University of Birmingham Toxicokinetics of silver nanoparticles in the mealworm Tenebrio molitor exposed via soil or food

S NPs for 21 days a 21-day elimination phase (clean soil/food). A one-compartment kinetics model with inert fraction (simulating a storage compartment, detoxi ed forms are located) was used to describe Ag accumulation. the uptake route in mealworms is dif ﬁ cult. For several approaches used, that and are valid uptake routes, different importance. ,butsigni uptake rate constants nm , different patterns for the different NPs. food exposure, particle characteristics bioavailability bioaccumulation. Ag 2 S NPs were taken up, elimination faster than for other Ag forms, the lowest inert fraction. cantly elimination rate constants that the mechanism of elimination may be the for AgNPs

Toxicokinetics of silver nanoparticles in the mealworm Tenebrio molitor exposed via soil or food Zahra Khodaparast a

H I G H L I G H T S
• Mealworms accumulate Ag 2 S NPs despite its low dissolution in soil pore water. • Ag 2 S NPs were accumulated less than pristine AgNPs and ionic Ag in mealworms. • Mealworms exposed via soil and via food showed different Ag uptake kinetics. • AgNPs dissolution in soil pore water influences its uptake in the mealworms. • Mealworms eliminate Ag through the faeces and by shedding exuviae.

G R A P H I C A L A B S T R A C T
a b s t r a c t a r t i c l e i n f o

Introduction
The production of nearly 1000 tons of silver engineered nanomaterials per year (Giese et al., 2018) raises concerns about the bioaccumulation of silver nanoparticles (AgNPs) in organisms living in exposed environments. The estimated consumption of AgNPs amounts about 360 to 450 ton year −1 (McGillicuddy et al., 2017), with the largest proportion being used in consumer electronics, textiles, and for medical purposes (Giese et al., 2018). A substantial proportion of the nanoparticles (NPs) used end up in the sewage system and therefore may accumulate in sewage sludge (Blaser et al., 2008). When NPs are discharged to the receiving environment, their characteristics will no longer be the same as those in the pre-exposure phase due to their interactions with and transformations by their surroundings (Svendsen et al., 2020). Thus, their behaviour will also change depending on the environmental conditions at discharge. The tendency of silver (Ag) to form complexes with organic matter, chloride or other compounds such as Mn/Fe oxides in the soil may further reduce its mobility and bioavailability. Coutris et al. (2012) reported that AgNO 3 may immobilize rapidly in the soil, nevertheless there are indications that AgNPs may behave like a source of bioavailable Ag following slow dissolution over time (Diez-Ortiz et al., 2015). Soil pH was reported to affect the transformation of ionic Ag, Ag-NP and AgCl-NPs in soil (Sekine et al., 2015). AgCl was found to be formed and persist in acidic conditions, but in neutral to alkaline soils sulfur-bound Ag was the dominate Ag form (Sekine et al., 2015). In addition to pH, the redox conditions also affect the silver transformation, while under anaerobic and aerobic conditions, Ag 2 S and AgCl x (x-1)are expected to form, respectively (Levard et al., 2012).
Because of the presence of a large proportion of sulfide and the reduced conditions during wastewater treatment, Ag 2 S is expected to be formed from Ag ions or AgNPs (Kim et al., 2010), decreasing the adverse effects of silver in the environment due to its low water solubility (Coutris et al., 2012). Ag 2 S NPs have high chemical stability and may still be stable following seven months incubation in neutral, alkaline and acidic soils (Sekine et al., 2015). One of the main applications of sewage sludge is as agricultural land fertilizer (Fytili and Zabaniotou, 2008). In European countries, over 10 million tons of dry sewage sludge are produced each year, about 40% of which is applied to agricultural lands as fertilizer (Roig et al., 2012). Giese et al. (2018) predicted concentrations of AgNPs in non-sludgeamended soils for 2017 of 30 pg kg −1 that may increase to 10 μg kg −1 soil by 2050, while sludge-amended soils are predicted to contain 1130 pg kg −1 in 2017 and to achieve 30-40 times higher values than non-sludge-amended soils by 2050 (Giese et al., 2018). Besides studying their toxic effects, understanding AgNP bioaccumulation potential and toxicokinetics is critical for decoding their behaviour, mechanism of toxicity and environmental risk (Baalousha et al., 2016;Petersen et al., 2019;Wang, 2011). Metal based NPs don't necessarily degrade like pesticides, leading to potential long-term exposures. Toxicokinetics approaches may help with assessing the potential for bioaccumulation of different sizes of AgNPs and different forms of Ag (different coatings, NP or ionic etc.) by deriving uptake and excretion rate contants (Argasinski et al., 2012). Internal concentrations reflect the bioavailable concentration for uptake from the environment (Hug Peter et al., 2018) linked with the bioaccessible fraction associated with the internalization of metals through the cell membrane (Loureiro et al., 2018). After ingestion, the gut may work as a barrier for the uptake of NPs and its conditions may lead to the dissolution or aggregation of NPs in the lumen (Van Der Zande et al., 2020). While dissolved metals may be taken up through solute transporters in the cell membrane of gut epithelial cells, NPs are too big to be taken up in this way (Van Der Zande et al., 2020). Endocytosis is the most likely uptake mechanism for NPs in the gut, however, intracellular digestion is another possible uptake route for invertebrates (Van Der Zande et al., 2020). So, if Ag NPs are bioaccumulated, different biodynamics may be expected from those observed for Ag + (Croteau et al., 2011). Therefore, NP behaviour in the environment (e.g. dissolution rate, or partitioning between the soil solid phase and the soil pore water) depends on their physicochemical properties and understanding and parameterisation of these processes are crucial to accurately assess their environmental risk (Oomen et al., 2018;Sørensen et al., 2019).
To improve the available hazard data for AgNPs towards soil organisms, it is essential to assess AgNP toxicity and bioaccumulation in different terrestrial organisms (Tourinho et al., 2016). The aim of this study was to address the knowledge gap still existing for the hazard assessment of NPs in soil, and focus on the toxicokinetics of Ag in mealworms exposed to different AgNPs, exploring the roles of coating, particle size and sulface chemistry (e.g. sufidation) and their interaction with AgNP dissolution. Ag 2 S NPs were used to simulate ageing of AgNPs in the waste water treatment plant. The mealworm Tenebrio molitor (Coleoptera: Tenebrionidade) was chosen as a test organism. This holometabolic insect lays its eggs in moist soil and has three larval stages usually inhabiting the top 15 cm of the soil (Tashiro, 1990). The number of exuviae (exoskeltons) they release during their growth is dependent on the specimen but also on the surrounding environmental conditions (Greenberg and Ar, 1996;Morales-Ramos et al., 2010). Mealworms are commonly used as food for pet animals (e.g. frogs or snakes) and nowadays several approaches are being developed to use them as food (Eriksson et al., 2020) (protein source) in aquaculture and poultry (Morales-Ramos et al., 2010). Because of their long-term larval stage, their active feeding behaviour and commercial availability, mealworms are a suitable species for assessing exposure through both soil and food. As there is some debate in the literature as to whether exposure via soil is an important route of exposure for mealworms, and which fraction (soil and/or pore water) plays the most important role, we explore this question in detail, and some preliminary toxicity tests and approaches to unravel the uptake route are also presented. The results are presented as a one-compartment kinetics model with inert fraction (simulating a storage compartment, where detoxified forms are located) and differences in both uptake and elimination rates for the different AgNPs and forms are elucidated.

Materials and methods
2.1. Test organisms, soil and silver forms Tenebrio molitor larvae were obtained from a commercial breeder and kept in the laboratory at 20 ± 2°C with a 16:8 (light:dark) h photoperiod. Mealworms were separated according to their weight, in order to use animals weighing 22 ± 6 mg (mean ± standard deviation (SD); n = 216). Considering the test duration (six weeks), the larval stage was chosen, corresponding to this weight range, to ensure that the animals did not enter the pupa stage and that their active behaviour was similar during the test.

