Acute and Sublethal Effects of Deltamethrin Discharges from the Aquaculture Industry on Northern Shrimp (Pandalus borealis Krøyer, 1838): Dispersal Modeling and Field Investigations

Pharmaceutical deltamethrin (Alpha Max), used as delousing treatments in aquaculture, has raised concerns due to possible negative impacts on the marine environment. A novel approach combining different scientific disciplines has addressed this topic. Acute (mortality) and sublethal effects (i.e., fitness, neurological, immunological, and oxidative responses) of exposure of northern shrimp (Pandalus borealis) were studied in laboratory experiments. Passive water sampling combined with sediment analyses revealed environmental concentrations. Finally, dispersal modeling was performed to predict environmental concentrations. Ecotoxicological analyses showed mortality in shrimp after 1 h of exposure to 2 ng L–1 (1000-fold dilution of treatment dose), revealing a high sensitivity to deltamethrin. Sublethal effects included induction of acetylcholinesterase and acyl CoA oxidase activities and oxidative impairment, which may be linked to neurotoxic responses. Field concentrations of 10–200 ng L–1 in water (100 m from the pens) and <LOD-0.19 ng g–1 dw in sediment (0–400 m from pens) were measured. Ecotoxicological values were compared with measured and modeled concentrations. They showed that concentrations higher than those causing mortality could be expected up to 4–5 km from point of release, in an area of 6.4 km2, with lethal concentrations remaining up to 35 h in some areas. Hence, the study demonstrates that there is a considerable risk for negative effects on the ecologically and commercially important shrimp.


Materials and methods, additional information
1.1 Additional information on the investigation of the effects in P. borealis of exposure to short (1 h) pulses of diluted deltamethrin in the Alpha Max ® formulation.

Collection of shrimps
Northern shrimps were collected with shrimp pots in fjords in northern Norway in October 2019. Fjords with no aquaculture activities were chosen for sampling. Shrimps were stored in tanks with constant flow of fresh sea water while onboard the ship. Collected individuals were visually inspected and transported to Akvaplan-niva's marine research station at Kvaløya, Tromsø, Norway. Shrimps were kept in tanks (600 L) supplied with a continuous flow (500 L/h) of filtered (60µm filter) and UVtreated bottom seawater (63 m depth) from the fjord Sandøysundet, adjacent to the research facility. All tanks were covered with a lightproof cover during acclimation and rearing. Temperature and oxygen saturation were monitored daily resulting in temperatures ranging from 2.0 to 5.1 C, a salinity of 34‰ and an O 2 -saturation >10 mg/L. Shrimp tanks were daily checked for dead individuals. The animals were acclimated for a minimum of 2 weeks before experimental start.

Exposure scenarios, experimental setup and stock preparations
Water quality parameters (oxygen saturation, temperature and salinity) were monitored in the beginning and end of exposure, and during the post-exposure period using a hand-held and portable Oxyguard®. As the salinity in the intake water from the fjord is stable at 34 ‰ throughout the year, salinity was not monitored. During the exposure pulse, the header solution was pumped through Teflon tubing into the seawater inlet to each exposure aquarium by a peristaltic pump with a multi channeling system (model 520, Watson and Marlow, Cornwall, UK). At the start of each exposure, the water flow was stopped, and 5 L water removed from each exposure tank, and replaced with 5L bath chemical in seawater at a concentration 9 times higher than the wanted exposure concentration (i.e. diluted 9 times in the tank) to ensure the correct exposure concentration from the beginning. The experimental set-up is based on the system used by Frantzen et al. (2020). The water flow was then restarted, and, at the same time, two peristaltic multi-channel pumps were started, providing stock solution to each tank to ensure a constant concentration of the treatment chemicals throughout the exposure period with a flow of 3 mL/min. After 1 hour the peristaltic pumps were stopped, and the stock solution tubes removed from the tanks (i.e. start of recovery). Control tanks were subjected to the same procedure.
Three replicate tanks with 5 shrimp in each tank for each treatment were used, including control. All control tanks were handled the same way as the exposure tanks (i.e. pumped with clean water). Due to the relatively long exposure period, oxygen and temperature levels were also monitored in the middle of the experiment.
Stock solutions of deltamethrin were prepared in 2 L Schott bottles using the commercial formulations Alpha Max® and distilled water ( Figure S1). The stock solutions were placed on magnetic stirrers. The stock solutions were used to achieve exposure concentration in the 2L headers, which contained 0.5ng/ml, 0.1ng/ml, 0.02 ng/mL of deltamethrin for the 1000 (high treatment), 5000 (middle treatment) 25000 (low treatment) dilutions respectively.

