Skip to main content

Effects of a realistic pesticide spraying sequence for apple crop on stream communities in mesocosms: negligible or notable?



Several large-scale studies revealed impacts and risks for aquatic communities of small rural lakes and streams due to pesticides in agricultural landscapes. It appears that pesticide risk assessment based on single products does not offer sufficient protection for non-target organisms, which are exposed repeatedly to pesticide mixtures in the environment. Therefore, a comprehensive stream mesocosm study was conducted in order to investigate the potential effects of a realistic spraying sequence for conventional orchard farmed apples on a stream community using pesticides at their regulatory acceptable concentrations (RACs). Eight 74-m-long stream mesocosms were established with water, sand, sediment, macrophytes, plankton and benthic macroinvertebrates. In total, nine fungicidal, four herbicidal and four insecticidal pesticides were applied in four of the eight stream mesocosms on 19 spraying event days in the period from April to July while the remaining four stream mesocosms served as controls. The community composition, the abundance of benthos, periphyton and macrophytes, the emergence of insects, physico-chemical water parameters, and drift measurements of aquatic invertebrates were measured.


The pesticide spraying sequence induced significant effects on invertebrates, periphyton, and macrophytes as well as on the water ion composition especially in the second half of the experiment. It was not possible to relate the observed effects on the community to specific pesticides applied at certain time points and their associated toxic pressure using the toxic unit approach. The most striking result was the statistically significant increase in variation of population response parameters of some taxa in the treated mesocosms compared to the controls. This inter-individual variation can be seen as a general disturbance measure for the ecosystem.


The pesticide spraying sequence simulated by using RAC values had notable effects on the aquatic stream community in the conducted mesocosm study. The results indicate that the current risk assessment for pesticides may not ensure a sufficient level of protection to the field communities facing multiple pesticide entries due to spraying sequences and other combined stress. Hence, there is still room for improvement regarding the prospective risk assessment of pesticides to further reduce negative effects on the environment.


The EU pesticide risk assessment (RA) is based on a tiered approach in order to sufficiently address the risk of plant protection products (PPP) for the environment (pesticide regulation, European Commission No. 1107/2009) [1]. The tier 1 RA is a laboratory-based ecotoxicological assessment, which includes several single species tests. More complex and environmentally realistic evaluations (tier 2 and 3) are usually proposed as refinements if the tier 1 RA indicates that the legally specified level of acceptability is not met [2].

A conceptual key element of the pesticide RA is the derivation of regulatory acceptable concentrations (RAC) of PPPs and/ or their active ingredients (AI). For tier 3, RACs are derived on the basis of negligible population effects or population effects with subsequent ecological recovery within a certain time period [2, 3]. In each tier, the RAC needs to be higher than the calculated predicted environmental concentration (PEC) for edge-of-field surface waters [2] in order to conclude on an acceptable risk. On the exposure side, the pesticide entries into the edge-of-field waterbodies, simulated by the PECs, can be reduced by applying mitigation measures such as, e.g. using drift-reducing spray nozzle technology or implementing no-spray-zones or vegetated buffer-strips [2].

The complexity of the higher tier pesticide regulation process increased over the years with the inclusion of multi-species tests, mesocosm and field studies, as well as effect modelling. Some of these studies are usually proposed to refine the risk by, in theory, introducing more realism into the risk assessment, e.g. by estimating more accurately the sensitivity and vulnerability of complex ecosystems [2, 4]. Also, mixture toxicity is included by considering the interactions between the AIs, safeners, synergists, and co-formulants of a single PPP. However, this implies that only technical mixtures such as PPP with one or more AI are included, while the so-called “coincidental’’ pesticide mixtures, like inter alia tank mixtures or pesticide spraying sequences during a growing season, are not explicitly considered [1, 5]. The application of a series of distinct PPPs with different modes of action, or the use of tank mixtures, is a common practice in current agriculture, assuring sufficient agricultural yield by avoiding the occurrence of different pests over the growing season [6,7,8]. Consequently, it is questionable if the risk assessment based on single PPP is adequate to protect non-target organisms from the negative effects of exposure to real-world mixtures and/or spray series of PPPs.

Analysis of field data for pesticides in smaller waterbodies of rural areas including detailed evaluation of large-scale aggregated pesticide data demonstrated increasing evidence that for a relevant number of waterbodies, impacts and risks for aquatic communities due to pesticides are indicated [e.g. 917]. Therefore, agricultural land use can be considered as a main input pathway of pesticides into streams [16]. In a comprehensive study, Junghans et al. [18] investigated the risk of pesticide mixtures for aquatic communities in five small agricultural streams in Switzerland. By use of stream water for biotests, a high risk for plants, invertebrates and fish at all investigated sites was confirmed for most of the water samples [18]. Herbicides and fungicides were the two pesticide classes most frequently detected in surface waters, and especially herbicides were widely present in aquatic ecosystems due to their high application rates [19, 20]. The analysis of environmental monitoring data from different countries displayed a mixture of three to five pesticides in median, but also exceeding a mixture of ten different compounds in some cases in surface waters. Herbicides were the dominant pesticide class found followed by fungicides and insecticides [21].

These findings demonstrate that there is a potential discrepancy between the claim and objectives of the plant protection regulation [1] and the actual situation in the field. There are several reasons why pesticide concentrations above RACs occur in edge-of-field surface waters. These reasons include an improper use of PPP, more frequent run-off events than anticipated, lack of or insufficient wetland vegetation at the riparian zones for input reduction, simultaneous PPP application of the same set of AIs on different crop areas within a river catchment, or application of a series of different PPPs within a crop-specific spraying sequence. Especially the environmental risk of spraying sequences of different PPPs has been ignored so far by the EU RA [8].

Therefore, the German Environment Agency (UBA) performed a stream mesocosm study in order to investigate the effects of a realistic spraying sequence, as applied to apple crop in the fruit growing area of Lake Constance in southern Germany, on an aquatic stream community for 150 days. Application data of a spraying sequence originally applied in 2010 were kindly provided by the Julius Kuehn Institute (Braunschweig, Germany), which documents the extent of spray series application of PPPs on conventional crop on the basis of annual surveys [22, 23].

Since pesticide concentrations in edge-of-field surface waters should not exceed the RAC for each AI, RACs were chosen to simulate pesticide entries in stream mesocosms due to apple crop spray treatments. The use of a RAC can be seen as a realistic approach since it was derived applying a safety factor that accounts for uncertainties such as extrapolations to other species or from laboratory test conditions to field situations [24, 25]. However, the occurrence of adverse effects as well as the regular exceedances of RACs in the field does frequently raise the question whether a series of PPP spraying events leading to concentrations that correspond to the RAC values (i.e. a sequence of RACs in time) would still be protective for stream communities [18, 26, 27]. Therefore, we hypothesised that the successive application of PPP, each at its RAC, within the spraying sequence of apple crop leads to adverse effects on the stream community. This hypothesis is based on the fact that the issue of multiple PPP applications is not explicitly considered in RAC derivation. Thus, there are some uncertainties in terms of the protection level ensured in the field. To test the hypothesis, the conducted mesocosm study focussed on endpoints such as species diversity, abundance of benthos, periphyton, macrophytes, emergence of insects, as well as physico-chemical water parameters, and additionally on behaviour (drift) of aquatic macroinvertebrates.


Stream mesocosm set-up and biological establishment

The study was conducted in eight indoor stream mesocosms at the field station of the UBA ( in Berlin (Germany). Eight flow-through circular stream mesocosms (length 74 m, width 1 m, water volume 20.5 m3, glass-fibre reinforced polyester material), each equipped with riffle sections and four separate pool sections per stream were set up [28]. The riffle sections were established with a sand bottom. The deeper pool sections were filled additionally with a 7-cm middle layer of a sand and natural sediment mixture (3:1) from lake Duckwitz (N 53.593894, O 12.345664; Mecklenburg-Western Pomerania, Germany) covered by a 5-cm top sand layer. The systems were filled with a mixture of treated ground water (de-ionised) and deionised water up to a water depth of 27 cm from the UBA field station´s own waterworks (conductivity approx. 500 µS−1 cm). Water flow was adjusted to 0.1 m s−1 and water loss due to evaporation was compensated by adding deionised water. For further details, see Mohr et al. [28, 29]. A sketch of the mesocosm set-up is shown in Additional file 1: Fig. S1. The systems were illuminated with fluorescent tubes (OSRAM LF72) providing a mean photosynthetically active radiation of about 50 and 120 μmol m−2 s−1 on the water surface of the riffles and pools, respectively. The light regime followed monthly the seasonal daylight length.

