Toxicity of microplastics and natural particles in the freshwater dipteran Chironomus riparius: Same same but different?

a r t i c l e i n f o


Introduction
The continuous release and break down of large plastic debris as well as direct emissions result in an accumulation of microscopic plastics, so-called microplastics (MP), in aquatic environments (Andrady, 2017;Conkle et al., 2018;Geyer et al., 2017;Hartmann et al., 2019). While a large number of studies address the abundance of MPs in aquatic systems, studies on their toxicity are less prevalent, especially in a freshwater context (Blettler et al., 2018). So far, toxicity studies cover a variety of outcomes, including null effects, reduced growth and reproduction, elevated mortality and inflammatory response (e.g., reviewed in Foley et al., 2018). MPs may act as vectors, transferring additives and sorbed pollutants to biota or affect the impacts of pollutants by altering their bioavailability (Wang et al., 2018). In theory, hydrophobic organic chemicals (high log K OW ) preferably sorb to MPs or other particulate matter, while less hydrophobic substances (low log K OW ) rather remain dissolved (Wang et al., 2018). Whereas most studies address the role of MP to act as a vector by using hydrophobic chemicals, effects by co-exposures to hydrophilic ones are scarce and mostly overlooked in MP research (Horton et al. 2018;Triebskorn et al., 2019).
In general, the toxicity of MPs depends on their physicochemical properties (material, size, shape etc.), the pollutant (e.g., log K OW ), and the species. Accordingly, generalizations with regards to the environmental risk of MPs is challenging as they represent a very diverse group of stressors (Lambert et al., 2017;Scherer et al., 2017b). Thus, it is unsurprising that the relevance of MPs as anthropogenic contaminants is discussed controversially (Burton, 2017;Kramm et al., 2018). One key question to assess the environmental risks of MPs is their toxicological impact compared with natural particulate matter (Backhaus and Wagner, 2018). Numerous studies have shown that suspended solids affect a range of species depending on the material and the particle size (e.g., reviewed in Bilotta and Brazier, 2008) The same is true for the composition of sediments. For example, organic carbon content, material composition as well as grain-size distribution of sediments has been shown to affect growth and survival of benthic invertebrates (Ankley et al., 1994;Bisthoven et al., 1998;Ristola et al., 1999). The presence of particulate matter in the water phase or in sediments also modulates the exposure to chemicals and their toxicity (Fleming et al., 1998;Landrum and Faust, 1994;Zhang et al., 2018). While this is relevant in an MP context, most toxicity studies do not compare the impacts of natural particles to that of MPs (Burns and Boxall, 2018;Connors et al., 2017;Scherer et al., 2017b). Accordingly, the context is missing to benchmark the impacts of MPs in environments in which natural particulate matter is abundant.
Thus, the aim of this study is to address the outlined knowledge gaps: First, we compare the toxicity of MPs to that of natural particulate matter (NPM) present in sediments or suspended in the water phase. Second, we investigate changes in toxicity of a hydrophilic chemical in the presence of MPs and NPM. Focussing on freshwater systems, we exposed the benthic deposit feeding larvae of Chironomus riparius to unplasticized polyvinyl chloride (PVC-MP) as well as kaolin and diatomite as NPM. We selected kaolin and diatomite as references materials based on the following reasoning: (1) Both are naturally occurring sediment components frequently applied in toxicity experiments (OECD, 2004). (2) Due to their high densities, they settle rapidly and are thus bioavailable for C. riparius. (3) Their physicochemical properties and associated effects are well-characterized (e.g., Hadjar et al., 2008;Murray, 2006;Nattrass et al., 2015;OECD, 2004). We have previously shown that C. riparius readily ingests MPs and, consequently, represent an ideal model to study the effect of dense particulate materials on benthic organisms (Scherer et al., 2017a).
In 28 d chronic toxicity experiments, we investigated the effects of the three fine particulate materials (FPM) kaolin, diatomite and PVC-MP on the emergence, development and mass of chironomids. Moreover, we evaluated the impact of kaolin and PVC-MP in chironomids co-exposed to the neonicotinoid pesticide imidacloprid in a multiple-stressor experiment. Imidacloprid is hydrophilic and, thus, sorption to FPM will be negligible. Therefore, an increased toxicity of imidacloprid is expected to be an additional stress induced by FPM rather than an effect of a higher bioavailability.