Acetylcholinesterase activity (AChE)
Samples of gills and muscles were homogenized in 0.1 M trizma base, pH 7.2 containing sucrose 0.25 M and centrifuged at 10 000 g for 10 minutes. Obtained supernatants were spectrophotometrically assayed by the Ellman's reaction at 18 ± 1 °C, λ = 412 nm, ε = 13.6 mM cm −1 using acetylthiocholine and 5,5-dithiobis-2-nitrobenzoic acid (DTNB) as substrates. This colorimetric procedure is based on the reaction of thiocholine (one of the products of enzymatic hydrolysis of acetylthiocholine) with 5,5dithiobis-2-nitrobenzoic acid (DTNB, Ellman's reagent) forming a yellow product (5-mercapto-2nitrobenzoic acid and its dissociated forms). The reaction medium was 0.1 M Trizma base (pH 7.4), DTNB 0.2 mM and acetylthiocholine 0.5 mM. After preincubation of DTNB and sample for 5 and 10 minutes for gills and muscles respectively, reactions were started by adding acetylthiocholine and readings were carried out for 1 minute at 412 nm (Ellman et al., 1961).

Activity of Acyl CoA oxidase (ACOX)
Samples of digestive gland were homogenized in 1 mM sodium bicarbonate buffer (pH 7.6) containing 1 mM EDTA, 0.1 % ethanol, 0.01 % Triton X-100 and centrifuged at 500 g for 15 min at 4 °C. The H 2 O 2 production was measured in a coupled assay by following the oxidation of dichlorofluorescein-diacetate (DCF-DA) catalyzed by an exogenous horseradish peroxidase (HRP). The reaction medium was 0.5 M potassium phosphate buffer (pH 7.4), 2.2 mM DCF-DA, 40 μM sodium azide, 0.01 % Triton X-100, 1.2 U mL −1 HRP in a final volume of 1 mL. After a preincubation at 25 °C for 5 min in the dark with an appropriate volume of sample, reactions were started adding the substrates palmitoyl-CoA at final concentrations of 30 μM and 100 μM for Acyl-CoA oxidase (ACOX) and readings were carried out against a blank without the substrates at 502 nm (Small et al., 1985).

