JSE
Journal of Systematics and Evolution
doi: 10.1111/jse.12275
Research Article
Evolution of floral zygomorphy in androecium and corolla in Solanaceae Jingbo Zhang1, Peter F. Stevens2, and Wenheng Zhang1* 1
Department of Biology, Virginia Commonwealth University, 1000 West Cary Street, Richmond, VA 23284, USA Department of Biology, University of Missouri-St. Louis, 1 University Boulevard, St. Louis, MO 63121, USA *Author for correspondence. E-mail:
[email protected]. Tel.: þ01-8048280798. fax: þ01-8048280503. Received 4 April 2017; Accepted 11 August 2017; Article first published online xx Month 2017
2
Abstract In Solanaceae, a group dominated by actinomorphic-flowered species, floral zygomorphy is frequently observed among the early-branching clades. Morphological studies indicated that a zygomorphic androecium is much more common than a zygomorphic corolla in the family. Ontogenic studies suggested the evolution of floral zygomorphy in these two whorls is independent. Here, we have examined the evolution of floral symmetry in the androecium and corolla in Solanaceae. The character states of floral symmetry were assembled for androecium and corolla separately, and ancestral state reconstructions were carried out at both the genus and species levels for Solanaceae and its outgroups. Correlation tests were performed to determine whether the presence of floral zygomorphy in the androecium and corolla is correlated. The ancestral state reconstructions suggest the flower of the most recent common ancestor of Solanaceae is likely zygomorphic in the androecium but actinomorphic in the corolla. Multiple losses and gains of floral zygomorphy in androecium and corolla explain the existing pattern of floral symmetry in Solanaceae. A significant positive correlation between the possession of floral zygomorphy in the androecium and corolla of Solanaceae was detected. Floral zygomorphy likely has evolved in androecium and corolla along separate evolutionary trajectories, and zygomorphy in the androecium may be a precursor for the many gains of zygomorphy in the corolla in Solanaceae. Key words: ancestral state reconstruction, character, character states, Convolvulaceae, correlated evolution, floral symmetry, floral zygomorphy, phylogeny, Solanaceae, Solanales.
1 Introduction Two main kinds of floral symmetry are recognized in angiosperms, zygomorphy (bilateral symmetry, monosymmetry) with a single plane of symmetry and actinomorphy (radial symmetry, polysymmetry) with several planes of symmetry (Neal et al., 1998). Floral zygomorphy is often thought of as an adaptive trait associated with large clades of angiosperms, such as Orchidaceae, core Lamiales, and Fabaceae (Ushimaru & Hyodo, 2005; Fenster et al., 2009; Ushimaru et al., 2009). The shape of zygomorphic flowers restricts pollinator movement and access to the flowers, which results in precise deposition of pollen and hence increased pollination efficiency (Jesson & Barrett, 2002, 2005; Fenster et al., 2009). Paleontological and phylogenetic studies have shown that floral zygomorphy has evolved independently multiple times from ancestors with actinomorphic flowers (Stebbins, 1974; Crepet, 1996; Donoghue et al., 1998; Endress, 2001; Westerkamp & Claßen-Bockhoff, 2007; Takhtajan, 2009; Citerne et al., 2010; Reyes et al., 2016). Understanding the evolution of floral zygomorphy in many lineages, however, awaits in-depth studies (Donoghue et al., 1998; Takhtajan, 2009). The evolution of floral symmetry in Solanaceae is highly dynamic. Recent studies indicate that it is a family with predominately XXX 2017 | Volume 9999 | Issue 9999 | 1–10
actinomorphic flowers, but zygomorphic-flowered species are commonly observed in many groups throughout the family, including its early-branching clades (Fig. 1; Knapp, 2002, 2010; Olmstead et al., 2008; Takhtajan, 2009). However, the evolutionary transitions that give rise to the diversity of zygomorphic patterns in this group are still unclear (Robyns, 1931; Knapp, 2002, 2010). One well-characterized synapomorphy of floral development in Solanaceae is that the plane of floral symmetry is oblique, 36° off the median plane (Fig. 2) (Robyns, 1931; Cocucci, 1995; Knapp, 2002). The species of the family usually have the two fused carpels oriented along this 36° oblique plane regardless of floral symmetry (Hunziker, 2001). Floral zygomorphy is always established along this oblique plane (Wydler, 1866; Eichler, 1878; Robyns, 1931). In Solanaceae, floral zygomorphy is observed in androecium, corolla, and/or calyx. The zygomorphic androecium can result from stamen abortion (e.g., Leptoglossis Benth., Salpiglossis Ruiz & Pav on, and Schizanthus Ruiz & Pav on), modifications in anther size and/or shape (e.g., common in Solanum L.), unequal filament length (e.g., Lycianthes Hassl., Fabiana Ruiz & Pav on, and Vestia Willd.), or filament curvature (e.g., Anthocercis Labill., Atropa L., and Duboisia R. Br.) (Cocucci, 1995; Hunziker, 2001) (Fig. 2). The development of the zygomorphic androecium in © 2017 Institute of Botany, Chinese Academy of Sciences
2
Zhang et al.
