Agneta Julia Borg

Home , Avicennia, Crossandra, Elytraria, Mendoncia, Nelsonioideae, Pseudocalyx

Evolutionary relationships in Thunbergioideae and other early branching lineages of Acanthac e a e Agneta Julia Borg

Evolutionary relationships in Thunbergioideae and other early branching lineages of Acanthaceae

Agneta Julia Borg

©Agneta Julia Borg, Stockholm 2012

Cover illustration: From top left, Mendoncia retusa, Thunbergia convolvulifolia, Pseudocalyx saccatus, Crossandra strobilifera, Avicennia bicolor, Elytraria marginata. Photos: Agneta Julia Borg and Jürg Schönenberger.

ISBN 978-91-7447-445-9

Printed in Sweden by Universitetsservice US-AB, Stockholm 2012 Distributor: Department of Botany, Stockholm University

Academic dissertation for the degree of Doctor of Philosophy in Plant Sys- tematics presented at Stockholm University 2012

Abstract Borg, A.J. 2012. Evolutionary relationships in Thunbergioideae and other early branching lineages of Acanthaceae.

Acanthaceae as circumscribed today consists of the three subfamilies Acan- thoideae (Acanthaceae sensu stricto), Thunbergioideae and Nelsonioideae, plus the genus Avicennia. Due to the morphological dissimilarities of Thun- bergioideae and Nelsonioideae, the delimitation of the family has been controversial. The mangrove genus Avicennia was only recently associated with Acanthaceae for the first time, based on molecular evidence, but without morphological support. In this thesis, phylogenetic analyses of nuclear and chloroplast DNA sequences were used to test the monophyly and exact posi- tions of Thunbergioideae and Nelsonioideae, and to infer detailed phylogenetic relationships within these subfamilies and among major lineages of Acanthaceae. Floral structure and development were comparatively studied in Avicennia and other Acanthaceae using scanning electron microscopy and stereo microscopy. Phylogenetic analyses strongly support monophyly of Thunbergioideae and Nelsonioideae, and place the latter clade with strong support as sister to all other plants treated as Acanthaceae. Thunbergioideae and Avicennia are moderately supported as sister taxa, and together they are sister to Acanthoideae. The general morphology of Avicennia can be easily accommodated in the Acanthaceae, and three synapomorphies support the suggested sister group relationship of Avicennia and Thunbergioideae: (1) collateral ovule arrangement, (2) vertical orientation of ovule curvature, and (3) an exposed nucellus that is contiguous with the ovary wall. Within Thun- bergioideae and Nelsonioideae, support values for major lineages are generally high. With some exceptions, the constituent genera are supported as monophyletic. Evolutionary relationships among and within genera are discussed in a morphological and biogeographical context.

Preface

This thesis is based on the following papers, referred to in the text by their Roman numerals.

I Borg, A.J., Schönenberger, J., and McDade, L.A. 2008. Mo- lecular phylogenetics and morphological evolution of Thun- bergioideae (Acanthaceae). Taxon 57: 811-822.

II Borg, A.J., and Schönenberger, J. 2011. Comparative floral development and structure of the black mangrove genus Avi- cennia L. and related taxa in the Acanthaceae. International Journal of Plant Sciences 172: 330-344.

III McDade, L.A., Daniel, T.F., Kiel, C.A., and Borg, A.J. Phy- logenetic placement, delineation, and relationships among genera of the enigmatic Nelsonioideae (Lamiales: Acanthace- ae). Accepted by Taxon.

IV Borg, A.J., and Schönenberger, J. Phylogenetic relationships in Acanthaceae based on nuclear and chloroplast sequences with particular focus on the Thunbergioideae. Manuscript.

Published papers are reproduced with permission from the publishers.

Papers I, II and IV were written by AJB with comments and suggestions from the co-authors. The papers were planned in collaboration with the co- authors. AJB has generated all sequences for Paper I, and all new sequences for Paper IV, and is responsible for the phylogenetic analyses in these papers. AJB has conducted the morphological analyses for Paper II in coopera- tion with the co-author. AJB took part in the preparation, laboratory work and writing of the manuscript of Paper III.

Contents

Introduction 11

Aims 15

Materials and methods 15

Results and discussion 17

Concluding remarks 21

Svensk sammanfattning (Swedish summary) 22

Acknowledgements 25

Literature cited 26

INTRODUCTION The flowering plant family Acanthaceae (Lamiales) consist of more than 4000 predominantly tropical and subtropical species. The vast majority of these belong to the subfamily Acanthoideae (sensu Scot- land & Vollesen 2000). Acanthoideae are characterized by having explosively dehiscent capsules with the few seeds borne on hook-like outgrowths of the funiculus, called retinacula. Retinacula are synapo- morphic for the subfamily, and Acanthoideae constitute a clearly de- fined and recognized group. In addition to these “core” Acanthaceae, two small lineages, the Thunbergioideae and Nelsonioideae sensu Scotland & Vollesen (2000), have traditionally been associated with Acanthaceae, though their markedly different morphology has resulted in various suggestions on how to classify these groups. Several authors have treated them outside Acanthaceae, and in some cases as distinct families (e.g. van Tieghem, 1908; Bremekamp, 1953; Dahlgren, 1980; Cronquist, 1981). In more recent years, the close relationship of thunbergioids and nelsonioids with core Acanthaceae has gained wide support from both molecular and morphological studies, although molecular studies have not included more than a few representatives of the two former subfamilies (e.g. Hedrén & al., 1995; McDade & Moody, 1999; Schönenberger & Endress 1998). More sur- prisingly, a molecular study by Schwarzbach & McDade (2002) showed that the black mangrove genus Avicennia, usually treated either as a family of its own in Lamiales, or within Verbenaceae, most likely belong in Acanthaceae.