Nanoparticle characterization
The stability of the NPs was examined in suspensions prepared in a weak ammonium acetate solution (0.01 M) in ultrapure water (UPW, pH: 7) as per the method by Thomassen et al. (2001), by measuring the hydrodynamic size, polydispersity index (PDI), Zeta-potential (Zpotential) and dissolution at different time points. The suspension concentration used for characterization for all materials tested was 1 mg Ag. Z. Khodaparast, C.A.M. van Gestel, A.G. Papadiamantis et al. Science of the Total Environment 777 (2021) 146071 L −1 . Analyses were performed using a Malvern Zetasizer (Nano ZS, equipped with a LASER of 632.8 nm and a scattering angle of 173°) and Sarstedt polystyrene cuvettes (Ref: 67.742, 10 × 4 × 45 mm) or Malvern Zetasizer (DTS0170, disposable folded capillary cells) for hydrodynamic size and Z-potential, respectively. Measurements were carried out at 20°C following 2 min equilibration time at 2, 4, 24 and 48 h post-dilution to match the timepoints used for the dissolution study, as described below. This allowed us to study potential physicochemical changes taking place, due to for example Ostwald ripening (Yao et al., 1993) as soon as the NPs were dispersed into the respective solutions. Malvern Zetasizer Software (version 7.13) was used by applying the built-in values for the refractive indices (η) and absorption coefficients (α) for Ag and Ag 2 S. The obtained results are the average of five consecutive measurements from 14 replicates per measurement. The protocol of Avramescu et al. (2017) was followed for monitoring Ag dissolution. This protocol is commonly used in metal bioaccessibility assays (Dodd et al., 2013) and is a modification of the European Committee for Standardization (CEN) guidance EN 71-3:2019 on "Safety of toys -Part 3: Migration of certain elements" (European's Committee for Standardisation (CEN). CEN/TC 52., 2019). The dissolved Ag levels were determined in UPW suspensions of 1 mg Ag.L −1 using three replicates for each NP and at 5 timepoints (0, 2, 4, 24 and 48 h). The timepoints were chosen to allow studying the kinetics of the NPs dissolution behaviour and to recreate the respective dissolution curves during the high activity phase following NP dispersion. Samples were picked randomly and part of each was filtered using 0.02 μm pore-diameter syringe filters (Anotop™, Whatman), while the rest was used for simultaneous hydrodynamic size and Z-potential measurements. Ag concentrations were determined using an ICP-MS (PerkinElmer, Nexion 3000) following acidification of samples with pure ICP-grade HNO 3 (Sigma Aldrich; CAS Number 7697-37-2) to a final acid concentration of 2%. The calibration curves (correlation coefficient >0.999) were made using ICP-grade standards. 103 Rh was used as internal standard and all samples and standards were spiked for QC purposes. Samples failing the QC thresholds were recalibrated and reanalyzed. The acquired results are presented as percentage of dissolved Ag (%dissolved Ag = ([Dissolved Ag + ] / [initial total Ag concentration]) * 100%), according to the OECD classification scheme (OECD, 2015). TEM (transmission electron microscopy) images of AgNPs 3-8 nm and AgNPs 50 nm have been reported by Ribeiro et al. (2014) and Baccaro et al. (2018), respectively. The stability of pristine AgNPs and Ag 2 S NPs in different aquatic media has been also reported by Silva et al. (2020).

Acute toxicity test with AgNO 3
To find a suitable exposure concentration for the bioaccumulation tests, two acute toxicity tests were carried out considering exposures via soil and food using AgNO 3 , as it was expected to be the most toxic Ag form. For the soil exposure, soil was spiked with AgNO 3 to final concentrations of 1, 10, 100, 1000 mg Ag kg −1 dry soil. Three replicates with five mealworms each were used for each concentration and control. The five organisms were weighted and transferred into each plastic container (135 mm x 85 mm) with 200 g moist soil (40% WHC). The organisms were fed with 5 pieces of oat flakes (~150 mg) twice a week, breaking them up to ensure low area and prevent mealworms from resting on the flakes and thus preventing exposure. Although acute toxicity tests usually do not include food, this methodology tried to combine an acute toxicity test with the conditions needed for the bioaccumulation test.
For the food exposure test, oat flakes were spiked at 1, 10, 100, 1000, 1500 mg Ag kg −1 dry food. Thirty mL of spiked suspension was added to 20 g of oat flakes in a plastic pot and mixed for a few minutes. The pot was then closed to let the oats absorb the suspension while the mixture was stirred 3 to 4 times. The size of the plastic pot was big enough to avoid overlap of oat flakes. Three replicates were used for each concentration and control. Five organisms were added to each plastic pot including 500 mg spiked food (about 17 pieces of oat flakes) on the surface of 200 g clean, non-spiked soil. The food was collected from the soil surface and replaced twice a week to be sure that enough food was always available.
Only water was added to the soil or food for the control group. Test containers were closed with lids with 5-6 small holes for aeration and incubated at 20 ± 1°C and 16:8 (light:dark) h photoperiod.
Mealworms were monitored each week for up to 3 weeks (21 days, T21) for weight changes and mortality. Mealworm biomass gain (%) was calculated (based on number of alive mealworms) as: where WeightT21 is the individual weight at day 21 and WeightT0 the weight at the start of the experiment.

Understanding soil as an exposure route for mealworms
To understand the importance of mealworm exposure through soil, several methods were used to discriminate between pore water exposure and soil particle ingestion. Mealworms were exposed to 200 g of Lufa 2.2 soil (40% WHC) for 6 days, using 3 replicates with 10 mealworms each. To infer on the ingestion of soil particles by the mealworms, two methods were used: 1) acid digestion and 2) burning at 500°C. For acid digestion, mealworms were digested individually in concentrated acids to find any undissolved particles, as acid digestion is unable to dissolve soil silica particles. More details can be found below, in the mealworm digestion method for Ag analysis. For the burning method, mealworms were collected from the soil and left to empty their guts for 0, 24 h, 36 h and 48 h. Mealworms not exposed to soil (from cultures) were also tested. The ash samples of depurated (up to 48 h) mealworms and non-depurated ones (T0) collected after burning were weighed and differences between mealworms soil-exposed and unexposed to soil were analysed. This experiment was also important to determine the proper time needed for mealworms to depurate their guts in the case of soil ingestion. In addition, to investigate whether depuration affects the retrieval of Ag concentrations in mealworms following exposure, organisms were exposed to 20 g soil (Lufa 2.2 soil, 40% WHC, individually in plastic pot ø 65 mm) spiked with 100 mg Ag kg −1 dry soil of AgNPs 3-8 nm for 21 days. Three mealworms were collected at each sampling point (1, 3, 6, 9, 15, 21 days) and frozen individually (−20°C) with or without depuration. For depuration, the organisms were kept in an empty plastic pot for 24 h as mealworms have been reported to have <24 h gut transit time (Yang et al., 2015) and also based on the optimal depuration time determined as described above. Following acid digestion, the Ag concentration in the mealworms was determined using graphite furnace Atomic Absorption Spectrometry (AAS; PinAAcle 900Z, PerkinElmer, Singapore).
Faeces were also collected after exposure to soil and acid digested to infer the presence of soil particles. For that, 20 mealworms were exposed to 100 mg Ag kg −1 dry soil (3-8 nm, 50 nm or 60 nm AgNPs, AgNO 3 , or Ag 2 S NPs) in each of three replicates in a plastic pot with 450 g soil for six days followed by 24 h depuration in an emptypot. In the acid digestion procedure for faeces, the final digested solution was centrifuged (3000 rpm, 10 min) and the pellet and the supernatant were separated. The pellet was dried at 60°C for two days and weighted to account for the amount of soil (percentage of soil in faeces = total soil dry weight * 100 / total faeces dry weight). The supernatant was measured for total Ag content in faeces by graphite furnace AAS (PinAAcle 900Z, PerkinElmer, Singapore).
To further explore the soil and/or pore water uptake, the soil was dyed at different moisture contents. Lufa 2.2 soil was coloured by mixing 28 mL of edible color (GLOBE GREEN DYE, for Confectionery) with 400 g soil. Then the soil was dried to a constant weight in an oven at 60°C for two days and finally, the moisture content was adjusted to 20% or 40% of the WHC or the soil was used completely dry. Mealworm intestines were dissected and monitored under a stereomicroscope after exposure to dyed soil from 2 to 48 h (2, 4, 6, 12, 24, 48 h) and the pictures recorded whereby color intensity could be detected.