Methods
Aquatic larvae of the diptera C. riparius were exposed to different concentrations of PVC-MP and NPM with and without coexposure to the neonicotinoid imidacloprid (CAS: 138261-41-3, Sigma-Aldrich) in 28 d, chronic toxicity experiments. C. riparius larvae were obtained from our in-house culture by transferring egg ropes into petri dishes containing Elendt M4 medium (OECD, 2004). The egg ropes were incubated at 20°C and a light:dark cycle of 16:8 h. Hatched larvae were monitored daily to collect chironomids < 24 h for the experiments.

Chronic exposures studies with Chironomus riparius
The experiments were performed in accordance with the OECD guideline 218 (OECD, 2004) with a modified sediment composition: While the guideline suggests a high proportion of FPM (fine quartz sand +20% kaolin), this would make it impossible to compare the toxicity of PVC-MP and natural particles as both would be present in a mixture. Additionally, increasing PVC-MP and constant quartz/kaolin concentrations would change the sediment characteristics by increasing the total amount of FPM. Therefore, we used sediments consisting of washed and sieved quartz sand (250-1000 mm), ground and sieved leaves of Urtica dioica and Alnus glutinosa (< 250 mm, 1:1 mixture based on mass) as carbon source and either PVC, kaolin or diatomite. Sediments for each replicate were individually prepared and mixed in 500 mL glass beakers.

Experiment I (exposure via sediment)
Each exposure vessel contained 100 g sediment (dry weight) consisting of sieved quartz sand, 1% leaves and 0.002, 0.02, 0.2, 2 and 20% (wt) of PVC, kaolin or diatomite. The sediments were homogenised and 400 mL Elendt M4 medium was carefully added per replicate to prevent suspension of particulate matter (n = 3 per treatment). Control treatments consisted of sediments without FPM.

Experiment II (exposure via water)
To evaluate the impact of a waterborne exposure, sediment compositions remained similar to experiment I except that the FPM were applied by suspending 0.0002, 0.002, 0.02, 0.2 and 2 g of PVC, kaolin or diatomite in 400 mL Elendt M4 medium (n = 3). FPM were weighed into 50 mL tubes per replicate, suspended in 50 mL Elendt M4 medium and shaken for 24 h in the dark (300 rpm, orbital shaker). One day later, the suspensions were transferred to the exposure vessels and the tubes were rinsed three times with fresh medium. The FPM were allowed to settle for 5 d before the toxicity experiment was started.

Experiment III (leaching of toxicity)
To investigate the toxicity of chemicals leaching from PVC-MP, we exposed chironomids to MP extracts (in solvent) and migrates (in medium). PVC-MP extracts were obtained by a 24 h Soxhlet extraction of 200 g PVC-MP using acetone. The extracts' volume was reduced to 50 mL using a rotary evaporator (Heidolph, Laborota 4000) and used to prepare diluted extracts equivalent to 0.2 and 20 g PVC-MP/10 mL acetone. An empty Soxhlet cartridge was extracted as blank. The sediments contained sieved quartz sand, 1% leaves and the corresponding amount of kaolin as FPM (0, 0.2 and 20%) and were spiked with 10 mL extracts, thoroughly mixed, followed by a 24 h period allowing the acetone to evaporate completely (n = 4). PVC-MP migrates were prepared by shaking a suspension of 20 g PVC-MP in 400 mL Elendt M4 medium for 33 d under the same conditions used for the toxicity experiments. The suspensions were filter using a glass fibre filter (1.5 mm pore size) and the filtrate was used as test medium. As a control, identically treated Elendt M4 medium without PVC-MP was used. The sediments contained sieved quartz sand, 1% leaves and 20% kaolin as FPM to have a similar size composition as in the other sediments (n = 4).

Experiment IV (pesticide co-exposure)
Chironomids were exposed to 0, 0.5, 1, 2, 4 and 8 mg L À1 imidacloprid (nominal concentration) and sediments containing sieved quartz sand, 1% leaves and 0, 0.2 and 20% kaolin or PVC-MP as FPM (n = 4). We excluded diatomite to ensure the feasibility of parallel testing. Water and sediment from an extra replicate per treatment was sampled after 5 d for chemical analysis.