Thunbergioideae The Thunbergioideae comprise five genera, the largest of which, Thunbergia, contains about 100 species restricted to tropical and subtropical regions of Africa, Madagascar, Asia, and Australia (Fig. 1). The second largest genus, Mendoncia (~60 species), is most diverse in Central and South America with some African representatives and a few species native to Madagascar (Fig. 1I). The remaining genera are Pseudocalyx with currently 7 species in Africa and Madagascar, monotypic Anomacanthus in tropical Africa, and monotypic Meyenia in India and Sri Lanka. The subfamily is characterized by a predominantly twining habit, enlarged prophylls (sensu Endress 2010), and a reduced calyx (Fig. 1). Furthermore, Thunbergioideae lack the retinacu- late fruits found in Acanthoideae. Flowers are often showy, and several Thunbergia species are cultivated as ornamentals. Particularly Thunbergia, but also Mendoncia, display a diversity of flower shapes

Fig. 1. Characteristics of Thunbergioideae, Avicennia, and Nelsonioideae. A-C, habitat; A, tropical forest clearance (e.g. Thunbergia convolvulifolia in Thunbergioideae); B, sunny space in rainforest near running water (e.g. Mendoncia retusa in Thunbergioideae); C, tropical coastal intertidal region (Avicennia). D-G, habit; D, small erect shrub (e.g. T. usambarica); twining around other plants (e.g. M. retusa); F, tree (A. bicolor); G, erect herb (E. imbricata in Nelsonioideae). H-K, flower morphology; H, corolla tube partially covered by prophylls, T. natalensis; I, corolla almost completely covered by prophylls, M. retusa; J, A. bicolor; K, E. imbricata. L-O, fruit types; L, immature capsule with persistent reduced calyx (T. convolvulifolia); M, drupe with persistent annular calyx and persistent prophylls (M. retusa); N, viviparous, leathery capsules (A. marina); O, dehisced capsule (Staurogyne merguensis Kuntze, Larsen & al. 44309, AAU). P-T, pollination syndromes and floral features; P, bee pollination

12 syndrome (T. grandiflora); Q, hawkmoth-pollination syndrome (T. guerkeana, Bally 14996, K); R, bird pollination syndrome (T. mysorensis); S, close-up of thecal awns (T. mysorensis); T, bristled anthers (M. retusa). and colors, reflecting adaptation to a variety of pollinators, including bees (e.g. in T. grandiflora, Faegri & van der Pijl, 1966; Fig. 1P), moths (T. guerkeana, Schönenberger, 1999; Fig. 1Q), birds (e.g. T. mysorensis, T. coccinea, Schönenberger, 1999; M. velloziana, Buzato & al. 2000, Snow & Teixeira 1982; Fig. 1R, S) and possibly even bats (Mendoncia, Vogel & al. 2004; M. bivalvis, M. glabra, Hequet, 2003.) There are no existing subgeneric classifications for Mendoncia or Pseudocalyx (but see Profice, 1988, for Mendoncia in Brazil). More attention has been directed towards Thunbergia. Lindau (1893) subdi- vided Thunbergia into four sections based on morphology and arrangement of flowers. Bremekamp (1955) revised and extended Lindau’s subdivision and proposed eight subgenera. His classification largely concurs with a recent study of floral development and structure in Thunbergia by Schönenberger (1999).

Avicennia The black mangrove genus Avicennia consists of around 8 species of trees which grow in the intertidal zone of coastal mangrove forests, ranging widely throughout tropical and warm temperate regions of the world (Tomlinson, 1995; Fig. 1C, F, J, N). Avicennia has always presented a problem to systematists. Like other mangroves, Avicennia exhibits a number of conspicuous adaptations to the mangrove habitat of which they are an important constituent. These include specialized erect, lateral roots developed as pneumatophores, longitudinal air chambers in the cortex of the pneumatophores, tolerance to high salt concentrations, xerophytic leaves with well-developed hypodermis and viviparous fruits which are sea-water dispersed (Fig. 1N). To a large extent, morphological characters in Avicennia appear to be con- trolled by environmental factors (Duke & al., 1998), making it a difficult task to trace the evolutionary history of the genus. However, there are also a number of traits, involving stem and root anatomy, pollen morphology, gynoecial anatomy, and embryology, which may or may not be related to the mangrove habitat (Sanders, 1997). Studies of these characters have led to a number of different suggestions of how to classify Avicennia: E.g., Bentham & Hooker (1876) placed Avicennia in Verbenaceae, Van Tieghem (1898) suggested a position near Santa- lales, Croizat proposed a place among Dipterocarpaceae (letters from

13 Croizat to Moldenke, in Moldenke, 1960) and Dahlgren (1975) placed the genus in Celastraceae. The results of the molecular study by Schwarzbach & McDade (2002), consistently placed Avicennia as sister to Thunbergioideae, although with moderate support, but they were unable to find morphological evidence to support the new findings. Several authors have given attention to the morphology of particular organs of Avicennia (e.g. pollen, Erdtman, 1945 and Saxena, 1981; embryology, Pad- manabhan, 1964 and Maheshwari & Kapil, 1966; gynoecium, Junell, 1934; wood anatomy, Zamski, 1979 and Carlquist, 1992), but did not compare them to lineages associated with Acanthaceae. The need to study Avicennia in greater detail in order to learn more about the taxonomic status and relationships of the genus has been pointed out several times (Padmanabhan, 1960; Schwarzbach & McDade, 2002).