Soil exposure: experimental setup for bioaccumulation test
Based on the results from the acute toxicity test (low toxicity) and the lack of information on the efficacy of exposure to Ag through soil, the concentration of 100 mg Ag kg −1 dry soil was chosen for soil exposures where low Ag uptake or high Ag elimination rates were expected. Although this concentration is far from the environmentally expected Ag concentration, it enabled the analysis of Ag and assessment and understanding of Ag toxicokinetics also at lower exposure concentrations. Therefore, Lufa 2.2 soil was spiked with 100 mg Ag kg −1 dry soil for 3-8 nm, 60 nm or 50 nm AgNPs and AgNO 3 , and 22 and 226 mg Ag kg −1 of Ag 2 S NPs. Different concentrations were used for Ag 2 S NPs, Because of an incorrect labelling of the original stock, the lower concentration (22 mg kg −1 ) was only about 25% of the nominal one. The higher concentration (226 mg Ag k −1 ), was included because of the expected low bioavailability and bioaccessibility of Ag 2 S NPs, also was a bit lower than anticipated because of the incorrect labelling of the original stock. All Ag forms were spiked into the soil as a suspension/solution. The soil was moistened up to 40% of its maximum WHC by adding deionised water. In the case of 3-8 nm and 60 nm AgNPs, where the concentration in the stock solution was rather low, the spiking was done in three steps with soil allowed to air-dry at 20°C between each step. Following three days of equilibration, soil exposures started. To determine the soil pH, 5 g of soil were shaken with 25 mL of 0.01 M CaCl 2 solution for 2 h, with the pH measured after 2 h settling of the suspension (Tourinho et al., 2016). To measure available silver concentrations in the soil, the CaCl 2 suspension was filtered through a 0.45 μm membrane filter after precipitation overnight and analysed by graphite furnace AAS (PinAAcle 900Z, PerkinElmer, Singapore). Mealworms were exposed to the spiked soil for a 21-day uptake phase and then transferred to clean soil for a 21-day elimination phase. In total 36 animals were used for each treatment, and 12 for the control. All were kept individually in plastic pots (ø 65 mm) containing about 20 g moist soil. Three replicates (with one mealworm each) were randomly collected per sampling point (1, 3, 6, 9, 15, 21 days) for both the uptake and elimination phases and frozen individually (−20°C). The test was carried out at 20°C and 16:8 h (light:dark) photoperiod and mealworms were fed with oat flakes (1 piece of oat which was replaced twice weekly).

Total and dissolved Ag in soil pore water
In a separate experiment, three replicate plastic pots were filled with 450 g spiked soil (100 mg Ag kg −1 dry soil) for each treatment (Ag form). At each sampling point (0, 1, 6, 15, 21 days) the soil pore water was extracted by saturating soil samples (the equivalent of 25 g dry weight) with deionised water for 24 h, followed by centrifugation through a 70 μm nylon filter (soaked in 0.1 M CuSO 4 ) at 2000 g for 1 h.
To determine the Ag dissolved during 21 days, 2 mL of the collected soil pore water was centrifuged at 3000 rpm for 1 h using a 3 kDa membrane filter. The soil pore water and its ultrafiltrate were acidified with concentrated HCl to a total concentration of 3% HCl and stored for Ag analysis by graphite furnace AAS (PinAAcle 900Z, PerkinElmer, Singapore).
The first-order decay model was used to calculate the decrease rate of total and ionic Ag in the soil pore water, as follows: where C(t) = concentration in the pore water at time t (μg Ag L −1 ), C 0 = concentration in the pore water at t = 0 (μg Ag L −1 ), K = decrease rate constant (day −1 ).

Food exposure: experimental setup for bioaccumulation tests
Oat flakes were spiked with a solution/suspension of the different Ag forms to reach a nominal concentration of 100 mg Ag kg −1 dry food as described above. For the control, only deionised water was added to the oats. Spiked food was air dried at 25°C for a few days. Plastic pots (ø 65 mm) containing 20 g moist (40% of WHC) Lufa 2.2 soil were used for the test. Mealworms were exposed to contaminated food for 21 days as the uptake phase and then offered clean food for another 21 days (elimination phase). A single oat piece (~30 mg) was added on the soil surface and replaced three times a week to ensure the permanent presence of food. Three mealworms were frozen (−20°C) at each sampling point (1, 3, 6, 9, 15 and 21 days) for both the uptake and elimination phase. For each treatment, 36 organisms were used, and another 12 for the control, and kept individually as in the soil exposure test (20°C and 16:8 (light:dark) h photoperiod). Each mealworm was allowed to depurate the gut for 24 h in an empty plastic pot after which the animal and exuviae (if available) were frozen, dried and stored for Ag analysis. Mealworms were monitored every day for moulting, so the complete exuviae could be collected after shedding and cleaned gently with a brush and Milli-Q water.

Total Ag analysis in mealworms
Mealworms were digested with a mixture of concentrated HNO 3 : HCl (3:1. v/v; PanReac AppliChem, Trace analysis) as described by Ribeiro et al. (2017). Concentrated HCl (37%) and HNO 3 (69%) were added to the dried samples for the digestion. Then the acids were allowed to evaporate to less than 2 mL and diluted with 1% HCl up to a volume of 45 mL. Blanks and reference material were analysed in triplicate at each digestion run. The digests were analysed for Ag by graphite furnace AAS (PinAAcle 900Z, PerkinElmer, Singapore). Analysis of the certified reference materials DOLT-3 and DOLT-5 showed acceptable recoveries of Ag (95% ± 12%, mean ± SD; n = 23). To find the limit of detection (LOD) for AAS measurements a blank was measured 20 times; the standard deviation was multiplied by three to get the detection limit. The LOD for graphite furnace AAS was 0.026 μg Ag L −1 .
Soil samples were dried at 50°C and then digested for 7 h in a mixture (4:1. v/v) of concentrated HCl and HNO 3 (J.T. Baker; 37% and 70%, respectively) in closed Teflon containers, in an oven (Binder ED 53) at 140°C. The digests were diluted with demineralized water up to 10 mL and analysed by flame AAS (AAnalyst 100, PerkinElmer, USA). Recovery of Ag from the DOLT-4 reference material was 90% ± 2.4% (n = 2).
Dried food (spiked oat) was digested using a microwave (Berghof speed wave) with the same mixture of concentrated acids as for the soil samples. Digestion was carried out applying a temperature ramp of 10 up to 180°C (25 min), then held at 180°C (25 min) and ramp 5 down to 100°C (10 min). The digests were diluted with Milli-Q water up to a volume of 10 mL. The recovery of Ag from the reference material (DOLT-5) for the digestion of food samples was acceptable (99% ± 7%, n = 3). Flame AAS (AAnalyst 100, PerkinElmer, USA) was used to measure the total Ag content.