Experimental set up
All beakers were covered with nylon meshes and constantly aerated to provide sufficient oxygen. After a 5-d incubation period at 20°C and a light:dark cycle of 16:8 h, aeration was stopped and 20 C. riparius larvae (< 24 h) were added per replicate (day 0). On the following day, the aeration was continued, and the larvae were fed with 0.2 mg fish food larva À1 d À1 (ground TetraMin suspended in Elendt M4 medium) during the first 7 d and 0.5 mg food larva À1d À1 afterwards. During 28 d, the number of emerged imagoes was recorded daily. Emerged imagoes were counted, removed and snap-frozen for further analysis. To measure their weight, frozen specimens were defrosted and dried at 60°C for 24 h. The dried individuals were weighted and sexed. Water quality parameters (oxygen level, pH and conductivity) were measured in 20 randomly selected vessels at the beginning and the end of the experiments. Here, validity of the experiments was given when the oxygen levels were > 85% and the pH was constant at 8.20 ± 0.35 in any vessel.

PVC-MP, phthalate and imidacloprid analysis
PVC-MP were analysed using thermal desorption gas chromatography coupled to mass spectrometry (TD-GC-MS, Multi-Shot Pyrolyzer EGA/PY-3030D coupled to an Agilent 7890B gas chromatograph and Agilent 5977B MSD, analytical details in Table S1). Imidacloprid, phthalates (dimethyl phthalate, diethyl phthalate, dibutyl phthalate, benzyl butyl phthalate, bis(2ethylhexyl) phthalate, di-n-octyl phthalate, diisononyl phthalate) and bis(2-ethylhexyl) adipate were quantified in water or sediments using liquid chromatography coupled to mass spectrometry (LC-MS, Agilent 1260 coupled to a QTrap 4500 AB Sciex, analytical details in Table S1). 10 mL water and 30 g sediment were sampled from (a) analytical replicates at day 0 and (b) one randomly selected replicate per treatment at day 28 of experiment IV. These samples were snap frozen and stored at À20°C until further processing (see SI for details).

Statistical analysis
Emergence, mean emergence time (EmT 50 ), mean development rate and the weight of male and females were used to evaluate the toxicity of PVC-MP, kaolin, diatomite and imidacloprid.
Mean development rates (x; d À1 ) were calculated for each replicate of every treatment in every experiment as: with m = the number of inspections, f i = number of emerged individuals since the last inspection, x i = time of inspection (d À1 ), and n e = total number of emerged individuals at the termination of the experiment. x i was calculated as: with day i = day of inspection since the beginning of the experiment (d) and l i = time interval since the last inspection (d) (OECD, 2004). All data were analysed and visualised using R (version 3.5.1; R Core Team, 2018). Generalized linear models (GLMs) were fit for every endpoint (emergence, weight, EmT 50 , and development rate) of each of the four experiments. For emergence and EmT 50 , the distribution family was binomial with a logit link, as the data was expressed as either success or non-success (1 or 0, respectively). For weight and development rate, an exponential Gamma family distribution with inverse link was used because the variance in data increased with increasing average values.
All GLMs were initially fit with full parametrization, that is, with all experimentally possible effect interactions. To avoid over-fitting, all individual model parametrizations were stepwise optimized for minimum values of Akaike's information criterion (AIC; Akaike, 1974). A detailed list of all initial and final parametrizations of all GLMs is given in the supporting information (Table S2-8, Figure S4-10). For each GLM, the remaining parameters were analysed for effect strength in independent single-step term deletions with subsequent analysis of variance, using v 2 tests with marginal (type III) sum-of-squares. An effect was considered significant for p < 0.05. Pseudo-R 2 values were calculated using McFadden's method (McFadden, 1979). Raw data and R scripts for data treatment, statistical analysis and visualisation can be made available upon request.

