Interpopulational differences in the nutritional condition of Aequiyoldia eightsii (Protobranchia: Nuculanidae) from the Western Antarctic Peninsula during austral summer

Sample collection

To assess nutritional condition of a key benthic species among localities with contrasting environmental conditions in Antarctica, adult individuals of the bivalve mollusk A. eightsii were collected from three roughly equidistant sites along the WAP. Samples were collected during austral summer by SCUBA diving at 10–15 m depth (Fig. 1). The individuals of the O’Higgins (63.3°19′S, 57°53′W; n = 24) and the Rothera stations (67.6°34′S, 68°07′W; n = 15) were collected during January 2018, while the individuals of the Yelcho station (64.9°52′S 63°35′W; n = 19) were collected during March 2017. Unfortunately, it was not logistically possible to obtain samples simultaneously from all three-study sites and the potential implications of this sampling design are discussed. After collection, all the samples were immediately preserved in 99% ethanol and maintained at –80 °C. Then, samples were transported to the UCSC Hydrobiological Resources laboratory at Concepción, Chile and kept under the same conditions until their analysis 4 weeks later. The collection permits were granted by the UK Government for JR17001 and JR18003 expeditions: 31/2017 and S6-2018/01. Also, a permit was granted for collection adjacent to Rothera Research Station: 33/2017.

Recently, potentially cryptic species have been documented, suggesting two different lineages of A. eightsii along the WAP (González-Wevar et al., 2019). Accordingly, to avoid biases in the biological response, samples from a single lineage have been used in the present study, following genetic analyses (Muñoz-Ramírez et al., 2020).

Shell length and body mass

These data were obtained as previously described in Bascur et al. (2020). Using Vernier calipers with 0.01 mm precision, we determined the individuals’ sizes, measured as the distance between the anterior and posterior edges of the shell (i.e., shell length). To determine the body mass of each individual, the soft tissue was separated from the shells and washed with abundant distilled water on a 0.2 mm sieve in order to remove salt and sediment. Then, samples were frozen at −20 °C for 24 h in independent labeled Eppendorf tubes and subsequently dried for 48 h at −80 °C by sublimation in a lyophilizer (FDU-701; Operon, Gimpo City, South Korea). Finally, using an analytical balance with a sensitivity of 0.1 mg (LA230S; Sartorius, Göttingen, Germany), body mass was determined as the dry mass of the individuals.

Proximate biochemical composition (lipid and protein content)

Following methods described in Bascur et al. (2020), the proximate composition was measured in 20 mg of homogenized dry mass for each individual and expressed in absolute values (mg 20 mg⁻¹), and then calculated in relative values of dry mass for each biochemical component [% dry mass, (DM) = (mg of component × 100)/mg of DM]. In order to improve the performance of the tests, samples were exposed for 15 min at 6 °C in an ultrasonic bath (AC-120H; MRC, Netanya, Israel) with distilled water (protein content) or dichloromethane: methanol (lipid content), and were then analyzed using the methods outlined below.

Lipid content was quantified both in the dry samples and in the ethanol in which the samples were preserved, following the gravimetric method of Folch, Lees & Stanley (1957), modified by Cequier-Sánchez et al. (2008). Each dried sample was homogenized in amber tubes with 5 mL of dichloromethane: methanol (2:1). Then, samples were combined with 4 mL of 0.88% potassium chloride, mixed for 15 s in a vortex (SBS100-2; Select Vortexer) and centrifuged (S-8; Boeco) for 5 min at 6 °C and 1,500 rpm. The precipitate of each sample was transferred to pre-weighed vials and dried through evaporation using ultrapure nitrogen gas (109A YH-1; Glass Col). Total lipid extract obtained by evaporating the solvent was weighed on a precision balance (120A, Precise) and was calculated by subtracting the weight of the empty vial from the weight of the vial with the lipid extract. A similar method was used to obtain the lipid content that potentially was released from the samples into the solvent in which they were preserved (i.e., ethanol). The ethanol from each sample was evaporated, in a previously weighed flask, through a rotary evaporator (RE-2000A; Winkler). Once the solvent has evaporated, the lipid content was obtained by subtracting the weight of the empty flask from the weight of the flask containing the lipid extract. Since we found a very small quantity of lipid in the single ethanol extracts (only 8–10% of the total individual lipid content) we decided to pool the lipid content found in each ethanol sample with the lipid content found in each individual. Finally, lipid extract of each sample was preserved at –80 °C in dichloromethane: methanol (2:1) with butylhydroxytoluene (BHT) as an antioxidant to avoid sample degradation.

