Impacts of crude oil on Arctic sea-ice diatoms modified by irradiance

The loss of summer sea ice in the Arctic is driving the development of marine resources and industries such as shipping lanes, fisheries, tourism, and petroleum exploration (Arctic Council, 2009; Conservation of Arctic Flora and Fauna, 2015; Eguíluz et al., 2016; Cao et al., 2022), which increases the risk of oil spills. Spilled oil may accumulate by pooling at the underside of the ice, saturate the highly porous skeletal layer, percolate through the brine channel system, and become encapsulated in growing ice (Arctic Monitoring and Assessment Programme [AMAP], 2007; Petrich et al., 2013; Oggier et al., 2020). Oil encapsulated in ice remains relatively unweathered, given limited accessibility to microbes and contact with liquid water, therefore retaining much of its original volatility (Boccadoro et al., 2018; Bakke et al., 2021). Encapsulated oil can remobilize under or within sea ice through physical processes such as ice deformation, melting, lead expansion or contraction, and spreading with strong currents (AMAP, 2007). Spills, and subsequent redistribution of oil through dynamic ice processes, render the Arctic ecosystem vulnerable to unweathered oil exposure at any time of the year, even after the point-source has been contained. The ability to respond to an oil spill in the Arctic is impeded by the presence of sea ice which acts as a physical barrier to traditional mechanical and chemical cleanup strategies, for example, boom and skim and dispersant application. A worst-case spill scenario, an oil well blowout, in ice-covered waters would likely persist unmitigated through winter until conditions were more favorable for containment and cleanup response during the following summer. Models show that oil released from an uncontained blowout could accumulate and travel with ice movements >1000 km and affect up to 1 million km² over the course of a year (Blanken et al., 2017). Toxicological research on the impact of crude oil on the marine species and biological communities of the Arctic is necessary to understand the ecological repercussions of an oil spill.

Average annual sea-ice extent has declined by approximately 2 million km² since 1979 (Onarheim et al., 2018), and summer sea ice is predicted to be absent from all Arctic shelf seas by 2050, regardless of emission scenario (Årthun et al., 2021). Loss of Arctic sea ice will change the relative roles of sympagic and pelagic primary producers dramatically (Tedesco et al., 2019; Lannuzel et al., 2020; and references therein). Despite this habitat loss, sea-ice algae will continue to grow in ice that forms each winter on an annual basis. Bacillariophyta (diatoms) proliferate in polar seas and dominate microalgal biomass and diversity within Arctic sea ice (Poulin et al., 2011; Campbell et al., 2016; Szymanski and Gradinger, 2016). The majority of ice-algal biomass is confined to the lowermost centimeters of sea ice where nutrient exchange with underlying seawater supports growth. The vernal ice-algal bloom generally occurs 1–3 months prior to the subsequent phytoplankton bloom (Hsiao, 1992; Ji et al., 2013; Mundy et al., 2014), fueling the base of the Arctic food web (Kohlbach et al., 2016; Kohlbach et al., 2019) by supporting dependent grazers like the copepod Calanus glacialis (Søreide et al., 2010) and under-ice amphipods (Poltermann, 2001; Gradinger and Bluhm, 2009). Key species like the Arctic cod (Boreogadus saida) link ice-algal production to higher trophic level consumers, for example, ringed seals, which are an important subsistence food with cultural significance in Arctic communities (Steiner et al., 2019). The accumulation of microalgae, their grazers, and the congregation of fish at the ice–water interface, renders the sea-ice ecosystem highly susceptible to crude oil exposure that would also accumulate at this interface. Not surprisingly, most toxicological studies on Arctic biota have focused on fauna with important trophic positions, but less attention has been paid to the photoautotrophs that form the base of the food web.

Crude oil is a complex mixture of organic compounds with large compositional differences between source oils (Faksness and Brandvik, 2008); each oil compound has a different solubility and toxicological impact, making the intercomparison of oil toxicity experiments inherently difficult. Exposure to hydrocarbons can result in a decline in growth rate and biomass for phytoplankton (Hsiao, 1978; Gilde and Pinckney, 2012; Bretherton et al., 2018) and sea-ice algae (Fiala and Delille, 1999; Dilliplaine et al., 2021), a decrease in primary productivity (Hsiao et al., 1978; Lemcke et al., 2019), loss of cell motility (Soto et al., 1975; Garr et al., 2014), and even stimulation of growth (Dunstan et al., 1975; Özhan et al., 2014; Bretherton et al., 2018). Part of the differing responses may be due to differences in source oils and interplay with environmental variables. Oil exposure responses also vary widely across microalgal groups, species, and even strains (Gaur and Kumar, 1981; Huang et al., 2011; Gilde and Pinckney, 2012; Özhan et al., 2014; Bretherton et al., 2020); attempts to generalize responses based on higher taxonomic classification or functional groups, for example, diatoms and dinoflagellates, have yielded conflicting results (Taş et al., 2010; Ozhan and Bargu, 2014; Bretherton et al., 2020), though smaller microalgae are generally more sensitive to crude oil exposure due to the increased surface area to volume ratio (Echeveste et al., 2010).

The mechanisms responsible for crude oil growth inhibition remain poorly understood, but wide-ranging damage has been reported, for example, to nucleic acid synthesis and structure (El-Sheekh et al., 2000; Parab et al., 2008), light-harvesting complex proteins (Kamalanathan et al., 2021), and membranes (Sikkema et al., 1995; Hook and Osborn, 2012; Dilliplaine et al., 2021), including limitations to nutrient uptake and gas exchange (Koshikawa et al., 2007). These cell structures may be damaged directly by contact with crude oil or indirectly via other compounds, such as reactive oxygen species induced by oil exposure (Ozhan et al., 2015; Kamalanathan et al., 2021). Species-specific sensitivity and changes to growth rate can alter microalgal community composition with implications for higher trophic levels (Gilde and Pinckney, 2012). Isolate experiments provide detailed response information and illuminate taxon-specific sensitivity thresholds and mechanisms of response to multiple stressors. The toxicological effects of crude oil have the potential to be enhanced when combined with other drivers such as ultraviolet radiation, salinity, and temperature (Sargian et al., 2007; Bender et al., 2021; DeLorenzo et al., 2021).

The near absence of light during the Arctic polar night limits photosynthesis (Berge et al., 2015). The seasonal transition from winter to spring coincides with the return of the sun above the horizon and supports the growth of sea-ice microalgae that are well adapted to low light, growing at irradiances as low as 0.17 µmols of photons m⁻² s⁻¹ (Hancke et al., 2018). This low-light efficiency, combined with their fixed location in the photic zone, provides a growth advantage which allows for accumulation of microalgal biomass in spring prior to the onset of the subsequent phytoplankton bloom (Tremblay et al., 2008; Leu et al., 2011; Song et al., 2016; Ardyna et al., 2020). These adaptations also make them particularly susceptible to photoinhibition and damage when irradiances are excessive (Lund-Hansen et al., 2014; Croteau et al., 2022). Significant decreases in diatom photosynthesis and growth rate due to photoinhibition at light levels above 50 µmol photons m⁻² s⁻¹ can occur in ice algae (Arrigo et al., 2010; Juhl and Krembs, 2010). Irradiance values increase at the ice–water interface by several orders of magnitude during the seasonal transition from snow-covered winter ice to summer melt pond formation (Perovich et al., 1998; Nicolaus et al., 2013).

