Near-future ocean acidification does not alter the lipid content and fatty acid composition of adult Antarctic krill

Euphausia superba (Antarctic krill) is a keystone species in the Southern Ocean, but little is known about how it will respond to climate change. Ocean acidification, caused by sequestration of carbon dioxide into ocean surface waters (pCO2), alters the lipid biochemistry of some organisms. This can have cascading effects up the food chain. In a year-long laboratory experiment adult krill were exposed to ambient seawater pCO2 levels (400 μatm), elevated pCO2 levels mimicking near-future ocean acidification (1000, 1500 and 2000 μatm) and an extreme pCO2 level (4000 μatm). Total lipid mass (mg g−1 DM) of krill was unaffected by near-future pCO2. Fatty acid composition (%) and fatty acid ratios associated with immune responses and cell membrane fluidity were also unaffected by near-future pCO2, apart from an increase in 18:3n-3/18:2n-6 ratios in krill in 1500 μatm pCO2 in winter and spring. Extreme pCO2 had no effect on krill lipid biochemistry during summer. During winter and spring, krill in extreme pCO2 had elevated levels of 18:2n-6 (up to 1.2% increase), 20:4n-6 (up to 0.8% increase), lower 18:3n-3/18:2n-6 and 20:5n-3/20:4n-6 ratios, and showed evidence of increased membrane fluidity (up to three-fold increase in phospholipid/sterol ratios). These results indicate that the lipid biochemistry of adult krill is robust to near-future ocean acidification.


Results
Effect of pco 2 on total lipid and phospholipid/sterol ratios in krill. Quantities of total lipid in krill in weeks 1-5 did not differ between pCO 2 treatments or weeks (pCO 2 ; p = 0.577, week; p = 0.097; pCO 2 *week; p = 0.165). During these first five weeks of the experiment, average quantities of total lipid in krill (Fig. 1A) were 57.4 ± 19.8 mg/g dry mass (DM; mean ± SD). During weeks 26-43 (Fig. 1A), there was a fourfold increase in average total lipid in krill to 273.8 ± 75.4 mg/g DM (mean ± SD), and the effect of pCO 2 on total lipid differed between weeks (Two Way ANOVA; pCO 2 *week, p = 0.052).
Krill in 4000 µatm pCO 2 seawater had lower quantities of total lipid than krill in all other pCO 2 treatments during week 26 (Tukey p < 0.003). They also had lower total lipid than krill in 400 and 2000 µatm pCO 2 during week 41 (Tukey p < 0.046). During weeks 39 and 43, the quantities of total lipid in krill did not differ between pCO 2 treatments (p > 0.930).
principal component analysis of fatty acid percentage composition. Fifty-eight fatty acids were found in krill. Only fatty acids at percentages of ≥0.5% of total fatty acids (17 fatty acids) are analysed and presented in the following results.
Fatty acid percentage data for adult krill collected during weeks 1, 2, 4 and 5 (summer) were similar, and data collected during weeks 26-43 (winter and spring) were similar, so data were combined into these two separate groups for principal component analysis (PCA). Results of PCA analyses for individual weeks can be found in Supplementary Figs S1 and S2.
Fatty acid percentage composition of krill did not differ between pCO 2 treatments during weeks 1-5 when analysed using PCA ( Fig. 2A). Principal component 1 (PC1) separated krill with higher percentages of LC-PUFA from those with higher percentages of 14:0 and medium-chain (C 16 -C 18 ) monounsaturated fatty acids (MUFA) and PUFA, but no separation of pCO 2 treatments was observed along PC1(x-axis) or principal component 2 (PC2; y-axis).