Toxicokinetic models
A one-compartment first order model with inert fraction (Fi) was used to describe the data. An inert fraction is the stored fraction of metal in the organism related to detoxification processes which is not excreted during the elimination phase. Fi ranged from 0 to 1, higher values show slow or slight elimination of metal from the storage compartment (Tourinho et al., 2016;Vijver et al., 2006). Eq. (3) was fitted to data from the uptake phase, while Eq. (4) was used for the elimination phase (van den Brink et al., 2019).
In these equations, Q (t) = internal Ag concentration in the mealworms at t days (μg Ag g −1 dry body weight); C 0 = background internal concentration at t = 0 (μg Ag g −1 dry body weight); K 1 = uptake rate constant (g soil or food g animal −1 day −1 ); K 2 = elimination rate constant (day −1 ); C exp = Ag exposure concentration (mg Ag kg −1 dry soil or food); t = time (days); t c = time at which the animals were transferred to clean soil or clean food (days); Fi = inert fraction; K growth = the growth rate constant calculated using the Von Bertalanffy model (Paine et al., 2012).
In case of soil exposure, the contribution of soil and pore water to total Ag uptake was taken into account and for that purpose Eqs. (5) and (6) were fitted to data from the uptake phase and the elimination phase, respectively (Ribeiro et al., 2017): In these equations, C exp soil = Ag exposure concentration in soil (mg Ag kg −1 dry soil); C exp SPW = Ag exposure concentration in soil pore water (mg Ag L −1 ); K soil = uptake rate constant from the soil (g soil g animal −1 day −1 ); K SPW = uptake rate constant from the soil pore water (L spw g animal −1 day −1 ).
Eqs. (7) and (8) were used to estimate the relative contribution of soil and soil pore water uptake to the total Ag uptake in the mealworms: Uptake from soil (%): Uptake from soil pore water (%):

Statistical analysis
The stability of the NPs over time was analysed using the Kruskal-Wallis H test (p < 0.05), with time as the independent variable and the respective physicochemical descriptor (Z-potential, hydrodynamic diameter) as the dependent variable. The Dunn-Bonferroni post-hoc test was used to test for significant changes between the different timepoints. Two-way analysis of variance (ANOVA) followed by Dunnett's post-hoc test (p < 0.05) was performed to determine statistical significance of differences between treatments and the control over time in the acute toxicity test for biomass gain. The significant differences of the total Ag concentration between depurated and nondepurated mealworms were checked with one-way ANOVA followed by Tukey's post-hoc test (p < 0.05). The one-way ANOVA followed by the Holm-Sidak method (p < 0.05) was used to demonstrate the significant differences of total Ag concentrations in soil, 0.01 M CaCl 2 extractable, and soil pore water between the spiked soil with different Ag forms. Two-way ANOVA followed by the Holm-Sidak method (p < 0.05) was used to analyze statistically significant differences over time in total Ag and dissolved Ag concentrations measured in the soil pore water extracted from the spiked soil with different Ag forms and in the mealworm body concentrations for food and soil exposure during the uptake and elimination phases for the different Ag treatments. SigmaPlot 14.0 software was used for determining normality, homoscedasticity of the data and running the ANOVA.
The toxicokinetics parameters were estimated by non-linear regression in SPSS (version 23). Akaike information criteria (AIC and AICc) were used to find the best-fitting model for each data set (Akaike, 1974). A Generalised Likelihood Ratio Test (Sokal and Rohlf, 2012), was used to test for significance of the differences between K 1 or K 2 values for the different Ag forms.

Nanoparticle characterization
NP stability over time was tested in UPW, with pH regulated to 7 using ammonium acetate. The monitored descriptors Z-potential, hydrodynamic diameter (including the PDI as a measurement quality parameter) and dissolution are presented in Fig. S1 in the Supplementary Information.
The hydrodynamic size of the 3-8 nm AgNPs initially decreased over time, which can be attributed to dissolution or potential deagglomeration of the particles from the stock solution. These differences are further supported from the pairwise timepoint comparison, as the size difference between 0 and 2 h was significant (Kruskal-Wallis H test followed by the Dunn-Bonferroni post-hoc test, p < 0.001). At later timepoints, significant size drops (4 versus 8 h: p = 0.004) were recorded, while from 8 h onward post-suspension no significant differences (p > 0.05) were observed. The 50 nm AgNPs presented a small, but significant, size-drop that was observed solely between 8 and 24 h (p = 0.007), with all other pairwise comparisons not being significant (p > 0.05).
The dissolution behaviour (Fig. S1E) of the AgNPs was consistent with the results for the Z-potential and hydrodynamic diameter. Ag 2 S NP dissolution was in all cases <0.02%. The 60 nm AgNPs presented consistent dissolution between 27.2%-29.3% at all timepoints. The 3-8 nm and 50 nm AgNPs showed rapidly increasing % dissolved Ag during the early post-dispersion period, reaching the expected plateau at 48 h post-dispersion. The observed % dissolved Ag in the time period between dispersion and 48 h post-dispersion were 6.5%-21.1% and 5.27%-26.1% for the 3-8 nm and 50 nm AgNPs, respectively.

Acute toxicity test
The acute toxicity test with AgNO 3 showed maximum mortalities of mealworms exposed to the spiked soil and food of 34% (±41%, mean ± SD, n = 3) and 27% (±11%, mean ± SD, n = 3), respectively, after 21 days. No significant difference (p > 0.05) in mealworm mortality was found compared to the control for any of the soil or food concentrations tested (Fig. S2). Mealworms exposed to 1000 mg AgNO 3 kg −1 via soil showed a significant decrease (Dunnett's test, p < 0.05) of biomass gain from the 7th day of exposure onwards (Fig. 1). Similarly, at 1500 mg Ag kg −1 dry food and after 21 days the mealworms also presented a significant decrease in their biomass gain (compared to the control) (Dunnett's test, p < 0.05) (Fig. 1).

Soil ingestion by mealworms
The acid digestion and burning/ashing methods used did not show any evidence of soil ingestion by the mealworms (Fig. S3). However, the mealworm intestine was totally coloured after just 2 h exposure to the dyed soil moistened at 40% of its WHC (Fig. S4). When kept on completely dry soil, the change of the intestinal color was not as clear as with the moist soil. Mealworm faeces showed a notable amount of soil, amounting (±SD, n = 3) to 9.6 ± 3.19 mg soil dry weight in pooled faeces from 20 mealworms, after the exposure to AgNO 3 spiked soil. The results of the acid digestion of mealworm faeces are shown in Table 1. The faeces contained 51 to 71% of soil, while the concentration of Ag in the faeces was at least 60.5 mg Ag kg −1 faeces for AgNPs 60 nm and maximum 194 mg Ag kg −1 faeces for Ag 2 S NPs. The total Ag concentration in the mealworms exposed to AgNO 3 was 21% lower (±26%, mean ± SD, n = 18) when correcting for the gut content: Fig. S5.
To check the difference between depurated and non-depurated animals, in the bioaccumulation test with the 3-8 nm AgNPs mealworms were analysed with and without their gut content. There was no significant difference in the measured total Ag concentrations in the mealworms (Tukey's post-hoc test, p > 0.05), except for the third day (Tukey's post-hoc test, p < 0.05) of the uptake phase of the bioaccumulation test (Fig. S6).