The primary controllers of light availability to sea-ice algae are snow thickness and ice thickness (Veyssière et al., 2022). The extinction coefficient of snow is over an order of magnitude greater than bare ice; therefore, snow cover has a disproportionately larger attenuative effect than ice thickness (Grenfell and Maykut, 1977). The general decline of ice and snow thickness (Kurtz et al., 2011; Webster et al., 2014; Stroeve et al., 2020) is leading to a brighter Arctic that can result in photoinhibition and damage, and may favor replacement of microalgal species that are better adapted for growth at higher irradiances (Croteau et al., 2022). In particular, the current sea-ice algal specialists, like the dominant Nitzschia frigida and ubiquitous Synedropsis hyperborea, are expected to be replaced in the ice by cryopelagic (abundant in both ice and water) taxa, for example, Fragilariopsis cylindrus (Lannuzel et al., 2020). Understanding the acclimatization potential of sea-ice algal specialists, and cryopelagic taxa, to the large range of irradiances encountered on an annual basis, in conjunction with crude oil exposure, is important to better predict how multiple drivers will impact sea-ice algal growth in a brighter Arctic.

The main objectives of our study were to investigate the photoacclimative ability of select sea-ice diatom isolates to realistic irradiances and to determine the effect of crude oil contamination on growth rate and cell concentration. These two easily-measured variables are important to consider with regard to the timing and magnitude of the spring ice-algal bloom. We hypothesized that the combination of the two anthropogenic drivers of increased irradiance and crude oil exposure would have interactive effects that differed by species. To test this hypothesis, we freshly isolated sympagic algae from sea ice collected near Utqiaġvik, Alaska, and exposed them to a range of water-accommodated fraction (WAF) concentrations of Alaska North Slope (ANS) crude oil over a range of irradiances. To contextualize our results, we simulated the toxicological effect of a crude oil spill prior to the onset of the sea-ice algal bloom on biomass and community composition of a simple microalgal community comprised of the four isolates used in this study.

Cores were collected from landfast sea ice near Utqiaġvik, Alaska (Chukchi Sea; 71°22' N, 156°35' W), on May 25, 2020. The bottom approximately 10 cm of core were melted slowly in the dark, with the addition of 0.2 µm filtered seawater to minimize osmotic shock (Garrison and Buck, 1986). Aliquots of the core melt material were diluted into f/25 +Si medium (Bigelow Labs, ME, USA) according to Guillard and Ryther (1962) using natural seawater (salinity of 36) collected from the Gulf of Alaska at a depth exceeding 500 m. In brief, seawater was filter-sterilized with a 0.2 µm Sterivex filter (MilliporeSigma, St. Louis, MO, USA), macronutrients, and trace metals were added, and the medium was autoclaved. To complete the preparation, filter-sterilized (0.2 µm) vitamins were added after cooling to avoid denaturation of the vitamins. Mixed sea-ice algal cultures were maintained at approximately 15 µmol photons m⁻² s⁻¹ in 25 mm culture tubes at 4°C with a 12:12 h light:dark cycle using a low temperature incubator (Percival, Perry, IA, USA) for propagation and isolation.

Sea-ice algae were isolated from these mixed cultures using 1.25% agar plates containing f/25 +Si medium following Kimura and Tomaru (2013). Briefly, 5 mL of diluted cultures were added to the agar plates allowing cells to settle for 12 h before decantation. Plates were left to incubate until colonies were visible, at approximately 1 month. Single colonies were selected using an inoculation loop, resuspended in f/25 +Si medium, and maintained by transfer to fresh medium every 2–4 weeks. Isolates were confirmed to be unialgal by light microscopy (Olympus BX51, Center Valley, PA, USA; Figure 1). Identification to the species level was performed using a combination of light microscopy, scanning electron microscopy (SEM; FEI Quanta 200), and 18S rRNA gene sequencing (Table S1).

To test the impact of crude oil contamination on growth physiology of these four sea-ice diatom isolates, oil-contaminated media were generated using ANS crude oil at environmentally relevant temperature. ANS crude oil, collected from Pump Station 1 at the Trans-Alaska Pipeline entry point (provided by Alyeska Pipeline Services), was used as a regionally representative oil for this experiment. The WAF of crude oil was prepared according to the Chemical Response to Oil Spills: Ecological Research Forum (CROSERF; Aurand and Coelho, 2005) and Singer et al. (2000). Briefly, 10.5 L of f/25 +Si medium were added to a 20 L acid-cleaned glass aspirator bottle, and 12 g L⁻¹ of ANS crude oil were gently added to the surface of the water. The mixture was stirred continuously for 48 h on a stir plate to produce a vortex of approximately 25% of the medium height, without the formation of droplets, at −1°C in the dark. WAF was collected from below the oil layer through the spigot and was free of any visible oil.

Samples for chemical analysis were collected in 200 mL acid-cleaned amber glass jars with Teflon-lined lids and stored at −20°C until further processing, approximately 12 months from collection. Concentrations of total petroleum hydrocarbons (TPH) were measured from the undiluted WAF at the onset of the experiment as technical replicates (n = 3; Figure S1). Chemical analyses of the stock WAF (100%) were conducted at Emax Laboratories, Inc. (Torrance, CA), using the Alaska Department of Environmental Conservation (ADEC) methods for n-alkanes in the diesel range organics (DRO; C₁₀–C₂₅), residual range organics (RRO; C₂₅–C₃₆), and TPH (C₈–C₄₀) using gas chromatography equipped with a flame ionization detector (GC-FID; Agilent G1530 N; Figure S1), according to methods AK 102/103 (ADEC, 2002a; 2002b). Petroleum concentrations in the dilutions were calculated from the stock concentration as it has been shown to be conserved linearly (Mcfarlin et al., 2011; Gardiner et al., 2013; unpublished data).

Experiments were conducted in a walk-in cold room (Environmental Growth Chambers, Chagrin Falls, OH, USA) located at the University of Alaska Fairbanks. Sea-ice algal isolates were transferred to optically clear 96-well flat bottom plates (Eppendorf, Hamburg, DE) and incubated at −1°C to approximate the temperature of the bottom skeletal layer of sea ice. Plates were set on rigid icepacks (40 × 40 × 1 cm; Sonoco ThermoSafe) with a melting point of −1°C to maintain stable temperature. Five light conditions, 3, 10, 20, 50, and 125 µmol photons m⁻² s⁻¹, were established using LED lights (54 W cool white; Lithonia Lighting, Conyers, GA, USA) with a 12:12 light:dark cycle. Photosynthetically active radiation (PAR) was measured with a Li-190 R planar light sensor (LI-COR, Lincoln, NE, USA), and intensity was tuned using a combination of voltage regulators and neutral density screens. Cultures were transferred to fresh medium approximately every 1–2 weeks, once exponential growth was achieved, and allowed to acclimate to light conditions over 70 days, ensuring that cultures were photoacclimated prior to the determination of growth physiology.

