Fig. 5

Mapping of diet contribution onto L. Tanganyika network. a Principal components analysis (PCA) of centered log-ratio (clr) transformed OTU counts for the Lake Tanganyika network. The PCA biplot shows both samples (i.e. individual fish, colored by diet), and OTUs (grey circles). Sample OTUs are grouped by cichlid species, as indicated by spider plots (lines connecting group samples to group centroid). The OTU coordinates in the PCA were used as a template to produce the network layout shown on panels b and c, as explained in the Methods. b Tanganyika network showing median node abundances for carnivorous (C), omnivorous (O) and herbivorous (H) diet. Node size is proportional to the median OTU abundance (after clr transformation) across individual specimens within each diet. Filled circles correspond to nodes with abundance above the geometric mean, whereas hollow circles indicate nodes with abundance below the geometric mean. c Tanganyika network showing node effect sizes for t-test pairwise comparisons between diets (estimated with ALDEx2). Nodes that significantly discriminated between diets (p < 0.05) are colored to phylum level, as in b. Hollow circles indicate nodes that tend to be relatively more abundant in the left-hand diet of the contrasted pair (e.g. C in C vs H), whereas filled circles show nodes that tend to be more abundant in the right-hand diet (e.g. H in C vs. H). Grey nodes were not significant