SYSTEMATICS, HYBRIDIZATION, AND CHARACTER EVOLUTION WITHIN THE SOUTHERN AFRICAN GENUS, ZALUZIANSKYA (SCROPHULARIACEAE S.S., TRIBE MANULEEAE)
DISSERTATION
Presented in Partial Fulfillment of the Requirements for
the Degree Doctor of Philosophy in the Graduate
School of The Ohio State University
By
Jenny Kay Archibald, B.S.
*****
The Ohio State University 2003
Dissertation Committee: Approved by
Dr. Andrea D. Wolfe, Adviser
Dr. John V. Freudenstein ______
Dr. John W. Wenzel Adviser Department of Evolution, Ecology, and Organismal Biology
ABSTRACT
Zaluzianskya (Scrophulariaceae s.s., tribe Manuleeae), a genus of 55 species of annual and perennial herbs endemic to southern Africa, includes a diverse array of morphological features – particularly in floral morphology. This dissertation examines the genus at both population and species levels. First, a molecular investigation of hybridization between two species in the genus was conducted. Although unexpected due to apparent ethological isolation, hybridization between day-flowering Z. microsiphon and night-flowering Z. natalensis has been proposed based on intermediate individuals found in sympatric populations of these two species. This putative hybridization was studied using molecular markers (ISSRs) and ordination of morphological traits. The species are very genetically similar, but intra- and interspecific variation in band-frequencies was found.
Eight of fifteen species-typical ISSR markers occurred at intermediate band frequencies in the putative hybrids compared to the average band frequency for each
“parental” species. The other seven markers occurred at extreme frequencies. However, they were not extreme when compared with individual population frequencies of one or both species. At minimum, the hybrid population always had comparable band frequencies to those of one or both sympatric “parental” population(s). These results are consistent with hybridization, although they could also indicate historical similarity.
ii Nine loci were present only in individuals of Z. microsiphon, the “hybrids,” and
sometimes the sympatric individuals of Z. natalensis (only one locus showed the reverse
pattern). This suggests unidirectional gene flow from Z. microsiphon to Z. natalensis,
which is also supported by detailed examinations of four marker bands with significant
differences in band frequencies across populations. Separate clusters for each species
were formed by PCA, with hybrid individuals located between the two clusters but often
closer to the Z. natalensis cluster. One hypothesis is that hybrids are backcrossing with
Z. natalensis, leading to introgression of Z. microsiphon genetic material.
A genus-wide investigation of phylogenetic relationships was also conducted,
using DNA sequences of ITS, rpl16, and trnL-trnF to produce the first broad
phylogenetic hypotheses for this genus and thus the first tests of the monophyly of the
genus and its sections. One of the “outgroups,” Reyemia nemesioides, was derived within
Zaluzianskya, with strong support. This is consistent with several morphological
characteristics of both species of Reyemia and it is thus proposed that Reyemia be merged
into Zaluzianskya. Sections Zaluzianskya and Holomeria are clearly not natural, whereas
the status of section Nycterinia is uncertain. The single species sampled from section
Macrocalyx was nested within section Nycterinia according to the chloroplast (and
combined) data, but sister to that section according to the ITS data. The status of several
species was also highlighted by the data. Populations of Z. villosa and Z. gracilis were derived from Z. affinis while Z. divaricata was derived from Z. pusilla. Finally, populations of the sole day-flowering species in section Nycterinia, Z. microsiphon, were separated on the phylogenies. The source of this division is unknown but could possibly be hybridization or convergent evolution via pollinator selection.
iii These phylogenetic hypotheses were also used to examine character evolution for eleven traits within Zaluzianskya, focusing on several traits considered important in the classification of the genus as well as floral traits that appear relevant for pollinator interactions. The evolution of habit was also examined along with its potential relation to distribution within different rainfall regimes. The placement of the former species of
Reyemia within Zaluzianskya is supported by calyx lobing, and the specific placement of
R. nemesioides is supported by stem indumentum and petal shape. The latter two traits generally support the division of Zaluzianskya into three major clades. Nectary adnation supports clade 3 as well as a subclade of clade 1 (Z. affinis + Z. villosa + Z. gracilis) also supported by stamen traits. Long corolla tubes define all of clade 3 except Z. mirabilis and intermediate tubes mark a subclade of clade 1. In contrast, floral symmetry and throat indumentum are largely homoplasious.
Several potential implications regarding pollination biology are discussed. First, several floral traits may have evolved in parallel within “Z. microsiphon” lineages due to pollinator selection (long-proboscid fly vs. the hawkmoth pollinators of other members of section Nycterinia), but this is unclear due to the uncertain status of this species. Several other traits found within clade 3, such as a ringlet of hairs around the throat and notched petals, may be beneficial for moth pollination. However, these traits also occur in some other clades and the function in those species, if there is any, is unknown.
Reconstructions of the distribution of species of Zaluzianskya within relatively arid vs. mesic regions appear similar to those of habit (annual vs. perennial). The plesiomorphic condition for this group is annual, distributed within arid regions. Subsequent migrations to mesic regions and derivations of the perennial habit both occur only within clade 3.
iv Dedicated to Mark and my parents
v ACKNOWLEDGMENTS
I would like to thank my advisor, Andrea Wolfe, and my committee members,
John Freudenstein and John Wenzel, for supporting me in my graduate work. I’d also like to thank Tom Waite for serving on my Candidacy Examination Committee as well as for his assistance with statistical analyses.
I thank Mark Mort for his constant encouragement, enthusiasm, generosity, and willingness to argue the minutiae of systematics with me.
I am indebted to many current and previous members of the systematics and ecology groups at O.S.U. for assistance with laboratory matters, analyses, editing, and for helpful and interesting discussions. In particular I would like to thank Chris Randle, for his help in all scientific arenas as well as his entertaining discourse; Shannon Datwyler for guidance, especially when I was new to the lab; Theresa Culley and Lisa Wallace, particularly for their insight into population biology; Sibyl Bucheli, Jeff Morawetz, and
Sarena Selbo for their encouragement; as well as Nidia Arguedas, Siri Ibarguen, Shawn
Krosnick, Paul Nunley, and Kristen Uthus for making O.S.U. a better place both scientifically and generally. Shannon, Andrea, Sarena, Shawn, Mark, and Sibyl are also due thanks for playing the arduous role of dog- and/or plant-sitters during my field work in Africa. I’d also like to thank Erich Grotewold and Allison Snow for allowing me to do
vi research with their lab groups. From the University of Kansas, I’d like to thank Dan
Crawford and Craig Freeman for helpful discussion and Kristopher Fairfield for assistance in the lab.
My work in South Africa was greatly aided by a number of researchers, including
Nigel Barker, Nicola Berg, Jo Beyers, Trevor Edwards, Craig Peter, Koos Roux, Frank
Smith, Dee Snijman, Kim Steiner, and Tony Verboom, among others. In particular, this work would not have been possible without the assistance and knowledge of Steve
Johnson with whom I collaborated for the hybridization studies. Multiple herbaria were also indispensable for my work, including NBG, BOL, GRA, and NU. These institutions were very generous with access to their specimens, resources, and staff.
Finally I wish to thank Paul Wolf and Bradley Kropp for giving me my start in scientific research; Audrey Seamons and David Marsolais for their friendship; and my family, for being the way that they are.
My work was made possible by funding from an ASPT Graduate Student
Research Award, a Molecular Life Science Graduate Fellowship, an Alumni Grant for
Graduate Research and Scholarship, Janice Carson Beatley Herbarium Awards, and
Graduate Student International Dissertation Research Travel Grants to myself, as well as grants from the National Science Foundation to my advisor (DEB 9708332 and DEB
0089640).
vii VITA
6 October 1974 …………………… Born – Logan, UT, USA
1997 ………………………………. B.S. Biology, Utah State University
1997 – 1998 ………………………. Researcher, Utah State University
1998 – present …………………….. Graduate Fellow and Teaching Associate, The Ohio State University
PUBLICATIONS
Archibald, J. K., M. E. Mort, D. J. Crawford. 2003. Bayesian inference of phylogeny: a non-technical primer. Taxon 52: 187-192.
Wolf, P. G., D. R. Campbell, N. M. Waser, S. D. Sipes, T. R. Toler, and J. K. Archibald. 2001. Tests of pre- and postpollination barriers to hybridization between sympatric species of Ipomopsis (Polemoniaceae). American Journal of Botany 88: 213-219.