Two sets of 96-well plates (experimental and unoiled) were prepared for each of the four isolates and five light conditions used in the experiment. Experimental plates contained a linear dilution gradient of WAF, from 0% to 90%, for a total of 12 treatments, each with 8 replicates per plate. Unoiled plates were used exclusively for cell harvest during the course of the experiment and contained oil-free medium (0% WAF). Photoacclimated cultures were transferred to each well, after pipetting to mix, at a 1:9 volume ratio of cell culture:WAF. The range of median initial cell densities within wells was 9.5–11.5 × 10³ cells mL⁻¹. Fluorescence was measured daily using a SpectraMax Gemini EM plate reader (Molecular Devices, San Jose, CA, USA) at wavelengths for in vivo chlorophyll a fluorescence (460 ex, 665 em) 8 h after lights were turned off. Experiments were ended after a leveling off, or decline, in fluorescence was observed (Figure S2).

Cells were harvested from the unoiled plates, immediately after fluorescent measurement, four times during the experiment; four replicate samples were collected from each plate and preserved with a 2% final concentration of Lugol’s solution and stored at 4°C until further processing (<6 months). Each fixed sample was a pool of four adjacent wells, with relative fluorescence averaged among these four wells for calculation of cell concentration. Cells were counted using a Sedgewick rafter counting chamber (Forestry Suppliers Inc., Jackson, MS, USA), with at least 100 cells, or 10 µL total volume, counted for each sample. Fragilariopsis cylindrus and Attheya septentrionalis were enumerated manually using an inverted Nikon Eclipse TE2000-U at 400×. Synedropsis hyperborea cells were enumerated from micrographs taken with a Teledyne Photometrics Prime BSI camera (Tucson, AZ, USA) with automated image analysis using the Cell Counter plugin (v. 3.0.0) for imageJ (Schneider et al., 2012). Image analysis counts closely matched manual counts with a slight underestimation of <5%. The relationships between relative chlorophyll a fluorescence and cell concentration were determined using linear regression with an intercept forced through 0 for each combination of isolate and irradiance (Figure S3). Relative fluorescence measured in the experimental plates was converted to cell concentration by multiplying the blank-corrected value by the empirically determined slope.

Maximum growth rate (µ_max, d⁻¹) was calculated as the log-linear change in cell concentration (Ce, cells mL⁻¹) over time (t) using Equation 1:

and the all_easylinear() function described by Hall et al. (2014; R package growthrates 0.8.2) for each individual well. Maximum cell concentration was calculated as the maximum observed cell concentration during the experiment. Error was propagated as the sum of squares using calculated errors associated with cell counts and biological replication.

Photoacclimated growth rate versus irradiance curves were generated for each of the four isolates using the unoiled control wells. The R package “phytotools” (Silsbe and Malkin, 2015) was used to fit the Eilers and Peeters (1988) growth-irradiance model and estimate growth parameters using the Levenberg-Marquardt fitting algorithm (Moré, 1978). This model was selected because it includes a term for photoinhibition, which we expected to occur within our range of irradiances for sea-ice microalgae. The saturation irradiance, e_k, was calculated as the intersection of µ and the initial slope of the growth-irradiance curve, α.

Dose-response curves were generated using species-specific maximum growth rates and concentrations using nonlinear models fit with the R package “drc” (Knezevic et al., 2007). Models with and without irradiance as a factor were tested for significance using ANOVA (p < 0.05); irradiance was significant and included as a factor in all models. Three curves (Attheya septentrionalis, irradiance level of 50 µmol photons m⁻² s⁻¹; Synedropsis hyperborea strain 1, irradiance level 10 µmol photons m⁻² s⁻¹; and S. hyperborea strain 2, irradiance level 20 µmol photons m⁻² s⁻¹) were removed before model fitting due to the presence of a positive trend resulting in model direction reversal and fit. A linear regression was fit to these treatments instead. The best fitting model was selected using Akaike’s information criterion (AIC), and visual inspection when values were close (AIC < 10; Table S2). Models for growth rate have a fixed lower limit of 0, with the assumption that there will be no growth at some concentration of WAF (not observed). The lower limit for dose-response curves generated using the maximum concentration was set to the median initial concentration per species. The concentrations required to inhibit growth rate or maximum cell concentration by 10% (IC₁₀) and 50% (IC₅₀) were calculated from the model fits (Knezevic et al., 2007). Due to the variability of the lower limit set point in models used between growth rate and cell concentration, IC values are absolute for growth rate and relative to initial cell concentrations for maximum cell concentration.

A simple numerical model was used to simulate the growth of a simplified ice algal community over a spring-bloom period in the presence and absence of crude oil exposure with realistic under-ice irradiance. The under-ice irradiance (UI_PAR) was calculated using the Beer-Lambert law (Equation 2) and forced with observations of incident irradiance and sea-ice thickness in the Alaskan Arctic (Figure S4). Incident irradiance (PAR; 30 min average) was downloaded from the National Ecological Observatory Network (NEON) at the Utqiaġvik, Alaska, station (BARR) for the years 2020 (Figure S4A) and 2022 (Figure S4B). PAR data were distilled to a single daily irradiance value (I) by averaging the irradiance within a 12 h window centered on solar noon from March 1 to May 15 (Figure S4C). Irradiance values were then averaged across years for the same date and time, while missing data were filled with values from the year with available data. A daily sea-ice thickness (Hi) value was calculated as the average of all daily point measurements (every 15 min) measured at a Sea Ice Mass Balance station in 2014 (Chukchi Sea; 71°22' N, 156°31' W) using an upward-facing acoustic transducer (NSF Arctic Data Center, 2009; Figure S4E). Fixed snow depth (Hs) was chosen to simulate under-ice light intensities under snow depths ranging from 0 to 20 cm (Figure S4D). The integrated PAR attenuation coefficients for snow (Ks, 11) and ice (Ki, 3.6) were derived from McDonald et al. (2015); Ki was adjusted to 2.6, within the reported confidence interval, to include upwelling irradiance as measured in situ (May 2022) using a spherical quantum sensor (LI-COR).

The simulated ice-algal community was composed of the four isolates investigated in this study seeded at equal starting concentrations. Nutrient concentrations and fluxes were not considered in this simplified model, as it is intended only to summarize the effect of oil on growth rate based on our experimentally determined values for our isolates, free of nutrient limitation. Therefore, biomass estimations should be considered relative. Cell abundance (A_d) was calculated for each day assuming exponential growth (Equation 3) of cell abundance from the previous time step (A_d−1). Isolate-specific growth rate (µ) was a function of irradiance (UI_PAR) and one of two oil concentrations determined empirically in this study: no oil contamination, and at the highest tested concentration (8.7 mg L⁻¹ TPH). A loss term (g) was fixed at a constant fraction of 0.10 of the taxon-specific growth rate (Lavoie et al., 2005). Cell abundance (A) at the initial time step was set at 1000 cells for each of the four algae. Cell concentrations were converted to carbon content (pg C) using the isolate-specific biovolume (BV, µm³) and the relationship derived from all data during exponential growth according to Lomas et al. (2019; Table S1; Text S1).

