A tangle of forms and phylogeny: Extensive morphological homoplasy and molecular clock heterogeneity in Bonnetina and related tarantulas.

Tarantula spider systematics has long been considered problematic. Species diagnosis and phylogenetic hypotheses have historically relied on morphological features, which are known to be relatively conserved and/or highly homoplastic across the family. Morphology-based attempts to clarify the phylogeny of the highly diverse New World Theraphosinae, have only been moderately successful, and the time-frame of tarantulas' evolution is nearly terra incognita. Here we present a molecular phylogenetic analysis of the Theraphosinae genus Bonnetina and related lineages, employing one mitochondrial (COI) and five nuclear (ITS1, EF1G, MID1IP1, MRPL44, and I3568) loci. We also perform ancestral state reconstruction of a newly formulated morphological data matrix. Our analysis includes 47 species placed in 17 genera and other undetermined lineages. We obtained well resolved and supported topologies. COI and EF1G substitution rates were much lower than the values generally accepted for mygalomorph evolution, with substantial rate heterogeneity among lineages. The origin of Theraphosinae was dated during the Late Cretaceous, followed by rapid diversification into the three recently proposed Theraphosinae tribes. North and Central American Hapalopini (including Bonnetina) form a monophyletic group that likely originated during the Oligocene to a dispersing ancestor from the then isolated South America. A clade that includes all but one Bonnetina species is estimated to have originated in the early Miocene and is the sister group of two morphologically divergent undescribed species. Morphological homoplasy is extensive across the tree. The two features that diagnose Bonnetina are homoplastic, but in combination still define the genus. Finally, we establish three groups of species within Bonnetina. Our results challenge the reliability of morphological characters for phylogenetic reconstruction in Theraphosinae, and indicate caution when interpreting Theraphosidae supra-specific classification in absence of a solid phylogenetic framework. They also question the dependability of universal substitution rates of COI and EF1G to calibrate phylogenetic analyses across Mygalomorphae.