Ag concentrations in soil and food
The background Ag concentration in Lufa 2.2 soil was below the detection limit while in the case of the oat flakes it was 0.01 mg Ag kg −1 .
The measured Ag concentrations in the spiked soil (total, CaCl 2 extractable and in pore water) and spiked food are given in Table 2. The measured concentrations were used in the toxicokinetic models to calculate Ag uptake and elimination rate constants in the mealworms.
The total and dissolved concentrations of Ag in the pore water from AgNO 3 spiked soil significantly decreased from day 6 onwards with decrease rates of 0.25 and 0.34 day −1 for total and dissolved Ag, respectively (Holm-Sidak method, p < 0.05) ( Table 3). No significant (p > 0.05) changes were detected in the total and dissolved porewater Ag concentrations for the 3-8 nm AgNPs, while only dissolved Ag concentration showed a decrease rate of 0.03 per day. Total Ag concentrations in the pore water of 50 nm AgNP spiked soil displayed no significantly (p > 0.05) differences during 21 days, however, the dissolved Ag concentration was significantly lower after 21 days compared to the start (Holm-Sidak method, p < 0.05). The 60 nm AgNPs showed significantly (Holm-Sidak method, p < 0.05) lower total Ag concentrations in the pore water after 21 days compared to days 0 and 1, while the dissolved Ag concentration was significantly (Holm-Sidak method, p < 0.05) lower than day 0 at all sampling times; decrease rates were 0.10 and 0.34 day −1 , respectively.
The total Ag concentration of pore water in case of Ag 2 S NPs after one day was significantly higher than the first and 15th day. The dissolved Ag concentrations in pore water of Ag 2 S NP spiked soil showed no significant changes during the 21-day incubation period (Table 3).
AgNO 3 spiked soil had significantly higher porewater Ag concentrations compared to all other Ag forms except for the 60 nm AgNPs (Holm-Sidak method, p < 0.05). Total Ag concentrations in pore water from soil spiked with pristine AgNPs and Ag 2 S NPs did not differ significantly (p > 0.05). Although the Ag concentration of the soil spiked with Fig. 1. Effect of AgNO 3 on the biomass gain (%) of the mealworm Tenebrio molitor following 21 days exposure in Lufa 2.2 soil (A) or food (B). * indicates significant difference compared to the control according to a two-way ANOVA, followed by Dunnett's post hoc test (p < 0.05). Error bars represent standard error.

Table 1
The measured total Ag concentrations in the Lufa 2.2 soil spiked with different Ag forms and in the faeces of mealworms (Tenebrio molitor) (pooled 20 mealworms) exposed to this soil. Also shown is the percentage of soil in the faeces of the mealworms. Data are shown as average ± SD; n is the number of replicates used. 64.3 ± 6.20 AgNPs 60 nm 101 ± 6.66 60.5 ± 17.6 (n = 2) 53.9 ± 26.8 (n = 2) Ag 2 S NPs 209 ± 17.5 194 ± 52.9 65.8 ± 8.9 Ag 2 S NPs was twice as high as for the other Ag forms, it had significantly (Holm-Sidak method, p < 0.05) lower dissolved Ag concentrations in the pore water at all sampling times (Table 3).

Soil exposure toxicokinetics
The pH CaCl2 of the Lufa 2.2 soil (Table S1) was slightly higher when spiked with AgNO 3 or 60 nm AgNPs, but was hardly affected by the other Ag forms. The pH of Ag spiked soils never differed more than 0.5 units compared to the control.
Although less than 20% mortality was observed during the bioaccumulation test with the different Ag forms, the mealworms gained little weight (growth constant rate (K growth ), 0.09 ± 0.11 per day, mean ± SD, n = 7) during the 42 days test period. Therefore, a growth constant rate (K growth ) was applied to the toxicokinetics model to consider the possible dilution of the body Ag concentration in the organism by growth (Ardestani et al., 2014). At the beginning of the bioaccumulation test (T 0 ), the background Ag concentration (C 0 ) in the mealworms was 0.24 ± 0.14 μg Ag g −1 dry body weight (mean ± SD, n = 12).
Uptake and elimination patterns and kinetics of Ag in the mealworms exposed to different Ag forms, estimated with the first model based on the soil exposure (Eqs. (3) and (4)), are displayed in Fig. 2 and Table 4. Total Ag concentrations in mealworms exposed to AgNO 3 reached steady state after 6 days with a level of 34 μg Ag g −1 dry body weight after 21 days of the exposure period. The Ag body concentration declined after transferring the animals to the clean soil, but did not completely return to the background level.
The pristine 3-8 nm and 60 nm AgNPs almost showed the same patterns of Ag uptake and elimination in the mealworms (Fig. 2). Steady state was reached after 6 days with body concentrations of 23.8 and 22.9 μg Ag g −1 dry body weight, respectively after 21 days of uptake. Also elimination rate constants for the Ag taken up from the 3-8 nm and 60 nm AgNPs were similar ( Table 4).
The 50 nm AgNPs were the only pristine AgNPs that pursued a different pattern, reaching steady state body Ag concentrations only after 15 days (Fig. 2). Although a lower Ag body concentration of 11.3 μg Ag g −1 dry body weight was reached at the end of the exposure period compared to other pristine AgNPs, it could not be eliminated as well as for the other pristine AgNPs leading to the highest inert fraction of all the treatments (Fi = 0.54).
After 21 days, the Ag concentration in mealworms exposed to Ag 2 S NPs at 22 and 226 mg Ag kg −1 dry soil was 2.6 and 15.4 μg Ag g −1 dry body weight, respectively. At both exposure concentrations, elimination was faster than for the other Ag forms. At the exposure concentration of 226 mg Ag kg −1 dry soil, which was twice than that for the other Ag forms, the body concentration at the end of the elimination phase was 0.28 ± 0.29 μg Ag g −1 dry body weight (mean ± SD, n = 3), close to the background Ag concentration.
The kinetics parameters were the same for AgNO 3 and for the 3-8 nm and 60 nm AgNPs with K 1 = 0.27-0.32 g soil g animal −1 day −1 and K 2 = 0.81-1.09 day −1 (Table 4). Uptake rate constant for the 50 nm AgNPs (K 1 = 0.06 g soil g animal −1 day −1 ) however, differed significantly (X 2 (1) > 3.84; p < 0.05) from all other Ag forms except for Ag 2 S NPs (226 mg Ag kg −1 dry soil), while the elimination rate constant (K 2 = 0.12 day −1 ) was significantly different (X 2 (1) > 3.84; p < 0.05) compared to all Ag forms. The parameters for the Ag uptake from Ag 2 S NPs (226 mg Ag kg −1 dry soil) were similar to all other Ag forms tested in this study (K 1 = 0.13 g soil g animal −1 day −1 , K 2 = 1.35 day −1 ), except for the K 2 for the 50 nm AgNPs. Only for the lower Ag 2 S NPs concentration Table 2 Nominal and measured total Ag concentrations and the 0.01 M CaCl 2 -extractable Ag concentrations in the Lufa 2.2 soil, total Ag concentration in soil pore water and in food spiked with different Ag forms (nominal concentration of 100 mg Ag kg −1 dry food). Data are shown as average ± SD for soil; only one replicate was used for food analysis. Statistically significant differences (one-way ANOVA followed by Holm-Sidak Method (p < 0.05)) within each column are shown in different capital letters.  Table 3 Total Ag and dissolved Ag concentrations measured in the pore water of Lufa 2.2 soil spiked at a nominal concentration of 100 mg Ag kg −1 with different Ag forms at each sampling time (days 0, 1, 6, 15, 21) and calculated decrease rate constants. Data are shown as average ± SD (n = 3). Statistically significant differences (two-way ANOVA followed by Holm-Sidak Method (p < 0.05)) within each column (treatments) and row (time) are shown in different capital letters and small letters in italics, respectively.  Table 4 Uptake and elimination kinetic parameters for Ag in mealworms (Tenebrio molitor) exposed to 100 mg Ag kg −1 dry soil of different Ag forms and 22 or 226 mg Ag kg −1 dry soil of Ag 2 S NPs. Model 1 describes results of a standard one-compartment model assuming uptake from soil, model 2 considered uptake through soil and soil pore water (SPW). K 1 is the uptake rate constant, K 2 the elimination rate constant, K soil the uptake rate constant from the soil, K SPW uptake rate constant from the soil pore water, and Fi the inert fraction. Corresponding 95% confidence intervals are given between brackets. Significant differences (X 2 (1) > 3.84; p < 0.05) between K 1 and K 2 values for the different Ag forms, as shown by a likelihood ratio test, are indicated with different letters within each column.

