soil, climate, alternative forage resources, as well as development of the OSR, (ii) ensuring the crop fidelity of the studied bees and (iii) the exposure of the bees to the focal neonicotinoid.
Similarity of environmental conditions and agronomic practice at the reference and test site
Land cover and land use types
The study sites were chosen based on the high OSR crop density for the region but also to resemble each other in important environmental conditions. Land use/land cover data (LULC) for the study area were obtained from high-resolution aerial images (Table 1). All types of arable fields as well as different landscape structures such as hedges, kettles (small hollows originating from buried dead ice after glacier retreat), and settlements were identified in the field and manually digitised from satellite images (Google Satellite, date taken 06.05.2011) using a computer-based geographical information system (Quantum GIS, Version 1.8.0 Lisboa). The exact coordinates of the location and shape of OSR fields in the study area as well as relevant landscape features inside the fields such as kettles, forest patches or shrubs were recorded with a GPS handheld receiver (Garmin eTrex 10).
The habitat mapping indicated a diverse distribution of different LULC types at both study sites and although field sizes are relatively large (up to several hundreds of ha), the whole area is well structured by a diversity of small forest patches and groves of trees, hedges, water bodies of different sizes and kettles (Fig. 1b). The most important land-use type was arable land, covering 49.5 % and 72.2 % of the core area of the reference site and the test site, respectively. The higher proportion of arable land at the test site was mainly due to the larger cropping area of maize and the lack of any larger water body at the test site (Fig. 1b, Fig. 2). OSR was the most common crop at both study sites (Fig. 2). At the core of the reference site, 17 study fields covered in total 614.6 ha with OSR, constituting 16.0 % of the area, whereas the test site comprised 791.7 ha (20.6 %) of OSR at 18 study fields (Fig. 1b, Fig. 2). The median size of OSR study fields did not differ between reference and test site (Table 2).
Table 2 Summary of parameter comparison between the study sites
Soil characterisation
To characterize the soil from each study field of the core areas, soil samples were taken before drilling of OSR seeds in August 2013. Study fields were subdivided into plots of 10 ha. Ten samples from equally spaced points were taken from the upper 10 cm of the soil in each plot. Plant material and other coarse contaminants were removed and all samples of one plot were combined and thoroughly mixed before the analyses. Characterisation of soil samples included the determination of pH (DIN ISO 10390 (2005)), total organic carbon (TOC, DIN ISO 10694 (1996)), water holding capacity (WHC, DIN EN ISO 11274 (1998)), and particle size (DIN 19683 (2012)). The pH, TOC and WHC were tested for differences between the study sites by fitting linear mixed models which included the study field ID as a random effect to account for the non-independence of sampled plots per study field. The soil type classification was analysed with a Fisher’s exact test. The soil characterisation indicated no significant difference between study fields at the reference and test site regarding the pH, the total organic carbon, and the water holding capacity (Table 2). The soil texture was identified to contain, on average, 67 % sand, 23 % silt, and 10 % clay and was classified accordingly as predominantly loamy sand both at the reference (98.3 %) and test site (98.7 %). Loamy sands are dominated by sand particles, but contain enough clay and silt to provide some structure and fertility.
Climatic conditions
To account for small scale climatic differences, weather conditions were measured at all study locations during the exposure phase. At each honey bee location, calibrated devices connected to two hive balances (CAPAZ GSM 200) measured the air temperature and relative humidity once per hour. From these double measurements an hourly average per location for both temperature and humidity was calculated. Wind speed and direction at 2 m height was recorded every 10 min by an anemometer (Davis Vantage Pro II) and stored as hourly mean and maximum. Additionally, hourly sums of precipitation were collected by a rain gauge (accessory of CAPAZ GSM 200). At each mason bee study location, the air temperature and relative humidity were collected by a validated data logger (Gemini TGP-45000) at 30 cm height which was protected against rain and direct sunlight. Similar to the honey bee study locations, an anemometer was set up at each mason bee study location to measure wind speed and direction. Daily sums of rainfall for all mason bee locations were obtained from the German Weather Service (DWD) of a local weather station at Goldberg, approximately 10 km east of the test site.
