Biological challenges to conclusions from molecular phylogenies: behaviour strongly favours orb web monophyly, contradicting molecular analyses.

This first-ever extensive review of the construction behaviour of orb webs, of webs secondarily derived from orbs, and of non-orbs shows that the evidence favouring monophyly over convergent evolution of orbs is stronger than previously appreciated. The two major orb-weaving groups, Uloboridae and Araneoidea, share 31 construction behaviour traits, 20 of which are likely to be both derived and to have feasible alternatives, making convergence an unlikely explanation. Convergence in two lineages seems unlikely, and convergence in five different lineages, as proposed in some recent molecular studies of phylogeny, is even less credible. A further set of seven shared responses in orb design to experimentally constrained spaces also supports orb monophyly. Finally, a 'control' case of confirmed convergence on similar 'pseudo-orbs' in a taxonomically distant group also supports this argument, as it shows a low frequency of behavioural similarities. I argue that the omission of behavioural data from recent molecular studies of orb web evolution represents a failure of the analytic techniques, not the data, and increases the risk of making mistakes. In general, phylogenetic studies that aim to understand the evolution of particular phenotypes can benefit from including careful study of the phenotypes themselves. [ABSTRACT FROM AUTHOR]