Temperate Versus Arctic: Unraveling the Effects of Temperature on Oil Toxicity in Gammarids

Shipping activities are increasing with sea ice receding in the Arctic, leading to higher risks of accidents and oil spills. Because Arctic toxicity data are limited, oil spill risk assessments for the Arctic are challenging to conduct. In the present study, we tested if acute oil toxicity metrics obtained at temperate conditions reflect those at Arctic conditions. The effects of temperature (4 °C, 12 °C, and 20 °C) on the median lethal concentration (LC50) and the critical body residue (CBR) of the temperate invertebrate Gammarus locusta exposed to water accommodated fractions of a fuel oil were determined. Both toxicity metrics decreased with increasing temperature. In addition, data for the temperate G. locusta were compared to data obtained for Arctic Gammarus species at 4 °C. The LC50 for the Arctic Gammarus sp. was a factor of 3 higher than that for the temperate G. locusta at 4 °C, but its CBR was similar, although both the exposure time and concentration were extended to reach lethality. Probably, this was a result of the larger size and higher weight and total lipid content of Arctic gammarids compared to the temperate gammarids. Taken together, the present data support the use of temperate acute oil toxicity data as a basis for assessing risks in the Arctic region, provided that the effects of temperature on oil fate and functional traits (e.g., body size and lipid content) of test species are considered. As such, using the CBR as a toxicity metric is beneficial because it is independent of functional traits, despite its temperature dependency. To the best of our knowledge, the present study is the first to report CBRs for oil. Environ Toxicol Chem 2024;43:1627–1637. © 2024 The Authors. Environmental Toxicology and Chemistry published by Wiley Periodicals LLC on behalf of SETAC.


INTRODUCTION
During the past decades, the Arctic has been warming up at a rate 4 times faster than the globe as a whole (Rantanen et al., 2022).With sea ice receding, Arctic regions are opening up for shipping activities, such as transport, fisheries, and tourism.As a result, a 1.3 times overall increase in shipping intensity by fishing and tourist vessels was observed in the Arctic region of Svalbard between 2012 and 2019, coupled with a stretched operational shipping season and an expansion of navigational areas (Stocker et al., 2020).Enhanced shipping activities increase the risks of shipping accidents, particularly in Arctic regions, which are challenging areas with low temperatures; the likely presence of sea ice; bad visibility due to storms, fog, and winter darkness; and often still unexplored shores.Shipping accidents can result in oil spills, of which the consequences are expected to be worse than in temperate areas because of slower oil spill responses due to the remoteness of the region and reduced radio coverage (Stocker et al., 2020).Furthermore, not all oil spill mitigation measures are suitable or feasible under Arctic conditions (Wilkinson et al., 2017), which further increases the likelihood of a prolonged environmental presence of oil after a spill.Therefore, it is crucial to assess the effects of oil spills in Arctic marine ecosystems.
Toxicity of oil to pelagic organisms, such as fish and invertebrates, is driven by the fraction of oil that dissolves in seawater.The mass and composition of the dissolved oil fraction depend on the solubility of the constituents present in the oil, the degree of weathering, as well as environmental conditions, including temperature (Brown et al., 2016;Payne et al., 1991;Perkins et al., 2005).At lower temperatures, the solubility of petroleum substances decreases, oil becomes more viscous, and low-molecular weight (LMW) oil constituents will evaporate less (Brown et al., 2016).Overall, at low temperatures, relatively more LMW hydrocarbons than higher-molecular weight hydrocarbons will dissolve.Because LMW constituents in particular are easily taken up and consequently cause toxicological effects in aquatic organisms (Brils et al., 2002;Jonker et al., 2006;Redman et al., 2017), temperature is expected to affect oil toxicity.
Exposure to oil has been shown to lead to several types of acute and chronic effects in marine organisms, including narcosis and developmental effects (see Gissi et al., 2021;De Laender et al., 2011;Vrabie et al., 2009).However, oil toxicity data for the Arctic and other cold-environment species are limited, creating a challenge in assessing the risks of oil spills for the polar regions (Chapman & Riddle, 2005;de Hoop et al., 2011;Olsen et al., 2011Olsen et al., , 2013)).Toxicity data for water accommodated fractions (WAFs; i.e., the oil fractions dissolved in water) of (crude) oil have been reported for Arctic copepods, mysids, and fish only (Bender et al., 2021;Gardiner et al., 2013;Hansen et al., 2011;Riebel & Percy, 1990).Some additional data exist for arthropods, mollusks, and other fish; but these are from the 1970s to 1990s, and access to them is limited (de Hoop et al., 2011, and references therein).
To fill this oil toxicity data gap, the accuracy of extrapolating toxicity data from temperate to Arctic conditions has been explored in modeling studies.These studies suggest that temperate data can be used as a surrogate or a first approximation of Arctic toxicity data because the variation in oil toxicity between temperate and polar organisms was considered negligible (within a factor of 3; Bejarano et al., 2017;de Hoop et al., 2011;Olsen et al., 2013).Still, because of the limited data availability, additional research was recommended to further support this extrapolation approach, also taking into account the specific Arctic species characteristics (Camus et al., 2014).
In the present study, oil toxicity to temperate and Arctic gammarid species was further investigated.First, the effects of temperature (4 °C, 12 °C, 20 °C) on WAF composition and the acute toxicity of oil, in terms of the median lethal concentration (LC50) and the critical body residue (CBR), were studied for the temperate gammarid Gammarus locusta.The CBR provides the internal exposure concentration at which organisms are lethally affected.It was included because it is believed to be a more robust toxicity metric (van der Heijden et al., 2015).Second, LC50 and CBR values were determined for the Arctic gammarid Gammarus sp. at 4 °C.This allowed a direct comparison between the sensitivity of temperate gammarids and that of Arctic gammarids at their natural ambient temperature.