In order to assess the potential for interacting anthropogenic impacts on the Alaskan Arctic ecosystem, sea-ice algae were isolated near Utqiaġvik, Alaska. Four diatoms, Attheya septentrionalis, Fragilariopsis cylindrus, and two strains of Synedropsis hyperborea were selected from among these isolates based on their prevalence in the original samples, pan-Arctic distribution (Hasle et al., 1994; Poulin et al., 2011; Figure 1), and dominance in Arctic ice, sub-ice, and water samples (Hegseth, 1992; von Quillfeldt, 1995; Melnikov et al., 2002). The four diatom isolates were diverse in their morphology (Figure 1). They varied in biovolume, with the largest being A. septentrionalis, followed by F. cylindrus and S. hyperborea (Table S1). The two strains of S. hyperborea varied in morphology and size, with S. hyperborea strain 1 being longer and thinner compared to the shorter and broader S. hyperborea strain 2 (Figure S5).

The sea-ice diatom isolates also varied in light-acclimated growth-irradiance curves (Figure 2). Ranges of ice-algal growth rates were 0.16–0.40 d⁻¹, 0.15–0.50 d⁻¹, 0.15–0.39 d⁻¹, and 0.15–0.36 d⁻¹ for Attheya septentrionalis, Fragilariopsis cylindrus, Synedropsis hyperborea strain 1 and S. hyperborea strain 2, respectively. Ranges fell within those previously reported for A. septentrionalis (Hsiao, 1978; Hegseth, 1992; Basionym = Chaetoceros septentrionalis Østrup) and Arctic strains of F. cylindrus (Pančić et al., 2015); these are the first reports for S. hyperborea. The irradiance at the onset of inhibition, e_phot, was indistinguishable with certainty between the two strains of S. hyperborea (Table 1). Other growth-irradiance model parameters differed by strain (Table 1). The parameter α was smallest for F. cylindrus, intermediate for S. hyperborea strains 1 and 2, and greatest for A. septentrionalis, while e_k was highest for F. cylindrus, with the lowest e_k calculated for S. hyperborea strain 2 (Table 1).

The mean concentration (± SD; n = 3) of TPHs was 9.7 (± 0.6) mg L⁻¹; of DROs, 3.0 (± 0.0) mg L⁻¹; and of RROs, 0.6 (± 0.3) mg L⁻¹. Approximately 60% of the TPHs were not attributed to the DRO or RRO fractions; nearly all of the remaining TPH concentration was attributed to the lowest molecular weight C_8–9 fraction, which is considered to be the range for gasoline organics (C₆–C₁₀; Figure S1).

Sea-ice diatom isolates were grown in 96-well plates with a gradient of ANS crude oil-contaminated media under five irradiances. There was interspecific variability in sea-ice algal growth rate responses to crude oil contamination that was modified by irradiance (Figure 3, Table 2). Attheya septentrionalis was relatively insensitive to ANS crude oil, except at the highest irradiance level where growth rates declined by approximately 19% at the highest tested TPH concentration (Figure 3A). The 10% growth inhibition concentration (IC₁₀) of ANS crude oil WAF for A. septentrionalis was calculable for the lowest (3 µmol photons m⁻² s⁻¹) and highest (125 µmol photons m⁻² s⁻¹) irradiance levels, with a significantly lower IC₁₀ at the highest irradiance level (Table 2, Figure S6). Values of IC₁₀ for Fragilariopsis cylindrus were calculated for all five irradiances, owing to its greater apparent sensitivity (Figure 3B) and to being the only isolate to exhibit inhibition of the growth rate by 50% (IC₅₀) within the tested concentrations (Table 2). There were significant differences in F. cylindrus IC₁₀ by irradiance (p < 0.05), with a lower IC₁₀ for the intermediate (10 and 20 µmol photons m⁻² s⁻¹) compared with the high (50 and 125 µmol photons m⁻² s⁻¹) irradiance levels. Values of IC₅₀, calculated for F. cylindrus at irradiance levels 10, 50, and 125 µmol photons m⁻² s⁻¹, differed significantly from one another (Table 2; Figure S6); the highest irradiance induced the lowest IC₅₀. Synedropsis hyperborea strain 1 was the least sensitive isolate when comparing calculable IC₁₀ values, though both strains of S. hyperborea exhibited relatively low sensitivity to ANS crude oil WAF (Figure 3; Table 2). IC₁₀ fell within the range of tested concentrations only at the lowest irradiance level (3) for both strains of S. hyperborea, with strain 2 exhibiting a significantly lower threshold than strain 1 (Table 2; Figure S6). Both strains of S. hyperborea exhibited a significant (p < 0.05) hormetic response, approaching 20% relative to the control for S. hyperborea strain 2 at some light levels. Three of the dose-response curves were fit with linear models, A. septentrionalis (irradiance of 50 µmol photons m⁻² s⁻¹), S. hyperborea strain 1 (irradiance of 10 µmol photons m⁻² s⁻¹), and S. hyperborea strain 2 (irradiance of 20 µmol photons m⁻² s⁻¹), because they exhibited an increase in growth rate with increasing TPH across the range of tested concentrations (Figure 3A, C, and D). Linear regressions were significant for S. hyperborea strains 1 and 2 (p = 0.0084, R² = 0.52; p = 0.026, R² = 0.41, respectively) but not for A. septentrionalis (p = 0.18, R² = 0.17).

While the isolates had varying growth rate responses to ANS crude oil WAF, we observed that maximum cell concentrations also differed between strains and treatments (Figure 4, Table 2). Not all growth curves reached a clear peak, with subsequent decline, by the end of the experiment (Figure S2). The dose-response curves for maximum cell concentration generally exhibited higher sensitivity to ANS crude oil WAF relative to those calculated for growth rate, as reflected by lower IC₁₀ concentrations (Table 2). The maximum cell concentrations of Attheya septentrionalis were the least sensitive to ANS crude oil contamination (Figure 4A); A. septentrionalis had a slight, but significant increase in maximum cell concentration with increasing TPH at irradiances below 50 µmol photons m⁻² s⁻¹, and a decrease in maximum cell concentration at higher irradiances. The IC₁₀ values for maximum cell concentration of A. septentrionalis were relatively low and did not differ significantly at irradiance levels of 50 and 125 µmol photons m⁻² s⁻¹ (Table 2, Figure S7). The maximum cell concentrations of Fragilariopsis cylindrus were strongly impacted by TPH exposure (Figure 4), particularly at the intermediate and highest irradiances. Again, F. cylindrus was the only isolate with measurable inhibition by 50% (IC₅₀; Figure S7, Table 2) over the range of TPH. IC₁₀ values for the maximum cell concentration of F. cylindrus differed significantly between two groups of irradiance levels; the IC₁₀ at irradiances of 3 and 50 µmol photons m⁻² s⁻¹ were significantly higher (p < 0.05) than for 10, 20, and 125 µmol photons m⁻² s⁻¹ (Table 2, Figure S7). The IC₅₀ for F. cylindrus mirrored the pattern of IC₁₀ grouping and significance, with the exclusion of irradiance of 20 µmol photons m⁻² s⁻¹ which fell outside the range of tested TPH concentrations (Table 2; Figure S7). Maximum cell concentrations varied between strains of Synedropsis hyperborea, with strain 2 growing to approximately half the concentration of strain 1. Although growth rates were relatively constant or even increasing with increasing TPH concentrations for S. hyperborea (Figure 3C and D), maximum cell concentration declined at all irradiance levels for both strains (Figure 4C and D). Nearly all irradiance levels had an IC₁₀ that fell within the range of tested TPH concentrations for both strains of S. hyperborea, reflecting the increased sensitivity of maximum cell concentration relative to growth rate (Table 2).