n (A), Schizanthus grahamii Gill. (B), Nicotiana alata Fig. 1. Zygomorphic flowers in Solanaceae. Schizanthus pinnatus Ruiz & Pavo Link & Otto (C), Browallia speciosa Hook. (D), Calibrachoa hybrid (E), and Nicotiana obtusifolia Martens & Galeotti (F) show different degrees of floral zygomorphy, but all display a single dorsal petal at maturity. Solanaceae follows Robyns’ rule, that is, the single ventral (abaxial) stamen along the 36° oblique plane of floral symmetry is always the first to be modified or lost (Robyns, 1931). On the other hand, the zygomorphic corollas are highly heterogeneous regarding the shape and size of the limb and tube. The limbs can be strongly zygomorphic with deeply fringed petals (e.g., Schizanthus), with distinctive color patterns in the dorsal and ventral regions (e.g., Salpiglossis), or only slightly different in size along the dorsoventral plane of floral symmetry (e.g., Nicotiana L.) (Hunziker, 2001). The tubes can be zygomorphic, such as Nicotiana, Hyoscyamus L., n, and some species of Solanum (e.g., Juanulloa Ruiz & Pavo Solanum herculeum Bohs) (Hunziker, 2001). Furthermore, the shapes of the tube are highly diverse: short and open (e.g., Solanum tridynamum Dunal), long and narrow (e.g., Nicotiana), cup-shaped (e.g., Solandra Sw.), or funnel-shaped (e.g., Petunia (Juss.) Wijsman) (Hunziker, 2001). The diversity of floral morphology in Solanaceae suggests that the flowers are associated with a broad range of animal pollinators, including bees, butterflies, birds, and bats (Cocucci, 1995; Knapp, 2010). Investigations of diverse lineages across Solanaceae revealed that zygomorphy is more common in the androecium than in the corolla (Robyns, 1931; Knapp, 2010). Robyns (1931) observed that while Hyoscyamus niger L., Petunia axillaris (Lam.) Britton, Stern & Poggenb., and Nicotiana alata Link & Otto share short dorsal (adaxial) stamens in their androecia, their zygomorphic corollas are modified in different ways, with dorsal petal enlargement in Hyoscyamus niger, but ventral petal enlargement in Petunia axillaris and Nicotiana alata. Robyns explained that the differences in zygomorphy in these two whorls are due to independent modifications in androecium and corolla. Detailed studies of floral ontogeny also imply a complicated evolutionary history. In androecia, J. Syst. Evol. 9999 (9999): 1–10, 2017
the zygomorphy seems to develop through either simultaneous primordium initiation with subsequent modifications, as in Lycium horridum Thunb. and Nierembergia repens Ruiz & Pav on, or sequential primordium initiation, as in Schwenckia browallioides Kunth and Schizanthus pinnatus Ruiz & Pav on (Ampornpan & Armstrong, 1988, 1990; Ampornpan, 1992). Similarly, the zygomorphic corolla may have either simultaneous primordium initiation as in Schizanthus pinnatus, or sequential primordium initiation as in Streptosolen jamesonii (Benth.) Miers (Ampornpan & Armstrong, 1988; Ampornpan, 1992). Moreover, the establishment of floral zygomorphy in different lineages can occur at different developmental stages (Ampornpan, 1992). Taken together, these observations suggest that floral zygomorphy in Solanaceae might have arisen several times and/or have been substantially modified in different lineages. Recently, the evolution of floral symmetry in Solanaceae has been reconstructed by Knapp (2010). In that study, floral symmetry was treated as a single character with two states, i.e., actinomorphy and zygomorphy, and a placeholder “Convolvulaceae” was used to root the solanaceous phylogeny. The study indicated that zygomorphy had evolved at least four times independently in the early-branching clades of the family, followed by many independent losses and subsequent reacquisitions (Knapp, 2010). This work laid the foundation for us to further explore the evolution of symmetry in Solanaceae. Here, we use a comprehensive strategy to examine the evolution of floral symmetry in Solanaceae. Particularly, our objective is to identify the evolutionary transitions that give rise to the unbalanced distributions of floral zygomorphy in androecium and corolla in this family. Species from 93 recognized genera of Solanaceae and representatives of www.jse.ac.cn
Floral symmetry evolution in Solanaceae
3
Fig. 2. Floral diagrams of Solanaceae and Convolvulaceae. The gynoecium of Solanaceae is positioned as 36° oblique to the median plane regardless of the floral symmetry (A). Floral zygomorphy always establishes along this plane. Four patterns of n stamen abortion are observed in Solanaceae (B), i.e., abortion of the single ventral stamen alone, as in Salpiglossis Ruiz & Pavo (B-1), abortion of the single ventral and shortening of the two dorsal stamens, as in Browallia L. (B-2), abortion of the single ventral and two lateral stamens, as in Leptoglossis Benth. (B-3), and abortion of the single ventral and two dorsal stamens, as in n (B-4). Flowers of Convolvulaceae are considered to be actinomorphic (C, Convolvulus arvensis L.). The Schizanthus Ruiz & Pavo only exception in Convolvulaceae is Humbertia madagascariensis Lam., which develops floral zygomorphy along a plane at 108° oblique to the median plane (D). The empty circle at the top of the floral diagram represents the position of the stem; three black arcs indicate the subtending bract (lower arc) and bracteoles (two lateral arcs); five gray arcs indicate sepals, the numbers on the sepals in A, indicating the order of initiation; five white arcs with black shading indicate the petals; circles with four lobes indicate the androecium: empty ¼ functional stamens, gray ¼ functional stamens with modified lengths of filaments, shaded ¼ staminodes; the gynoecium is positioned in the center; dotted lines indicate the median planes of flowers, which are defined by the stem and subtending bract; arrows across the flower indicate the plane of floral symmetry, for both the actinomorphic zygomorphic species illustrated in A based on the orientation of gynoecium, for zygomorphic-flowered species in B based on both the orientation of gynoecium and floral zygomorphy that overlap; dotted arcuate arrows indicate the resupination of the flower at anthesis. Floral diagrams are modified from Robyns (1931), Cocucci (1995), Eichler (1878), and Deroin (1992). major lineages of Solanales (Soltis et al., 1997; Olmstead & Bohs, 2007) were analyzed in a phylogeny, largely based on the dataset in S€arkinen et al. (2013). We included the major clades of Solanales to be able to examine the diversity of symmetry in the order as a whole. Specifically, we included the genus Humbertia Lam. of Convolvulaceae, sister to all other genera in this family, that exhibits floral zygomorphy in both androecium and corolla (Fig. 2) (Deroin, 1992). The ancestral state reconstructions (ASRs) were carried out considering floral symmetry as either a single character or two characters, zygomorphy of the androecium and corolla being treated separately, and using both genus and species as units in the character analyses. We show that considering floral zygomorphy to be a single character does not allow one to understand www.jse.ac.cn
the evolution of symmetry in Solanaceae, and the proper delimitation of characters and character states providing new insights into the evolution of symmetry in this clade.