Test species
For the experiments conducted at temperate conditions, the amphipod G. locusta was used.This species lives on sediment and macrophytes, occurs on west European coasts (Costa et al., 2004), and is frequently used in eco(toxico)logical research (see Szczybelski et al., 2018).Organisms were collected in the intertidal area at two sites in the Oosterschelde, The Netherlands (Goese Sas [51°32′43.36″N,3°55′28.79″E]and Kattendijke [51°31′45.71″N,3°57′57.07″E]), between March and June 2016.Individuals with a length of 6 to 15 mm (from head to end of tail) were selected for the experiments.
For the Arctic experiments, Gammarus sp. were collected in June 2016 and July 2017 in the intertidal area of Ny-Ålesund harbor, Kongsfjorden, northwest Svalbard (78°55′39.5″N,11°56′17.5″E).A few individuals were examined and identified as G. setosus, an arctoboreal gammarid with a circumpolar distribution (Opalinski & Węsławski, 1989).This species has a life span of over 3 years and is believed to mature in its third year (Węsławski & Legeżyńska, 2002).However, a mixture of G. setosus and G. oceanicus has previously been observed in the Kongsfjorden, and the presence of G. oceanicus is increasing (Węsławski et al., 2018).Therefore, it cannot be excluded that G. oceanicus individuals were also present in the batch of test organisms.Gammarus oceanicus is a boreal gammarid species with a North Atlantic distribution and a life span of 3 to 4 years (Opalinski & Węsławski, 1989;Węsławski et al., 2018).To acknowledge this uncertainty in test species, the test species name for the toxicity tests was set to "Gammarus sp."Because the test organisms were collected in the harbor with the potential of site pollution, an additional control sample was collected outside of the harbor for chemical analysis.All test organisms were 16 to 20 mm in length.
To obtain further information on the functional traits of the test species (size, weight, and lipid content), additional batches of temperate and Arctic gammarids were collected from the field.After collection, these gammarids were rinsed with demineralized water, blotted dry, weighed, pooled per size class (1-7, 8-15, 16-20, 21-25, >26 mm), and stored at −20 °C until further processing.Frozen pooled gammarid samples were freeze-dried and homogenized using a mortar and pestle.Total lipids in these samples were extracted with dichloromethane-methanol-water (8:4:3), following the protocol of Folch et al. (1957), with some modifications.

Test oil and WAF production
The oil used in the experiments was distillate marine grade A (DMA; Shell marine gas oil), a low-sulfuric (<1.5%) diesel petroleum, containing a blend of refined (distilled) oils and having a density of 0.87 kg/L at 20 °C.It has been previously used in other toxicity studies (Brils et al., 2002;Jonker et al., 2006;Szczybelski et al., 2018;Vrabie et al., 2009).Water accommodated fractions of DMA oil were prepared in 5-L glass Duran ® bottles, equipped with a glass-bottom tap.Standard WAF production using the variable loading approach was slightly adapted by minimizing headspace volume to reduce evaporation during WAF production.Each bottle was filled with 5200 mL of artificial seawater (ASW), and a glass stir bar was added.Artificial seawater was used to eliminate potentially interfering components that may be present in natural seawater and to standardize test conditions across experiments.To obtain the ASW, Pro Reef Salt Mix (Colombo ® ) was dissolved in Milli-Q water to a salinity of 34‰ to 36‰.A predetermined volume of DMA oil was carefully pipetted to the glass wall of the Duran bottle, just above the water surface (see Supporting Information, Table S1, for applied loadings).Each bottle received a specific predetermined oil loading, to obtain a range in WAF exposure concentrations.Then, the bottle was closed and gently stirred on a magnetic stirrer, creating a small vortex of 1 to 2 cm.Stirring was done in the dark for 48 h, which was considered sufficient for obtaining an equilibrium WAF of a light fuel oil like DMA at all test temperatures (Faksness et al., 2008;Mitusova et al., 2005).The stirring conditions created a stable system, consisting of two clearly separated layers without the formation of visible droplets or a dispersion.Each WAF was produced in a temperature-controlled room, at the same test temperature that was maintained during the LC50 and CBR experiments.