Archibald, J. K., P. G. Wolf, V. J. Tepedino, and J. B. Bair. 2001. Genetic relationships and population structure of the endangered steamboat buckwheat, Eriogonum ovalifolium var. williamsiae (Polygonaceae). American Journal of Botany 88: 608- 615.
FIELDS OF STUDY
Major Field: Evolution, Ecology, and Organismal Biology
viii TABLE OF CONTENTS
Page Abstract …………………………………………………………………………………...ii Dedication ………………………………………………………………………………...v Acknowledgments ……………………………………………………………………….vi Vita ……………………………………………………………………………………..viii List of Tables ……………………………………………………………………………..x List of Figures …………………………………………………………………………...xii
Chapters:
1. Introduction ……………………………………………………………………….1
2. Hybridization and gene flow between a day- and night-flowering species of Zaluzianskya .………………………………………..…………………………7
Introduction ……………………………………………………………….7 Methods …………………………………………………………………10 Results …………………………………………………………………...16 Discussion ……………………………………………………………….19
3. Phylogenetic relationships within Zaluzianskya and their impact on classification, based on sequences from multiple genomes ……………………..38
Introduction ……………………………………………………………...38 Methods ………………………………………………………………….40 Results …………………………………………………………………...44 Discussion ……………………………………………………………….49
4. Character evolution within Zaluzianskya: implications for classification, floral evolution, and biogeography ……………………………………………...67
Introduction ……………………………………………………………...67 Methods ………………………………………………………………….73 Results …………………………………………………………………...76 Discussion ……………………………………………………………….81
List of References ……………………………………………………………………...109
ix LIST OF TABLES
Table Page
2.1 Several of the floral differences between Z. microsiphon and Z. natalensis ……28
2.2 Populations of Zaluzianskya included in this study ……………………………..28
2.3 Morphological traits measured in sympatric individuals of Z. microsiphon, Z. natalensis, and their putative hybrids ………………………………………...29
2.4 Band frequencies of the fifteen species-typical loci (differ by at least 25% between the putative parental species). M, H, and N indicate the average band frequencies for the Z. microsiphon, putative hybrid, and Z. natalensis individuals, respectively ………………………………………………………...30
2.5 Band frequencies for ten loci that occur only in one proposed parental species, the putative hybrids, and sometimes in the sympatric population of the other proposed parental species. The two loci that are also considered species-typical (see Table 2.4) are underlined …………………………………..30
2.6 P-values for pair-wise population comparisons of band frequencies as determined by a modified Fisher’s exact test, for four ISSR loci: I-7, II-4, II-9, and III-6. Significant values (α = 0.05) are shown in bold. Those values still considered significant with a Bonferroni adjusted alpha level (α = 0.05/45 = 0.0011) are underlined. All standard errors were less than 0.0015, the average S.E. was 0.0004 ……………………………………………31
2.7 Character loadings for the first three PCA axes …………………………………32
3.1 Several characteristics of the sections and subsections of Zaluzianskya ………..61
3.2 The 28 populations of Zaluzianskya and 11 outgroup taxa included in the final phylogenetic analyses ……………………………………………………...62
3.3 The number and types of characters in each data set as well as characteristics of the resulting most parsimonious trees (MPTs) for the final analyses including 39 populations ………………………………………………………...63
x 3.4 Number of additional steps found for the most parsimonious trees when the designated groups were constrained to be monophyletic for each data set. Percentage of total number of steps given in parentheses ………………………64
4.1 Characters and character states examined within Zaluzianskya and the outgroup taxa …………………………………………………………………..100
4.2 Ancestral character states (coding indicated in parentheses, see Table 4.1) for Zaluzianskya and number of changes within the genus for each character examined. Ancestral state refers to the basal condition for Zaluzianskya, not for the outgroups (e.g., for throat indumentum, a full ring of hairs around the corolla throat [1] is basal for Zaluzianskya, but the ancestral state for the entire phylogeny is having hairs inside the throat [0]) ………………………...101
xi LIST OF FIGURES
Figure Page
1.1 Representatives of some of the morphological diversity within Zaluzianskya: (A) Z. pulvinata, (B) Z. villosa, (C) Z. elongata, (D) Z. peduncularis, (E) Z. benthamiana, (F) Z. violacea, (G) Z. pusilla …………………………………..6
2.1 Flowers of (A) Z. microsiphon, (B) a putative hybrid, and (C) Z. natalensis. A long-proboscid fly (Nemestrinidae) and hawkmoth (Sphingidae) are shown pollinating Z. microsiphon and Z. natalensis, respectively (photographs of pollinators and Z. natalensis provided by S. D. Johnson) ………………………33
2.2 Map of the Kwazulu-Natal province of South Africa showing the location of each sampled population. Population abbreviations are as in Table 2.2 …….34
2.3 Neighbor-joining dendrogram inferred for 170 individuals with 52 ISSR loci, using Nei and Li’s (1979) similarity coefficient. The majority of individuals within each smaller, circled group are from one or two populations as labeled; exceptions are individually labeled (e.g. “G” within the Z. microsiphon group indicates an individual from the MG population). The single individual of Z. natalensis that was placed within the Z. microsiphon group is enclosed in a square. Putative hybrids are indicated by an “H” with a gray branch ………….35
2.4 Population band frequencies for four ISSR loci: I-7, II-4, II-9, and III-6. Populations with statistically indistinguishable (α = 0.05) frequencies are labeled with the same letter. Column color indicates populations of Z. microsiphon (black), putative hybrids (gray), or Z. natalensis (white) …………36
2.5 Plot of the first two PCA axes for the morphological traits listed in Table 2.3. Only the first PCA axis clearly describes significant variation as determined by the broken-stick model. Individuals of Z. microsiphon, Z. natalensis, and putative hybrids are indicated by squares, circles, and triangles, respectively ….37
xii 3.1 Representatives from each infrageneric taxon in Zaluzianskya: (A) section Nycterinia, Z. pulvinata (left) and Z. capensis (right); (B) sect. Macrocalyx, Z. mirabilis; (C) sect. Holomeria, Z. divaricata (top) and Z. benthamiana (bottom); (D) sect. Zaluzianskya subsect. Zaluzianskya, Z. violacea (top left), Z. affinis (top right), and Z. gracilis (bottom); and (E) subsect. Noctiflora, Z. peduncularis ………………………………………………………………….65
3.2 One of the MPTs for analyses of (A) ITS, (B) chloroplast, and (C) combined data sets with 39 accessions. Bootstrap support values greater than 50% are indicated above the branches followed by decay values; branch-lengths are provided below. Asterisks mark those nodes that are not recovered in the strict consensus. Major clades are labeled on the combined phylogeny ………..66
4.1 The distribution of Zaluzianskya (shaded) within southern Africa (Hilliard 1994) …………………………………………………………………102
4.2 One of the most parsimonious trees for combined analyses of ITS, rpl16, and trnL-F with 39 accessions. Bootstrap support values greater than 50% are indicated above the branches and decay values are provided below. Asterisks mark those nodes that are not recovered in the strict consensus. Three major clades within Zaluzianskya are marked ………………………….103
4.3 Evolution of flowering time within Zaluzianskya, (A) shown on the combined strict consensus tree. Several possible resolutions of this trait were possible, depending on the topology within clade 3. Arrows mark the locations of night-flowering taxa. Three of the potential resolutions within clade 3 are shown (bars indicate synapomorphies): (B) four derivations of night flowering and one reversal to day flowering, (C) a single derivation of night flowering but four reversals to day flowering, and (D) a single derivation of night flowering with a one reversal to day flowering …………...104
4.4 Evolution of eight characters within Zaluzianskya. Character type is indicated by letter; character state is indicated by shade (see Table 4.1). Potential synapomorphies are marked on the combined strict consensus tree; coding is shown on the right. Some of these traits are only synapomorphies in some reconstructions. The only such traits affecting major clades are “c” and “h,” e.g. for “c” if “1” is basal in Zaluzianskya, “2” is a synapomorphy for clades 1 and 3; if “2” is basal, “1” is a synapomorphy for clade 2. The precise location of synapomorphies for “a” and “b” may be shifted slightly depending on the placement of Z. microsiphon. The ITS placement for Z. mirabilis may shift basal reconstructions in clade 3 for “e” and “h” (see text) ..105
xiii 4.5 Evolution of habit, i.e. annual (0) or perennial (1), and distribution within relatively arid (0) or mesic (1) regions of South Africa, shown on the combined strict consensus tree. Character coding is indicated next to the taxon names and the three major clades are marked by boxes ………………...106
4.6 Diagram of rainfall regions within South Africa, as described by both seasonality and amount of rainfall. The area to the left of the gray line is the winter rainfall region (WRR); the area to the right is the summer rainfall region (SRR). Shaded areas indicate relatively mesic regions, unshaded areas are relatively arid (modified from Linder and Kurzweil 1999) ………….107
4.7 Photographs showing some of the character variation within Zaluzianskya for throat indumentum (A, B) and petal shape (C, D). (A) full ring of eglandular hairs around mouth of Z. villosa; (B) glabrous mouth of Z. microsiphon; (C) entire petal lobes of Z. peduncularis; (D) bifid petal lobes of Z. capensis …………………………………………………………………..108
xiv CHAPTER 1
INTRODUCTION
Zaluzianskya F. W. Schmidt (Scrophulariaceae s.s., tribe Manuleeae) is a genus of flowering plants endemic to southern Africa. It is distributed mainly within South Africa; the 55 species of the genus occur in all major climatic regions of the country. These species include plants known only from a single population as well as species spread throughout several provinces of South Africa. Although largely endemic to South Africa and Lesotho, several species also extend to the neighboring countries of Swaziland and
Namibia. Two species occur outside this range: Z. tropicalis, distributed mainly within
Zimbabwe and Mozambique, and Z. elgonensis, with a disjunct distribution within
Uganda and Tanzania (Fig. 4.1; Hilliard 1994).