Nelsonioideae The Nelsonioideae include about 170 mainly tropical species occurring on all continents (Wenk & Daniel, 2009). They are currently di- vided into seven genera, and can be distinguished from other Acan- thaceae by a combination of characters, including absence of retinacula, numerous ovules and descending-cochlear corolla aestivation. They are herbs and generally have small flowers when compared to many Acanthoideae and Thunbergioideae (Fig. 1G, K, H). Nelsonioi- deae have long been associated with Acanthaceae, but due to their deviating morphology, some authors have treated them as a family of their own (Sreemadhavan, 1977; Lu, 1990), or as a tribe in Scrophu- lariaceae (Bremekamp, 1953, 1965). Hossain (1971) presented the most detailed morphological study of the Nelsonioideae to date. He suggested that nelsonioids should be treated as a tribe within Acan- thaceae, and he also proposed a number of taxonomic changes within the group. Molecular studies of Acanthaceae have placed Nelsonioi- deae as sister to all other lineages (e.g. Hedrén & al., 1995; Scotland & al., 1995; McDade & Moody, 1999; McDade & al., 2008; Paper I). However, too few representatives of Nelsonioideae, as well as of other lineages within Lamiales, were included to test rigorously the hypoth- esis that nelsonioids are most closely related to Acanthaceae, or that Nelsonioideae or any of its constituent genera are monophyletic.

14 AIMS While general agreement currently prevails regarding the delimitation of Acanthaceae, this had not been tested in any study including more than a few representatives of Thunbergioideae and Nelsonioide- ae. The exact relationships among the major lineages in Acanthaceae, as well as more detailed relationships within Thunbergioideae and Nelsonioideae, have been little studied, and no morphological evidence to support the position of Avicennia in Acanthaceae has been presented. The main goals of paper I are to place Thunbergioideae among the other acanthaceous lineages, and to test earlier morphology-based hypotheses about relationships among and within the larger genera of the subfamily, based on chloroplast DNA sequences. Paper II aims at providing a comparative investigation of floral structure and development in Avicennia in order to shed light on the evolutionary history of the genus in relation to other Acanthaceae. Paper III aims at testing the monophyly and exact position of the Nelsonioideae, as well as the monophyly of the seven currently recognized nelsonioid genera, and explores evolutionary relationships among and within the genera. Based on the phylogenetic analysis of chloroplast and nuclear sequences as well as a much broader taxon sampling as in Paper I, paper IV, aims to infer more detailed hypotheses for phylogenetic relationships within Thunbergioideae and to extend our understanding of the evolution and diversification of the thunbergioid taxa.

MATERIALS AND METHODS In paper I, 30 species from the three major genera of Thunbergioi- deae (Thunbergia, Mendoncia, and Pseudocalyx) were included, cov- ering the main distribution areas of the genera. The sampled Thunber- gia species (21 species) represent all 8 subgenera circumscribed by Bremekamp (1955). The sampling also includes T. arnhemica, the only Thunbergia species native to Australia. Two species from each of the two tribes of the Acanthoideae; the Acantheae and Ruellieae (sensu Scotland & Vollesen 2000), were sampled along with single representatives of three out of six genera of Nelsonioideae. Three out of 8 species of Avicennia (sensu Tomlinson, 1995) were included and chosen with respect to geographic distribution. The genus Schlegelia which is possibly sister to Acanthaceae (71% jackknife support in Bremer & al., 2002) was chosen as out-group. All nucleotide sequences from three chloroplast DNA regions (rps16, rpl16, trnT-trnL) were newly produced for this study, and were analyzed using maximum

15 parsimony and Bayesian methods, and parsimony bootstrap support was estimated (see paper I for details). In paper II, floral development and structure of three out of eight species of Avicennia (sensu Tomlinson, 1995) were studied comparatively with representatives of other acanthaceous lineages. Taxon sampling of Avicennia reflects the biogeographic distribution of the genus: A. germinans occurring along the coasts of the Americas and the Atlantic coast of Africa, A. bicolor occurring along the pacific coast of Central America, and A. alba occurring along the coasts of south-eastern Asia and islands of the south-western Pacific. Micro- tome sections of old flower buds were prepared and studied with light microscope, and flowers in different developmental stages were prepared for and studied with scanning electron microscopy (see paper II for details). In paper III, we sampled representatives of all lineages of Lamiales that were shown by Schäferhoff & al. (2010) to be closely related to Acanthaceae, as well as members of all major clades within Acan- thaceae, including five out of seven nelsonioid genera (after testing the impact of different sampling strategies). This sampling was used to test the position of Nelsonioideae among Lamiales. To test relationships among Nelsonioideae, representatives of all genera that have been treated as nelsonioids were included. The number of sampled species reflects the relative size of each genus, and also mirrors morphological diversity and geographic ranges of the genera. Two different data sets were compiled, with genic regions offering various rates of evolution, in order to answer both the broad-scale questions (position of nelsonioids among Lamiales) and questions regarding detailed relationships among and within nelsonioid genera. For the broad-scale study, many sequences were retrieved from Genbank or EMBL, while new sequences were generated for the analyses of relationships within Nelsonioideae. Data sets were analyzed using maximum parsimony, Bayesian inference, and maximum likelihood methods. In paper IV, we sampled about three times as many thunbergioid taxa compared to Paper I, representing 87 species. Notably, monotypic Anomacanthus was sequenced for the first time, and several species of Pseudocalyx were included. For some species with variable morphology or uncertain taxonomic determination, two or more accessions were selected. Representatives of the other acanthaceous lineages were also included. In addition to the three chloroplast markers used in Paper I, we added one nuclear DNA region (the external transcribed spacer, ETS). Beside the main data sets, one subset of taxa was com-

16 piled using additional marker regions with sequences downloaded from Genbank. This subset was used to investigate the position of one nelsonioid taxon that was placed with Avicennia in some of the initial analyses. We analyzed all data sets separately, as well as in combinations, using maximum parsimony and Bayesian methods for phylogenetic inference.