Toxicity experiments
The temperate organisms were exposed to the WAFs at 3 temperatures: 4 °C, 12 °C, and 20 °C.Experiments were conducted between March and June 2016, when the water temperature in the Oosterschelde ranged from 6 °C (March) to 17 °C (June).The experiments were conducted in the order 4 °C, 12 °C, and 20 °C, at temperatures similar (within a maximum of 5 °C) to the Oosterschelde water temperature.The testing organisms thus had been acclimatized to the test temperature in a natural, gradual way.After collection, prior to the start of the experiments, gammarids were kept in the climate room for 2 days in aerated seawater without feeding.The Arctic specimens were tested at 4 °C, with a recorded ambient water temperature of 6.9 °C in Ny-Ålesund harbor on July 20, 2017.
Gammarids were exposed in 1-L glass bottles to the WAFs prepared from different DMA oil loadings, as described above.The bottles were gently filled with 900 mL of the respective WAF, using the bottom taps of the WAF bottles.This was done at the desired temperature and at low light intensity to minimize any photodegradation or photoactivation of oil constituents.Eight preselected gammarid specimens were sieved from 50-mL beakers filled with ASW and added to the 1-L test bottles.The bottles were immediately closed with aluminumlined caps to minimize any evaporation of chemicals.They were randomly divided over the climate rooms, and survival was scored every 24 h during a period of 5 days (120 h), starting at t = 0.During the experiments, lids were not removed, to avoid evaporation.For each temperature, 5 to 7 WAF "concentrations" were tested at the same time, including a control that was prepared as described above but without the addition of oil.Each concentration was simultaneously tested in 5 (temperate species) or 3 (Arctic species) replicates.Experiments were run twice for each temperature.Experiments were performed in the climate rooms of Wageningen Marine Research in Yerseke, The Netherlands (temperate conditions), and KingsBay Marine Laboratory in Ny-Ålesund, Svalbard (Arctic conditions).To ensure comparability between the tests performed at the two locations, the experiments were conducted using the same approach, test protocols, oil, ASW, and, as much as possible, personnel.
Dissolved oxygen concentrations (milligrams per liter) in the WAF solutions were measured in the replicates before and after (120 h) the experiments, using a Hach © HQ-40d multimeter equipped with a luminescent dissolved oxygen probe.

CBR experiments
Based on the results of the toxicity experiments, 3 oil loadings were selected for use in the CBR experiments: a control loading (C-0); a medium loading, which had been demonstrated to lead to mortality within 48 h (C-medium); and a high loading (C-high; Supporting Information, Table S1).For the CBR experiments performed at temperate conditions, the C-high loading was 0.3 mL of DMA/L of water, whereas the Cmedium loadings differed per temperature: 0.1 mL/L (4 °C), 0.03 mL/L (12 °C), and 0.0065 mL/L (20 °C).For the Arctic CBR experiment, the oil loadings for C-medium and C-high were 1 and 2 mL/L, respectively.These higher loadings were required to reach gammarid mortality within the test duration.Each loading was tested in 5 (temperate species) or 3 (Arctic species) replicates.
Test bottles were filled with WAF and eight gammarids, as described above for the LC50 experiments.Directly after the start of the experiments, the gammarids were observed every 15 to 30 min (day and night), for 48 h in total.The CBR experiment with the Arctic gammarids was prolonged up to 72 h, to collect a sufficient number of dead organisms.Upon death, the individual was immediately removed from the test bottle, rinsed with demineralized water, and blotted dry with a clean tissue.The time of death, weight (milligrams), and length (millimeters) were recorded.All dead individuals removed from the experimental systems were pooled per replicate in 20-mL scintillation vials and frozen at −20 °C.Prior to chemical analysis, organisms were pooled per oil concentration per test and freeze-dried for 16 h.

Chemical analysis
Following the preparation of the 1-L exposure bottles, 500 mL of the remaining 700 mL of WAF in each of the 5-L Duran bottles was gently drained into an amber-colored 500-mL bottle, containing 50 mL of n-hexane (Biosolve).The bottles were firmly closed with aluminum-lined lids and stored in a dark room at or below room temperature for several weeks until chemical analysis.
The dissolved total petroleum hydrocarbon (TPH) concentrations in the WAFs were determined by liquid-liquid extraction with n-hexane.The bottles containing WAF and nhexane were shaken on a heavy-duty reciprocal shaker for 1 h, after which the hexane phase from each bottle was transferred to a calibrated pointed flask with a glass pipette.The WAF was extracted a second time with an additional 20 mL of hexane, and the extracts were pooled and concentrated to 0.5 mL with a modified Kuderna-Danish setup and a gentle flow of nitrogen gas, respectively.Then, 1.0 mL of n-heptane was added, after which the extract was concentrated under nitrogen to 0.4 mL.Finally, 100 µL of internal standard solution (500 mg/L of decane [C 10 ] and 300 mg/L of tetracontane [C 40 ] in n-heptane) was added, and the resulting extract was transferred to an autosampler vial.
Concentrations of TPH in the gammarids from the CBR experiment were determined as follows.Approximately 150 to 300 mg of freeze-dried organism material was Soxhletextracted with n-hexane/acetone (3:1) for 16 h in preextracted glass fiber extraction thimbles.The extracts were concentrated as described above to 1 mL and cleaned using basic/acidtreated silica gel columns with n-hexane to minimize the presence of biogenic material in the extracts, which would interfere during instrumental chemical analysis (Muijs & Jonker, 2009a).The hexane eluates were concentrated and solventexchanged to 0.4 mL of n-heptane, and 100 µL of internal standard was added as described above.For both WAF and biota extraction series, 4 blank and 4 recovery determinations were included.Total petroleum hydrocarbon concentrations in the extracts were quantified by gas chromatography-flame ionization detection, as described before (Muijs & Jonker, 2009a), using nine calibration standards (measured in fourfold) for each analysis series (see Supporting Information, Text S1, for further details).Because biota extracts were cleaned using silica gel, any polar metabolites may not be included in the CBR values.However, the contribution of polar metabolites to the overall CBR is considered negligible because of the relatively limited biotransformation capacity of amphipods (Camacho-Jiménez et al., 2023) and the fact that any biotransformation will diminish quickly upon exposure as a result of narcosis (inhibiting enzyme activity).
Final TPH concentrations in WAFs and gammarids were blank-and recovery-corrected and expressed as micrograms of TPH (i.e., C 10 -C 40 ) per liter of water and milligrams of TPH per kilogram dry or lipid weight in organisms, respectively.Blank (control) correction was done by subtracting the averaged TPH concentrations in the controls from the exposed samples (WAF and gammarids) to correct for any remaining background signal.
To obtain information on chemical profiles (hydrocarbon fraction distributions), final concentrations in WAFs and gammarids were expressed in terms of hydrocarbon blocks (HBs; i.e., boiling point fractions): C 10 to C 12 , C 12 to C 14 , C 14 to C 16 , C 16 to C 18 , C 18 to C 22 , C 22 to C 28 , and C 28 to C 40 .Distillate marine grade A oil is a light diesel oil, and a C 28 to C 40 fraction is therefore not expected to be present.The fact that very small C 28 to C 40 fractions were nevertheless observed in the pure DMA oil, the gammarids, and the WAFs (Supporting Information, Table S8) is the result of the integration approach (i.e., drawing a horizontal baseline from C 10 -C 40 ), which cannot fully exclude the tailing effect of the previous fractions.
Total lipid contents of the gammarids from the CBR experiment were quantified according to Bligh and Dyer (1959).
Samples of approximately 200 mg of freeze-dried material were extracted 3 times with a water/methanol/chloroform mixture.The chloroform phases were pooled and evaporated to dryness, after which lipid weights were determined gravimetrically.