To better understand how a crude oil spill might alter the diversity and biomass of sea-ice algae in the Arctic, we incorporated the dose-response data for these strains into a simple numerical model to simulate the impact of an oil spill prior to the spring bloom. The model was forced with measurements of daily irradiance and sea-ice thickness from the Alaskan Arctic coast, and the under-ice irradiances were calculated assuming three fixed snow depths (Figure S4). The seasonal increase of incident irradiance was similar between the two years of in situ data (Figure S4A and B), so a single daily irradiance value was used to force the model (Figure S4C). The mean daily irradiance ranged from 30% to 89% of the daily maximum incident irradiance (Figure S4A–C). The ice thickness data that was used to force this model reached a maximum thickness of 1.37 m in late April (Figure S4E). Calculated under-ice irradiances reached maximum values of approximately 28, 9, and 3 µmol photons m⁻² s⁻¹ under snow depths of 0, 0.1, and 0.2 m, respectively (Figure S4D). The impacts of two oil conditions on sea-ice algal production were evaluated: no oil contamination (0.0 mg L⁻¹ of TPH) and oil contamination equal to the highest tested concentration from this study (8.7 mg L⁻¹ of TPH). Results from the model show that snow depth and crude oil exposure altered simulated community composition and biomass (Figure 5). The relative abundances were distributed more evenly under moderate-to-high snow cover in the absence of crude oil contamination (Figure 5A). Under low-snow conditions, without oil contamination, Fragilariopsis cylindrus dominated the community (Figure 5A), but remained low in the presence of oil (<5%; Figure 5B). Predicted biomass declined with increasing snow depth and in the presence of oil (Figure 5C–E).

This study focused on exploring the potential consequences of dual anthropogenic stressors on sea-ice diatoms in the Alaskan Arctic. The combination of two anthropogenic drivers, increasing irradiance from declining snow/ice thickness and crude oil spills, are increasingly likely with a warming Arctic Ocean (Webster et al., 2014; Veyssière et al., 2022). To better understand the impacts of these anthropogenic changes on the sea-ice ecosystem in the Alaskan Arctic, we isolated four diatoms from landfast ice near Utqiaġvik, Alaska, and characterized their responses to a fully factorial gradient of irradiance and crude oil contaminated conditions. The four sea-ice diatoms were freshly isolated for this study to minimize the evolutionary drift that can occur in microalgae over time (Schaum et al., 2013) and were chosen to represent ecologically relevant species with a range of physiological and morphological traits (Figure 1).

Sea-ice algae are exposed to extreme shifts in light intensity, spectrum, and duration, ranging from seasonal variations (e.g., changes in day length and ice thickness) to diel cycles related to solar and lunar zenith angle, to rapid meteorologically driven changes (e.g., cloud cover and precipitation; Johnsen et al., 2021). Photoacclimation is the result of physiological changes that compensate for these shifts in irradiance based on genetic adaptations (Dubinsky and Stambler, 2009). To determine the photophysiology of the four sea-ice diatom isolates, we acclimated cultures to five irradiances and determined their exponential growth rates (Figure 2). We chose ecologically relevant irradiances for our experiment, 3–125 µmol photons m⁻² s⁻¹, where the lower end of this range corresponds to returning light near the end of winter and to late spring irradiances under high snow cover (Campbell et al., 2018), while the highest irradiances tested are comparable to bare ice and open water in late spring (Massicotte et al., 2020).

The maximum growth rate of Attheya septentrionalis determined in this study (approximately 0.40 d⁻¹; Figure 2A) was similar to that reported by Hegseth (1992; approximately 0.37 d⁻¹ at −0.5°C) but far greater than reported by Lomas et al. (2019; 0.03 d⁻¹ at 2°C) at an irradiance of 20 µmol photons m⁻² s⁻¹; these are the only previously reported growth rates for this species. Lomas et al. (2019) used axenic cultures, which may explain in part the relatively low growth rates recorded for all eleven species tested in that study, as diatoms are known to have beneficial interactions with heterotrophic bacteria (Amin et al., 2012). Our estimates of the maximum growth rate of Fragilariopsis cylindrus (approximately 0.5 d⁻¹; Figure 2B) were similar to those from previous research using Arctic strains (Pančić et al., 2015), but well exceeded those typically reported for Antarctic strains (<approximately 0.31 d⁻¹; Mock and Hoch, 2005; Arrigo et al., 2010; Jabre and Bertrand, 2020; Croteau et al., 2022), with the exception of Sommer (1989) who reported a specific growth rate of 0.69 d⁻¹. We suggest that the vast geographic separation of this bipolar species, or the age of the culture, yield conclusions that differ from more recently collected, regionally specific isolates. To our knowledge, the growth rates reported here are the first to be published for Synedropsis hyperborea; they fall within a typical range of growth rates for Arctic sea-ice diatoms (Lacour et al., 2017).

Arctic diatoms exploit environmental niches, partly defined by light availability, through differences in their photophysiology, for example, the sea-ice environment experiences overall low light levels with low amplitude changes relative to open water (Croteau et al., 2021). Sympagic algae that are well adapted to their environment have a high a and low e_k, demonstrating the capacity to utilize photons efficiently to support growth at incredibly low irradiances, that is, 0.17 µmol photons m⁻² s⁻¹ (Hancke et al., 2018). In contrast, planktonic taxa have a lower a and higher e_k, which shifts their niche preference toward higher light conditions and is generally coupled with mechanisms that allow them to grow uninhibited by supersaturating irradiances, for example, non-photochemical quenching (Croteau et al., 2021). Cryopelagic growth-irradiance parameters can be expected to fall between these two extremes, balancing the contrasting irradiance niches. Fragilariopsis cylindrus is a cryopelagic taxon (Poulin et al., 2011; Hop et al., 2020; Croteau et al., 2022), while Attheya septentrionalis and Synedropsis hyperborea are considered to be sympagic and are often found attached to strands of Melosira arctica or colonies of Nitzschia fridida as epiphytes (Poulin et al., 2011; Poulin et al., 2014; Campbell et al., 2018). We found a range of growth-irradiance parameters, suggesting that all of our strains are adapted to growth in low-light conditions compared to other polar phytoplankton, with e_k falling into the lowest quartile out of 18 polar diatoms compiled by Lacour et al. (2017). Of the species tested, F. cylindrus was the only cryopelagic taxon and was distinct from A. septentrionalis and S. hyperborea, with the lowest a and highest e_k (Table 1). Different ice types and features (e.g., multi-year vs first year ice, ridges, melt ponds, sediment inclusions) increase heterogeneity of light transmittance through ice and may act as an environmental filter that influences community structure from the bottom up. Knowledge of growth-irradiance parameters for particular taxa may therefore improve our understanding of their fine-scale distribution as it relates to irradiance microhabitats.