FPM characterization and chemical analysis
On a microscopic scale, diatomite had a heterogeneous particle shape including sharp and needle-like fragments (Fig. 1). Kaolin and PVC-MP were more uniform and rounded. In suspension, all materials formed agglomerates. The size distributions (2-60 mm) confirm that the majority of FPM is < 16 mm ( Fig. 1G-I). The mean sizes as well as the particle size distributions of kaolin (mean 2.80 mm with 90% of particles < 4.38 mm) and PVC-MP (mean 2.77 mm with 90% of particles < 4.25 mm) are similar. In contrast, diatomite particles are slightly larger (mean 4.44 mm with 90% of particles < 8.19 mm) and the size distribution lacks the exponential increase of smaller particles. According to the density of the materials, we expected the highest particle concentration for PVC-MP (1.4 g cm À3 ), followed by diatomite (2.3 g cm À3 ) and kaolin (2.6 g cm À3 ). We experimentally determined the particle concentrations (numbers) per mass of 2-60 mm particles and observed a different order with kaolin (1.28 Â 10 10 particles g À1 ) containing more particles per mass than PVC-MP (8.40 Â 10 9 particles g À1 ) and diatomite (5.25 Â 10 9 particles g À1 , Figure S1A). We used these particle concentrations to calculate their mass based on the density and assuming spherical particles ( Figure S1B). Indeed, predicting the mass of heterogeneously shaped and porous fragments is biased. However, the discrepancy in estimated and observed mass demonstrates that particles in the size range of 2 to 60 mm do not represent the total number and mass of particles and point to a missing fraction below the limit of quantification (0.68 mm) for kaolin (recovery of 85.2 ± 6.7%) and PVC-MP (recovery of 55.4 ± 11 .6%). This is supported by the exponentially increasing concentrations of particles < 2 mm for PVC-MP and kaolin ( Figure S1C). Here, filtered (1.5 mm pore size) suspension of 50 g L À1 kaolin, diatomite and PVC-MP contain 1.59 Â 10 4 , 3.92 Â 10 3 and 4.04 Â 10 6 particles mL À1 in the size range of 0.68-2 mm, respectively.
According to the results of the thermal desorption GC-MS, PVC-MP does not contain additives. In addition, the measured concentrations of phthalates in the water and sediment samples are in the same range as blanks and controls, even at the highest PVC-MP concentration ( Figure S2). The electrical conductivity differs between the 20% PVC-MP treatment (1347 ± 31 mS cm À1 ) and all other treatments (1015 ± 44 mS cm À1 ).
The weight of emerged imagoes depends on the sex with females being heavier than males (v 2 = 9170, p < 0.001, Fig. 2B). Exposures to kaolin (v 2 = 4.3, p < 0.05), diatomite (v 2 = 31.2, p < 0.001) and PVC-MP (v 2 = 46.1, p < 0.001) significantly affect the weight of male and female chironomids (Table 2). Here, increasing concentrations reduce the weight of male and female chironomids by 1.14% and 10.6% for diatomite and 0.23% and 16.7% for PVC-MP, respectively, compared to the treatments without FPM (Fig. 2B). Although the impacts on weight do not follow a clear dose-response pattern, the GLM suggests a slight decrease in weight with increasing kaolin concentrations (v 2 = 4.3, p < 0.05, Fig. 2B, Table S6).
In summary, the impacts of FPM are minor at concentrations < 2% and become more pronounced at the highest concentration. At 20% FPM, the emergence of chironomids exposed to PVC-MP or diatomite is reduced by 32.7% and 12.7%, respectively, and generally delayed for PVC-MP (male: 1.43 d, female: 0.05 d) and diatomite (male: 0.24 d, female: 0.5 d), compared to the corresponding kaolin exposure. In addition, the weight of chironomids exposed to diatomite (male: À3.36%, female: À9.56%) and PVC-MP (male: À2.48%, female: À15.7%) is lower than in the kaolin treatments.

Chemical analysis
The recovery rate of imidacloprid is 94.7 ± 12.7% at day 0 and 87.4 ± 14% at day 28 ( Figure S3). Thus, nominal concentrations were used for data analysis. During 28 d, we observed only a minor change in the partitioning of imidacloprid: At day 0, 88.4 ± 4% is dissolved in the water phase and 11.6 ± 4% associated with the sediment. At the end of experiment IV, the imidacloprid concentration in water decreases to 80 ± 4% and increases to 20 ± 7% in the sediment. This observation was especially pronounced in treatments containing high FPM levels (20% kaolin/PVC, Figure S3B). A maximum partitioning of 27.5% of imidacloprid to sediments containing 20% FPM indicates that sorption is minor and chironomids are mainly exposed to dissolved imidacloprid.

Interactions of FPM and imidacloprid
There are significant interactions of FPM and imidacloprid for emergence success and time as well as weight in the GLM. This implies that co-exposure to FPM modulates the toxicity of imidacloprid. For instance, the significant interaction between kaolin: imidacloprid (v 2 = 8.58, p < 0.01) and PVC:imidacloprid (v 2 = 17.3, p < 0.001) on the emergence success is mirrored by the EC 50 values for emergence. The latter demonstrate that the joint toxicity depends on the material with kaolin being less toxic than PVC-MP (Table 2). Here, the influence of sand, 0.2% kaolin, 0.2% PVC-MP and 20% kaolin is minor (overlapping confidence intervals). In contrast, co-exposure to 20% PVC-MP markedly increases the imidacloprid toxicity (Fig. 3A).  For the emergence time, the GLM shows significant interactions between kaolin:imidacloprid (v 2 = 10.6, p < 0.01) and PVC:imidacloprid (v 2 = 5.7, p < 0.05). The EmT 50 values indicate that a higher proportion of kaolin reduces the delaying effect on the emergence induced by imidacloprid whereas a higher proportion of PVC-MP exacerbates this effect (Table S8). For instance, increasing the concentration of imidacloprid from 0 to 2 mg L À1 and the proportion of FPM from 0.2 to 20%, reduces the EmT 50 by 0.56 d (males) for kaolin and increases the EmT 50 by 0.53 d (males) for PVC-MP. In addition, the interaction between PVC-MP and imidacloprid significantly affect the weight of chironomids (v 2 = 21.4, p < 0.001, Fig. 3B, Table 1). For instance, increasing the concentration of imidacloprid from 0 to 2 mg L À1 and the proportion of PVC-MP from 0.2 to 20%, results in a reduction of weight by 32.5% (males) and 38.2% (females).