Data analysis
Median lethal loading (LL50; cf.Wheeler et al., 2020) values with 95% confidence intervals (CIs) were determined by fitting a dose-response model with variable slope (Prism, Ver.9.5.0) to the experimental data, using system loading (milliliters of oil per liter of water) as the exposure metric and percentage of survival (0%-100%) as the response (Supporting Information, Text S2 and Table S4a).For these calculations, the results from all tests performed at a specific temperature with the same species were combined.Especially for the higher oil loadings (>0.0025 mL/L) at 20 °C, oxygen demand in the test systems was relatively high, resulting in dissolved oxygen concentrations <30% at the end of the 5-day exposure period.For the majority of these systems, 100% mortality occurred.From the oxygen levels in systems without any mortality it could be concluded that the gammarids survived as long as the oxygen saturation level was >30%.At lower oxygen levels, the observed mortality could however (partly) be caused by oxygen deficiency; therefore, these data were excluded from further data analysis (dose-response curve fitting).At 20 °C, this concerned the majority of loadings of 0.0025 mL/L and all oil loadings >0.003 mL/L (resulting in the elimination of 36 out of a total of 69 data points).At 12 °C, three systems with loadings of 0.03, 0.1, and 0.3 mL/L were excluded for the same reason (i.e., <5% of the 65 data points).At 4 °C, the oxygen levels remained >30% in all 52 cases.
In addition to LL50 values (i.e., the toxicity metric based on oil loadings), LC50 values (i.e., the toxicity metric based on TPH concentrations in the WAFs) were calculated (Supporting Information, Table S4b).The latter were expected to better reflect actual toxicity because they take into account the differences in oil solubility at different temperatures.However, several WAF concentrations were not determined (excluded a priori because of budget limitations) or were missing as a result of artifacts that occurred during WAF collection.A 1:1 translation from loadings to WAF concentrations was therefore not possible.Specifically, in some cases, unexpectedly high TPH concentrations were found in combination with deviating TPH profiles, being inconsistent with expected aqueous concentrations.Most probably, contamination of the WAF with the oil layer occurred in these cases because the WAF aliquots used for chemical analysis were collected from what was left in the 5-L glass bottles, after draining them for the toxicity experiments.On one occasion, a thin oil layer was observed in the exposure jar of one of the replicates of the 12 °C, 0.3 mL/L oil loading in the CBR experiment.The results of the chemical analysis of all WAF samples were therefore carefully evaluated, and data points with a TPH profile similar to that of pure DMA oil were excluded (Supporting Information, Table S8).Still, variability was observed in the measured TPH concentrations in WAF samples produced at the same DMA loading and temperature (see Supporting Information, Table S3 and Figure S1).Hence, to be able to translate oil loadings to WAF concentrations, linear, temperature-specific relationships between logtransformed oil loading and log-transformed measured TPH concentration in the WAFs were constructed, using all remaining data (also including previously obtained pilot experimental data; Supporting Information, Figure S1 and Table S2).Because the final LC50 values were obtained by an indirect approach, they should be considered indicative.
We used R (Ver.4.2.2;2022) to test for significant differences among LL50 and LC50 values for the 4 different experiments (i.e., exposure of the temperate G. locusta species at 4 °C, 12 °C, and 20 °C and of Arctic Gammarus sp. at 4 °C).Because assumptions of normality and homogeneity of the residuals were violated, a permutation test of independence was used via the independence_test function from the coin package (Hothorn et al., 2006).Afterward, the pairwisePermu-tationTest function from the rcompanion package (Mangiafico, 2023) was used to perform post hoc pairwise comparisons with false discovery rate-adjusted p values.
Finally, the average time of death for each temperate experiment was calculated.After confirmation of normal distribution based on a D'Agostino and Pearson test, a one-way analysis of variance (ANOVA) was performed, followed by Tukey's multiple comparisons test (Prism, Ver.9.5.0) to test for statistical differences between experiments.