Model
Ag form K 1 (g soil g animal −1 day −1 ) K 2 (day −1 ) K soil (g soil g animal  (22 mg Ag kg −1 dry soil), K 1 and K 2 values were much higher but not significantly (X 2 (1) < 3.84; p > 0.05) different due to a large variation. Moreover, the lowest Fi value of all Ag forms was found for the higher Ag 2 S NPs concentration (226 mg Ag kg −1 dry soil) ( Table 4).
When accounting for both exposure routes (model 2, Table 4), AgNO 3 showed the highest uptake rate constant from soil pore water (K spw ) of all Ag forms. Calculating the relative contribution (Table 5) of exposure routes revealed also higher Ag uptake from soil pore water (62%) than soil in mealworms exposed to AgNO 3 . Of the pristine AgNPs, 60 nm AgNPs had higher K spw and also higher Ag uptake from pore water than soil, similar to AgNO 3 . The uptake of Ag from the 50 nm AgNPs showed equal contributions from the soil and soil pore water, while for AgNPs 3-8 nm Ag uptake was highest from the soil (57%). For Ag 2 S NPs, 90% of the Ag uptake was from the soil ( Table 5).
The statistical comparison of the raw experimental data (two-way ANOVA with the Holm-Sidak post-hoc test) demonstrated that significant difference exists for the combined effect of the uptake and elimination phases (p < 0.001) and when we consider the different treatments (p < 0.001). The Holm-Sidak post-hoc test demonstrated that significant difference exists between the Ag 2 S (22 mg Ag kg −1 dry soil) with all the other Ag treatments (p = 0.02 for the 50 nm Ag NPs and p < 0.001 for the rest), while no other statistical significant difference (p > 0.05) was observed during any other pairwise comparison.
During the uptake phase (0-21 days), statistical significant difference was observed (two-way ANOVA and Holm-Sidak post-hoc test) for both the combined effect of the different treatments over time (p = 0.011) and the pairwise treatments comparisons (p < 0.001). The pairwise description of the various treatments using the Hom-Sidak post-hoc test demonstrated significant differences between the Ag 2 S (22 mg Ag kg −1 dry soil) with the other Ag treatments (p < 0.001 for all), with the exception of the 50 nm AgNPs treatment (p = 0.412). From the remaining pairwise comparisons, significant difference was observed in the case of the 50 nm AgNPs with the 60 nm AgNPS (p = 0.002) and the AgNO 3 (p = 0.014) treatments.
During the elimination period (22-42 days), the statistical analysis of two-way ANOVA with the Holm-Sidak post-hoc test demonstrated no significant differences for the combined effect of the different treatments over time (p = 0.808), but there is significant differences between the individual treatments (p < 0.001). The pairwise comparison of the different treatments using the Holm-Sidak post-hoc test, demonstrated that no significant difference (p = 0.998) exists between the Ag 2 S (22 mg Ag kg −1 dry soil) and the Ag 2 S (226 mg Ag kg −1 dry soil) treatments, but the Ag 2 S (22 mg Ag kg −1 dry soil) is significantly different with all other treatments (3-8 nm AgNPs: p = 0.06, 50 nm AgNPs: p < 0.001, 60 nm AgNPs: p = 0.019, AgNO 3 : p < 0.001). The Ag 2 S (226 mg Ag kg −1 dry soil) differs significantly with the 50 nm AgNPs (p < 0.001) and the AgNO 3 (p < 0.001) treatments. No other statistical significant differences (p > 0.05) were observed between the remaining pairwise comparisons.

Food exposure toxicokinetics
The Ag concentration in mealworms fed with AgNO 3 spiked food reached steady state after 3 days of exposure and was 10.0 μg Ag g −1 dry body weight after 21 days (Fig. 3). After feeding with clean food, Ag body concentration decreased slowly (Fi = 0.57) ( Table 6).
The 3-8 nm and 60 nm AgNPs followed a similar pattern to reach steady state body Ag concentrations in the mealworms after one day of exposure to spiked food (Fig. 3). The Ag concentrations in mealworms exposed to 3-8 nm and 60 nm AgNPs were 10.8 and 10.0 μg Ag g −1 dry body weight, respectively after 21 days of exposure. The elimination patterns of 3-8 nm and 60 nm AgNPs were also comparable with a considerable fraction of the Ag retained at the end of the depuration period (Fi = 0.69-0.72). The uptake of Ag from the 50 nm AgNPs reached steady state after 3 days (Fig. 3), but showed a higher elimination than for the other pristine AgNPs (Table 6). In mealworms exposed to Ag 2 S NPs, steady state Ag concentration was reached after one day of exposure, but elimination was faster than for all other Ag forms with the body concentrations decreasing to the background level (Fig. 3, Table 6).
For AgNO 3 and the 50 nm AgNPs, the K 1 and K 2 values were similar and much lower than for the other Ag forms ( Table 6). The estimated K 1 values for Ag uptake in the mealworms exposed to food spiked with 3-8 nm and 60 nm AgNPs and Ag 2 S NPs were equal at 1.96-2.18 g food g animal −1 day −1 (Table 6) and so were the K 2 values (13.5-21.7 day −1 ). The Fi value for AgNO 3 (0.57) was similar to that for the 3-8 nm and 60 nm AgNPs (0.69-0.72), while the Fi for Ag 2 S NPs was the lowest among all Ag forms ( Table 6). The Fi value for AgNPs 50 nm (0.31) was different to that for the other pristine AgNPs.
When comparing the combined uptake and elimination patterns, no significant differences could be observed either between the different Ag treatments (two-way ANOVA, p = 0.098) or for the combined effect of the over time period of the different treatments (p = 0.353). Similarly, no statistically significant differences were observed during the uptake phase (days 1-21), with p values for the individual treatments and the differences of the treatments over time being 0.876 and 0.465, respectively. During the elimination phase, statistically significant differences were observed between the different treatments (p < 0.001). Based on the Holm-Sidak post-hoc test, Ag elimination in Ag 2 S NP exposures differed significantly from all other treatments (3-8 nm AgNPs: p < 0.001; 50 nm AgNPs: p = 0.003; 60 nm AgNPs: p < 0.001; AgNO 3 : p < 0.001). The 3-8 nm AgNPs elimination phase did not significantly (p > 0.05) differ with other than the Ag 2 S NPs treatments, while the 50 nm AgNPs (besides Ag 2 S NPs) differed significantly with the 60 nm AgNPs (p = 0.005) and AgNO 3 (p = 0.014). The 60 nm AgNPs and AgNO 3 elimination differed significantly from the Ag 2 S and 50 nm AgNPs as presented above, which means that these can be grouped together considering the elimination rate constants.
Total Ag concentrations in the exuviae recovered from exposed mealworms are reported in Table 6. Exuviae from mealworms exposed to Ag 2 S NPs had the highest Ag amount (71.8% of the body burden), followed by those from AgNO 3 (6.4%), and AgNP (1.2-6.2%) exposures.