RESULTS AND DISCUSSION Paper I. Both parsimony and Bayesian analyses of chloroplast data strongly suggest that Thunbergioideae are monophyletic, as has been implied in earlier molecular studies based on more limited taxon sampling (e.g. Hedrén & al. 1995; Scotland & al. 1995; McDade & al. 2000). The results also agree with the intrafamilial position of Thunbergioi- deae proposed in earlier higher-level studies (e.g. Scotland & al. 1995; McDade & Moody, 1999; both studies did not include Avicennia). In our analyses, Avicennia is consistently placed as sister to Thunbergi- oideae although with only moderate support, as was also found by Schwarzbach & McDade (2002). Acanthoideae is sister to Thunbergi- oideae plus Avicennia, and Nelsonioideae is sister to all other Acan- thaceae. Our results further suggest that Thunbergia and Mendoncia are both monophyletic, and that Mendoncia is sister to Thunbergia plus Pseudocalyx. Relationships within the two largest genera Men- doncia and Thunbergia are highly resolved and most branches are strongly supported by molecular data. Also morphological characters support the results. Besides the general characteristics of the subfamily (see introduction; Fig. 1), the sister taxa Thunbergia and Pseudo- calyx both have two fertile locules in the gynoecium and the fruit is a dry capsule, whereas in Mendoncia only one of two initiated locules develops fully and the fruit is a fleshy drupe (Schönenberger & En- dress, 1998; Fig. 1L, M). However, there are also characters shared by Mendoncia and Pseudocalyx, including stigma shape and indumentum (Brummitt, 1989; Schönenberger & Endress, 1998). Subgeneric relationships in Mendoncia reflect the geographic distribution of species. The genus displays a trans-Atlantic disjunction, and the results from the molecular analyses support a clade consisting entirely of American species, and two lineages of African species. Some of the African species have earlier been ascribed to two isolated genera, i.e. Monachochlamys Baker and Afromendoncia Gilg. Howev- er, in our analyses, neither of these two taxa turned out monophyletic

17 and our data therefore support Benoist’s (1944) suggestion that the two genera should be included in Mendoncia. Molecular data partially support the existing morphology-based classification of Thunbergia proposed by Bremekamp (1955), but not all eight subgenera are monophyletic and the classification needs revi- sion. Based on reconstructions of character-evolution in Thunbergia, we found a twining habit to be ancestral for that genus. The thecal awns, characteristic for many Thunbergia species (Schönenberger, 1999; Fig. 1R, S), have probably evolved from unicellular bristles. Generally, growth habit, calyx morphology, and anther dehiscence patterns seem to reflect molecular evolutionary relationships particularly well among Thunbergia species, whereas the presence or absence of anther appendages and stigma structure are somewhat more labile and most likely are closely correlated with differences in pollination biology.

Paper II. Phylogenetic relationships of the small black mangrove genus Avicennia have been difficult to trace due to the presence of many convergent characters (Duke & al, 1998). Our study of early floral development and floral structure documents that the general morphology of Avicennia fits well among other Acanthaceae. Most importantly, we found three synapomorphies supporting the suggested sister group relationship of Avicennia and Thunbergioideae: (1) collateral ovule arrangement, (2) vertical orientation of ovule curvature, and (3) an exposed nucellus that is contiguous with the ovary wall at least during early stages of ovule development. Additionala number of similarities between Avicennia and Thunbergioideae, including include a basically pentamerous floral ground plan with a dimerous gynoecium, a persisting calyx, left contort corolla aestivation, presence of five stamen primordia of which one degenerates, swollen filament bases, glandular hairs on filament bases, presence of pollen sac placentoids, collateral ovule arrangement, and, and unequal stigma lobes. Other similarities include articulated stems and fruit morphology. Several of these characters as well as other structural features and additional characters are shared with Acanthaceae in general (e.g. corolla aestivation, inflorescence structure and endosperm development (Pad- manabhan, 1964; Mohan Ram & Wadhi, 1965) and some are even commonly found in various families of the Lamiales, such as a pentamerous floral ground plan with a dimerous gynoecium, pollen sac

18 placentoids, and a strong reduction or even a complete loss of the dor- sal (median) stamen.

Paper III. The results provide strong support for the monophyly of Nelsonioi- deae and a placement of the group sister to all other Acanthaceae. We were unable to identify the closest relatives of Acanthaceae, but as is well known among students of Lamiales, resolving relationships among lineages of this order is challenging. Very large data sets (in terms of numbers of informative characters) will likely be necessary. Among Nelsonioideae, our results indicate that Nelsonia, Elytraria and Anisosepalum are monophyletic with strong support. Possibly, Nelsonia is the first branching clade and Elytraria sister to the remaining nelsonioids. An African clade comprising monospecific Saintpau- liopsis sister to Anisosepalum is sister to a clade that includes all sampled members of pantropical Staurogyne. Monotypic Gynocraterium and Ophiorrhiziphyllon, comprising either one or two species, are nested within Staurogyne. This clade is strongly supported in our analyses, and our data confirm the results of Hossain’s (2004) morphological study, indicating that these genera are very closely related. The Guyanan genus Gynocraterium is sister to all sampled members of American Staurogyne; this last clade is sister to a clade comprising the other sampled Staurogyne plus Asian Ophiorrhiziphyllon, which is nested among Asian Staurogyne. Continued recognition of Gynocrate- rium and Ophiorrhiziphyllon would require description of additional genera to accommodate clades within Staurogyne. While Nelsonia has a wider range than many large lineages of Acanthoideae, and its native range is uncertain, other Nelsonioideae appear to have originated in Africa with at least two dispersal events to the American continent, three to Madagascar and two to Asia. Our results suggest that Nelsonioideae in general have a complex history of inter-continental dispersals.

