Evolution of floral morphology and pollination - Ohio University

2 downloads 142 Views 1MB Size Report
tion of specific shifts in floral morphologies allows us to estab- lish when significant .... of evolution of the overall flower morphology, and (3) the patterns of evolu-.
American Journal of Botany 97(5): 782–796. 2010.

EVOLUTION OF FLORAL MORPHOLOGY AND POLLINATION SYSTEM IN BIGNONIEAE (BIGNONIACEAE)1 Suzana Alcantara2 and Lúcia G. Lohmann Universidade de São Paulo, IB, Departamento de Botânica, Cidade Universitária, Rua do Matão 277, São Paulo, SP, CEP 05508-090, Brazil The radiation of angiosperms is associated with shifts among pollination modes that are thought to have driven the diversification of floral forms. However, the exact sequence of evolutionary events that led to such great diversity in floral traits is unknown for most plant groups. Here, we characterize the patterns of evolution of individual floral traits and overall floral morphologies in the tribe Bignonieae (Bignoniaceae). We identified 12 discrete traits that are associated with seven floral types previously described for the group and used a penalized likelihood tree of the tribe to reconstruct the ancestral states of those traits at all nodes of the phylogeny of Bignonieae. In addition, evolutionary correlations among traits were conducted using a maximum likelihood approach to test whether the evolution of individual floral traits followed the correlated patterns of evolution expected under the “pollination syndrome” concept. The ancestral Bignonieae flower presented an Anemopaegma-type morphology, which was followed by several parallel shifts in floral morphologies. Those shifts occurred through intermediate stages resulting in mixed floral morphologies as well as directly from the Anemopaegma-type morphology to other floral types. Positive and negative evolutionary correlations among traits fit patterns expected under the pollination syndrome perspective, suggesting that interactions between Bignonieae flowers and pollinators likely played important roles in the diversification of the group as a whole. Key words: Bignonieae; discrete traits; evolutionary correlations; evolutionary specialization; floral traits; morphological transitions; pollinator shifts.

Understanding how reproductive traits evolved in the past leads to important insights into how organisms adapted. In sexually reproducing organisms, traits associated with outcrossing are thought to be under strong selection because of their direct effect on reproductive success; the radiation of floral morphologies in angiosperms represents a great example of this (e.g., Darwin, 1871; Lloyd and Webb, 1992). Specifically in the case of animal-pollinated plant species, pollinator preference represents a strong selective pressure to flower traits (Darwin, 1862; Fægri and van der Pijl, 1966; Stebbins, 1970; Schemske and Bradshaw, 1999). Attraction of pollinators and successful pollen transfer represent the primary targets of selection during flower evolution, leading to repeated evolutionary shifts between pollinators and, consequently, to the diversification of floral forms (Darwin, 1862; Fægri and van der Pijl, 1966; Stebbins, 1970, 1974; Harder and Barrett, 2006). The association between particular pollinators and specific floral traits is thought to have led to the evolution of pollination syndromes, which correspond to suites of floral traits that are associated with the attraction of specific pollinators (Vogel, 1954; Fægri and van der Pijl, 1966; Stebbins, 1970; Fenster et al., 2004). In this context, homoplastic evolution of similar flower morphologies are thought to have resulted from strong selection pressures by similar pollinators (Fægri and van der Pijl, 1966). Despite this, differing opinions exist regarding the 1

applicability of the syndrome concept (see Herrera, 1996; Strauss and Whittall, 2006; Waser, 2006). Nonetheless, the recognition of pollinators as functional groups defined by their ecological similarities is generally considered a valuable concept for the study of flower specialization to pollinators (Fenster et al., 2004). Evolutionary specialization leads to shifts in floral morphologies that are associated with the use of a subset of pollinators compared to those visiting the ancestral morphology (Armbruster et al., 2000; Fenster et al., 2004). The identification of specific shifts in floral morphologies allows us to establish when significant evolutionary changes took place, as well as to test specific hypotheses associated with the processes that may have led to such morphological shifts (Grant and Grant, 1968; Armbruster and Webster, 1982; Armbruster et al., 1994; Johnson and Steiner, 1997; Hansen et al., 2000; Fenster et al., 2004 and references therein). Shifts among floral morphologies resulting from the selection exerted by specific pollinator groups can occur in three different ways (Armbruster, 1993): (1) gradual quantitative shifts that correspond to those shifts proposed by Darwin’s coevolutionary race model (1862); (2) gradual qualitative shifts with intermediate stages among flowers pollinated by different pollinators; and (3) qualitative shifts without intermediate morphologies (Stebbins, 1970, 1974; Armbruster, 1993). Despite the transient nature of the intermediate phases among floral morphologies, these morphologies may persist in plant populations when the frequencies of the effective pollinators fluctuate (Stebbins, 1970, 1974). This condition is thought to represent the rule, rather than the exception, for shifts between pollination modes (Stebbins, 1970). On the other hand, drastic variation in the frequencies of pollinators might lead to rapid changes in traits that determine pollinator specificity, leading to shifts without intermediate morphologies. Little evidence is available as to how floral morphologies have shifted over time, making it difficult to evaluate the direction

Manuscript received 25 June 2009; revision accepted 25 February 2010.

The authors thank B. Loeuille, M. Kaehler, R. Ree, S. Branco, R. Olmstead, S. Graham, and an anonymous reviewer for comments that greatly improved this manuscript. This paper is part of the thesis of S.A., which was supported by FAPESP (Grant 06/59916-0) and MBG (Elizabeth E. Bascom Fellowship). 2 Author for correspondence (e-mail: [email protected]) doi:10.3732/ajb.0900182

American Journal of Botany 97(5): 782–796, 2010; http://www.amjbot.org/ © 2010 Botanical Society of America

782

May 2010]

Alcantara and Lohmann—Evolution of floral morphology in Bignonieae

of evolution in individual floral traits and the number of shifts in pollination systems (Armbruster, 1992, 1993; Johnson et al., 1998; Armbruster and Baldwin, 1998; Wilson et al., 2004). Information on the evolution of individual floral traits might provide particularly useful insights into the processes that may have driven the evolution of flower morphologies as a whole. Specifically, a better understanding of the order of evolution and the lability of individual traits and of the patterns of correlated evolution among those traits could greatly enhance our understanding of the processes that may have led to current variation in floral morphologies (Pérez et al., 2006). Some studies have addressed the patterns of evolution of individual floral traits that are associated with attractiveness to animals (reviewed by Fenster et al., 2004; with subsequent studies by Ackermann and Weigend, 2006; Pérez et al., 2006, 2007, Whittall and Hodges, 2007; Okuyama et al., 2008; Smith et al., 2008; Tripp and Manos, 2008; Yamashiro et al., 2008). However, no study has ever evaluated the exact order of evolution of individual traits in a temporal context. In this study, we review the pollination systems of the tribe Bignonieae, a clade of neotropical Bignoniaceae and evaluate the evolution of floral traits in this group. Bignonieae includes approximately 380 widely distributed species, mostly lianas but with some shrubs (Lohmann, 2003, 2006). The ecological and morphological diversity of this clade makes it an appropriate model for studies on evolutionary and ecological diversification (Gentry, 1990; Lohmann, 2006). Specifically, their pollination strategies are thought to have played major roles in the diversification of the group, with changes in floral structure being associated with shifts in pollinator guilds and phenological restraints delimiting differential use of shared pollinators (Gentry, 1974b, 1990). Flowers of Bignonieae were previously classified into seven floral types (Gentry, 1974a; Fig. 1) that agree with currently recognized pollination syndromes (sensu Fægri and van der Pijl, 1966). Specifically, the Anemopaegma, Amphilophium, and Pithecoctenium floral types (Fig. 1) are pollinated by large and medium-sized solitary bees, the prevalent pollinators of Bignonieae (Gentry, 1974a; references in Appendix 2). More specifically, the Anemopaegma-type floral morphology corresponds to the classical syndrome of open-mouthed flowers pollinated by large to medium-sized bees, mainly euglossine bees and anthophorids (described by Fægri and van der Pijl, 1966), while Pithecoctenium-type flowers are described as xylocopid-specialized flowers, with thick texture and curved corollas limiting nectar robbing by hummingbirds and xylocopids (Gentry, 1974a). In addition, Amphilophium-type flowers represent an extreme specialization to avoid nectar robbing because the corolla lobes are closed at anthesis and can only be opened by strong, large-bodied pollinators, usually xylocopids (Gentry, 1974a). The Martinella-type flowers were originally described as representing the typical hummingbird-flowers, with red-orange to deep violet flowers, exserted stamens, and long tubular corollas (Gentry, 1974a). Hawkmoth pollination is thought to be associated with Tanaecium-type flowers (Gentry, 1974a), while the Cydista-type flowers are considered morphologically and phenologically specialized to a “mimetic” pollination mode by large to medium-sized bees. Cydista-type flowers lack a nectar disk, but have colorful corollas with conspicuous nectar guides (Gentry, 1974a). Butterflies and small bees putatively pollinate the small Tynanthus-type floweredspecies (Gentry, 1974a), which are considered to be generalists given that the small bee syndrome includes pollination by a broad spectrum of small insects.