Protein content was quantified using a microplate adaptation of the BIO-RAD colorimetric assay of Lowry et al. (1951). This kit included three reagents: S (aqueous solution of sodium dodecyl sulfate), A (alkaline copper tartrate solution) and B (diluted Folin solution). The dry samples of 4 mg for each individual were homogenized in 200 μL of ultrapure water (Mili-Q, Bedford, MA, USA). Then, 5 μL of the mixture was transferred to a 96-well microplate with 200 μL of Reagent B and 25 μL of Reagent A′ (mixture of 20 μL of Reagent S and 1 mL of Reagent A). Subsequently, the samples were shaken for 15 s in a vortex (SBS100-2, Select Vortexer) and incubated in the microplates for 15 min at room temperature. Finally, the absorbance was measured with a spectrophotometer at a wavelength of 750 nm (ELx808; Biotek, Winooski, VT, USA). The concentration of each sample was obtained using a calibration curve for proteins, created by diluting different concentrations of bovine serum albumin (500-0111; Bio-Rad).

Energy content

The energy content (J 20 mg DM⁻¹) was estimated using a bioenergetics equivalent from the biochemical composition data (lipid and protein), as formerly described in Bascur et al. (2020). The bioenergetics equivalents were calculated through conversion coefficients: (a) 1 mg of lipids = 39.54 J, (b) 1 mg of protein = 23.64 J. An approximation of the total energy content for each individual was calculated by adding the energy equivalents of the biochemical composition (total energy = J mg lipid + J mg protein) (Winberg, 1971; Urzúa et al., 2012; Bascur et al., 2017).

Fatty acid composition

Fatty acid profile was determined through standard methods (Urzúa & Anger, 2011; Bascur et al., 2018; Bascur et al., 2020). Specifically, fatty acid methyl esters (FAMEs) were measured after preparation using the sample’s lipid extract. Lipid extracts were esterified at 70 °C for 1 h in a Thermo-Shaker (DBS-001; MRC, Netanya, Israel) using sulfuric acid (1% in methanol) incubations. Then, each sample was vortexed (SBS100-2, Select Vortexer) with 3 mL of n-hexane and centrifuged for 15 s. This process was repeated 3 times and the supernatant was transferred to labeled tubes. Finally, using a nitrogen evaporator (109A YH-1; Glass Col), fatty acids were concentrated. The measurement of FAMEs was performed using a gas chromatograph (Agilent, model 7890A) at set temperature equipped with a DB-225 column (J&W Scientific, 30 m in length, 0.25 internal diameter, and 0.25 µm film). Using chromatography software (Agilent ChemStation, Santa Clara, CA, USA), individual FAMEs were identified by comparison to known standard fatty acids of marine origin (certificate material, Supelco 37 FAME mix 47885-U (Malzahn et al., 2007; Urzúa & Anger, 2011). Each sample was quantified using a calibration curve for fatty acids, diluting different concentrations of Supelco 37 FAME mix standard.

Statistical analysis

Statistical analyses were performed based on standard methods (Sokal & Rohlf, 1995; Clarke & Gorley, 2006; Zuur, Ieno & Graham, 2007) in the STATISTICA V8 and PRIMER V6 (+ PERMANOVA) software packages with a 95% confidence level (p < 0.05). The assumptions of the ANOVA analysis were evaluated with Kolmogorov-Smirnov tests for normality and Levene test for homogeneity of variances. Considering that sample sizes for each group were different, a type 3 sum of squares was used for ANOVA and PERMANOVA analyses. Besides, when significant differences were detected for ANOVA or Kruskal-Wallis test, post hoc Tukey HSD or multiple range tests with a Bonferroni correction were performed to assess differences among localities, respectively. All analyzes were performed with locality as a factor [with 3 levels: O’Higgins station (northern WAP), Yelcho station (middle WAP) and Rothera station (southern WAP)].