Paper IV. Our results strongly supported Anomacanthus as sister to Mendon- cia, and these two together are sister to the remaining thunbergioids. The clade with Anomacanthus and Mendoncia is unique among Acan- thaceae in having fleshy drupes (Fig. 1M). While Mendoncia came out monophyletic, Thunbergia and Pseudocalyx did not, due to the strongly supported placement of Thunbergia heterochondros with Pseudo- calyx. This taxon was only recently transferred from Pseudocalyx to

19 Thunbergia (Vollesen, 2008), and our morphological examinations reveal that the species indeed has much in common with the latter genus. Also one other taxon is worth special attention regarding the delimitation of Pseudocalyx and Thunbergia. The Asian species Thunbergia colpifera shares many features with both genera, and is sister to all other Thunbergia species (except T. heterochondros). We included several accessions of some Thunbergia species, and also included pairs of species, both in Mendoncia and Thunbergia, considered synonymous by some authors. Our results show that several of the species are not monophyletic, and the data do not provide evidence to support synonymy of certain species pairs. Our data sets generally contained a high percentage of parsimony- informative characters (36 % for the combined chloroplast and nuclear data set), but maximum parsimony and Bayesian analyses of the combined data set failed to resolve relationships among the large Ameri- can Mendoncia clade. Many of these species are at least superficially similar (Wasshausen, 1989), and we interpret this as the result of a rapid colonization and diversification in South and Central America. The few species described in Pseudocalyx occur in tropical regions of Africa, with one species occurring both in continental Africa and on Madagascar. Also Thunbergia and Mendoncia seem to have dispersed once each to Madagascar, with the subsequent evolution of several endemic species. In Thunbergia, we observe a more complex pattern of geographic distribution than in the other genera. Although the geographic origin of the subfamily as a whole appears to be in tropical regions of Africa, the origin of Thunbergia is uncertain. Attempts to trace early dispersal events are likely to be strongly influenced by sampling strategy, as the first branching lineages of the genus covers a wide geographic range. Despite the fact that we were unable to retrieve DNA from Mey- enia, we agree with earlier authors who consider this genus synonymous to Thunbergia. There are no substantial morphological differences between the genera, and an examination of herbarium speci- mens imply that Meyenia belongs to one of the early diverging lineages of Thunbergia. Regarding relationships among major lineages of Acanthaceae, our results are congruent with earlier studies (e.g. Schwarzbach & McDade, 2002; paper I; Paper III), and suggest Avicennia is the closest relative of Thunbergioideae.

20 CONCLUDING REMARKS Acanthaceae sensu stricto were long recognized as a natural group consisting of mainly tropical herbs with explosively dehiscent fruits and seeds borne on retinacula, corresponding to Acanthoideae sensu Scotland and Vollesen (2000). In addition to these, some smaller lineages (Thunbergioideae and Nelsonioideae, sensu Scotland and Vollesen, 2000) have been regarded as related to Acanthaceae s.str. However, they were considered substantially different morphological- ly from Acanthoideae, and were not always accepted as “true” Acan- thaceae. The development of molecular methods has revolutionized the field of plant systematics, and earlier family circumscriptions have under- gone thorough revaluations. Sometimes, results have been rather un- expected. There are also many examples of cases where molecular data have failed to resolve problems regarding evolutionary relationships. Studies of morphological structure and development have prov- en valuable in elucidating many controversial interpretations of floral structures and in understanding systematic relationships (e.g., Endress, 2003; von Balthazar & al., 2004; Schönenberger & Grenhagen, 2005; Schönenberger & Endress, 1998¸ Manktelow, 2000; Wortley & al., 2005). Morphological approaches remain indispensable in systematic research, but for many groups where DNA sequences have been used to assess phylogenetic questions, detailed morphological studies are still lacking. The implications of molecular and morphological methods for Ac- anthaceae have resulted in an altered, but fairly robust, definition of the family and its major lineages. To begin with, Thunbergioideae and Nelsonioideae have been shown with strong statistical support, based on solid sampling strategies, to belong in Acanthaceae (Paper I; Paper III). Although superficially different from other Acanthaceae, detailed studies of morphology and floral development have shown that Thun- bergioideae fit well among the other lineages (Schönenberger & En- dress, 1998; Paper I; Paper II). Next, Schwarzbach & McDade’s (2002) molecular study presented Avicennia as part of Acanthaceae, and possibly as sister to Thunbergioideae. Their results have been reproduced in later molecular surveys (e.g. McDade & al., 2008; Paper I; Paper III and Paper IV in this thesis), but although results have been consistent across studies, support values are moderate. Several studies of particular organs have been presented earlier for Avicennia (see introduction), but more thorough analyses of floral structure and development had not been attempted. Furthermore, earlier comparisons

21 with e.g. Verbenaceae, or even groups outside Lamiales, are of limited value when trying to interpret floral morphology in Avicennia in relation to other Acanthaceae. As shown in Paper II, the general morphology of Avicennia can be easily accommodated in the Acanthaceae, but more importantly, a number of synapomorphies supporting a sister group relationship between Avicennia and Thunbergioideae were re- vealed. As discussed in paper III and elsewhere (e.g. Schäferhoff & al., 2010), all efforts to find the closest relatives of Acanthaceae have failed, making it difficult to find unambiguous synapomorphies for the family as a whole. Nonetheless, Acanthaceae are well supported by DNA evidence as well as morphological data, and our understanding of the evolutionary history of this diverse and fascinating family has improved considerably during the last decade.

SVENSK SAMMANFATTNING (SWEDISH SUMMARY) Blomväxtfamiljen Acanthaceae består av över 4000 arter som växer framförallt i tropiska och subtropiska delar av världen. De flesta av dessa ingår i underfamiljen Acanthoideae (sensu Scotland & Vollesen, 2000), som sedan länge erkänts som en naturlig grupp och som kännetecknas av en speciell typ av fruktkapsel. Några mindre grupper, Thunbergioideae (cirka 170 arter i fem släkten) och Nelsonioideae (cirka 170 arter fördelade på sju släkten; sensu Scotland & Vollesen, 2000) har också ansetts stå nära Acanthoideae. På grund av att de saknar den speciella frukttypen och i övrigt också skiljer sig morfologiskt från Acanthoideae, har vissa forskare ansett att de inte hör hemma i samma familj, och även inbördes släktskap och klassifikation av Thunbergioideae har ifrågasatts (bland andra van Tieghem, 1908; Bremekamp, 1953; Dahlgren, 1980; Cronquist, 1981). Schönenberger & Endress (1998) visade med stöd av komparativa blomutvecklingsstudier att Thunbergioideae utgör en naturlig grupp. Sedan DNA-sekvenser (molekylära data) började användas i större omfattning för att besvara frågor om släktskap, har flera studier visat på att Thunbergioideae och Nelsonioideae trots allt står så nära Acanthoideae att de kan klassificeras i samma familj, men bara ett fåtal representanter av de senare har tagits med i dessa studier (till exempel Hedrén & al., 1995; McDade & Moody, 1999). Ganska överraskande presenterade Schwarzbach & McDade (2002) resultat baserade på DNA-data som visade att mangrovesläktet Avicennia med sina omkring åtta arter också hör hemma i Acanthaceae. Ingen hade