783

Most floral types of Bignonieae are thought to represent modifications from the basic Anemopaegma-type, except the Amphilophium-type, which is thought to have evolved from Pithecoctenium-type flowers (Gentry, 1974a). In a previous study, the flower types defined by Gentry (1974a) were used to infer pollination syndromes that were mapped onto the phylogeny of Bignonieae as a single multistate trait, indicating several homoplastic origins for pollination systems in the group (Lohmann, 2003). Because the various traits that determine the individual flower types can show different levels of evolutionary correlation, a more detailed evaluation of the patterns of evolution of individual floral traits can contribute important information on the evolution of flower morphologies and pollination systems as a whole. In particular, an understanding of the evolutionary patterns of individual floral traits of Bignonieae allows for a test of the hypothesis that the ancestral flower of Bignonieae was a “specialized” flower, pollinated by a narrow group of pollinators (i.e., an Anemopaegma-type flower, as suggested by Gentry, 1974a), from which successive shifts in pollinators subsequently occurred. In this study, we specifically address (1) the patterns of evolution of individual floral traits, (2) the pattern of evolution of the overall flower morphology, and (3) the patterns of evolutionary correlation among pairs of individual floral traits. First, we used a likelihood approach to reconstruct the ancestral states of individual flower traits in Bignonieae. Second, we identified the overall floral morphology reconstructed at each ancestral node by compiling the ancestral states of each individual floral trait (as defined by Gentry, 1974a); these data were then used as a basis for mapping the inferred floral types as a single multistate character and to reconstruct the evolutionary pattern of the overall flower morphology in the group. The information gathered through ancestral-state reconstructions of individual floral traits and overall floral morphology permits a detailed evaluation of the sequence of evolution in reproductive traits of Bignonieae over time, allowing us to test specific predictions associated with the pollination syndrome concept. In particular, we tested whether shifts in floral morphologies occurred through intermediate phases (i.e., mixed morphologies). Last, we test for correlated patterns of evolution among individual floral traits. The analyses of evolutionary correlation allows us to evaluate whether transitions in particular traits are associated with transitions in other traits, as expected under the pollination syndrome hypothesis. MATERIALS AND METHODS Taxon and trait sampling—Taxon sampling was identical to that used in the combined molecular phylogeny of Lohmann (2006). Specifically, we included 104 Bignonieae species, representing 20 of the 21 genera recognized by Lohmann (in press); only the monotypic Callichlamys was not included in the analysis. Species were selected to include all the morphological diversity of the group. In particular, this sampling strategy included representatives of all major morphological shifts in the group (see Appendix S1 in the Supplemental Data with the online version of this article). Specific floral traits used are described in detail in Lohmann (2003) and Lohmann et al. (in press). Of the 95 discrete characters originally coded by Lohmann (2003), 12 floral characters were selected for the present study, seven of which were coded as binary and five as multistate (Appendix 1). These 12 discrete floral traits were chosen because of their association with Gentry’s floral types (Table 1; see Evolution of overall floral morphology for further details). Phylogenies—A well-supported phylogeny of Bignonieae based on plastid (ndhF) and nuclear (PepC) markers including 104 species of the tribe

784

American Journal of Botany

[Vol. 97

Fig. 1. Morphological types of Bignonieae as classified by Gentry (1974a). (A, B) Anemopaegma-type. (A) Anemopaegma chamberlaynii (Sims) Bureau & K. Schum., (B) Fridericia candicans (Rich.) L. G. Lohmann. (C–F) Martinella-type. (C) Dolichandra cynanchoides Cham., (D) Lundia cordata (Vell.) DC., (E) Adenocalymma dichilum A. H. Gentry, (F) Pyrostegia venusta (Ker Gawl.) Miers. (G) Amphilophium-type, Amphilophium paniculatum (L.) Kunth. (H) Pithecoctenium-type, Amphilophium crucigerum (L.) L. G. Lohmann. (I) Cydista-type, Bignonia corymbosa (Vent.) L. G. Lohmann. (J) Tynanthus-type, Tynanthus cognatus (Cham.) Miers. (K) Tanaecium-type, Tanaecium jaroba Sw. (Lohmann, 2006) reconstructed monophyletic groups that represent 21 genera in a new generic classification of Bignonieae (Lohmann, in press). Two fossils identified as Callichlamys (Chaney and Sanborn, 1933) and Paragonia/Arrabidaea—currently Tanaecium/Fridericia—(Graham, 1985) were placed at nodes 32 and 102 of the resulting phylogeny (see Appendix S2 in the online Supplemental Data), with estimated ages of 34.8 ± 0.22 Myr and 35.35 ± 1.65 Myr, respectively. Complementary analyses using these calibration points and a penalized likelihood (PL) approach (Sanderson, 2002) led to a maximum likelihood (ML) tree with branch lengths proportional to time (L. G. Lohmann; C. Bell [University of New Orleans], and R. C. Winkworth [South Pacific], unpublished results). Confidence intervals associated with the age of representative nodes were obtained through bootstrapping (Sanderson, 2002). Additionally, the five polytomies encountered in the PL tree were randomly resolved using the treefarm Package implemented in the program Mesquite 1.12 (Maddison and Maddison, 2006). To evaluate how alternative tree resolutions affected ancestral-state reconstructions, we created an additional set of 500 trees with polytomies randomly

resolved. The shortest possible branch lengths were also assigned to the six branch lengths created to resolve the five polytomies. These time-calibrated, fully resolved trees were used in all analyses. To assess the potential impact of branch lengths to the analyses, we modified branch lengths of the time-calibrated tree (L. G. Lohmann et al., unpublished data) in Mesquite 1.12 so that nodes reached the maximum or minimum confidence intervals associated with each node. That is, branch lengths were modified so that the minimum age estimated for each node would represent the youngest or oldest age indicated by the confidence intervals. For convenience, hereafter we refer to these trees as those presenting minimum possible branch lengths and maximum possible branch lengths, respectively. Overall, three sets of 500 trees were produced from the PL Bignonieae tree as follows: (1) trees including the original branch lengths assigned by the PL analyses, (2) trees with minimum possible branch lengths, and (3) trees with maximum possible branch lengths. Evolution of individual floral traits—Ancestral character state reconstructions were carried out in Mesquite 1.12 (Maddison and Maddison, 2006) using

May 2010]

Alcantara and Lohmann—Evolution of floral morphology in Bignonieae

785

Table 1.

Gentry’s floral types described for Bignonieae followed by individual traits associated with each floral type and their putative pollination syndrome; traits associated with each morphology were compiled from Gentry (1974a) and from published pollination studies (see Appendix 2). Double calyx

Corolla color

Nectar guides

Corolla shape

Corolla Corolla curvature texture

Amphilophium

Cupular Present

Magenta

Absent

Tubular

Straight

Coriac.

Anemopaegma

Cupular, Absent Magenta, Absent tubular yellow, white

Infund.

Straight

Memb., Rounded Not-bil. Opened Included rarely coriac.

Cydista

Cupular Absent

Magenta

Present

Infund.

Straight

Memb.

Martinella

Spath., Absent tubular, urceolate

1. Red, magenta 2. White

Absent

Tubular, infund., urceolate

Straight

Flattened Not-bil. Opened Included abaxially Coriac., Rounded Not-bil. Opened Exserted rarely memb.

Pithecoctenium Cupular Absent

Yellow, white White

Absent

Infund.

Curved

Coriac.

Rounded Not-bil. Opened Included

Cupular, Absent Absent tubular Cupular Absent Magenta, Present white

Tubular

Straight

Coriac.

Rounded Not-bil. Opened Exserted

Infund.

Straight

Memb.

Rounded Bilabiate Opened Exserted, rarely inserted

Floral types

Tanaecium Tynanthus

Calyx shape

Corolla tubes

Corolla Corolla lobes mouth*

Anther position

Rounded Bilabiate Closed

Included

Nectar disk

Pollination syndrome

Present 1. Large to medium-sized bees 2. Bats Present; 1. Large to absent in medium-sized Lundia bees 2. Small bees and insects Absent Large to mediumsized bees Present 1. Hummingbirds (long tube) 2. Bats (short tube, see text) Present Large to mediumsized bees Present Hawkmoths Absent, Small bees and reduced insects, butterflies

Notes: *, at anthesis; spath., spathaceous; infund., infundibuliform; coriac., coriaceous; memb., membranous; not-bil., not bilabiate. a ML approach. This methodology finds the ancestral states for each node that maximize the probability of the observed states in the terminal taxa under a stochastic model of evolution (Pagel, 1999). The assignment of individual ancestral states is made from the likelihood estimated for each character state according to a decision threshold value, with the states that present the lowest likelihood values being rejected (Pagel, 1999). This method allows the incorporation of additional parameters (i.e., phylogenetic branch lengths, varying rates of evolution and evolutionary models), which leads to increased accuracy on the ancestral state reconstructions (Pagel, 1997). The assignment of probabilities to all possible states at each node is particularly interesting for traits with a high number of transitions. All morphological characters were equally weighted and considered unordered in all analyses. For the binary traits, two evolutionary models were tested: the Markov chain (Mk) with 1 parameter and the Mk with 2 parameters, representing forward and backward rates of changes between character states. Models were chosen using the likelihood ratio test (LRT) following the recommendation of Posada and Buckley (2004). For multistate traits, reconstructions using the Mk model with 1 parameter were conducted. We used a decision threshold value 2.0 when ancestral states were assigned to a given node (Maddison and Maddison, 2006). These ancestral character state reconstructions were used to assess the rates of transition of floral traits and the shifts among floral morphologies in Bignonieae. To further assess the effect of alternative topologies and branch lengths to the reconstructed ancestral states, we used the three sets of trees obtained (i.e., dated trees, trees with minimum branch lengths possible, and trees with maximum branch lengths possible) as a basis for carrying out the ancestral-state reconstructions for the 12 characters studied. One tree was randomly chosen from all trees analyzed (Fig. 2). Results from this tree were compared to the results obtained from all other trees and the following information were recorded: (1) the percentage of trees that presented a particular node and (2) the percentage of trees in which the ancestral states were reconstructed unambiguously for a given node (see Case et al., 2008). These results were reported for the five nodes that were randomly resolved (one of these corresponding to a supra-generic node), for the 35 suprageneric nodes, and for the 20 infrageneric nodes associated with shifts in floral morphology, leading to a total of 59 nodes analyzed overall. Evolution of overall floral morphology—The original description of floral types in Bignoniaceae (Gentry, 1974a) was contrasted with the morphological matrix of Lohmann (2003) and Lohmann et al. (in press) to identify characters that were associated with the floral types described by Gentry (1974a). Twelve morphological characters fit these criteria and were used to classify the overall flower morphology of each species into one of the Gentry’s floral types or into a