Discussion
The acute toxicity test showed that the mealworm Tenebrio molitor is tolerant to silver, exposed either via spiked soil or spiked food and no increased mortality was seen even at very high concentrations of AgNO 3 . The high tolerance to silver brought questions about whether or not this Table 5 The relative contribution of uptake from soil and soil pore water (SPW) to the total uptake of Ag in mealworms exposed to different Ag forms in Lufa 2.2 soil. K soil is uptake rate constant from the soil, K SPW uptake rate constant from the soil pore water, C exp soil Ag exposure concentration in soil, and C exp SPW Ag exposure concentration in soil pore water.
Ag form K soil (g soil g animal −1 day −1 ) K SPW (L spw g animal −1 day −1 ) C exp soil (μg Ag g −1 soil) C exp SPW (μg Ag L spw −1 ) Uptake from soil (%) Uptake from SPW (%) terrestrial organism is exposed in its larval stage. Bioaccumulation tests may provide an answer to that, along with other approaches that can help unravel exposure routes through soil, soil pore water and food. These tests were run with soil or food spiked with different silver forms, including different pristine AgNPs, ionic Ag (AgNO 3 ) and Ag 2 S NPs to simulate the ageing of AgNPs in wastewater treatment plants. Vijver et al. (2003) performed different tests (including visual inspection of soil particles in the dissected organisms and in digests of organisms, and also ashing of the lyophilized animals at 500°C) and concluded that mealworms do not ingest soil. Our results contradict these findings, which were only possible by using different complementary methods. The first results obtained with the acid digestion and Table 6 Uptake and elimination kinetic parameters for different Ag forms in mealworms (Tenebrio molitor) exposed to 100 mg Ag kg −1 food. K 1 is the uptake rate constant, K 2 the elimination rate constant and Fi the inert fraction. 95% confidence intervals are given in brackets. Statistically significant differences (X 2 (1) > 3.84; p < 0.05) in K 1 and K 2 values between the Ag forms, as determined by a likelihood ratio test, are shown by different letters within each column. The average amount of Ag (±SD) in the exuviae of mealworms exposed to 100 mg Ag kg −1 in food spiked with different Ag forms are given in the last column. No common letter showed a significant difference between kinetics.
Z. Khodaparast, C.A.M. van Gestel, A.G. Papadiamantis et al. Science of the Total Environment 777 (2021) 146071 burning were in agreement with the study of Vijver et al. (2003) and almost no soil particles were obtained from mealworms exposed to soil. Burning/ashing showed no difference between the mealworms exposed and non-exposed to soil. However, adding the soil dyeing method showed that mealworm guts were totally coloured, which was confirmed by the acid digestion of faeces. In spite of this finding, the Ag concentrations in depurated and non-depurated mealworms following exposure to 3-8 nm AgNPs showed no differences. Therefore, we can conclude that soil may have a low residence time in mealworms and the soil present in the gut by itself does not make a significant contribution to the final Ag concentration in the mealworms. This may suggest that the amount of ingested soil is small and not contributing much to the Ag body concentrations during the bioaccumulation test. Pore water ingestion may be the more likely route, as dermal uptake seems to be unlikely due to the hard wax-coated cuticle (water-impermeable integument) (Vijver et al., 2003). Water plays a crucial role in the growth and development of T. molitor larvae and it is therefore the more plausible exposure route in soil (Murray, 1968).

Soil exposure
The uptake and elimination parameters (K 1 , K 2 and Fi) obtained from the 3-8 nm and 60 nm AgNPs and ionic Ag exposures were almost the same, and show the ability of the mealworms to eliminate Ag after being transferred to clean soil and detoxify this metal primarily by excretion (Ardestani and Van Gestel, 2013). On the other hand, Waalewijn-Kool et al. (2014) exposed Folsomia candida to the same 3-8 nm AgNPs and reported a significantly higher bioaccumulation factor for AgNO 3 exposure, which was due to the higher elimination rate constant for nanoparticulate than for ionic Ag.
Moreover, AgNP characteristics affected their bioavailability and their bioaccumulation in the mealworms. The uptake (K 1 ) and elimination (K 2 ) rate constants for the citrate-capped 50 nm AgNPs were significantly lower than those for other AgNPs (paraffin and PVP coated, and Ag 2 S) and ionic Ag, while Fi was higher. This remarkable difference may suggest a different behaviour of the 50 nm AgNPs and also different Ag handling by the mealworms. The significant differences in the toxicokinetics between the 50 nm and 60 nm AgNPs may be due to the low stability of 60 nm AgNPs compared to 50 nm AgNPs as was shown by monitoring the NP stability in the stock solutions in water. This was also supported by the higher dissolution of Ag in the soil pore water for 60 nm AgNPs when compared to the 50 nm AgNPs, despite the fact that the PVP-coated NPs should be more sterically stable. The 50 nm AgNPs were stabilised with sodium citrate while the 60 nm AgNPs were coated with PVP and the behaviour of AgNPs is related to their size, shape, size distribution, and coating. Some types of coatings, like citrate-based, are of low molecular weight and protect the NPs by electrostatic repulsion, but a hydrophilic high molecular weight polymer like polyvinylpyrrolidone (PVP) imparts steric stabilization (Burkowska-But et al., 2014). Coating has an impact on dissolution and agglomeration of NPs and depends on the media in which they are released (Topuz and van Gestel, 2017). Besides, AgNPs' bacterial efficacy is often related to their coating and, consequently, dissolution rate. Burkowska-But et al. (2014) showed that PVP-coated AgNPs had a substantially higher antibacterial efficacy than citrate-coated AgNPs (Burkowska-But et al., 2014), which agrees with the lower dissolution rate of citrate coated 50 nm AgNPs when compared to the one reported for the 60 nm AgNPs. Thus, the significantly lower uptake rate constant of the 50 nm AgNPs is most likely related to its characteristics. Confirming the findings for the mealworms, a study with differently coated 25 nm AgNPs in Enchytraeus crypticus showed similar uptake of PVP-coated AgNPs and ionic Ag while the uptake of citrate-coated AgNPs was much lower (Topuz and van Gestel, 2015). Studying the effect of different coatings (bovine serum albumin, chitosan and PVP) and sizes (20, 35 and 50 nm) on AgNP bioaccumulation in Lumbricus rubellus revealed that only for PVP-coated particles, size had an influence on Ag uptake but not for the other coatings. Therefore, the effect of the size of AgNPs (20-50 nm) on its uptake seems to be limited compared with the impact of the type of coating (Makama et al., 2016).
The toxicokinetics of Ag 2 S NPs were different at the high and low concentrations tested (22 and 226 mg Ag kg −1 dry soil). While for the lower concentration of Ag 2 S NPs the estimated uptake rate constant (K 1 ) was higher than that for the other Ag forms, the inert fraction was similar leading to a higher elimination constant rate (K 2 ). The K 1 for the higher concentration of Ag 2 S NPs used was lower than that for the lower Ag 2 S NPs concentration, but similar to what was observed for the other Ag forms. Hence, Ag 2 S NPs could be taken up by mealworms even though they were expected not to be bioaccessible compared to the other Ag forms as they had a very low CaCl 2 -extractable Ag concentration (0.001 mg Ag kg −1 soil) ( Table 2). This observed low bioaccessibility for Ag 2 S NPs revealed that the uptake of AgNPs is more related to the surface speciation than to their release of Ag ions, so the sulphadised form was taken up but was also eliminated more rapidly resulting in low overall internalization. Moreover, this was confirmed by the fact that Ag 2 S NPs could be taken up by mealworms, but they did eliminate them faster than the other Ag forms with the lowest Fi at 226 mg Ag kg −1 dry soil (0.08, corresponding to 8% of the amount accumulated). The elimination rate constants in the mealworm were similar for all different Ag forms except for the 50 nm AgNPs; such a K 2 similarity for different Ag forms has been reported before for the isopod Porcellionides pruinosus (Tourinho et al., 2016). However, mealworms can eliminate silver more efficiently than isopods, which are able to accumulate silver as granules in the S-cells of the hepatopancreas (Tourinho et al., 2016). No significant difference between uptake and elimination rate constants of ionic Ag and 50 nm AgNPs was seen in the earthworm Eisenia fetida in the study of Baccaro et al. (2018). Moreover, Baccaro et al. (2018) reported a remarkably lower uptake of Ag from Ag 2 S NPs in earthworms compared to ionic Ag, which is in contrast to what was found for the mealworms. In addition, a low bioavailability of Ag 2 S NPs was also observed in the soil arthropod F. candida (Collembola) and no uptake in the isopods Porcellio scaber (Talaber et al., 2020). Nevertheless, in the study by Kampe et al. (2018) Ag 2 S NPs showed to be bioavailable for P. scaber exposed to AgNP spiked sludge amended soil. To compare the bioaccumulation of pristine and aged silver NPs, Velicogna et al. (2017) exposed Eisenia andrei to AgNP and AgNO 3 spiked soil and to soil amended with spiked biosolid-AgNPs. The bioaccumulation factors were 0.89 and 0.74 for AgNPs and ionic Ag, respectively and 0.12 for biosolid-AgNP amended soil (Velicogna et al., 2017), which is in agreement with our study. Vijver et al. (2003) reported that feeding on soil organic matter rather than soil pore water is the main exposure route for metals in mealworms (Vijver et al., 2003). Our results, however, show that the soil pore water was slightly more important than soil for the total uptake in the mealworms, for the Ag forms that showed significant dissolution resulting in high concentrations of ionic Ag in the soil pore water. In this regard, AgNO 3 and the 60 nm AgNPs, with 60-62% of Ag uptake from the soil pore water, showed nearly 90% of the total Ag was present as ionic Ag in the soil pore water. The 50 nm AgNPs displayed an equal uptake of Ag from the soil and soil pore water with about 50% of ionic Ag in soil pore water. The 3-8 nm AgNPs and Ag 2 S NPs were taken up more from the soil with 43% and 10% Ag uptake from the soil pore water and with 17% and 0.6% of ionic Ag in the soil pore water. Therefore, mealworms were indeed exposed to the soil and soil pore water and what made them so tolerant to Ag during these life stages could be their ability to eliminate Ag by different mechanisms such as shedding the exuvia (although no exuvia was collected in the soil exposure) or through the faeces.
The results suggest that there is a NPs stability phenomenon driving the uptake and elimination phases, as the insoluble Ag 2 S are in all cases significantly different with the other treatments. This is also supported by the faster kinetics observed for the Ag 2 S NPs compared to the rest of the treatment that presented significantly higher dissolution (see Section 3.1 and Fig. S1E), while the differences observed between the Ag 2 S treatments can be attributed to the substantially higher variation observed in the raw data for the Ag 2 S (226 mg Ag kg −1 dry soil) treatment. Furthermore, the 50 nm AgNPs uptake results are also partly supported by the kinetics results, as they were found to be significantly different from the 60 nm AgNPs and AgNO 3 , which present the higher ionic content, but not with the both the Ag 2 S and 3-8 nm AgNPs treatments.