The measured weather conditions at the study locations coincided with official measurements (Statistisches Amt Mecklenburg-Vorpommern 2015) indicating a warm and relatively dry period during the third pentad of April, followed by lower temperatures at the beginning of May which increased again towards the end of May. Rainy periods were concentrated during the second and third pentads in May (Statistisches Amt Mecklenburg-Vorpommern 2015). In general, no weather extremes occurred during the exposure phase. Although they are not representative for the whole study area, weather data collected at the honey bee study locations were analysed for differences between the study sites. There were no significant differences between the study sites in the daily mean temperature, the daily mean of relative humidity, the daily sum of precipitation, and the daily mean wind speed (Table 2, Fig. S1).
Agronomic practice
Information about agricultural practices at the study fields, such as treatment with other PPPs and their application rates was gathered from the farmers for the period of the monitoring study as well as additional details of the variety, drilling rate, and origin of OSR seeds. Apart from the seed dressing and the request not to apply any further neonicotinoids between drilling in August 2013 and harvest in July 2014, the farmers were allowed to decide for themselves about all agricultural practices including the application of other PPPs.
The study was conducted with the cooperation of independent farmers who made the decisions about all agricultural practices, the seed types and the PPP applications and so there was some variation between study fields. In order to comply with local conditions and to optimally schedule agricultural activities, several OSR varieties were used. In total, 33 different OSR varieties were drilled at the study fields of which the most common were Genie (R: 40.6 % of crop area, Rapool-Ring GmbH), Sherpa (R: 25.3 %, T: 22.2 %, Rapool-Ring GmbH), and Xenon (T: 21.4 %, Rapool-Ring GmbH). The diversity of OSR varieties was larger at the test site mainly because the study field T13 was used for a variety demonstration and, thus, contained 22 different varieties sown in stripes each less than 1 ha in size. Two of these demonstration varieties were dressed with thiamethoxam (3 g/kg seeds) instead of clothianidin. The test fields T1, T9, T10, T14 and T15 also contained more than one OSR variety (Table S1). Grouped by their anticipated time of flowering, early varieties dominated at the test site while intermediate flowering varieties dominated at the reference site (Table S1). However, the difference in the period of full flowering between early and intermediate varieties constitutes 3–4 days only and nectar and pollen are available beyond that period. Furthermore, small scale microclimatic conditions may cause a higher variability in the flowering time of OSR. The seeds had an average thousand seed weight (TSW) of 6.2 ± 1.3 g. The General Linear Model revealed that the TSW differed due to the OSR variety (F31, 32 = 2.42, p = 0.007) and the amount of seed dressing (F1, 32 = 5.39, p = 0.027), but not between the reference and test sites (Table 2). The drilling rate of OSR seeds averaged 3.4 ± 1.1 kg/ha and was significantly higher in study fields at the test site compared to the reference site (Table 2, Table S1). However, compared at the landscape level and weighted by the field size of study fields, the drilling rate did no longer differ between the treatments (Table 2). Based on differences in the TSW and the drilling rate, the average number of seeds per square meter was significantly higher at test fields compared to the reference fields (Table 2).
During the development of the OSR plants, they received on average 4.8 ± 0.4 and 4.1 ± 0.9 insecticide spray applications at the reference and test site, respectively. This difference was statistically significant (Table 2) and was due to a significantly higher number of applications in autumn at the reference site (Table 2). This was because the OSR plants lacked the insecticidal seed treatment and most of the study fields at the reference site received an additional pyrethroid spray treatment in autumn 2013 to control cabbage stem flea beetles (Psylliodes chrysocephalus) and cabbage root fly (Delia radicum). The number of additional insecticide applications in spring 2014 did not differ statistically significant between the sites. The most frequently applied compounds were the pyrethroids etofenprox (Trebon 20 EC®) and beta-cyfluthrin (Bulldock®). The oxadiazine indoxacarb (Avaunt®) which is classified as harmful to honey bees and bumble bees when exposed to direct treatment (DuPont 2004; van der Steen and Dinter 2008), was applied at seven reference fields in March and the beginning of April 2014. This was well in advance of the establishing of the bumble bee hives at the study locations (by at least 2.5 weeks). Pymetrozine (Plenum®), the only triazine used on the study fields, was applied at T7–T10 and T13–T15 at least 3 weeks prior to the start of the exposure phase of the bees.