“mixed” morphology type. Mixed morphologies were assigned to classify overall flower morphologies whenever flowers included traits that were associated with multiple floral types described by Gentry (1974a). To assess the evolution of the overall floral morphology, we compiled the ancestral states of the individual traits reconstructed at each node and used the same procedure described above to classify each ancestral flower into one of Gentry’s flower types (1974b) or into a mixed morphology. For binary traits, the assignment of floral types was solely based on the unambiguous reconstructions. For multistate traits, several characters were reconstructed ambiguously. However, in those cases, the probabilities assigned to each character allowed us to rule out particular character states that were improbable for a particular node; this information was used to establish the morphological types associated with particular nodes. A literature review indicated that the pollination system of 46 species of Bignonieae had been studied in the field (Appendix 2). For these 46 species, we were able to establish a direct association between Gentry’s floral types (1974b) and their respective pollinator groups. This information was then used to predict the most likely pollination mode for the remaining species for which the pollination system had not been studied in the field. Furthermore, these studies, provided sufficient information for us to categorize the individual floral morphologies into specialized or generalized flowers (sensu Fenster et al., 2004). Whenever a floral morphology was associated with a single functional guild of pollinator, that particular species was classified as a specialist. On the other hand, whenever a floral morphology was pollinated by two or more functional groups, the species was considered to be a generalist. Based on the association between floral types and the pollination systems described for the tribe (see Table 1), we were able to classify all 104 species of Bignonieae into specialists and generalists. Similarly, comparisons with current pollination systems were extrapolated to infer the most likely pollination mode for the ancestral flowers. To evaluate whether the evolution of floral morphologies included an intermediate phase (i.e., mixed morphology), we considered whether changes were concentrated within a single branch or whether transitions were spread out over multiple branches, resulting in branches with mixed floral types. Transitions according to the former category were considered as punctuated, while transitions according to the latter type were considered as sequential. To account for the potential effect of ambiguous reconstructions of some characters at some nodes, we considered all possible reconstructions while classifying transitions as punctuated and sequential. We then carried a Kolmogorov–Smirnov onesample test (appropriate for small samples) to test whether the number of punctuated and sequential shifts differed significantly (Zar, 1999). To assess whether the number of traits used to define the floral types had any impact on the assessment of the punctuated and sequential shifts, we tested whether punctuated shifts were caused by a higher number of traits with

786

American Journal of Botany

transitions being concentrated within a single branch of the phylogeny and whether sequential shifts (with mixed morphologies) occurred due to a sequential change of a lower number of traits at each branch. We carried out a t test for two samples with unequal variances (Zar, 1999) and estimated the probability values using 1000 replicates in a random resampling of the data (using the resampling probability estimates; Lowry, 2009). We used this same procedure to test whether there was any relationship between branch length and the occurrence of punctuated vs. sequential changes. Specifically, we tested whether punctuated shifts were associated with longer periods of time, which could have resulted from the accumulation of character transitions within a given lineage and/or from the extinction of lineages with intermediate stages given the increased time. Evolutionary correlation among floral traits—We tested for correlated evolution among floral traits using Pagel’s likelihood approach for discrete characters (1994) implemented in the Correl Package (Midford and Maddison, 2006) for Mesquite (Maddison and Maddison, 2006). Three traits originally coded as multistate by Lohmann (2003) (i.e., calyx shape, corolla shape, and corolla color) were recoded as binary, so that each state was transformed into a new trait coded as presence/absence. Double calyx and nectar disk were also recoded as binary, with the states “only reminiscent present” and “reduced” coded as absent. This coding scheme was necessary to assess possible correlations among gains and losses of each character state of the individual multistate characters because Mesquite does not allow evolutionary correlations to be performed between multistate characters. In total, we evaluated potential evolutionary correlations among 20 binary traits. For all 190 pairwise correlations tested, 100 searches were carried out, with the P value being estimated from 10 000 repeated simulations. We applied a Bonferroni adjustment (corrected alpha: 0.00027) to account for the use of multiple tests (Zar, 1999). Paired correlations between the presence/absence states of the multistate traits were evaluated to verify the directionality of changes among the states of those characters. Hypotheses of character correlations were accepted whenever a model with eight-parameters presented a better fit than a simpler model of evolution with four parameters (Midford and Maddison, 2006). The eightparameter model differed from the four-parameter model in the rates of transition among states of two binary traits. Under the eight-parameter model, asymmetric rates of transitions were associated with each state of a character, varying according to the state of the other trait that was being evaluated. Whenever an eight-parameter model presented a better fit than a four-parameter model, the asymmetric rates of transition associated with the character transition were associated with any directionality among the correlated traits.

RESULTS Evolution of individual floral traits— The ancestral condition of most traits analyzed was unambiguously reconstructed on the completely resolved, penalized likelihood tree used in the current study (data available upon request). Even for corolla color, the most homoplastic trait analyzed, the ancestral condition was unambiguously assigned to most nodes (Appendix S3 in the online Supplemental Data). The 12 binary traits considered presented a better fit to the Mk1 model of evolution (Appendix 1). Rates of character change on the time-calibrated phylogeny varied. For example, corolla curvature and double calyx showed lower rates of evolution than anther position in relation to corolla lobes, corolla shape, and corolla color (Appendix 1). The alternative resolutions of the five polytomies encountered on the penalized likelihood tree did not affect the ancestral state reconstructions (see online Appendix S4). Variations

[Vol. 97

in the estimation of branch lengths also had very little impact on the reconstruction of the ancestral character state reconstructions (Appendix S4). For example, nectar guides, corolla curvature, corolla texture, corolla tube, corolla lobes, corolla mouth at anthesis, double calyx, and nectar disk presented identical and unambiguously reconstructed ancestral states in all three sets of trees evaluated (data not shown). The ancestral state reconstruction of anther position in relation to corolla lobes was also not affected by branch length (Appendix S4). Corolla shape presented identical ancestral state reconstructions in two sets of trees: trees with the original branch lengths (estimated by the PL analysis) and trees with maximum branch lengths. In the trees with the minimum possible branch lengths, ambiguities were encountered in two nodes. Calyx shape had the root node unambiguously reconstructed as cupular in the analyses that considered the minimum and maximum branch lengths; however, a cupular/tubular condition was assigned in the set of trees with the original branch lengths. As far as corolla color is concerned, alternative branch lengths affected the ancestral state reconstructions of three of the 59 nodes considered. In two of those three ambiguously reconstructed nodes, the ambiguities present in the trees with the original branch lengths were unambiguously assigned in the tree with minimum branch lengths. The third node was unambiguously reconstructed in the trees with the original branch lengths and in the tree with maximum branch lengths; this same node was ambiguously reconstructed in the tree with minimum branch lengths (see Appendix S4). Evolution of the overall floral morphology in Bignonieae— The diagnostic combination of traits presented in Table 1 allowed us to classify species of Bignonieae as belonging to a specific floral type (e.g., online Appendix S1). Among the 104 species sampled in the phylogeny, the Anemopaegma-type flower represented the most common condition (53 species), followed by the Martinella-type flower (11 species); nine species were reported as presenting mixed floral morphologies (Fig. 2). In most cases, field studies on the pollination biology of Bignonieae (see Appendix 2) corroborated the pollination syndromes originally associated with floral types (Table 1). The only two exceptions were the recent description of visitation by bats in Amphilophium-type flowers (L. G. Lohmann, personal observation) and bat pollination in Adenocalymma dichilum, a species we classified as Martinella-type flower (Fig. 1). Despite this, all species studied that presented Amphilophium, Cydista, Martinella, Pithecoctenium, and Tanaecium flower morphologies were associated with a single pollinator group and were hence categorized as specialized. On the other hand, the Tynanthus-type morphology was associated with multiple pollinator groups and hence considered to represent a generalized morphology. The open-mouthed bee syndrome of the Anemopaegmatype was confirmed by most studies; however, a few studies also reported visitation by multiple pollinator groups in species with Anemopaegma-type flowers (Appendix 2). Generalist species ®

Fig. 2. The completely resolved penalized likelihood tree of Bignonieae used for the maximum likelihood (ML) ancestral state reconstructions, showing the evolution of floral morphologies in the tribe. Floral morphologies mapped in this phylogeny were inferred from the ancestral state reconstructions of the 12 discrete traits showed in Table 1. Branch lengths are proportional to time (Myr). Black branches marked with an asterisk (*) represent branches in which the reconstructed morphologies could not be assigned due to uncertainties in the reconstruction of individual character states (see Table 2). Representations of the extant species as well as the ancestral morphology of the tribe reconstructed by ML are shown (see Fig. 1 for further information on the species represented).

May 2010]

Alcantara and Lohmann—Evolution of floral morphology in Bignonieae

787

788

[Vol. 97

American Journal of Botany

usually presented smaller and magenta corollas, with inflorescences displaying a large number of flowers (common in Fridericia). On the other hand, species in which a specialized pollination by large to medium-sized bees were reported, usually presented larger and yellow corollas (common in Adenocalymma and Anemopaegma). Even though Anemopaegma-type flowers (as described by Gentry, 1974a) fit better a specialized pollination system, the wide variation in flower morphologies associated with the Anemopaegma-type flowers indicates that some Anemopaegma-type flowers might be generalists. Hence, both possibilities (generalists and specialists) were considered for this morphological type. Ancestral character state reconstructions indicated that the ancestral flower of Bignonieae presented an Anemopaegmatype flower (Fig. 2). This ancestral flower was likely magenta (53% probability); however, the exact condition was ambiguously reconstructed (magenta/yellow/white). Among the 12 individual floral traits associated with floral types (Table 1), seven did not present ambiguities in any of the 59 nodes considered (Table 2). Calyx shape, corolla color, nectar guides, corolla shape, and anther position in relation to corolla lobes led to some ambiguously reconstructed nodes. Such uncertainty makes the identification of the ancestral morphology of six internal nodes ambiguous; these nodes could be assigned as Anemopaegma-type flowers or other alternative type morphologies depending on the assigned state of the ambiguously reconstructed characters (Table 2). It was still possible to identify seven internal nodes that were not affected by ambiguous reconstructions and in which the most likely morphology was characterized by a mixed pattern (Table 2). Overall, we identified 29 shifts in floral morphologies, most of which were concentrated next to terminal branches in the phylogeny (Fig. 2). Generally, shifts from the basic Anemopaegma-type into one of Gentry’s flower types were encountered; less commonly, shifts into mixed floral traits, not associated with any of Gentry’s types, were encountered. The transition from an Anemopaegma-type to a Martinella-type flower was the most common, having occurred at least eight times. The repeated evolution of a Tanaecium-type flower from an ancestor presenting an Anemopaegma-type flower was also common. Amphilophium-type flowers, on the other hand, are always inferred to have evolved from ancestors with Pithecoctenium-type flowers, while Tynanthus-type flowers evolved a single time from an ancestor presenting an Anemopaegma-type flower and once from an ancestor with a Cydistatype flower. Among the 29 shifts encountered in floral morphologies, 10 showed transitions through mixed morphologies (sequential), while 19 showed a punctuated pattern (i.e., without intermediate stages). Even though the ancestral morphology of six of these 19 shifts were ambiguous, all most likely resolutions of these ambiguous assignments resulted in punctuated shifts (Table 2). Overall, the number of sequential and punctuated shifts did not differ (Kolmogorov–Smirnov test: Dmax = 0.155, D0,05 = 0.27), and no difference was detected between the length of branches associated with punctuated or sequential shifts in floral morphologies (t test = 0.953, df = 27, P = 1). Similarly, no difference was encountered between the numbers of traits involved in shifts with or without intermediate floral types (t test = 0.1567, df = 27, P = 0.88). Evolutionary correlations among floral traits in Bignonieae— Among the 190 pairwise correlation tests carried out