For the Arctic strain of Fragilariopsis cylindrus used in this study, the onset of photoinhibition occurred at a much higher irradiance, 100 µmol photons m⁻² s⁻¹, than in previous work using an Antarctic strain of F. cylindrus, <50 µmol photons m⁻² s⁻¹ (Arrigo et al., 2010; Croteau et al., 2022). This two-fold difference is notable because much of the work being conducted on F. cylindrus utilizes the model strain (CCMP1102), which was collected from the Antarctic but is being used for Arctic-focused studies. For example, Croteau et al. (2022) presents a detailed study using sea-ice, cryopelagic and planktonic microalgae, but their use of the Antarctic strain of F. cylindrus (CCMP1102) demonstrated photoinhibition and a maximum growth rate more similar to the sea-ice algal endmember, Nitzschia frigida, used in their study. Using Arctic-specific isolates of F. cylindrus may better explain its importance as part of under-ice algal blooms in the Arctic (Arrigo et al., 2012) due to a higher maximum growth rate and photoinhibition occurring only at a higher irradiance, as measured in this study. Growth-irradiance data from Hegseth (1992) showed no signs of photoinhibition for Attheya septentrionalis, measured to an irradiance of 400 µmol photons m⁻² s⁻¹. Our limited irradiance curves, ending at 125 µmol photons m⁻² s⁻¹, present an important caveat, as they introduce uncertainty surrounding the degree of photoinhibition incurred. Attheya spp. have been shown to accumulate rapidly in sea ice with low snow cover, dominating relative abundance and outcompeting pennate diatoms like Nitzschia frigida (Campbell et al., 2018), which suggests that they may occupy a high-light niche within sea-ice associated communities. Together, these data suggest that A. septentrionalis and F. cylindrus are better adapted to an Arctic with thin or bare sea ice than species that are photoinhibited at such high light levels like Synedropsis hyperborea (this study) and N. frigida (Hegseth, 1992; Croteau et al., 2022).

The relative toxicity of crude oil depends on its chemical composition and concentration, as well as environmental conditions, such as temperature which impacts solubility and viscosity (Atlas, 1981; Das and Chandran, 2011). Variability in source oil composition and WAF preparation methodology, for example, temperature, mixing energy, chemical dispersants, light, and salinity, influence the concentration and composition of the resulting WAF of crude oil. Utilizing the source oil most likely to contaminate a region of interest, as well as preparing the WAF and toxicity tests under environmentally relevant conditions, are therefore important. We prepared our WAF using ANS crude oil at a temperature of −1°C to closely approximate the source oil and conditions under which we might expect an oil spill in the Alaskan Arctic, that is, in the Chukchi or Beaufort seas. The freezing point of seawater, at approximately −1.8°C, is the temperature at the ice–water interface where much of the ice-algal biomass is concentrated. The choice of a slightly warmer temperature was necessary to avoid freeze-thaw cycles due to the small liquid volumes used in the microwell plates and the difficulty in maintaining such a precise temperature in the incubator. Concentrations of TPH generated in our WAF were similar to ranges generated using similar conditions and source oil (Gardiner et al., 2013).

The chemical composition of WAF generated in this study was dominated by short-chain hydrocarbons that can be lost due to volatilization (Das and Chandran, 2011). Crude oil, or contaminated seawater, under sea ice can be expected to experience less volatilization due to low temperatures (Atlas, 1975) and the presence of ice, which inhibits gas exchange (Rutgers van der Loeff et al., 2014). The presence of this lighter fraction of crude oil has been shown to increase toxicity and inhibit biodegradation (Atlas, 1975; Chen et al., 2008; Rodriguez Martinez et al., 2008; Sherry et al., 2014). Our study utilized microwell plates sealed with parafilm to reduce water loss due to evaporation but the relatively large headspace may have allowed some volatilization of crude oil compounds with additional loss by adsorption to the container walls (Berrojalbiz et al., 2011), potentially reducing the effective concentrations during the experiment. Recent studies have also shown considerable biodegradation of petroleum hydrocarbons using Arctic sea-ice, sub-ice, and open-water microbial communities, ranging from 48% to 94% (McFarlin et al., 2014; Garneau et al., 2016; Bakke et al., 2021). Our sea-ice diatom isolates were non-axenic and therefore had associated communities of heterotrophic bacteria which are thought to be the primary source of biodegradation. The anticipated decline in TPH during the study could result in culture recovery, such as was observed by Kamalanathan et al. (2021); however, Garneau et al. (2016) found that sea-ice microbial communities degrade oil much more slowly than the sub-ice community under identical conditions. We did not measure hydrocarbon loss during our experiments due to the small sample volumes, but both biotic and abiotic losses of TPH likely occurred. Therefore, our study presents potentially low estimates of the impacts of ANS crude oil on the sea-ice diatom isolates investigated. Despite this potential underestimation, we observed substantial impacts on microalgal growth.

We used exponential growth rate and maximum cell concentrations of microalgae, which integrate cellular physiology (Sunda and Huntsman, 1995), to provide robust indicators of sublethal crude-oil toxicity. A common approach for investigating toxicological impacts on microalgae is to measure and analyze response variables at single time points (Gilde and Pinckney, 2012; Dilliplaine et al., 2021; Putzeys et al., 2022); a major drawback of this approach is that the conclusions may be time-dependent and conclusions will vary based on the termination date. We used a combination of methods to assess toxicological impacts without a pre-determined termination date. The maximum growth rate was determined by finding the maximum exponential growth over the log-linear portion of the growth curve; the maximum cell concentration was determined from the whole growth curve and occurred after several days to several weeks. Factors that can influence maximum cell concentration include changes in cell nutrient quotas, self-shading, death, and toxicological inhibition of growth. For example, cellular repair from oil-induced damages may increase nutrient quotas leading to a decrease in maximum cell concentration independent of the growth rate. Growth rate and maximum cell concentration therefore represent distinct measures providing information regarding the toxicological impact of crude oil on sea-ice algae. Changes to growth rate and cell concentration can be used to infer how an oil spill may alter the timing and magnitude of the vernal ice algal bloom. We hypothesized that declines in growth rate and maximum cell concentration would be greatest for treatments combining the highest irradiance level and most concentrated crude oil WAF due to the combined stress. The sea-ice diatoms used in this study had a range of crude-oil sensitivities and a complex interaction with irradiance levels that did not always support this hypothesis. In our study, Fragilariopsis cylindrus was the most sensitive isolate assayed, with calculable values for IC₁₀ of the growth rate and cell concentrations for all treatments, and it was the only isolate with measurable values for IC₅₀ (Table 2). IC₁₀ and IC₅₀ are measures used for reporting toxicity data, with the former signifying sensitivity (Beasley et al., 2015). These findings contrasted with the other isolates, Attheya septentrionalis and both strains of Synedropsis hyperborea, which had relatively high tolerances to crude oil in our study with few measurable values for IC₁₀ (Table 2). This variability in sensitivity is similar to previous studies. Özhan et al. (2014) measured near-complete growth inhibition in 5 phytoplankton taxa by a concentration of 10 mg L⁻¹ TPH exposed to Louisiana sweet crude, in contrast to Bretherton et al. (2020) who measured some taxa showing stimulation to growth exceeding 20 mg L⁻¹ TPH of chemically dispersed Macondo crude oil. Our experiments were conducted simultaneously with the same treatment conditions and therefore represent considerable taxon-specific variability.