The agronomic practice of non-OSR fields at the study sites are also not expected to have any confounding effect on this study. Although sowing of maize fields overlapped with the flowering of OSR and the dust from sowing operations of neonicotinoid dressed maize was shown to adversely affect honeybees under specific exposure conditions (Pistorius et al. 2009), neonicotinoid dressings of maize are not authorized in Germany since 2008 (Federal Office of Consumer Protection and Food Safety 2009) and the sown maize only contained a fungicide or no dressing at all (personal communication with farmers). Furthermore, due to the restrictions on neonicotinoid use since December 2013 (European Commission 2013), confounding effects of neonicotinoid applications at adjacent fields can also be excluded.
OSR development
At the study fields in the core area, the development of the OSR crops was surveyed seven times between November 2013 and the end of the exposure phase in May 2014. Corresponding BBCH-stages were determined based on the adjusted code for OSR development (Federal Biological Research Centre for Agriculture and Forestry 2001). The rate of emerged OSR plants and the respective plant density was assessed prior to the stem elongation in March 2014. For methodological details of the density estimation see Rolke et al. (2016b).
Drilling of the OSR seeds took place between 13 and 29 August 2013 with a peak on 18 and 19 August 2013 which was similar for study fields at the reference and test sites (Table 2). The rate of plants surviving the winter averaged 68 ± 25 % and was equal at study fields of the reference and test sites (Table 2). However, due to the differences in the drilling rate, the OSR plant density was higher at test fields compared to reference fields (Table 2). The OSR crops at the study sites developed homogenously across all seven assessments based on the BBCH stages (Table 2). By the first assessment on 21 November 2013, almost all OSR plants had reached BBCH stage 19 (“9 or more leaves unfolded”). More importantly, a few days before the start of the exposure phase of the bees (21 April 2014) at least 30 % of flowers were open on all study fields (BBCH 63), ensuring sufficient food was available for the bees. Full flowering of OSR (BBCH 65) lasts for three to five weeks. Accordingly, by 22 May 2014 flowers at the majority of study fields had withered and only 5 % of the plants were still with flowers. The exposure phase was terminated at this stage because the OSR plants did not provide sufficient amounts of nectar and pollen for foraging bees any more.
The OSR yield standardised to the field size was significantly higher at the test site than at the reference site (Table 2). This difference was in line with yield differences in previous years probably due to a slightly more productive soil in the south of the study area. This difference could also have been due to early losses of OSR plants at the reference site which lacked the neonicotinoid seed treatment although it was compensated for by the additional pyrethroid spray application. The lower plant density at the reference fields may also have contributed to the difference in yields. However, the different plant densities did not affect the availability of OSR nectar and pollen for the investigated bees because the coverage of OSR at the study sites provided food in excess and all bee hives developed very well during the exposure phase (Rolke et al. 2016a; Sterk et al. 2016). Furthermore, the average yields at both study sites were close to the average yield of winter OSR of 37.5 dt/ha in the district of Ludwigslust-Parchim in 2014 (Statistisches Amt Mecklenburg-Vorpommern 2015).
In summary, the environmental and agronomic conditions at the reference and test site were largely similar with the exception of the insecticide treatment. Thus, local environmental conditions were considered not to have any significant confounding effect on the results of the monitoring study.
Crop fidelity of bees
Three measures were applied to ensure that the investigated bees foraged at the OSR that was grown from seeds either treated with or without a clothianidin dressing. Firstly, the study sites were selected to provide a high density of OSR crops but did not include any other mass flowering crop which was suitable as bee forage during OSR flowering. Furthermore, as described above, the size of the study sites was intended to cover the foraging flights of the investigated bees.
Secondly, bees may also use weeds and flowering plants at field margins, forest edges, settlements and grasslands as pollen and nectar resource apart from cultivated crops (Stanley et al. 2013a; Stanley and Stout 2014). Accordingly, a detailed assessment of the abundance of alternative bee forage (pollen and nectar provided by other than OSR plants) during OSR flowering was obtained by a semi-quantitative survey of non-crop habitats at both study sites. During OSR flowering, 10 representative hedges, kettles, and forest edge habitats were visited once and the abundance of flowering plants assessed along a transect of at least 150 m in length (Fig. 3). Taking the variations in importance for the different bee species into account, each flowering plant species was rated on an ordinal scale as following: 0—no occurrence of plant species; 1—very few flowers present, to be neglected as food source for bees; 2—few flowers present, sufficient as food source for individual bees; 3—numerous flowers present, sufficient as food source for bees; 4—abundant flowers present, sufficient as food source for bees, very attractive. A similar classification was carried out for grasslands and field margins from photographs taken during OSR flowering. Ratings of all plant species present at a sampling site were averaged and used to calculate a mean for each habitat type per study site. The resulting value per habitat was weighted by the area of the habitat according to Eq 1.