Percentages (mean ± SD) of the eight fatty acids with the highest PCA loadings (

Discussion
Krill reared in ambient seawater pCO 2 levels (400 µatm pCO 2 ), and those levels predicted for the near-future (100-300 years; 1000-2000 µatm pCO 2 ) did not have significantly different quantities of total lipid (mg g −1 DM) or ratios of PL/ST. We observed no effects of near-future pCO 2 on fatty acid composition during weeks 1-5 of the experiment (summer). In winter and spring (weeks [26][27][28][29][30][31][32][33][34][35][36][37][38][39][40][41][42][43], elevated percentages of C 18 fatty acids 18:4n-3 and 18:3n-3 were measured in krill in 400-2000 µatm pCO 2 and were highest in krill in 1500 µatm pCO 2 . Krill in extreme pCO 2 (4000 µatm), had a different lipid composition to those in 400-2000 µatm pCO 2 treatments during winter and spring. Some krill had lower quantities of total lipid, higher PL/ST ratios and MCL, and all krill had consistently lower ratios of n-3/n-6 fatty acids (22:6n-3/20:4n-6 and 18:3n-3/18:2n-6). The absence of any pCO 2 effect on krill biochemistry during the first five weeks of the experiment suggests that it may take some time before changes can be detected in the lipid profile of adult krill. Numerous short-term ocean acidification studies have, however, detected changes in the lipid and fatty acid profile of other organisms, over time periods substantially shorter than or equal to five weeks 23,27,30,34,35 . Krill metabolism is controlled by endogeneous rhythms which are cued by seasonal changes in photoperiod, and krill have higher metabolic rates during summer 42 . A higher metabolic rate may enable krill to more efficiently regulate acid-base balance and other vital functions such as their lipid biochemistry. This could explain why effects of extreme pCO 2 on krill biochemistry were not observed during summer, but were most evident in winter (week 26) when metabolic rates are lowest. The interaction between seasonal metabolic rates, lipid biochemistry and increased pCO 2 at different time scales is a topic for further study.
Many ocean acidification studies to date have found no effect of near-future pCO 2 on total lipid levels in organisms 11,31,33,34,[43][44][45] . Lipids are an important energy source and essential for physiological function and survival, therefore, organisms are likely to maintain relative lipid levels unless they are under substantial physiological stress.
Like other laboratory studies 46,47 , krill in our study displayed seasonal fluctuations in lipid mass even when given a constant food supply. This occurred in all pCO 2 treatments, indicating that endogenous rhythms entrained by the seasonal light cycle were the dominant driver controlling lipid deposition in krill 47 .
Our finding that near-future pCO 2 did not have a significant effect on total lipid mass in adult krill suggests that ocean acidification does not affect their ability to feed or store fat. This corresponds well with a recent study 41 , which indicates that physiological processes in adult krill are unaffected by near-future acidification. Animals may preferentially retain lipids and utilize protein as an energy source when exposed to near-future pCO 2 43 , or maintain lipid and protein levels but grow at a slower rate 33 . Adult krill in near-future pCO 2 , however, do not display slow or delayed growth compared with those in ambient seawater 41 . A previous study found that krill exposed to 750 µatm pCO 2 for 24 hours had slightly lower protein content than krill in ambient pCO 2 seawater 48 , suggesting that they may switch from lipid to protein catabolism in high pCO 2 conditions. Near-future pCO 2 did not significantly alter the composition of fatty acids associated with immune function (n-3/n-6 ratios) and cell membrane fluidity (MCL, PUFA/SFA and PL/ST) in krill. This is a further indication that these levels of pCO 2 do not induce physiological stress. Cell membrane fatty acid composition is tightly regulated by temperature 49 and may be driven more by the cold temperatures krill are adapted to 14 . Ambient seawater temperatures (0.5 °C) were used in this study, which could explain the stability of these fatty acid ratios. Elevated www.nature.com/scientificreports www.nature.com/scientificreports/ seawater temperatures may influence fatty acid composition more than acidification, although previous studies indicate that krill lipids are not altered by temperatures up to 4 °C above ambient 46 .
Decreases in total lipid and increases in levels of inflammatory n-6 PUFA in krill reared in 4000 µatm pCO 2 , suggest that unlike krill in 400-2000 µatm pCO 2 , these krill were physiologically stressed. Lipid depletion observed during selected weeks in winter and spring corresponds with decreases in quantities of storage lipid (triacylglycerol) in these krill 41 . Physiological processes such as growth and maturation, along with acid-base regulation required in extreme seawater pCO 2 , are energetically expensive 50 and these processes may have depleted lipid reserves. Feeding in these krill may have also been compromised and caused a decrease in total lipid, although feeding rates were not measured in this study.
Krill in 4000 µatm pCO 2 seawater may have been storing 20:4n-6 for production of inflammatory eicosanoids, and for ion transport 15,51 , in an attempt to regulate immune responses and maintain intra-and extra-cellular pH in this extreme environment. Such increases in n-6 fatty acids have been observed in fish 28 and shrimp 30 exposed to acidification. Inflammation is important for organism health and tissue repair, but excessive inflammation is maladaptive 52 .