among the 20 selected Bignonieae traits, 74 were significant (38.95%; Table 3).Most correlations were negative, suggesting that particular flower traits never occur together within the same flower. For example, the negative correlation between yellow corolla and exserted position of the anther indicates that yellow corollas generally present inserted anthers. Similarly, a negative correlation between red and curved corollas indicates that red-flowered corollas generally present straight tubes. Nectar guides were also negatively correlated with tubular corollas, suggesting that whenever nectar guides are present, the state tubular corolla is not. On the other hand, strong positive correlations were observed between urceolate and red corollas, tubular and white corollas, presence of nectar guides and infundibuliform corollas, and presence of double-calyx and tubular corollas, indicating that these traits usually occur together. Other positive correlations were found between double-calyx and magenta corollas, closed corolla mouth at anthesis and magenta corollas, spathaceous calyx and red corollas, tubular and red corollas. Among the three characters originally coded as multistate by Lohmann (2003, in press) but recoded as presence/absence for the analyses of correlated evolution (i.e., calyx shape, corolla shape and color), negative correlations among the states of the same character were prevalent because a single species cannot present multiple states of the same character. For example, the corolla of a species cannot be tubular and urceolate simultaneously. Overall, the assigned rates of change between the states of these three traits and the reconstructed ancestral states (Appendix S5 in the online Supplemental Data) indicate the following transitions: (1) Calyx shape: cupular shape often precedes the evolution of all other calyx shapes, with tubular and spathaceous calyces always evolving from ancestors with cupular calyces (see Appendix S5-A). Furthermore, spathaceous and urceolate calyces are negatively correlated and occur in distantly related lineages, with tubular calyces having evolved multiple times. (2) Corolla shape: tubular and urceolate corollas most often evolved from ancestors with infundibuliform corollas (see Appendix S5-B). (3) Corolla color: yellow, white, and red corollas often evolved from ancestors with magenta flowers, with white and red corollas being negatively associated with yellow corollas (see Appendix S3). Moreover, the assigned rates of transition in other characters that are correlated with flower color are generally affected by the color of flowers and not the other way around. DISCUSSION The radiation of angiosperms is greatly associated with the diversification of flower forms, which is thought to have resulted from natural selection, leading to adaptation to different pollination strategies (Darwin, 1862; Fægri and van der Pijl, 1966; Stebbins, 1970). Despite the prevalent concept that shifts in floral morphology are driven by selective processes (e.g., Stebbins, 1970, 1974; Harder and Barrett, 2006), we still lack evidence of how floral morphologies have shifted over time (see introduction). Thus, we assessed in this paper the evolutionary pattern of (1) transitions in the individual floral traits, (2) shifts in the overall floral morphology, and (3) correlations between paired floral traits in the neotropical tribe Bignonieae (Bignoniaceae). We discuss below our major results attempting for an association between floral shifts and putative pollinator groups in this clade.

May 2010]

Alcantara and Lohmann—Evolution of floral morphology in Bignonieae

789

Table 2.

Ancestral states of traits that determine Gentry’s (1974a) morphologies in Bignonieae reconstructed using a maximum likelihood (ML) approach. Node numbers are indicated in Appendix S2 (see the online version of this manuscript) and correspond to the 35 suprageneric nodes and 24 infrageneric nodes considered here (see Material and Methods, Evolution of individual floral traits for further information). The six nodes created to resolve the five polytomies encountered in the penalized likelihood (PL) tree are indicated by (*).

Node 2 3 4 5 6* 10 26 29 30 31 32 33 34 35 38* 39 41 42 56 57 58 66 75 76 79 87 88 94 97 101 102 103 104 107 113 116 117 121 122 123 124 127 129* 140* 144* 150 154 160 163* 164 165 169 178 185 188 195 197 202 205

Calyx shape

Double calyx

Absent Absent Absent Absent Absent Absent Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Present Cupular Present Cupular Present Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Tubular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Absent Tubular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Cupular Absent Tubular Absent Urceolate Absent Cupular Cupular Cupular Cupular Cupular

Corolla color M/Y Y Y Y Y M/Y M/Y M/Y M/Y M M M M M M M M M/Y Y M M M M M M R/M/Y R/Y M/Y M/Y M/W M/W M/W M/W M/W M/W M/W M/W M/W M/W M/W M M R/M M/W M M/W M/W M M W W W M/Y Y M M

Nectar guides

Corolla shape

Corolla curvature

Corolla texture

Corolla tubes

Corolla lobes

Corolla mouth*

Anther position

Nectar disk

Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent

Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Tubular Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund.

Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Curved Curved Curved Curved Curved Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight Straight

Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Coriac. Coriac. Coriac. Coriac. Coriac. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Coriac. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb. Memb.

Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Flattened Flattened Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded Rounded

Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Bilabiate Bilabiate Bilabiate Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Bilabiate Not-bil. Not-bil. Not-bil. Not-bil. Not-bil. Not-bil.

Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Closed Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened Opened

Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included

Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Absent Absent Absent Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Present Reduced Absent Absent Present Present Present Present

Present Present Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Absent Present Present Absent Absent Absent Absent

Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Infund. Tubular Infund. Infund.

Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Included Exserted Included Included Included Included Exserted Included

Floral type Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemop./Martinella Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemopaegma Pithecoctenium Pithecoctenium Pithecoc.-Amphilop. Pithecoc.-Amphilop. Amphilophium Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemopaegma Cydista Cydista Anemop.-Cydista Anemop./Martinella Anemop.-Martinella Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemop./Tanaecium Anemop./Tanaecium Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemop.-Martinella Anemop./Tanaecium Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemopaegma Tynanthus Anemopaegma Anemopaegma Anemop./Tan./Martin Anemop.-Tan.-Martin Martinella Anemop.-Martinella

Note: M, magenta; R, red; Y, yellow; W, white; infund., infundibuliforme; memb., membranous; coriac., coriaceous; not-bil., not bilabiate; blank cells correspond to ambiguous character state reconstructions by ML. Despite the ambiguity for the reconstructions of the character corolla color, states are reported whenever only two states were assigned to a given node. A slash (/) between floral types means that different floral types could be assigned to the node depending on the alternative resolutions of one or more ambiguous reconstructed traits; a dash (-) means that morphologies were reconstructed with mixed character states.

790

[Vol. 97

American Journal of Botany

Table 3.

Evolutionary correlations between floral traits of Bignonieae. Above diagonal: differences among –logL of the four-parameter models (independent evolution) and eight-parameter models (correlated evolution among traits). Below diagonal: P values calculated by 10 000 simulations. Significant differences among the –logL are marked with an asterisk (*), indicating a better fit of the data to the model of correlated evolution between two traits.

Trait

01

02

03

01. Nectar 0.357 0.302 guides 02. Corolla 0.4 0.0 curvature 03. Corolla 0.91 10.67* texture 04. Corolla 8.97* 0.32 0.92 tube 05. Corolla 0.44 21.25* 1.43* lobes 06. Corolla 1.6 4.92* 4.58* mouth at anthesis 07. Anther 5.72 1.85 6.7* position 08. Double 0.38914.6* 6.5 calyx 09. Corolla 5.43* 3.73 1.31 tubular 10. Corolla 1.08* 0.75 1.71 infundibuliform 11. Corolla 0.3 0.48 0.15 urceolate 12. Nectar disk10.8* 0.55* 0.95* 13. Calyx 1.05 2.04* 2.01 cupular 14. Calyx 0.86 8.71* 9.21* tubular 15. Calyx 1.08 2.63 3.2 spathaceous 16. Calyx 0.3 2.23 2.96 urceolate 17. Corolla 2.35 1.93* 4.07* red 18. Corolla 5.8* 0.61* 21.48* magenta 19. Corolla 2.4 0.69 1.15 yellow 20. Corolla 4.06 2.5 2.31 white

04

05

06

0.0001 0.002 0.109

07

08

0.065 0.393

09

10

11

12

13

14

15

16

17

18

19

20

0.0002 0.0003 0.548 0.0

0.002 0.538 0.003 0.548 0.082 0.0001 0.164 0.054

0.283

0.0

0.0002 0.154 0.0

0.001

0.141 0.046 0.0

0.0

0.0

0.021 0.012 0.0002 0.0

0.359 0.01

0.17

0.0

0.0002 0.0

0.486

0.026 0.172 0.0

0.271 0.0

0.005 0.072 0.0002 0.0

0.434 0.099

0.034

0.78

0.02

0.65 0.083 0.25

0.018

0.28 0.31

0.0001 0.0

0.0

0.0002 0.16

0.24 0.0001 0.145

0.0

0.04 0.002 0.31 0.03

0.049 0.03 0.0004 0.0

0.0

0.63 0.28

0.191 0.17 0.109 0.112 0.126 0.079 0.28

0.0

0.002 0.005

0.0

0.055 0.0

2.09

5.94*

1.36

1.29* 1.14

0.21

5.32* 6.88*

1.2

1.97

2.95* 1.63

8.12* 0.64

2.02

1.13 1.54

12.62* 0.44

50.34*

0.48

2.32 0.25

3.04 1.56

0.63

2.57*

7.27* 0.001

0.84 1.56 2.13* 1.4

2.25* 0.46 5.14 1.42

2.51 9.2*

2.61 3.83

0.88

5.79* 1.09

9.49* 1.14

5.59*

2.93* 0.73

0.77* 30.17*

0.88

2.78* 0.72

4.76 0.95

3.47

1.58

0.07

1.47 11.07* 2.55*

0.25

0.54 0.61

3.07 2.75

1.59

50.34

0.44

0.36 5.15* 1.25 3.19

0.91

1.28 0.69

7.23* 2.08

9.36* 10.96* 2.14

0.43 4.43

2.9* 1.95* 2.45

5.03* 18.37* 3.22*

4.39* 11.6*

7.31*

7.06* 0.63

3.65 1.86

0.94 1.67 1.5* 8.74*

1.86

0.62 3.47*

6.82* 1.96

5.28*

6.12* 3.04

2.96 3.54

1.72 3.86* 1.07 8.15* 9.76*

3.13

3.67* 3.9*

2.52 2.86

4.89*

5.06* 1.72

3.63* 2.51

4.74* 1.32 1.29 5.51* 9.76* 13.22*

0.249

0.32

0.76

0.0002 0.0001 0.007 0.0 0.577

Evolution of individual floral traits— The unambiguous assignment of most ancestral states of Bignonieae was surprising given that previous studies found high levels of ambiguity while reconstructing ancestral states of floral traits (see Case et al., 2008). Furthermore, given the high levels of homoplasy in floral traits of Bignonieae (Lohmann et al., in press), difficulties in the reconstruction of the ancestral states of those traits had been anticipated. However, the homoplasy in floral traits of Bignonieae was shown to be restricted to specific lineages and/or character types (Lohmann et al., in press). Specifically, most binary-coded traits were unambiguously reconstructed, while multistate characters were shown to be more homoplasious, with the ambiguities being concentrated within particular genera (Appendix S5-A with the online version of this article). Alternative resolutions of polytomies did not affect the ancestralstate reconstructions. Similarly, the alternative branch lengths evaluated also did not impact the ancestral-state reconstructions. Thus, we are confident that the phylogenetic uncertainty gener-