One year before the start of the spraying sequence, the macrophyte species Elodea canadensis, Potamogeton natans, Myriophyllum spicatum, and Sparganium erectum were planted in the four pool sections as structural elements in the streams. Eight weeks before the first PPP application (start of the experiment), equal aliquots of a periphyton suspension obtained from a mesotrophic lake (Britzer Garten, N 52.4332, E 13.4160, Berlin, Germany) were added to each stream for inoculation. To ensure sufficient nutrient supply for the growth of macrophytes and periphyton, trace elements and nutrients were added in equal concentrations to each of the streams before and during the experiment: 1 × mixture of trace elements (0.1 mg L−1 H3BO3, 0.02 mg L−1 MnCl2*4H2O, 0.001 mg L−1 Na2MoO4* 2H2O, 0.011 mg L−1 ZnCl2, CuCl2* 2H2O, 0.084 mg L−1 FeCl3*6H2O), 5 × phosphate (0.02 mg L−1 PO4-P), 5 × nitrate (1 mg L−1 NO3-N), and 9 × silicate (4 mg L−1 Si). Given values for addition refer to water concentrations in the mesocosms).

In order to stock the stream mesocosms with benthic macroinvertebrates, attraction devices consisting of polyethylene net bags (32 cm × 20 cm, mesh opening 4 mm) were filled with 100 g of loose organic triticale straw (Eco-village Brodowin, N 52.5439, E 13.5741, Germany). Bags were exposed in three different reference streams for two weeks and subsequently transferred to the stream mesocosms (with an equal number of bags per stream) one to two weeks prior to the start of the experiment (for details see supplemental information Additional file 1: Table S1). As reference sites, the sub-lacustrine creek Barolder Fließ (N 51.9912, E 14.2209, South-Brandenburg, Germany), the stream Lieberoser Mühlenfließ (N 51.9864, E 14.3522, South-Brandenburg, Germany), and the ditch Lietzengraben (N 52.6551 E 13.4741, North of Berlin, Germany) were chosen in order to guarantee a broad variety of benthic macroinvertebrate species. The ecological condition of the reference streams was characterised as good to moderate [30]. After 2–3 days of acclimatisation in the stream mesocosms, straw bags were opened to allow the loose straw to accumulate into 10-cm-deep cross-drains, which were evenly distributed over the sediment bottom acting as hiding spots and further detrital based food supply. Further details can be found in Mohr et al. [29]. In addition, macroinvertebrates were collected from the creeks Rabener Plane (N 52.0463, E 12.5719, South-Brandenburg, Germany) and Bardenitzer Fließ (N 52.0675, E 12.9320, South-Brandenburg, Germany) by means of two polyethylene net bags (38 × 65 cm, mesh size 4 mm), which were filled with ca. 300 g of dried leaves of Alnus glutinosa and exposed for two weeks. Macroinvertebrates were then separated from leaves using a sieve (10 mm mesh) and divided into nine subsamples. Each stream mesocosm was stocked with one subsample and the ninth subsample was immediately fixed in 96% ethanol as a reference sample for initial stocking. Furthermore, the mesocosms were colonised with different species of Trichoptera larvae and Asellus aquaticus collected by dip nets or hand from the stream Fredersdorfer Mühlenfließ (N 52.5675, E 13.7890, Brandenburg, Germany), as well as with Cloeon dipterum and different snail taxa (Theodoxus fluviatilis, Potamopyrgus antipodarum, Gyraulus sp., Physella sp., Lymnaea stagnalis) from outdoor stream and pond mesocosms at the UBA field station. Details of stocking dates and number of introduced organisms are given in Additional file 1: Table S1.

Eleven weeks (day 77) after the experimental start with first PPP application on day 0, all stream mesocosms were colonised one more time with eight straw bags per system, which had previously been exposed for two weeks in the Lieberoser Mühlenfließ in order to compensate for emerged insects and to simulate re-colonisation with the next generation of insect larvae. For each stocking event, three bags were immediately fixed in 96% ethanol to determine the initial macroinvertebrate community in the mesocosms (Additional file 1: Table S1).

Spray sequence scenario

The original apple crop application scheme, which was provided by Julius Kuehn Institute, comprised 25 application days in the period from March to September 2010 with a total of 23 different PPPs with 21 AI. Fungicidal AIs were used 39 times, insecticidal AIs 13 times, and herbicidal AIs four times. The spray sequence tested in this study was slightly adjusted by excluding a virus, a pheromone, an antibiotic, and a growth regulating product. In addition, the sequence was shortened due to technical and organisational reasons by considering that all remaining PPPs were applied at least once. Thus, 17 PPPs composed of 17 different AIs (nine fungicidal, four herbicidal and four insecticidal AI) were applied on 19 spraying event days in the period from April to July 2015. Information on all applied PPPs, their AIs, mode of action, the producers, used RACs, and sampling dates are listed in Table 1 and Additional file 1: Table S2. Four stream mesocosms were used as treatments by applying the specific RAC of each PPP and by chronologically following the original application sequence as close as possible. The four remaining stream mesocosms served as controls. Control and treatment stream mesocosms were allocated randomly.

Table 1 Concentrations and recovery of applied pesticides for active ingredients and application dates

Application of plant protection products

The procedures of stock solution preparation are described in Additional file 1: Sect. 1. For a precise application of low pesticide concentrations, high precision micro-dosing rotary piston pumps (IP65 MCP-CPF with pump head FMI205 QP.Q1.CSC/9003, Ismatec, Switzerland), equipped with ceramic pistons and cylinders, PTFE seals, stainless-steel tubes, and fittings were used. During application, pesticide solutions were permanently stirred to avoid separation or sedimentation inside the flask before entering the pump. For rapid mixing of applied PPP over the whole stream water body, the duration of the application was adjusted to one complete water rotation in the circular streams (about 12 min). The control streams were dosed with the same volume of tap water by use of a watering can.

Sampling and analysis of active ingredients

Water sampling for pesticide analysis in all streams was carried out at the beginning and end of the study, to monitor all applied AIs (except dodine). After each application event, water sampling started at t = 240 min using stationary fixed rotary piston pumps automatically filling up 1-L samples of stream water during one water rotation in the circular streams. When necessary, additional buffer solution was added. When a PPP was applied on several application dates, additional sampling shortly before re-application was carried out to assess, if there were any residues from the previous application. All samples were transported directly to the laboratories in dark and cool conditions. The AIs were analysed by Eurofins Sofia GmbH (Berlin, Germany) and two chemical laboratories of the UBA. A detailed description of special PPP solution preparations, sampling, and analysis can be found in Additional file 1: Section 1.

The RACs of the AIs were used as references for both, dosing solutions and stream water, concentrations to calculate chemical recovery (Table 1). In the case of the second application of penconazole, the actual concentration was corrected, based on the measured preload concentration. Specific calculational conversions were made in case of copper-oxychloride, glyphosate-IPA, and MCPA-DMA, because analytes (Cu, glyphosate, and MCPA) differed from the AI in the PPP (e.g. Cu-oxychloride). For dodine, no water analysis was carried out, since rapid adsorption on all organic surfaces in the stream mesocosms was expected, due to its high soil partition coefficient Kd (range 2,200 to 18,000 kg L−1) and no sensitive analytical method was available at the time.

Standard water quality parameters

Standard water parameters namely oxygen, pH, electrical conductivity, and water temperature were measured weekly with a multi-parameter probe system (Multi 3430, probes FDO 925, SensoLyt 900, TetraCon 925; each WTW, Germany). For nutrient and major ion analysis, water samples were taken every two weeks. Water samples were filtered (0.45 µm, cellulose-nitrate filters, Schleicher & Schuell, Germany) for further analysis. The major ions fluoride, chloride, bromide, sulphate, lithium, sodium, potassium, magnesium, and calcium were analysed using standard methods (TitrIC-system, Metrohm, Switzerland) [31,32,33]. Concentrations of o-phosphate, ammonia, nitrite, nitrate, and silicate were quantified photometrically using continuous flow analysis (San + + , Skalar, Netherlands) [34,35,36,37]. Dissolved organic carbon was analysed according to DIN EN 1484 [38] as non-purgeable organic carbon by catalytic combustion at 680 ℃ and NDIR-detection by use of a TOC 5000A with ASI 5000A (Shimadzu, Japan) with 3-point calibration and sample injection.

Periphyton and macrophyte sampling and analysis

For periphyton sampling, each stream was equipped with a special rack consisting of a stainless-steel frame holding 56 microscopic slides (76 × 52 mm, soda-lime glass), which was adjusted in parallel to the water current at 13 cm below water surface. For uniform illumination, eight rows of LED strip lights (neutral white; ~ 122 µmol m−2 s−1 PhAR at water surface; Co. Nobilé, Germany) were mounted under a stainless-steel flat roof.