$${\mathrm{Availability}}\,{\mathrm{per}}\,{\mathrm{habitat}} = \frac{{{\sum} {{\mathrm{mean}}\,{\mathrm{rating}}\,{\mathrm{per}}\,{\mathrm{sampling}}\,{\mathrm{site}}} }}{{{\mathrm{number}}\,{\mathrm{of}}\,{\mathrm{sampling}}\,{\mathrm{sites}}}} \\ \times {\mathrm{area}}\,{\mathrm{of}}\,{\mathrm{habitat}}$$
(1)
In total, 38 plant species were found in hedges, kettles and forest edges which are attractive to bees during OSR flowering and may have represented a forage source for at least one of the bee species studied. Of these plants, 7.2 species occurred on average per surveyed site. The Wild Chervil (Anthriscus sylvestris), Common Oak (Quercus robur), Archangel Fair (Lamium album), and Dandelion (Taraxacum spec.) were relatively common and occurred at more than half of the surveyed sites. Only trees and shrubs were highly available as food resource for foraging bees. Hedges were on average more diverse than kettles and forest edges. For grassland, field margins and urban areas the coverage with flowers suitable as bee forage were estimated to constitute 10 % which is transformed to a rating of 1.5. Compared to OSR, which is highly attractive at least to honey bees and available at 13.9 % of the area of the study sites, the alternative foraging resources play only a minor role (Fig. 4).
However, this assessment of alternative forage for bees at non-crop habitats can only give an approximate estimate of the availability of alternative food resources. Therefore, a third approach to ensure that the investigated bees fed on the OSR of the study sites, was to analyse the composition of the pollen collected by the honey bees, earth bumble bees and red mason bees, as well as the nectar and honey collected by honey bees. For methodological details see the respective papers in this issue (Peters et al. 2016; Rolke et al. 2016a; 2016b; Sterk et al. 2016). This investigation confirmed that all three bee species foraged on OSR although to different degrees. In particular red mason bees collected pollen from a diversity of plants available in the close vicinity of the study locations. Nevertheless, the exposure of the bees to OSR at the study site was proven and the difference in the amount of OSR among the utilized food resources reflect the typical exposure of the different bee species to OSR under natural field conditions.
History of study fields
Crop and PPP history of study fields
The study area has a long agricultural history which might contribute confounding effects on the monitoring study because PPPs from agricultural applications seem to occur ubiquitously in the environment (Stewart et al. 2014, Cutler et al. 2014). Hence, the difficulty arises to find control sites which coincide with the test sites in their background levels off PPPs without confounding the study results (e. g., Cutler et al. 2014; Rundlöf et al. 2015). In order to reduce the possibility that PPPs applied to the study fields in recent years might have affected the outcome of the monitoring study, detailed information on the agricultural practice at the study fields within the five years previous to the monitoring, including cultivated crops, treatment with PPPs and their application rate were obtained from the farmers.
The crop history of the study fields between the harvest years 2009 and 2013 indicated common crop rotations following good agricultural practices at both reference and test site (Fig. 5). The most common crops cultivated at the study fields during the previous five years were wheat and maize. The previous OSR cultivation at the study fields dated back between three and more than five years. At least 83.3 % of previously cultivated OSR contained a clothianidin dressing, and for 12.5 % of the former OSR cultivated there was no information of seed dressing available. Seed dressings of crops other than OSR were primarily fungicides and did not contain clothianidin or other neonicotinoids as an active ingredient except for sugar beet cultivated at the test field T10 in 2012 which was dressed half with clothianidin and half with thiamethoxam, a neonicotinoid of which the primary metabolite is clothianidin (Nauen et al. 2003). Insecticides applied to the study fields during the five years previous to the study were mainly pyrethroids (70 % of insecticide applications) and neonicotinoids (21 %). Oxadiazines (5 %), carbamates (2 %), and organophosphates (2 %) played only a minor role. Neonicotinoids were primarily applied as seed dressings of which 96 % contained clothianidin and 4 % thiamethoxam. Thiacloprid accounted for 96 % of all neonicotinoid spray applications whereas acetamiprid was applied once. Imidacloprid had not been applied during the last five years. At all study fields, the last applications of neonicotinoids were in 2011 or earlier, hence, at least 3 years prior to the monitoring study.