0.091 0.0

0.003 0.011 0.0

0.0001 0.0002 0.0001 0.0

0.0002 0.062

0.919 0.075 0.88 0.041 0.36 0.045 0.012 0.103 0.0

0.19

0.159

0.0

0.545 0.14 0.0002 0.0

0.011 0.301 0.0

0.0

0.0

0.0002

0.0002 0.36 0.0007 0.015 0.03 0.026 0.0

0.0

0.0

0.0002

0.07 0.206 0.148 0.231 0.137 0.002 0.199 0.059 0.25 0.58 0.73

0.23 1.46

0.0 0.0

0.34 0.01 0.88 0.23 0.2 0.0 0.0 0.0 0.006 0.053 0.179 0.237 0.0

0.382 0.0001 0.005 0.521 0.0001 0.007 0.0002 0.201 0.0 0.015 0.0 0.0

0.41

0.381 0.061 0.0

0.0

0.0

0.0001 0.0

ated by these variables did not affect the reconstructions reported here. Four of the five lowest rates of evolution were encountered in the following traits: corolla curvature, corolla tube, corolla mouth at anthesis, and double calyx. The character corolla mouth at anthesis and double calyx are associated with the Amphilophium-type flowers. On the other hand, curved corolla and corolla tube are associated with the floral types Pithecoctenium and Cydista, respectively. Despite the utility of these characters for the identification of clades within genera, these traits were homoplasious and did not represent synapomorphies of any genera (Lohmann, 2006). Nectar guides, corolla texture, corolla lobes, anther position in relation to corolla lobes, nectar disk, corolla shape, and calyx shape were also homoplastic (online Appendix S5). Yet, the state transitions were in general concentrated in terminal branches that were sparsely distributed along the phylogeny (e.g., anther position in relation to corolla lobes, see online Appendix S3).

May 2010]

Alcantara and Lohmann—Evolution of floral morphology in Bignonieae

The exception to the unequivocal reconstructions of ancestral states is the character corolla color. This trait had several ambiguous reconstructed suprageneric nodes and was also used in the definition of flower types in Bignonieae (Gentry, 1974a). Despite the ambiguous reconstruction of flower color of the ancestral Bignonieae, there is a higher probability of a magenta ancestral flower. Yet, evolutionary correlations indicated higher rates of change from magenta flowers to any other flower color. In addition, the patterns of color diversity illustrate an asymmetry in color transitions in this group. Specifically, most extant species of Bignonieae are magenta or yellow, with these colors rarely co-occurring within the same genus. On the other hand, red and white corollas have evolved multiple times within different lineages, even within those characterized by yellow or magenta flowers (Appendix S3). Seven of eight studies of flower color shifts compiled by Rausher (2008) indicate that changes in color determined by anthocyanins (i.e., deep purple, magenta, and red flowers) were caused by loss of function (LOF) mutations in genes associated with the synthesis of those pigments. This pattern implies an unequal rate of color transitions from blue to red and from pigmented to white flowers in a great number of taxa (Rausher, 2008). Multiple LOF mutations on anthocyanin pathways (associated with magenta flowers) may have led to the various origins of white flowers in Bignonieae. In turn, LOF mutations in genes associated with the production of anthocyanins combined with the activation of flavonoids/carotenoid production (e.g., Ono et al., 2006) may have been associated with the evolution of yellow and orange flowers. However, to test hypotheses concerning the causes of extreme color lability in Bignonieae, we would need a better understanding of the biosyntheses of flower pigments in this group, combined with field studies on the impact of color polymorphism on the variation of pollinator attractiveness. Corolla color, corolla lobes, corolla shape, and calyx shape are important features in the attractiveness of flowers to pollinators (Vogel, 1954; Fægri and van der Pijl, 1966). In a review of comparative studies on the shifts in pollination syndromes, Fenster et al. (2004) showed that changes in floral traits are associated with pollinator functional groups in ways that are easily predicted by traditional pollination syndromes (Fægri and van der Pijl, 1966). Different traits seem to show different levels of association with different pollinator groups, with flower morphology responding consistently to all functional groups. In Bignonieae, corolla color, corolla shape, and anther position provide excellent examples of high evolutionary lability. In addition, pollination studies indicate that those traits are also associated with particular pollinator groups (e.g., references in Appendix 2). Despite this, no empirical evidence is available to support the hypothesis that the lability in those traits has been caused by pollinator pressures. Indirect selection coupled with pleiotropy resulting in floral exaptations to several pollinator groups represents an alternative explanation to this pattern. It seems unlikely, however, that the reappearance of major floral features associated with specific pollinator groups represents a series of exaptations. Evolution of the overall floral morphology in Bignonieae— Particular floral traits of Bignonieae have been traditionally interpreted as specializations to specific pollinators (Gentry, 1974a). The general trends encountered allowed us to formulate reasonable hypotheses about the pollinator groups and strategies associated with unstudied species. Exceptions are the Anemopaegma-type flowers, which are associated with both specialist

791

and generalist pollination modes. As far as the Anemopaegmatype flowers are concerned, specific floral traits (e.g., flower size, number of flowers per inflorescence, and corolla color) allow us to identify which of the present species are associated with a specialized or a generalized pollination mode (see Results). Unfortunately, however, these features did not allow us to identify the pollination mode associated with the ancestral morphologies here reconstructed as Anemopaegma-type. The Anemopaegma-type flower represents the predominant condition among current species of Bignonieae and was here reconstructed as the ancestral flower of the tribe. This pattern suggests conservatism in floral shape, despite recurrent evolutions of different morphologies in the tribe. Given that the ancestral Anemopaegma-type flowers could not be associated with a generalized or specialized morphology, at least two evolutionary scenarios can be hypothesized. First, a specialized ancestral Bignonieae flower may have been followed by several shifts among pollinator groups, several losses of specialization in some Anemopaegma-type species and generalization through morphological shifts from Anemopaegma to Tynanthus-type flowers. Alternatively, a generalized ancestral morphology may have been followed by several specializations along the history of Bignonieae with at least two shifts to Tynanthus-type flowers, maintaining a generalist pollination system. These alternative scenarios assume two extremes of a continuum concerning the specialization–generalization of flowers to pollinators; however, we cannot ignore all possible intermediate levels of pollinator association (Waser, 2006). In particular, intermediate stages may have occurred several times during the evolutionary history of Bignonieae, even among groups that maintained the Anemopaegma-type flowers, as exemplified by the present specialized and generalized species sharing this morphology. Most comparative studies have encountered great lability in the degree of specialization or generalization of flowers in different plant groups (Armbruster and Baldwin, 1998; Fenster et al., 2004; Tripp and Manos, 2008). While some evolutionary changes may narrow the spectrum of pollinators, e.g., Calochortus lilies (Dilley et al., 2000; Patterson and Givnish, 2004), and some clades of Ruellia (Tripp and Manos, 2008), others seem to broaden the spectrum, e.g., Aphelandra (McDade, 1992) and Dalechampia (Armbruster and Baldwin, 1998). Nevertheless, most studies reflect shifts from one functional group of pollinators to another (Goldblatt and Manning, 1996; Johnson et al., 1998; Steiner, 1998; Wilson et al., 2004; Pérez et al., 2006; Okuyama et al., 2008; Smith et al., 2008; Tripp and Manos, 2008). Similar patterns of change are also observed during the diversification of Bignonieae (Fig. 2). Below we discuss shifts in morphologies associated with specific pollination modes in Bignonieae, but highlight that this pattern might represent an underestimation due to our incomplete sampling. Clades within which a more comprehensive taxon sampling might lead to a significant impact on the results here encountered are discussed in further detail. It is frequently suggested that the relationship between pollinators and flowers mediated by floral traits is prone to parallelisms and reversals (Armbruster, 1993; Goldblatt and Manning, 1996; Bruneau, 1997; Johnson et al., 1998; Patterson and Givnish, 2004; Pérez et al., 2006; Thomson and Wilson, 2008). In Bignonieae, we observe that the ancestral morphology (i.e., Anemopaegma-type) led to multiple evolutions of more derived flower types (Fig. 2), corroborating the hypothesis of parallel evolution. A careful observation of the flower morphology of species of Bignonieae that were not included in