For periphyton sampling, four colonised glass slides were taken at each sampling date (11 sampling dates in total). Prior to sampling, the slide positions were chosen randomly and the chosen positions were then taken out equally in all streams. The periphyton on both sides of a slide was scraped and rinsed off with a plastic Japan spatula and tap water. Periphyton of all four slides per stream was merged in one sample representing a total surface area of 237.12 cm2. The suspension was transferred to a glass bottle and filled with tap water up to a volume of 1 L for further analysis.

Species identification of periphyton samples preserved with Lugol’s solution or formalin was done by use of light microscopy with differential interference contrast (BX51, Olympus, Japan) using standard preparation techniques [39] and latest identification keys. Species specific cell or filament volumes were given as size classes based on microscopic size measurements and by use of corresponding geometric formulas. Biovolume was calculated as the product of species number and its specific specimen volume. Development of the macrophyte P. natans was monitored once a week by counting all floating leaves in each stream, starting 10 days before and ending 151 days after the first application event.

Macroinvertebrates sampling

Determination of macroinvertebrate species and their abundances in the stream mesocosms was carried out using traps stuffed with different substrates in order to attract diverse taxonomic groups. Straw (7 g dry weight) and small ceramic tubes (Eheim, Germany; 400 g) were filled into two polyethylene net bags (13 cm × 17 cm, 3 × 4 mm mesh size), respectively. For each sampling, two benthos traps were placed at two positions in each stream mesocosm for colonisation. After two weeks, the traps were sampled and rinsed with tap water over a sieve (200 µm mesh size) until all organisms were washed out and then stored in ethanol (80%) for species determination.

Emergence samples were taken at four different locations of each stream as described in Mohr et al. [29]. Each emergence trap covered a water surface area of 1 m2 (polyethylene, 0.5 mm mesh size). Starting four weeks before the first application of PPP, emergence of merolimnic insects was quantified weekly for a period of 24 weeks by emptying and refilling the plastic beaker with 30 ml of “Renner fixation solution” (mixture of ethanol, distilled water, glycerine, acetic acid, and detergent) on the open top of the roof-like trap.

Macroinvertebrate drift was measured as described by Berghahn et al. [40]. Briefly, 30 min after pesticide dosing, two drift nets (size of the opening: 15 cm × 7.5 cm; mesh size of the funnel: 283 µm; total length: 140 cm) were placed in the middle of the stream bottom above the sediment surface (distance between nets approximately 30 m) for 4 h. Furthermore, night drift was measured employing drifts nets for 14 h (6 pm to 8 am). Drift was measured on three application dates with estimated high toxic pressure for benthic macroinvertebrates: (1) day 55: application of dithianon, penconazole and fenoxycarb; (2) day 66: application of sulphur, dithianon and spirotetramat; and (3) day 83: application of sulphur, dithianon, fluquinconazole and pyrimethanil (Tables 1, 3). Macroinvertebrates caught in the drift nets were preserved in 80% ethanol for further evaluation.

Taxonomic identification of macroinvertebrates (samples of initial stocking, sampling and emergence traps as well as drift net catches) were performed using magnifying glasses and a stereomicroscope (Stemi 2000-C; Zeiss, Germany). All macroinvertebrates were identified to lowest practical taxonomic level (mostly genus or family level, see Additional file 1: Table S1) using the latest identification keys (see Additional file 1: section 2). In the case of samples gained from emergence traps, male chironomids were identified to subfamily level, whereas female chironomids were pooled.

Statistical analysis

All statistical analyses were conducted using the statistical software package R Development Core Team (R version 3.6.1) [41] with the additional packages vegan [42] and mvabund [43]. Generalized linear models for multivariate data (GLMmv) as described in Szöcs et al. [44] were used to evaluate the effects of multiple exposures of low-level applications of pesticides (control vs. treatment) on the temporal development of major ions (i.e. water concentration of measured ions), periphyton community (i.e. biovolume of identified periphyton taxa), macroinvertebrate community (i.e. density of identified invertebrate taxa), and cumulative emergence of insects (i.e. density of identified taxa of emerging insects). Treatment effects were expressed and visualised as deviations from the control (y = 0) over time (x-axis). A general treatment effect considering the whole experimental duration was analysed by comparison of the model including the time × treatment interaction to the model with time as sole variable. Significance level was set to α = 0.05 and level for a tendency to α = 0.1. Since the GLMmv analyses only potential effects on the multivariate level, also species weights of the GLMmv model (as percentage contribution to the deviance of single taxa, Table 2) were used as tool for the selection of taxa used for further analyses. In order to evaluate potential effects of PPPs into more detail, all taxa/parameter with species weight greater than 5% were used for further univariate analysis. As density data are usually skewed and widespread, log(x + 1) transformation was applied before further analysis [45]. Differences in the development of the selected taxa over time between control and treatment were analysed using generalized linear models (GLM, error distribution = Gaussian). If overdispersion was detected using the function dispersiontest within the R package AER [46], the error distribution 'quasi-Poisson' was used to overcome overdispersion. In case of a significant treatment effect (control vs. treatment) without a significant time × treatment effect, additionally sampling date(s) before the first application of a pesticide were compared. This was done in order to exclude that a significant difference between control and treatment already occurred before the first application of PPP for the respective cases (cf. Table 2). In doing so, a Mann–Whitney test (in case of only one sampling date before first PPP application) or a GLM (in case of more than one sampling date before first PPP application) was used to evaluate if start conditions were comparable between control and treatment replicates. For all relevant cases, no significant difference was found before the first application of PPP (p > 0.05). We further analysed the variation of the development of the selected taxa between control and treatment replicates. For each sampling date and taxon, the variation was calculated as the absolute difference between the mean value and single value per replicate of the control and treatment mesocosms (non-transformed data), respectively. Statistical analyses were then carried out as described above for the analyses of the development of the selected taxa over time.

Table 2 Statistical analysis of the development of ions, periphyton, benthos, and emerging insects as mean values and variation for single taxa over time between control and treatment

Similarity of drifting invertebrates for each sampling date was analysed by non-metric multidimensional scaling (nMDS). When analysing Bray–Curtis similarities, which compare ranked similarities for differences between defined groups, square root transformation of abundance was used to reduce the effect of dominant taxa [47]. Differences in invertebrate drift were determined using PERMANOVA (permutational analysis of variance), which analyses multivariate data on the basis of distance measurements using permutations [48, 49].

Species weight (as percentage contribution) of single species to the differences on the community level in drifting community between control and treatment was estimated using SIMPER [42]. We selected the three species contributing the most to the differences in drift between control and treatment for further analyses. For these three species and for the development in the number of floating leaves of the macrophyte Potamogeton natans the same univariate statistical procedures as described above for standard benthos samples were used.

Toxic units

The toxic units (TU) approach was applied in order to describe the “toxic pressure” emanating from the pesticide concentrations dosed in this study. TU is commonly defined as the ratio between the concentration and a defined effective concentration as determined by ecotoxicity testing of the respective compound (e.g. EC50). Hence, a toxic unit of 1 for a given compound would describe the exposure concentration, at which there is a 50% effect when dealing with e.g. an acute EC50 of a certain biological endpoint. The TU approach can be used to calculate the toxicity of mixtures for different organisms [2], assuming concentration–additive behaviour of mixture components as default to obtain a theoretical “mixture toxicity indicator”. In this study, the TU approach was used to describe the theoretical toxicity as overall impact of the successive applications of the PPPs on the main biological components (periphyton, benthos) of the overall species community:

$${\text{sumTU}} = \mathop \sum \limits_{i\, = 1}^{n} \frac{{{\text{ci}}}}{{{\text{ECxi}}}}.$$

The calculation of TU (TU for each AI) was performed with “ci”, which refers to the nominal exposure concentration (RAC for each AI tested) in the stream water at the start of application, corrected by the measured concentration in the dosing solution. For TU, the exposure concentration at the time of application was thus divided by the 48 h LC50 from acute Daphnia test and the 72 h EC50 from algae test (as given in EFSA conclusions and lists of endpoints for the respective AIs; Both assessments were made in parallel. The sum of toxic units (sumTU) was calculated weekly as a measure for toxic pressure over time, which means that TUs can contain PPPs of several application dates. TU for PPPs containing sulphur were not included in the analysis since sulphur is considered toxic to sediment organisms only.


