Fig. 1: The geographical and environmental context of our metabarcoding samples.

A map of sampling locations is shown in panel (a), where the pink line separates samples collected on the Western Antarctic Shelf and those collected in the Powell Basin. The samples were attributed to water masses based on a salinity vs. temperature diagram (panel c, surface water freezing temperatures are shown with the magenta line), and five water masses sampled by us are color-coded according to the legend in the lower right corner. Sampling depths and the dependence of oxygen concentration on depth are shown in panel (b). Tight non-linear relationship of oxygen concentration and salinity is illustrated in panel (d). Distributions of these four key abiotic variables (depth, oxygen concentration, salinity, and temperature) across the five water masses sampled are shown in panel (e) in the form of boxplots, along with results of relevant dispersion analyses showing if water masses are significantly different according to a certain variable (multiple ANOVA in the case of oxygen concentration and the non-parametric Kruskal-Wallis test followed by post-hoc Dunn’s test in the case of depth, salinity, and temperature). The statistical tests were applied to sample sets of the following sizes: 6 samples (Surface Water), 30 samples (Shelf Water), 20 samples (Modified Warm Deep Water), 24 samples (Warm Deep Water), 30 samples (Weddell Sea Deep Water). Ecological boundaries inferred for pico-nano eukaryotic communities along the depth, oxygen concentration, salinity, and temperature gradients are shown with vertical and horizontal gray lines in panels (b), (c), (d). The source data underlying this figure see in Suppl. Data 1.