Discussion
Chronic exposures to PVC-MP and natural particles significantly affect the emergence and weight of Chironomus riparius. These effects are material-specific with PVC-MP being more toxic than the natural kaolin and diatomite. In addition, kaolin slightly reduces while PVC-MP exacerbates the toxicity of imidacloprid by acting as an additional stressor. Given the high concentrations needed to induce adverse effects, C. riparius seems to tolerate PVC-MP exposures at levels much higher than currently expected in the environment.
Overall, it remains challenging to determine which particle and material properties drive the toxicity of FPM in chironomids. Here, we tried to match the multiple properties (Table 3) and show that PVC-MP and kaolin are very similar regarding their shape and size distribution. Accordingly, the higher toxicity of PVC-MP may be caused by the presence of very small particles, its surface chemistry or its chemical composition. In contrast, diatomite is generally larger and contains less particles < 2 mm than PVC-MP and kaolin, which leads to the assumption that the spiny and needle-like shape as well as the high porosity contribute to the toxicity.
Taken together, this study empirically supports previous calls to consider the heterogeneous properties of MP and other types of particulate matter when assessing their toxicity (Rochman et al., 2019;Hartmann et al., 2019;Lambert et al., 2017). A detailed discussion of our findings and potential mechanisms for FPM toxicities in chironomids are provided in the following section. Exposure to diatomite and PVC-MP significantly reduces the emergence and weight of chironomids (Fig. 2). In contrast, kaolin positively affects the chironomids by increasing the emergence success and decreasing the emergence time. The beneficial effect of kaolin is in line with its application as sediment component in standardized toxicity testing (OECD, 2004) and highlights that chironomids depend on FPM present in sediments. Moreover, because its particle size distribution and shape are similar to PVC-MP, kao- lin appears to be a suitable reference material for this type of MP. At the same time, diatomite may be a suitable counterpart when assessing particle related toxicities in chironomids because it represents different physical properties (sharpness and porosity).
PVC-MP is more toxic than natural particles. Adverse effects on the emergence and weight of chironomids become evident at the highest applied concentrations (20% FPM exposed via sediment and 2% exposed via water) with PVC-MP being more toxic than diatomite and kaolin (Fig. 2). The same is true when comparing the joint effects of particles in combination with imidacloprid (Fig. 3). Diatomite induced adverse effects (see above) which were lower than the ones of PVC-MP at the same mass-based concentration. However, when comparing numerical concentrations instead, diatomite was more toxic because it contains less particles per mass (Table 3).

The toxicity of kaolin, diatomite and PVC-MP is concentrationdependent
As indicated by the GLMs, FPM gradients significantly affect the probability and time of emergence as well as the weight of emerged chironomids in a material-specific manner (Table 1, Fig. 2). This is in accordance with literature on MPs and natural FPM reporting that the results of an exposure strongly depends on the dose and material (reviewed in Scherer et al., 2017b). In general, increasing FPM concentrations will increase the encounter rates (physical) and/or exposure concentrations (chemical) and, thus, increase material-related toxicities. Consequently, low FPM concentrations in mixed sediments (< 2%) correspond to low encounter rates and thus, minimize particle-related effects. Such dilution in the sediment does not occur for exposures via the water phase because, during the 5 d incubation period, the FPM settled on the sediment surface resulting in a top layer consisting almost exclusively of FPM at the highest concentration (2%). This might explain the significantly affected weight of imagoes in those treatments. Accordingly, the spiking method strongly influences the outcomes of toxicity studies with MPs.