In addition to variation in distributions, the diversity within this genus includes variation in habit, floral morphology, and phenology (Fig. 1.1, 3.1). Both annual and perennial plants occur within Zaluzianskya, ranging from 1 cm tall herbs to spikes over
60 cm, from mat-like plants growing over sheets of rock to laxly branched shrublets growing in sand or grasslands. Floral morphology includes a variety of sizes, shapes, colors, patterns, and assemblages – all variations on the basic theme of spikes of flowers with a narrow fused corolla tube and five petal lobes. Diversity in phenology occurs both
1 in seasonality of flowering and daily flowering patterns. Depending mainly on distribution (i.e., within the winter or summer rainfall regime of southern Africa, or both), most species of Zaluzianskya flower either within the range of November to March or from July to October; only a few species flower year-round. During their flowering season, the flowers of these species open and close daily with some species’ flowers opening during the day and others opening at night.
Originally named Nycterinia D. Don, Zaluzianskya was first studied by Bentham
(1836) who divided the 16 species known at the time into two sections, Zaluzianskya (12 species with bifid petal lobes) and Holomeria (four species with entire petal lobes).
Although Hiern (1904) expanded the number of species within the genus to 32, most of these have since been moved to other genera or demoted to synonymy (Hilliard 1994).
Taxonomy within Zaluzianskya and affiliated genera was reexamined by Hilliard and
Burtt (1983), and then more exhaustively by Hilliard (1994), resulting in a detailed monograph of the tribe Manuleeae. Hilliard recognized 55 species of Zaluzianskya, grouped into four sections and two subsections: Nycterinia, Macrocalyx, Holomeria, and
Zaluzianskya subsections Zaluzianskya and Noctiflora. To date, her monograph is the most extensive and current look at relationships within the genus – however, it does not include any type of phylogenetic framework and was based mainly on herbarium specimens.
Other studies within the genus have been scarce. The variation in floral morphology and phenology is likely of importance in pollinator interactions within the genus – potentially resulting in a diversity of pollinator relationships. However, there is little known of the breeding systems or pollination biology of Zaluzianskya. The only
2 studies done thus far were focused within section Nycterinia, which includes 19 night-
flowering species as well as a single day-flowering species (Z. microsiphon). McGregor
(1989), in an undergraduate honors thesis, considered pollination syndromes in four
species (Z. microsiphon, Z. glareosa, Z. ovata, and Z. spathacea). She found that Z. microsiphon, Z. glareosa, and Z. spathacea appear to be outcrossing (Z. ovata was not examined in the breeding system study). McGregor hypothesized, based on floral morphology, that the night-flowering species were mainly pollinated by nocturnal moths but was only able to verify this through pollinator observations for one species, Z. glareosa. No major pollinators were noted for Z. microsiphon, although small predaceous beetles (Hispidae) appeared to have potential as occasional, accidental pollinators. Johnson et al. (2002) were able to conclude more decisively that several night-flowering species within this section (Z. natalensis, Z. elongata, and Z. pulvinata) are pollinated by hawkmoths whereas long-proboscid flies pollinate the day-flowering species (Z. microsiphon). Using crossing studies, they also found that both Z. microsiphon and Z. natalensis are obligate outcrossers.
Floral morphology within section Nycterinia is relatively constant when compared with the rest of the genus. Future studies in other species within Zaluzianskya are likely to reveal additional types of pollinator interactions and breeding systems. For example, Hilliard (1994) noted that some species appear at least in general morphology to be self-fertilizing and some have both chasmogamous and cleistogamous flowers. Both
Hilliard (1994) and Johnson et al. (2002) mention the probability for hybridization between some species of Zaluzianskya, although prior to the studies discussed here, no work had been done to investigate this possibility.
3 I have used phylogenetic analyses and molecular markers to gain insight within
this diverse genus, along three main lines of inquiry (discussed in the following chapters).
Each of these studies have been written as separate manuscripts for future publication and
then compiled into this dissertation. First, two species (Z. microsiphon and Z. natalensis) that are believed to hybridize were examined at the population level. Although these species appeared to be reproductively isolated by flowering time (day flowering and night flowering, respectively), putative hybrids have been collected in an area where individuals of each species occur together. Both ISSR markers and morphological traits were examined to look for patterns of gene flow between the species and their putative hybrids.
Second, DNA sequences from the nuclear and chloroplast genomes were analyzed to resolve the first broad phylogenetic trees for this genus and to test the current classification for the group. The monophyly of both the genus and its sections and subsections had previously never been tested, and relationships among the sections were largely unknown. Results of these analyses indicate that several of these groups are not monophyletic; implications of this are discussed in Chapter 3.
Finally, the phylogenies were also used in an investigation of character evolution within Zaluzianskya. Characters relevant to taxonomy or pollinator interactions were examined in addition to a preliminary inspection of the relationship between habit and biogeography. Due to the reorganization of the genus indicated by the molecular phylogenies, it is important to seek morphological synapomorphies for the newly resolved infrageneric groups. Additionally, the floral diversity within this genus bears investigation both to gain understanding of the evolution of this diversity and also to
4 combine with future work on pollinator interactions in the group. The evolution of habit and changes in distribution are also discussed. Hilliard (1994) noted a trend within
Zaluzianskya of annual species occurring in the western portion of southern Africa and perennial species occurring in the east. This potential association was examined by mapping habit and distribution (as it relates to rainfall regimes) onto the phylogenetic trees.
The various species of Zaluzianskya are frequent members of many of the diverse ecosystems within southern Africa. This region is known for its remarkable botanical diversity, much of which has yet to be studied (Goldblatt and Manning 2002, G. Smith et al. 1996). This research represents the first phylogeny-based investigations across this distinct genus, as well as the first genetic study of hybridization within this group.
5 Figure 1.1. Representatives of some of the morphological diversity within Zaluzianskya: (A) Z. pulvinata, (B) Z. villosa, (C) Z. elongata, (D) Z. peduncularis, (E) Z. benthamiana, (F) Z. violacea, (G) Z. pusilla. 6 CHAPTER 2
HYBRIDIZATION AND GENE FLOW BETWEEN A DAY- AND NIGHT-
FLOWERING SPECIES OF ZALUZIANSKYA.
INTRODUCTION
Hybridization has been shown to be an important evolutionary process in plants.
It can have implications on the status of species themselves, both through the creation of hybrid species and through the effects of hybrid individuals on gene flow between their parental species (reviewed by Rieseberg 1995, Rieseberg and Wendel 1993, Arnold
1992). When hybridization is discovered between species, two key questions are: (1) what is incomplete about their isolating mechanism and (2) what effect is this having on the genetic structure of these species and their hybrids? Isolating mechanisms can occur at many levels, ranging from distributional, ecological, or structural differences between the parental species to hybrid inviability or sterility (reviewed in Stebbins 1950). As a result, a breach in these barriers can occur in multiple ways. For example, isolation of parental species may be violated through range expansion or habitat alteration, either natural or anthropogenic (e.g., Capula 2002, Wolf et al. 2001, Francisco-Ortega et al.
2000, Runyeon-Lager and Prentice 2000, Levin et al. 1996). Plant-animal interactions
7 may also be an important factor in hybridization. For example, an herbivore in one part of the parental distributions could alter the relative fitness of hybrids versus parental species in those regions, or a new pollinator could be introduced and facilitate gene exchange between two species that were previously separated by ethological isolation.