TPH concentrations and profiles in WAFs
Total petroleum hydrocarbon concentrations in the produced WAF series ranged from 44 to 4210 μg/L (considering all temperatures and loadings, control-corrected), with ASW containing 41 ± 26 μg/L as background level (Supporting Information, Text S3).The relationships between oil loading and THP concentration obtained by fitting the experimental data were different for different temperatures, generally implying lower TPH concentrations in the 20 °C compared to the 4 °C and 12 °C systems (Supporting Information, Figure S1).Also, the oil profiles of the WAFs differed per temperature (Supporting Information, Figure S2).The relative contribution of the smallest HBs (C 10 -C 16 ) to the overall TPH concentration decreased with increasing temperature, whereas the relative contribution of the larger HBs (consequently) increased with increasing temperature.

Test organism characteristics
On average, the Arctic gammarids were larger and had a higher individual weight (factor 2-3) than the temperate gammarids (Supporting Information, Figure S4 and Table S6).In addition, in particular the smaller Arctic individuals had a higher lipid percentage than the temperate gammarids (Supporting Information, Figure S5).Total lipid percentage of the Arctic gammarids decreased with size (linear regression, p < 0.05), whereas for the temperate gammarids the opposite trend was observed (not significant, p = 0.19).

Toxicity experiments
The LL50 values for the temperate G. locusta decreased with increasing temperature.The LL50 at 4 °C (0.0056 mL/L) was approximately 2 times higher than those at 20 °C (0.0022 mL/L) and 12 °C (0.0028 mL/L; see Figure 1; Supporting Information, Table S4a, for 95% CIs).For the Arctic Gammarus sp., an LL50 value of 0.095 mL/L was found, which is 17 times higher than the LL50 for the temperate G. locusta exposed at the same temperature (Supporting Information, Table S4a).A permutation test of independence indicated a significant difference in LL50 values among experiments (maxT = 12.67, p < 2.2e−16), with all pairwise comparisons being significantly different, except for the comparison between the 12 °C and 20 °C experiments (Supporting Information, Table S5a).
When expressed as (indicative) LC50s in micrograms of TPH per liter, the values were 114, 261, and 321 for G. locusta at 20 °C, 12 °C, and 4 °C, respectively (Supporting Information, Table S4b for 95% CIs).As such, the LC50 at 20 °C was approximately 3 times lower than that at 4 °C and 2 times lower than that at 12 °C.For the Arctic Gammarus sp.(4 °C), an indicative LC50 value of 825 µg TPH/L was obtained, which is 3 times higher than the LC50 for the temperate G. locusta when exposed at the same temperature.A permutation test of independence showed a significant difference in LC50 values among experiments (maxT = 16.99,p < 2.2e−16), with all pairwise comparisons being significantly different (Supporting Information, Table S5b).

CBR experiments
The CBR for the temperate G. locusta ranged between 387 and 781 mg/kg dry weight (Supporting Information, Table S6) and between 5209 and 10,521 mg/kg when expressed on a lipid weight basis (Figure 2; Supporting Information, Table S7).The CBR was the lowest at 20 °C (5266 ± 79 mg/kg lipid wt; n = 2) and the highest at 4 °C (8995 ± 2157 mg/kg lipid wt; n = 2).The CBR of the 12 °C experiment was in between these values (7861 mg/kg lipid wt; n = 1).
The CBR for the Arctic Gammarus sp. at 4 °C was 923 ± 78 mg/kg dry weight (Supporting Information, Figure S6) and 6833 ± 577 mg/kg lipid weight (Figure 2).These values are similar to those for G. locusta at 4 °C, but the Arctic gammarids have a slightly higher dry weight-based CBR than the temperate gammarids (factor of 1.1-1.8difference) and a lower CBR when expressed on a lipid weight basis (factor of 1.0-1.6).The average total lipid content of the Arctic gammarids (13.5%; n = 3) was 1.8 times higher than that of the temperate gammarids at 4 °C (7.4%, n = 2).Whether the differences in CBR were significant could not be tested because of the limited number of replicates.Note that one replicate for the 12 °C experiment was omitted because of the presence of an oil phase in the test system (see Materials and Methods).
The TPH profiles in all exposed gammarids demonstrated a preferential uptake of the C 14 to C 16 fraction because this fraction was enriched by a factor of 1.6 to 2.5 compared to the general profile in the WAFs and by a factor of 2.0 to 2.5 compared to the DMA oil profile (Supporting Information, Table S8 and Figure S2).The TPH profiles of the exposed organisms were similar for all temperatures, although the C 18 to C 22 and C 22 to C 28 fractions were somewhat more abundant in the organisms exposed at 4 °C and 12 °C compared to those exposed at 20 °C (Supporting Information, Table S8).The TPH profiles in Arctic and temperate gammarids (4 °C) of the CBR experiment were equal (Supporting Information, Figure S3).
At the higher exposure temperatures, organisms died sooner (Figure 3).A significant difference in the time of death was observed between temperate gammarids exposed at different temperatures to a loading of 0.3 mL/L (one-way ANOVA p < 0.0001, F = 45.25;Tukey's multiple comparison tests p < 0.0001).The average time of death was 8.4 ± 2.8 h at 20 °C, 16.4 ± 6.5 h at 12 °C, and 23.8 ± 9.9 h at 4 °C, with in the latter experiment eight individuals still being alive after 48 h of exposure.A direct comparison between the time of death at 4 °C and that in the Arctic experiment was not possible because a different (higher) oil loading was applied in the latter experiment.The exposure time of the Arctic gammarids in the CBR experiment at 4 °C was also extended by another day, up to 72 h in total, such that mortality was caused in all organisms and CBRs could be determined.