As levels of n-6 fatty acids in organisms increase, levels of n-3 fatty acids decrease, as the elongation-desaturation pathways for n-3 and n-6 fatty acids compete for the same enzymes 53 . The lower n-3/n-6 ratios in krill in 4000 µatm pCO 2 during winter and spring may, therefore, correspond to a shift in elongation-desaturation pathways used by these krill. The increase in n-3 PUFA in krill up to 1500 µatm pCO 2 , followed by a decrease down to 4000 µatm pCO 2 , suggests that 1500 µatm pCO 2 may be the point at which krill fatty acid composition switches from an anti-inflammatory status (more n-3 PUFA) to a pro-inflammatory status (more n-6 PUFA).
Cell membrane alteration via homeoviscous adaptation has been most commonly explored with respect to changing temperatures 10,54 , but other factors such as salinity, hypoxia 18 , and pH 11,21 can alter membrane structure. The higher ratios of PL/ST in krill in 4000 µatm pCO 2 in winter and spring suggests that krill may have been actively increasing membrane fluidity, to enable more efficient exchange of ions across their cell membranes and control acid-base balance. Alternatively, the ability of krill in 4000 µatm pCO 2 to maintain an optimal ST composition may have been compromised. This could lead to membrane 'hyper-fluidity' and disrupt cellular function 11 . Under hypercapnic stress, homeoviscous adaptation through regulation of lipid class ratios (e.g. PL/ST) may be more energy efficient than modification of PUFA/SFA ratios and MCL, which remained more stable in krill in 4000 µatm pCO 2 .
The fatty acid profile of krill in our laboratory study also reflects their aquarium diet and does differ to that of wild krill. Ratios of 22:6n-3/20:4n-6, 20:5n-3/22:6n-3, and 18:3n-3/18:2n-6 are higher in wild krill 3,36 than were  www.nature.com/scientificreports www.nature.com/scientificreports/ observed for our laboratory reared krill, indicating that wild krill have higher n-3/n-6 ratios. The diet of wild krill is not replicable in the laboratory 55 , but the higher n-3/n-6 ratios of these krill may influence and even further enhance their resilience to elevated pCO 2 , due to their higher levels of anti-inflammatory fatty acids. Levels of krill prey in the Southern Ocean also fluctuate spatially and temporally 37 , and krill in our study were fed a constant supply of food. Krill increase their feeding rates when exposed to high pCO 2 48 , possibly to maintain enough energy for physiological processes under pCO 2 stress 41,48 . Changing food levels both seasonally, regionally and with climate change may, therefore, also influence how wild krill respond to ocean acidification.
Krill will be exposed to multiple climate change stressors in the future, in addition to ocean acidification 4 . Rapid warming is already evident in the West Antarctic Peninsula region 56 , both at the sea surface 57 , and in the deep ocean (Antarctic Bottom Water 58 ). In laboratory studies, simulated ocean warming significantly affects the fatty acid composition of some organisms 11,26,29,59 . A previous long-term laboratory study found only minor differences between the lipid and fatty acid composition of krill reared in −1 °C, 1 °C and 3 °C 46 . The temperature range used in this study was within the range that krill experience in their natural environment (krill are abundant at South Georgia where seawater temperatures reach 5 °C 37 ), therefore, the temperatures may not have been high enough to detect significant temperature effects. Further studies are needed to establish whether the combined effects of increased seawater temperature and pCO 2 affect the lipid and fatty acid composition of krill.
conclusions Lipid mass and fatty acid composition in adult krill were unaffected when krill were exposed to near-future levels of pCO 2 (1000-2000 µatm) in the laboratory. Extreme pCO 2 altered the lipid and fatty acid content and composition of krill, although consistent differences were not observed across all experimental weeks. Extreme pCO 2 had no effect on krill lipid biochemistry during summer, but during selected weeks in winter and spring, krill in 4000 µatm pCO 2 had elevated levels of inflammatory omega-6 fatty acids and showed evidence of increased membrane fluidity. These observations suggest that krill may be less able to tolerate elevated pCO 2 conditions during winter and spring, when metabolic rates are lower and reproductive maturation occurs. Seawater pH levels are also lower in the Antarctic in the winter than summer 60 , and prey availability is lower in winter in some areas of the Southern Ocean 37 . Collectively, these factors may influence how krill respond to near-future pCO 2 in the wild, and determine their resilience in a future high CO 2 world. Methods experimental conditions. Experimental conditions are described in detail in a previous manuscript 41 .
Briefly, krill were collected from the Southern Ocean (66-03°S, 59-25°E and 66-33°S, 59-35°E) on the research and supply vessel (RSV) Aurora Australis, using a mid-water trawl net (sampled within the top 100 m of the water column). They were held in shipboard aquaria using standard husbandry methods 61 , and transported to the Australian Antarctic Division Krill Aquarium in Tasmania.