Analytical recovery of active ingredients in stream mesocosm water following the spray sequence scenario for apple crop

Overall, achieving similar and very low AI concentrations corresponding to the RAC (some as low as 0.16 µg L−1) for 17 PPPs with different physico-chemical properties in the water of the four treated stream mesocosms was challenging. Analysis of stream water samples, taken directly after PPP application, disclosed that 22 of 35 analysed AIs were within the recovery range with less than 20% deviation from nominal concentrations. Hence, the desired concentrations (RACs) for these PPPs were reached (Table 1). However, a deviation of more than 20% above or below RAC was found for 13 AI in stream water. This was the case for some hydrolytically unstable AIs (dithianon, captan) and for highly adsorptive AIs (copper oxychloride, sulphur). For these substances, target concentrations refer to the respective dosing concentration (Table 1). In case of dodine, dosing solutions reached the RAC in most cases, but recovery in stream water was not analysable (see method section). For the substance dithianon, it was not possible to achieve concentrations corresponding to the RAC in most of the dosing solutions. Stream water samples were below limit of quantification of dithianon (Table 1).

Before the first PPP application, analytical screening of all organic AIs in the eight stream mesocosms revealed that background concentration levels were always below the limit of quantification. Only three of the organic AIs were found in water samples of the four treatment replicates at the end of the study after 144 days: dimethenamid with 0.15 µg L−1, penconazole with 1.7 µg L−1, and methoxyfenozide with 1.8 µg L−1. Possibly these AIs led to higher chronic stress responses than the more short-lived AIs.

Effects on physico-chemical water parameters

During the experimental period, water temperature increased from 12.7 ℃ (day-28) in April to 18.5 ℃ at the end of May (day 48), and varied between 19 to 24.5 ℃ until the end of August (day 145). Data of physico-chemical water parameters for each sampling date are summarised in Additional file 1: Tables S3, S4; and Additional file 1: Fig. S2. GLMmv analysis of the major ion composition (i.e. concentration of the analysed ions per replicate and sampling date) in the stream water of control and treatment was based on 13 single ions and alkalinity as sum parameter (Fig. 1A). Considering the whole experimental duration, no general treatment effect was found (GLMmv, p = 0.22), but significant differences in the treatments were found on sampling date day 88 (p = 0.024) and day 101 (p = 0.013; Fig. 1A). Calculation of the contribution of single parameters to differences between control and treatment (as proportion of the deviance of a respective ion to the overall deviance) revealed eight ions with a contribution of more than or equal to 5% (nitrate, nitrite, ammonium, silicate, bromide, sodium, potassium, and phosphate), but none of these single parameters showed a significant difference between control and treatment (Table 2, Additional file 1: Fig. S2).

Fig. 1
figure 1

Visualisation of results from generalized linear models for multivariate data (GLMmv) of the community of A ion composition [i.e. water concentration of measured ions]; B periphyton [i.e. biovolume of identified periphyton taxa]; C benthic invertebrates [i.e. density of identified invertebrate taxa], and D emerging insects [i.e. density of identified taxa of emerging insects]. Open circles indicate sampling dates with significant differences (α = 0.05) in the treatment mesocosms from the controls. Grey circles indicate a level for a tendency to α = 0.1. Day 0 represents the day of the first application of a pesticide. Vertical dashed lines indicate the start (left) and the end (right) of the PPP application period

Effects on periphyton and macrophytes

During the experimental period, 34 taxa of primary producers were identified in the periphyton biofilm and then grouped to analyse effects of the applied pesticide spraying sequence on periphyton biovolume. A significant general treatment effect (i.e. biovolume of identified periphyton taxa per replicate [control vs. treatment] and sampling date) was found over the whole experimental period (GLMmv, p = 0.043, Fig. 1B). Furthermore, GLMmv analysis revealed a tendency for differences on day 75 (p = 0.078) and significant differences on day 18, day 89, day 103, and day 117 (p < 0.05, Fig. 1B) in periphyton community composition.

Six periphyton species contributed to differences between control and treatment with more than 5% (Table 2). A significantly higher variation in biovolume for the species Synechococcus spp. and Chaetophorales was found over time in the treatment mesocosms, compared to the controls (Table 2, Fig. 2A-B). In turn, for Fragilaria fasciculata/acus and Nitzschia spp., there was a significantly higher variation in the control mesocosms.

Fig. 2
figure 2

Development of biovolume of the periphyton taxa A Synechococcus spp; B Chaetophorales; development of abundance of the benthic invertebrate taxa; C Turbellaria; D Gammarus spp; cumulative emergence of the merolimnic insect taxa; E Baetidae; F female Chironomidae during the experimental period. Grey and white boxplots represent treatment and control, respectively. Day 0 represents the day of the first pesticide application. X-axes are with different scaling for better visual resolution

Since macrophytes standing stock should not be disturbed, only effects on the non-destructive endpoint leaf development of P. natans were analysed. At the start of the experiment in spring, P. natans shoots had not yet developed floating leaves (Fig. 3). The number of floating leaves increased over time and was significantly higher in the control compared to the treatment (GLM, treatment: p = 0.016, treatment × time interaction: p < 0.001), indicating a higher net production rate of control macrophytes.

Fig. 3
figure 3

Development (mean ± SD) of floating leaves of Potamogeton natans during the experimental period. Grey and white points represent treatment and control, respectively. Day 0 represents the day of the first pesticide application

Overall, the toxic pressure of specific PPP applications expressed as sumTU (Fig. 4) for each application date could not be related to effects on the periphyton community (Fig. 1B) nor to effects on P. natans (Fig. 3).

Fig. 4
figure 4

Sum of toxic units (sumTU) over time for primary producers based on the 72 h EC50 from algae test (dark bars) and invertebrates based on the 48 h EC50 from acute Daphnia test (bright bars). SumTU values are given as weekly average and consider the concentration at the time of PPP application (Note: RAC corrected by recovery of dosing solution; values for sulphur are excluded since they are considered as toxic toward sediment organisms only.)

Effects on the macroinvertebrate community

For statistical analysis, the identified invertebrate taxa from benthos traps were grouped to higher taxonomic levels. For the resulting 16 taxa, effects of the pesticide spraying sequence were analysed. No general treatment effect (i.e. densities of identified macroinvertebrate taxa per replicate [control vs. treatment] and sampling date) was found for the whole experimental time (GLMmv, p = 0.41, Fig. 1C). Furthermore, GLMmv analysis revealed a significant difference in the community (i.e. densities of identified macroinvertebrate taxa as described above) between the treatment and the control at the end of the study on day 131 (p = 0.049, Fig. 1C).

Ten taxa of benthic macroinvertebrates contributed to the differences between controls and treatments with more than or equal to 5% (Table 2). The variation in number of Orthocladiinae, Trichoptera (Table 2), and Turbellaria (Fig. 2C), were higher in the treatment mesocosms, while the variation in the number of Gammarus spp. showed a significant increase in the treatment mesocosms over time (Table 2, Fig. 2D).

Drift of invertebrates

Drift of benthic macroinvertebrates was measured after three application dates. A significant treatment effect was detected on the community level only for the pesticide application event on day 83 (sulphur, dithianon, fluquinconazole and pyrimethanil; PERMANOVA, p = 0.03, Table 3), but no significant differences in drift of the three taxa Tanypodinae, Gammarus spp., and Baetidae, which contributed the most to the community effect, were found (Table 3).

Table 3 Statistical analysis of drifting invertebrates for day and night drift catches (time) after three different application events

Emergence of aquatic insects

In order to analyse effects of the pesticide spraying sequence on the emergence of aquatic insects, hatched insects were grouped into 18 taxa. No general treatment effect (i.e. densities of identified taxa of emerging insects per replicate [control vs. treatment] and sampling date) was found for the whole experimental duration (GLMmv, p = 0.32, Fig. 1D). Furthermore, GLMmv analysis revealed significant differences in the community composition of emerging insects at the first sampling date of the experiment before the first pesticide application at day-30 (p < 0.05, Fig. 1D) and in the treatment on day 75, day 102, and day 124 (p < 0.05). At the first sampling date (day-30) only few individuals were found in the emergence traps indicating stronger random effects.

Ten taxa of emerging insects with a contribution of more than 5% to differences between control and treatment were identified (Table 2). None of the ten taxa revealed a significant difference in the cumulative number of emerging insects in the treatment over time, when compared to the control. However, variations in the cumulative emergence of some chironomid taxa (i.e. female Chironomidae, male Orthocladiinae, male Chironominae), and cumulative emergence of all taxa (sum) showed increasing variation over time in the treatment (Table 2). Furthermore, cumulative emergence of Baetidae and female Chironomidae displayed increasing variation over time in the treatment (Fig. 2E, F) and the variation in cumulative emergence of male Tanypodinae was significant (Table 2).