Antioxidant responses and oxidative damage
The antioxidant response and oxidative damage were analysed by measuring the total antioxidant capacity (TOSC assay towards peroxyl and hydroxyl radicals) and lipid peroxidation (malondialdehyde levels). For the total oxyradical scavenging capacity (TOSC) analysis, digestive glands were homogenized in 100 mM of potassium phosphate buffer (pH 7.5) containing NaCl (2.5 %), 0.1 mg mL −1 bacitracin and 0.008 TIU mL −1 aprotinin, 0.1 mg mL −1 leupeptin, 0.5 mg mL −1 pepstatin as protease inhibitors. After centrifuging at 100,000g for 70 min at 4°C, supernatants were collected and used for the analyses. The total oxyradical scavenging capacity (TOSC) assay measured the overall capability of cellular antioxidants to absorb different forms of artificially generated oxyradicals, thus inhibiting the oxidation of 0.2 mM α-keto-γ-methiolbutyric acid (KMBA) to ethylene gas (Regoli and Winston, 1999). Peroxyl radicals (ROO•) were generated by the thermal homolysis of 20 mM 2,2'-azo-bis-(2-methylpropionamidine)-dihydrochloride (ABAP) in 100 mM K-phosphate buffer, pH 7.4. Hydroxyl radicals (•OH) were produced by the Fenton reaction of iron-EDTA (1.8 mM Fe3+, 3.6 mM EDTA) and ascorbate (180 mM) in 100 mM potassium-phosphate buffer. Under these conditions, the different oxyradicals produced quantitatively similar yields of ethylene in control reactions, thus allowing the comparison of the relative efficiency of cellular antioxidants toward a quantitatively similar radical flux. Ethylene formation reactions were analysed at 12 min time intervals (total time: 96 min) by gas chromatographic separation and analyses by a detector (Agilent Technologies, Santa Clara, CA, USA) and the TOSC values were quantified by equation 1: where ʃSA and ʃCA are the integrated areas from the kinetic curves for samples (SA) and control (CA) reactions. For all samples, a specific TOSC (normalized to content of protein) was calculated by dividing the experimental TOSC values by the relative protein concentration contained in the assay.
Protein concentrations were measured according to the Lowry method, using bovine serum albumin (BSA) as standard (Lowry et al., 1951).
1.2 Field sampling of sediment and water 1.2.1. Evaluation and validation of passive sampling technique and partitioning coefficient experiments K pw was measured for deltamethrin and cypermethrin using the co-solvent method (Booij et al., 2017;Pintado-Herrera et al., 2016;Smedes et al., 2009). As additional experiments, partitioning coefficients were also established for cypermethrin and the in-feed pharmaceuticals diflubenzuron and teflubenzuron. These measurements were undertaken for two types of passive samplers (SSP and Altesil™) and 8 water samples (n=16) simultaneously by equilibrating water/methanol solutions with deltamethrin and cypermethrin-spiked silicone rubber in laboratory-based batch experiments. The cosolvent methodology was first used to load the chemicals into the two types of silicone rubber (Booij et al., 2002) in a similar way to the spiking with PRCs (performance reference compounds). The compounds were added to methanol in a small vial together with the silicone rubber sheets cut to an appropriate size. The vial was shaken at 150 rpm on an orbital shaker and ultrapure water was added gradually over time to reach a 50:50 proportion of water and methanol over a one-week period. Silicone rubber sheets were then collected, dried with a lint-free tissue and place in a clean jar at -20 C until use. All glassware was cleaned by heating in a furnace at 550 C. Replicate assays with no methanol were conducted in 2 L-glass jars with just under 2 L of ultrapure water in contact with 50 mg of each of the silicone rubber. Experiments with a proportion of methanol (mol/mol) of 4 % (1 L solution and 100 mg silicone rubber), 9 % (1 L solution and 100 mg silicone rubber), 19 % (250 mL solution and 100 mg silicone rubber) and 45 % (250 mL solution and 2.2 g silicone rubber) were prepared. For the experiments with 0 and 4 % methanol, the glass wall of the jar was spiked with the two chemicals in order to shorten the time to equilibrium of these two assays. In short, the compounds were spiked in a small amount of methanol:water (20:80) that was subsequently added to the glass jars, the solution was rolled in the glass jar for some minutes. The solution was removed, and the bottle was rinsed twice with small amounts of ultrapure water to ensure no methanol was left in the 0 % methanol assays. The glass jars were then wrapped in foil and placed on an orbital shaker for 3 months at 150 rpm (slightly longer than planned due to covid-19 lock-down). Upon termination of the experiments, the solution (different volumes for different experiments) was placed in a decanter. In cases where the solutions contained significant amounts of methanol, water was added to dilute the methanol and ensure high recovery for the extraction of the chemicals. These samplers followed the same analytical protocol as the samplers exposed in field. Analyses and quantification were done as described in section 2.2.3.

Sediment samples:
The sediment samples were clean-up and analyzed based on the method in Tucca et al (2017). Briefly, the samples were freeze dried and a subsample of 2-3 g was collected. D6cyfluthrin was added as internal standard and the samples were extracted twice (ultrasonic bath) with 20 mL of dichloromethane and evaporated to 0.5 ml. Further clean-up was conducted with SPE-Florsil eluted with 20 % diethyl ether in iso-hexane followed by PSA and filtration through Costar spinex 0.2µm nylon filter.
All field samples were separated by an Agilent 7890B gas chromatograph (GC). This GC was equipped with two 15 m x 0.25 mm HP-5MS-UI columns with 0.25 µm film thickness and connected to a 7010B Triple Quadrupole mass spectrometer ((MS/MS) for detection and quantification of deltamethrin. One µl of sample extract was injected in splitless mode (injection temperature: 280°C). The oven temperature was 60°C and held for 1-minute, increased 40°C/min to 170°C followed by an increase of 10°C/min to 310°C and held for 4 min. PAS were also analyzed for the reference PRCs by using GC The clean-up procedure for diflu-and teflubenzuron for K pw experiments followed the same procedure. More information about the analytical procedure can be found in Langford et al. (2014).

Quality control
One PAS field blank was exposed to air during the deployment. Laboratory blanks were analysed together with both water and sediment samples and all blanks were <LOD. Two pre-spiked PAS (200 ng, i.e., in line with amount deltamethrin measured in the samples) were analysed parallel with the sediment samples from site 1 (133 % and 137 % recovery, respectively). The general, additive analytical uncertainty in the PAS analyses were estimated to be 40-60 %, considering analyses of both PRCs and deltamethrin. Three subsamples of the sediment samples, from site 2 were additionally spiked with 0.5, 2.5 and 50 ng/g, respectively and analyzed in parallel to make sure the analytical methods were working also for the quantification of sediment.