Once hybridization has occurred, there is a wide range of possible consequences, ranging from a small, inert zone of hybrid individuals to the complete merging of parental species. Hybridization is often suspected because of some degree of morphological intermediacy and examination of morphological traits can be very useful in studies of hybrids. However, morphology does not always reflect the degree to which genomes have become contaminated. Molecular markers may allow a more accurate estimation of the permeability of species barriers (e.g., Hodges and Arnold 1994).
This study involves the putative hybridization occurring between a day- and night-flowering species of Zaluzianskya F. W. Schmidt (Scrophulariaceae s.s., tribe
Manuleeae): Z. microsiphon (O. Kuntze) K. Schum. and Z. natalensis Krauss, respectively. Previous field studies have shown apparent pollinator specificity for each species, with long-proboscid flies (Nemestrinidae) pollinating Z. microsiphon and hawkmoths (Sphingidae) pollinating Z. natalensis (Johnson et al. 2002). Despite this seeming ethological isolation, putative hybrids have been observed in an area where the two species occur sympatrically (Johnson et al. 2002). The intermediacy of these individuals in morphology and flowering time lead one to expect that they are hybrids, despite the brevity of overlap in daily flowering times for the presumed parental species.
The goals of this study were to use ISSR markers to test the hybrid status of these individuals and to investigate potential gene flow between the species, both in the area of
8 sympatry and in several isolated populations of each species. In addition, several
morphological traits were analyzed to determine if ordination of these characters also
supported the hybrid hypothesis.
Study group
Zaluzianskya includes 55 species of annual and perennial herbs and shrublets
(Hilliard 1994). Both Z. microsiphon and Z. natalensis are placed within section
Nycterinia (D. Don) Hilliard, and this placement is supported by phylogenetic analyses of
both nuclear and cpDNA sequences (see Chapter 3). Section Nycterinia consists of 20
species. Pollination biology for this section has been examined in five species, all of
which are pollinated by insects (hawkmoths, or, in the case of Z. microsiphon, long-
proboscid flies; Johnson et al. 2002, McGregor 1989). Members of this section are
morphologically distinct from the rest of the genus, sharing a suite of floral characters
including a long, narrow corolla tube, five notched petal lobes, and contrasting red and
white petal coloration (Fig. 2.1). However, there are several notable differences between
the flowers of Z. microsiphon and most of the other species within this section, including
Z. natalensis (Table 2.1). These characteristics appear to be derived (see Chapter 4,
Johnson et al. 2002) and several seem adaptive for pollination by long-proboscid flies.
Being the sole day-flowering species in this section is obviously adaptive for pollination
by day-flying pollinators. Additionally, several features of Z. microsiphon (e.g., lack of scent and zygomorphic flowers) are shared with unrelated species utilizing the same
9 pollinator, such as species of Pelargonium, Gladiolus, and Disa. This suggests that these characteristics may be part of a pollination syndrome for long-proboscid flies (Goldblatt and Manning 2000).
Although mainly open at different times of the day, there is some potential overlap in time and space between Z. microsiphon and Z. natalensis that could provide opportunity for hybridization. The distributions of these species overlap within the
Kwazulu-Natal province of South Africa, although currently I am only aware of one set of locally sympatric populations. Zaluzianskya microsiphon occurs generally at higher altitudes than Z. natalensis (although again with overlap: 1530-2750 m vs. 600-1700 m, respectively), but habitat preferences appear to be similar; each species occurs on hillside grasslands. Both species flower generally between January and March (Hilliard 1994).
METHODS
ISSR survey
Ten populations were included in this study: sympatric populations of both species and the putative hybrids, five isolated populations of Z. microsiphon, and two of
Z. natalensis (Table 2.2). All populations were within the Kwazulu-Natal province of
South Africa (Fig. 2.2). Leaf material (~20 mg) was collected on silica gel from 11-20 individuals within each population. Voucher specimens were deposited in the Ohio State
University (OS) and University of Natal (NU) herbaria. DNA was extracted from leaf samples using standard CTAB methods (Doyle and Doyle 1987) and cleaned using an
EluKwik DNA purification kit (Schleicher and Schuell, Keene, NH).
10 ISSR markers are generated by single-primer PCR, using primers designed from a microsatellite motif with a one to three nucleotide anchor on either the 3’ or 5’ end (to eliminate strand-slippage artifacts; Wolfe et al. 1998, Wolfe and Liston 1998). ISSRs have been used successfully in numerous population genetic studies for natural populations and also specifically in the study of hybridization (e.g., Datwyler 2001,
Wolfe et al. 1998). They are usually as, or more, variable than RAPDs, while being more robust to slight changes in DNA concentration and buffers (Esselman et al. 1999, Wolfe et al. 1998). In addition, they retain the benefits of other PCR-based techniques such as the need for very little template material.
Forty-three ISSR primers were screened for the presence of variation and species- marker bands. Four primers were found suitable after optimization and used in this
study: (CA)7YG, (CTC)7RC, (CA)6RG, and (AG)8RG. For simplicity, these primers will be referred to as primer I, II, III, and IV, respectively. PCR reactions (25 µl) consisted of
1 µl DNA, 1 µM primer, 1X buffer (0.02 M Tris, 0.05 M KCl, and 0.001% Tween-20),
2.9 mM MgCl2, 0.2 mM dNTPs, and 0.5-1.25 U Taq DNA polymerase (Invitrogen; 0.5 U for I and IV, 0.75 U for III, and1.25 U for II). An Eppendorf MasterCycler Gradient thermocycler was used with the following settings: 110 s at 94˚C; 35 cycles of 45 s at
94˚C, 45 s at 51˚C, and 110 s at 72˚C; 10 min at 72˚C. All reactions were replicated and only those bands that were visible in both replicates were scored for analysis. Each thermocycler run included a negative control reaction of 25 µl with all of the reagents except for the DNA.
ISSR reactions were electrophoresed on 1.5% agarose gels in 1X TAE buffer.
The entire reaction volume was loaded into prepared wells and each gel was run until the
11 bromophenol blue indicator dye had traveled 10 cm. Afterwards the gels were stained in ethidium bromide and visualized using UV light. An imaging system (Alpha Innotech
Corporation) was used to record the gels as a TIFF file that was transferred to a
PowerMac 7500 and examined using Kodak 1D image analysis software (Eastman
Kodak Company). The 1 kb plus ladder size standard (Invitrogen) was used to estimate band sizes. Loci were designated based on fragment size; bands were scored as diallelic
(1 = band present, 0 = band absent).
ISSR data analyses
Genetic clustering of populations and species. Nei and Li’s (1979) coefficient
was used to assess pair-wise individual similarity: 2NAB/(NA+NB), where NAB is the number of shared bands, NA is the total number of bands in taxon A, and NB is the total number of bands found in taxon B. This coefficient is appropriate for analyses of ISSR data because it does not include shared absences in the calculation of similarity values.
As with RAPDs, absences in ISSR banding patterns can occur for a variety of reasons and thus should not be considered homologous (Wolfe and Liston 1998). Distance matrices were created using !WXDNL (Vera Ford, University of California, Davis) and
RAPDPLOT 3.0 (using option “S;” William C. Black, IV, Colorado State University).
These programs produce identical results but the former is formatted for use with NTSYS
(Rohlf 1998) whereas the latter is formatted for PHYLIP’s NEIGHBOR (Felsenstein
1993). A neighbor-joining analysis was run in NTSYS to determine whether there were tied trees (Backeljau et al. 1996), but no such trees were found and the remainder of the neighbor-joining analyses were run in NEIGHBOR. RAPDPLOT has the capability of
12 producing a distance matrix along with a set of bootstrapped distance matrices from the original data. These latter matrices can then be inputted into NEIGHBOR using the multiple datasets option to produce a set of bootstrap trees. After saving these trees in nexus format using TREEVIEW (Page 1996), their majority rule consensus was calculated in PAUP* 4.0b10 (Swofford 2003) to determine the level of bootstrap support for neighbor-joining groups. All runs of NEIGHBOR employed the jumble option to randomize taxon addition order prior to construction of the tree, this also randomizes taxon order between bootstrap replicates (Farris et al. 1996). Neighbor-joining analyses were run with and without hybrid individuals included.
A Mantel test (Mantel 1967) was performed using NTSYS to determine if genetic distance was correlated with geographic distance. Geographic distances were calculated based on GPS coordinates for each population (Table 2.2). Genetic distances were derived from population similarity values calculated using !WAVSIML (V. Ford). A normalized Mantel statistic was used and significance was tested with the maximum number of random permutations (9,999).