Toxic pressure over time

The TU approach showed relatively high toxic pressure on primary producers during the first part of the spraying sequence (weeks 1 to 4; Fig. 4). SumTU exceeded a value of 0.1 (1/10th EC50) during week 1, 3, and 4 (Fig. 4), which corresponds to the time of application of the two fungicides copper and dodine, both being more toxic to primary producers than to invertebrates and delivering rather high sumTUs than the four herbicides applied during the second part of the experiment. In addition, in week 12 and 18, sumTU was also above a value of 0.1.

The TU approach also indicated a toxic pressure on invertebrates in the first part (weeks 1 to 6) of the experiment (Fig. 4). However, the overall toxic pressure exerted on invertebrates was much lower than on primary producers with a maximum sumTU value of 0.035, at no time exceeding a value of 0.1.


Effects of the apple crop spraying sequence on the stream mesocosm community

The applied pesticide concentrations on the basis of regulatory acceptable concentrations (RAC values) following an apple crop spraying sequence scenario led to significant differences between control and treatment streams over time, especially towards the end of the study. This was the case for the endpoints: variation in biovolume of selected periphyton species, abundance and variation of abundance of selected benthic macroinvertebrates and insect emergence, as well as the number of floating leaves of P. natans (Figs. 1, 2 and 3, Table 2). Changes in physico-chemical parameters can be used as proxy of the sum of all photosynthetic and respiratory activities and therefore act as an indicator for changes in the ratio of respective processes. In our experiment, it was not possible to identify a relationship between specific PPP pulses and their associated toxic pressure exerted over time and significant biological and/ or ecological effects. Indeed, the TU approach relates to exposure concentrations at the time of treatment and acute ecotoxicity (from laboratory testing, i.e. under standardised laboratory conditions, Fig. 4). Thus, its predictive power is limited since it does not consider (i) potential long-term or delayed effects of fast dissipating substances or chronic effects of substances persisting in the water (e.g. copper) nor (ii) effects of species interactions and additional environmental stressors exerted on field communities and uncertainties related to more sensitive species than Daphnia and algae. Nevertheless, the community response as shown in Figs. 13 indicates that the repeated low-level pesticide applications resulted in measurable cumulative stress for several species and groups, confirming the findings of other studies. In a mesocosm study conducted by Polazzo et al. [50], field-relevant peak exposure concentrations of pesticides resulted in a changed community composition of zooplankton only at the end of the experiment after 50 days. Our results are also in line with the review by Altenburger et al. [6] about mixture effects of substances with different modes of action at concentrations below single species effect thresholds.

The most striking result of the present study was the statistically significant increase in variation of population response parameters of selected taxa in the treatment mesocosms compared to the controls (Table 2, Fig. 2). On the one hand, variation in species responses has always been a big issue in ecotoxicology and species tests [51,52,53], as it hinders the statistical evaluation of true effects from background noise with regard to mean values. On the other hand, there is increasing evidence that inter-replicate variation of population development parameters as a result of genotype plasticity of individuals, can be an indicative endpoint for chemical pollution [54]. In general, ecotoxicological tests are designed to minimise the variation of possible influencing factors, other than the stressor under assessment [55]. In mesocosm studies with multiple populations of different trophic levels acting together, the likelihood of facing significant variation in species responses is rather high [56]. In the present study, data variation of most endpoints at the beginning of the experiment was low in the control and treatment mesocosms. This indicates that the actions taken to synchronise the community development in the mesocosms by a thorough and homogenous biological establishment were successful. Therefore, the increased variation in abundance of selected species observed in the treatment mesocosms might be considered as an effect of the repeated PPP pulses rather than a general mesocosm variation.

The observed variation in species response to the 19 times repeated low-dosed PPP application might have been induced by genotype plasticity of respective species. The higher the genotype plasticity of species or communities, the more complex the responses will be in terms of physiological adaptations such as detoxification processes or increased escape responses [53]. In addition, small differences in concentrations of some pesticides (see standard deviation of recovery of AI in streams; Table 1) at the start of the experiment may have further promoted the variation in species responses. The slightly different pesticide concentrations in the four treatment mesocosms may have led to different individual stress responses, causing cascadic effects in the populations. This so-called butterfly effect has been described as the sensitive dependence on initial conditions, in which a small change in one state of a system can result in large differences in a later state [57]. In the present study, the concentration differences might have caused indirect or hormesis effects on the species communities indicated by higher abundance values in the treatments (Fig. 2). Especially hormesis effects have been shown to be trigged by very low doses of a range of chemicals and physical agents [58, 59]. In a review by Agathokleous [60], numerous plant and animal species responses to various stressors have been linked to hormesis at low stressor concentrations.

Overall, the increased variation in species response in the treatment mesocosms of the present study should be discussed as a general disturbance measure [53, 61], even if it was not possible to link the results to specific pesticide pulses, to genotype plasticity of the species or the differences in exposure between replicates. When assessing population health, e.g. in the risk assessment of pesticides, the parameter “inter-replicate variation” should also be considered. Studies with high variation should not per se be considered as unreliable, as this parameter might in fact reflect system resistance and resilience against perturbations or it might serve as an early warning sign of potential disturbances of communities in the field [62,63,64,65,66]. Hence, such results of inter-individual variation can also be seen as valuable source of information for the risk assessment [61].

Relevance of chosen apple spray scenario and implementation for the risk assessment

In Germany, 48.8% of the fruit cultivation area was used for apple production in 2008 [22], and in 2017 about 10 Mio tons of apples were harvested from approx. 473.550 ha [67]. The worldwide production of apples was 83.1 Mio tons in 2017 [68]. Apple orchards are regularly treated with pesticides, often only for a high-quality visual appearance of the fruits to increase the sales value. Fungicides represent the most dominant pesticide group with more than 80% of AI applications per growing season [8, 69, 70] used mainly to prevent apple scab and other fungal infections. Captan and dithianon have been listed as two of the most frequently used fungicides [8, 71] and were also applied several times in the spraying sequence of the present study. In a survey on application of chemical pesticides in apple farming, Roßberg and Harzer [71] reported that the treatment frequency index (TFI), which quantifies the number of registered PPP applied to apple crop, was 28 TFI in 2001 and increased continuously to 33 in 2013. In France, between 2006 and 2008, the mean TFI was 35 for conventional and 26 TFI for organic apple orchards with scab-susceptible apple cultivar [72]. For the shortened apple spraying scenario used in this study, 19 treatments with 17 different AI were applied, which can be considered as a “light to moderate” case scenario for standard practice in Germany [8, 71]. Thus, potential effects on natural communities in streams and rivers in agricultural areas facing multiple stressors in addition to the PPP spraying scenarios such as other chemicals, habitat degradation, and nutrient stress over the years, might be more pronounced [e.g. 73]. Also, the exposure concentrations limited to the RAC values used in this study may not represent a realistic exposure scenario of pesticide entries in rural areas. Liess et al. [27] identified an exceedance of RAC values for at least one pesticide concentration in agricultural streams in Germany for 81% of the water samples tested.

The present study reinforces mounting evidence, that the current environmental risk assessment in the context of the authorisation of PPP may not be safe for ecosystems and their communities, when facing multiple pesticide entries of spraying sequences applied year after year. Indeed, risk assessment is currently restricted since it refers to individual PPP. In other words, it is questionable whether the use of RAC values derived for individual PPPs are enough in view of the typical agricultural practice of applying pesticide tank mixtures and/or spray series. This issue has also been raised by, e.g. Weber et al. [74], Junghans et al. [18], or Covert et al. [75]. Unfortunately, experimental studies investigating effects of realistic spray sequences scenarios are still very rare since it is very difficult to get real application scheme data from farmers. In laboratory studies, Panico et al. [76] discovered strong effects to invertebrates in soils, facing mixtures of pesticides conventionally used on agricultural fields. To our knowledge, until now only three mesocosm studies have assessed complex realistic exposure to pesticides used on bulb, potato, and apple crop, respectively [77,78,79]. Wijngaarden et al. [77] and Arts et al. [78] revealed significant effects on pond/ditch communities at higher spray application rates, but concluded that risk assessment based on single pesticides seems to sufficiently protect freshwater ecosystems. Both studies had a different experimental approach using pesticide concentrations based on a realistic application and spray drift scenario, but the amount of pesticides and application times were lower than in the present study. Talk et al. [79] used a similar set-up as in this study, but focussed on fungal endpoints in pond mesocosms. The authors could not identify effects on fungal communities or leaf litter decomposition. Overall, more experimental studies applying realistic exposure scenarios are needed, even though implementing them is a challenge and very time consuming. An intensive field study by Schäfers et al. [26] investigating invertebrate communities in differently treated apple orchards in the German region Altes Land revealed that invertebrate communities were significantly affected by pesticide use, depending on exposure pressure. This was confirmed in a larger monitoring study in Germany showing that (i) 83% of agricultural streams did not meet the pesticide-related ecological targets; (ii) agricultural nonpoint source pesticide pollution was the major driving force in the reduction of vulnerable insect populations in aquatic invertebrate communities and (iii) the risk resulting from exposure to pesticide mixtures and to frequent successive pesticide applications needs to be better addressed [17, 27].


