4 | DISCUSSION
The convergent evolution of phenotype in correlation with behavioral niche is clearly a pervasive trend in the evolution of mygalomorph spiders. Their adaptive landscape is simple and constrained at two extremes: at one end are opportunistic taxa that inhabit existing spaces and construct capture webs, and at the other are taxa that construct their own burrow or nest, and structurally modify the entrance, for example with a trapdoor (Fig. 2). A spectrum exists between these extremes, but most intermediate taxa still burrow, or show facultative burrowing habits, but do not structurally modify the entrance. Within these constraints, changes in the niche occupied have been common in the evolution of the infraorder, and have occurred in both directions (Fig. 1). For example, the general trend in both the Atypoidea and the Avicularioidea is that burrowing, trapdoor-building taxa have evolved from opportunistic, web-building ancestors, yet in (at least) the Venom Clade and the Nemesioidea, the opportunistic, web-building niche has evolved again, independently (Fig. 1). Adaptation to different optima in this narrow adaptive landscape is one of the primary forces shaping somatic morphology in the Mygalomorphae, and this trend is clear in both overall morphology (Fig. 2) and in those morphological features that are intuitively adaptive (Tables 1,2; Fig. 3). The historical use of these characters to infer phylogenetic relationships explains, at least in part, the conflict between traditional morphological hypotheses and new molecular ones. Indeed, it is now clear that the “Dipluridae”sensu lato and the previous higher classification “Rastelloidina” are both artificial groups lumping together taxa from either end of the mygalomorph adaptive landscape (Raven, 1985).