Research reviewUsing phylogenetics to detectpollinator-mediated floral evolutionAuthor for correspondence:Stacey DeWitt SmithTel: +1 919 684 3378Email: [email protected]: 28 February 2010Accepted: 6 April 2010Stacey DeWitt SmithDepartment of Biology, Duke University, Durham, NC 27708, USANew Phytologist (2010) 188: 354–363doi: 10.1111/j.1469-8137.2010.03292.xKey words: ancestral state reconstruction,diversification, floral evolution, phylogeneticgeneralized least squares, phylogenetics,pollination system, stochastic mapping,transition rate.SummaryThe development of comparative phylogenetic methods has provided a powerfultoolkit for addressing adaptive hypotheses, and researchers have begun to applythese methods to test the role of pollinators in floral evolution and diversification.One approach is to reconstruct the history of both floral traits and pollination sys-tems to determine if floral trait change is spurred by shifts in pollinators. Lookingacross multiple shifts, it is also possible to test for significant correlations betweenfloral evolution and pollinators using parsimony, likelihood and Bayesian methodsfor discrete characters or using statistical comparative methods for continuouscharacters. Evolutionary shifts in pollinators and floral traits may cause changes indiversification rates, and new methods are available for simultaneously studyingcharacter evolution and diversification rates. Relatively few studies have yet usedformal comparative methods to elucidate how pollinators affect floral evolutionacross the phylogeny, and fruitful directions for future applications are discussed.IntroductionComparative analysis has a long history in the study ofplant–pollinator interactions. By examining how floralfeatures serve to attract pollinators and enhance reproduc-tive success in different species of angiosperms, pollinationbiologists have sought to identify floral traits that are adap-tations for particular modes of pollination. For example,Darwin (1877) examined how the shape, size and orienta-tion of floral parts varies among orchid species in relation tothe type of pollinators that visit them. He proposed thatthese differences across species, acquired since divergencefrom the common ancestor of orchids, are adaptations topromote cross-pollination given each species’ ‘conditions oflife’. Looking more broadly across angiosperms, pollinationbiologists have noted combinations of floral features (‘syn-dromes’) that have arisen independently in distantly relatedlineages in association with particular modes of pollination,consistent with the idea that these syndromes reflect floraladaptation (Stebbins, 1970; Fenster et al., 2004). Thesecomparative studies of floral morphology and pollinationsystems across lineages have provided the foundation forour understanding of the ecology and evolution of plant–animal interactions and have inspired numerous studiesdetailing the mechanisms by which pollinators have acted asagents of selection (e.g. Cresswell & Galen, 1991;Campbell et al. , 1996; Mele´ndez-Ackerman et al., 1997).Despite the rich history of comparing floral featuresacross species in the context of their pollination systems, theapplication of formal phylogenetic approaches to the studyof pollinator-mediated floral evolution has only just begun.Modern phylogenetic methods make it possible not only toaddress basic questions such as how many times a particu lartrait has arisen but also to test broader questions abou t howfloral traits respond to pollinator shifts and how such evolu-tionary transitions affect the fates of lineages. To date,NewPhytologistReview354 New Phytologist (2010) 188: 354–363www.newphytologist.com! The Author (2010)Journal compilation ! New Phytologist Trust (2010)however, the vast majority of phylogenetic studies of floralradiations have used comparative analyses only to mapchanges of traits on trees, and often these reconstructionshave been limited to floral traits or pollination modes alone.Studies that extend comparative methods to estimat ingevolutionary correlations among traits and pollinators ortesting for associations between trait evolution and diversifi-cation rates remain rare. This gap betwee n what is possibleand what has been done may reflect the fact that manystatistical comparative methods are relatively new(e.g. Pagel, 1994; Martins & Hansen, 1997; Maddisonet al., 2007) and also that there are few groups in whichthe floral trait variation, the pollination ecology and thephylogeny have been well studied.This review aims to outline how phylogenetic methodsand particularly statistical comparative methods can be usedto test the role of pollinator shifts in floral evolution.Although we are increasingly discov ering that floral traitsare shaped by a multitude of biotic and abiotic forces(McCall & Irwin, 2006; Strauss & Whittall, 2006; Kessleret al., 2008; Armbruster et al., 2009), the central roleplayed by pollinators seems undeniable (Schemske &Bradshaw, 1999; Fenster et al., 2004). I will first revi ewhow the history of floral traits and pollination systems canbe reconstructed to ask whether the order of evolutionaryevents is consistent with floral adaptation in response topollinator shifts. Next, I consider how one can test forcorrelated evolution by looking across multiple evolutionarytransitions. In addition, I will briefly consider the impact oftransitions in floral traits and pollination systems on thediversification of lineages, and how differential diversifica-tion might affect tests of pollinator-mediated floralevolution. Throughout the review, I focus on examplesfrom the literature that employ independently estimatedphylogenies (not based on floral morphology) and thatcharacterize pollination systems using field studies or obser-vations (as opposed to inference from floral morphology).This approach minimizes the potential for circularity andallows for robust tests of the relationship betweenpollinators and floral trait evolution across the tree.Although my goal here is to examine how pollinatorshave shaped the evolution of flowers, the methods andapproaches I describe could be used to test the importanceof any putative selective factor acting on any trait.Reconstructing the history of trait evolutionPhylogenetic methods for ancestral state reconstruction canbe applied to any trait that is heritable. Although geneticmaterial may be the first character that jumps to mind,phylogenetic analysis has been