22 tidigare förknippat Avicennia med Acanthaceae eftersom de till synes inte alls liknar varandra, men å andra sidan hade forskare länge diskuterat var Avicennia hör hemma. Släktet uppvisar många karaktärer som är gemensamma för alla växter som lever i mangroveskogar, men som är ett resultat av anpassningen till livsmiljön snarare än ett resultat av evolutionärt släktskap med dessa. Syftet med den här avhandlingen har varit att 1) placera Thunbergioideae i Acanthaceae, undersöka om släktena i underfamiljen är monofyletiska, samt vidga kunskapen om inbördes släktskapsförhållanden (artikel I och IV), 2) ta reda på om det finns morfologiska karaktärer som förenar Avicennia med andra akantacéer och särskilt Thunbergioideae (artikel II), och 3) undersöka om Nelsonioideae är en monofyletisk grupp och om den hör hemma i Acanthaceae, ta reda på om släktena är monofyletiska, samt utöka kunskapen om inbördes släktskapsförhållanden (artikel III). För att uppnå syftena har olika molekylärbiologiska och morfologiska metoder använts. Olika genregioner har extraherats från kloroplast- och cellkärns-DNA och därefter sekvenserats och analyserats med parsimonimetoder och Bayesianska metoder för att ta fram släktskapshypoteser (fylogenier). Morfologiska preparat har tillverkats av blommor och blomknoppar i olika utvecklingsstadier och från olika taxa för jämförande studier med ljusmikroskopi och svepelektronmikroskopi. Resultaten som presenteras i artikel I, och som är baserade på information från kloroplast-DNA, visar att Thunbergioideae med stor sannolikhet är närmast släkt med Avicennia, och att dessa tillsammans utgör systergrupp till Acanthoideae. Inom Thunbergioideae är Mendoncia syster till Thunbergia plus Pseudocalyx. De två senare släktena har frukter som är kapslar, liksom de flesta andra akantacéer, medan Mendoncia utmärker sig genom att ha stenfrukter. Både Thunbergia och Mendoncia är monofyletiska (endast en representant från Pseudocalyx finns med i analyserna). En jämförelse med en tidigare klassifikation av släktet Thunbergia (Bremekamp, 1955) visar att denna klassifikation delvis stämmer överens med molekylära data. Vi gjorde karaktärsoptimering av morfologiska data som tyder på att några av de karaktärer Bremekamp (1955) använde för sin klassifikation verkar återspegla evolutionärt släktskap, exempelvis foderbladens utseende och typ av ståndaröppning. Andra karaktärer verkar vara starkt påverkade av pollinationsbiologiska faktorer som inte speglar släktskapsförhållanden i Thunbergia, till exempel förekomst och typ av ståndarbihang samt märkets utseende.

23 I artikel II gör vi en jämförande studie av blomutveckling och blommorfologi hos tre arter av Avicennia och representanter från de tre underfamiljerna i Acanthaceae. De generella morfologiska dragen hos Avicennia passar väl in i Acanthaceae (exempelvis blomställning och endospermutveckling; Padmanabhan, 1964; Mohan Ram & Wad- hi, 1965). Särskild vikt läggs vid jämförelsen med Thunbergioideae, och vi hittar tre synapomorfier i fruktämnet som stödjer ett nära släktskap mellan Avicennia och Thunbergioideae. Därutöver hittar vi fler likheter mellan dessa två klader och mellan Avicennia och Acanthaceae generellt, men eftersom dessa inte är tillräckligt väl kända hos andra närstående grupper går det inte att bedöma om de utgör synapomorfier. Artikel III presenterar en molekylär studie av Nelsonioideae. Med en gedigen representation av taxa från Acanthaceae och andra familjer inom Lamiales kan vi ställa utom rimligt tvivel att Nelsonioideae är närmast släkt med andra akantacéer. Dessutom ger analyserna starkt stöd för att underfamiljen verkligen är monofyletisk. Inom Nelsoniodeae däremot visar våra data att de två små släktena Gynocraterium och Ophiorrhiziphyllon hamnar inuti det största släktet Staurogyne. Dessa tre släkten delar stora morfologiska likheter, och Hossain (1971, 2004) ifrågasatte giltigheten av släktstatus för de två mindre. Övriga släkten är monofyletiska. Våra data tyder på att Nelsonia är syster till alla andra Nelsonioideae, men det statistiska stödet är inte övertygande. Medan övriga släkten verkar ha sitt ursprung i Afrika med senare spridning till Madagaskar, Asien och den amerikanska kontinenten, är den vitt utbredda Nelsonias geografiska urspung, och därmed hela underfamiljens och akantacéernas geografiska ursprung, fortfarande höljt i dunkel. I den sista artikeln, IV, mer än fördubblar vi antalet taxa och vi utökar dataseten från artikel I med nukleär DNA-information. Det monotypiska släktet Anomacanthus sekvenseras för första gången och flera representanter av Pseudocalyx analyseras. Detta resulterar i flera nya upptäckter. Studien ger starkt statistiskt stöd för att Anomacanthus är syster till Mendoncia, vilket även stöds morfologiskt av det faktum att dessa båda släkten har stenfrukt. Våra data tyder på att Mendoncia efter spridning från Afrika till Central- och Sydamerika har genomgått en snabb expansion och artbildning. I Thunbergia avslöjar sig ett biogeografiskt mer komplext mönster än i första artikeln, och släktets geografiska ursprung är osäkert, medan underfamiljen som grupp verkar ha uppstått i Afrika. Ett par arter, en av dem placerad med Pseudocalyx och den andra som syster till alla andra Thunbergia-arter,