A comparison with the literature points to species-and materialspecific effects
Our results are in accordance with literature, as many studies on MPs do not observe adverse effects in freshwater invertebrates at low concentrations (Imhof et al., 2017;Imhof and Laforsch, 2016;Weber et al., 2018). In contrast, Ziajahromi et al. (2018) observed size-specific effects at much lower MP concentrations in Chironomus tepperi. Exposure to 500 PE-MP kg À1 sediment, Table 3 Comparison of the properties of fine particulate materials used in this study.
Measurements of mean size [mm] and particle concentration, classification of shape via microscopy, surface characteristics and additional information of kaolin (Murray, 2006) and diatomite (Hadjar et al., 2008) from the literature. reduced survival, emergence, body length and growth for 10-27 mm MP, while 100-126 mm PE-MP did not affect the survival but increase the emergence time. Indeed, species-specific sensitivities (C. riparius vs C. tepperi), material-specific toxicities (PVC vs PE) as well as the different experimental setup (28 vs 10 d) might provide an explanation. In addition, a recalculation of MP concentrations based on the given density (0.98-1.02 g cm À3 ), size (1-126 mm) and mass (300-500 mg L À1 ) of the PE-MP used by Ziajahromi et al. (2018) reveals a discrepancy. Theoretically, the exposure concentrations for PE-MP < 27 mm are 48 to 57,500 times higher than stated in the study. However, when comparing the effects of the two studies based on mass not particle concentrations, it becomes clear that C. tepperi is either more sensitive to MP exposure than C. riparius or that PE-MP is more toxic than PVC-MP. Here, exposure studies with C. riparius and PE-MP confirm both assumptions (Silva et al. 2019). However, exploring this interesting question would require identical experimental setups (exposure regimes and materials). Nevertheless, previous studies did not include natural reference particles or adapted sediment compositions to characterise MP-specific effects. In this regard, our results highlight that material-specific properties of PVC-MP induce a stronger toxicity than natural particles.

PVC-MP affects C. Riparius at concentrations much higher than currently detected in the environment
We observed a reduced and delayed emergence as well as a reduced weight of imagoes at high, environmentally irrelevant exposures. While reports on mass-based MP concentrations in freshwater sediments are scarce, concentrations between 2.6 (18 particles kg À1 ) and 71.4 mg kg À1 (514 particles kg À1 ) have been reported for riverine sediments (Rodrigues et al., 2018). The low numerical compared to the given mass concentrations indicate that mainly larger MPs have been analysed. In general, environmental levels of MPs < 63 mm remain largely unknown due to the methodological challenges of separation and identification (Burns and Boxall, 2018;Conkle et al., 2018). Nonetheless, sediment contamination with 20% MP is highly unlikely. However, we used such high concentrations of MP not to mimic environmental conditions but to better understand how different FPM affect chironomids. Overall, C. riparius tolerates exposures to high concentrations of FPM, including unplasticised PVC-MP.

The availability of utilisable FPM affects chironomids
The autecology of chironomids is an important factor when assessing the toxicity of FPM. As rather non-selective deposit feeders, they actively ingest various particles including detritus, algae, silt and MPs in the size range of 1-90 mm (Nel et al., 2018;Rasmussen, 1984;Scherer et al., 2017a;Ziajahromi et al., 2018). They mainly interact with particles on sediment surfaces and are tolerant to various sediments as long as FPM for tube building and foraging are available (Brennan and McLachlan, 1979;Naylor and Rodrigues, 1995;Suedel and Rodgers, 1994). Since we removed the fine fraction from the sediments, only the ground leaves and the FPM are in the utilizable size range. At low FPMconcentrations, larvae utilized grounded leaf fragments for tube building and with increasing concentrations mostly FPM. Thus, modifying the sediment compositions is an appropriate tool to highlight the material-specific toxicities in chironomids.
The overall beneficial effect of kaolin on the emergence of chironomids is most likely related to the increased availability of utilizable particles. In theory, these reduce the time and energetic effort of larvae to build and maintain stable tubes and, thus, increase the time available for foraging on the sediment surface (Naylor and Rodrigues, 1995). Our findings support this assump-tion as chironomids develop faster and obtain higher emergence rates when exposed to 20% kaolin (experiment I and IV). In contrast, 20% kaolin reduces the weight of imagoes (experiment IV) probably by decreasing the dietary uptake by the larvae (Ristola et al., 1999;Sibley et al., 2001). Both findings highlight that natural FPM can be beneficial for some life history parameters of chironomids by providing tube-building materials as well as negative for others by reducing food availability.

