Fig. 5

Examining on the negative relationship between plasticity evolution and gene–gene interactions. Here we focus on genes of the whole transcriptomes, with plasticity persistence and plasticity evolution inferred as in Fig. 3B. We applied the WGCNA analyses [61] to the pooled samples of the two extant lowland populations acclimated to either low- or high-altitude common gardens, so to identify co-expression modules. A To examine whether plasticity evolution is associated with fewer gene–gene interactions compared to plasticity persistence, we evaluate the Spearman correlation between the module size and the proportions of genes with plasticity evolution in the modules. B To examine whether plasticity evolution is associated with weaker interactions than is plasticity persistence, we evaluate their difference in the intra-modular connectivity (k_IM; see the text for details) using the Kruskal–Wallis test (*** indicates P < 0.001). Compared to genes with persisting expression plasticity, those with evolved plasticity (A) interact with similar numbers of other genes while (B) showing weaker gene–gene interactions in the ancestral population