The shell length of A. eightsii individuals collected at the three study localities was analyzed through a one-way ANOVA as assumptions of normally distributed data and homogeneity of variances were fulfilled. Because these assumptions (normality and homogeneity) were not fulfilled for soft tissue dry mass, lipid and protein content (mg and %DM) and energy content of A. eightsii individuals captured at the three study localities, these variables were analyzed by non-parametric Kruskal-Wallis tests. Also, the assumptions of normality and homogeneity of variances were evaluated for the quantity of each fatty acid (e.g., C16: 0) and for the total values of each group of fatty acids (e.g., total saturated fatty acids, SFA) among the three study localities. The vast majority of fatty acid comparisons were analyzed with a Kruskal–Wallis test because they did not fulfill ANOVA assumptions. The exceptions analyzed with a one-way ANOVA after a log (x + 1) data transformation, were C18:0 (normality: KS = 0.11, p > 0.20; homogeneity: F = 2.91, p = 0.06), C22:6n-3 (normality: KS = 0.14, p > 0.20; homogeneity: F = 2.50, p = 0.09) and the total of SFA (normality: KS = 0.12, p > 0.20; homogeneity: F = 1.89, p = 0.16). On the other hand, the fatty acids C18:2n-6c and C22:1n-9 were only found in two localities, and as they did not fulfill the assumptions of normality and homogeneity, they were analyzed with a Mann–Whitney U test.

In addition, multivariate analyses were conducted to compare fatty acid composition. A one-way permutational multivariate analysis of variance (PERMANOVA) analysis based on Bray-Curtis similarity and fourth root data transformation was performed to evaluate the complete fatty acids data set. Moreover, a similarity percentage analysis (SIMPER) was carried out to observe the percentage of contribution of each fatty acid to dissimilarity among localities. Last, a principal component analysis (PCA) based on Bray-Curtis similarity and square root data transformation was used to visualize the spatial distribution of data and the fatty acid with the highest contribution at each locality (Pearson Correlation > 0.9).

Shell length and body mass

Shell length (mm ind.⁻¹) showed no significant differences among the three study localities (Fig. 2A, Table S1). In contrast, body mass (mg ind.⁻¹) was significantly different between the study localities, since individuals around O’Higgins and Yelcho stations had a higher body mass than individuals from Rothera station (Fig. 2B, Table S2).

Proximate biochemical composition and energy content

Significant variation among locations were found for lipid content (mg 20 mg DM⁻¹; Fig. 3A, Table S2), lipid percentage (% DM; Fig. 3B, Table S2), protein content (mg 20 mg DM⁻¹; Fig. 3C, Table S2), protein percentage (% DM; Fig. 3D, Table S2) and energy content (J 20 mg DM⁻¹; Fig. 4, Table S2). In all these cases, higher values occurred at O’Higgins station compared to Yelcho and Rothera stations.

Fatty acid composition

One-way ANOVA results showed some significant differences among the fatty acid profiles at the three study localities. The amount of total fatty acid, saturated fatty acid, monounsaturated fatty acid, polyunsaturated fatty acid n−6, polyunsaturated fatty acid n−3, and total polyunsaturated fatty acid was higher in O’Higgins station individuals than those from Yelcho and Rothera station (Table 1).

PERMANOVA analysis, which compares the complete fatty acid profile, showed significant statistical differences among locations (Pseudo-F_{2, 55} = 206.68; p < 0.001; 9999 permutations; Table S3). Those differences also displayed a clear separation in the spatial distribution among the three populations in the PCA plot (Fig. 5). This is consistent with the SIMPER analysis, since the contribution to the dissimilarity was driven by different fatty acids for each comparison between localities (Table 2).

Limitations and future directions

The absence of information on gonadal maturation or development of the analyzed individuals is considered an important limitation in this study. Our samples were collected in summer, temporally distinct from the spawning season described for A. eightsii in the southern WAP as during winter (Lau et al., 2018). This suggests gonad maturation would form a minor (if any) component of the variation between locations, especially since we are comparing a quite narrow biogeographic range. However, it is necessary to take into account that there could be spatial variation of the reproductive period in this species at different locations in the WAP. In this context, continuous reproductive analysis (i.e., gonadal maturation) using A. eightsii at a number of sites along the WAP environmental gradient should be conducted in future studies, since there is a generalized lack of information on this topic within Antarctic marine invertebrates.

