Macroevolutionary patterns of defense SPECIAL FEATURE and pollination in Dalechampia vines: Adaptation, exaptation, and evolutionary novelty W. Scott Armbrustera,b,c,1, Joongku Leed,2, and Bruce G. Baldwind aSchool of Biological Sciences, University of Portsmouth, Portsmouth PO1 2DY, United Kingdom; bInstitute of Arctic Biology, University of Alaska, Fairbanks, AK 99775-7000; cDepartment of Biology, Norwegian University of Science and Technology, NO-7491 Trondheim, Norway; and dDepartment of Integrative Biology and Jepson Herbarium, University of California, Berkeley, CA 94720-2465 Edited by Anurag A. Agrawal, Cornell University, Ithaca, NY, and accepted by the Editorial Board August 29, 2009 (received for review June 25, 2009) We conducted phylogenetically informed comparative analyses of One advantage of taking a long-term approach to evolutionary 81 taxa of Dalechampia (Euphorbiaceae) vines and shrubs to assess analysis is that it sometimes permits detection of the causes of the roles of historical contingency and trait interaction in the evolutionary novelty, such as invasion of new adaptive zones (9) evolution of plant-defense and pollinator-attraction systems. We or escape from enemies through novel defenses, which can lead asked whether defenses can originate by exaptation from preex- to subsequent evolutionary radiation (10, 11). An explicitly isting pollinator attractants, or vice versa, whether plant defenses historical approach also allows evaluation of the role of historical show escalation, and if so, whether by enhancing one line of contingency in evolutionary change (12)—for example, phylo- defense or by adding new lines of defense. Two major patterns genetic lag, genetic constraint, and preaptations (13–15). Phy- emerged: (i) correlated evolution of several complementary lines of logeny-based approaches allow tests of associations between defense of flowers, seeds, and leaves, and (ii) 5 to 6 losses of the traits or partnerships (16, 17) and whether particular traits or resin reward, followed by redeployment of resin for defense of relationships influence the evolution of others (18). male flowers in 3 to 4 lineages, apparently in response to herbi- Historical contingency is implicit in Ehrlich and Raven’s vore-mediated selection for defense of staminate flowers upon escape-and-radiate hypothesis of plant–herbivore coevolution EVOLUTION relaxation of pollinator-mediated selection on resin. In all cases, (10) and defensive escalation (11). Other historically contingent redeployment of resin involved reversion to the inferred ancestral evolutionary scenarios include consistent sequences of trait arrangement of flowers and resiniferous bractlets. Triterpene resin change (‘‘ordered change’’) (18); exaptation (13) (e.g., coopting has also been deployed for defense of leaves and developing preexisting compounds for new defense or reward functions); seeds. Other unique defenses against florivores include nocturnal and evolutionary ‘‘novelty’’ through regulatory gene–based trait closure of large involucral bracts around receptive flowers and reversals (‘‘atavisms’’) (19–21). Previous macroevolutionary permanent closure around developing fruits (until opening again studies of plant–herbivore interactions have shown patterns upon dehiscence). Escalation in one major clade occurred through of escalation and decline in the intensity and effectiveness of an early dramatic increase in the number of lines of defense and in different defense systems (22–24) and specific sequences of the other major clade by more limited increases throughout the evolutionary change in plant defense systems (25, 26). group’s evolution. We conclude that preaptations played impor- Macroevolutionary studies of the interactions between plant tant roles in the evolution of unique defense and attraction defense and attraction systems are few, although the importance systems, and that the evolution of interactions with herbivores can of this link has long been recognized. In considering pollination be influenced by adaptations for pollination, and vice versa. of primitive angiosperms, Pellmyr and Thien (27) hypothesized that the secretion of essential oils by flowers originated as floral resin ͉ florivory ͉ plant defense ͉ resin defense defense in response to selection generated by herbivores and/or pathogens. These mostly toxic compounds now play roles in t is accepted that most plant species experience a variety of advertisement and attract pollinators. The origin of floral ad- Iantagonistic and mutualistic relationships with animals simul- vertisements by exaptation [sensu Gould and Vrba (13) and taneously or sequentially over the course of their lives. To date, Arnold (28)] has been invoked as a key innovation in angiosperm however, most research has focused on only 1 type of interaction evolution (27). It seems likely that many of these compounds at a time (e.g., herbivory ignoring pollination or pollination today play dual roles in attraction and defense [see also Lev- ignoring herbivory) (1). Less frequently considered are the Yadun (29)], that is, are ‘‘addition exaptations,’’ whereby a new interactions between these partnerships (e.g., how pollination function is added to, rather than replaces, the prior function (13, and herbivory might interact evolutionarily). 