Hybridization and gene flow. The distribution of each ISSR locus was examined in all populations to ascertain patterns of gene flow. Ideally, species-specific marker bands (i.e. bands that are present in all individuals of one species and none of the other) would be found and morphologically intermediate individuals would be examined for additivity of these bands to test the individuals’ hybrid status (e.g., Steen et al. 2000,
Cruzan and Arnold 1993). However, it is frequently not possible to find such marker bands, possibly due to the often-close relationship of the parental species or to introgression. Several studies of hybridization have therefore used less stringent methods
13 for defining marker bands, employing bands that are more common in one taxon versus the other, to some degree (Feliner et al. 2002, Datwyler 2001, Neuffer et al. 1999, Wolfe et al. 1998, Allan et al. 1997). In this study, gene flow was traced in three ways. First, those bands that occurred in one species in at least a 25% higher frequency than the other were considered species-typical bands. Band frequencies in the proposed hybrids were then examined to determine if they were intermediate to those found in the two species.
This is a modification of the method used by Wolfe et al. (1998). Second, bands that occurred in only one species plus the hybrids, or sometimes also extended into the sympatric population of the other species, were noted regardless of the band frequency difference between Z. microsiphon and Z. natalensis. Third, band frequencies for each locus were inspected at the population level to determine whether there were statistically significant differences between populations. This was tested using a modification of
Fisher’s exact test (Raymond and Rousset 1995), implemented in TFPGA (Miller 1997) with 10,000 permutations, 50 batches, and 5,000 dememorisation steps.
Morphological survey of sympatric populations
Nine morphological characters were assessed in the sympatric populations of Z. microsiphon, Z. natalensis, and their putative hybrids (Table 2.3). These traits were chosen due to their apparent or potential distinctness in each species. Eighteen individuals of Z. microsiphon, 19 of Z. natalensis, and 11 hybrid individuals were examined (48 total). Ten of these individuals from each population were also included in the ISSR survey. Seven of the morphological traits were measured for all 48 individuals
14 whereas two (plant height and floral orientation) were measured for about half of the individuals of each population. It was not possible to measure these two traits in the other half due to the condition of the plant collections.
Morphological data analyses
Principal components analysis (PCA) was used to explore the morphological variation of sampled individuals and populations. These analyses were conducted in
NTSYS using standardized data (subtracting the average and dividing by the standard deviation) and a matrix of correlations. A Mantel test was also run for those individuals that were sampled for both morphological and ISSR data (including ten individuals from each population) to determine whether the distance matrices for these data were correlated. These matrices were compared on an individual level, rather than the population level used for the Mantel test between genetic and geographic distance. The
ISSR distance matrix was constructed from a similarity matrix produced by !WXDNL.
The morphological distance matrix was created in NTSYS using standardized data and the Manhattan distance coefficient. The plant height character was excluded in these latter analyses because this trait was not measured in most of the individuals that were sampled for both data sets (floral orientation, in contrast, was sampled for nearly all of these individuals). The Mantel test was conducted as described above.
15 RESULTS
Genetic clustering of populations and species.
Neighbor-joining analyses of the 52 scored ISSR loci produced a single, unrooted tree both for the dataset with putative hybrids (170 individuals total) and without (159 individuals). Although there were some differences in the placement of individuals between these two trees, the main relationships indicated are similar and so only the tree including all individuals will be shown (Fig. 2.3). Bootstrap support for groups within this tree was very low (not shown). Only 15 groups on the tree had any support greater than 50%, all but one of which consisted of only two individuals (the exception included three individuals).
Populations were not resolved as separate groups within the neighbor-joining tree, but several geographically proximate populations formed mixed clusters. For example, individuals from the MR and MS populations were largely intermixed in one group.
These two populations are both located close to Sentinal Peak and are in closer proximity to each other than any of the other sampled populations of Z. microsiphon. Other major groups were formed, each largely consisting of one to two populations (Fig. 2.3).
However, for each population at least a few individuals occurred outside of that population’s main cluster, and individuals from MP were found within multiple groups.
With a few exceptions, the two species were resolved as separate groups. One exception was a small cluster of individuals from both species along with putative hybrid individuals – all from the sympatric Mt. Gilboa populations (MG, NG, and HG).
Additionally, one individual of Z. natalensis (from the sympatric NG population)
16 clustered within the main Z. microsiphon group, although this did not occur in the analysis without hybrids. Hybrid individuals were scattered in the tree, clustering with sympatric individuals from both Z. microsiphon (MG) and Z. natalensis (NG).
A positive and highly significant correlation was found between genetic distance
(Nei and Li 1979) and geographic distance (r = 0.73155, P = 0.0002, Mantel test). This correlation was also found when only looking at populations of Z. microsiphon (r =
0.84532. P = 0.003) and when looking at those populations plus the hybrid population (r
= 0.81041, P = 0.0008). Thus, the correlation between genetic and geographic distance was seen at the population level and was not due solely to differences between the species.
Hybridization and gene flow.
No completely species-specific ISSR loci were found (i.e., present in all individuals of one species and no individuals of the other). However, 15 loci occurred with at least a 25% difference in band frequencies between Z. microsiphon and Z. natalensis (Table 2.4); these loci were used to examine gene flow. For eight of these
“species-typical” loci the putative hybrids had band frequencies intermediate to those found in the presumed parental species, corroborating the hypothesis that they are in fact hybrids. In four of the other seven loci, the hybrids had higher band frequencies than either of the two species, whereas the hybrids lacked the other three loci despite their presence in both parental species. These seven loci do not necessarily support the hybrid hypothesis although there are several possible explanations for this pattern, regardless of whether these individuals are hybrids or not (see discussion).
17 Two species-typical loci (III-6 and III-12) suggest unidirectional gene flow because they are present only in Z. microsiphon, the putative hybrids, and the sympatric population of Z. natalensis. Seven other loci, although differing in less than 25% frequency between the two species, showed a similar pattern, and one locus showed the reverse pattern (Table 2.5).
Further examination of gene flow between these two species was accomplished by statistically comparing band frequencies for each locus across populations using a modification of Fisher’s exact test. Four loci were found to demonstrate a significant pattern of band frequencies relevant to the question of hybridization (Fig. 2.4, P-values given in Table 2.6). Although there was intraspecific variation for each species, there were significant differences between many populations of Z. microsiphon and Z. natalensis, with hybrid individuals often intermediate between the two.
Morphological analyses
All morphological analyses were conducted both for the full morphological data set and for a data set excluding the two characters with missing data (plant height and flower orientation). The results of these two sets of analyses were similar and thus only those for the larger data set are discussed. The first PCA axis accounted for 58.0% of the variation, the second for 20.9%, and the third for 8.0%. According to comparisons with the broken-stick model (Jackson 1993, Joliffe 1986, Frontier 1976), only the first axis, and possibly to a small extent the second axis, accounted for a significant amount of variation. The variation shown in the first axis was clearly significant, whereas the percent variation accounted for by the second axis was nearly identical to that produced
18 under the broken-stick model and thus of marginal or no significance according to that
model. Many of the variables contributed to the loading on the first axis, especially floral
orientation (coefficient = -0.950), throat indumentum (0.924), floral symmetry (0.887),
node length (-0.840), and tube length (0.821). Thus, individuals were divided generally
into those with vertically oriented petal lobes, glabrous throats, zygomorphic flowers,
longer distances between flowers on inflorescences, and shorter corolla tubes (i.e.,
morphology associated with Z. microsiphon) and those with the opposite: horizontally
oriented petal lobes, a ring of hairs around the throat, actinomorphic flowers, shorter
distances between flowers on inflorescences, and longer corolla tubes (i.e., morphology
associated with Z. natalensis). Only leaf length (0.778) and plant height (0.720) had
coefficients greater than 0.7 for the second axis (Table 2.7). A plot of the first two axes
resulted in spatial separation of the two species along the first axis but not the second.
The hybrids are distributed within the area between the two species but are more
concentrated towards individuals of Z. natalensis (Fig. 2.5).
Genetic (Nei and Li) and morphological (Manhattan) distances were significantly
correlated (r = -0.19203, P = 0.0012, Mantel test), but the relationship was weak and
slightly negative.
DISCUSSION
Zaluzianskya microsiphon and Z. natalensis are seemingly reproductively isolated species, separated by pollinator preferences due to differences in flowering-time along with differences in floral symmetry, scent, orientation, and indumentum (Table 2.1).