792

American Journal of Botany

the current study, associated with a careful evaluation of their putative position within the phylogeny (online Appendix S1), suggest that an even higher number of transitions in floral morphologies must have occurred during the diversification of Bignonieae. In addition, this observation also indicated that the Anemopaegma-type morphology could subsequently have reevolved from a Tanaecium-type morphology, a hawkmoth-pollinated flower. Because the sampling for this study was designed to encompass maximum diversity, many Tanaecium-type species within Tanaecium were not sampled. Unfortunately, the lack of a complete sampling in Tanaecium may have biased ancestral state reconstructions in this clade. Despite the uncertain reconstructions in some internal nodes of Tanaecium and the assignment of an Anemopaegma-type flower as the ancestral condition for this genus, it seems unlikely that the Anemopaegma-type flower would represent the ancestral condition in those internal nodes or the ancestral condition for Tanaecium. It has been previously suggested that the evolution of hawkmoth pollination might represent an evolutionary dead end (Tripp and Manos, 2008). In Ruellia (Acanthaceae), another neotropical and ecologically diverse group in terms of pollination systems, the evolution of hawkmoth and bat pollination never led to the evolution of different pollination systems (Tripp and Manos, 2008). Reversions of Anemopaegma-type flowers from Tanaecium-type ancestors would imply an absence of constraints in Tanaecium, contrasting with the results of Tripp and Manos (2008); the occurrence of reversions would not corroborate their being an evolutionary dead end. In contrast to hawkmoth-pollinated flowers, bee-, insect-, and hummingbird-pollinated species of Ruellia have been shown to represent very labile floral morphologies (Tripp and Manos, 2008). This lability was not observed in the hummingbird-pollinated flowers here (i.e., Martinella-type flowers). Even though Martinella-type flowers appear repeatedly on the phylogeny of Bignonieae, this flower type never led to reversals into other floral types. Variants of the Martinella-type morphology that are associated with bat-pollination are restricted to three species of Bignonieae (i.e., Adenocalymma dichilum, Fridericia tynanthoides, and Pachyptera ventricosa). Even though these species were not included in the current study, they occur in unrelated genera, also implying a homoplastic evolution. Amphilophium-type flowers represent another flower type putatively pollinated by bats. Interestingly, Amphilophium-type flowers also never led to the evolution of different flower types, suggesting that hummingbird and bat pollination might, instead, represent evolutionary dead ends in Bignonieae. In Penstemon (Wilson et al., 2007) and Costus (Kay and Schemske, 2003), two genera with many hummingbird-pollinated species, multiple origins of ornithophily never led to subsequent evolutions of any other flower types. However, transitions from hummingbird to bee pollination and from hummingbird to hawkmoth pollination (Whittall and Hodges, 2007; Kulbaba and Worley, 2008; Tripp and Manos, 2008) have been documented in other plant groups (see references cited in Tripp and Manos, 2008). The transition of hummingbird to hawkmoth pollination, associated with the directional increase in tube length in Aquilegia provides an excellent example of the latter (Whittall and Hodges, 2007). This pattern was not observed in Bignonieae, a group where hummingbird- and hawkmoth-pollinated flowers seem to have both evolved independently from bee-pollinated ancestors. The Tynanthus-type flower, pollinated by a great variety of insect groups (Gentry, 1974a), evolved at least twice in Bignonieae (Gentry also reported this flower type in Godmania, tribe

[Vol. 97

Tecomeae, Bignoniaceae.). Thus, the morphological shift to a Tynanthus-type flower might represent an example of an “adaptive wandering” (Thomson and Wilson, 2008) in which different morphologies related to generalized pollination systems may have emerged from diverse evolutionary responses to different environments. As a result, this adaptive wandering may increase the morphological diversity in flowers of a particular group, yet maintain a broad pollinator use range. This situation might apply to Bignonieae if the ancestral of the tribe presented a generalist flower. At least one generalization to the Tynanthus-type flower, however, must have occurred from the specialized Cydista-type. The patterns of evolution of the traits that determine floral morphologies in Bignonieae suggest that shifts among floral types took place both with and without intermediate stages. Assuming an association between floral types and pollination modes, this pattern of change implies a combination of gradual (i.e., sequential evolution of correlated traits) and punctuated (i.e., concurrent evolution of correlated traits along a branch) shifts in pollination systems in Bignonieae. Shifts with intermediate stages are thought to have been prevalent in most plant groups, with the intermediate stages being pollinated by two pollen vectors (Stebbins, 1970). Two species of Bignonieae characterized by flowers with mixed morphologies, Stizophyllum perforatum and Tanaecium pyramidatum, were visited by two groups of putative pollinators (Gentry, 1974a; Amaral, 1992; Carvalho et al., 2003). Additional field studies conducted within a phylogenetic context are needed to assess whether these intermediate stages among Gentry’s floral types could be considered transient between different adaptive peaks. On the other hand, the dissociation between the mode of transition and branch length suggests that the reconstructed punctuated shifts do not represent an artifact created by putative extinction of lineages over time. In addition, the dissociation between the number of traits and the mode of transitions among morphologies indicates that some traits must be more important in the determination of shifts among morphologies than others. This observation is in agreement with the qualitative nature of the pollination syndrome definition and with criticism concerning the unclear boundaries of the individual syndromes (e.g., Fægri and van der Pijl, 1966; Fenster et al., 2004; Waser, 2006). Under a quantitative perspective, one should associate punctuated changes to changes in several traits simultaneously, while gradual changes should be associated with changes in a lower number of traits. However, we did not find any correlation between the number of traits and mode of change. Further analyses on the phenotypic space occupied by flowers of Bignonieae and quantitative approaches to assess the association between morphologies and pollinators should provide important insights in this direction (S. Alcantara, F. B. de Oliveira [Universidade de São Paulo], and L. G. Lohmann, unpublished data). The dissociation between branch lengths and number of traits needed to characterize the mode of morphological shifts (i.e., sequential or punctuated) provide interesting insights into the evolution of floral morphologies in Bignonieae. These patterns suggest that punctuated and sequential shifts in floral morphologies could both have led to the evolution of current floral types in this group. It further suggests that the shifts in floral morphologies are determined by complex evolutionary dynamics. Evolutionary correlations among floral traits in Bignonieae— Negative correlations among discrete traits suggest that some combinations of characters are developmentally prohibited

May 2010]

Alcantara and Lohmann—Evolution of floral morphology in Bignonieae

or maladaptive. This finding agrees with the hypothesis that “non-adaptive” phenotypes would be excluded under a syndrome scenario. For example, lack of exserted anthers in flowers with yellow corollas and curved corollas in red flowers are in agreement with the described syndromes of melittophily by large to medium-sized bees, and ornithophily by hummingbirds, respectively (Thomson and Wilson, 2008). The absence of flowers with exserted anthers and curved corollas might represent a functionally maladaptive combination because tubular, curved flowers might prevent hummingbirds from accessing the nectar disk, while exserted anthers might favor pollen transfer by hummingbirds (Grant and Grant, 1968). Among the positive correlations observed, some correspond directly to expectations under the pollination syndrome concept. In particular, nectar guides are positively correlated with infundibuliform corollas (and negatively correlated with tubular corollas), as expected by melittophily. Tubular corollas are negatively correlated with nectar guides and positively correlated with red and white corollas, as expected by hummingbird and bat-hawkmoth pollination, respectively. In Aquilegia, the switch to white tubular corollas pollinated by hawkmoths tends to occur from red tubular corollas pollinated by hummingbirds, which in turn evolved from flowers pollinated by bees (Whittall and Hodges, 2007). However, we did not find any indication of this directionality in Bignonieae (see above). Interestingly, rates of transition in flower characters that are correlated are generally affected by the color of flowers and not the other way around, suggesting that flower color generally guides the transition in other morphological changes. This pattern is in agreement with the proposition that reward and signaling characters might drive shifts between syndromes (Wilson et al., 2006; Thomson and Wilson, 2008). Changes in morphology and pollen presentation would follow to improve the effectiveness of the new pollinators (Thomson and Wilson, 2008). In this case, detailed studies on the flower color transitions in Bignonieae would greatly contribute to a better understanding of the evolution of pollination syndromes in the tribe. These studies, with a historic context provided by upcoming species-level phylogenies (L. G. Lohmann, unpublished data), would be a fruitful approach to understanding pollination shifts at lower taxonomic levels in this group, as well as the ecological processes involved in those shifts. Despite these interesting results, we highlight that several correlations were not significant after the application of the Bonferroni’s adjustment. This correction is recognized to be greatly conservative (Zar, 1999). Hence, putatively weak but significant correlations may have been ignored. The phylogenetic patterns of correlation among floral traits in Bignonieae are being carefully evaluated with a quantitative, multivariate perspective (S. Alcantara et al., unpublished data). We expect that these further analyses on the evolution of flowers of Bignonieae will give us further insights on the correlated nature of the evolution of floral traits. Conclusions and perspectives—Our results indicate an increasing diversification of pollinator systems in Bignonieae, departing from an entomophilic ancestral state and passing through several and repeated shifts in floral morphology. The overall pattern observed indicates an evolution of more derived flower types from an ancestor with Anemopaegma-type flowers. It further suggests repeated specializations, in combination with shifts to generalized Tynanthus-type flowers. These shifts in floral morphologies followed both gradual and punctuated modes, suggesting that shifts occurred under variable evolutionary dynamics.

793

An evaluation of the phylogenetic uncertainty in the tree used for the current study indicates that our results are robust, regardless of the phylogenetic uncertainties. Similarly, the balanced sampling of floral morphologies of Bignonieae in the phylogeny of Lohmann (2006) suggests that the incomplete taxon sampling would have a minor impact on the general shifts in floral morphologies inferred here. The incomplete taxon sampling does, however, likely contribute to an overall underestimation of the number of shifts in floral morphologies. Positive and negative evolutionary correlations among flower traits are in agreement with the evolutionary patterns expected under the pollination syndrome concept. These features suggest that a functional relationship between Bignonieae flowers and pollinator groups must have played a fundamental role in the morphological diversification of the tribe. Studies on the multivariate aspects of evolution of floral phenotypes and a characterization of the ecological context under which the flowers of Bignonieae evolve will provide further evidence on the processes that may have driven the evolution of floral morphologies in this group (S. Alcantara et al., unpublished data). Given the diversity of floral types and plant–pollinator interactions represented in Bignonieae, these analyses should greatly improve our knowledge on the processes that drive morphological evolution of flowers and plant–pollinator interactions in the tropics as a whole.