Residues in soil before drilling
The collected soil samples were analysed for clothianidin residues because of the predominant use of this active ingredient and its relatively long half-life in soil (Krupke et al. 2012). Based on the information gained by the agricultural practices of the farmers, the analysis of the soil was restricted to clothianidin because other neonicotinoid insecticides were not applied or the application dated back a multiple of the respective half-life. Samples of the test field T18 were only taken after drilling of OSR seeds and therefore were not suitable for the analysis of background clothianidin residue levels. Preparation of soil samples was based on QuEChERS methods (“Quick Easy Cheap Effective Rugged Safe”, DIN EN 1552 (2008); Lehotay 2006). To determine clothianidin concentrations, liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) was applied. The chromatographic system used was a high performance liquid chromatograph with a reversed phase chromatography (Zorbax Eclipse C18, 50 × 2.1 mm, 1.8 μm column) coupled with tandem mass spectrometry and electrospray ionisation (AB Sciex API 6500 Triple Quadruple Mass Spectrometer, Analyst version 1.6.2). The coefficient of determination for calibration curves was above 0.996. In soil, the limit of quantification (LOQsoil) was 5 μg/kg dry weight soil and the limit of detection (LODsoil) was 1.5 μg/kg.
In 82 % of the 134 samples no clothianidin was detected (<LODsoil), whereas residues below the LOQsoil were found in 6 and 18 plots of study fields at the reference (R6, R7, R9, R12) and test sites (T3, T4, T7, T8, T13, T14), respectively. Calculations based on the upper limits of LODsoil and LOQsoil revealed a very conservative estimate for average soil residues of 2.1 ± 1.3 μg/kg which did not differ significantly between the study sites (Table 2). This residue concentration is in the range of clothianidin residues reported for agricultural soils with repeated drilling of clothianidin dressed corn (2–11.2 μg/kg (de Perre et al. 2015), 2013: 4.0 ± 1.1 μg/kg, 2014: 5.6 ± 0.9 μg/kg (Schaafsma et al. 2015), 7.0 ± 4.2 μg/kg (Xu et al. 2016)) and OSR seeds (5.7 ± 4.0 μg/kg (Xu et al. 2016)). The distribution of plots with clothianidin residues among the study fields did not reveal any correlation with the crop or PPP history during the previous five years. Since clothianidin is known for its ageing behaviour in soil and has also a low bioavailability of 6–10 % (Xu et al. 2016), the non-quantifiable soil residues on 18 % of all sampled plots were considered not to contribute to translocation into bee-relevant matrices (nectar and pollen). This assumption has been confirmed by the later residue findings in nectar and pollen samples on the reference sites under confined conditions (>90 % of all residue samples not detectable at LODnectar/pollen/honey = 0.3μg/kg).
Because crop history and PPP applications resembled each other at the two study sites and based on the applied PPPs no adverse long-term effects are to be expected.
Ensuring exposure to focal PPP
Clothianidin loading of OSR seeds
Another uncertainty in field studies is whether the investigated free flying bees are exposed to the PPP under consideration. To be able to quantify the potential exposure, OSR seed samples taken before drilling were analysed for their clothianidin loading. Before sowing, samples of approximately 500 g were taken from all OSR seeds for the analysis of clothianidin loadings. The seed samples were pre-processed by mixing with an acetonitrile/water solvent mix (4/1, v/v) to extract the clothianidin and analysed by liquid chromatography coupled with tandem mass spectrometry similar to the analysis of residues in soil samples described above. The coefficient of determination for calibration curves was again above 0.996. The LOQseeds of clothianidin residues in OSR seeds was 1.0 mg/kg.