Diatomite is a suitable reference material for shape-specific effects
The adverse effects of diatomite on chironomids are not surprising as it is used as a natural pesticide in agriculture (Kavallieratos et al., 2018;Korunic, 1998). The spiny and porous fragments are supposed to damage the digestive tract and disrupt the functionality of the cuticle by sorption and abrasion (Korunic, 1998). Therefore, the observed effect of diatomite seems plausible and, in addition, is in accordance with literature (Bisthoven et al., 1998). While this mode of action is plausible for diatomite, kaolin and PVC-MP have rounder shapes (Fig. 1, Table 3). Thus, other properties than particle shape drive the toxicity of PVC-MPs in chironomids.

Leaching chemicals are of minor importance for the toxicity of unplasticised PVC-MP in chironomids
The leaching of chemicals is a commonly addressed issue in toxicity studies with PVC (Lithner et al., 2011(Lithner et al., , 2012b. We used unplasticised PVC on purpose to minimize the complexity of the experiments and confirm that the PVC-MP does not contain additives commonly used in plasticized PVC. Along that line, PVC extracts simulating a worst-case leaching induced limited toxicity in C. riparius (Fig. 2E-F). However, volatile or inorganic compounds (e.g., vinyl chloride monomer, Cd, Zn, Pb) may migrate from PVC to water (Ando and Sayato, 1984;Benfenati et al., 1991;Lithner et al., 2012a;Whelton and Nguyen, 2013). Due to constant aeration and the 5 d acclimatisation period, concentrations of volatile substances in our experiments are expected to be low. Nonetheless, chironomids are sensitive to heavy metals and migrated metal ions might contribute to the observed effects (Béchard et al., 2008;Timmermans et al., 1992). The slight increase in electric conductivity of treatments with either PVC-MP-migrates or sediments with 20% PVC-MP indicates a higher concentration of ions compared to the other treatments. As C. riparius is tolerant to high salinities, a general effect of high ion concentrations is unlikely (Bervoets et al., 1996). Accordingly, leaching chemicals did not contribute significantly to the toxicity observed for PVC-MP in our experimental setup.
However, the surface chemistry of PVC-MP might have induced toxicities in chironomids. In contrast to the hydrophilic silicate minerals kaolin (Al 2 H 4 O 9 Si 2 ) and diatomite (mainly SiO 2 ), the surface of PVC-MP consists mainly of CAC, CAH and CACl moieties rendering it hydrophobic (Bakr, 2010;Murray, 2006;Asadinezhad et al., 2012). As surface charges and polarities are known to influence particle-particle and particle-biota interactions (e.g. aggregation, wettability), an effect induced by the hydrophobicity of PVC-MP on C. riparius seems likely (Nel et al., 2009;Potthoff et al., 2017).

Particles in the nanometre size range might have affected the toxicity of PVC-MP
In contrast to leachates, PVC-MP migrates produced under much milder conditions, significantly affected the emergence of chironomids. Since the concentrations of chemicals will be much lower than in the extracts, we speculate that particles in the nanometre size-range might have caused the observed effects (Table 1, Figure S1). Notwithstanding the similarities in shape and mean size, PVC-MP suspensions contain higher numbers of particles < 2 mm than kaolin (Table 3). Accordingly, this very small size fraction, including nanoplastics, present in the PVC-MP migrates might have induced the strong toxicity. While these particles would be also present in the MP suspensions used in the other experiments, the lower toxicity observed there may be due to a reduced bioavailability of nanoplastics attached to larger MP. The migrate toxicity highlights that the relevance of the smaller size fraction of particles present when using polydisperse MP mixtures. While it is technically challenging to remove nanoplastics from those, a comprehensive characterization will facilitate the interpretation of results.

PVC-MP >2 mm affects the weight of chironomids
In contrast to suspended PVC-MP, extracts and migrates significantly affected the emergence but not the weight of chironomids (p < 0.001, Table 1). Thus, the weight of imagoes is affected only if PVC-MP > 2 mm are present. Here, hydrophobic PVC-MP might have affected the weight by interfering with tube building and maintenance or with locomotion or digestion (Silva et al. 2019). A reduced feeding by food dilution in the sediment or the digestive tracts appears plausible, especially at high particle concentrations (Rist et al., 2016;Scherer et al., 2017b;Ziajahromi et al., 2018). However, the same mechanisms would be true for kaolin which did not reduce the body weight in the same way as PVC-MP (Figs. 2, 3). Accordingly, PVC-MP must have a more specific mechanism of action (e.g., hydrophobicity, chemical composition) affecting growth apart from a general physical toxicity.  (Matsuda et al., 2001;Casida, 2005, 2003). Therefore, the reduced growth and affected emergence are mostly linked to an impaired foraging (Alexander et al., 2007;Roessink et al., 2013).