Calculation of PAS deltamethrin concentration
The uptake of deltamethrin in PAS are based on Rusina et al. (2010). In equation 1, where R S is the insitu sampling rate (L/d) and is estimated based on dissipation of spiked PRCs by the model from Rusina et al. (2010). β is a parameter that depends on e.g. temperature and flow-rate and was estimated based on R S calculated from dissipation of the PRCs by plotting the retained PRC-fraction as a function of K sw using non-linear least squares estimation (Booij and Smedes, 2010).
The sampler water partition coefficient (K sw ) used in this study (6.2) to calculate C W were obtained from experiments as described further up. K sw is in general similar to K ow , which is 5.9 for deltamethrin. Water concentrations are calculated according to equation 2. t is exposure time (d). The sampling rate estimated from the suite of PRCs (acenaphthene-d 10 , fluorene-d 10 , Phenanthrene-d 10 , fluoranthene-d 10 , chrysene-d 12 , and benzo(a)pyrene-d 12 ) ranged between 1.8-5.9 L/d (log = 0.93 L 1.08 kg 0.08 d -1 ; log se = 0.05) for an AlteSil silicone rubber with a mass of 16.5 g and surface area of 500 cm 2 deployed for 6 days.
Deltamethrin was distributed into the water after the PAS were deployed. The release of PRCs therefore occurred during six days, while the uptake of deltamethrin only occurred after the salmon delousing treatment began. The delousing took place over a few days, and the total exposure (days) of deltamethrin to the PAS may therefore vary between a few hours to a theoretical maximum of time between beginning of delousing and time for collection of samplers (3.5 days).

Oceanographic modelling
During bath treatment, the cages are normally raised to about 10 m depth and covered by a tarpaulin before the de-lousing agent is added. After treatment the tarpaulin is removed, and the treatment water is released into the ambient water. To simulate this release in the model all the cells within a circle of the same horizontal size as the cage, and down to about 10 m, are initialized with concentrations equal to the treatment dose. The resulting plume of deltamethrin will then be transported and diluted in the model.  Figure S2C. To simulate a realistic delousing operation, the treatment procedure performed at the farm is followed as closely as possible. Hence, in the simulations, we have picked 7 of the 10 cages at the model site and released deltamethrin sequentially from each of these 7 cages at 12-hour intervals. The entire simulated operation then spans 3 days, and the simulation is run for 4 more days to ensure that concentrations of deltamethrin are negligible. To study the development of the discharge in time, the model output is saved at 10-minute intervals.
The model input builds mainly on oceanographical and hydrological data in the relevant area, and the input regarding deltamethrin is the amount used for the delousing at the site and timing for release and treatment of each cage. The hydrodynamic model FVCOM (Chen et al, 2003) has the possibility to adapt the grid size locally, which makes it ideal for modelling the currents along a complex coastline such as in Norway (ex. In Børve et al., 2021 and. Additionally, in this study we have exploited this flexibility of the model grid to increase resolution around the fish farm in order to study the detailed spreading pattern and dilution of the delousing chemical. The hydrodynamical model is coupled to a tracer module in FABM (Bruggemann and Bolding 2014), and the coupled FVCOM-FABM system used in this study is a state-of-the-art modelling tool that has been specifically developed and applied in several studies concerning the spreading of delousing chemicals (e,g. Refseth et al., 2016, 2019, Refseth and Nøst 2018.

Statistical analyses
Levels of effect on shrimp observed after exposure of deltamethrin were compared statistically to levels of effect observed to untreated control handled in the same manner. Level of significance was set at p < 0.05. Mortality and behavior data analyses were performed in PAST v 17 (Hammer, 2001), using the Wilcoxon Rank sum test. Analysis of variance (ANOVA) was applied for all the other investigated shrimp parameters to test differences among experimental concentrations after experimental and recovery periods. Homogeneity of variance was checked by Cochram C and posthoc comparison (Newman Keuls) was used to discriminate between means of values. Analysis of variance (2-way ANOVA) was applied to test differences among experimental conditions, experimental periods and interactions "exposure concentrations × periods" (level of significance at p < 0.05). All statistical analyses were performed using Rstudio (version 0.99.491).

Figure S6. Total Oxyradical Scavenging Capacity (TOSC) toward peroxyl (ROO•) (A) and hydroxyl (•OH) radicals (B) and malondialdehyde levels in digestive gland (C).
Lowercase letters indicate significant differences between groups of means at the end of exposure time; capital letters indicate significant differences between groups. Data are given as mean values ± standard deviations, n =15. Ns= not significant variations; n.a.= not analysed.