For ocean acidification experiments, five 300 L tanks were equilibrated to five pCO 2 levels; 400 μatm pCO 2 (pH 8.1 control treatment), 1000 μatm pCO 2 (pH 7.8), 1500 μatm pCO 2 (pH 7.6), 2000 μatm pCO 2 (pH 7.4) and 4000 μatm pCO 2 (pH 7.1). Seawater temperature of all tanks was held at 0.5 °C (±0.2). Seawater chemistry for the duration of the experiment is reported in the supplementary material of a previous manuscript 41 . Observational units (CO 2 treatment tanks) could not be replicated, due to the large tank size required to achieve the best possible animal husbandry for this pelagic species, and the limited space and resources available for these large tanks over such a long-term study. Tanks were inspected daily, and there was no visual evidence to suggest that tank effects were confounding our experimental results.
Two hundred krill were randomly assigned to each tank on the first day of the experiment (25 th January 2016), and reared in these pCO 2 treatments until the experiment ended on the 12 th December 2016. Light was controlled in the laboratory to mimic the seasonal Southern Ocean light regime (66°S, 30 m depth) and krill were fed six days per week with a mixed microalgal diet of the Antarctic species Pyramimonas gelidicola (2 × 10 4 cells mL −1 ), and Reed Mariculture Inc. (USA) cultures of Thalassiosira weissflogii (8.8 × 10 3 cells mL −1 ), Pavlova lutheri (4.5 × 10 4 cells mL −1 ) and Isochryisis galbana (5.5 × 10 cells mL −1 ).

Sample collection and lipid extraction.
Krill were sampled from the pCO 2 treatment tanks in experimental weeks 1, 2, 4 and 5 (summer), 26 (winter), and 39, 41 and 43 (spring). Five to ten krill were sampled from each tank during each sampling week (only three krill were sampled from the 4000 μatm pCO 2 tank due to increased mortality in that tank and lower overall numbers of krill 41 ). Individual krill were placed in cryo-tubes and frozen immediately at −80 °C until needed for lipid analysis.
Krill were weighed (wet mass), and the length of each specimen was measured from the tip of the rostrum to the tip of the uropod using measurement 'Standard Length 1' 62 . To prevent sample degradation, krill were kept frozen during the measuring process. A dry mass (g) for each krill sample was obtained by multiplying the wet mass by 0.2278 to account for the 77.2% water content in krill 2 .
Krill specimens were added to separatory funnels and extracted using a modified Bligh and Dyer method 63 , consisting of a methanol:dichloromethane:water (MeOH:CH 2 Cl 2 :H 2 O) solvent mixture (20:10:7 mL), and overnight extraction. Phase separation was carried out the following day by adding 10 mL CH 2 Cl 2 and 10 mL saline MilliQ H 2 O to each separatory funnel, giving a final MeOH:CH 2 Cl 2 :H 2 O solvent ratio of 1:1:0.85. The lower layer was drained into a round bottomed flask, and the total solvent extract was concentrated using rotary evaporation. The concentrated extract was transferred into a pre-weighed 2 mL vial and the solvent was blown down under nitrogen (N 2 ) gas to obtain a total lipid extract (TLE) weight. Solvent (CH 2 Cl 2 ) was added until further procedures were carried out to avoid oxidation. www.nature.com/scientificreports www.nature.com/scientificreports/ Lipid class analysis. TLE were used to obtain the lipid class composition of each sample. Aliquots (1 μl) of each TLE were spotted on chromarods and developed in a solvent bath of hexane:diethyl-ether:acetic acid (90:10:0.1 mL, v-v:v) for 25 min, before drying in an oven at 50 °C for 10 min. Chromarods were placed in an Iatroscan MK-5 TLC/FID analyser (Iatron Laboratories, Tokyo, Japan) for analysis. A standard solution of known quantities of wax esters (WE), triacylglycerols (TAG), free fatty acids (FFA), sterols (ST), and phospholipids (PL) was used to confirm peak identities and to calibrate the flame ionisation detector. Lipid class peaks were labelled using SIC-480II Iatroscan Integrating Software v.7.0-E, quantified using predetermined linear regressions, and expressed as mg per g of krill dry mass (mg g DM −1 ). Triacylglycerol data is presented in an earlier manuscript 41 . Only the PL to ST ratio is presented in this manuscript as we were primarily interested in investigating homeoviscous adaptation in krill. fatty acid analysis. To prepare fatty acid methyl esters (FAME), a subsample of the TLE was transferred to a glass test tube fitted with a Teflon lined screw cap, and treated with 3 mL methylating solution (MeOH: CH 2 Cl 2 : HCl (hydrochloric acid), 10:1:1, v-v:v). The sample was then heated at 90-100 °C for 1 hr 15 mins. Samples were cooled and 1 mL of H 2 O and 1.8 mL of C 6 H 14 (hexane): CH 2 Cl 2 solution was added to extract the FAME. Samples were then centrifuged for five minutes at 3000 rpm, and the upper layer containing FAME was transferred to a vial. An additional 1.8 mL of C 6 H 14 :CH 2 Cl 2 was added to the test tube and samples were centrifuged again. This process was repeated three times in total, and samples were blown down using N 2 gas in between transfers. FAME samples were made up to 1.5 mL with CH 2 Cl 2 and stored at −20 °C until further analysis. Prior to analysis, samples were blown down again using N 2 gas and 1.5 mL of internal injection standard (23:0 FAME) was added to each vial.