The repeated tiny dose makes the poison—meaning that the results of this mesocosm study provide further evidence for notable effects on aquatic stream communities following exposure to repeated low pesticide concentrations simulating a realistic spraying sequence for apple crop. Indeed it induced disturbances for some invertebrates and primary producers. Albeit the observed differences between control and treatment were rather small, our results support concerns that the current risk assessment for PPPs may not ensure a sufficient level of protection to field communities facing multiple pesticide entries from spraying sequences or other combined pesticide stress. The current framework for single PPP allows risk assessment refinements that claim for more realism. However, it ignores important aspects such as multiple exposure and effects of additional stressors, both natural and anthropogenic, which typically increase the sensitivity of organisms to pesticides [80, 81]. Therefore, it is still very challenging to realistically assess the risk. Obviously, the actual regulatory convention focussing on RAC values derived for individual AIs only, is not sufficient to ensure that no adverse effects on populations might occur in the field in view of multiple chemical (pesticides and other chemicals) exposures as well as further environmental stressors. As a consequence, legally binding protection goals may not be fulfilled. Applying for example an additional mixture assessment factor (MAF) in risk assessment, which has frequently been suggested [82], is one option to better address the issue of mixture exposure occurring in the field. Given the inherent and remaining uncertainties of any (however complex) risk assessment, our findings might support the importance of an ambitioned pesticide reduction strategy as recently brought forward in the European Green deal with the Farm-to-Fork strategy [83, 84], too.

Availability of data and materials

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.



Active ingredient


Effect concentration


Generalized linear models


Mixture assessment factor


Predicted environmental concentration


Plant protection product


Risk assessment


Regulatory acceptable concentration


Supplemental information




Toxic unit


German Environment Agency


  1. EC (2009) European Commission (EC) Regulation No 1107/2009 of the European parliament and of the council of 21 October 2009 concerning the placing of plant protection products on the market and repealing council directives 79/117/EEC and 91/414/EEC

  2. EFSA (2013) European Food Safety Authority Guidance on tiered risk assessment for plant protection products for aquatic organisms in edge-of-field surface waters. EFSA J 11(7):268

    Google Scholar 

  3. Brock TCM, Bhatta R, van Wijngaarden RPA, Rico A (2016) Is the chronic tier-1 effect assessment approach for insecticides protective for aquatic ecosystems? Integr Environ Assess Manag.

    Article  Google Scholar 

  4. EPRS (2019) European parliamentary research service. farming without plant protection products. In-depth analysis. panel for the future of science and technology. scientific foresight unit (STOA). PE 634416, European parliament.

  5. Frische T, Matezki S, Wogram J (2014) Environmental risk assessment of pesticide mixtures under regulation 1107/2009/EC: a regulatory review by the German federal environment agency (UBA). J Verbr Lebensm 9:377–389

    Article  CAS  Google Scholar 

  6. Altenburger R, Arrhenius Å, Backhaus T, Coors A, Faust M, Zitzkat D (2013a) Ecotoxicological combined effects from chemical mixtures. 1: relevance and adequate consideration in environmental risk assessment of plant protection products and biocides, Umweltbundesamt, p 219

  7. Altenburger R, Backhaus T, Boedeker W, Faust M, Scholze M (2013) Simplifying complexity: mixture toxicity assessment in the last 20 years. Environ Toxicol Chem 32(8):1685–1687

    Article  CAS  Google Scholar 

  8. Knillmann S, Liess M, Scholz-Starke B, B, D, Ottermanns R, Schäffer A, Sybertz A and Roß-Nickoll M (2021) Environmental risks of pesticides between forecast and reality: how reliable are results of the environmental risk assessment for individual products in the light of agricultural practice (tank mixtures, spray series)? Texte 82/2021, Project No. (FKZ) 3715 63 407 0 p. 196, Umweltbundesamt

  9. Süß A, Bischoff G, Mueller ACW, Buhr L (2006) Chemisch-biologisches Monitoring zu Pflanzenschutzmittelbelastungen und Lebensgemeinschaften in Gräben des Alten Landes. Nachrichtenbl Deut Pflanzenschtzd 58(2):28–42

    Google Scholar 

  10. Schriever CA, Liess M (2007) Mapping ecological risk of agricultural pesticide runoff. Sci Total Environ 384(1–3):264–279

    Article  CAS  Google Scholar 

  11. Malaj E, von der Ohe PC, Grote M, Kühne R, Mondy C, Usseglio-Polatera P, Brack W, Schäfer R (2014) Organic chemicals jeopardize the health of freshwater ecosystems on the continental scale. Proc Natl Acad Sci 111:9549

    Article  CAS  Google Scholar 

  12. Stehle S, Schulz R (2015) Agricultural insecticides threaten surface waters at the global scale. Proc Natl Acad Sci 112(18):5750–5755

    Article  CAS  Google Scholar 

  13. Knauer K (2016) Pesticides in surface waters: a comparison with regulatory acceptable concentrations (RACs) determined in the authorization process and consideration for regulation. Environ Sci Eur 28(1):13

    Article  Google Scholar 

  14. Liess M, Foit K, Schäfer RB, Liess H-D (2016) Predicting the synergy of multiple stress effects. Sci Rep 6:32965

    Article  CAS  Google Scholar 

  15. Lorenz S, Rasmussen JJ, Süß A, Kalettka T, Golla B, Horney P, Stähler M, Hommel B, Schäfer RB (2017) Specifics and challenges of assessing exposure and effects of pesticides in small water bodies. Hydrobiologia 793(1):213–224

    Article  CAS  Google Scholar 

  16. Szöcs E, Brinke M, Karaoglan B, Schäfer RB (2017) Large scale risks from agricultural pesticides in small streams. Environ Sci Technol 51:7378–7385

    Article  Google Scholar 

  17. Weisner O, Frische T, Liebmann L, Reemtsma T, Roß-Nickoll M, Schäfer RB, Schäffer A, Scholz-Starke B, Vormeier P, Knillmann S, Liess M (2021) Risk from pesticide mixtures—the gap between risk assessment and reality. Sci Total Environ 796:149017

    Article  CAS  Google Scholar 

  18. Junghans M, Langer M, Baumgartner C, Vermeirssen E, Werner I (2019) Ökotoxikologische Untersuchungen: Risiko von PSM bestätigt. NAWA-SPEZ-Studie 2017 zeigt Beeinträchtigung von Gewässerorganismen. Aqua Gas 99(4):26–34

    Google Scholar 

  19. Moschet C, Wittmer I, Simovic J, Junghans M, Piazzoli A, Singer H, Stamm C, Leu C, Hollender J (2014) How a complete pesticide screening changes the assessment of surface water quality. Environ Sci Technol 48(10):5423–5432

    Article  CAS  Google Scholar 

  20. Spycher S, Mangold S, Doppler T, Junghans M, Wittmer I, Stamm C, Singer H (2018) Pesticide risks in small streams-how to get as close as possible to the stress imposed on aquatic organisms. Environ Sci Technol 52(8):4526–4535

    Article  CAS  Google Scholar 

  21. Schreiner VC, Szöcs E, Bhowmik AK, Vijver MG, Schäfer RB (2016) Pesticide mixtures in streams of several European countries and the USA. Sci Total Environ 573:680–689

    Article  CAS  Google Scholar 

  22. Roßberg D (2013) Erhebungen zur Anwendung von Pflanzenschutzmitteln in der Praxis im Jahr 2011. J Kult 64(4):141–151

    Google Scholar 

  23. Dachbrodt-Saaydeh S, Sellmann J, Strassemeyer J, Schwarz J, Klocke B, Krengel S, Kehlenbeck H (2018) Netz vergleichsbetriebe pflanzenschutz Zwei-jahresbericht 2015 und 2016—analyse der ergebnisse der Jahre 2007 bis 2016 berichte aus dem Julius Kühn-institut. julius kuehn institute, federal research center for cultivated plants

  24. Chapman PM, Fairbrother A, Brown D (1998) A critical evaluation of safety (uncertainty) factors for ecological risk assessment. Environ Toxicol Chem 17:99–108

    Article  CAS  Google Scholar 

  25. Schäfer R, Liess M, Altenburger R, Filser J, Hollert H, Roß-Nickoll M, Schaeffer A, Scheringer M (2019) Future pesticide risk assessment narrowing the gap between intention and reality. Environ Sci Eur.