2 Material and Methods 2.1 Summarizing characters of floral symmetry We scored floral symmetry either as a single character of the flower, i.e., a flower is scored as zygomorphic if zygomorphy is observed in either the androecium or corolla, or as two separate characters, zygomorphy in the androecium and zygomorphy in the corolla. Four states of symmetry were identified in the species sampled, i.e., actinomorphic, J. Syst. Evol. 9999 (9999): 1–10, 2017
4
Zhang et al.
zygomorphic with a 36° oblique plane, zygomorphic with a 108° oblique plane (Humbertia), and asymmetric. The ASRs of floral symmetry were examined at both the genus and species levels (Tables S1, S2). For analysis at the generic level, the character states were summarized based on the previous morphological studies (e.g., Hunziker, 2001; Knapp, 2010; see Table S1). Since some genera have both zygomorphic and actinomorphic flowers, the floral symmetry was determined by either the Majority Role Count (MRC) or the Diversity Count (DC) methods. For the MRC method, a genus is considered to be zygomorphic if the majority (more than 60%) of its species are zygomorphic, while for the DC method, a genus is considered to be zygomorphic if any of its species is zygomorphic. For the character analysis at the species level, we included 1054 species that represent ca. 40% of the species diversity of Solanaceae. The floral symmetry of these species was determined based on descriptions or botanical drawings in the literature (a complete list of references is given in Table S2).
2.2 Reconstruction of molecular phylogenies A species level phylogeny including 1054 species from 93 genera of Solanaceae and 10 outgroup species was reconstructed. This dataset was mostly based on a published DNA sequence supermatrix by S€arkinen et al. (2013), which comprised two nuclear (waxy and ITS) and five plastid regions (matK, ndhF, trnS-G, trnL-F, and psbA-trnH) including 1075 species of Solanaceae and three species of outgroups, i.e., Convolvulus L. and Ipomoea L. from Convolvulaceae and Montinia Thunb. from Montiniaceae. Based on this dataset, we updated species names and removed the duplicated species following Dupin et al. (2016). Other changes include the merging of Larnax Miers into Deprea Raf. (Carrizo Garcıa et al., 2015; Deanna et al., 2015) and the renaming of Leucophysalis viscosa (Schrad.) Hunz. as Schraderanthus viscosus (Schrad.) Averett (Averett, 2009). In addition, we added published data of an additional genus of Solanaceae, Tubocapsicum anomalum (Franchet & Savatier) Makino and seven species from Humbertia (Convolvulaceae), Grevea Baill., and Kaliphora Hook. f. (Montiniaceae), Sphenoclea Gaertn. (Sphenocleaceae), and Hydrolea L. (Hydroleaceae), representing diverse clades of Solanales (Table S3). The molecular markers for these taxa were downloaded from Genbank and manually aligned with the existing matrix (S€arkinen et al., 2013). The ambiguous regions of this alignment were removed according to the method described by S€arkinen et al. (2013). The phylogeny was inferred using maximum-likelihood (ML) implemented by RAxML v 8.2.6 (Stamatakis, 2006). Bootstrap supports were assessed by the rapid-bootstrapping method with 1000 replicates. The GTRCAT approximation was used for all partitions as recommended (Stamatakis, 2006). These analyses were carried out at the CIPRES Science Gateway (Miller et al., 2010). To obtain the generic level phylogeny, the species level phylogeny was pruned. For the genera that were monophyletic, each genus was represented by a single tip. Genera that were paraphyletic were represented as more than one tip to reflect their phylogenetic placements, or as a single tip if they were nested within an unresolved clade. Then, branch lengths were re-estimated on the final topology by the pml and optim. J. Syst. Evol. 9999 (9999): 1–10, 2017
pml functions in the package of Phangorn version 2.1.1 (Schliep, 2011) using R (R Foundation, Vienna, Austria). 2.3 Reconstruction of ancestral states of floral symmetry ASRs at the genus level were conducted using both parsimony and Bayesian approaches (see Table S4 for data matrix). Both methods are able to incorporate polytomies in the phylogeny. The parsimony analyses were carried out in Mesquite version 3.10 (Maddison & Maddison, 2016) under the unordered states model that counts any change of a character state as one step. Compared with the parsimony method, the Bayesian approach allows exploring the uncertainty in branch lengths and modeling transition rates among character states (Pagel & Meade, 2006). For the Bayesian inference, stochastic character mapping (SCM) (Nielsen, 2002; Huelsenbeck et al., 2003) was used, which was implemented as the make.simmap function in phytools version 0.5-38 (Bollback, 2006; Revell, 2012). SCM works for discrete characters using a continuoustime reversible Markov model with a prior matrix of transition state rates (Huelsenbeck et al., 2003). The best-fitting model for each dataset was first determined among the three models, i.e., equal rates (ER), symmetric rates (SYM), or all rates different (ARD), by using the fitMK function in phytools version 0.5-38 (Revell, 2012). Then, the transition state rates are estimated based on this model. The posterior distribution of the transition rate matrix was then determined using a Markov chain Monte Carlo (MCMC) simulation run for 100 000 generations and sampled every 100 generations. 1000 simulations were carried out to simulate stochastic character histories at each node (Nielsen, 2002; Huelsenbeck et al., 2003). We also conducted ASRs at the species level (see Table S2 for data matrix) using the parsimony method under the unordered states model and the ML method under the Mk1 model (Lewis, 2001) in Mesquite v. 3.10 (Maddison & Maddison, 2016). 2.4 Testing for correlation between floral zygomorphy in androecium and corolla We tested if the occurrences of floral zygomorphy in androecium and corolla are correlated using BayesTraits version 2.0 (Pagel, 1994; Pagel & Meade, 2007). Correlated and uncorrelated models were run based on the MRC and DC datasets. Analyses were carried out using reversible jump MCMC (Pagel & Meade, 2006) for 10 000 000 iterations with sampling every 1000 iterations, of which the first 500 000 iterations were discarded as burn-in. The log Bayes Factor (log BF) value was calculated using the formula: 2 [log marginal likelihood (two traits dependent model)–log marginal likelihood (two traits independent model)] (Kass & Raftery, 1995). The result was interpreted based on the logBF value, that is, the correlated model is favored if 5 > logBF > 2, strongly favored, if 10 > logBF > 5, or significantly favored, if logBF > 10 (Pagel & Meade, 2006).