LITERATURE CITED Abreu, C. R. M., and M. F. Vieira. 2004. Os beija-flores e seus recursos florais em um fragmento florestal de Viçosa, sudeste brasileiro. Lundiana 5: 129–134. [in Portuguese]. Ackermann, M., and M. Weigend. 2006. Nectar, floral morphology and pollination syndrome in Loasaceae subfam. Loasoideae (Cornales). Annals of Botany 98: 503–514. Amaral, M. E. C. 1992. Ecologia floral de dez espécies da tribo Bignonieae (Bignoniaceae) em uma floresta semidecídua no município de Campinas, SP.). Ph.D. dissertation, Universidade Estadual de Campinas–Campinas, São Paulo, Brazil. Araújo, A. C., and M. Sazima. 2003. The assemblage of flowers visited by hummingbirds in the “capões” of southern Pantanal, Mato Grosso do Sul, Brazil. Flora 198: 427–435. Armbruster, W. S. 1992. Phylogeny and the evolution of plant–animal interactions. Bioscience 42: 12–20. Armbruster, W. S. 1993. Evolution of plant pollination systems: Hypotheses and tests with the neotropical vine Dalechampia. Evolution 47: 1480–1505. Armbruster, W. S., and B. G. Baldwin. 1998. Switch from specialized to generalized pollination. Nature 394: 632. Armbruster, W. S., M. E. Edwards, and E. M. Debevec. 1994. Floral character displacement generates assemblage structure of Western Australian triggerplants (Stylidium). Ecology 75: 315–329. Armbruster, W. S., C. B. Fenster, and M. R. Dudash. 2000. Pollination “principles” revisited: Specialization, pollination syndromes, and the evolution of flowers. Det Norske Videnskapsakademia. I. Matematisk Naturvidenskapelige Klasse, Skrifter, Ny Serie 39: 139–148. Armbruster, W. S., and G. L. Webster. 1982. Divergent pollination systems in sympatric species of South American Dalechampia (Euphorbiaceae). American Midland Naturalist 108: 325–337. Baker, H. G., I. Baker, and S. A. Hodges. 1998. Sugar composition of nectars and fruits consumed by bats in the tropics and subtropics. Biotropica 30: 559–586. Barrows, E. M. 1977. Floral maturation and insect visitors of Pachyptera hymenaea (Bignoniaceae). Biotropica 9: 133–134. Bianchi, M. B., S. A. Harris, P. E. Gibbs, and D. E. Prado. 2005. A study of the mating system in Dolichandra cynanchoides (Bignoniaceae): An

794

American Journal of Botany

Argentinian Chaco woodlands liane with a late-acting self-incompatibility. Plant Systematics and Evolution 251: 173–181. Bruneau, A. 1997. Evolution and homology of bird pollination syndromes in Erythrina (Leguminosae). American Journal of Botany 84: 54–71. Carvalho, E. O., M. F. Vieira, and L. A. O. Campos. 2003. Polinizadores de espécies da tribo Bignonieae (Bignoniaceae) na região de Viçosa, Minas Gerais. Proceedings of the XX Congresso Nacional de Botânica in Belém, Pará, Brazil. Sociedade Botânica do Brasil, Brazil. [abstract, in Portuguese]. Case, A. L., S. W. Graham, T. D. Macfarlane, and S. C. H. Barrett. 2008. A phylogenetic study of evolutionary transitions in sexual systems in Australasian Wurmbea (Colchicaceae). International Journal of Plant Sciences 169: 141–156. Chaney, R. W., and E. L. Sanborn. 1933. The Goshen flora of west central Oregon. Carnegie Institution of Washington, Washington, D.C., USA. Correia, M. C. R., M. C. B. Pinheiro, and H. A. de Lima. 2005. Biologia floral e polinização de Arrabidaea conjugata (Vell.) Mart. (Bignoniaceae). Acta Botanica Brasilica 19: 501–510 [in Portuguese]. Correia, M. C. R., M. C. B. Pinheiro, and H. A. de Lima. 2006. Biologia floral e polinização de Anemopaegma chamberlaynii Bur. & K. Schum. (Bignoniaceae). Lundiana 7: 39–46 [in Portuguese]. Darwin, C. R. 1862. On the various contrivances by which orchids are fertilized by insects. John Murray, London, UK. Darwin, C. R. 1871. The descendant of man, and selection in relation to sex. John Murray, London, UK. Dilley, J. D., P. Wilson, and M. R. Mesler. 2000. The radiation of Calochortus: Generalist flowers moving through a mosaic of potential pollinators. Oikos 89: 209–222. Fægri, K., and L. van der Pijl. 1966. The principles of pollination ecology. Pergamon, Oxford, UK. Fenster, C. B., W. S. Armbruster, P. Wilson, M. R. Dudash, and J. D. Thomson. 2004. Pollination syndromes and floral specialization. Annual Review of Ecology Evolution and Systematics 35: 375–403. Galetto, L. 1995. Nectary structure and nectar characteristics in some Bignoniaceae. Plant Systematics and Evolution 196: 99–121. Gentry, A. H. 1974a. Coevolutionary patterns in Central American Bignoniaceae. Annals of the Missouri Botanical Garden 61: 728–759. Gentry, A. H. 1974b. Flowering phenology and diversity in tropical Bignoniaceae. Biotropica 6: 64–68. Gentry, A. H. 1990. Evolutionary patterns in neotropical Bignoniaceae. Memoirs of the New York Botanical Garden 55: 118–129. Gobatto-Rodrigues, A. A., and M. N. S. Stort. 1992. Biologia floral e reprodução de Pyrostegia venusta (Ker-Graw.) Miers (Bignoniaceae). Revista Brasileira de Botânica 15: 37–41. Goldblatt, P., and J. C. Manning. 1996. Phylogeny and speciation in Lapeirousia subgenus Lapeirousia (Iridaceae: Ixioideae). Annals of the Missouri Botanical Garden 83: 346–361. Graham, A. 1985. Studies in neotropical paleobotany IV. The Eocene communities of Panama. Annals of the Missouri Botanical Garden 72: 504–534. Grant, K. A., and V. Grant. 1968. Hummingbirds and their flowers. Columbia University Press, New York, New York, USA. Gusman, A. B., and G. Gottsberger. 1996. Differences in floral morphology, floral nectar constituents, carotenoids, and flavonoids in petals of orange and yellow Pyrostegia venusta (Bignoniaceae) flowers. Phyton 36: 161–171. Hansen, T., W. S. Armbruster, and L. Antonsen. 2000. Comparative analysis of character displacement and spatial adaptations as illustrated by the evolution of Dalechampia blossoms. American Naturalist 156: S17–S34. Harder, L. D., and S. C. H. Barrett. 2006. The ecology and evolution of flowers. Oxford University Press, Oxford, UK. Herrera, C. M. 1996. Floral traits and plant adaptation to insect pollinators: A devil’s advocate approach. In D. G. Lloyd and S. C. H. Barrett [eds.], Floral biology: Studies on floral evolution in animal-pollinated plants, 65–87. Chapman & Hall, New York, New York, USA. Johnson, S. D., H. P. Linder, and K. E. Steiner. 1998. Phylogeny and radiation of pollination systems in Disa (Orchidaceae). American Journal of Botany 85: 402–411.

[Vol. 97

Johnson, S. D., and K. E. Steiner. 1997. Long-tongued fly pollination and evolution of floral spur length in the Disa draconis complex (Orchidaceae). Evolution 51: 45–53. Kay, K. M., and D. W. Schemske. 2003. Pollinator assemblages and visitation rates for 11 species of neotropical Costus (Costaceae). Biotropica 35: 198–207. Kulbaba, M. W., and A. C. Worley. 2008. Floral design in Polemonium brandegei (Polemoniaceae): Genetic and phenotypic variation under hawkmoth and hummingbird pollination. International Journal of Plant Sciences 169: 509–522. Lloyd, D. G., and C. J. Webb. 1992. The selection of heterostyly. In S. C. H. Barrett [ed.], Evolution and function of heterostyly, 179–207. Springer-Verlag, Berlin, Germany. Lohmann, L. G. 2003. Phylogeny, classification, morphological diversification and biogeography of Bignonieae (Bignoniaceae, Lamiales). Ph.D. dissertation, University of Missouri–St. Louis, St. Louis, Missouri, USA. Lohmann, L. G. 2006. Untangling the phylogeny of neotropical lianas (Bignonieae, Bignoniaceae). American Journal of Botany 93: 304–318. Lohmann, L. G. In press. A new generic classification of Bignonieae (Bignoniaceae) based on molecular phylogenetic data and morphological synapomorphies. Annals of the Missouri Botanical Garden. Lohmann, L. G., P. F. Stevens, and E. A. Kellogg. In press. Twists in the phylogeny of the neotropical Bignonieae (Bignoniaceae): Insights from morphology and molecules. Systematic Botany. Lopes, A. V., S. Vogel, and I. C. Machado. 2002. Secretory trichomes, a substitutive floral nectar source in Lundia A. DC. (Bignoniaceae), a genus lacking a functional disl. Annals of Botany 90: 169–174. Lowry, R. 2009. Resampling probability estimates [online application available in VassarStats: Website for Statistical Computation]. Website http://faculty.vassar.edu/lowry/VassarStats.html [accessed 13 October 2009]. Machado, I. C., and S. Vogel. 2004. The north-east-Brazilian liana, Adenocalymna dichilum (Bignoniaceae) pollinated by bats. Annals of Botany 93: 609–613. Maddison, W. P., and D. R. Maddison. 2006. Mesquite: A modular system for evolutionary analysis, v1.12 [computer program]. Website http://mesquiteproject.org. McDade, L. A. 1992. Pollinator relationships, biogeography, and phylogenetics. BioScience 42: 21–26. Midford, P., and W. P. Maddison. 2006. Pagel 94 analysis module of Mesquite, v1.12. Website http://mesquiteproject.org/mesquite/ CharacterEvolution/Pagel94.html. Okuyama, Y., O. Pellmyr, and M. Kato. 2008. Parallel floral adaptations to pollination fungus gnats within the genus Mitella (Saxifragaceae). Molecular Phylogenetics and Evolution 46: 560–575. Ono, E., M. Fukuchi-Mizutani, N. Nakamura, Y. Fukui, K. YonekuraSakakibara, M. Yamaguchi, T. Nakayama, et al. 2006. Yellow flowers generated by expression of aurone biosynthetic pathway. Proceedings of the National Academy of Sciences, USA 103: 11075–11080. Pagel, M. 1994. Detecting correlated evolution on phylogenies: A general method for the comparative analysis of discrete characters. Proceedings of the Royal Society, B, Biological Sciences 255: 37–45. Pagel, M. 1997. Inferring evolutionary processes from phylogenies. Zoologica Scripta 26: 331–348. Pagel, M. 1999. The maximum likelihood approach to reconstructing ancestral character states of discrete characters on phylogenies. Systematic Biology 48: 612–622. Patterson, T. B., and T. J. Givnish. 2004. Geographic cohesion, chromosomal evolution, parallel adaptive radiations, and consequent floral adaptations in Calochortus (Calochortaceae): Evidence from a cpDNA phylogeny. New Phytologist 161: 253–264. Pérez, F., M. T. K. Arroyo, and R. Medel. 2007. Phylogenetic analysis of floral integration in Schizanthus (Solanaceae): Does pollination truly integrate corolla traits? Journal of Evolutionary Biology 20: 1730–1738. Pérez, F., M. T. K. Arroyo, R. Medel, and M. A. Hershkovitz. 2006. Ancestral reconstruction of flower morphology and pollination systems in Schizanthus (Solanaceae). American Journal of Botany 93: 1029–1038.