The mechanisms of crude oil growth inhibition/stimulation are poorly understood despite the widespread observance of these responses. Oxidative stress by the formation of reactive oxygen species is produced during normal metabolic activity but can cause extensive damage and mortality when antioxidant defenses are overwhelmed (Hansel and Diaz, 2021). Ozhan et al. (2015) reported oxidative stress induced by crude oil exposure and large differences in antioxidant efficiency between two phytoplankton taxa, but the role of oxidative stress in growth inhibition remained unknown. Kamalanathan et al. (2021) suggested that oxidative damage is secondary to growth reduction from the damaged light-harvesting complex and electron transport proteins. Differences in light-harvesting complex proteins or susceptibility of antioxidants to disruption by crude oil compounds across species could explain the range of sensitivities displayed here, as well as interactive effects or lack thereof, between irradiance and WAF.

Excessive irradiance alone can impact growth negatively (Section 4.1); we expected to observe the most significant inhibition at the highest irradiance level tested due to an additive or multiplicative effect from combined stressors. Attheya septentrionalis had a significantly lower growth rate IC₁₀ at the highest light level (Table 2, Figure S6), indicating higher sensitivity and supporting our hypothesis. We were unable to determine whether oxidative stress played a role in this enhanced inhibition. We expected the greatest combinatorial effect to be exhibited by the Synedropsis hyperborea isolates because they exhibited photoinhibition in the absence of oil (Figure 2C and D). In contrast, the IC₁₀ of growth rate for S. hyperborea strains indicated that these diatoms were relatively insensitive to crude oil WAF (Figure 3C and D) at all light levels and appeared to be more sensitive to ANS crude oil contamination at the lowest light level tested (Table 2, Figure 3C, D, G and H; Figure S6). This finding directly refutes our hypothesis that combined stressors would enhance growth rate inhibition. Maximum cell concentration of S. hyperborea declined at all irradiance levels, indicating a greater susceptibility to crude oil exposure than indicated by growth rate alone (Figure 4C, D, G and H), possibly indicating a greater metabolic burden in the presence of oil leading to increased nutrient demand. Interestingly, an increase in the maximum cell concentration was observed for A. septentrionalis, with increasing oil for irradiance levels of 3–20 µmol photons m⁻² s⁻¹, in spite of declines in growth rate (Figures 3A, E and 4A, E). While not measured directly in the oiled treatments, a decline in cell size or nutrient quota possibly supported a greater cell concentration at the end of the experiment. Mutualistic interactions between microalgae and bacteria might also explain this discrepancy. Bacteria are known to provide microalgae with iron (Amin et al., 2009), vitamin B12 (Croft et al., 2005), CO₂, and protection from reactive oxygen species (Hünken et al., 2008), while microalgae provide bacteria with organic carbon and oxygen. Additionally, the bacterial degradation of crude oil results in the production of sugars, lipids, and amino acids that may stimulate growth of microalgae (Goutx et al., 1984). Differences between the bacterial community composition, activity, or abundance could lead to different nutritional availability for microalgal growth leading to the observed differences in maximum cell concentration. Microbial interactions and/or changes in cell nutrient quota may be responsible for the observed discrepancies in A. septentrionalis growth rate and cell concentration data, but were not determined and are beyond the scope of this study.

The mechanisms of crude-oil-induced damage proposed by Kamalanathan et al. (2021) include large scale damage to light-harvesting complex proteins, inhibiting photosynthesis, along with impacts on downstream processes such as the ability to catabolize organic carbon. The greater inhibition of relative growth rate by Synedropsis hyperborea strains at the lowest light level could indicate subtle damage to the light-harvesting complex or catabolic pathway, leading to energy limitation in the presence of oil that could be overcome with greater irradiances. The values for IC₁₀ of Fragilariopsis cylindrus growth indicated the greatest sensitivity at the intermediate irradiances (10 and 20 µmol photons m⁻² s⁻¹; Table 2), though they did not differ significantly from the low irradiance (3 µmol photons m⁻² s⁻¹; Table 2), and may indicate an energy-dependent growth disruption alleviated at even higher irradiances. The cellular stress response counteracts macromolecular damage, regardless of the stressor, by induction of proteins such as molecular chaperones and removal of macromolecular debris through ubiquitination, and so on, and can convey cross-tolerance to different stressors (Kültz, 2003). Alternatively, deployment of cellular stress response mechanisms by S. hyperborea in the presence of excessive irradiance could stress-harden the taxon preventing combinatorial inhibitory effects. However, threshold-based sensitivity, that is, IC₁₀, provides a limited view of the dose response, especially when applied to biphasic curves such as those of F. cylindrus (Stebbing, 2009).

The shape of dose-response curves also differed by taxa, revealing differences in their underlying physiological responses to crude oil exposure. Monophasic dose responses, like the common sigmoidal log-logistic regression, have a single region of decline in response to a toxic compound. Prior to the decline, a homeodynamic response can compensate for low doses of a toxic compound until a critical threshold is reached (Stebbing, 1981; 2009). Multi-phasic models include the capacity to capture multiple steps in a curve resulting from multiple homeodynamic responses with individual thresholds, or the overemployment of a homeodynamic response that can result in the stimulation of the measured variable (hormesis; Stebbing, 2009). The growth rate and maximum cell concentration of Attheya septentrionalis and cell concentration of Synedropsis hyperborea (Table S2) were best fit with monophasic dose-response curves, suggesting that a single homeodynamic response was overwhelmed over this range of crude oil exposure. In contrast, the dose-response curves of Fragilariopsis cylindrus showed two distinct inhibitory phases, an initial minor inhibition with a plateau phase followed by a steep tipping point at approximately 6 mg L⁻¹ (Figures 3B and 4B). The expectation that cryopelagic taxa, particularly F. cylindrus, will play a greater role in a brighter Arctic may render the microalgal communities here more susceptible to crude oil exposure (Lannuzel et al., 2020). This bi-phasic dose response of F. cylindrus suggests at least two distinct inhibitory mechanisms or homeodynamic responses (Di Veroli et al., 2015). We hypothesize that some cellular function of F. cylindrus is deployed (e.g., repair mechanisms as part of the cellular stress response) by low levels of TPH, partially countering inhibitory effects, until damage becomes excessive and homeostasis is lost.