DISCUSSION
The results of the present study demonstrate that temperature influenced the acute toxicity of the WAF of oil in  temperate gammarids, expressed as LC50, CBR, and time of death.Furthermore, a threefold difference in the LC50 values between Arctic and temperate gammarids exposed at 4 °C was observed, whereas the CBR was similar for both species.

The effect of temperature on oil toxicity in temperate gammarids
Although the present LC50 values were obtained indirectly (see above, Data analysis), the observed inverse relationship between the metric and exposure temperature is in agreement with earlier study findings.Higher temperatures (32 °C vs. 25 °C) significantly increased oil toxicity to larvae of shrimp, snails, and fish (DeLorenzo et al., 2021).Likewise, LC50 values of a copepod and a rotifer were inversely related to temperature when exposed to chemicals such as copper and dichlorodiphenyltrichloroethane (Li et al., 2014).Several mechanisms may cause or contribute to the observed effect of temperature.First, temperature influences the aqueous solubility and evaporation rate of organic compounds from water, and thereby their bioavailability.Both solubility and evaporation are temperature-dependent and increase with increasing temperature (Brown et al., 2016;Payne et al., 1991;Perkins et al., 2005).In the present experiments, TPH concentrations in WAFs were lower at higher temperatures, and the relative contribution of the LMW hydrocarbons to the overall TPH concentration was lower.This may point to increased evaporation of the (more volatile) LMW compounds at higher temperatures (Brown et al., 2016), during either WAF production or the toxicity experiments.Second, the activity of the test organisms increased with increasing temperature.This has been observed in other studies as well.For instance, an increase in test temperature from 15 °C to 20 °C resulted in an accelerated and condensed life cycle of G. locusta (Neuparth et al., 2002), whereas a copepod and a rotifer displayed dormant behavior when tested at a lower temperature of 4 °C (Li et al., 2014).The increased activity at 20 °C may have resulted in faster chemical uptake, for example, due to increased ventilation.This also implies a higher (biological) oxygen consumption.Probably in combination with an increased chemical oxygen demand at high oil concentrations, this resulted in low oxygen levels at the end of the exposure of particularly the experiments at high oil loadings and high temperature (Supporting Information, Figure S7).Although all data points below the threshold level of 30% oxygen saturation were excluded from further data analysis, stress due to reduced oxygen levels (just) above 30% saturation in the 20 °C experiment cannot be ruled out.Hypoxia may have resulted in a higher sensitivity, as has previously been shown for other amphipod species (Gorokhova et al., 2010(Gorokhova et al., , 2013)), which may have led to an overestimation of oil toxicity at 20 °C.However, because the observed effects at 20 °C are in line with those at 4 °C and 12 °C, where oxygen depletion was not a substantial issue, it is unlikely that the present results were significantly affected by this phenomenon.All in all, oil fate (dissolution and evaporation), as well as the physiology and activity of the test organism, are all influenced by temperature, potentially explaining the observed differences in gammarid sensitivity at the different exposure temperatures.
In addition, a temperature-dependent physiology of the test organisms may have influenced the sensitivity toward oil during the CBR experiment.A toxic effect level is reached when internal concentrations exceed the CBR.In the case of a general acutely toxic, narcotic effect, the concentration of organic contaminants within the lipid bilayer of the cell membranes of an organism exceeds a certain molar threshold concentration (Redman et al., 2022;van Wezel & Opperhuizen, 1995).Petroleum hydrocarbons generally exert such a nonpolar narcotic mode of action (De Laender et al., 2011).In addition to partitioning into membranes, hydrophobic contaminants accumulate in the storage lipids of organisms (van Wezel & Opperhuizen, 1995), where they, however, will not exert any toxicity.Lipid-water partition coefficients may be different for membrane and storage lipids, but both are inversely related to temperature (van Wezel & Opperhuizen, 1995;van Wezel et al., 1996), just like bioaccumulation factors (Muijs & Jonker, 2009b).Furthermore, membrane composition may be adapted by organisms in response to a changing temperature, to maintain membrane fluidity (Hazel, 1997).Considering these processes, the higher CBR values at lower temperature observed in the present, as well as a previous, study ( van Wezel & Jonker, 1998) may be explained as follows.Uptake kinetics are slower at lower temperatures because of reduced molecular diffusion and organism activity, explaining the observed longer times of death (Figure 3).At lower temperatures, partitioning of oil constituents to lipids is enhanced because of a decreased aqueous solubility and the accompanying increased lipophilicity (Muijs & Jonker, 2009b).This results in increased chemical mass transport and transport times, adding to an increase in the time of death.During the longer exposure time, petroleum hydrocarbons will also accumulate more in storage lipids, thereby increasing internal concentrations and the final CBR.Note that part of the accumulated mass thus does not concern a "toxic mass" (i.e., the chemicals associated with the nontarget storage lipids), but this part is included in the total lipid-based CBR (van der Heijden et al., 2015).
The higher contribution of the C 14 to C 16 fraction in gammarids compared to that in the WAFs and pure DMA oil points to enhanced uptake of this fraction in the organisms (Supporting Information, Table S8 and Figure S3).In previous studies, the C 10 to C 16 (Jonker et al., 2006) and C 10 to C 19 (Brils et al., 2002) fractions were suggested to be the most toxic TPH fractions for benthic invertebrates and bacteria.These fractions were also most abundant in the gammarids from the present study, showing summed C 10 to C 16 and C 10 to C 18 fractions of 62% and 74% for 4 °C, 55% and 73% for 12 °C, and 68% and 86% for 20 °C (Supporting Information, Table S8).
To the best of our knowledge, the present study is the first to report CBRs for TPH.To contextualize the obtained CBRs in terms of narcotic effects, the values were translated to lipidbased molar concentrations by assuming the C 14 to C 16 fraction to be the most abundant HB in the exposed gammarids (31%-40% of the overall TPH concentrations in the exposed gammarids).Using the molecular weight of n-pentadecane (C 15 ) as a representative of this HB, this resulted in estimated averaged overall CBRs of 25, 37, and 42 mmol/kg lipid for 20 °C, 12 °C, and 4 °C, respectively.These CBRs differ by less than a factor of 2 and are at the lower end of the 40 to 160 mmol/kg lipid range, as defined for polar and nonpolar narcotic compounds (van Wezel & Opperhuizen, 1995).Although they are somewhat low, these values make sense and, considering the crudeness of the calculation, can be considered acceptable.Note that CBR values to a certain extent are species-specific, which may be related to differences in, for example, body composition and biotransformation capacity (van der Heijden et al., 2015;van Wezel & Opperhuizen, 1995).

