Interactions between a plant and its pollinators exert great influence on its morphological and physiological adaptations. To the extent that such adaptations are evolutionarily conservative, they can be traced on a phylogeny at least if traits can be unambiguously coded and species are reasonably densely represented in the phylogeny. Studies using historic reconstructions have inferred the evolution of long-tubed flowers adapted to pollinators with long mouthparts (Whittall and Hodges 2007), radial or bisymmetric flowers adapted to different pollinators (Knapp 2010), or perfume-producing flowers adapted to oil-collecting bees (Renner and Schaefer 2010). In orchids, it has also been possible to infer evolutionary shifts between different types of mimicry and deception (Vereecken et al. 2012) and between deceptive and rewarding pollination systems (Johnson et al. 2013). Such shifts between antagonistic and mutualistic pollination interactions are of interest because theory predicts that mutualistic interactions are vulnerable to cheaters (Bronstein 2001). Over evolutionary time, this is expected to lead to the repeated replacement of mutualistic interactions by interactions in which plants deceive their pollinators by attracting them to consistently rewardless flowers. Yet of the 415 families of flowerings plants, only 32 have evolved pollination modes that rely on deceit (Renner 2006). Quantitatively most important among these families are the Orchidaceae, in which thousands of species are thought to rely on deceptive pollination, often involving sexual deception. The first phylogenetic studies addressing evolutionary transitions between deceptive and rewarding pollination in orchids, however, revealed unexpected results. Thus, in the Orchidinae, the ancestral state seems to have been bee-pollinated, food-deceptive flowers, with sexual deception evolving later (Inda et al. 2012),