28). Later studies based on phylogenetic approaches have also Interactive effects of herbivory and pollination on plant suggested that protection or defense functions often precede the reproductive success have been detected in a number of micro- attractive functions of biosynthetic products (4, 7, 30). For evolutionary studies; these reveal surprisingly strong effects and example, previous work suggested that the resin-reward system complex, sometime counterintuitive, responses (2, 3). For ex- seen in Dalechampia vines and shrubs (Euphorbiaceae) and ample, the evolution of flowers may be influenced by selection generated as much by nonpollinating agents as by pollinators Author contributions: W.S.A. designed research; W.S.A., J.L., and B.G.B. performed re- (4–6; but see ref. 7), in contrast to traditional expectations that search; W.S.A. and B.G.B. analyzed data; and W.S.A. and B.G.B. wrote the paper. pollinators alone drive floral evolution (8). Research on inter- The authors declare no conflict of interest. actions between various plant–animal relationships has naturally This article is a PNAS Direct Submission. A.A.A. is a guest editor invited by the Editorial focused on ecology (e.g., the immediate growth and/or repro- Board. ductive outcomes of complex interactions). Although an evolu- 1To whom correspondence should be addressed. E-mail: [email protected]. tionary perspective underlies these studies, few studies have 2Present address: Korea Research Institute of Bioscience and Biotechnology, #52 Eoeun- explicitly considered the long-term evolutionary dynamics of dong, Yuseong-gu, Daejeon 305–333, South Korea. plant interactions with multiple partners (but see articles cited This article contains supporting information online at www.pnas.org/cgi/content/full/ below). 0907051106/DCSupplemental. www.pnas.org͞cgi͞doi͞10.1073͞pnas.0907051106 PNAS ͉ October 27, 2009 ͉ vol. 106 ͉ no. 43 ͉ 18085–18090 Downloaded by guest on September 27, 2021 upon by specialist butterfly larvae, Dynamine spp. in the neo- tropics and Neptidopsis in the paleotropics, as well as generalists, such as tettigoniid grasshoppers/katydids (Orthoptera: Tettigo- niidae), especially at night (36). Rewards produced for pollina- tors include pollen (34), oxygenated triterpenes secreted by a condensed resin gland associated with the staminate subinflo- rescence (31, 32), and monoterpene fragrances secreted by either the stigmatic surface of the pistillate flowers (37) or a gland homologous with the resin gland (38, 39). Results Bayesian posterior probabilities and maximum-parsimony boot- strap values from phylogenetic analyses of chloroplast DNA (cpDNA), internal transcribed spacer (ITS), and external tran- Fig. 1. Blossom inflorescences (pseudanthia) in flower and fruit. (A) Recep- scribed spacer (ETS) sequence data provided strong support for tive pseudanthium of D. stipulacea, a species that is pollinated by resin- numerous species groups and 2 major clades: species with 4 collecting euglossine bees. Pseudanthia have resin glands (yellow arrow) branches in the male subinflorescence (‘‘4-armed clade’’) and formed by asymmetrical clusters of resiniferous staminate bractlets. The floral species with 5 branches (‘‘5-armed clade’’; Fig. 2 and supporting resin gland secretes a mixture of oxygenated triterpene ketones and alcohols information Fig. S1). Maximum-likelihood optimization of traits (32). This species and its relatives also secrete the same oxygenated triterpene alcohols from capitate glands along the margins of stipules (green arrow), onto the ultrametric Bayesian trees resampled from 22,500 leaves, and involucral bracts (white arrow). (B) Capsular fruits of D. scandens, retained ITS trees of the posterior distribution indicated 1 to 2 showing capitate glands (white arrow) on pistillate sepals, which secrete origins of stinging crystalliferous trichomes on vegetative parts oxygenated-triterpene resin. Note also the involucral bracts, which are par- (depending on whether Tragia and Dalechampia share this trait tially closed around the fruit. As the fruits mature these bracts begin to open by common descent; see Fig. 2, first line of defense). Optimi- (as here) in preparation for explosive dispersal of the