Some dose-response curves for Synedropsis hyperborea strains 1 and 2 displayed a hormetic response; that is, a stimulation of growth at low concentrations of TPH (Figure 3C, D). According to Stebbing (1981), hormesis is a common characteristic of dose-response curves, and may exceed 200% relative to the control (Agathokleous and Calabrese, 2020). Stimulation of the growth rate of S. hyperborea remained below 20% for both isolates. Previous studies have found microalgal growth to be stimulated to approximately 1 mg L^−l of crude oil concentration, with inhibition occurring thereafter (Huang et al., 2011; Ozhan et al., 2014), while some studies have seen stimulation to growth in excess of 20 mg L⁻¹ (Bretherton et al., 2020). Stimulation of S. hyperborea growth rate generally peaked around 0.8–1.6 mg L⁻¹ TPH but remained higher than the control for all concentrations at irradiances of 10 µmol photons m⁻² s⁻¹ (strains 1 and 2) and 20 µmol photons m⁻² s⁻¹ (strain 2; Figure 3C, D, G, and H). There is no clear indication as to why intermediate irradiances were the most resistant to inhibition for this species nor has a mechanism for light-mediated toxicity tolerance been described, but S. hyperborea may possibly employ an energy-dependent strategy in response to crude oil exposure. No hormetic response was observed for maximum cell concentration in dose-response curves of S. hyperborea (Figure 4C and D); instead, there was a general decline at all light levels. These findings suggest that increased growth rate, stimulated by crude oil exposure, may require a larger nutrient cell quota resulting in a quicker nutrient limitation. This combination of crude oil tolerance for growth rate at light levels common under sea ice suggests that S. hyperborea may be positioned physiologically to experience little interruption after exposure to crude oil when nutrients are replete, but may be more sensitive to sublethal inhibition during winter to early spring when light is restricted and during the spring bloom when nutrients may become limiting. Further research is required to determine the mechanisms behind the various phases of toxicity for these taxa and which cellular processes are disrupted or deployed to counter toxicity-induced cellular stress.

We used a simple numerical model to simulate the biomass accumulation and relative compositional change of our isolates in sea ice under varying snow depths, with and without oil, to contextualize our results over a spring growth season in the Alaskan Arctic (Figure 5). The light environment was forced with seasonal irradiance and ice thickness data from Utqiaġvik, with variable snow depth; attenuation by snow is much greater than ice and is the controlling factor for transmitted irradiance (Figure S4). In the absence of oil, the diatom community shifts in response to snow depth from dominance by Fragilariopsis cylindrus to Attheya septentrionalis to Synedropsis hyperborea with increasing snow depth (Figure 5A). At the end of the simulation, biomass declined with increasing snow depth and was always lower when exposed to oil at the same snow depth (Figure 5C–E). The impact of oil on community structure and biomass was largest in the zero-snow scenario (Figure 5B and C). For example, F. cylindrus was nearly entirely absent from the community exposed to oil, except at the deepest snow depth where irradiance was also limiting to the other isolates. This model result shows the relative sensitivity of F. cylindrus to crude oil exposure and highlights the threat of a crude oil spill to a key Arctic species.

This model has several important caveats and limitations. For example, the model does not take into account nutrient limitation, changes in salinity, or radiative absorption. Attheya spp. have been shown to outcompete pennate diatoms at higher irradiances when combined with low-salinity and low-nitrogen conditions (Campbell et al., 2018). Nutrient limitation could also impact the final biomass, as a previous study found lower biomass under bare ice in the Canadian Arctic (Campbell et al., 2015). However, in a more locally relevant study in the Alaskan Arctic, removal of thin snow layers (4–5 cm) during peak bloom stimulated algal growth in landfast sea ice from Utqiaġvik, supporting the assumption in our model that nutrients are not limiting in the spring (Juhl and Krembs, 2010). This model is also limited in considering only four strains of sea-ice algae. We expect the true sea-ice algal community to be more diverse, as over 1000 species of algae have been documented in sea ice (Poulin et al., 2011); for example, Synedropsis hyperborea has not been observed to dominate the sea-ice community even under thick snow cover. The oil concentration of 8.7 mg L⁻¹ TPH in our model, and our experiments, was a static exposure. Oil concentrations are likely to decline over time as dilution and or biodegradation occur, but sustained high concentrations of crude oil may be encountered where brine is in contact with oil or when continuously resupplied as during a blowout event. A worst-case Arctic oil spill, as a well blowout, could affect as much as 1.1 million km² with high concentrations of oil, representing a significant threat to sea-ice microalgal biomass (Blanken et al., 2017), comparable to our model assumptions.

Our study highlights the importance of toxicological studies that use high-resolution doses in order to determine responses and thresholds at low concentrations. We found evidence of stimulation and inhibition that was dependent on sea-ice algal species, oil concentration, and irradiance level. The interaction of crude oil exposure and irradiance was complex and not generalizable across all taxa. Growth rate and maximum cell concentration represent useful measurements because they integrate the physiological responses of an organism and are straightforward to measure, making them ideal candidates for high-throughput assays. The reduction of growth rate and maximum cell concentration may delay the timing and magnitude, respectively, of a spring sea-ice algal bloom. However, growth rate and maximum cell concentration alone do not provide a mechanistic understanding of crude oil stimulation/inhibition, which is required to make broader predictions for unmeasured taxa. We provide evidence that the toxicological impact of crude oil exposure on the sea-ice algal community could shift community structure and inhibit biomass accumulation. The degree of growth inhibition is species-specific; of our isolates tested, the abundant and widespread cryopelagic taxon, Fragilariopsis cylindrus, was the most impacted. The predicted increase in the relative importance of F. cylindrus in a brighter Arctic may make the Arctic ecosystem more prone to disruption from oil spills in the future.

The following datasets were generated:

• DNA sequences: Genbank accessions OR037227 and OR037228

• R scripts: Publicly available on GitHub at https://github.com/kbdilliplaine/Isolate-Growth.git

The supplemental files for this article can be found as follows:

Figures S1–S7. Tables S1–S2. Text S1. Docx

The authors thank the Ukpeaġvik Iñupiat Corporation (UIC) Science for their logistical support in Utqiaġvik during field work and willingness to core and ship ice during a global pandemic. The authors thank Jim Janoso for his microscopy expertise and the two anonymous reviewers whose comments and suggestions improved this manuscript.

This project was funded by a research grant provided through the Coastal Marine Institute (CMI M20AC10007), a joint institute between the Bureau of Ocean Energy Management (BOEM) and the University of Alaska Fairbanks. Further support and match funding was provided by the College of Fisheries and Ocean Sciences at the University of Alaska Fairbanks. Micrographs were captured at the UAF Molecular Imaging Facility, NIGMS P20GM130443.

The authors indicate that they have no known competing interests.

• Contributed to conception and design: KD, GH.

• Contributed to acquisition of data: KD.

• Contributed to analysis and interpretation of data: KD, GH.

• Drafted and/or revised the article: KD, GH.

• Approved the submitted version for publication: KD, GH.

How to cite this article: Dilliplaine, K, Hennon, G. 2023. Impacts of crude oil on Arctic sea-ice diatoms modified by irradiance. Elementa: Science of the Anthropocene 11(1). DOI: https://doi.org/10.1525/elementa.2023.00074

Domain Editor-in-Chief: Jody W. Deming, University of Washington, Seattle, WA, USA

Associate Editor: Kevin R. Arrigo, Department of Earth System Science, Stanford University, Stanford, CA, USA

Knowledge Domain: Ocean Science