Food exposure
The toxicokinetics parameters K 1 and K 2 for 3-8 nm and 60 nm AgNPs were similar, while the K 1 and K 2 values for mealworms fed on food spiked with 50 nm AgNPs and AgNO 3 were also similar. The K 1 and K 2 values for the Ag 2 S NPs were seemed to be similar to those of the 3-8 nm and 60 nm AgNPs and gave the lowest Fi (0.01). However, the statistical analysis of the mealworms body concentration showed that Ag taken up from the Ag 2 S NPs was eliminated significantly different compared to other Ag forms. The latter suggests that the Ag 2 S NPs could be eliminated completely and may be differently handled by the mealworms compared to the other silver forms. Metals have a high tendency to adsorb to organic matter and their desorption during gut passage depends strongly on the binding affinity to the organic matter (Tourinho et al., 2016), and also to the gut conditions such as pH or enzymes present. Ag 2 S NPs may not or only slightly bind to the food as they are stable. As a consequence, they may have detached from the food and be absorbed faster during gut passage but also eliminated faster compared to pristine AgNPs. Truzzi et al. (2019) reported that Cd, Ni, and As were taken up by T. molitor larvae feeding on metal-contaminated substrates but not accumulated (BAF close to 1), while Hg was bioaccumulated.
These results of the statistical comparison of the raw experimental data seem to be consistent with the long-term dissolution results (48 h timepoint where the dissolution curves start reaching the expected plateau) presented in Fig. S1E in the Supplementary Information and further support the observations during the soil exposure experiments presented earlier, where a NPs-stability effect was observed. The 60 nm AgNPs present, in absolute numbers, higher dissolution (29.2 ± 0.33%) and being closer to the ionic AgNO 3 , followed by the 50 nm AgNPs (26.1 ± 6.02%), the 3-8 nm AgNPs dissolution (48 h dissolution: 21.1 ± 4.04%) being closer to the mean (19.1 ± 1.4%) of all the NP treatments and the more stable Ag 2 S NPs presenting the lowest dissolution (0.01 ± 0.003%), as expected due to it being in practice insoluble (Sekine et al., 2015). Further study is needed to verify these results, by extending the dissolution monitoring period to the entire 42-day uptake and elimination periods.
Coleoptera have a short straight and tubular foregut, while the midgut is the main site of digestion. The hindgut has a permeable cuticle to prevent loss of useful substances and starts with the ileum and then continues into the colon, rectum and anus (Sarwade and Bhawane, 2013). In T. molitor, the lifetime of the midgut epithelium cells is about 4 days after which they are discharged into the gut lumen and excreted through the faeces (Lindqvist and Block, 1995). If any metal taken up could not be eliminated by excretion, it may be stored by the organism to prevent further toxic effects. Pedersen et al. (2007) reported the presence of Cd-binding proteins without high cysteine contents in mealworms, although the mechanism of detoxification of metals and the role of this protein is not clear yet. Bednarska and Świątek (2016) separated the cellular components of mealworms into three fractions (S1 or cytosolic fraction including organelles, heat-sensitive and heat-stable proteins, S2 or cellular debris and the G or metal-rich granule fraction) to estimate the risk of food chain transfer of stored metals. About 30% of Cd and Zn was found to be in the S1 fraction, which has the possibility to be transferred to higher trophic levels in a food web. For predators with a low gut pH, the metals stored in granules in the mealworms could also become bioavailable (Bednarska and Świątek, 2016).
The results presented also highlighted an extra route of excretion of Ag by moulting, although it seems that the amount of Ag eliminated in exuviae depends on the Ag form/speciation and its behaviour inside the organism. Lindqvist and Block (1995) found that the main amount of discharged Cd during the moulting of mealworms was from the gut epithelium. Metals present in other tissues could not be eliminated by moulting, except for the part bound to the integument (Lindqvist and Block, 1995). Because of the low concentration of cadmium and zinc in mealworm exuviae, Vijver et al. (2003) reported that moulting may not be an important elimination route for these metals. Discharge through moulting was reported also for CeO 2 NP elimination from Daphnia pulex (Auffan et al., 2013), and for the elimination of zinc, cadmium and copper from chironomids (Diptera, Chironomidae) (Timmermans and Walker, 1989).
Among the different Ag forms, Ag 2 S NPs had the highest concentration in the mealworm exuviae, suggesting they were mainly attached to the gut epithelium and did not reach other tissues, which was confirmed by the low Ag body concentrations found. But some caution has to be taken considering this result due to the low number of exuvia collected.
The estimated kinetics in the mealworms exposed via the soil and via the food showed a lower inert fraction upon soil exposure for all different Ag forms except for the 50 nm AgNPs. Maybe because of the different type and content of organic matter in the soil compared to the food, the ingestion time and uptake mechanisms were different resulting in different Ag storage levels in the mealworms. Tourinho et al. (2016), exposing isopods to the same 3-8 nm AgNPs via soil and via food, found higher uptake rate constants from soil exposure and similar elimination rate constants for both exposures. This is in contrast with our results, probably because isopods avoided contaminated food or because of the higher ability of mealworms to eliminate Ag when compared to isopods.

Conclusions
The larvae of Tenebrio molitor present several routes of Ag uptake, including food and soil. In the latter case, the water fraction is most important (i.e. soil pore water). Ingested soil particles are excreted through feaces, not contributing much to Ag accumulation.
The bioavailability and bioaccumulation of AgNPs are affected by their characteristics (surface coating, speciation) and related stability in the receiving environment, as Ag uptake and elimination rate constants differed for differently coated AgNPs of similar size. Ag 2 S NPs, which are the most environmentally relevant and the most stable AgNP form, could be taken up by the mealworms but were also eliminated faster than the other Ag forms for both exposure routes. Therefore, bioaccumulation of different NPs, including aged forms, should be considered in the risk assessment of NPs.

Declaration of competing interest
There are no conflicts to declare. Z. Khodaparast, C.A.M. van Gestel, A.G. Papadiamantis et al. Science of the Total Environment 777 (2021) 146071