19 However, as noted by Johnson et al. (2002), apparent hybrids occur in the sympatric Mt.
Gilboa populations of these species. These putative hybrid individuals appear
intermediate in flowering time and general morphology (personal observation, Johnson et
al. 2002). The results of the molecular analyses support the possibility of gene flow
between the two species. These data are somewhat inconclusive due to the genetic
similarity of the proposed parental species, however their combination with the
morphological analyses strengthens the case for introgression.
The neighbor-joining tree suggests a geographic component to genetic diversity
for the two species and their putative hybrids, as would be expected if gene flow is
occurring between rather than just within populations. Adjacent populations merged into
mixed groups in the tree rather than forming distinct clusters, and this occurred both
within and between species (Fig. 2.3). The correlation between genetic distance and
geographic distance was verified by the highly significant results of the Mantel test.
Thus, if hybridization is occurring in a few sympatric populations, the effects could be
spreading to nearby allopatric populations. Alternatively, it is possible that the genetic
similarity between populations could be due to recent colonization from a common
ancestor. A choice cannot be made between these hypotheses of historical similarity vs.
current gene flow based on these data.
Zaluzianskya microsiphon and Z. natalensis largely cluster separately within the neighbor-joining tree. However, there is some mixing, particularly in a small group of individuals from the Mt. Gilboa populations, including hybrids and members of both species. The fact that these individuals cluster together rather than with other members of their own species corroborates the hypothesis of interspecific gene flow, although this
20 tree has very low bootstrap support. Additional hybrids are found in several groups across the tree; however, they are always found with individuals from the sympatric populations (MG and/or NG). The placement of these hybrids on the neighbor-joining tree and the morphological PCA plot were compared but no obvious pattern was apparent. Those hybrids that clustered with a particular species on the tree did not necessarily fall closest to individuals of that species on the PCA plot. Instead they were spread across the range between the two species on the PCA plot (Fig. 2.5).
The ISSR loci were also examined individually, first at the species level and then at the population level. Fifteen species-typical loci were found (Table 2.4). As explained above, band frequencies at a particular locus must differ by at least 25% between the two species for the locus to be considered species-typical. By this definition, a species- typical locus does not necessarily need to be exclusively found in one species. In the literature, species markers have been defined in a variety of ways; this is a similar, although less stringent, definition to those used by several other studies (e.g., Datwyler
2001, Neuffer et al. 1999, Wolfe et al. 1998). Ironically, the difficulty in locating more conclusive species-typical markers could be due to the very hybridization and gene flow that I am trying to detect. In fact, if the sympatric Mt. Gilboa populations are excluded from the analyses, the number of loci that are only found within a single species is raised from ten to 19 (including two species-typical loci). In contrast, if any two other populations are removed from the analyses (one from each species), this number is only raised to 13 at the most.
However, in addition to hybridization, overlap in loci could also be due simply to a high level of genetic similarity between the species. Excluding floral characters, these
21 species are nearly identical, and parsimony analyses of morphology place these taxa in a clade, along with another very similar species, Z. spathaceae (Johnson et al. 2002).
Regardless, it is possible to see some trends in the data at hand given the caveat that these marker bands are not truly species-specific.
Putative hybrids had intermediate band frequencies relative to Z. microsiphon and
Z. natalensis for eight of the fifteen marker bands (Table 2.4), lending support to the hypothesis of hybrid origin. The remaining seven marker bands showed “extreme” frequencies in the hybrids, that is, the hybrid band frequencies were either higher or lower than either of the putative parental species. Only the three loci with lower hybrid band frequencies were not additive in the hybrids, in that bands for those loci were completely absent in the hybrids while being present in at least some individuals of both proposed parental species. Incomplete additivity in these types of markers (i.e., RAPDs and ISSRs) for hybrids has been found in multiple previous studies (e.g., Feliner et al.
2002, Huang et al. 2000, Steen et al. 2000, J. Smith et al. 1996, Huchett and Botha 1995).
The extreme band frequency values, including the absence of those three loci in hybrid individuals, could possibly be explained by several factors, regardless of whether those individuals truly are hybrids. The fact that ISSRs are dominant markers means that a heterozygous individual will show the same banding pattern as a dominant homozygous individual. Thus, some hybrids could lack a particular band due to inheriting the absent
“allele” from two heterozygous parents. Probably of more importance in this study is the fact that both proposed parental species are polymorphic for all marker bands. The band frequency for one population of putative hybrid individuals was compared to the average band frequency for several populations of each presumed parental species. However, due
22 to intraspecific variation, the average band frequency for each species did not always reflect population-level band frequencies. For example, the band frequencies of the two
Sentinal populations of Z. microsiphon (MR and MS) were often distinct from those found in other populations of that species. This difference was statistically significant for eight of the 52 ISSR loci (results shown for III-6, see Fig. 2.4). After comparing the hybrid population band frequencies to those found in each population of each species, I found that all seven marker bands with supposedly “extreme” values were actually well within the normal range of population-level frequencies found in one or both parents.
This was verified statistically for all seven bands (results shown for III-6, see Fig. 2.4).
The band frequencies for the putative hybrids were statistically indistinguishable from at least some of the populations of one or both proposed parental species. Also, one or both of the sympatric populations (MG and NG) were always among those populations with similar band frequencies to the hybrids – again congruent with the possibility of hybridization. This was true for all 15 species-typical loci.
When examining populations individually, four of the 15 marker bands were found to have a statistically significant pattern relevant to hybridization (Fig. 2.4 and
Table 2.6). Although, again, variation within species complicates interpretation, several trends are visible in these four loci. The band frequencies for the putative hybrids appear to be intermediate between the two species. In all four loci, the populations of Z. microsiphon have (in general) higher band frequencies than those of Z. natalensis, with the hybrids falling in-between. For three of these loci (I-7, II-9, and III-6), the hybrid population is statistically indistinguishable from several populations of Z. microsiphon and the sympatric population of Z. natalensis (NG), while being significantly different
23 from the other populations of Z. natalensis. At the other locus (II-4) the hybrid band frequency is statistically indistinguishable from one allopatric Z. natalensis population
(NA) and the sympatric Z. microsiphon population, while the other populations of both species are significantly different. One potential explanation for these patterns is that gene flow is occurring, possibly preferentially from Z. microsiphon to Z. natalensis.
This interpretation is also supported by the data shown in Table 2.5. Although these loci (with the exception of III-6 and III-12) do not show a >25% difference in band frequencies between the species, they do demonstrate a pattern that could indicate gene flow. In nine of these ten loci, ISSR bands were found in Z. microsiphon and in the hybrids, but not in Z. natalensis – except, in six cases, for the sympatric population of Z. natalensis (NG). This pattern could again be explained by differential gene flow from the day-flowering Z. microsiphon into the hybrids and the night-flowering Z. natalensis.
One locus (IV-1) shows the reverse pattern – supporting the hypothesis of hybridization but in this case implying gene flow from Z. natalensis into the hybrids.
Hybrids were identified in the field by their generally intermediate morphology and flowering time. However, when nine morphological traits were examined more closely using PCA (Table 2.3), they indicated that although the putative hybrid individuals occurred across the range between the clusters of Z. microsiphon and Z. natalensis individuals, most were placed very close to individuals of the latter species
(Fig. 2.5). This is particularly noticeable if you ignore the second axis, whose significance was questioned by comparisons with a broken-stick model. Thus, the ISSR data suggest that the hybrids are potentially more similar to Z. microsiphon whereas the morphological data indicate that many of them are more similar to Z. natalensis. Mantel
24 tests comparing these two data sets also show a significant (P = 0.0012) negative correlation, albeit a very weak one (r = -0.19203). One possible conclusion based on the ordination analyses alone is that many of these “hybrids” are misidentified individuals of
Z. natalensis. However, the differences seen in these individuals extend beyond the generally excepted range of variation for this species (Hilliard 1994), and this would also conflict with the ISSR data. A stronger hypothesis is that these individuals are backcrosses with Z. natalensis. It appears that the Z. natalensis-like individuals are more genetically similar to Z. microsiphon (at least in some portions of their genomes), supporting the possibility of introgression of Z. microsiphon genetic material into the night-flowering species. Perhaps introgression of morphological characters of Z. microsiphon is more strongly selected against, as seen in Hodges and Arnold’s (1994) study of Aquilegia and Goulson and Jerrim’s (1997) study of Silene. Further investigation is necessary to examine that possibility. Although additional molecular and morphological data should aid in clarifying the degree and direction of gene flow, the unidirectional nature of this gene flow is corroborated by these data.