Sensitivity of temperate versus Arctic gammarids
The LC50 value for the Arctic amphipods derived in the present study (0.8 mg TPH/L) is similar to, though somewhat lower than, earlier published LC50 values for oil in polar marine species.Previously reported Arctic values include 2.3 to 9.6 mg TPH/L for copepods (Hansen et al., 2013), 1.6 to 4.0 mg TPH/L for copepods and larval and juvenile fish (Gardiner et al., 2013), and 4.6 ± 2.9 mg TPH/L for arthropods (Camus et al., 2014).Furthermore, the LC50 values for Antarctic gammarids were 0.4 to 1.0 mg TPH/L (Brown et al., 2017) and <0.2 mg TPH/L for Antarctic copepod species (Payne et al., 2014).
The Arctic gammarids had a 3 times higher LC50 value than the temperate gammarids when tested under the same conditions at 4 °C.This difference is equal to the previously reported differences for marine species, as obtained by modeling (Bejarano et al., 2017;de Hoop et al., 2011;Olsen et al., 2013).Also, for the Arctic copepod Calanus glacialis, LC50s were 3 to 6 times higher than for the boreal Calanus finmarchicus when exposed to marine diesel oil at 2 °C and 10 °C (Hansen et al., 2013), and a similar difference was observed for Arctic and temperate copepod and shrimp species, when exposed at their ambient temperature to produced water (i.e., the discharge from oil and gas production; Camus et al., 2015).However, a smaller difference (factor of 1.3) was found when comparing C. glacialis and C. finmarchicus upon exposure to a WAF produced from weathered crude oil (Hansen et al., 2011) and when studying the effects of produced water on multiple Arctic and temperate species (Camus et al., 2015).Overall, despite these differences, the present study, literature reports, and modeling results all suggest a factor of 2 to 3 difference in LC50s between Arctic and temperate counterpart species.
In contrast, the CBR of the Arctic gammarids in the present study was similar to that of the temperate gammarids, but the test setup had to be adapted to be able to observe effects; that is, a longer exposure time and a higher oil loading were required to reach the CBR.A similar need for modifications of test methods has been reported in previous studies when testing species from the polar regions (Chapman & Riddle, 2005;Hansen et al., 2011;Olsen et al., 2011).The differences in sensitivity and the required test modifications are probably related to the morphological and physiological adaptations of Arctic crustaceans to their cold environment, such as larger body size, higher lipid content, altered lipid composition, and differences in activity.These factors affect the accumulation rate and total chemical mass flux, leading to acute toxicological responses of the animals to organic contaminants such as oil (Camus et al., 2014).The Arctic gammarids in the present study had a 2 to 3 times higher body weight than their temperate equivalents.A larger body weight implies a longer time needed before a critical oil mass has partitioned into the cell membranes to reach the CBR and produce narcotic effects.Still, in a recent modeling study, only a limited influence of body weight (or size) on the final hydrocarbon concentrations in organisms was found; however, it should be noted that this conclusion was drawn based on pooled data from different literature sources (Redman et al., 2022).
In addition to a higher weight, the total lipid percentage of the Arctic Gammarus sp. in the present study was almost 2 times higher than in the temperate gammarids.Similarly, other researchers reported Arctic copepods and arctoboreal krill species to have a 2 to 5 times higher lipid percentage than their equivalent temperate and subtropical counterparts (Huenerlage et al., 2016;Kattner & Hagen, 2009).Moreover, the lipid composition of Arctic and temperate invertebrate species may differ.Arctic and arctoboreal crustaceans have been observed to contain a higher percentage of wax esters and triacylglycerol (i.e., nonpolar storage lipids, which are used for energy storage) compared to their counterparts from lower latitudes (Falk-Petersen et al., 2000;Huenerlage et al., 2016).Approximately 65% of the total lipid pool in the arctoboreal krill species consisted of wax esters and triacylglycerol, whereas this was only approximately 15% in the boreal-subtropic counterpart (Huenerlage et al., 2016).The higher total lipid content, the higher storage-to-membrane lipid ratio, and the difference in membrane composition in Arctic crustaceans, compared to their temperate counterparts, will have consequences for the CBR and the time needed for uptake and distribution of narcotic compounds within the organisms up to the CBR level (Hazel, 1997;van der Heijden et al., 2015).Further studies assessing the role of lipid composition are therefore recommended to advance the understanding of compound distribution and toxicity in temperate and Arctic species (van der Heijden & Jonker, 2011;van der Heijden et al., 2015).

