Fig. 1: Homologous caste differentiation trajectories in two ant species.
From: Canalized gene expression during development mediates caste differentiation in ants
a, M. pharaonis developmental trajectories reconstructed from 568 individual transcriptomes covering all developmental stages and visualized with the Fruchterman–Reingold algorithm in an undirected network (Methods), so that individuals with similar transcriptomic profiles cluster together. Shading of connecting lines reflects the strength of mutual transcriptome correlations, ranging from 0.8 (light grey) to 1.0 (black). Symbol colours differentiate between embryos within eggs (white and blue), larvae (yellow), pupae (red) and adults (brown). Images courtesy of L.P.18. b, The first three PC axes for individual transcriptomes of A. echinatior and M. pharaonis, from first instar larvae to adults, constructed from 979 individual transcriptomes across the two ant species, using 7,838 one-to-one orthologous genes. Upper panel: probability density function (PDF) of PC1 values separating the two ant species. Lower panel: PC2 and PC3 jointly distinguish between developmental stages and caste phenotypes across individual transcriptomes, plotted separately for the two ant species. Lines connect the median PC values for each caste across development stages, showing that individuals follow very similar trajectories regardless of species identity. Symbol colours and shapes as in a. c, Between-species transcriptome similarities comparing M. pharaonis (ngynes = 171; nworkers = 154) and A. echinatior (ngynes = 139; nworkers = 319) (left) and M. pharaonis and D. melanogaster females (n = 123) (right), based on stage-specific Spearman correlation coefficients and plotted separately for gynes (red) and workers (blue). Between-species similarities peaked in third instar larvae where gynes and workers were similar but similarities were always lower in earlier and later stages, particularly for workers. The box plot shows the median (centre line), 25% and 75% quartiles (boxes), outermost values (whiskers) and data points (overlapping with boxes and whiskers).