It appears plausible that pollen flow is also unidirectional in these taxa. As noted above, long-proboscid flies pollinate Z. microsiphon while Z. natalensis is pollinated primarily by hawkmoths. Only one exception to this pollinator specificity has been observed; a single long-proboscid fly was seen probing flowers of Z. natalensis after visiting Z. microsiphon, during the brief period at dusk when flowers of both species are open (Johnson et al. 2002). This pattern of visiting Z. natalensis after Z. microsiphon may be typical of natural crosses between these species. Flowers of Z. microsiphon become less visible than those of Z. natalensis as the light fades. These flowers are not
25 only beginning to close, but also their upper petal surface is duller than flowers of Z. natalensis (greenish-white vs. white; Hilliard 1994). The lower petal surface of both species is dark red, seeming to camouflage closed and closing flowers. While Z. microsiphon flowers are closing, flowers of Z. natalensis are beginning to open and becoming clearly visible in the low-intensity light. Thus, differential pollen flow could be occurring towards Z. natalensis when late afternoon turns to evening, with some pollinators picking up pollen from Z. microsiphon before those flowers close and carrying that pollen to the increasingly more visible and prevalent flowers of Z. natalensis. The reverse flow of pollen could be less likely because as more Z. natalensis flowers are open and available for pollen-donation, more of the Z. microsiphon flowers are closed (or at least are less conspicuous). It is unlikely that these species are also hybridizing in the morning hours. Long-proboscid flies are not active in the morning until after flowers of Z. natalensis are completely closed, and flowers of Z. microsiphon do not open early enough in the morning to be affected by moth activity.
Differential pollen flow has been shown in previous studies to result in unidirectional gene flow. For example, in Aesculus, migrating hummingbirds disperse pollen far north of the range of one parental species – leading to introgression only with a single parental species (dePamphilis and Wyatt 1989, 1990). Seed dispersal for these
Aesculus species appears to be localized and thus does not play a role in gene flow between the species. Unidirectional gene flow in several species of Penstemon is also hypothesized to be due to differential pollen flow, as a result of both the timing of pollinator visits to different species as well as differences in floral morphology (Wolfe and Elisens 1995). Concluding whether an analogous mechanism is influencing gene
26 flow in these species of Zaluzianskya requires further study of pollinator interactions between these populations. This will help to determine the potential causes for unidirectional gene flow, such as hybrids crossing more easily with individuals of Z. natalensis, possibly through differences in fruit abortion, pollen germination rates, floral mechanics, or pollinator behavior (Tiffin et al. 2001, Ellis and Johnson 1999, Barton and
Hewitt 1985).
The intermediacy of putative hybrids for some morphological and especially phenological traits could have implications for hybrid fitness. Many flowers of these individuals remain partially open throughout the day and evening, but are never fully open. This would be expected to be unattractive to either species’ pollinators, possibly leading to a “dynamic-equilibrium” hybrid zone, which is maintained by a balance between dispersal and hybrid inferiority (Arnold 1992, Barton 1979). However, the preliminary data given here suggests that the hybrids may contribute to further gene flow, with some introgression of Z. microsiphon genetic material into Z. natalensis. Field studies are necessary to determine if pollinators are visiting the putative hybrids. The data in this study are consistent with the hypothesis that hybridization is occurring with introgression. However, the genetic similarity of these two species makes it difficult to more conclusively support these hypotheses over alternatives such as historical similarity.
Future work will focus on increasing population sampling and developing other markers for examining these questions.
27 Species Anthesis Symmetry Petal limbs Scent Throat fringe Z. microsiphon Diurnal Zygomorphic Vertical None Absent Z. natalensis Nocturnal Actinomorphic Horizontal Strong Present
Table 2.1. Several of the floral differences between Z. microsiphon and Z. natalensis.
Taxon Pop. Location Voucher Sample Label Size Z. microsiphon MR Witsiehoek Resort, near Sentinal, SJ 19ii00 20 Royal Natal National Park s. n. (S28°41.161’ E28°54.007’) MS Road to Sentinal, JKA 100 15 Royal Natal National Park (S28°42.072’ E28°53.836’) MP Sani Pass, JKA 98 13 Ukhahlamba-Drakensberg Park (S29°35.857’ E29°18.712’) MC Garden Castle, JKA 89 12 Ukhahlamba-Drakensberg Park (S29°44.500’ E29°11.900’) MB Bushman’s Nek, JKA 84 20 Ukhahlamba-Drakensberg Park (S29°50.410’ E29°12.853’) MG Mt. Gilboa, SJ 6ii00 20 Karkloof Mt. Range s. n. (S29°17.228’ E30°17.561’) Putative hybrid HG Mt. Gilboa SJ 6ii00 11 s. n.
Z. natalensis NG Mt. GilboaSJ 20 6ii00 s. n., JKA H NK Krantzkloof Nature Reserve, JKA 114 20 Durban (S29°49.072’ E30°48.156’) ND Near Kloof Country Club, JKA 117 19 Durban (S29°48.182’ E30°49.750’)
Table 2.2. Populations of Zaluzianskya included in this study. 28 Morphological Abbreviation Definition trait
Floral Orientation FO Angle between the corolla tube and upper petal lobes
Throat TI Unicellular hairs around throat: none to few (0), Indumentum partial ring (1), or complete ring (2)
Floral Symmetry FS Top two petal lobes overlap (0) or separate (1)
Floral Node NL Average length from base of one flower to another Length (mm)
Corolla Tube TL Length from top of ovary to throat of flower (mm) Length
Petal Width PW Width at widest point of lowest petal lobe
Inflorescence IL Length from base of lowest bract to top of highest Length bract (mm)
Plant Height PH Length from base of plant to top of bracts (mm)
Leaf Length LL Length of highest leaf (mm)
Table 2.3. Morphological traits measured in sympatric individuals of Z. microsiphon, Z. natalensis, and their putative hybrids.
29 Loci with intermediate “hybrid” band frequencies I-2 I-7 I-8 II-4 II-9 III-2 III-12 IV-7 M 0.69 0.79 0.41 0.82 0.85 0.47 0.35 0.65 H 0.82 0.55 0.45 0.36 0.73 0.64 0.09 0.91 N 0.95 0.21 0.75 0.12 0.27 0.76 0.03 1.00 Loci with extreme “hybrid” band frequencies II-1 II-2 II-3 III-1 III-5 III-6 IV-3 M 0.12 0.37 0.73 0.29 0.33 0.53 0.11 H 0.00 0.00 1.00 0.00 1.00 0.64 0.45 N 0.49 0.10 0.98 0.03 0.92 0.15 0.36
Table 2.4. Band frequencies of the fifteen species-typical loci (differ by at least 25% between the putative parental species). M, H, and N indicate the average band frequencies for the Z. microsiphon, putative hybrid, and Z. natalensis individuals, respectively.
Loci found only in Z. microsiphon Locus found only in Z. natalensis and “hybrids” and “hybrids” I-9 I-10 IV-5 IV-1 M 0.06 0.09 0.17 0.00 H 0.09 0.09 0.18 0.27 N 0.00 0.00 0.00 0.14 Loci found in Z. microsiphon, “hybrids,” and the sympatric population of Z. natalensis II-5 II-6 II-10 III-6 III-7 III-12 M 0.29 0.18 0.16 0.53 0.18 0.35 H 0.45 0.55 0.09 0.64 0.18 0.09 N 0.08 0.07 0.05 0.15 0.03 0.03
Table 2.5. Band frequencies for ten loci that occur only in one proposed parental species, the putative hybrids, and sometimes in the sympatric population of the other proposed parental species. The two loci that are also considered species-typical (see Table 2.4) are underlined.