Samples were analysed via gas chromatography (GC-FID) using an Agilent Technologies 7890 A GC System (Palo Alto, California USA) equipped with a non-polar Equity ™ −1 fused silica capillary column (15 m × 0.1 mm internal diameter and 0.1 µm film thickness). Samples (0.2 µl) were injected in splitless mode at an oven temperature of 120 °C with helium as the carrier gas. The oven temperature was raised to 270 °C at a rate of 10 °C per minute, then to 310 °C at 5 °C per minute. Agilent Technologies ChemStation software was used to quantify fatty acid peaks, with initial identification based on comparison of retention times with known (Nu Chek Prep mix) and laboratory (fully characterised tuna oil) standards. Fatty acid peaks were expressed as a percentage of the total fatty acid area. Fatty acid quantities (in mg g −1 DM and mg g −1 lipid) were calculated using the internal injection standard (C23:0) of known concentration. Confirmation of component identification was performed by gas chromatography-mass spectrometry (GC-MS) of selected samples and was carried out on a ThermoScientific 1310 GC coupled with a TSQ triple quadruple. Samples were injected using a Tripleplus RSH auto sampler using a non polar HP-5 Ultra 2 bonded-phase column (50 m × 0.32 mm i.d. × 0.17 µm film thickness). The HP-5 column was of similar polarity to the column used for GC analyses. The initial oven temperature of 45 °C was held for 1 min, followed by an increase in temperature of 30 °C per minute to 140 °C, then at 3 °C per minute to 310 °C, where it was held for 12 minutes. Helium (He) was used as the carrier gas. The operating conditions of the GC-MS were: electron impact energy 70 eV; emission current 250 µamp, transfer line 310 °C; source temperature 240 °C; scan rate 0.8 scan/sec and mass range 40-650 Da. Thermo Scientific Xcalibur TM software (Waltham, MA, USA) was used to process and acquire mass spectra.
Mean fatty acid chain length (MCL) was calculated using the equation from reference 11 : where MCL ( mg fatty acid g lipid C)/total mg fatty acid g lipid C number of carbon atoms 1 1 Statistical analyses. Principal component analyses (PCA) were carried out in PRIMER 6 (http://www. primer-e.com). Pearson correlation was used due to differences in fatty acid variances, and data were transformed (log x + 1) before analysis. All other statistical analyses were carried out in RStudio (v 1.1.453; www.rstudio.com). Total lipid, specific fatty acids, lipid class and fatty acid ratios were analysed using Two Way ANOVA with pCO 2 and week as main effects, and a pCO 2 *week interaction term. Tukey comparisons or Dunnett's tests were used to compare levels of pCO 2 with one another. On visual assessment of the data, weeks 1-5 were analysed as a group, and weeks 26-43 were analysed as a separate group, as the groups had heterogeneous variances and represented two distinct data sets. Type 3 Sums of Squares were used as the sampling regime was unbalanced. Data were visualised using Q-Q plots and residuals versus fitted values plots, to verify that they met the assumptions of normality and homogeneity of variances. Log or square root transformations were applied when assumptions of normality and/or homogeneity of variances were not met. For Two Way ANOVA of total lipid data, one outlier was removed from the statistical analysis in order to meet assumptions of homogeneity of variances. All tests were two tailed with α = 0.05. Principal component figures were created in PRIMER 6, and all other figures were created using the RStudio packages ggplot2, plyr and dplyr.

Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.