    Article  Google Scholar 

  26. Schäfers C, Hommen U, Dembinski M, Gonzales-Valero JF (2006) Aquatic macroinvertebrates in the Altes land, an intensely used orchard region in Germany: correlation between community structure and potential for pesticide exposure. Environ Toxicol Chem 25(12):3275–3288

    Article  Google Scholar 

  27. Liess M, Liebmann L, Vormeier P, Weisner O, Altenburger R, Borchardt D, Brack W, Chatzinotas A, Escher B, Foit K, Gunold R, Henz S, Hitzfeld KL, Schmitt-Jansen M, Kamjunke N, Kaske O, Knillmann S, Krauss M, Küster E, Link M, Lück M, Möder M, Müller A, Paschke A, Schäfer RB, Schneeweiss A, Schreiner VC, Schulze T, Schüürmann G, von Tümpling W, Weitere M, Wogram J, Reemtsma T (2021) Pesticides are the dominant stressors for vulnerable insects in lowland streams. Water Res 201:117262

    Article  CAS  Google Scholar 

  28. Mohr S, Feibicke M, Ottenströer T, Meinecke S, Berghahn R, Schmidt R (2005) Enhanced experimental flexibility and control in ecotoxicological mesocosm experiments—a new outdoor and indoor pond and stream system. Environ Sci Pollut Res 12(1):5–7

    Article  Google Scholar 

  29. Mohr S, Berghahn R, Schmiediche R, Hübner V, Loth S, Feibicke M, Mailahn W, Wogram J (2012) Macroinvertebrate community response to repeated short-term pulses of the insecticide imidacloprid. Aquat Toxicol 110–111:25–36

    Article  Google Scholar 

  30. LFU (2015) Landesamt für Umwelt Brandenburg. managementplanung natura 2000 im land brandenburg, 2015. managementplan für das gebiet “dobberburger mühlenfließ” landesinterne melde Nr. 661, EU-Nr. DE 4051–302. Accessed 19 Jan 2023

  31. DIN En ISO 10304–1 (2009) Determination of dissolved anions by liquid chromatography of ions—part 1: determination of bromide, chloride, fluoride, nitrate, nitrite, phosphate and sulfate (ISO 10304–1:2007). German version EN ISO 10304–1:2009

    Google Scholar 

  32. DIN En ISO 14911 (1999) Determination of dissolved Li+, Na+, NH4+, K+, Mn2+, Ca2+, Mg2+, Sr2+ and Ba using ion chromatography—Method for water and waste water (ISO 14911:1998). German version EN ISO 14911:1999

    Google Scholar 

  33. DIN En ISO 9963–1 (1996) Determination of alkalinity—part 1: determination of total and composite alkalinity (ISO 9963–1:1994). German version EN ISO 9963–1:1995

    Google Scholar 

  34. DIN En ISO 11732 (2005) Determination of ammonium nitrogen - Method by flow analysis (CFA and FIA) and spectrometric detection (ISO 11732:2005). German version EN ISO 11732:2005

    Google Scholar 

  35. DIN En ISO 13395 (1996) Determination of nitrite nitrogen and nitrate nitrogen and the sum of both by flow analysis (CFA and FIA) and spectrometric detection (ISO 13395:1996). German version EN ISO 13395:1996

    Google Scholar 

  36. DIN En ISO 15681–2 (2005) Determination of orthophosphate and total phosphorus contents by flow analysis (FIA and CFA)—part 2: method by continuous flow analysis (CFA) (ISO 15681–2:2003). German version EN ISO 15681–2:2004

    Google Scholar 

  37. DIN En ISO 16264 (2004) Determination of soluble silicates by flow analysis (FIA and CFA) and photometric detection (ISO 16264:2002). German version EN ISO 16264:2004

    Google Scholar 

  38. DIN En 1484 (1997) Guidelines for the determination of total organic carbon (TOC) and dissolved organic carbon (DOC). German version EN 1484:1997

    Google Scholar 

  39. Utermöhl H (1958) Zur vervollkommnung der quantitativen phytoplankton-methodik. Mitt Int Verein Limnol 9:1–38

    Google Scholar 

  40. Berghahn R, Mohr S, Hübner V, Schmiediche R, Schmiedling I, Svetich-Will E, Schmidt R (2012) Effects of repeated insecticide pulses on macroinvertebrate drift in indoor stream mesocosms. Aquat Toxicol 122–123:56

    Article  Google Scholar 

  41. R development core Team (2019) R: a language and environment for statistical computing. Vienna, Austria: R foundation for statistical computing

  42. Oksanen J, Blanchet FG, Kindt R, Legendre P, Minchin PR, O’Hara RB, Simpson GL, Solymos P, Henry H, Stevens H, Wagner, H (2013) vegan: community ecology package R package. Accessed 08 May 2023

  43. Wang Y, Naumann U, Wright S, Warton D (2012) mvabund–an R package for model-based analysis of multivariate abundance data. Methods Ecol Evol 3:471–474

    Article  Google Scholar 

  44. Szöcs E, Van den Brink PJ, Lagadic L, Caquet T, Roucaute M, Auber A, Bayona Y, Liess M, Ebke P, Ippolito A, ter Braak CJ, Brock TC, Schäfer RB (2015) Analysing chemical-induced changes in macroinvertebrate communities in aquatic mesocosm experiments: a comparison of methods. Ecotoxicology 24(4):760–769

    Article  Google Scholar 

  45. Borcard D, Gillet F, Legendre P (2018) Numerical ecology with R. Springer, Zurich

    Book  Google Scholar 

  46. Kleiber C, Zeileis A (2008) Applied Econometrics with R. Springer publisher, New York. ISBN 978-0-387-77316-2. Accessed 18 Jan 2023

  47. Anderson MJ, Crist TO, Chase JM, Vellend M, Inouye BD, Freestone AL, Sanders NJ, Cornell HV, Comita LS, Davies KF, Harrison SP, Kraft NJB, Stegen JC, Swenson NG (2011) Navigating the multiple meanings of β diversity: a roadmap for the practicing ecologist. Ecol Lett 14(1):19–28

    Article  Google Scholar 

  48. Anderson MJ (2001) A new method for non-parametric multivariate analysis of variance. Austral Ecol 26(1):32–46

    Google Scholar 

  49. McArdle BH, Anderson MJ (2001) Fitting multivariate models to community data: a comment on distance-based redundancy analysis. Ecology 82(1):290–297

    Article  Google Scholar 

  50. Polazzo F, dos Anjos TBO, Arenas-Sánchez A, Romo S, Vighi M, Rico A (2022) Effect of multiple agricultural stressors on freshwater ecosystems: the role of community structure, trophic status, and biodiversity-functioning relationships on ecosystem responses. Sci Total Environ 807(3):151052

    Article  CAS  Google Scholar 

  51. Calow P (1996) Variability: noise or information in ecotoxicology? Environ Toxicol Pharmacol 2(2–3):121–123

    Article  CAS  Google Scholar 

  52. Nikinmaa M, Tjeerdema R (2013) Environmental variations and toxicological responses. Aquat Toxicol 127:1.

    Article  CAS  Google Scholar 

  53. Devin S, Giamberini L, Pain-Devin S (2014) Variation in variance means more than mean variations: what does variability tell us about population health status? Environ Int 73:282–287

    Article  Google Scholar 

  54. Nikinmaa M, Anttila K (2019) Individual variation in aquatic toxicology: not only unwanted noise. Aquat Toxicol 207:29–33

    Article  CAS  Google Scholar 

  55. Höss S, Ahlf W, Fahnenstich C, Gilberg D, Hollert H, Melbye K, Meller M, Hammers-Wirtz M, Heininger P, Neumann-Hensel H, Ottermanns R, Ratte HT, Seiler TB, Spira D, Weber J, Feiler U (2010) Variability of sediment-contact tests in freshwater sediments with low-level anthropogenic contamination—determination of toxicity thresholds. Environ Pollut 158(9):2999–3010

    Article  Google Scholar 

  56. Knauer K, Hommen U (2012) Sensitivity, variability, and recovery of functional and structural endpoints of an aquatic community exposed to herbicides. Ecotox Environ Saf 78:178–183