3 Results 3.1 Statistics on floral zygomorphy in Solanaceae At the generic level, floral zygomorphy is quite common in Solanaceae with 42% genera (39 out of the 93), or 53% genera www.jse.ac.cn
Floral symmetry evolution in Solanaceae (49 out of 93) having zygomorphic flowers based on MRC and DC, respectively (Fig. 3; Table S1). Within these zygomorphicflowered genera, zygomorphy can occur in both androecium and corolla, or only in the androecium, or only in the corolla, with a ratio of 22: 16: 1 (MRC) or 25: 22: 2 (DC). Zygomorphy is also described in the calyx in 38 genera (Table S1).
3.2 Phylogeny of Solanaceae and its outgroups Our species level phylogeny was rooted by the clade Montiniaceae þ (Sphenocleaceae þ Hydroleaceae), which is the sister of the Convolvulaceae þ Solanaceae clade based on previous work (Soltis et al., 1997). Solanaceae and Convolvulaceae form a sister group with 100% support. Within the Solanaceae, the most early-branching clade is the genus Schizanthus, followed successively by Duckeodendron Kuhlm., Reyesia Gay, and Goetzeoideae. The relationships of the remaining clades are (Schwenckieae, Petunioideae, (Benthamielleae, Cestroideae), (Nicotianoideae þ Solanoideae)). These results are largely congruent with those from previous molecular systematic studies (Olmstead et al., 2008; S€arkinen et al., 2013). To convert the species level phylogeny into a genus level phylogeny, the clades with support rates less than 45% were collapsed. Four large clades then included polytomies: (i) a clade within Anthocercideae (including Anthocercis, Cyphanthera Miers, Grammosolen Haegi, Anthotroche Endl., Crenidium Haegi, and Duboisia) (Garcia & Olmstead, 2003; Olmstead et al., 2008; S€arkinen et al., 2013), (ii) the clade of Benthamielleae (including Benthamiella Speg., Pantacantha Speg., and Combera Sandw.) (Olmstead et al., 2008; S€arkinen et al., 2013), (iii) the clade of Cestreae (including Vestia, Sessea n, and Cestrum L.) (Olmstead et al., 2008; S€arkinen Ruiz & Pavo et al., 2013), and (iv) a clade including Dunalia Kunth., Saracha n, Vassobia Rusby, Eriolarynx Hunz., and some Ruiz & Pavo species of Iochroma Benth. (Fig. S1) (Smith and Baum et al., 2006; Olmstead et al., 2008; S€arkinen et al., 2013). The paraphyletic genera in these four clades are kept as single tips because of unresolved relationships. Moreover, three other paraphyletic genera outside these four clades, i.e., Iochroma (Smith & Baum, 2006), Physalis L. (Whitson et al., 2005; Zamora-Tavares et al., 2016), and Witheringia L’ Heritier (Stone & Pierce, 2005; Olmstead et al., 2008) are represented by more than one tip based on the phylogeny we obtained. The
5
finalized genus-level phylogeny includes 104 tips covering 93 solanaceous and eight outgroup genera (Fig. 4). 3.3 ASRs of floral symmetry evolution in androecium and corolla For ASRs on the genus-level phylogeny using the SCM inference, SYM was found to be the optimal model for all datasets as it scores the lowest AIC value. When symmetry was treated as a single character, both parsimony and SCM inferences generally suggest that the flower of the most recent common ancestor (MRCA) of Solanaceae is zygomorphic with a 36° oblique plane (Tables 1, S4; Figs. S2, S4, S5). When symmetry in the androecium and corolla were treated as two separate characters, the results from both parsimony and SCM inferences suggest that the androecium and the corolla of the MRCA of Solanaceae are zygomorphic with a 36° oblique plane and actinomorphic, respectively (Table 1; Figs. 4, S3, S6, S7). The results of ASRs on the generic level tree based on MRC and DC are largely congruent, except that MRC results from parsimony inferences tends to be unresolved at more nodes, i.e., all character states are inferred with the same possibility at a node (Tables 1, S4; Figs. 4, S2, S3). The ASRs for the species level phylogeny provided similar results as the inferences based on genus level phylogeny, but support was stronger in the ML analyses (Tables 1, 2; Figs. S8, S9). The results based on the Bayesian and ML methods mostly agreed with the parsimony results, but the former also provided support values for the unresolved nodes, which are helpful in interpreting the results. The results also suggest that zygomorphic corollas in diverse lineages of Solanaceae are the outcome of independent origins (Figs. 4, S3, S7). Parsimony analyses indicate that a zygomorphic corolla had evolved independently 17 (MRC) or 20 (DC) times. Most of these independent gains of zygomorphic corollas happened at the levels of the genus or tribe, e.g., Schizanthus, Reyesia, Espadaea Rchb., Petunieae, Schwenckieae, and Browallieae, representing the early-branching lineages, while others were at the species level, e.g., in some species of Markea and Solanum, representing laterbranching clades. On the other hand, zygomorphic androecia have been lost 24 (MRC), or 26 (DC) times (Figs. 4, S3, S6). These losses in androecium zygomorphy mainly occurred in four large clades, i.e., Benthamielleae, Cestreae, Juanulloeae, and a clade including Datureae, Physaleae, Capsiceae, and
Fig. 3. Summary of numbers of genera in Solanaceae having zygomorphy in both androecium and corolla, in androecium alone, or in corolla alone based on MRC (A), and DC (B). www.jse.ac.cn
J. Syst. Evol. 9999 (9999): 1–10, 2017
6
Zhang et al.
Fig. 4. ASRs of floral symmetry in androecium (A), and corolla (B), using parsimony and Bayesian methods based on DC dataset on the genus level phylogeny of Solanaceae and outgroups. Four states of floral symmetry, i.e., actinomorphic, zygomorphic along a 36° oblique plane, zygomorphic along a 108° oblique plane, and asymmetric, are indicated by white, green, red, and black, respectively. Two pie charts at each node represent the results inferred from Bayesian and parsimony methods. Gray indicates unresolved nodes based on parsimony approach. The percentage of area occupied in the pie chart shows the likelihood of a particular character state at a node.
Table 1 Summary of ASRs of floral symmetry based on genus level phylogeny MRCA of Solanaceae
Whole flower MRC
Parsimony SCM
Androecium DC
MRC
Corolla DC
MRC
DC
Not resolved Zygo 36° Not resolved Zygo 36° Actino Actino Zygo 36° (67.7%) Zygo 36° (83.3%) Zygo 36° (65.7%) Zgyo 36° (84.7%) Actino (93.7%) Actino (90.9%)
MRCA, most recent common ancestor; SCM, stochastic character mapping; MRC, majority role count; DC, diversity count; Actino, Actinomorphy; Zygo, Zygomorphy.
J. Syst. Evol. 9999 (9999): 1–10, 2017
www.jse.ac.cn
Floral symmetry evolution in Solanaceae
7
Table 2 Summary of ASRs of floral symmetry based on species level phylogeny MRCA of Solanaceae Parsimony ML
Whole flower
Androecium
Corolla
Not resolved Zygo 36° (89.5%)
Not resolved Zygo 36° (90.7%)
Actino Actino (97.6%)
MRCA, most recent common ancestor; ML, maximum likelihood; Actino, Actinomorphy; Zygo, Zygomorphy.
Solaneae, and in several genera, i.e., Duckeodendron, Tsoala Bosser & D’ Arcy, Goetzea Wydler, Protoschwenkia Soler., Jaborosa Juss., Anisodus Link, Przewalskia Maxim., and Physochlaina G. Don. 3.4 Tests of correlated evolution of floral zygomorphy in androecium and in corolla We detected a significant positive correlation between the presence of floral zygomorphy in androecium and corolla with the log BF value larger than 10 (Table 3).
4 Discussion 4.1 Identification of characters and character states is key to our understanding of character evolution Although the results of ASR can be influenced by many factors, such as sample size, phylogenetic resolution, and the evolutionary mode of traits, proper selection and identification of characters and character states is the sine qua non of the analysis (e.g., Harvey & Pagel, 1991; Stevens, 1991; Wiens, 2001; Sereno, 2007). A good character is thought to be an independent unit that plays a role in a biological process, regardless of the level of observation, gene, cell, or organ (Wagner, 2000; Fristrup, 2001). In Solanaceae, we hypothesize that the evolution of floral symmetry in different floral whorls may have separate trajectories based on the unbalanced distribution of zygomorphy observed in androecium and corolla. Furthermore, previous ontogenic studies also support the idea that the changes in androecium and corolla are likely dissociated (Ampornpan, 1992; Ampornpan & Armstrong, 2002). Therefore, we treat the floral symmetry in the androecium and corolla as two separate units to help us understand the evolution of zygomorphy in Solanaceae. On the other hand, whether the similar morphologies are determined by the same or different mechanisms is the key to determining whether they should be treated as the same or different character states (Hawkins et al., 1997; Freudenstein, 2005; Wake et al., 2011). In the case of Solanaceae versus Convolvulaceae, although floral zygomorphy occurs in both groups, the traits develop through distinct processes. In Convolvulaceae, most species have an actinomorphic corolla but filaments of unequal lengths, but the genus Humbertia,
the sister to all other members of the family, has a zygomorphic corolla and five fertile stamens with unequal length filaments that curve towards the dorsal/adaxial side of the flower (Deroin, 1992; Takhtajan, 2009). The zygomorphic plane of Humbertia, however, is oblique at 108° to the median plane compared to the 36° in Solanaceae (Fig. 2; Robyns, 1931; Deroin, 1992; Knapp, 2002). So, floral zygomorphy in Solanaceae and Convolvulaceae likely had independent origins. ASRs that do not distinguish the two types of zygomorphy give a similar estimation for the MRCA of Solanaceae (data not shown). But they also suggest a single origin of floral zygomorphy in the MRCA of Solanaceae and Convolvulaceae instead of independent origins. Therefore, it is critical to treat different developmental mechanisms of floral zygomorphy as separate character states when interpreting floral symmetry evolution in Solanales. 4.2 Separate evolutionary trajectories of floral symmetry in androecium and in corolla of Solanaceae When floral symmetry is treated as a single character, floral zygomorphy with a 36° oblique plane is seen to have evolved in the MRCA of Solanaceae based on our analysis (Tables 1, S4; Fig. S2, S4, S5). In contrast, Knapp (2010) suggested floral actinomorphy in the MRCA of the family, while zygomorphy had evolved at least four times independently based on the parsimony approach. Both studies agree that gain(s) of floral zygomorphy were followed by many losses and subsequent reacquisitions. These results, however, largely reflect the pattern of symmetry evolution in androecium, which dominates the symmetry patterns of the dataset. The pattern illustrated by the single character ASRs, however, cannot provide precise explanations on how the diverse zygomorphic flower morphologies were achieved in Solanaceae due to the uncertainty of the symmetry pattern in the androecium and corolla of the MRCA of the family. Three evolutionary scenarios may be suggested by the results from the single character ASRs. The first scenario proposes that the MRCA of Solanaceae is zygomorphic in both androecium and corolla, followed by independent losses in both whorls, especially more losses in the corolla. The second scenario proposes that the MRCA of Solanaceae had zygomorphy in the androecium alone, followed by the independent gains of zygomorphy in the corolla but the independent losses of
Table 3 Correlation tests for the co-occurrence of floral zygomorphy in corolla and androecium in Solanaceae MRC DC
Independent model
Dependent model
Difference
P
logBF
117.54 124.43
100.03 112.25
17.51 17.78
2 suggests positive correlation; 10 > logBF > 5 suggests strong positive correlation; logFB > 10 suggests very strong positive correlation. www.jse.ac.cn
J. Syst. Evol. 9999 (9999): 1–10, 2017
8
Zhang et al.
zygomorphy in the androecium. The third scenario proposes that the MRCA of Solanaceae had zygomorphy in the corolla alone, followed by the independent gains of zygomorphy in the androecium but the independent losses of zygomorphy in the corolla. When the floral symmetry in androecium and corolla are treated as two characters, the ASRs suggest that the MRCA of Solanaceae had a zygomorphic androecium but an actinomorphic corolla, which is consistent with the second scenario. We propose that the origin of floral zygomorphy in the androecium preceded the origin of floral zygomorphy in the corolla, which will serve as a working hypothesis to test the molecular mechanisms underlying these dynamic floral symmetry patterns in Solanaceae.
4.3 Correlated evolution of zygomorphy in androecium and corolla in Solanaceae Interestingly, not only do we show that the origin of corolla zygomorphy is after the origin of androecial zygomorphy in Solanaceae, but also we find that the evolution of floral zygomorphy in these two whorls is correlated. One of the explanations for this correlation is that the origin of the corolla zygomorphy more likely took place in the clades that already had zygomorphic androecia. Recent advances in developmental genetics have revealed that CYCLOIDEA2 (CYC2)-like genes from the TCP gene family play crucial roles in regulating the development of zygomorphic flowers in several lineages of core eudicots (e.g., Luo et al., 1996; Busch & Zachgo, 2007; Zhang et al., 2010; Howarth et al., 2011). The shifts of the expression of CYC2-like genes often correlate with modifications of floral zygomorphy within the same or across different whorls of zygomorphic flowers (Hileman et al., 2003; Zhang et al., 2016). For examples, in Mohavea confertiflora (Benth.) A. Heller (Plantaginaceae), the abortion of lateral stamens is associated with the expansion of the expression of CYC2-like genes from the single dorsal stamen to the regions of two lateral stamens (Hileman et al., 2003). In the case of Hiptage benghalensis (L.) Kurz (Malpighiaceae), the expression of CYC2-like genes also shifted, androecial zygomorphy being associated with the expansion of CYC2-like genes from the dorsal corolla into the dorsal androecium (Zhang et al., 2016). Further investigation is required to reveal the molecular mechanisms connecting the evolution of the zygomorphic androecium and the zygomorphic corolla in Solanaceae. In summary, delimitation of characters and character states based on evidence from morphology, ontogeny, and molecular analyses can be helpful for better understanding evolution through providing more precise models of change. Floral zygomorphy has been estimated to have evolved at least 130 times in angiosperms (Reyes et al., 2016). It is usually considered as a single character when studying floral symmetry evolution since the trait is often gained and lost in all whorls of a flower simultaneously. Here, we show that the evolution of floral symmetry in the androecium and corolla is likely along separate evolutionary trajectories in Solanaceae. It is critical to realize that the evolution of floral symmetry can be complicated, which may reflect the responses to different evolutionary and developmental contexts in individual clades.
J. Syst. Evol. 9999 (9999): 1–10, 2017
Acknowledgements The authors thank Drs. Sandra Knapp and Lynn Bohs for the critical review of the manuscript. We thank Sarah McClanahan for helping with the summarization of the morphological data, and Dr. An-Ming Lu for valuable suggestions on this work. This research was supported by VCU and NSF DEB1355109 to Wenheng Zhang.
References Ampornpan L. 1992. Ontogeny of zygomorphic flowers in the Solanaceae. Ph. D. Dissertation. Normal: Illinois State University. Ampornpan L, Armstrong JE. 1988. The floral ontogeny of Schizanthus, a zygomorphic member of the Solanaceae. American Journal of Botany 75: 54. Ampornpan L, Armstrong JE. 1990. The floral ontogeny of Schwenckia (Solanaceae). American Journal of Botany 76: 64. Ampornpan L, Armstrong JE. 2002. Floral ontogeny of Salpiglossis (Solanaceae) and the oblique gynoecium. Journal of the Torrey Botanical Society 129: 85–95. Averett JE. 2009. Schraderanthus, a new genus of Solanaceae. Phytologia 91: 54–61. Bollback JP. 2006. Simmap: Stochastic character mapping of discrete traits on phylogenies. BMC Bioinformatics 7: 88. Busch A, Zachgo S. 2007. Control of corolla monosymmetry in the Brassicaceae Iberis amara. Proceedings of the National Academy of Sciences USA 104: 16714–16719. Carrizo Garcıa C, Wahlert G, Orozco CI, Barboza GE, Bohs L. 2015. Phylogeny of the Andean genus Deprea (Physalideae, Solanaceae): Testing the generic circumscription. Phytotaxa 238: 71–81. Citerne H, Jabbour F, Nadot S, Damerval C. 2010. The evolution of floral symmetry. Advances in Botanical Research 54: 85–137. Cocucci A. 1995. Floral mechanisms in the tribe Salpiglossidae (Solanaceae). Plant Systematics and Evolution 194: 207–230. Crepet WL. 1996. Timing in the evolution of derived floral characters: Upper Cretaceous (Turonian) taxa with tricolpate and tricolpatederived pollen. Review of Palaeobotany and Palynology 90: 339–359. Deanna R, Gonz alez SL, Barboza GE. 2015. Changes in the circumscription of Deprea (Physalideae, Solanaceae): Thirty two new combinations. PhytoKeys 46: 73–87. Deroin T. 1992. Anatomie florale de Humbertia madagascariensis Lam. Contribution a la morphologie compar ee de la fleur et du fruit des Convolvulaceae. Bulletin du Museum National d’Histoire Naturelle. Section B, Adansonia 14: 235–255. Donoghue MJ, Ree RH, Baum DA. 1998. Phylogeny and the evolution of flower symmetry in the Asteridae. Trends in Plant Science 3: 311–317. Dupin J, Matzke NJ, S€ arkinen T, Knapp S, Olmstead RG, Bohs L, Smith SD. 2016. Bayesian estimation of the global biogeographical history of the Solanaceae. Journal of Biogeography 44: 887–899. Eichler AW. 1878. Bl€ uthendiagramme. Leipzig: Wilhelm Engelmann. Endress PK. 2001. Evolution of floral symmetry. Current Opinion in Plant Biology 4: 86–91. Fenster CB, Armbruster WS, Dudash MR. 2009. Specialization of flowers: Is floral orientation an overlooked first step? New Phytologist 183: 502–506.
www.jse.ac.cn
Floral symmetry evolution in Solanaceae Freudenstein JV. 2005. Characters, states and homology. Systematic Biology 54: 965–973. Fristrup KM. 2001. A history of character concepts in evolutionary biology. In: Wagner GP ed. The character concept in evolutionary biology. Boston: Academic Press. 13–35. Garcia VF, Olmstead RG. 2003. Phylogenetics of tribe Anthocercideae (Solanaceae) based on ndhf and trnl/f sequence data. Systematic Botany 28: 609–615. Harvey PH, Pagel MD. 1991. The comparative method in evolutionary biology. Oxford: Oxford University Press. Hawkins JA, Hughes CE, Scotland RW. 1997. Primary homology assessment, characters and character states. Cladistics 13: 275–283. Howarth DG, Martins T, Chimney E, Donoghue MJ. 2011. Diversification of CYCLOIDEA expression in the evolution of bilateral flower symmetry in Caprifoliaceae and Lonicera (Dipsacales). Annals of Botany 107: 1521–1532. Hileman LC, Kramer EM, Baum DA. 2003. Differential regulation of symmetry genes and the evolution of floral morphologies. Proceedings of the National Academy of Sciences USA 100: 12814–12819. Huelsenbeck JP, Nielsen R, Bollback JP. 2003. Stochastic mapping of morphological characters. Systematic Biology 52: 131–158. Hunziker AT. 2001. Genera solanacearum: The genera of Solanaceae illustrated, arranged according to a new system. Ruggell: A.R.G. Gantner Verlag K. G. Jesson LK, Barrett SC. 2002. Enantiostyly: Solving the puzzle of mirrorimage flowers. Nature 417: 707. Jesson LK, Barrett SC. 2005. Experimental tests of the function of mirror-image flowers. Biological Journal of the Linnean Society 85: 167–179. Kass RE, Raftery AE. 1995. Bayes factors. Journal of the American Statistical Association 90: 773–795.
9
Proceedings of the Sixth International Solanaceae Conference, Madison, Wisconsin, 23–27 July, 2006. Leuven: International Society for Horticultural Science. 255–268. Olmstead RG, Bohs L, Migid HA, Santiago-Valentin E, Garcia VF, Collier SM. 2008. A molecular phylogeny of the Solanaceae. Taxon 57: 1159–1181. Pagel M. 1994. Detecting correlated evolution on phylogenies: A general method for the comparative analysis of discrete characters. Proceedings of the Royal Society of London B: Biological Sciences 255: 37–45. Pagel M, Meade A. 2006. Bayesian analysis of correlated evolution of discrete characters by reversible-jump Markov chain Monte Carlo. The American Naturalist 167: 808–825. Pagel M, Meade A. 2007. Bayestraits version 2.0.2 [online]. Available from www.evolution.rdg.ac.uk/BayesTraitsV2.0.2.html [accessed March 2016]. Revell LJ. 2012. Phytools: An R package for phylogenetic comparative biology (and other things). Methods in Ecology and Evolution 3: 217–223. Reyes E, Sauquet H, Nadot S. 2016. Perianth symmetry changed at least 199 times in angiosperm evolution. Taxon 65: 945–964. Robyns W. 1931. L’organisation florale des Solanac ees zygomorphes. Memories de l’Academie Royale de Belgique, Classe des Sciences 11: 1–84. S€ arkinen T, Bohs L, Olmstead RG, Knapp S. 2013. A phylogenetic framework for evolutionary study of the nightshades (Solanaceae): A dated 1000-tip tree. BMC Evolutionary Biology 13: 1471–2148. Schliep KP. 2011. Phangorn: Phylogenetic analysis in R. Bioinformatics 27: 592–593. Sereno PC. 2007. Logical basis for morphological characters in phylogenetics. Cladistics 23: 565–587. Smith SD, Baum DA. 2006. Phylogenetics of the florally diverse Andean clade Iochrominae (Solanaceae). American Journal of Botany 93: 1140–1153.
Knapp S. 2002. Floral diversity and evolution in the solanaceae. In: Cronk Q, Batemen R, Hawkins J eds. Developmental genetics and plant evolution. New York: Taylor & Francis. 267–297.
Soltis DE, Soltis PS, Nickrent DL, Johnson LA, Hahn WJ, Hoot SB, Sweere JA, Kuzoff RK, Kron KA, Chase MW. 1997. Angiosperm phylogeny inferred from 18S ribosomal DNA sequences. Annals of the Missouri Botanical Garden 84: 1–49.
Knapp S. 2010. On ‘various contrivances’: Pollination, phylogeny and flower form in the Solanaceae. Philosophical Transactions of the Royal Society of London B: Biological Sciences 365: 449–460.
Stamatakis A. 2006. RAxML-VI-HPC: Maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics 22: 2688–2690.
Lewis PO. 2001. A likelihood approach to estimating phylogeny from discrete morphological character data. Systematic Biology 50: 913–925.
Stebbins GL. 1974. Flowering plants: Evolution above the species level. Cambridge: The Belknap Press of Harvard University.
Luo D, Carpenter R, Vincent C, Copsey L, Coen E. 1996. Origin of floral asymmetry in Antirrhinum. Nature 383: 794–799. Maddison W, Maddison D. 2016. Mesquite: A modular system for evolutionary analysis. Version 3.10. [online]. Avaliable from www. mesquiteproject.wikispaces.com [accessed February 2016] Miller MA, Pfeiffer W, Schwartz T. 2010. Creating the cipres science gateway for inference of large phylogenetic trees. In: Proceedings of the Gateway Computing Environments Workshop (GCE), New Orleans, Louisiana, 14 November, 2010. USA: Institute of Electrical and Electronics Engineers (IEEE). 1–8. Nielsen R. 2002. Mapping mutations on phylogenies. Systematic Biology 51: 729–739. Neal PR, Dafni A, Giurfa M. 1998. Floral symmetry and its role in plantpollinator systems: Terminology, distribution, and hypotheses. Annual Review of Ecology and Systematics 29: 345–373. Olmstead RG, Bohs L. 2007. A summary of molecular systematic research in Solanaceae: 1982–2006. In: Spooner DM, Bohs L, Giovannoni J, Olmstead RG, Shibata D eds. Acta Horticulturae. Proceedings of Solanaceae VI: Genomics Meets Biodiversity.
www.jse.ac.cn
Stevens PF. 1991. Character states, morphological variation, and phylogenetic analysis: A review. Systematic Botany 16: 553–583. Stone J, Pierce S. 2005. Rapid recent radiation of S-RNase lineages in Witheringia solanacea (Solanaceae). Heredity 94: 547–555. Takhtajan A. 2009. Flowering plants. Berlin: Springer Science. Ushimaru A, Hyodo F. 2005. Why do bilaterally symmetrical flowers orient vertically? Flower orientation influences pollinator landing behaviour. Evolutionary Ecology Research 7: 151–160. Ushimaru A, Dohzono I, Takami Y, Hyodo F. 2009. Flower orientation enhances pollen transfer in bilaterally symmetrical flowers. Oecologia 160: 667–674. Wagner GP. 2000. Characters, units and natural kinds: An introduction. In: Wagner GP ed. The character concept in evolutionary biology. Boston: Academic Press. 1–10. Wake DB, Wake MH, Specht CD. 2011. Homoplasy: From detecting pattern to determining process and mechanism of evolution. Science 331: 1032–1035. Westerkamp C, Claßen-Bockhoff R. 2007. Bilabiate flowers: The ultimate response to bees? Annals of Botany 100: 361–374.
J. Syst. Evol. 9999 (9999): 1–10, 2017
10
Zhang et al.
Whitson M, Manos PS, Plunkett GM. 2005. Untangling Physalis (Solanaceae) from the Physaloids: A two-gene phylogeny of the Physalinae. Systematic Botany 30: 216–230. Wiens JJ. 2001. Character analysis in morphological phylogenetics: Problems and solutions. Systematic Biology 50: 689–699. Wydler H. 1866. Morphologische mitteilungen. Flora 49: 513–525. Zamora-Tavares M del P, Martınez M, Magall on S, Guzman-Davalos L, Vargas-Ponce O. 2016. Physalis and Physaloids: A recent and complex evolutionary history. Molecular Phylogenetics and Evolution 100: 41–50. Zhang W, Kramer EM, Davis CC. 2010. Floral symmetry genes and the origin and maintenance of zygomorphy in a plant-pollinator mutualism. Proceedings of the National Academy of Sciences USA 107: 6388–6393. Zhang W, Kramer EM, Davis CC, Prenner G. 2016. Differential expression of CYC2 genes and the elaboration of floral morphologies in Hiptage, an Old World genus of Malpighiaceae. International Journal of Plant Sciences 177: 551–558.
Supplementary Material The following supplementary material is available online for this article at http://onlinelibrary.wiley.com/doi/10.1111/ jse.12275/suppinfo: Fig. S1. Species level phylogeny of Solanaceae reconstructed by ML method. The phylogeny is rooted by species from Sphenocleaceae, Montiniaceae, and Hydroleaceae. Bootstrap values are shown at nodes. “#” indicates the bootstrap value is 100%. The supergeneric groups named by Olmstead et al. (2008) are labeled. The paraphyletic genera are indicated with superscript numbers which match the labels in the genus level phylogeny. Fig. S2. Parsimony ASRs of floral symmetry as a single character based on MRC (A), and DC (B), datasets at genus level in Solanaceae and its outgroups. Four states of floral symmetry, i.e., actinomorphic, zygomorphic along a 36° oblique plane, zygomorphic along a 108° oblique plane, and asymmetric, are indicated by white, green, red, and blue, respectively. Grey indicates unresolved nodes. Fig. S3. Parsimony ASRs of floral symmetry in androecium (A), and corolla (B), based on MRC datasets at genus level in Solanaceae and its outgroups. Four states of floral symmetry, i.e., actinomorphic, zygomorphic along a 36° oblique plane, zygomorphic along a 108° oblique plane, and asymmetric, are indicated by white, green, red, and blue, respectively. Grey indicates unresolved nodes. Fig. S4. Bayesian ASRs of floral symmetry as a single character based on MRC datasets at genus level in Solanaceae and its outgroups. Four states of floral symmetry, i.e., actinomorphic, zygomorphic along a 36° oblique plane, zygomorphic along a 108° oblique plane, and asymmetric, are indicated by white, green, red, and blue, respectively. The percentage of area
J. Syst. Evol. 9999 (9999): 1–10, 2017
occupied in the pie chart shows the likelihood of a particular character state at a node. Fig. S5. Bayesian ASRs of floral symmetry as a single character based on DC datasets at genus level in Solanaceae and its outgroups. Four states of floral symmetry, i.e., actinomorphic, zygomorphic along a 36° oblique plane, zygomorphic along a 108° oblique plane, and asymmetric, are indicated by white, green, red, and blue, respectively. The percentage of area occupied in the pie chart shows the likelihood of a particular character state at a node. Fig. S6. Bayesian ASRs of floral symmetry in androecium based on the MRC dataset at genus level in Solanaceae and its outgroups. Four states of floral symmetry, i.e., actinomorphic, zygomorphic along a 36° oblique plane, zygomorphic along a 108° oblique plane, and asymmetric, are indicated by white, green, red, and blue, respectively. The percentage of area occupied in the pie chart shows the likelihood of a particular character state at a node. Fig. S7. Bayesian ASRs of floral symmetry in corolla based on the MRC dataset at genus level in Solanaceae and its outgroups. Three states of floral symmetry, i.e., actinomorphic, zygomorphic along a 36° oblique plane, and zygomorphic along a 108° oblique plane, are indicated by white, green, and red, respectively. The percentage of area occupied in the pie chart shows the likelihood of a particular character state at a node. Fig. S8. Species level ASRs of floral symmetry in the whole flower (A), androecium (B), and corolla (C), in Solanaceae and its outgroups using parsimony. Four states of floral symmetry, i.e., actinomorphic, zygomorphic along a 36° oblique plane, zygomorphic along a 108° oblique plane, and asymmetric, are indicated by white, green, red, and blue, respectively. Grey indicates unresolved nodes. Fig. S9. Species level ASRs of floral symmetry in the whole flower (A), androecium (B), and corolla (C), in Solanaceae and its outgroups using ML inference. Four states of floral symmetry, i.e., actinomorphic, zygomorphic along a 36° oblique plane, zygomorphic along a 108° oblique plane, and asymmetric, are indicated by white, green, red, and blue, respectively. The percentage of area occupied in the pie chart shows the likelihood of a particular character state at a node. Table S1. Floral symmetry in androecium, corolla, and calyx at the genus level in Solanaceae and its outgroups. Table S2. Floral symmetry in androecium and corolla at the species level in Solanaceae and its outgroups. Table S3. GenBank accession numbers of DNA sequences added to the supermatrix from S€arkinen et al. (2013) for reconstructing the species level phylogeny. Table S4. Character state matrixes and results of ASRs using Bayesian method. 0 ¼ actinomorphic, 1 ¼ zygomorphic along a 36° oblique plane, 2 ¼ zygomorphic along a 108° oblique plane, 3 ¼ asymmetric.
www.jse.ac.cn