The amount of clothianidin in the seed coating of treated OSR seeds averaged 7.8 ± 1.5 g/kg and ranged between 43.9 % and 108.2 % of the nominal concentration of 10 g/kg (Table S1). Traces of clothianidin were also found in the OSR seeds of the reference site with a median loading of 0.02 g/kg (range 0.001–0.226 g/kg). These low amounts of clothianidin arose from residues in commercial facilities for seed treatment. Though OSR seeds used on the reference site were not treated with clothianidin, the seeds were processed in common seed treatment facilities for dressing with the fungicides thiram and dimethomorph. The study fields R17 and R18 at the edge of the reference site (Fig. 1) also contained a clothianidin seed dressing. However, this did not diminish the value of the monitoring study because the nearest study locations of bees were 2.9 km apart (compare Rolke et al. 2016b) and the Great Sternberg Lake in between most likely formed a natural barrier for the bees. Accordingly, the lack of clothianidin residues in pollen and nectar from the closest study locations verified that the bees at the reference site did not forage on these two study fields where the seeds had been treated with clothianidin (Rolke et al. 2016b) probably due to the ample food available in the vicinity of the hives.
Based on the clothianidin-loading of seeds and drilling rates, on average 28.8 ± 10.0 g/ha of clothianidin were applied to the study fields of the test site and 0.19 ± 0.25 g/ha at the reference site during drilling. If we assume an equal distribution of OSR seeds and clothianidin at the field and an average soil density of 1.5 kg/L, the clothianidin concentration at the test and reference site amounted to 19.2 ± 6.7 μg/kg and 0.13 ± 0.17 μg/kg, respectively, in the uppermost 10 cm of the soil after drilling. The contamination at the reference site is below the average residue concentration in the soils before drilling and, thus, considered very unlikely to have a confounding effect on the study results.
Clothianidin residues in bee forage
To further verify the exposure to clothianidin of the bees at the study locations, pollen collected by honey bees, earth bumble bees and red mason bees as well as nectar and honey collected by honey bees were analysed for clothianidin residues and its metabolites thiazolylmethylurea and thiazolylnitroguanidine. For methodological details see Rolke et al. (2016b). In an additional semi-field tunnel-tent study (bees confined to the test crop in insect-proof cages at all study fields), nectar and pollen samples were collected from honey bees foraging exclusively on OSR. The residues in pollen and nectar (Npollen = Nnectar = 39) from the tunnel-tent study indicated a clear exposure to the neonicotinoid at the test site, whereas at the reference site no residues were detectable in the majority of samples (96 % for pollen, 100 % for nectar, Npollen = Nnectar = 34).
Similar results were obtained for the investigated bee species at the study locations which could freely forage. Neither clothianidin nor its metabolites were detected (LODnectar/pollen/honey = 0.3 μg/kg) in any pollen sample collected by honey bees (N = 96), bumble bees (N = 6) and mason bees (N = 6) at the reference site, whereas a few nectar samples (5.6 %, N = 96) and 62.5 % of honey samples (N = 48) contained non-quantifiable amounts (LOQnectar/pollen/honey = 1.0 μg/kg) of clothianidin. In contrast at the test site, clothianidin residues were detected in the majority of pollen (Nhoney bee = 96, Nbumble bee = 6, Nmason bee = 6), nectar (N = 96) and honey samples (N = 48), mainly at concentrations below the LOQnectar/pollen/nectar but also at clearly quantifiable concentrations with a maximum of 2.7 μg/kg in pollen, 1.6 μg/kg in nectar, and 2.1 μg/kg in honey (Rolke et al. 2016b). These results clearly demonstrate that the investigated bees were exposed to clothianidin while foraging on OSR grown from clothianidin dressed seeds at the test site.
Residues in soil after harvest
The half-life of clothianidin in soil was reported to range between 13.3 and 305.4 days under field conditions (mean: 120.1 days, European Commission 2005). In order to assess the persistence of clothianidin in the soil of the study area after applying a known amount of the neonicotinoid as seed dressing, the central study fields T7, T8, and T10 of the treatment site were sampled again after the harvest of OSR plants in August 2014. This time, one soil sample was taken per subplot (N = 15). The analysis of clothianidin residues was conducted as described above for the previous analysis of clothianidin residues in soil. The residues of clothianidin were below the LOQsoil (5 μg/kg) in 11 of the 15 samples and the maximum concentration found was 5.9 μg/kg at study field T10. These concentrations are considerably lower than the amount of clothianidin in the soil after drilling which was calculated as above to constitute 15, 9.4 and 16.9 μg/kg on the study fields T7, T8 and T10, respectively. The analysed residue concentrations indicate clothianidin concentrations to dissipate by 50 % in 0.5 years or less which corresponds well with the reported dissipation of clothianidin in agricultural soil (Schaafsma et al. 2015).