Ideally, future studies should also consider the collection of environmental parameters (e.g., seawater temperature, salinity, etc.) in order to evaluate any potential relationship between biological and environmental data. There are few research centers along the WAP with the capacity to obtain long-term environmental data (e.g., Carlini, Palmer, Rothera). Unfortunately, in the case of the Chilean bases O’Higgins and Yelcho, there are no oceanographic monitoring programs and data could not be taken by other means. For this reason, it was not possible in our study to include environmental data to provide an overall picture at the three study sites. In this context, we emphasize the urgent need to obtain long-term oceanographic data in the northern WAP. In this way, a more representative monitoring of the effect of regional warming on the WAP should improve our understanding of the impacts of climate change on the biology of Antarctic marine invertebrates.

Another consideration is that Yelcho samples were collected eight months earlier than O’Higgins and Rothera samples due to logistical difficulties related to working in isolated and strongly seasonal ecosystems with limited access. This region may experience significant interannual variability, driven by the Southern Annular Mode (SAM) and El Niño-Southern Oscillation (ENSO) (Martinson et al., 2008; Santamaría-del-Ángel et al., 2021), which can translate into biotic variability. In this context, the oceanographic variables such as temperature and salinity at the southern area of Anvers Island (where Yelcho is located) indicated only limited (but significant) interannual variation between the summer seasons of 2017 and 2018 (Fig. S1). On the contrary, chlorophyll-a did not display significant differences between summer seasons of 2017 and 2018 (Fig. S1). Those differences, especially in temperature between the 2 years at Yelcho, while not being lethal to adults, could influence metabolism (e.g., Davenport, 1988b) and therefore the balance between energy gains and costs, modifying energy storage and growth (e.g., Morley et al., 2016). Furthermore, temperature can alter the composition of phytoplankton communities (Schofield et al., 2017) and the nutritional properties of the organic matter stored in the sediment (e.g., Malinverno & Martinez, 2015), causing a change in the type of food available for benthic species. In turn, this limitation also makes it difficult to relate biological aspects to environmental variability, given the lack of information on precise gonadal cycle of Antarctic species. Therefore, differences found in our study might not only be driven by spatial variability, but also by a mixed spatio-temporal variability that should be carefully considered in futures studies.

A final limitation is that we did not analyze glycogen content, even though it is an important body component of bivalves. Glycogen is used mainly as an energy source for oocyte production within the gonads (Mathieu & Lubet, 1993). Thus, by analyzing this component, we would have had insights into the stage of gonadal maturation (e.g., mature or immature stage) of individuals. In this context, despite the fact that proteins, lipids and fatty acids are also an important part of the biochemical composition of organisms, we suggest that related future studies prioritize the evaluation of glycogen content and its relationship with the reproductive cycle of A. eightisii.

In spite of limitations mentioned above, the biochemical and energetic results shown here are within previously published ranges for Antarctic marine invertebrates (Heine et al., 1991; McClintock et al., 1991; McClintock et al., 1992). There is also agreement with the predominance of protein content above lipid content, which in our case was almost three times as much protein (13.11–22.34% DM) as lipid (4.60–8.30% DM). Furthermore, differences were found between the O’Higgins and Rothera samples even though they were captured on exactly the same date. Only the Yelcho data should be interpreted with caution due to the difference in the date of collection, which could potentially be affected by interannual environmental differences. Therefore, we suggest that our study represents a valuable first step, highlighting the importance of evaluating the relationship between physiological and regional oceanographic processes, influencing the nutritional condition of benthic marine invertebrates along the WAP. This will add spatial context to high resolution temporal sampling that is currently undertaken at Rothera (Lau et al., 2018). Additional testing with other taxa and a more comprehensive spatial distribution of study sites can evaluate whether A. eightsii proves to be a good example of how biochemistry of Antarctic marine invertebrates responds to changes in environmental conditions.