May 2010]

Alcantara and Lohmann—Evolution of floral morphology in Bignonieae

Posada, D., and T. R. Buckley. 2004. Model selection and model averaging in phylogenetics: Advantages of Akaike information criterion and Bayesian approaches over likelihood ratio tests. Systematic Biology 53: 793–808. Rausher, M. D. 2008. Evolutionary transitions in floral color. International Journal of Plant Sciences 169: 7–21. Sanderson, M. J. 2002. Estimating absolute rates of molecular evolution and divergence times: A penalized likelihood approach. Molecular Biology and Evolution 19: 101–109. Schemske, D. W., and H. D. Bradshaw. 1999. Pollinator preference and the evolution of floral traits in monkeyflowers (Mimulus). Proceedings of the National Academy of Sciences, USA 96: 11910–11915. Smith, S. D., C. Ané, and D. A. Baum. 2008. The role of pollinator shifts in the floral diversification of Iochroma (Solanaceae). Evolution 62: 793–806. Stebbins, G. L. 1970. Adaptive radiation of reproductive characteristics in angiosperms. I. pollination mechanisms. Annual Review of Ecology and Systematics 1: 307–326. Stebbins, G. L. 1974. Flowering plants: Evolution above the species level. Belknap Press, Cambridge, Massachusetts, USA. Steiner, K. E. 1998. The evolution of beetle pollination in a South African orchid. American Journal of Botany 85: 1180–1193. Strauss, S. Y., and J. B. Whittall. 2006. Nonpollinator agents of selection on floral traits. In L. D. Harder and S. C. H. Barrett [eds.], The ecology and evolution of flowers, 120–138. Oxford University Press, Oxford, UK. Thomson, J. D., and P. Wilson. 2008. Explaining evolutionary shifts between bee and hummingbird pollination: Convergence, divergence, and directionality. International Journal of Plant Sciences 169: 23–38. Appendix 1. Characters and respective character states from Lohmann (2003) and Lohmann et al. (in press). N: number of states, Rate of transition: per Myr assigned by maximum likelihood (ML) reconstructions in the MLcompletely resolved and dated tree.

Character 01. Nectar guides 02. Corolla curvature 03. Corolla texture 04. Corolla tube 05. Corolla lobes 06. Corolla mouth at anthesis 07. Anther position in relation to corolla lobes 08. Double calyx 09. Corolla shape 10. Nectar disc 11. Calyx shape

12. Corolla color

States

N

Rate of transition

1. Absent 2. Present 1. Straight 2. Curved 1. Membranous 2. Coriaceous 1. Rounded 2. Flattened abaxially 1. Bilabiate 2. Not bilabiate 1. Opened 2. Closed 1. Included 2. Exserted 1. Absent 2. Present 3. Only reminiscent present 1. Tubular 2. Infundibuliform 3. Urceolate 1. Absent 2. Reduced 3. Present 1. Cupular 2. Tubular 3. Spathaceous 4. Urceolate 1. Red 2. Magenta 3. Yellow 4. White

2

0.00162

2

0.00051

2

0.00157

2

0.00107

2

0.00103

2

0.00103

2

0.00773

3

0.00051

3

0.00566

3

0.00105

4

0.00303

4

0.00798

795

Tripp, E. A., and P. S. Manos. 2008. Is floral specialization an evolutionary dead-end? Pollination system transition in Ruellia (Acanthaceae). Evolution 62: 1712–1737. Vogel, S. 1954. Blütenbiologische Typen als Elemente der Sippengliederung, dargestellt anhand der Flora Südafrikas. Botanische Studien 1: 1–338. Waser, N. M. 2006. Specialization and generalization in plant–pollinator interactions: A historical perspective. In N. M. Waser and J. Ollerton [eds.], Plant–pollinator interactions: From specialization to generalization, 3–17. University of Chicago Press, Chicago, Illinois, USA. Whittall, J. B., and S. A. Hodges. 2007. Pollinator shifts drive increasingly long nectar spurs in columbine flowers. Nature 447: 706–709. Wilson, P., M. C. Castellanos, J. N. Hogue, J. D. Thomson, and W. S. Armbruster. 2004. A multivariate search for pollination syndromes among penstemons. Oikos 104: 345–361. Wilson, P., M. C. Castellanos, A. D. Wolfe, and J. D. Thomson. 2006. Shifts between bee- and bird-pollination among penstemons. In N. M. Waser and J. Ollerton [eds.], Plant–pollinator interactions: From specialization to generalization, 47–68. University of Chicago Press, Chicago, Illinois, USA. Wilson, P., A. D. Wolfe, W. S. Armbruster, and J. D. Thomson. 2007. Constrained lability in floral evolution: Counting convergent origins of hummingbird pollination in Penstemon and Keckiella. New Phytologist 176: 883–890. Yamashiro, T., A. Yamashiro, J. Yokoyama, and M. Maki. 2008. Morphological aspects and phylogenetic analyses of pollination systems in the Tylophora-Vincetoxicum complex (Apocynaceae-Asclepiadoideae) in Japan. Biological Journal of Linnean Society 93: 325–341. Zar, J. H. 1999. Biostatistical analysis, 4th ed. Prentice Hall, Upper Saddle River, New Jersey, USA.

796

American Journal of Botany

Appendix 2. Pollination systems of 45 species of Bignonieae. Species names follow Lohmann (in press). Flower types were classified according to Gentry

(1974b). Legitimate visitors include all functional groups reported as pollinators in the reference cited herein. Other visitors include illegitimate visitors such as nectar and pollen robbers or visitors that were not observed as resource-thieves. Species

References

Flower type

Legitimate visitors

Adenocalymma bracteatum Adenocalymma comosum Adenocalymma dichilum

Amaral 1992 Porsch 1929 in Machado & Vogel 2004 Machado & Vogel 2004

Anemopaegma Anemopaegma Martinella

Large to medium-sized bees Hummingbirds Bats

Adenocalymma marginatum Amphilophium askersonii Amphilophium buccinatorium Amphilophium crucigerum Amphilophium paniculatum Amphilophium pannosum Anemopaegma chamberlaynii Anemopaegma orbiculatum Bignonia aequinoctialis Bignonia corymbosa Bignonia diversifolia

Amaral 1992 Lohmann 2003 Baker et al. 1998 Gentry 1974b; Amaral 1992 Amaral 1992; L.G. Lohmann pers. obs. Gentry 1974b Amaral 1992; Correia et al., 2006 Gentry 1974b Gentry 1974a, b Gentry 1974a Gentry 1974a

Anemopaegma Amphilophium Martinella Pithecoctenium Amphilophium Amphilophium Anemopaegma Anemopaegma Cydista Cydista Cydista

Large to medium-sized bees Bats, large to medium-sized bees Hummingbirds Large to medium-sized bees Bats, large to medium-sized bees Large to medium-sized bees Large to medium-sized bees Large to medium-sized bees Large to medium-sized bees Large to medium-sized bees Not observed

Bignonia hyacinthina Callichlamys latifolia Dolichadra cynanchoides Dolichandra unguis-cati Fridericia candicans Fridericia caudigera Fridericia chica Fridericia conjugata

Gentry 1974b Gentry 1974b Galetto 1995; Bianchi et al. 2005 Gentry 1974b Gentry 1974a, b Carvalho et al. 2003 Gentry 1974a, b Gentry 1974a, b; Correia et al. 2005

Tynanthus Anemopaegma Martinella Anemopaegma Anemopaegma Anemopaegma Anemopaegma Anemopaegma

Small bees Large to medium-sized bees Hummingbirds Large to medium-sized bees Large to medium-sized bees Large to medium-sized bees Large to medium-sized bees Large to medium-sized bees

Fridericia corallina Fridericia florida Fridericia japurensis Fridericia mollissima Fridericia patellifera

Gentry 1974a, b Gentry 1974a, b Gentry 1974b Gentry 1974a, b Gentry 1974a, b

Anemopaegma Tynanthus Cydista/Tanaecium Anemopaegma Anemopaegma

Large to medium-sized bees Small bees Large to medium-sized bees Large to medium-sized bees Bees

Fridericia pubescens

Araujo & Sazima 2003; Carvalho et al. 2003 Amaral 1992 Abreu & Vieira 2004 Amaral 1992 Lopes et al., 2002

Anemopaegma

Large to small bees

Anemopaegma Martinella Anemopaegma Martinella

Large to medium-sized bees Hummingbirds Large to medium-sized bees Hummingbirds

Anemopaegma Anemopaegma Anemopaegma Martinella Tanecium Martinella

Small hawkmoths Large to medium-sized bees Large to medium-sized bees Hummingbirds Large to medium-sized bees Hummingbirds

Stizophyllum perforatum

Gentry 1974b Amaral 1992 Barrows 1977 Gentry 1974b Gentry 1974b Gobatto-Rodrigues & Stort 1992; Gusman & Gottsberger 1996 Amaral 1992; Carvalho et al. 2003

Stizophyllum riparium

Gentry 1974b

Butterflies; large to medium-sized bees Large to medium-sized bees

Tanaecium jaroba Tanaecium pyramidatum

Gentry 1990 Gentry 1974b; Carvalho et al. 2003

Tanaecium selloi Tynanthus croatianus

Amaral 1992 Gentry 1974b

Anemopaegma/ Martinella Anemopaegma/ Martinella Tanaecium Anemopaegma/ Pithecoctenium Anemopaegma Tynanthus

Xylophragma mirianthum Xylophragma seemanianum

Carvalho et al. 2003 Gentry 1974b

Anemopaegma Anemopaegma

Large to medium-sized bees Large to medium-sized bees

Fridericia samydoides Fridericia speciosa Fridericia triplinervia Lundia cordata Lundia corymbifera Lundia obliqua Mansoa hymenaea Martinella obovata Pachyptera kerere Pyrostegia venusta

Other visitors

Large to medium-sized bees

Beetles (mainly), small bees Small insects

Small bees Small bees, butterflies and hummingbirds Butterflies Small bees, butterflies Hummingbirds, bees, butterflies, wasps, flies Hummingbirds

Small bees, flies, wasps, hummingbird Wasps, ants, small bees Large to medium-sized bees

Hawkmoth Large to medium-sized bee, wasps Bees Small bees

Small insects, butterflies Small insects, large to medium-sized bees