24 visar sig dela morfologiska karaktärsdrag med båda släkten, och aktualiserar frågan om dessa två släkten verkligen kan särskiljas morfologiskt. Flera arter inom Thunbergia visar sig ha oklara gränser, och ytterligare morfologiska studier och molekylära studier behövs för att reda ut artgränser och släktskapsförhållanden. I Mendoncia tycks situationen vara närmast den omvända, och olika exemplar från arter som synonymiserats får inget stöd för att vara monofyletiska med våra dataset. Vi lyckades inte sekvensera det monotypiska släktet Meyenia, men en granskning av herbariematerial tyder på att släktet i själva verket hör hemma inom Thunbergia, där den enda arten också först beskrevs (Wallich, 1826), och troligen bland någon av de grupper som divergerat tidigast. Som familjen avgränsas idag är Acanthaceae en artrik och divers familj. Den finns representerad på alla kontinenter och i många olika livsmiljöer, från mycket torra inlandsregioner till fuktiga regnskogar och saltvattenområden. Med den stora morfologiska variationen mellan de tre underfamiljerna samt Avicennia, och på grund av att det ännu är osäkert vilka växter som är närmast släkt med Acanthaceae, är det svårt att peka ut morfologiska synapomorfier för hela familjen. Däremot finns det nu ett starkt molekylärt stöd för familjen och större klader inom den, och många morfologiska likheter som också vittnar om släktskapet. Det finns fortfarande mycket att arbeta vidare med, men vår kunskap om denna fascinerande familj har genom morfologiska och molekylära studier ökat markant under det senaste decenniet.

ACKNOWLEDGEMENTS First of all, I am most grateful to my supervisor Jürg Schönenberger for sharing his knowledge and passion for plant systematic research, for giving new insights and constructive feedback, and for always be- ing ready to help and support me. Next, I want to thank all my col- leagues at the Department of Botany, including researchers, other staff and students, for the many stimulating discussions, practical advice and invaluable support. I also want to thank Lucinda McDade for re- ceiving me in her lab, and for help and inspiration. Throughout this journey, I have received constant support and en- couragement from my friends, my Swedish and Danish families, my parents, and most of all from Marc. I send you all my love.

25 LITERATURE CITED Benoist, R. 1944. Contribution à la connaissance des Acanthacées africaines et malgaches. Notul. Syst. 11: 137-151. Bentham, G. 1876. Acanthaceae. Pp. 1060--1122 in: Bentham, G. & Hook- er, J.D. (eds.), Genera plantarum, vol. 2. Reeve, London. Bremekamp, C.E.B. 1953. The delimitation of the Acanthaceae. Proc. Kon. Ned. Akad. Wetensch. C 56: 533--546. Bremekamp, C.E.B. 1955. The Thunbergia species of the Malesian area. Verh. Kon. Akad. Wetensch., Afd. Natuurk., Sect. 2. part 50(4): 1--90. Bremekamp, C.E.B. 1965. Delimitation and subdivision of the Acanthace- ae. Bull. Bot. Surv. India 7: 21--30. Bremer, B., Bremer, K., Heidari, N., Erixon, P., Olmstead, R.G., Anderberg, A.A., Källersjö, M. & Barkhordarian, E. 2002. Phyloge- netics of asterids based on 3 coding and 3 non-coding chloroplast DNA markers and the utility of non-coding DNA at higher taxonomic levels. Molec. Pylogen. Evol. 24: 274--301. Buzato, S., Sazima, M. & Sazima, I. 2000. Hummingbird-pollinated floras at three Atlantic Forest sites. Biotropica 32: 824--841. Carlquist, S. 1992. Wood anatomy of sympetalous dicotyledon families: A summary, with comments on systematic relationships and evolution of the woody habit. Ann. Missouri. Bot. Gard. 79 (1): 303-332. Dahlgren, R. 1975. A system of classification of the angiosperms to be used to demonstrate the distribution of characters. Bot. Not. 128: 119-147. Duke, N.C., Benzie, J.A.H., Goodall, J.A. & Ballment, E.R. 1998. Genetic structure and evolution of species in the mangrove genus Avicennia (Av- icenniaceae) in the Indo-West Pacific. Evolution 52(6): 1612-1626. Endress, P.K. 2003. Early floral development and the nature of the calyptra in Eupomatiaceae. Int. J. Pl. Sci. 164: 489--503. Endress, P.K. 2010. Disentangling confusions in inflorescence morphology: Patterns and diversity of reproductive shoot ramification in angiosperms. J. Syst. Evol. 48: 225--239. Erdtman, G. 1945. Pollen morphology and plant taxonomy. IV. Labiatae, Verbenaceae and Avicenniaceae. Svensk. Bot. Tidskr. 39: 279-285. Faegri, K., van der Pijl, F. 1979. The principles of pollination ecology. Pergamon Press, Oxford. Hedrén, M., Chase, M.W. & Olmstead, R.G. 1995. Relationships in the Acanthaceae and related families as suggested by cladistic analysis of rbcL nucleotide sequences. Pl. Syst. Evol. 194: 93--109. Hequet, V. 2003. Pollen atlas of bat-pollinated plants of central French Gui- ana. http://www.nybg.org/botany/tlobova/hequet/pollen_atlas.html Hossain, A.B.M.E. 1971. Studies in the classification and affinities of Acan- thaceae. Dissertation, Edinburgh University, Edinburgh. Hossain, A.B.M.E. 2004. Taxonomic structure of the Nelsonioideae (Acan- thaceae) - I: Generic circumscriptions and their interrelationships. Bang- ladesh J. Pl. Taxon. 11: 19--30.

