Figure 1.
Major Southern Ocean habitats and related distribution of diatom species.
The seasonal cycle of sea ice formation and decay produces three major habitats: sea ice, meltwater and pelagic. Sea ice brine channels contain cold, highly saline water within which extremophile microalgae, such as F. cylindrus, are able to flourish (visible as brown coloured regions within the sea ice). The melting sea ice produces a stable layer of cool, low salinity water to which species such as P. subcurvata have adapted. Once the stratified meltwater layer has broken down, deeper mixing returns and the pelagic habitat prevails. C. simplex is most successful in the pelagic and meltwater habitats, but also manages to survive the winter months within the sea ice. F. cylindrus is abundant in all three habitats but frequently dominates the sea ice microalgal community.
Figure 2.
Average second derivative spectra.
The average second derivative spectra of F. cylindrus (a), C. simplex (b) and P. subcurvata (c) from meltwater (blue), pelagic (pink) and sea ice (black) treated cells. Labels indicate the macromolecular pool represented by peaks in each spectrum. Standard error ranged from 9.16 × 10-7 to 1.76 × 10-4 absorbance units for the pooled dataset.
Figure 3.
Partial Least Squares Discriminant Analysis (PLSDA).
PLSDA modelling was used to classify samples by meltwater (●), pelagic (▲) and sea ice (■) treatment conditions based on their infrared spectra. Scores plots for F. cylindrus and C. simplex (a & d) clearly show three clusters of cell spectra, indicating that the macromolecular composition was distinctly different between treatments, whereas P. subcurvata showed minimal plasticity (g). Because the data were second derivative transformed, a positive peak in the loadings plot indicates a decrease in absorbance for points that have positive scores in the PLSDA scores plots.
Figure 4.
Cellular concentrations of macromolecules.
Concentration is proportional to the area under the peak in the infrared spectrum corresponding to each macromolecule, according to the Beer-Lambert Law proportionality between concentration and absorbance. For all parameters excluding protein, bars show the mean peak area. Protein is expressed as the percentage of total carbon contributed by protein, determined by mass spectrometry for a subset of C. simplex samples (pelagic and sea ice treated samples). This mass spectrometry data was used as a Partial Least Squares Regression (PLSR) calibration data set (499 spectra) to predict protein content in the remaining samples based on their FTIR spectrum (1328 spectra). Key: meltwater (white), pelagic (light grey) and sea ice (dark grey) treatments. Error bars indicate the standard error. Results from Scheirer-Ray-Hare two-way non-parametric statistical tests are shown in the top right corner of each plot.