    Article  CAS  Google Scholar 

  57. Lorenz EN (1963) Deterministic nonperiodic flow. J Atmos Sci 20(2):130–141

    Article  Google Scholar 

  58. Calabrese EJ, Baldwin LA (2003) The hormetic dose-response model is more common than the threshold model in toxicology. Toxicol Sci 71(2):246–250

    Article  CAS  Google Scholar 

  59. Calabrese EJ (2007) Hormesis: why it is important to toxicology and toxicologists. Environ Toxicol Chem 27(7):1451–1474

    Article  Google Scholar 

  60. Agathokleous E (2018) Environmental hormesis, a fundamental non-monotonic biological phenomenon with implications in ecotoxicology and environmental safety. Ecotox Environ Saf 148:1042–1053

    Article  CAS  Google Scholar 

  61. Depledge MH (1990) New approaches in ecotoxicology: can inter-individual physiological variability be used as a tool to investigate pollution effects? Ambio 19(5):2

    Google Scholar 

  62. Cairns J Jr (1977) Aquatic ecosystem assimilative capacity. Fisheries 2(2):5–7

    Google Scholar 

  63. Matthews RA, Landis WG, Matthews GB (1996) The community conditioning hypothesis and its application to environmental toxicology. Environ Toxicol Chem 15(4):597–603

    Article  CAS  Google Scholar 

  64. Landis WG, Markiewicz AJ, Matthews RA, Matthews GB (2000) A test of the community conditioning hypothesis: persistence of effects in model ecological structures dosed with the jet fuel jp-8. Environ Toxicol Chem 19(2):327–336

    Article  CAS  Google Scholar 

  65. Death RG (2010) Disturbance and riverine benthic communities: what has it contributed to general ecological theory? River Res Appl 26(1):15–25

    Article  Google Scholar 

  66. Steudel B, Hector A, Friedl T, Löfke C, Lorenz M, Wesche M, Kessler M (2012) Biodiversity effects on ecosystem functioning change along environmental stress gradients. Ecol Lett 15(12):1397–1405

    Article  Google Scholar 

  67. Eurostat (2019) News release. Structure of orchards in 2017. Two-thirds of the EU’s fruit plantation area is concentrated in Spain, Italy and Poland, 32/2019. Accessed 18 Jan 2023

  68. Kreisel A, Schlossberger F (2019) Obst, Agrarmärkte 2018. Bayerisches Landesamt für Landwirtschaft, Germany, Schwäbisch Gmünd, pp 91–117

    Google Scholar 

  69. Roßberg D (2007) NEPTUN or how many treatments are done. Gesunde Pflanzen 59:55–65

    Article  Google Scholar 

  70. Roßberg D(2009) NEPTUN 2007—obstbau berichte aus dem julius kühn-institut, julius kühn institute, federal research Centre for cultivated plants, Braunschweig, Germany, ISSN 1866–590X, pp 73

  71. Roßberg D, Harzer U (2015) Survey on application of chemical pesticides in apple farming. J für Kulturpflanzen 67(3):6

    Google Scholar 

  72. Simon S, Brun L, Guinaudeau J, Sauphanor B (2011) Pesticide use in current and innovative apple orchard systems. Agron Sustain Dev 31:541–555

    Article  CAS  Google Scholar 

  73. Schäfer R, Kühn B, Malaj E, König A, Sahm R (2016) Contribution of organic toxicants to multiple stress in river ecosystems. Freshw Biol 61:2116–2128

    Article  Google Scholar 

  74. Weber G, Christmann N, Thiery A-C, Martens D, Kubiniok J (2018) Pesticides in agricultural headwater streams in southwestern Germany and effects on macroinvertebrate populations. Sci Total Environ 619–620:638–648

    Article  Google Scholar 

  75. Covert SA, Shoda ME, Stackpoole SM, Stone WW (2020) Pesticide mixtures show potential toxicity to aquatic life in U.S streams, water years 2013–2017. Sci Total Environ 745:141285

    Article  CAS  Google Scholar 

  76. Panico SC, van Gestel CAM, Verweij RA, Rault M, Bertrand C, Menacho Barriga CA, Coeurdassier M, Fritsch C, Gimbert F, Pelosi C (2022) Field mixtures of currently used pesticides in agricultural soil pose a risk to soil invertebrates. Environ Pollut 305:119290

    Article  CAS  Google Scholar 

  77. Van Wijngaarden RPA, Cuppen JGM, Arts GHP, Crum SJH, Van den Hoorn MW, Van den Brink PJ, Brock TCM (2004) Aquatic risk assessment of a realistic exposure to pesticides used in bulb crops: a microcosm study. Environ Toxicol Chem 23(6):1479–1498

    Article  Google Scholar 

  78. Arts GHP, Buijse-Bogdan LL, Belgers JDM, van Rhenen-Kersten CH, Van Wijngaarden RPA, Roessink I, Maund SJ, Van den Brink PJ, Brock TCM (2006) Ecological impact in ditch mesocosms of simulated spray drift from a crop protection program for potatoes. Integr Environ Assess Manag 2(2):105–125

    Article  CAS  Google Scholar 

  79. Talk A, Kublik S, Uksa M, Engel M, Berghahn R, Welzl G, Schloter M, Mohr S (2016) Effects of multiple but low pesticide loads on aquatic fungal communities colonizing leaf litter. J Environ Sci 46:116–125

    Article  CAS  Google Scholar 

  80. Sigmund G, Ågerstrand M, Brodin T, Diamond ML, Erdelen WR, Evers DC, Lai A, Rillig MC, Schäffer A, Soehl A, Torres JPM, Wang Z, Groh KJ (2022) Broaden chemicals scope in biodiversity targets. Science 376(6599):1280.

    Article  CAS  Google Scholar 

  81. Shahid N, Liess M, Knillmann S (2019) Environmental stress increases synergistic effects of pesticide mixtures on Daphnia magna. Environ Sci Technol 53(21):12586–12593.

    Article  CAS  Google Scholar 

  82. Kortenkamp A, Faust M (2018) Regulate to reduce chemical mixture risk. Science 361(6399):224–226

    Article  CAS  Google Scholar 

  83. European commission (2020a) factsheet: from farm to fork: our food, our health, our planet, our future. Brussels: European commission. Accessed 18 Jan 2023

  84. European commission (2020b) “a farm to fork strategy for a fair, healthy and environmentally friendly food system.” COM (2020) 381 final. Brussels: European Commission. Accessed 18 Jan 2023

Download references


Special thanks go to the staff of department IV 2.5 of UBA for their great help in preparing and conducting the experiment. Bonny Haueisen, Ina Janthur, Stefan Loth, and Ronny Schmiediche are especially acknowledged for the contributions in experimental set-up preparations, sampling and water chemistry analysis. The help of Claudia Kohls, Dagmar Schnee, and Wolfgang Mailahn for the pesticide analysis is highly appreciated. The authors thank Luettig & Friends and Claus Orendt for quantitative and qualitative periphyton and benthos analysis. The Federal Office of Consumer Protection and Food Safety (BVL) is thanked for the analytical analysis of grinded plant production product granulates. Thomas Graeff (UBA) is appreciated for his help in statistical analysis as well as Romeo Herr (UBA) for his help in conceptual planning of this study. Joern Wogram, Valeska Contardo-Jara, and Ulrike Scholz (both UBA) are highly acknowledged for their valuable comments on a former version of the MS.


Open Access funding enabled and organized by Projekt DEAL. No funding.

Author information

Authors and Affiliations



SM, StM, MF and RS designed and performed the experiment, evaluated data and wrote the original manuscript. SD and TF were involved in designing the experiment and wrote and edited the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Silvia Mohr.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

No, I declare that the authors have no competing interests as defined by Springer, or other interests that might be perceived to influence the results and/or discussion reported in this paper.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1

. Section S1. Chemical analysis; Table S1. Date, origin and number of stocked macroinvertebrates for each stream mesocosm based on estimates from stored reference samples; Table S2. List of active ingredients. International union of pure and applied chemistry name; Table S3. Water parameters of control and treatment stream mesocosmsfor each sampling date; Table S4. Water-chemical parameters of control and treatment stream mesocosmsfor each sampling date; Fig. S1. Sketch of stream mesocosm set-up, Fig. S2. Development of the concentration of the nutrients A) potassium, B) phosphate, C) nitrate, and D) nitrite during the experiment.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Mohr, S., Meinecke, S., Feibicke, M. et al. Effects of a realistic pesticide spraying sequence for apple crop on stream communities in mesocosms: negligible or notable?. Environ Sci Eur 35, 35 (2023).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: