bs_bs_banner

Botanical Journal of the Linnean Society, 2013, 171, 80–102. With 6 figures

REVIEW ARTICLE

Phylogeny and biogeography in Solanaceae, Verbenaceae and Bignoniaceae: a comparison of continental and intercontinental diversification patterns

RICHARD G. OLMSTEAD*

Department of Biology and Burke Museum, University of Washington, Box 355325, Seattle, WA 98195, USA

Received 10 January 2012; revised 31 July 2012; accepted for publication 25 August 2012

Recent molecular phylogenetic studies of Solanales and Lamiales show that Solanaceae, Verbenaceae and Bignoniaceae all diversified in South America. Estimated dates for the stem lineages of all three families imply origins in the Late Cretaceous, at which time South America had separated from the united Gondwanan continent. A comparison of clades in each family shows (1) success in most clades at dispersing to, and diversifying in, North America and/or the Caribbean, (2) a mix of adaptation to novel ecological zones and niche conservation, (3) limited dispersal to continents outside of the western hemisphere, and, where this has occurred, (4) no association between long-distance dispersal and fleshy, animal-dispersed fruits. Shared patterns among the three families contribute to a better understanding of the in situ diversification of the South American flora. © 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102.

ADDITIONAL KEYWORDS: biogeography – long-distance dispersal – Neotropics – South America.

INTRODUCTION & Antonelli, 2005; Andersson, 2006; Hughes & East- wood, 2006; Erkens, Maas & Couvreur, 2009; O’Leary South America is one of the great crucibles of plant et al., 2009; Lu-Irving & Olmstead, 2012) than about diversity, with some of the most diverse plant ecosys- patterns of diversification of indigenous groups that tems on earth populated in high proportion by plant originated and have a long history in South America. A groups that originated and diversified in situ (Gentry, few examples are available that seem to fit these 1982; Pennington & Dick, 2004 Pennington et al., criteria, including Gesneriaceae subfamily Gesnerio- 2010; Antonelli & Sanmartin, 2011). That said, ideae (Smith et al., 1997; Zimmer et al., 2002; Weber, perhaps more has been written about the biogeography 2004; Perret et al., 2012), Bromeliaceae (Givnish et al., of immigrants to South America (Renner, Clausing & 2004, 2011), a large clade of Cactaceae (Edwards, Meyer, 2001; Davis et al., 2002; Lavin et al., 2004; Nyffeler & Donoghue, 2005) and Asteraceae (Funk Richardson et al., 2004; Bell & Donoghue, 2005; Lavin, et al., 2005). A well-documented example is Malpighi- Herendeen & Wojciechowski, 2005; Särkinen et al., aceae, which has a circumtropical distribution, but has 2007; McDade, Daniel & Kiel, 2008; Soza & Olmstead, been shown to have diversified initially in South 2010a, b), and groups that have radiated recently in America at least 64 Mya (Davis et al., 2002) and has South America (e.g. Richardson et al., 2001; Andersson colonized the Old World as many as six times. Three other large clades that originated and exhibited their *E-mail: [email protected] early diversification in South America are the subject

80 © 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 SOLANACEAE, VERBENACEAE AND BIGNONIACEAE 81 of this paper: Solanaceae (Olmstead et al., 2008), 20 Mya, even if it did not persist. An earlier connec- Bignoniaceae (Olmstead et al., 2009) and Verbenaceae tion, if confirmed, would require re-evaluation of (Marx et al., 2010). classic explanations for mammal (Simpson, 1980; Clade diversification may entail both the geographi- Marshall et al., 1982) and, more recently, bird (Smith cal spread from a point of origin and adaptation to & Klicka, 2010) distributions, which have been linked novel environments along the way. The processes to the more recent date. involved in successful migration and adaptation Over-water, long-distance dispersal requires a present a continuum from migration and diversifica- vector; wind, rafting via ocean currents and transport tion in closely similar environments, referred to as by birds are the three most commonly postulated ‘niche conservatism’ (Harvey & Pagel, 1991; Wiens & mechanisms (Carlquist, 1967; Renner, 2004; Nathan Donoghue, 2004), or ‘biome conservatism’ when con- et al., 2008; Baldwin & Wagner, 2010). A summary of sidered on a global scale (Crisp et al., 2009), to adap- mechanisms for transport of plants to Hawaii, one of tation to rather different environmental settings, the most remote oceanic island archipelagos, which referred to as ‘niche evolution’ (Wiens & Donoghue, was populated (prior to arrival of the first Polyne- 2004). Niche conservation has been shown to be asso- sians) entirely by long-distance dispersed propagules, ciated with recently diversified groups in South indicates that transport by birds can account for c. America (e.g. Inga Mill. – Richardson et al., 2001; 75% of all immigrants, with wind and ocean drift Coursetia DC – Lavin et al., 2003; Lavin, 2006; Vale- accounting for the remainder (Gillespie et al., 2012). rianaceae – Bell & Donoghue, 2005; Lupinus L. – Disjunctions between North and South America have Hughes & Eastwood, 2006; Cinchoneae – Andersson also been attributed mainly to transport by birds & Antonelli, 2005; Calceolaria L. – Andersson, 2006) (Carlquist, 1967; Raven, 1972; Wen & Ickert-Bond, and may be a major determinant in the pattern 2009; Cody et al., 2010). However, Renner (2004) recognized by Gentry (1982) that there is a significant reviewed the putative long-distance dispersed dis- taxonomic division between Andean and Amazonian juncts between tropical Africa and South America and floras. At the other end of this continuum, niche concluded that dispersal by rafting on ocean currents evolution, the adaptive shifts between ecological is likely to have been more common than dispersal by zones, has been important in the adaptation of plants birds or wind. Whereas wind can be an effective into geologically new ecosystems, such as oceanic mechanism for seed dispersal in some species, islands (Carlquist, 1974; Baldwin et al., 1998) and the reviews suggest that it is rarely effective for long- cerrado of South America (Simon et al., 2009), but distance, over-water dispersal (Renner, 2004; Nathan may prove to be a barrier to entry into established et al., 2008; Gillespie et al., 2012). From a local, eco- ecosystems with well-developed floras (Antonelli logical perspective, animal-dispersed fruits/seeds, et al., 2009; Antonelli & Sanmartin, 2011). Under- including fleshy fruits or fruits/seeds with mecha- standing the course of diversification of large clades nisms for sticking to the body of an animal, are and the balance between ecological shifts and diver- considered to be more effective for medium- to long- sification in similar habitats can offer insight into the distance seed dispersal than are fruits/seeds lacking relative importance of these processes. any such mechanism (Nathan et al., 2008), although As reviewed by Pennington & Dick (2004) and in such fruits/seeds may have an alternative dispersal the present volume by Christenhusz & Chase (2012), mechanism, such as hitch-hiking on a rafted sub- explanations for intercontinental disjunctions involv- strate, and modelling long-distance dispersal is ing South American plant groups with distributions fraught with difficulty (Levin et al., 2003). on other continents typically involve either relict South America has a long history of geographical Gondwanan distributions (Gentry, 1982), migration isolation from other continental land masses and has routes involving northern hemisphere continental a biota that reflects that isolation, exemplified most connections (Renner et al., 2001; Davis et al., 2002) or iconically by its mammal fauna (Simpson, 1980). Its over-water, long-distance dispersal (Givnish et al., link to Gondwana was severed about 100 Mya when 2004; Renner, 2004; Särkinen et al., 2007). A continu- South America and Africa separated and began to ous land connection to North America has existed drift apart (Goldblatt, 1993; Burnham & Graham, with certainty only since the closure of the Isthmus of 1999; Morley, 2003), even though east–west running Panama (c. 3 Mya), so, for groups that originated in archipelagos (Walvis Ridge/Rio Grande rise and South America after the split up of Gondwana, except Sierra Leone rise) derived from volcanic hotspots on for very recent migrants to North America, virtually or near the mid-Atlantic ridge may have permitted all emigration required over-water dispersal (Cody stepping stone migration perhaps until c. 70 Mya (e.g. et al., 2010). However, recent evidence (Farris et al., Morley, 2003; Pennington & Dick, 2004), when the 2011; Montes et al., 2012) suggests that there may mid-Atlantic ridge moved off the stationary hotspot have been a continuous land connection as early as c. that formed the Walvis Ridge/Rio Grande rise

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 82 R. G. OLMSTEAD

(O’Connor & Duncan, 1990). After the opportunities sphere, (2) the importance of niche conservatism and for short-distance, over-water migration with Africa adaptive shifts between ecological zones for explain- waned in the Cretaceous and early Tertiary, such ing current distribution patterns, (3) the success of opportunities arose again, this time with North transoceanic, long-distance dispersal to other conti- America, in the mid Tertiary when blocks of the nents and, where possible, (4) the association of fleshy Caribbean Plate (proto-Greater Antilles) emerged fruits (conventionally associated with transoceanic, between South and North America at c. 50–40 Mya, long-distance dispersal via animal vectors) with via GAARlandia (Greater Antilles and Aves Ridge) at dispersal to other continental areas. c. 35–33 Mya (Morley, 2003; Pennington & Dick, 2004; Pirie et al., 2006) or via a Central American MATERIAL AND METHODS peninsula that reached close to north-western South America at c. 20–15 Mya (Cody et al., 2010; Montes To infer the ancestral geographical area for Solanales, et al., 2012). Finally, a continuous land connection for which there is currently no single densely sampled arose between South and North America when the study, a tree was assembled by combining results Isthmus of Panama formed, connecting the two con- from several plastid DNA studies of Asteridae tinents at c. 3 Mya (Keigwin, 1982; Coates & Obando, (Bremer et al., 2002; Soltis et al., 2011; N. Refulio & 1996; Burnham & Graham, 1999). Unlike the very R. Olmstead, unpubl. data), Solanaceae (Olmstead limited dispersal capabilities of mammals, however, et al., 2008) and Convolvulaceae (Stefanovic, Krueger the potentially much longer history of plant dispersal & Olmstead, 2002). For Solanaceae, Bignoniaceae and to and from South America makes understanding Verbenaceae, phylogenetic trees from previously pub- the origin of contemporary floras more complicated lished studies were used (Olmstead et al., 2008, 2009, (Pennington & Dick, 2004; Renner, 2004; Wen & and Marx et al., 2010, respectively). Ickert-Bond, 2009; Cody et al., 2010). Distribution ranges were assembled from mono- Phylogenetic studies have been published for three graphs and global treatments for families (e.g. plant families that are widely distributed throughout Gentry, 1980, 1992a; Hunziker, 2001; Atkins, 2004; South America, Solanaceae (Olmstead et al., 2008), Fischer, Theisen & Lohmann, 2004) and specimen Bignoniaceae (Olmstead et al., 2009) and Verbenaceae databases (e.g. GBIF – http://data.gbif.org; TROPI- (Marx et al., 2010). Each family has substantial COS – http://www.tropicos.org). Northern and south- although much smaller representation in North ern distribution limits in the New World were America, and each is also represented on other con- determined for each clade (not only those exemplars tinents. All three families occur in a broad range included in the phylogenetic studies) to the nearest of ecosystems and exhibit a range of dispersal 5° latitude (Table 1). Effort was made to exclude attributes, including animal and non-animal vectors. non-native ranges, where weedy species may have The phylogenetic hypotheses for these three major expanded their distributions through anthropogenic clades of plants suggest that all three originated and means. diversified initially in South America, thus providing natural replicate ‘experiments’ for investigating plant RESULTS diversification in South America and their spread to North America and beyond. Whereas well-sampled, The individual phylogenetic analyses included nearly dated phylogenies are not yet available for any of complete generic level sampling (Solanaceae 91%, these three families, higher level dating studies that Bignoniaceae 85%, Verbenaceae 97%), but sampling have included members of these clades suggest that at the species level was only representative and the crown ages of each are younger than the separa- ranged from 7.5% (Solanaceae) to 12% (Bignoniaceae) tion of South America from Gondwana (Wikström, and 13% (Verbenaceae). This inevitably means that Savolainen & Chase, 2001; Bremer, Friis & Bremer, the diversity of large genera is underrepresented. 2004). In each family, many of these species-rich groups Here we use these phylogenies to explore common have been subject to more fine-grained phylogenetic patterns of geographical distribution and establish studies, but for the purposes here, general patterns explanatory hypotheses that can form the basis for were inferred from the available sampling in pub- future studies. Major clades identified in previously lished higher-level phylogenetic analyses with addi- published phylogenies for Solanaceae (Olmstead tional inferences drawn from species-level studies et al., 2008), Bignoniaceae (Olmstead et al., 2009) and where needed. Verbenaceae (Marx et al., 2010) here provide replicate Solanales comprise three main clades (Fig. 1), with case studies for a series of coarse-scale macro- a small clade composed of Montiniaceae, Sphenoclea evolutionary comparisons, including (1) the ability Gaertn. and Hydrolea L. sister to a much larger clade to extend distributions throughout the western hemi- comprising Convolvulaceae sister to Solanaceae.

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 SOLANACEAE, VERBENACEAE AND BIGNONIACEAE 83

Table 1. Major clades of Solanaceae, Bignoniaceae and Verbenaceae, with presence in South America, North America and outside of the New World, total number of species in each clade, and latitudinal distribution in the western hemisphere estimated to the nearest 5° (Y = present; N = absent)

S. N. Old No. of Latitudinal Clade Am. Am. World species range

Solanaceae Schizanthus Y N N 12 40°S–30°S Goetzeoideae/Metternichia/Tsoala Y Y Y 9 25°S–25°N Cestroideae Y Y N 206 35°S–30°N Benthamielleae Y N N 15 55°S–30°S Petunieae Y Y N 145 50°S–30°N Schwenckieae Y Y N 28 30°S–30°N Nicotiana Y Y Y 76 50°S–50°N Anthocercideae N N Y 32 Lycieae s.l. Y Y Y 229 55°S–40°N Hyoscyameae N N Y 43 Juanulloeae s.l. Y Y N 50 25°S–25°N Mandragora NNY 2 Datureae Y Y N 18 30°S–30°N Salpichroina Y Y N 6 35°S–30°N Cuatresia Y Y N 11 15°S–20°N Withaninae Y N Y 42 30°S–15°S Physalinae/Iochrominae Y Y Y 187 35°S–50°N Capsiceae Y Y Y 231 35°S–35°N Solaneae Y Y Y (~5¥) 1378 50°S–50°N Bignoniaceae Jacarandeae Y Y N 52 30°S–20°N Tourrettieae Y N N 4 40°S–10°N Tecomeae/Argylia YYY(2¥) 57 45°S–40°N Delostoma Y N N 4 15°S–5°N Palaeotropical clade N N Y 151 Tabebuia alliance Y Y N 155 30°S–30°N Oroxyleae N N Y 6 Catalpeae N Y Y 11 20°N–45°N Bignonieae Y Y N 392 35°S–35°N Verbenaceae Petreeae Y Y N 12 25°S–25°N Duranteae Y Y Y 191 30°S–30°N Casellieae Y Y N 20 35°S–25°N Citharexyleae Y Y N 134 30°S–30°N Priveae Y Y Y 21 35°S–30°N Rhaphithamnus Y N N 2 45°S–30°S Neospartoneae Y N N 7 45°S–25°S Verbeneae Y Y Y 177 55°S–50°N Lantaneae Y Y Y (3¥) 276 50°S–40°N

Solanaceae include about 100 genera and c. 2500 species reside in one genus, Solanum L., which is species (Olmstead & Bohs, 2007) and are distributed one of the largest genera of flowering plants and worldwide, but by far the greatest diversity occurs which has a nearly cosmopolitan distribution. in the Neotropics, especially in South America. The Solanaceae have evolved a diversity of alkaloids for plants are mostly small trees, shrubs and herbs, resistance to herbivory, which have been exploited and several of the world’s most important non-cereal for medicinal (atropine), psychotropic (nicotine) and crops belong to Solanaceae (e.g. potatoes, tomatoes, culinary (capsaicin) purposes, among others (Heiser, chili peppers, eggplants). Approximately half of the 1969).

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 84 R. G. OLMSTEAD

Figure 1. Phylogenetic tree of Solanales (see text for sources used to depict relationships). The three main branches are each inferred to have their earliest crown divergence in one of the three main Gondwanan continental fragments.

Divergence time estimates for the crown clade Solanum extend these results by providing additional Solanaceae range from a minimum of c. 35 Mya phylogenetic and geographical resolution in the genus (Dillon et al., 2009), c. 40 Mya (Wikström et al., 2001) (e.g. Bohs & Olmstead, 2001; Bohs, 2005; Levin, or c. 51 Mya (Paape et al., 2008) and for stem Myers & Bohs, 2006; Weese & Bohs, 2007). Other Solanaceae a minimum of c. 65 Mya (Wikström et al., genera that were substantially underrepresented, but 2001) or 85 Mya (Bremer et al., 2004). The fact that which also have more detailed phylogenies, include the closest relatives of Solanaceae diversified initially Nicotiana L. (Clarkson et al., 2004), Nolana L. ex on the different continental plates derived from the L.f. (Dillon et al., 2007, 2009), Jaltomata Schldl. (R. initial separation of Gondwana (Fig. 1) suggests that Miller et al., 2011), Physalis L. (Whitson & Manos, diversification of Solanaceae coincided with the isola- 2005) and Lycium L. (J. Miller, Kamath & Levin, tion of South America from the Late Cretaceous 2009; J. Miller et al., 2011). through the Tertiary. Bignoniaceae include about 80 genera and c. 800 Phylogenetic analyses of Solanaceae (Olmstead & species (Lohmann & Ulloa, 2007; Olmstead et al., Palmer, 1992; Olmstead & Sweere, 1994; Olmstead 2009). They are primarily trees, shrubs and lianas et al., 1999, 2008) have resolved generic relationships and are important components of Neotropical commu- and provided the basis for a revised classification for nities, with secondary distributions in tropical regions the family (Olmstead & Bohs, 2007). For the purposes of Africa and Asia. The large clade, Bignonieae, con- here, 19 clades, ranging in size from as few as two to sists largely of lianas, which form an important com- as many as c. 1400 species, are recognized (Table 1; ponent of Neotropical forest canopies. The family also Fig. 2). Sixteen of the 19 clades are represented in is abundant in tropical savannas, including the the New World, with three clades found only outside cerrado of South America, but only a few lineages the New World (Anthocercideae – Australia; Hyo- have established in temperate zones in the northern scyameae – Eurasia; Mandragora – Eurasia). All of and southern hemispheres. Species of several genera the 16 remaining clades have South American repre- (e.g. Tabebuia Gomes ex D.C., Jacaranda Juss., sentatives; of these, 13 have representatives on the Spathodea P.Beauv.) are widely planted tropical orna- North American continent or in the Greater Antilles mentals and many species have had indigenous uses and seven have Old World representatives. One small (Gentry, 1992b). clade, Withaninae, with about nine genera, but Estimated ages for the crown clade Bignoniaceae only 42 species, occurs in South America, Africa, the are a minimum of c. 40 Mya (Wikström et al., 2001) to Mediterranean basin, East Asia and three widely c. 45 Mya (Nie et al., 2006), although Lohmann, spaced oceanic islands or archipelagos (Olmstead Winkworth & Bell (2012) provide an age estimate for et al., 2008). Since the emphasis on sampling was for the included clade Bignonieae of 40–68 Mya, and generic diversity, Solanum, with about half of all minimum ages of the stem clade range from 47 Mya species in the family, is here very underrepresented. (Wikström et al., 2001) to 68 Mya (Bremer et al., However, numerous detailed phylogenetic studies of 2004). Bignoniaceae belong in a part of Lamiales

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 SOLANACEAE, VERBENACEAE AND BIGNONIACEAE 85

Figure 2. Phylogenetic tree of Solanaceae adapted from Olmstead et al. (2008). Clades in Solanaceae and their distributions are indicated. Arrows indicate Old World dispersal events.

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 86 R. G. OLMSTEAD where resolution is still lacking (Schäferhoff et al., clear sister group in the small west African genus 2010; Soltis et al., 2011; N. Refulio & R. Olmstead, Thomandersia (Wortley, Harris & Scotland, 2007; unpubl. data) and it is still unclear what constitutes Soltis et al., 2011; N. Refulio & R. Olmstead, unpubl. the sister group of Bignoniaceae. data, but see McDade et al., 2012). Phylogenetic analyses of Bignoniaceae (Spangler & Phylogenetic analyses of Verbenaceae (Yuan & Olmstead, 1999; Zjhra, Sytsma & Olmstead, 2004; Olmstead, 2008a; O’Leary et al., 2009; Marx et al., Lohmann, 2006; Grose & Olmstead, 2007a; Olm- 2010; Yuan et al., 2010; Lu-Irving & Olmstead, 2012) stead et al., 2009) have established generic relation- have resolved generic relationships and a new classi- ships and a new classification for Bignoniaceae is fication for Verbenaceae has been developed (O’Leary emerging (Grose & Olmstead, 2007b; Olmstead et al., et al., 2009; Marx et al., 2010), with the exception of 2009; Lohmann, in press). Nine clades are recog- generic circumscriptions and relationships in tribe nized for the purposes of the present study, ranging Lantaneae, which are in progress (P. Lu-Irving & in size from four to nearly 400 species (Table 1; R. Olmstead, unpubl. data). Nine well-supported Fig. 3). Seven of the nine clades are represented in clades ranging in size from two to c. 275 species are the New World, with two exclusively Palaeotropical recognized for the purposes of the present study in distribution (Oroxyleae – SE Asia; Palaeotropical (Fig. 4; Table 1). All nine clades exhibit their greatest clade – Africa, Asia, Australia). Of the seven clades diversity in South America, with seven having dis- with New World distributions, six are represented in tributions reaching North America. Four clades South America and, of these, three include repre- (Duranteae, Priveae, Verbeneae, Lantaneae) have sentatives in North America or the Antilles and one both New and Old World representatives. More includes Old World representatives. The seventh detailed phylogenetic studies are available for two of clade, Catalpeae, occurs exclusively in the northern the most species-rich groups, Verbeneae (Yuan & hemisphere and includes North American, Antillean Olmstead, 2008a, b; O’Leary et al., 2009) and Lanta- and Asian species. As with Withaninae (Solanaceae), neae (Lu-Irving & Olmstead, 2012). one small clade of Bignoniaceae, tribe Tecomeae (12 Distributional ranges for each clade in all three genera, c. 55 species), is exceptionally widely distrib- families were plotted on north and south latitudinal uted, occurring in South America, North America, limits (Fig. 5). Most clades fall within latitudinal Africa, Central Asia, SE Asia and Australia (Olm- limits of 35° north and south of the equator. A handful stead et al., 2009). More densely sampled phyloge- of outliers represent mostly very small clades netic analyses are available for several of the larger restricted to South America (e.g. Schizanthus Ruiz & clades, including tribe Bignonieae (Lohmann, 2006), Pav. – Solanaceae, Tourrettieae – Bignoniaceae and the Tabebuia alliance, including Crescentieae (Grose Neospartoneae – Verbenaceae) and one small north- & Olmstead, 2007a), and tribe Coleeae (Zjhra et al., ern hemisphere clade (Catalpeae – Bignoniaceae). A 2004). second set of outliers represent those clades that have Verbenaceae include about 35 genera and c. 900 extended their ranges into higher latitudes; these species (Atkins, 2004; Marx et al., 2010) of trees, clades typically are large (e.g. Solaneae – Solanaceae, shrubs, lianas and herbs. They are particularly Tecomeae – Bignoniaceae, Lantaneae – Verbenaceae) important components of arid to semi-arid communi- and include Old World representatives. ties in North and South America, but also are present in wet and dry tropical forests, high Andean grass- DISCUSSION lands and cloud forests. A secondary centre of distri- bution is found in Africa. A few species [e.g. Lantana Well-sampled phylogenetic analyses of large families camara L., Stachytarpheta jamaicensis (L.) Vahl.] increasingly permit inferences about the geographical have become widespread weeds in tropical regions, origins of those clades, and, thus, to important and several genera are grown as ornamentals (e.g. insights into the detailed biogeographical history of Glandularia J.Gmelin, Aloysia Palau, Lantana L., the earth’s flora. Digging deeply into the past becomes Duranta L.). increasingly difficult with age, because very old clades The crown age of Verbenaceae has been estimated often have either broad or relictual distributions that to be a minimum of c. 40 Mya (Nie et al., 2006), reflect complicated histories of plate tectonics, migra- whereas the stem age ranges from a minimum of tion and extinction. 48 Mya (Wikström et al., 2001) to 62 Mya (Bremer The difficulty in resolving the branching order of et al., 2004). Verbenaceae belong to the same part of the four primary lineages of Lamiidae (Lamiales, Lamiales as Bignoniaceae, where resolution is still Solanales, Gentianales and Boraginaceae: Schäferhoff lacking (Schäferhoff et al., 2010; Soltis et al., 2011; et al., 2010; Soltis et al., 2011; N. Refulio & R. Olm- N. Refulio & R. Olmstead, unpubl. data). However, stead, unpubl. data) suggests that Solanales and unlike Bignoniaceae, Verbenaceae seem to have a Lamiales have approximately similar stem ages,

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 SOLANACEAE, VERBENACEAE AND BIGNONIACEAE 87

Figure 3. Phylogenetic tree of Bignoniaceae adapted from Olmstead et al. (2009). Clades in Bignoniaceae and their distributions are indicated. Arrows indicate Old World dispersal events; asterisk indicates dispersal back to South America.

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 88 R. G. OLMSTEAD

Figure 4. Phylogenetic tree of Verbenaceae adapted from Marx et al. (2010). Clades in Verbenaceae and their distribu- tions are indicated. Arrows indicate Old World dispersal events. estimated at a minimum of c. 80 Mya (Wikström (Olmstead et al., 2001; Schäferhoff et al., 2010), et al., 2001; Magallón & Castillo, 2009) or c. 106 Mya whereas Solanaceae represent one of three early (Bremer et al., 2004). Bignoniaceae and Verbenaceae diverging lineages in Solanales (Bremer et al., 2002; represent two clades nested well within core Lamiales N. Refulio & R. Olmstead, unpubl. data).

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 SOLANACEAE, VERBENACEAE AND BIGNONIACEAE 89

Northern limit Antarctica) and Solanaceae in South America. Thus (latitude) Solanales may present a classic case of vicariance 60ºN following Gondwanan breakup, with its descendants

50 on each of the three initial continental fragments giving rise to the three extant clades. 40 Solanaceae, Bignoniaceae and Verbenaceae thus represent replicated comparative datasets for inves- 30 tigating the diversification of plants with origins in 20 South America. Divergence time estimates for the crowns of all three family clades (going as far back as 10 the common ancestor of all extant representatives, as Southern limit distinct from the stem clades, which go back to the 30ºN 20 10 10 20 30 40 50 60ºS (latitude) divergence of each lineage from its nearest living 10 relatives) are inferred to be in the Late Cretaceous or early Tertiary, after the separation of South America 20 from Africa (Wikström et al., 2001; Bremer et al., 2004), with Solanaceae inferred to be somewhat older 30ºS than Bignoniaceae or Verbenaceae (Wikström, Savol- Figure 5. Graph depicting the latitudinal limits for the ainen & Chase, 2001; Bremer et al., 2004). Thus each 32 New World clades found in Solanaceae (red), Bignon- family represents a lineage probably derived from iaceae (green) and Verbenaceae (blue). The bold dashed Gondwanan ancestors the diversification of which has line represents the equal northern and southern latitudi- coincided with the geological and climatological nal limits, with the fine dashed lines representing 5° changes that South America has experienced during latitude on either side. Lines at 35° latitude roughly its long existence spanning almost 100 million years represent the 10 °C minimum average monthly tempera- as an isolated continent (Gentry, 1982; Burnham & ture for the coldest month (January in the north; July in Graham, 1999; Antonelli et al., 2009; Antonelli & San- the south). martin, 2011). Phylogenetic studies of each family form the basis for revised classifications that recognize well- Lamiales also represent a phylogenetically chal- supported clades as taxa. Although these clades are lenging clade, in which resolving the branching order all of different ages, the absence of dated phylogenetic has been difficult (Olmstead et al., 2001; Bremer trees constrains the potential for testing hypotheses. et al., 2002; Schäferhoff et al., 2010; Soltis et al., 2011; However, these multiple, more or less independent N. Refulio & R. Olmstead, unpubl. data). However, groups can still be used to explore broad continental- the geographical origins of several large family-level scale geographical patterns of diversification within crown clades of Lamiales (Oleaceae – Wallander & and beyond South America. The latitudinal extent to Albert, 2000; Acanthaceae – McDade et al., 2008; which multiple clades have been able to spread, the Bignoniaceae – Olmstead et al., 2009; Verbenaceae – ability to spread into North America and to other Marx et al., 2010) have been determined and can continents and the degree to which niche conserva- contribute to our understanding of plant diversifica- tism constrains the habitat diversity occupied by tion on a global scale. Two of these clades, Bignon- members of each clade can be examined for each iaceae and Verbenaceae, have unequivocally South clade. These patterns provide more explicit hypoth- American ancestral areas, and have diversified eses for testing with more densely sampled, time- throughout the New World, including North America, calibrated phylogenies. and have established toeholds on other continents, too, primarily in the southern hemisphere (Olmstead et al., 2009; Marx et al., 2010). SOLANACEAE Solanales represent a much smaller clade of similar South American origin and diversification age to Lamiales. However, the relationship among the The 16 clades of Solanaceae present in the New World three primary clades of this order are well resolved all have their origins in South America. The one and robustly supported (Fig. 1), with each of the three possible exception is the Goetzeoideae/Metternichia/ crown clades inferred to have different southern Tsoala clade, which has one lineage in Brazil, one in hemisphere ancestral areas, the Montiniaceae/ the Greater Antilles and one in Madagascar. However, Sphenoclea/Hydrolea clade in Africa, Convolvulaceae evidence suggests that Amazonian Duckeodendron most likely in the continental fragment known Kuhlm. is sister to this clade (Santiago-Valentin & as East Gondwana (SE Asia/India/Madagascar/ Olmstead, 2003; Olmstead et al., 2008), and all other

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 90 R. G. OLMSTEAD early diverging lineages of Solanaceae in this part of the ecological preferences of the plants in each the phylogenetic tree are also South American in clade. origin, so a South American ancestral area for this clade is also likely. Most of the early diverging line- Niche conservatism vs. adaptation to novel ages in Solanaceae are either temperate South ecological zones in the diversification America [Schizanthus, Benthamielleae, Petunieae, of Solanaceae Lycieae sensu lato (s.l.)] or Andean (Cestroideae, Nico- Solanaceae can be found in virtually all terrestrial tiana) in origin. Tropical members of these clades and ecosystems in South America, evidence of a history clades with primarily tropical distributions are all that involved numerous ecological shifts during their relatively derived in Solanaceae. diversification. Similarly, many of the clades identi- fied here for the purposes of comparison include members found in two or more distinct habitats. Distribution in North America and the Caribbean Most of the larger clades (e.g. Solaneae, Capsiceae, Only three clades (Schizanthus, Benthamielleae, Physalinae/Iochrominae, Petunieae, Nicotiana) have Withaninae) have not reached North America (Fig. 5) significant representation in both xeric and mesic and all three are small, with c. 12–15 species each in ecological zones and broad latitudinal ranges, reach- South America (Withaninae are also represented in ing into cool temperate climates, providing further Africa and Asia). Schizanthus and Benthamielleae evidence that adapting to novel ecological settings are found only in far southern South America, has been an important process in diversification of whereas Withaninae (Athenaea Sendt. and Aureliana Solanaceae. However, at the generic level, most Sendt.) are found only in the Atlantic coastal forests groups do exhibit significant levels of niche conserva- of Brazil. tism, and even a couple of the larger clades show There is a striking consistency with respect to the consistent, if contrasting, environmental preferences. northern and southern distribution limits for the The Lycieae/Nolana/Sclerophylax Miers clade is clades that have successfully reached North restricted to arid habitats, with Lycium mostly in cool America. Given that all of the 16 New World clades temperate habitats, distantly disjunct between North apparently originated in South America and south- and South America, Sclerophylax Miers in the arid ern South America in many cases, the same distri- high Andes in northern Argentina, and Nolana occu- bution limits seem to have been reached in most pying coastal lomas. Cestreae, by contrast, are clades, regardless of clade age. Of the 13 clades that restricted to mostly mesic, tropical settings. As have reached North America, ten have north/south sampling for broad, family-level phylogenetic treat- latitudinal limits that are within 5° of each other on ments of Solanaceae necessarily under-samples both sides of the equator (Table 1; Fig. 5), and all species diversity within genera, the relative impor- except Lycieae have more or less continuous distri- tance of niche conservation vs. niche evolution will butions through the Tropics. Lycieae exhibit a dis- probably be underestimated by examining trends on junct distribution between the arid regions of generic-level phylogenetic trees. Much more densely temperate South and North America (a pattern seen sampled phylogenetic trees will be needed to quantify again in Verbenaceae). Exceptions include Phy- the balance between these two contrasting modes of saleae, which has diversified in colder regions of diversification in Solanaceae. temperate North America, where Physalis extends to 50°N (c. 15° further north than the clade occurs Transoceanic dispersal in the south), and Petunieae, which only reaches c. Ten clades of Solanaceae have successfully colonized 30°N, but in which Fabiana Ruiz & Pav. extends as other parts of the world outside of North and South far as 50°S. Whereas several clades do extend to America (prior to European exploration), including latitudes further than c. 30–35° north or south of the three clades that do not occur in the New World the equator, a relatively small number of species, (Anthocercideae, Hyoscyameae, Mandragora L.), each perhaps no more than 125, or c. 5% of New World of which is inferred to have descended from a New species occur at those latitudes. To the extent that World immigrant. A total of c. 15–17 pre-Columbian latitude is a placeholder for ensemble environmental dispersal events are needed to account for the present conditions, or something as simple as the latitudinal distribution, with about five to seven events in limits of a frost-free zone (Thompson, Anderson & Solanum (Bohs, 2005; Olmstead et al., 2008) and two Bartlein, 2000; Wiens & Donoghue, 2004), the in Lycium (J. Miller et al., 2009, 2011). In the absence distribution limits for the clades of Solanaceae of a dated tree, the best one can do is infer relative suggest that for most clades, successful migration age of dispersal events on the basis of depth in the in the New World has reached its full poten- tree where the event was inferred to occur (Fig. 2). tial geographical ranges, within the constraints of Many of the postulated dispersal events are clearly

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 SOLANACEAE, VERBENACEAE AND BIGNONIACEAE 91 recent in origin (a single species of Physalis, mum Jacq. in Africa; S. lasiocarpum Dunal and P. alkekengi L., in East Asia, a Hawaiian endemic S. repandum G.Forst. in the South Pacific; S. spirale subspecies, Lycium carolinianum Walt. subsp. sand- Roxb. in Australia/SE Asia), which in some cases may wicense (Gray) C.L.Hitchcock, of a North American be early anthropogenic dispersal events. Thus, there species, and a species of Solanum section Geminata appear to have been at least four and possibly as many (G.Don) Walp. in Australia, an otherwise entirely as seven or eight pre-Columbian long-distance disper- New World group of c. 140 species), whereas others sal events in Solanum. represent potentially relatively old dispersal events Using simple parsimony to determine the direction (e.g. Tsoala in Madagascar, Anthocercideae in Aus- of dispersal events overstates the confidence with tralia, Hyoscyameae in Eurasia). which we can discuss long-distance dispersal history Withaninae present a microcosm of Solanaceae in any clade, and Solanaceae are no exception. dispersability. With c. 40 species assigned to nine However, based on current evidence, long-distance genera, their distribution includes Brazil, Africa, the dispersal in Solanaceae appears to have been both to western Mediterranean, Central Asia and China, as the east and west (Fig. 6). At least four events are well as remote oceanic islands, such as the Canaries, postulated to have been to the east from South Hawaii and St Helena. The clade probably originated America to Africa or Madagascar: (1) Tsoala to Mada- in tropical South America and dispersed east to Africa gascar; (2) Lycium to Africa (and on to Asia and or the Mediterranean initially, and then on to East Australia); (3) Withaninae to Africa (and on to Asia Asia and Hawaii (Olmstead et al., 2008), although two and Hawaii); and (4) the Solanum non-spiny African/ Asian genera have not been sampled in previous Normania/Archaesolanum to Africa (and on to Asia phylogenetic studies. Despite an origin in South and Australia). At least five events seem to have been America, and an apparent vagility that has permitted to the west: (1) Nicotiana subgenus Suaveolentes to colonization of remote oceanic islands and widely Australia; (2) the ancestor of Anthocercideae to Aus- disjunct localities in Africa and Eurasia, the distri- tralia; (3) Lycium carolinianum to Hawaii (where it is bution of Withaninae in the New World remains recognized as L. carolinianum var. sandwicense); (4) restricted to the Atlantic coastal forest of Brazil, in Physalis alkekengi to East Asia; and (5) a clade within contrast to so many other clades that have colonized Lycianthes (Dunal) Hassler to SE Asia. Hyoscyameae, much of the New World. Mandragora and at least three of the clades in The phylogeny of Solanum is under active investi- Solanum have distributions that extend across gation (Bohs, 2005; Levin et al., 2006; Weese & Bohs, Eurasia or Africa/Asia/Australia, such that no simple 2007; Stern, Agra & Bohs, 2011; L. Bohs, pers. comm.), inferences are possible as to the directionality of the but it is too early to delineate all postulated dispersal dispersal events responsible for them. The recent events in that group confidently. The dispersal event individual species dispersals in Solanum, if they out of the New World that has resulted in the greatest are not of anthropogenic origin, have gone in both radiation in the Old World has occurred in subgenus directions. Leptostemonum (Dunal) Bitter, the spiny solanums, which are worldwide in distribution and include a Association of seed dispersal with clade of c. 175 species widely distributed in the Old long-distance dispersal World. However, in addition to this major diversifica- There is a fundamental split in Solanaceae between tion in Leptostemonum following dispersal to the Old fleshy, animal-dispersed fruits and dry dehiscent cap- World, the relatively ancient clade comprising the sular fruits with no evident means of seed dispersal. non-spiny African species, the Normania Lowe group Along with a few other characteristics, this trait from the Mediterranean and the Canary Islands (as defines the two subfamilies Solanoideae (fleshy fruits) segregate genera until Bohs & Olmstead, 2001) and and Cestroideae (dry fruits) in traditional classifica- the Australian ‘kangaroo apples’ (Symon, 1994), tions (although Cestrum L., itself, has berries). Phy- Solanum section Archaesolanum Marzell (Weese & logenetic studies of Solanaceae (e.g. Olmstead & Bohs, 2007; Poczai, Hyvönen & Symon, 2011) also Palmer, 1992; Olmstead et al., 2008) have shown that represents a significant Old World lineage. The latter Cestroideae form a paraphyletic grade from which clade appears to have involved initial radiation in Solanoideae are derived. Each of the 19 clades con- Africa, followed by dispersal east to Australia, whereas sidered here can be characterized by either fleshy or the direction of dispersal is uncertain in section Lep- dry dehiscent fruits, with only minor exceptions. At tostemonum, where resolution remains poor. Addi- first glance, it appears that there is a close association tional dispersal events in Solanum are likely to be between long-distance, intercontinental dispersal and relatively recent and involve small groups in the fleshy fruits, with 12 of the c. 15 events (80%) occur- dulcamaroid (Eurasia) and Morelloid (Europe, Africa ring in fleshy-fruited clades. However, approximately and Asia) clades, or individual species (S. aculeatissi- 87% of species in Solanaceae belong to fleshy-fruited

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 92 R. G. OLMSTEAD

Figure 6. Direction of transoceanic migration events in the history of Solanaceae (red), Bignoniaceae (green) and Verbenaceae (blue) divided into those inferred to originate in North America and South America. Only events where direction can be inferred are indicated; direction for six to eight additional events in Solanaceae and one in Bignoniaceae cannot be inferred. lineages, so the distribution of long-distance dispersal three of the earliest diverging lineages of Bignon- among clades of Solanaceae does not seem to reflect a iaceae, with Bignonieae being more derived. Jacaran- bias towards fleshy-fruited clades. deae, sister to the rest of the family, are widespread in lowland tropical wet and dry forest and savannas throughout the Neotropics. Their phylogeny is poorly BIGNONIACEAE resolved (R. Farias & L. Freitas, unpubl. data), and South American origin and diversification includes two widespread South American species, the Of the seven New World clades in Bignoniaceae, four ranges of which extend into southern Central can confidently be inferred to be of South American America, and one clade of six Antillean species. origin (Tourrettieae, Tecomeae/Argylia D.Don, Delos- Although the phylogenetic analysis does not permit toma D.Don, Bignonieae). These clades represent an unambiguous ancestral area, the Antillean species

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 SOLANACEAE, VERBENACEAE AND BIGNONIACEAE 93 belong to the putatively derived Jacaranda Juss. sub- (Table 1; Fig. 5). The two largest such clades, Bigno- genus Jacaranda (= Monolobos DC.), characterized by nieae and the Tabebuia alliance, each have nearly the derived condition of having one anther sac identical northern and southern limits (35°S/N and missing; all species of subgenus Dilobos DC. are 30°S/N, respectively), whereas the Tecomeae/Argylia restricted to South America. Therefore, the geographi- clade has northern and southern limits of 45°S/40°N cal origin of Jacarandeae was most likely in South and Jacarandeae extends c. 10° further south of the America. Argylia (sister to Tecomeae), Tourrettieae equator than north (30°S/20°N). Bignoniaceae are and Delostoma are all endemic Andean groups. apparently the most constrained of these three fami- Tecomeae (exclusive of Argylia) are of New World lies to warm tropical regions, with perhaps no more origin, although most of the species diversity is in than seven or eight species (c. 1% of New World Australasia, Africa and the Himalayas. The first two diversity) occurring further than c. 35° north or south diverging branches of Tecomeae are a temperate of the equator. This pattern is consistent with the northern hemisphere taxon (Campsis Lour.), with one tropical niche conservation model of Wiens & Dono- species in North America and one in east Asia, and a ghue (2004). Catalpeae are the only clade recognized primarily Andean taxon (Tecoma Juss.; one wide- for the purposes of this study in any of the three spread species extends into Central America), with families that are restricted to the northern hemi- the remainder of the clade in the Old World (with one sphere; their distribution includes the deserts of exception, see below), thus the geographical origin south-west North America, the Greater Antilles and seems uncertain. However, with modest support for mesic forests in eastern North America and east Asia. the southern Andean taxon Argylia as the sister to Lack of resolution prevents determining what the Tecomeae and the evidence that all the other basal closest relatives of Catalpeae are, with Oroxyleae branches of Bignoniaceae (Jacarandeae, Tourrettieae from SE Asia as a possible candidate (Fig. 3) and and the isolated Delostoma) are South American in Bignonieae and the Tabebuia alliance/Palaeotropical origin would also place the origin of Tecomeae in clade as other possibilities. Thus, it is uncertain at South America. this point whether it is derived directly from South The remaining clade found in South America is the American ancestors or not. Tabebuia alliance, which comprises the large taxon The pattern observed in Bignoniaceae suggests that Tabebuia s.l., now split into three genera, Handroan- relatively old dispersals to North America have thus Mattos, Roseodendron Miranda and Tabebuia occurred, with many recent, post-Panama land bridge (Grose & Olmstead, 2007b), Crescentieae and a few dispersals. Relatively ancient dispersals include Cat- other small Neotropical genera. Crescentieae are alpeae, Campsis, one lineage of Jacaranda in the nested within this clade and are Central American in Antilles (R. Farias & L. Freitas, unpubl. data) and origin. There is also a Caribbean radiation in Tabe- perhaps three lineages of the Tabebuia alliance buia sensu stricto (s.s.), but much of the remaining (Ekmanianthe Urb. and a large clade of Tabebuia s.s. diversity, including two small lineages that emerge at in the Antilles and Crescentieae in Central America). or near the base of the clade (Sparattosperma Mart. Putative post-isthmian dispersals include one wide- ex Meisn. and the Cybistax Mart. ex Meisn./ spread Tecoma sp., two Jacaranda spp., eight species Godmania Hemsl./Zeyheria Mart. clade) are South of Tabebuia s.l. and several widespread species of American. The sister to this clade is a strictly Pal- Bignonieae spanning both South and Central America aeotropical clade, but poor resolution in this part (Gentry, 1982, 1992a). Also, three or four species of of the phylogenetic trees for Bignoniaceae means Central American Crescentieae have ranges that that the sister to the inclusive clade remains extend into northern South America, perhaps the unknown. Thus, there remains uncertainty about the result of mammal migrations made possible by the ancestral area of this clade, but it is probably also closure of the Isthmus of Panama, which assisted South American. with dispersal of their large, fleshy, mammal- dispersed fruits (Janzen & Martin, 1982; Grose & Distribution in North America and the Caribbean Olmstead, 2007a). Of the six clades of Bignoniaceae with putative South American ancestral areas, four also include elements Niche conservatism vs. adaptation to novel in North America. The exceptions are two small ecological zones in the diversification Andean clades, Tourrettieae, which includes three or of Bignoniaceae four species of vines found in the Andes from Chile to Much of the early diversification in Bignoniaceae Colombia, and Delostoma, with four species of trees in occurred in Andean South America, perhaps in the the north-central Andes. As with Solanaceae, the seasonally dry habitats where Tourrettieae, Argylia, clades that have reached North America seem to have Tecoma and Delostoma all occur today (Gentry, similar northern and southern latitudinal limits 1992a). Even though much of the Andes represent a

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 94 R. G. OLMSTEAD relatively recent physiographic province in South nieae and Catalpeae. Its origin is most likely New America (Antonelli et al., 2009; Antonelli & Sanmar- World and South American. Resolution at the base of tin, 2011), there is both phylogenetic and fossil evi- the Palaeotropical clade is poor and sampling of Asian dence for older plant diversification in the Andes, taxa, in particular, is limited, so it is uncertain especially for Andean dry forest clades (Pennington whether dispersal originally went east or west from et al., 2004, 2010; Särkinen et al., 2012). That being South America. said, Jacarandeae, which comprise the sister to the The two clades that originated first in the New rest of Bignoniaceae, are largely absent from the World and now have both New and Old World Andes, where only two species occur, and are found in members each have unique biogeographical histories. a variety of low-elevation ecosystems, including wet Crown Catalpeae originated in North America, with Atlantic coastal and Amazonian forests and dry or Chilopsis D.Don (and perhaps Astianthus D.Don) in fire-prone ecosystems such as cerrado and the edaphi- south-western USA and arid regions of Central cally dry coastal restingas (Gentry, 1992a). However, America, and Macrocatalpa (Griseb.) Britton in the limited resolution and support in the phylogenetic Antilles (Li, 2008; Olmstead et al., 2009) as succes- trees for Jacarandeae (R. Farias & L. Freitas, unpubl. sively closer sister groups to Catalpa Scop., which data) currently provide few insights into the histori- occurs in eastern North America and East Asia. The cal biogeography of this clade. The Tabebuia alliance Asian species of Catalpa represent two lineages (Li, is also found in diverse ecosystems, including wet 2008), suggesting either two dispersal events to East and dry forests, and savannas, including the cerrado. Asia, or one dispersal to Asia followed by dispersal Bignonieae are common elements of the liana flora in back to North America. wet tropical forests, including Amazonian and Atlan- Tecomeae are similar to Withaninae (Solanaceae), tic coastal forests, and probably originated in those in that a small clade of c. 12 genera and 50 species ecosystems, but have successfully colonized the dry, has dispersed to all habitable continents except fire-prone cerrado of Brazil, most probably once, but Europe (South and North America, Africa, Himalayan perhaps multiple times (Lohmann et al., 2012). Thus Asia, SE Asia, Australia). In the New World, the the history of diversification of Bignoniaceae appar- Tecomeae/Argylia clade is represented by Argylia and ently represents a complex of radiations within Campsidium Seem. in the southern Andes, Tecoma in biomes (e.g. Andean seasonally dry forests, wet tropi- the central/northern Andes (with one widespread cal forests, cerrado), punctuated by ecological shifts species extending into Central America), and Campsis into new habitats, including several lineages in the in eastern North America. Two dispersal events to cerrado (Jacaranda, Bignonieae, Tabebuia) that Asia are inferred to have given rise to the Old World mostly probably represent niche shifts from more taxa. One dispersal to Asia occurred in Campsis,in mesic habitats (Simon et al., 2009). Despite the which C. grandiflora (Thunb.) K.Schum. is sister to apparent ability of several clades of Bignoniaceae to C. radicans (L.) Seem. of eastern North America. The shift ecological habitats, very few lineages, perhaps second dispersal was most likely in the southern five, representing a handful of species in Tecomeae/ hemisphere and gave rise to the clade that includes Argylia, one in Tourrettieae, one in Bignonieae and c. the Himalayan Incarvillea Juss. and a radiation of 10 in Catalpeae occur outside the tropics or subtrop- about five genera in SE Asia and Australia. One ics, where cold tolerance is thought to present one of lineage of this clade has reached Africa [including the more difficult adaptive barriers to cross (Wiens & Tecomaria (Endl.) Spach, incorrectly included in Donoghue, 2004; Donoghue, 2008). Tecoma by some authors, e.g. Gentry (1992a)]. Another lineage is postulated to have dispersed back Transoceanic dispersal across the South Pacific from Australia to Chile where Five to six long-distance dispersal events need to be it is represented by the monotypic Campsidium. invoked to account for the Old World distribution of Campsidium represents the only hypothesized disper- Bignoniaceae (Fig. 6). Two clades of Bignoniaceae are sal back from the Old to the New World in Bignon- Old World in distribution, Oroxyleae, with four iaceae, Solanaceae or Verbenaceae. genera and six species restricted to SE Asia, and a larger Palaeotropical clade distributed in Africa, Aus- Association of seed dispersal with long-distance tralasia and Madagascar. As noted above, poor reso- dispersal lution renders the origin of Oroxyleae uncertain, Dispersal in Bignoniaceae is typically facilitated by although it undoubtedly represents an independent dry dehiscent fruits with winged seeds, which are dispersal event from the New World. The Palaeotropi- readily dispersed by wind, and which are found in cal clade is sister to the New World Tabebuia alliance, most members of the clade. However, three lineages with which it is part of a poorly resolved section of the in the family have animal-dispersed, indehiscent tree also including Oroxyleae and New World Bigno- fruits. None of these lineages originated in South

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 SOLANACEAE, VERBENACEAE AND BIGNONIACEAE 95

America (Crescentieae in Central America, Coleeae in ence between the northern and southern limits. In Madagascar and Kigelia DC. in Africa), and thus they the case of Lantaneae, the 10° latitude greater all represent lineages with a history of long-distance distribution in the southern hemisphere can be dispersal. However, seed dispersal of these taxa is accounted for by one species, Acantholippia seriphio- believed to be facilitated by mammals (Janzen & ides (Gray) Moldenke, which is the only species in the Martin, 1982; Zjhra et al., 2004; Grose & Olmstead, clade to occur south of 40°S. In total, approximately 2007a) and mammals are rarely thought to be vectors 65 species (c. 7.5% of New World diversity) occur of long-distance seed dispersal. The phylogeny of further than c. 35° north or south of the equator. Bignoniaceae suggests that each case could represent Glandularia, Verbena L. (Verbeneae) and Aloysia more recent in situ evolution of animal-dispersed (Lantaneae) all exhibit amphitropical disjunctions of fruits, after the lineage to which each group belongs several thousand kilometres between arid North and was established in the continent in which they occur South America, a pattern observed in numerous unre- today. Thus, all long-distance dispersal to the Old lated groups, suggesting long-distance dispersal World is inferred to have been by plants with seeds (Lewis & Oliver, 1961; Solbrig, 1972; Yuan & Olm- presumed to be wind dispersed. stead, 2008a; P. Lu-Irving & R. G. Olmstead, unpub- lished), although there is evidence for Verbena suggesting that a dispersal route along the Andes, VERBENACEAE where seasonal dry forests have existed for a long South American origin and diversification time in a series of discontinuous patches (Pennington, Every clade of Verbenaceae recognized by Marx Lavin & Oliveira-Filho, 2009), may be responsible et al. (2010) originated in South America and has (Marx et al., 2010; V. Thode & R. Olmstead, unpubl. its greatest species diversity in South America data). today. The initial diversification is inferred to have occurred in the tropics, where Petreeae are part of Niche conservatism vs. adaptation to novel the diverse Amazonian liana flora. The next three ecological zones in the diversification diverging clades, Duranteae, Casselieae and Cith- of Verbenaceae arexyleae, also occur primarily in tropical habitats As with Solanaceae and Bignoniaceae, diversification in wet tropical forests and Andean cloud forests. in Verbenaceae represents a complex of ecological Citharexylum L. exhibits a pattern of diversification shifts between ecosystems and radiations within eco- first in lowland wet tropical forests, and sub- systems. Verbenaceae most likely originated in wet sequently radiating in Andean montane forests tropical forests, but following a successful shift to arid (C. argutedentatum Moldenke and C. ilicifolium ecosystems in temperate South America (possibly Kunth in Fig. 4). This pattern appears to be coincident with the expansion of such habitats during repeated in Duranta L. (V. Thode & R. Olmstead, uplift of the Andes), diversified significantly in unpubl. data). However, Verbenaceae exhibit their those environments. Phylogenetic niche conservation greatest radiation in arid, temperate South America, apparently has dominated diversification following presumably as the Andean uplift contributed this shift to arid ecosystems. This can be seen in two to those environments in what is now Argentina rather different manifestations. In one, amphitropical (Blisniuk et al., 2004; Barreda & Palazzesi, 2007; disjunctions are found between arid zones of temper- Graham, 2010, 2011). Priveae, Neospartoneae, Ver- ate and subtropical North and South America in beneae and Lantaneae are all found today or Verbeneae and Lantaneae. The fact that these clades initially diversified primarily in such habitats. have hardly colonized geographically adjacent, more mesic habitats, yet have succeeded through dispersal Distribution in North America and the Caribbean in colonizing geographically distant, but ecologically Whereas all of the clades of Verbenaceae identified comparable arid habitats in both hemispheres is con- here originated in South America, only two are sistent with Donoghue’s (2008) ‘easier to move than to endemic to that continent, Neospartoneae, with three evolve’ hypothesis. genera and seven species, and Rhaphithamnus Miers, In another instance, the Lantana/Lippia L. complex with two species (one in the Patagonian Andes and has successfully colonized the relatively young, fire- one endemic to the Juan Fernández islands). The adapted savannas of central Brazil and eastern other seven clades all exhibit more or less continuous Bolivia, where they have diversified extensively in the distributions between South and North America. As cerrado (Lu-Irving & Olmstead, 2012). In contrast, in Solanaceae and Bignoniaceae, most of these clades Stachytarpheta Vahl. has its roots in the wet tropics reach similar latitudinal limits in the northern and during the early diversification of Verbenaceae, but southern hemispheres (Table 1; Fig. 5). Only two of has also diversified extensively following one or more these clades exhibit more than a 5° latitudinal differ- niche shifts into the Cerrado, where it co-occurs with

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 96 R. G. OLMSTEAD the Lippia/Lantana complex. The xeric vs. mesic rence of fleshy, animal-dispersed fruits in Verben- divide seems to present a strong adaptive barrier in aceae, only one of the long-distance, over-water Verbenaceae, with the repeated occurrence of phylo- dispersal events appears to coincide with a fleshy- genetically niche-conserved dry or wet-adapted line- fruited disperser (Lantana in Africa). Plants of Coe- ages across the family. For example, the central locarpum in Madagascar also have fleshy fruits, but Andes (northern Bolivia to central Ecuador) are home the lineage is inferred to have arisen from dry-fruited to groups of Verbenaceae that apparently arose in ancestors (Lu-Irving & Olmstead, 2012), so it is not arid regions of temperate South America (Verbeneae, possible to infer whether the dispersing colonizers Lantaneae), and which also occupy seasonally dry had fleshy or dry fruits. Similarly, the lineages with habitats at mid to high elevations, and to groups that amphitropical desert distributions are all dry-fruited arose in wet tropics (Duranta, Citharexylum), and groups (Glandularia, Verbena, Aloysia). Thus, to an which occur in wet Andean cloud forests today. This even greater extent than in Solanaceae, fleshy-fruited has been observed in many other Neotropical groups lineages are under-represented relative to dry-fruited that are diverse in arid habitats (reviewed by Pen- ones among long-distance dispersers. nington et al., 2004, 2009; Lavin, 2006). CONCLUSIONS Transoceanic dispersal Four clades of Verbenaceae have representatives Solanaceae, Bignoniaceae and Verbenaceae represent outside the New World and six long-distance dispersal three ecologically and floristically important South events are required to account for them (Fig. 6). Most American plant families, the origin and diversifica- of these Old World occurrences represent individual tion of which occurred primarily in situ. By breaking species or small clades within genera with primarily down each of the three families into constituent New World distributions and most likely represent clades, 37 in total, and looking at their geographical recent dispersal events. Verbena officinalis L. occurs distributions in the context of the phylogeny of each in Eurasia, but is the only member of Verbena found family, patterns emerge that contribute to a greater outside the New World. It is phylogenetically nested understanding of the historical development of South in a derived group of North American Verbena spp. American floristic diversity. and represents a relatively recent dispersal event to The patterns observed suggest that a large majority Europe (Marx et al., 2010). The other dispersal events of clades in each family, regardless of age, have suc- all follow a pattern of South America to Africa; ceeded in colonizing North America, which had no Lantana, Lippia and Priva Adans. all have a small contiguous land connection throughout most of the number of species in Africa. Chascanum E.Meyer is history of the South American continent since its found only in the Old World (Africa, Arabia, Indian split from Gondwana c. 100 Mya. Few apparent con- subcontinent), but is a close relative of South Ameri- straints on migration/dispersal seem to exist for can Bouchea Cham. and Stachytarpheta. Coelocar- plants in these groups in the western hemisphere pum Balf.f., a Madagascan genus of five species, is (Cody et al., 2010). Clade limits within the New World sister to the rest of Lantaneae, and thus represents a seem to be primarily ecological, rather than strictly relatively older dispersal event than those within geographical, suggesting that adaptive barriers that Lantana and Lippia, although Lantaneae themselves limited geographical spread and constrain mesic/xeric are a relatively recent group within Verbenaceae. All transitions have been important factors shaping the six inferred dispersal events appear to have been diversification of these three families, in line with from west to east. recent finding for other groups (Lavin et al., 2004; Donoghue, 2008). Inability to evolve cold tolerance Association of seed dispersal with seems to restrict most clades to within a range of c. long-distance dispersal 30–35°N/S latitude (Fig. 5), roughly coinciding with Dry fruits are inferred to be ancestral in Verbenaceae the 10 °C average temperature for the coldest month, (O’Leary et al., in press), but fleshy, putatively a placeholder for the frost-free zone on most conti- animal-dispersed fruits have arisen in several clades nents (Thompson et al., 2000). Only an estimated 5, 1 (Duranteae, Citharexyleae, Neospartoneae, Rhaphi- and 7.5% of New World Solanaceae, Bignoniaceae and thamnus, Lantaneae), including multiple times in the Verbenaceae, respectively, occur beyond c. 35°N/S, species-rich Lantaneae, in which Lantana, tradition- thus fitting the tropical niche conservation model ally defined by the presence of fleshy fruits, is (Wiens & Donoghue, 2004; Donoghue, 2008). Those polyphyletic (Lu-Irving & Olmstead, 2012). Fleshy clades that have succeeded in reaching high latitudes fruits are found in groups inhabiting both wet and dry tend to be species-rich (although most of the diversity habitats and represent about 33% of all species of is still tropical) and also to have colonized the Old Verbenaceae. Despite the relatively frequent occur- World much more often than the clades restricted to

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 SOLANACEAE, VERBENACEAE AND BIGNONIACEAE 97 lower latitudes. Bromeliaceae provide an interesting evolution is important in the history of each family, as contrast. They are even more highly constrained to it is likely to be in any large clade of ancient origin frost-free latitudes, where they are largely restricted (e.g. Fabaceae – Schrire, Lavin & Lewis, 2005) and to mesic ecosystems, with rare shifts to arid habitats; even limited niche shifting can lead, with time, to following such shifts, the resulting clades remain successful establishment in many biomes. constrained to arid ecosystems (Givnish et al., 2011). Despite the near universal success of multiple inde- Solanaceae and Verbenaceae also include notable pendent clades within each family to colonize North arid temperate radiations (e.g. Lycium in Solanaceae; America, in many cases most likely preceding the Aloysia, Glandularia and Verbena in Verbenaceae) closing of the Isthmus of Panama, long-distance dis- that occupy distantly disjunct distributions between persal to colonize Old World continents apparently deserts of North and South America. Amphitropical has been much more limited, and diversification fol- disjuncts of this kind are relatively common between lowing those events typically has been modest (these arid habitats of North and South America in coastal, are, by definition, relatively recent events). The spiny Mediterranean ecosystems in California/Chile and in solanums (Solanum section Leptostemonum) have interior arid habitats in the south-west USA and been the most successful diversification outside the northern Mexico/central and northern Argentina New World in any of these three families, with c. 175 (Raven, 1972; Wen & Ickert-Bond, 2009). Coastal, species in the Old World. Elsewhere in Solanaceae, Mediterranean climate disjuncts appear to be pre- Hyoscyameae have about 40 species in Eurasia, dominantly the result of north to south migration Lycium has about 35 Old World species distributed in events (Wen & Ickert-Bond, 2009), whereas the inte- Africa, Asia and Australia, and Anthocercideae and rior desert disjuncts exhibit a mix of both south to Nicotiana have about 30 and 16 species in Australia, north events (e.g. Larrea Cav. – Lia et al., 2001; respectively. In Bignoniaceae, the Palaeotropical Hoffmannseggia Cav. – Simpson, Tate & Weeks, clade has also diversified with c. 150 species, and 2005), and north to south events (e.g. Tiquilia Pers. – Tecomeae in Australasia and the Himalaya with c.40 Moore, Tye & Jansen, 2006; Astragalus L. – Scherson, species. In Verbenaceae, Chascanum (Duranteae) Vidal & Sanderson, 2008). The four lineages in includes about 25 species in Africa, Arabia and the Solanaceae and Verbenaceae in which this pattern is Indian subcontinent, whereas Lantana and Lippia found all represent South America to North America (Lantaneae) each have c. 15 species in Africa result- dispersals (Yuan & Olmstead, 2008a; J. Miller et al., ing from independent dispersal events (Lu-Irving & 2011; P. Lu-Irving & R. Olmstead, unpubl. data). Olmstead, 2012). Again, these patterns provide good evidence for The patterns exhibited by these three families, of large-scale phylogenetic niche or biome conservatism primary radiations within South America, and limited sensu Donoghue (2008). Despite the prevalence of colonization of the Old World continents by clades phylogenetic niche conservatism, major habitat shifts that have not themselves radiated extensively, have are also apparent within each family both at the similarities and one stark contrast with Asteraceae, origin of clades and within clades, resulting in another large clade that is thought to have originated present distributions for each family that include all and initially diversified in South America. In similar major biomes in South America. Indeed, niche shifts ways to the families described here, many of the early within clades may account for some impressive recent diverging lineages of Asteraceae also expanded to radiations, e.g. the diversification of Stachytarpheta colonize North America (Funk et al., 2005). However, (Duranteae: Verbenaceae) and Bignonieae (Bignon- the vast majority of Asteraceae are derived primarily iaceae) in the cerrado vegetation of Brazil. Although from a transatlantic colonization event that led to an evolution of the Cerrado has been shown to have explosive radiation in Africa, with subsequent spread involved recruitment of lineages from wet tropical and radiations on other continents, especially North forest, dry tropical forest and subtropical pampa America, from which many lineages eventually rec- ancestors (Simon et al., 2009), the diversification of olonized South America (Funk et al., 2005). Aster- the Lantana/Lippia complex in the cerrado represents aceae were tremendously successful at colonizing a diversification derived from arid-adapted ancestors temperate biomes, unlike any of the three families in temperate South America (Lu-Irving & Olmstead, considered here. Another large family, Bromeliaceae, 2012). Solanum seems to have adapted to virtually all in contrast, exhibited relatively limited colonization of habitable Neotropical ecosystems, even though it North America and only a single transatlantic disper- suffers the same limited ability to colonize cold tem- sal event (Givnish et al., 2011). perate regions as observed in all three families. The It is not possible with the present level of taxon fact that habitat outliers within clades are often the sampling in the phylogenetic analyses for these three exception suggests that niche conservation prevails families to infer all instances of dispersal from South to within most clades in all three families, but niche North America or all biome shifts and, without time-

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 98 R. G. OLMSTEAD calibrated trees, it is impossible to rigorously test insights on the origin and evolution of oceanic island plants. hypotheses of dispersal routes and timing (e.g. over- In: Soltis PS, Soltis DE, Doyle JJ, eds. Molecular systemat- land through the Isthmus of Panama vs. an over-water ics of plants II: DNA sequencing. Boston: Kluwer, 410– dispersal route). Work is planned or in progress on 441. dating phylogenies of Solanaceae (S. Knapp, pers. Baldwin BG, Wagner WL. 2010. Hawaiian angiosperm comm.) and Bignoniaceae (L. Lohmann, pers. comm.) radiations of North American origin. Annals of Botany 105: and is also anticipated in Verbenaceae. Dated phylog- 849–879. enies for these clades will permit more rigorous testing Barreda V, Palazzesi L. 2007. Patagonian vegetation turnovers during the Paleogene–Early Neogene: origin of of explicit biogeographical hypotheses of the sort arid-adapted floras. The Botanical Review 73: 31–50. described here on the basis of undated phylogenies and Bell CD, Donoghue MJ. 2005. Phylogeny and biogeography species distribution data. Enhanced phylogenies using of Valerianaceae (Dipsacales) with special reference to supermatrix and supertree approaches to increase the South American valerians. Organisms, Diversity and taxon sampling will also permit much finer-scale eco- Evolution 5: 147–159. logical assessment of niche conservation vs. ecological Blisniuk PM, Stern LA, Chamberlain CP, Idleman B, shifts in the history of diversification of these three Zeitler PK. 2004. Climatic and ecologic changes during iconic South American plant families. Miocene surface uplift in the southern Patagonian Andes. Earth and Planetary Science Letters 230: 125–142. Bohs L. 2005. Major clades in Solanum based on ndhF ACKNOWLEDGEMENTS sequence data. In: Keating RC, Hollowell VC, Croat TB, eds. A festchrift for William G. D’Arcy: the legacy of a taxonomist, Thanks to the many students and collaborators (in Vol. 104. Monographs in systematic botany. St. Louis: Mis- particular L. Bohs, L. Lohmann, S. Grose, N. O’Leary, souri Botanical Garden Press, 27–49. M. Múlgura) who participated in the phylogenetic Bohs L, Olmstead RG. 2001. A reassessment of Normania studies of Bignoniaceae, Solanaceae and Verbenaceae, and Triguera (Solanaceae). Plant Systematics and Evolution and to the many herbaria, Botanical Gardens and 228: 33–48. individuals who provided plant material for those Bremer B, Bremer K, Heidari N, Erixon P, Olmstead studies. Thanks to A. Antonelli and T. Pennington for RG, Anderberg AA, Kallersjo M, Barkhordarian E. inviting me to speak in the symposium at IBC 2011, 2002. Phylogenetics of asterids based on 3 coding and 3 to C. Stromberg and A. Graham for responses to non-coding chloroplast DNA markers and the utility queries, and to P. Lu-Irving, L. Lohmann, L. Bohs, of non-coding DNA at higher taxonomic levels. Molecular Colin Hughes, John Klicka and two anonymous Phylogenetics and Evolution 24: 274–301. reviewers for discussions on these issues and/or Bremer K, Friis EM, Bremer B. 2004. Molecular phyloge- comments on the manuscript. Funding for these netic dating of asterid flowering plants shows early Creta- studies came from numerous sources, including NSF ceous diversification. Systematic Biology 53: 496–505. grants DEB0309065, DEB0309065, DEB0542493, Burnham RJ, Graham A. 1999. The history of Neotropical DEB0710026 and EF-0431184 to R.G.O. vegetation: new developments and status. Annals of the Missouri Botanical Garden 86: 546–589. Carlquist S. 1967. The biota of long-distance dispersal. V. REFERENCES Plant dispersal to Pacific Islands. Bulletin of the Torrey Botanical Club 94: 129–162. Andersson L, Antonelli A. 2005. Phylogeny of the tribe Carlquist S. 1974. Island biology. New York: Columbia Cinchoneae (Rubiaceae), its position in Cinchonioideae, and University Press. description of a new genus, Ciliosemina. Taxon 54: 17–28. Christenhusz MJM, Chase MW. 2012. Biogeographical pat- Andersson S. 2006. On the phylogeny of the genus Calceo- terns of plants in the Neotropics – dispersal rather than plate laria (Calceolariaceae) as inferred from ITS and plastid tectonics is most explanatory. Botanical Journal of the matK sequences. Taxon 55: 125–137. Linnean Society 171: doi: 10.1111/j.1095-8339.2012.01301.x Antonelli A, Nylander JAA, Persson C, Sanmartin I. Clarkson JJ, Knapp S, Aoki S, Garcia VF, Olmstead RG, 2009. Tracing the impact of the Andean uplift on Neotropi- Chase MW. 2004. Phylogenetic relationships in Nicotiana cal plant evolution. Proceedings of the National Academy of (Solanaceae) inferred from multiple plastid DNA regions. Sciences of the United States of America 106: 9749–9754. Molecular Phylogenetics and Evolution 33: 75–90. Antonelli A, Sanmartin I. 2011. Why are there so many Coates AG, Obando JA. 1996. The geologic evolution of the plant species in the Neotropics? Taxon 60: 403–414. Central American isthmus. In: Jackson JBC, Budd AF, Atkins S. 2004. Verbenaceae. In: Kadereit JW, ed. The fami- Coates AG, eds. Evolution and environment in tropical lies and genera of flowering plants, Vol. 7. Berlin: Springer, America. Chicago: University of Chicago Press, 21–56. 449–468. Cody S, Richardson JE, Rull V, Ellis C, Pennington RT. Baldwin BG, Crawford DJ, Francisco-Ortega J, Kim 2010. The great American biotic interchange revisited. S-C, Sang T, Stuessy TF. 1998. Molecular phylogenetic Ecography 33: 326–332.

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 SOLANACEAE, VERBENACEAE AND BIGNONIACEAE 99

Crisp MD, Arroyo MTK, Cook LG, Gandolfo MA, Jordan framework for hypothesis testing. Trends in Ecology and GJ, McGlone MS, Weston PH, Westoby M, Wilf P, Evolution 27: 47–56. Linder HP. 2009. Phylogenetic biome conservatism on a Givnish TJ, Barfuss MHJ, Van Ee B, Riina R, Schulte K, global scale. Nature 458: 754–756. Horres R, Gonsiska PA, Jabaily RS, Crayn DM, Smith Davis CC, Bell CD, Mathews S, Donoghue MJ. 2002. AC, Winter K, Brown GK, Evans TM, Holst BK, Luther Laurasian migration explains Gondwanan disjunctions: evi- H, Till W, Zizka G, Berry PE, Sytsma KJ. 2011. Phyl- dence from Malpighiaceae. Proceedings of the National ogeny, adaptive radiation, and historical biogeography in Academy of Sciences of the United States of America 99: Bromeliaceae: insights from an eight-locus plastid phylog- 6833–6837. eny. American Journal of Botany 98: 872–895. Dillon MO, Tu T, Quipuscoa Silvestre V, Wen J. 2009. Givnish TJ, Millam KC, Evans TM, Hall JC, Pires JC, Biogeographic diversification in Nolana (Solanaceae), a Berry PE, Sytsma KJ. 2004. Ancient vicariance or ubiquitous member of the Atacama and Peruvian Deserts recent long-distance dispersal? Inferences about phylogeny along the western coast of South America. Journal of and South American-African disjunctions in Rapateaceae Systematics and Evolution 47: 457–476. and Bromeliaceae based on ndhF sequence data. Dillon MO, Tu T, Soejima A, Yi T, Nie Z, Tye A, Wen J. International Journal of Plant Sciences 165 (Suppl): S35– 2007. Phylogeny of Nolana (Nolaneae, Solanoideae, S54. Solanaceae) as inferred from granule-bound starch synthase Goldblatt P. 1993. Biological relationships between Africa I (GBSSI) sequences. Taxon 56: 1000–1011. and South America. New Haven: Yale University Press. Donoghue MJ. 2008. A phylogenetic perspective on the dis- Graham A. 2010. Late Cretaceous and Cenozoic history of tribution of plant diversity. Proceedings of the National Latin American vegetation and terrestrial environments. St. Academy of Sciences of the United States of America 105: Louis: Missouri Botanic Garden Press. 11549–11555. Graham A. 2011. The age and diversification of terrestrial Edwards EJ, Nyffeler R, Donoghue MJ. 2005. Basal New World ecosystems through Cretaceous and Cenezoic cactus phylogeny: implications of Pereskia (Cactaceae) para- time. American Journal of Botany 98: 336–351. phyly for the transition to the cactus life form. American Grose SO, Olmstead RG. 2007a. Evolution of a charismatic Journal of Botany 92: 1177–1188. Neotropical clade: molecular phylogeny of Tabebuia s.l., Erkens RHJ, Maas JW, Couvreur TLP. 2009. From Africa Crescentieae, and allied genera (Bignoniaceae). Systematic via Europe to South America: migrational route of a species- Botany 32: 650–659. rich genus of Neotropical lowland rain forest trees (Guatte- Grose SO, Olmstead RG. 2007b. Taxonomic revisions in the ria, Annonaceae). Journal of Biogeography 36: 2338–2352. polyphyletic genus Tabebuia s.l. (Bignoniaceae). Systematic Farris DW, Jaramillo CA, Bayona GA, Restrepo-Moreno Botany 32: 660–670. SA, Montes C, Cardona A, Mora A, Speakman RJ, Harvey PH, Pagel MD. 1991. The comparative method in Glasscock MD, Reiners P, Valencia V. 2011. Fracturing evolutionary biology. New York: Oxford University Press. of the Panamanian isthmus during initial collision with Heiser CB. 1969. Nightshades, the paradoxical plants. San South America. Geology 39: 1007–1010. Francisco: Freeman and Company. Fischer E, Theisen I, Lohmann LG. 2004. Bignoniaceae. Hughes C, Eastwood R. 2006. Island radiation on a In: Kadereit JW, ed. The families and genera of flowering continental scale: exceptional rates of plant diversification plants, Vol. 7. Berlin: Springer, 9–38. after uplift of the Andes. Proceedings of the National Funk VA, Bayer RJ, Keeley S, Chan R, Watson L, Academy of Sciences of the United States of America 103: Gemeinholzer B, Schilling E, Panero JL, Baldwin BG, 10334–10339. Garcia-Jacas N, Susanna A, Jansen RK. 2005. Every- Hunziker AT. 2001. Genera solanacearum: the genera of where by Antarctica: using a supertree to understand the Solanaceae illustrated, arranged according to a new system. diversity and distribution of the Compositae. Biologiske Ruggell: A.R.G. Gantner Verlag. Skrifter 55: 343–373. Janzen DH, Martin PS. 1982. Neotropical anachronisms – Gentry AH. 1980. Bignoniaceae partI–Tribes Crescentieae the fruits the gomphotheres ate. Science 215: 19–27. and Tourretieae. Flora Neotropica Monograph 25: 1–131. Keigwin L. 1982. Isotopic paleoceanography of the Caribbean Gentry AH. 1982. Neotropical floristic diversity: phytogeo- and East Pacific: role of Panama uplift in Late Neogene graphical connections between Central and South America, time. Science 217: 350–353. Pleistocene climatic fluctuations, or an accident of the Lavin M. 2006. Floristic and geographical stability of discon- Andean orogeny? Annals of the Missouri Botanical Garden tinuous seasonally dry tropical forests explains patterns of 69: 557–593. plant phylogeny and endemism. In: Pennington RT, Ratter Gentry AH. 1992a. Bignoniaceae – part II (Tribe Tecomeae). JA, Lewis GP, eds. Neotropical savannas and seasonally dry Flora Neotropica Monograph 25: 1–370. forests: plant biodiversity, biogeography and conservation. Gentry AH. 1992b. A synopsis of Bignoniaceae ethnobotany Boca Raton: CRC Press, 433–437. and economic botany. Annals of the Missouri Botanical Lavin M, Herendeen PS, Wojciechowski MF. 2005. Evo- Garden 79: 53–64. lutionary rates analysis of Leguminosae implicates a rapid Gillespie RG, Baldwin BG, Waters JM, Fraser CI, diversification of lineages during the Tertiary. Systematic Nikula R, Roderick GK. 2012. Long-distance dispersal: a Biology 54: 575–594.

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 100 R. G. OLMSTEAD

Lavin M, Schrire B, Lewis G, Pennington T, Delgado- and classification of Verbenaceae. American Journal of Salinas A, Thulin M, Hughes C, Beyra Matos A, Botany 97: 1647–1663. Wojciechoswki MF. 2004. Metacommunity process rather McDade LA, Daniel TF, Kiel C. 2008. Toward a compre- than continental tectonic history better explains geographi- hensive understanding of phylogenetic relationships among cally structured phylogenies in legumes. Philosophical lineages of Acanthaceae s.l. (Lamiales). American Journal of Transactions of the Royal Society London (Biological Sci- Botany 95: 1136–1152. ences) 359: 1509–1522. McDade LA, Daniel TF, Kiel CA, Borg AJ. 2012. Lavin M, Wojciechowski MF, Gasson P, Hughes C, Phylogenetic placement, delineation, and relationships Wheeler E. 2003. Phylogeny of robinioid legumes among genera of the enigmatic Nelsonioideae (Lamiales: (Fabaceae) revisited: Coursetia and Gliricidia recircum- Acanthaceae). Taxon 61: 637–651. scribed, and a biogeographical appraisal of the Caribbean Miller JS, Kamath A, Damashek J, Levin RA. 2011. Out endemics. Systematic Botany 28: 387–409. of America to Africa or Asia: inference of dispersal histories Levin RA, Myers NR, Bohs L. 2006. Phylogenetic relation- using nuclear and plastid DNA and the S-RNase self- ships among the ‘spiny solanums’ (Solanum subgenus Lep- incompatibility locus. Molecular Biology and Evolution 28: tostemonum, Solanaceae). American Journal of Botany 93: 793–801. 157–169. Miller JS, Kamath A, Levin RA. 2009. Do multiple tortoises Levin SA, Muller-Landau HC, Nathan R, Chave J. 2003. equal a hare? The utility of nine noncoding plastid regions The ecology and evolution of seed dispersal: a theoretical for species-level phylogenetics in tribe Lycieae (Solanaceae). perspective. Annual Review of Ecology, Evolution, and Sys- Systematic Botany 34: 796–804. tematics 34: 575–604. Miller RJ, Mione T, Phan H-L, Olmstead RG. 2011. Color Lewis WH, Oliver RL. 1961. Cytogeography and phylogeny by numbers: nuclear waxy gene phylogeny of Jaltomata of the North American species of Verbena. American Journal (Solanaceae), the sister genus to Solanum, supports three of Botany 48: 638–643. clades differing in fruit color. Systematic Botany 36: 153– Li J. 2008. Phylogeny of Catalpa (Bignoniaceae) inferred from 162. sequences of chloroplast ndhF and nuclear ribosomal DNA. Montes C, Cardona A, McFadden R, Morón SE, Silva CA, Journal of Systematics and Evolution 46: 341–348. Restrepo-Moreno RDA, Hoyos N, Wilson J, Farris D, Lia VV, Confalonieri VA, Comas CO, Hunziker JH. 2001. Bayona GA, Jaramillo CA, Valencia V, Bryan J, Flores Molecular phylogeny of Larrea and its allies (Zygophyl- JA. 2012. Evidence for middle Eocene and younger land laceae): reticulate evolution and the probable time of creo- emergence in central Panama: implications for Isthmus sote bush arrival to North America. Molecular Phylogenetics closure. Geological Society of America Bulletin 124: 780– and Evolution 21: 309–320. 799. Lohmann LG. 2006. Untangling the phylogeny of Neotropi- Moore MJ, Tye A, Jansen RK. 2006. Patterns of long- cal lianas (Bignonieae, Bignoniaceae). American Journal of distance dispersal in Tiquilia subg. Tiquilia (Boraginaceae): Botany 93: 304–318. implications for the origins of amphitropical disjuncts and Lohmann LG. In press. A new generic classification of Galápagos islands endemics. American Journal of Botany Bignonieae (Bignoniaceae) based on molecular phylogenetic 93: 1163–1177. data and morphological synapomorphies. Annals of the Mis- Morley R. 2003. Interplate dispersal paths for megathermal souri Botanical Garden. angiosperms. Perspectives in Plant Ecology, Evolution and Lohmann LG, Ulloa CU. 2007. Bignoniaceae in iPlants Systematics 6: 5–20. prototype checklist. [online]. Available at: http://www. Nathan R, Schurr FM, Spiegel O, Steinitz O, Trakhten- iplants.org [accessed 15 December 2011]. brot A, Tsoar A. 2008. Mechanisms of long-distance seed Lohmann LG, Bell CD, Calió MF, Winkworth RC. 2012. dispersal. Trends in Ecology and Evolution 23: 638–647. Pattern and timing of biogeographic history in the Neotropi- Nie Z-L, Sun H, Beardsley PM, Olmstead RG, Wen J. cal tribe Bignonieae (Bignoniaceae). Botanical Journal of 2006. Evolution of biogeographic disjunction between the Linnean Society 171: doi: 10.1111/j.1095-8339.2012. eastern Asia and eastern North America in Phryma (Phry- 01311.x. maceae). American Journal of Botany 93: 1343–1356. Lu-Irving P, Olmstead RG. 2012. Investigating the O’Connor JM, Duncan RA. 1990. Evolution of the Walvis evolution of Lantaneae (Verbenaceae) using multiple loci. Ridge-Rio Grande Rise Hot Spot System: implications for Botanical Journal of the Linnean Society 171: doi: 10.1111/ African and South American plate motions over plumes. j.1095-8339.2012.01305.x. Journal of Geophysical Research 95: 17475–17502. Magallón S, Castillo A. 2009. Angiosperm diversifica- O’Leary N, Calviño CI, Martinez S, Olmstead RG, tion through time. American Journal of Botany 96: 349– Múlgura ME. In press. Evolution of morphological traits 365. in Verbenaceae. American Journal of Botany. Marshall LG, Webb DD, Sepkoski JJ, Raup DM. 1982. O’Leary N, Yuan Y-W, Chemisquy A, Olmstead RG. 2009. Mammalian evolution and the Great American Interchange. Reassignment of species of paraphyletic Junellia s.l. to Science 215: 1351–1357. the new genus Mulguraea (Verbenaceae) and new circum- Marx HE, O’Leary N, Yuan Y-W, Lu-Irving P, Tank DC, scription of genus Junellia: molecular and morphological Múlgura M, Olmstead RG. 2010. A molecular phylogeny congruence. Systematic Botany 34: 777–786.