FPM modulate the toxicity of imidacloprid with material-specific outcomes
High concentrations of PVC-MP significantly exacerbated the adverse impacts of the neonicotinoid while kaolin has a slight compensatory effect (Tables 1 and 2, Fig. 3). Such modulation of toxicities by FPM is in accordance with literature. Here, elevated or reduced toxicities of chemicals are commonly associated with changes in bioavailability (Avio et al., 2015;Beckingham and Ghosh, 2017;Fleming et al., 1998;Landrum and Faust, 1994;Ma et al., 2016;Rochman et al., 2013). However, because we used a hydrophilic chemical not sorbing in relevant amounts to FPM, a modulation of imidacloprid bioavailability plays a minor role in the combined exposures.
PVC-MP and imidacloprid both affect the emergence and weight of chironomids (Fig. 3, Table 2). Here, co-exposure to 20% PVC-MP results in a 1.7-fold reduction of the EC 50 of the neonicotinoid (Table 2). Considering the hypothesis that 20% PVC-MP directly (digestion) or indirectly (locomotion and tube building) interfere with foraging and feeding (4.2), a higher toxicity of a neu-rotoxic insecticide in the presence of an additional, particulate stressor is not surprising. In contrast, kaolin significantly interacts with imidacloprid by increasing the probability of emergence (Table 1, Fig. 3). As imidacloprid remains primarily dissolved in the water (> 72.4%) independent of the presence of FPM, any alterations in toxicity of imidacloprid are linked to an additional effect of the FPM. The slightly decreasing concentrations in the water and increasing concentrations of imidacloprid in the sediment over time are in accordance with the literature (Capri et al., 2001). Degradation (e.g., photolysis) might explain the slightly reduced imidacloprid concentration at day 28 ( Figure S2B) whereas the high concentration of FPM probably affect the sorption capacity of sediments by increasing the reactive surface and potential binding sites (Capri et al., 2001;Cox et al., 1998). However, due to the high water solubility (610 mg L À1 ) and low log K OW (0.57), imidacloprid is expected to have a low affinity to bind on particulate matter and to accumulate in sediments (EC, 2006). In general, the capacity of PVC-MP to adsorb chemicals is considered to be low compared to other polymers (e.g., polyethylene) as the high crystallinity of PVC affects the diffusivity and, thus, the sorption of chemicals (Bakir et al., 2016(Bakir et al., , 2012O'Connor et al., 2016;Teuten et al., 2009Teuten et al., , 2007. Thus, our findings highlight that a coexposure of MP and chemicals induces a stronger toxicity that is not caused by changes in bioavailability of the chemical stressor. Here, the beneficial effect of kaolin mitigates the toxicity of imidacloprid whereas PVC-MP enhances the toxicity at sub-lethal concentrations by acting as an additional stressor (Table 2).

Conclusion
This study shows that PVC-MP and natural particulate matter affects the emergence, development and weight of Chironomus riparius in a concentration-dependent manner. Kaolin and diatomite as natural reference for MP induced beneficial and negative effects, respectively. This demonstrates that natural FPM are not benign per se and that their impacts strongly depend on their physico-chemical properties, such as shape. PVC-MP induced more severe effects than both natural FPM highlighting that synthetic particles are more toxic than natural particles. As this was unrelated to leaching chemicals, other parameters, such as the presence of very small plastic particles or hydrophobicity may drive the toxicity of PVC-MP. Importantly, PVC-MP was toxic in high concentrations currently not detected in the aquatic environment. Same is true for co-exposures to the neonicotinoid imidacloprid. In a multiple-stressor experiment, high PVC-MP and kaolin concentrations increased and decreased the toxicity of imidacloprid in chironomids, respectively. Since imidacloprid is hydrophilic, other factors than bioavailability contributed to the combined effect of FPM and a chemical stressor. Although we observed a higher toxicity of PVC-MPs compared to natural particles, our results indicate that chironomids are very tolerant to an exposure to unplasticised PVC particles.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
PVC-MP and Christel Möhlenkamp for chemical analysis of water and sediment samples.

Contributions
C.S. designed the experiments and conducted the experimental work. J.V. produced the PVC-MP extracts and assisted in the experiments. C.S. and R.W. analysed the data and prepared the figures. C. S. and M.W. wrote the manuscript. N.B., F.S. and G.R. provided their feedback on the manuscript and guided the overall project. All authors reviewed the manuscript.