Figure 3. Global trends in lipid and primary metabolite data.

To visualize overall patterns in the metabolome of different experimental groups, we performed principal component analysis (PCA) first on all lipids and primary metabolites (A,B), then on individual categories thereof (C,D). Samples tend to cluster primarily by tissue of origin, in line with studies from mammals (Ma et al., 2015). Shading is by population (B) or feeding state (C). Categorical breakdown of primary metabolites (C) and lipids (D) reveals that cavefish tend to cluster closer to one another than to surface. (E) Legend for C and D.