Influence of test setup on study results
Oil is a challenging substance for toxicity testing, and test results may therefore be associated with a variety of uncertainties (Faksness et al., 2008).In the present study, the chemical analyses of the WAFs suffered from some artifacts (see Materials and Methods), illustrating the challenging nature of oil toxicity testing.The occurrence of artifacts cannot always be confirmed visually, but interestingly, oil in particular provides the opportunity to reconstruct any artifacts a posteriori by distinguishing different HBs.In the present study this allowed for excluding unreliable data but also for confirming the quality of the remaining data and the absence of, for instance, microdroplets.The formation of microdroplets during WAF production may occur because of inconsiderate stirring, but the clearly different TPH profiles of the pure test oil, on the one hand, and the WAFs and gammarids, on the other, confirmed the absence of a dispersion in the present study (exempting the one case in the CBR experiment as discussed).The confirmed biased data were omitted, and missing data were dealt with by constructing temperature-specific oil loading-WAF TPH relationships.Although the LC50 values which were derived with the help of these relationships should be considered indicative, there is no reason to doubt the observed trends with temperature, all the more so because they agree with previous findings.
In addition, working with biota generally introduces data variability.In this respect, it is crucial to test with well-defined species.Unfortunately, in the present study it could not be excluded that our Arctic test species pool was contaminated with G. oceanicus.However, the potential presence of this species is unlikely to have had major consequences for the interpretation of the data.Despite the fact that the maximum body size of G. oceanicus (30 mm on Svalbard) is slightly smaller than that of G. setosus (35 mm on Svalbard; Węsławski et al., 2020), test individuals were sorted into a size class; and therefore, differences in size will not have played a role in the uptake and distribution of oil constituents.Still, lipid content may differ slightly between these species, but this characteristic was directly measured in the present study.The lipid content of the test organisms, and thereby the final results of toxicity tests, may also be influenced by the season, for example, because of the link with the reproductive cycle (Skogsberg et al., 2022).For instance, G. setosus showed the highest total lipid content between June and August (>2% based on wet wt) and lower lipid percentages in early spring (<2%; Skogsberg et al., 2022).Considering and correcting for these differences is important when interpretating toxicity data of substances accumulating in these specific biological matrices.

Concluding remarks
The observed differences in sensitivity between temperate and Arctic gammarids were within a factor of 3, when expressed in terms of LC50.Such variability is considered acceptable for risk-assessment purposes in modeling studies (Hendriks et al., 2001).In addition, the differences in sensitivity in terms of the CBR were less than a factor 2. Therefore, the present study further supports the use of temperate acute oil toxicity data as a proxy for the sensitivity of species living in the Arctic region.This finding is important because the risks of oil spills in the Arctic are increasing, while at the same time empirical oil toxicity data for the Arctic are limited.However, when applying temperate toxicity data to the Arctic, the effects of temperature on the availability of petroleum hydrocarbons, as well as organism traits, should be considered.
The present study only concerned acute, narcotic effects; and it should be noted that oil can also cause adverse effects by smothering organisms, by inducing specific toxicity, or by causing sublethal effects upon chronic exposure to low concentrations (Stark et al., 2017;Szczybelski et al., 2018).
Evidently, all such potential effects should be included when assessing or modeling the potential risks of oil in Arctic environments.
Supporting Information-The Supporting Information is available on the Wiley Online Library at https://doi.org/10.1002/etc.5897.

FIGURE 2 :
FIGURE 2: Averaged critical body residues of total petroleum hydrocarbons in the temperate species Gammarus locusta (dotted bars) after exposure to distillate marine grade A oil at test temperatures of 4 °C (n = 2), 12 °C (n = 1), and 20 °C (n = 2) and in Arctic Gammarus sp.(n = 2; striped bar) at 4 °C.Numbers in parentheses above the bars indicate the number of replicates; error bars represent standard deviations.CBR = critical body residue; TPH = total petroleum hydrocarbons.

FIGURE 3 :
FIGURE 3: Cumulative time of death for the individual gammarids in the critical body residue experiment with a test duration of 48 h (A) and a box and whiskers plot, showing median, 25th-75th percentile, and minimum and maximum values for the time of death for the temperate Gammarus locusta, upon exposure to a water accommodated fraction prepared based on a loading of 0.3 mL oil/L at 4 °C, 12 °C, and 20 °C (B).