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 SOLANACEAE, VERBENACEAE AND BIGNONIACEAE 101

Olmstead RG, Bohs L. 2007. A summary of molecular sys- Pirie MD, Chatrou LW, Mols JB, Erkens RHJ, Oosterhof tematic research in Solanaceae: 1982–2006. In: Spooner J. 2006. ‘Andean-centered’ genera in the short-branch clade DM, Bohs L, Govannoni J, Olmstead RG, Shibata D, eds. of Annonaceae: testing biogeographical hypotheses using Solanaceae VI: genomics meets biodiversity. Proceedings of phylogeny reconstruction and molecular dating. Journal of the Sixth International Solanaceae Conference. Acta Horti- Biogeography 33: 31–46. culturae 745, Leuven: International Society for Horticul- Poczai P, Hyvönen J, Symon DE. 2011. Phylogeny of tural Science, 255–268. kangaroo apples (Solanum subg. Archaesolanum, Olmstead RG, Bohs L, Abdel Migid H, Santiago-Valentin Solanaceae). Molecular Biology Reports 38: 5243–5259. E, Garcia VF, Collier SM. 2008. A molecular phylogeny of Raven P. 1972. Plant species disjunctions: a summary. the Solanaceae. Taxon 57: 1159–1181. Annals of the Missouri Botanical Garden 59: 234–246. Olmstead RG, Palmer JD. 1992. A chloroplast DNA phyl- Renner SS. 2004. Plant dispersal across the tropical Atlantic ogeny of the Solanaceae: subfamilial relationships and char- by wind and sea currents. International Journal of Plant acter evolution. Annals of the Missouri Botanical Garden Sciences 165 (Suppl): S23–S33. 79: 346–360. Renner SS, Clausing G, Meyer K. 2001. Historical bioge- Olmstead RG, dePamphilis CW, Wolfe AD, Young ND, ography of Melastomataceae: the roles of Tertiary migration Elisens WJ, Reeves PA. 2001. Disintegration of the Scro- and long-distance dispersal. American Journal of Botany 88: phulariaceae. American Journal of Botany 88: 348–361. 1290–1300. Olmstead RG, Sweere JA. 1994. Combining data in phylo- Richardson JE, Chatrou LW, Mols JB, Erkens RHJ, genetic systematics: an empirical approach using three Pirie MD. 2004. Historical biogeography of two cosmopoli- molecular data sets in the Solanaceae. Systematic Biology tan families of flowering plants: Annonaceae and Rham- 43: 467–481. naceae. Philosophical Transactions of the Royal Society Olmstead RG, Sweere JA, Spangler RE, Bohs L, Palmer London (Biological Sciences) 359: 1495–1508. JD. 1999. Phylogeny and provisional classification of the Richardson JE, Pennington RT, Pennington TD, Holl- Solanaceae based on chloroplast DNA. In: Nee M, Symon ingsworth PM. 2001. Rapid diversification of a species-rich DE, Jessup JP, Hawkes JG, eds. Solanaceae IV, Advances in genus of Neotropical rain forest trees. Science 293: 2242– biology and utilization. Kew: Royal Botanic Gardens, 111– 2245. 137. Santiago-Valentin E, Olmstead RG. 2003. Phylogenetics of Olmstead RG, Zjhra ML, Lohmann LG, Grose SO, Eckert the Antillean Goetzeoideae (Solanaceae) and their relation- AJ. 2009. A molecular phylogeny and classification of Bigno- ships within the Solanaceae based on chloroplast and ITS niaceae. American Journal of Botany 96: 1731–1743. DNA sequence data. Systematic Botany 28: 452–460. Paape T, Igic B, Smith SD, Olmstead RG, Bohs L, Kohn Särkinen TE, Newman MF, Maas PJM, Maas H, Poulsen JR. 2008. A 15-Myr-old genetic bottleneck. Molecular AD, Harris DJ, Richardson JE, Clark A, Hollings- Biology and Evolution 25: 655–663. worth M, Pennington RT. 2007. Recent oceanic long- Pennington RT, Dick CW. 2004. The role of immigrants in distance dispersal and divergence in the amphi-Atlantic the assembly of the South American rainforest tree flora. rain forest genus Renealmia L.f. (Zingiberaceae). Molecular Philosophical Transactions of the Royal Society London Phylogenetics and Evolution 44: 968–980. (Biological Sciences) 359: 1611–1622. Särkinen TE, Pennington RT, Lavin M, Simon MF, Pennington RT, Lavin M, Oliveira-Filho A. 2009. Woody Hughes CE. 2012. Evolutionary islands in the Andes: per- plant diversity, evolution, and ecology in the tropics: per- sistence and isolation explain high endemism in Andean spectives from seasonally dry tropical forests. Annual dry tropical forests. Journal of Biogeography 39: 884– Review of Ecology, Evolution, and Systematics 40: 437– 990. 457. Schäferhoff B, Fleishmann A, Fischer E, Albach DC, Pennington RT, Lavin M, Prado DE, Pendry CA, Pell Borsch T, Heubl G, Müller KF. 2010. Towards resolving SK, Butterworth C. 2004. Historical climate change and Lamiales relationships: insights from rapidly evolving chlo- speciation: neotropical seasonally dry forest plants show roplast sequences. BMC Evolutionary Biology 10: 352. doi: patterns of both Tertiary and Quaternary diversification. 10.1186/1471.2148-10-352. Philosophical Transactions of the Royal Society London Scherson RA, Vidal R, Sanderson MJ. 2008. Phylogeny, (Biological Sciences) 359: 515–537. biogeography, and rates of diversification of New World Pennington RT, Lavin M, Särkinen T, Lewis GP, Klit- Astragalus (Leguminosae) with an emphasis on South gaard BB, Hughes CE. 2010. Contrasting plant diversifi- American radiations. American Journal of Botany 95: 1030– cation histories within the Andean biodiversity hotspot. 1039. Proceedings of the National Academy of Sciences of the Schrire BD, Lavin M, Lewis GP. 2005. Global distribution United States of America 107: 13783–13787. patterns of the Leguminosae: insights from recent phylog- Perret M, Chautems A, Araujo AO, Salamin N. 2012. enies. Biologiske Skrifter 55: 375–422. Temporal and spatial origin of the Gesneriaceae in the New Simon MF, Grether G, de Queiroz LP, Skema C, Pen- World inferred from plastid DNA sequences. Botanical nington RT, Hughes CE. 2009. Recent assembly of the Journal of the Linnean Society 171: doi: 10.1111/j.1095- cerrado, a Neotropical plant diversity hotspot, by in situ 8339.2012.01303.x evolution of adaptations to fire. Proceedings of the National

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102 102 R. G. OLMSTEAD

Academy of Sciences of the United States of America 106: important trees and shrubs in North America. Washington: 20359–20364. United States Geological Survey Professional Paper 1650-A. Simpson BB, Tate JA, Weeks A. 2005. The biogeography of Wallander E, Albert VA. 2000. Phylogeny and classification Hoffmannseggia (Leguminosae, Caesalpinoideae, Caes- of Oleaceae based on rps16 and trnL-F sequence data. alpinieae): a tale of many travels. Journal of Biogeography American Journal of Botany 87: 1827–1841. 32: 15–27. Weber A. 2004. Gesneriaceae. In: Kadereit JW, ed. The fami- Simpson GG. 1980. Splendid isolation: the curious history of lies and genera of flowering plants. Berlin: Springer, 7: South American mammals. New Haven: Yale University 63–158. Press. Weese TL, Bohs L. 2007. A three gene phylogeny of Smith BT, Klicka J. 2010. The profound influence of the the genus Solanum (Solanaceae). Systematic Botany 32: Late-Pliocene Panamanian uplift on the exchange, diversi- 445–463. fication, and distribution of New World Birds. Ecography 33: Wen J, Ickert-Bond SM. 2009. Evolution of the Madrean– 333–342. Tethyan disjunctions and the North and South American Smith JF, Wolfram JC, Brown KD, Carroll CL, Denton amphitropical disjunctions in plants. Journal of Systematics DS. 1997. Tribal relationships in the Gesneriaceae: and Evolution 47: 331–348. evidence from DNA sequences of the chloroplast gene ndhF. Whitson M, Manos PS. 2005. Untangling Physalis Annals of the Missouri Botanical Garden 84: 50–66. (Solanaceae) from the physaloids: a two-gene phylogeny of Solbrig OT. 1972. The floristic disjunctions between the the Physalinae. Systematic Botany 30: 216–230. ‘Monte’ in Argentina and the ‘Sonora Desert’ in Mexico and Wiens JJ, Donoghue MJ. 2004. Historical biogeography, the United States. Annals of the Missouri Botanical Garden ecology and species richness. Trends in Ecology and Evolu- 59: 218–223. tion 19: 639–644. Soltis DE, Smith SA, Cellinese N, Wurdack KJ, Tank DC, Wikström N, Savolainen V, Chase MW. 2001. Evolution of Brockington SF, Refulio-Rodriguez NF, Walker JB, the angiosperms: calibrating the family tree. Proceedings of Moore MJ, Carlsward BS, Bell CD, Latvis M, Crawley the Royal Society of London. Series B, Biological Sciences S, Black C, Diouf D, Xi Z, Gitzendanner MA, Systma 268: 2211–2220. KJ, Qiu Y-L, Hilu KW, Manchester SR, Davis CC, Wortley AH, Harris GJ, Scotland RW. 2007. On the tax- Sanderson MJ, Olmstead RG, Judd WS, Donoghue onomy and phylogenetic position of Thomandersia. System- MJ, Soltis PS. 2011. Angiosperm phylogeny: 17-genes, 640 atic Botany 32: 415–444. taxa. American Journal of Botany 98: 704–730. Yuan Y-W, Liu C, Marx HE, Olmstead RG. 2010. An Soza VL, Olmstead RG. 2010a. Molecular systematics of empirical demonstration of using PPR (pentatricopeptide tribe Rubieae (Rubiaceae): evolution of major clades, devel- repeat) genes as phylogenetic tools: phylogeny of Verben- opment of leaf-like whorls, and biogeography. Taxon 59: aceae and the Verbena complex. Molecular Phylogenetics 755–771. and Evolution 54: 23–35. Soza VL, Olmstead RG. 2010b. Evolution of breeding Yuan Y-W, Olmstead RG. 2008a. A species-level phyloge- systems and fruits in New World Galium and relatives netic study of the Verbena complex (Verbenaceae) indicates (Rubiaceae). American Journal of Botany 97: 1630–1646. two independent intergeneric chloroplast transfers. Molecu- Spangler RE, Olmstead RG. 1999. Phylogenetic analysis of lar Phylogenetics and Evolution 48: 23–33. the Bignoniaceae based on the cpDNA gene sequences rbcL Yuan Y-W, Olmstead RG. 2008b. Evolution and phylogenetic and ndhF. Annals of the Missouri Botanical Garden 86: utility of the PHOT gene duplicates in a recently diversified 33–46. group: dramatic intron size variation and footprint of ances- Stefanovic S, Krueger LE, Olmstead RG. 2002. tral recombination. American Journal of Botany 95: 1166– Monophyly of the Convolvulaceae and circumscription 1176. of their major lineages based on DNA sequences of multiple Zimmer EA, Roalson EH, Skog LE, Boggan JK, Idnurm chloroplast loci. American Journal of Botany 89: 1510– A. 2002. Phylogenetic relationships in the Gesnerioideae 1522. (Gesneriaceae) based on nrDNA ITS and cpDNA trnL-F Stern S, Agra M, Bohs L. 2011. Molecular delimitation of and trnE-T spacer region sequences. American Journal of clades within New World species of the ‘spiny solanums’ Botany 89: 296–311. (Solanum subg. Leptostemonum). Taxon 60: 1429–14441. Zjhra ML, Sytsma KJ, Olmstead RG. 2004. Delimitation of Symon DE. 1994. Kangaroo apples. Adelaide: D. Symon. tribe Coleeae and implications for fruit evolution in Bigno- Thompson RS, Anderson KH, Bartlein PJ. 2000. Atlas of niaceae inferred from a chloroplast DNA phylogeny. Plant relations between climatic parameters and distributions of Systematics and Evolution 245: 55–67.

© 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 171, 80–102