26 Junell, S. 1934. Zur Gynäceummorphologie und Systematik der Verbenaceen und Labiaten. Symb. Bot. Upsal. 4: 1-219. Lindau, G. 1893. Übersicht über die bisher bekannten Arten der Gattung Thunbergia L. f. Bot. Jahrb. Syst. 41: 31--41. Lu, A.M. 1990. A preliminary cladistic study of the families of the superor- der Lamiiflorae. Bot. J. Linn. Soc. 103: 39--57. Maheshwari, P. & Kapil, R.N. 1966. Some indian contributions to the embryology of angiosperms. Phytomorphology 16: 239-291. Manktelow, M. 2000. The filament curtain: A structure important to systematics and pollination biology in the Acanthaceae. Bot. J. Linn. Soc. 133: 129--160. McDade, L.A., & Moody, M.L. 1999. Phylogenetic relationships among Acanthaceae: Evidence from noncoding trnL-trnF chloroplast DNA sequences. Amer. J. Bot. 86: 70--80. McDade, L.A., Masta, S.E., Moody, M.L. & Waters, E. 2000. Phyloge- netic relationships among Acanthaceae: Evidence from two genomes. Syst. Bot. 25: 106-121. Mohan Ram, H.Y. & Wadhi, M. 1965. Embryology and the delimitation of the Acanthaceae. Phytomorphology 15 (1-4): 201-205. Moldenke, H.N. 1960. Materials toward a monograph of the genus Avicen- nia. Phytologia 7 (3-5): 123-168, 179-232, 259-293. Padmanabhan, D. 1960. The embryology of Avicennia officinalis. 1. Floral morphology and gametophytes. Proc. Indian Acad. Sci. Pl. Sci. B52: 131-145. Padmanabhan, D. 1964. The embryology of Avicennia officinalis. 2. En- dosperm. Phytomorphology 14: 442-451. Profice, S.R. 1988. Mendoncia Vell. ex Vand. (Acanthaceae): Espécies ocorrentes no Brasil. Arq. Jard. Bot. Rio J. 29: 201--279. Sanders, R. W. 1997. The Avicenniaceae in the Southeastern United States. Harvard Pap. Bot. 10: 81-92. Saxena, M.R. 1981. Contribution to the palynotaxonomy of Avicenniaceae Endl. J. Indian Bot. Soc. 60: 28-31. Schäferhoff, B., Fleischmann, A., Fischer, E., Albach, D.C., Borsch, T., Heubl, G., Müller, K.F. 2010. Towards resolving Lamiales relationships: insights from rapidly evolving chloroplast sequences. BMC Evol. Biol. 10: 352--374. Schönenberger, J. & Endress, P.K. 1998. Structure and development of the flowers in Mendoncia, Pseudocalyx, and Thunbergia (Acanthaceae) and their systematic implications. Int. J. Pl. Sci. 159: 446-465. Schönenberger, J. 1999. Floral structure, development and diversity in Thunbergia (Acanthaceae). Bot. J. Linn. Soc. 130: 1-36. Schönenberger, J., & Grenhagen, A. 2005. Early floral development and androecium organization in Fouquieriaceae (Ericales). Pl. Syst. Evol. 254: 233--249.

27 Schwarzbach, A. E., & McDade, L. A. 2002. Phylogenetic relationships of the mangrove family Avicenniaceae based on chloroplast and nuclear ri- bosomal DNA sequences. Systematic Botany 27: 84-98. Scotland, R. W., & Vollesen, K. 2000. Classification of Acanthaceae. Kew Bulletin 55: 513-589. Snow, D.W. & Teixeira, D.L. 1982. Hummingbirds and their flowers in the coastal mountains of southeastern Brazil. J. Ornithol. 123: 446--450. Sreemadhavan, C.P. 1977. Diagnosis of some new taxa and some new combinations in Bignoniales. Phytologia 37: 412--416. Tomlinson, P.B. 1995. The botany of mangroves. Cambridge University Press, Cambridge. van Tieghem, M.P. 1898. Avicenniacées et Symphorémacées: Place de ses deux nouvelles familles dans la classification. J. Bot. (Morot) 12: 345- 365. van Tieghem, M.P. 1908. Structure du pistil et de l'ovule, du fruit et de la graine des Acanthacées. Ann. Sci. Nat., Bot. 7: 1--24. Vogel, S., Machado, I.C., & Lopes, A.V. 2004. Harpochilus neesianus and other novel cases of chiropterophily in neotropical Acanthaceae. Taxon 53: 55-60. Vollesen, K. 2008. Acanthaceae (Part 1). Pp 1-- 286 in: Beentje, H.J. & Ghazanfar, S.A. (eds.), Flora of Tropical East Africa. Royal Botanic Gardens, Kew. von Balthazar, M., Alverson, W.S., Schönenberger, J., & Baum, D.A. 2004. Floral development and androecium structure of Malvoideae (Malvaceae s.l.). Int. J. Pl. Sci. 165: 445--473. Wallich, N. 1826. Tentamen Florae Napalensis Illustratae Consisting of Botanical Descriptions and Lithographic Figures of Select Nipal Plants, fasc. 2. Asiatic Lithographic Company’s Press, Calcutta and Serampore. Wenk, R. & Daniel, T.F. 2009. Molecular phylogeny of Nelsonioideae (Acanthaceae) and phylogeography of Elytraria. Proc. Calif. Acad. Sci. 60: 53--68. Wortley, A.H., Scotland, R.W., Rudall, P.J. 2005. Floral anatomy Tho- mandersia (Lamiales), with particular reference to the nature of the reti- naculum and extranuptial nectaries. Bot. J. Linn. Soc. 149: 469--482. Zamski, E. 1979. The mode of secondary growth and the three-dimensional structure of phloem in Avicennia. Bot. Gaz. 140(1): 67-76.