30 I-7 MR MS MP MC MB MG HG NG NK ND MR 0.4924 0.0276 0.419 0.0065 0.2680 0.6973 0.5073 0.0002 0.0003 MS 0.4924 0.0066 0.2174 0.0010 0.0683 1.0000 1.0000 0.0041 0.0051 MP 0.0276 0.0066 0.2200 1.0000 0.2523 0.0107 0.0040 0.0000 0.0000 MC 0.4190 0.2174 0.2200 0.1331 1.0000 0.1941 0.1292 0.0000 0.0000 MB 0.0065 0.0010 1.0000 0.1331 0.1060 0.0028 0.0005 0.0000 0.0000 MG 0.2680 0.0683 0.2523 1.0000 0.1060 0.1041 0.0783 0.0000 0.0000 HG 0.6973 1.0000 0.0107 0.1941 0.0028 0.1041 1.0000 0.0058 0.0071 NG 0.5073 1.0000 0.0040 0.1292 0.0005 0.0783 1.0000 0.0034 0.0033 NK 0.0002 0.0041 0.0000 0.0000 0.0000 0.0000 0.0058 0.0034 1.0000 ND 0.0003 0.0051 0.0000 0.0000 0.0000 0.0000 0.0071 0.0033 1.0000 II-4 MR MS MP MC MB MG HG NG NK ND MR 1.0000 1.0000 1.0000 0.6040 0.0000 0.0008 0.0000 0.0001 0.0000 MS 1.0000 1.0000 1.0000 0.2435 0.0000 0.0005 0.0000 0.0001 0.0000 MP 1.0000 1.0000 1.0000 0.2614 0.0001 0.0009 0.0000 0.0002 0.0000 MC 1.0000 1.0000 1.0000 0.2745 0.0001 0.0014 0.0000 0.0005 0.0000 MB 0.6040 0.2435 0.2614 0.2745 0.0011 0.0135 0.0000 0.0028 0.0000 MG 0.0000 0.0000 0.0001 0.0001 0.0011 1.0000 0.0198 1.0000 0.0196 HG 0.0008 0.0005 0.0009 0.0014 0.0135 1.0000 0.0103 1.0000 0.0120 NG 0.0000 0.0000 0.0000 0.0000 0.0000 0.0198 0.0103 0.0082 1.0000 NK 0.0001 0.0001 0.0002 0.0005 0.0028 1.0000 1.0000 0.0082 0.0081 ND 0.0000 0.0000 0.0000 0.0000 0.0000 0.0196 0.0120 1.0000 0.0081 II-9 MR MS MP MC MB MG HG NG NK ND MR 1.0000 0.0058 1.0000 1.0000 0.0012 0.0371 0.0001 0.0000 0.0000 MS 1.0000 0.0129 1.0000 1.0000 0.0044 0.0639 0.0006 0.0000 0.0000 MP 0.0058 0.0129 0.0389 0.0240 1.0000 0.6788 0.4818 0.0678 0.0052 MC 1.0000 1.0000 0.0389 1.0000 0.0120 0.0936 0.0015 0.0000 0.0000 MB 1.0000 1.0000 0.0240 1.0000 0.0083 0.1144 0.0014 0.0000 0.0000 MG 0.0012 0.0044 1.0000 0.0120 0.0083 0.4509 0.7527 0.1061 0.0061 HG 0.0371 0.0639 0.6788 0.0936 0.1144 0.4509 0.2589 0.0210 0.0011 NG 0.0001 0.0006 0.4818 0.0015 0.0014 0.7527 0.2589 0.3210 0.0308 NK 0.0000 0.0000 0.0678 0.0000 0.0000 0.1061 0.0210 0.3210 0.4084 ND 0.0000 0.0000 0.0052 0.0000 0.0000 0.0061 0.0011 0.0308 0.4084 III-6 MR MS MP MC MB MG HG NG NK ND MR 0.3684 0.0000 0.0000 0.0000 0.0313 0.0031 0.0313 0.4873 0.4878 MS 0.3684 0.0033 0.0062 0.0012 0.3141 0.1086 0.3123 0.0267 0.0290 MP 0.0000 0.0033 1.0000 1.0000 0.0322 0.3566 0.0321 0.0000 0.0000 MC 0.0000 0.0062 1.0000 1.0000 0.0622 0.3707 0.0624 0.0000 0.0000 MB 0.0000 0.0012 1.0000 1.0000 0.0192 0.2100 0.0188 0.0000 0.0000 MG 0.0313 0.3141 0.0322 0.0622 0.0192 0.4576 1.0000 0.0012 0.0010 HG 0.0031 0.1086 0.3566 0.3707 0.2100 0.4576 0.4567 0.0001 0.0001 NG 0.0313 0.3123 0.0321 0.0624 0.0188 1.0000 0.4567 0.0013 0.0013 NK 0.4873 0.0267 0.0000 0.0000 0.0000 0.0012 0.0001 0.0013 1.0000 ND 0.4878 0.0290 0.0000 0.0000 0.0000 0.0010 0.0001 0.0013 1.0000
Table 2.6. P-values for pair-wise population comparisons of band frequencies as determined by a modified Fisher’s exact test, for four ISSR loci: I-7, II-4, II-9, and III-6. Significant values (α = 0.05) are shown in bold. Those values still considered significant with a Bonferroni adjusted alpha level (α = 0.05/45 = 0.0011) are underlined. All standard errors were less than 0.0015, the average S.E. was 0.0004.
31 Trait 1 2 3
Floral Orientation -0.950 -0.331 0.293
Throat Indumentum 0.924 -0.083 0.102
Floral Symmetry 0.887 -0.087 0.347
Floral Node Length -0.840 0.052 0.177
Corolla Tube Length 0.821 0.281 -0.122
Petal Width 0.675 0.351 -0.145
Inflorescence Length -0.644 0.655 0.248
Plant Height -0.574 0.720 -0.415
Leaf Length 0.321 0.778 0.447
Table 2.7. Character loadings for the first three PCA axes.
32 Figure 2.1. Flowers of (A) Z. microsiphon, (B) a putative hybrid, and (C) Z. natalensis. A long-proboscid fly (Nemestrinidae) and hawkmoth (Sphingidae) are shown pollinating Z. microsiphon and Z. natalensis, respectively (photographs of pollinators and Z. natalensis provided by S. D. Johnson). 33 Figure 2.2. Map of the Kwazulu-Natal province of South Africa showing the location of each sampled population. Population abbreviations are as in Table 2.2.
34 Figure 2.3. Neighbor-joining dendrogram inferred for 170 individuals with 52 ISSR loci, using Nei and Li’s (1979) similarity coefficient. The majority of individuals within each smaller, circled group are from one or two populations as labeled; exceptions are individually labeled (e.g. “G” within the Z. microsiphon group indicates an individual from the MG population). The single individual of Z. natalensis that was placed within the Z. microsiphon group is enclosed in a square. Putative hybrids are indicated by an “H” with a gray branch.
35 Figure 2.4. Population band frequencies for four ISSR loci: I-7, II-4, II-9, and III-6. Populations with statistically indistinguishable (α = 0.05) frequencies are labeled with the same letter. Column color indicates populations of Z. microsiphon (black), putative hybrids (gray), or Z. natalensis (white).
36 Figure 2.5. Plot of the first two PCA axes for the morphological traits listed in Table 2.3. Only the first PCA axis clearly describes significant variation as determined by the broken-stick model. Individuals of Z. microsiphon, Z. natalensis, and putative hybrids are indicated by squares, circles, and triangles, respectively.
37 CHAPTER 3
PHYLOGENETIC RELATIONSHIPS WITHIN ZALUZIANSKYA AND THEIR
IMPACT ON CLASSIFICATION, BASED ON SEQUENCES FROM MULTIPLE
GENOMES.
INTRODUCTION
Zaluzianskya F. W. Schmidt (Scrophulariaceae s.s., tribe Manuleeae) is a genus of annual and perennial herbs endemic to southern Africa and primarily distributed within
South Africa. This botanically rich country contains plant diversity comparable to some tropical regions (Goldblatt and Manning 2002). It encompasses a wide variety of habitats and includes, despite its relatively small size, what has been considered one of only six floristic regions in the world (Takhtajan 1986, Good 1974). Although research is being done on some of the interesting plant groups in this region (e.g., Christian 2001, Bell and
Ojeda 1999, Johnson et al. 1998, Linder and Mann 1998), the sheer diversity in the country has left many taxa largely unexplored (G. Smith et al. 1996). An example of this is Zaluzianskya, which is only beginning to be examined in a phylogenetic and ecological context (Johnson et al. 2002). This genus comprises taxa with a high degree of diversity in addition to being widely distributed throughout all major climatic regions of South
38 Africa. Species of Zaluzianskya occur in a range of habitats from coastal dunes to
meadows in the highest mountains of South Africa. The diversity within this genus
includes variation in life cycle (annuals and perennials), vegetative form (including
sprawling mats, laxly branching herbs, simple herbs with 10 to 600 mm spikes, and
shrublets; Fig. 1.1), and leaf morphology (with variation in shape, succulence, margins,
and indumentum). However, floral variation in Zaluzianskya is especially prominent
(Fig. 3.1, Table 3.1). While all Zaluzianskya flowers have a fused, narrow corolla tube
and five petal lobes, they vary greatly in characteristics such as tube length, floral
coloration patterns, lobe shape, and indumentum, among others. Additional characteristic
features of Zaluzianskya include inflorescences that are usually terminal spikes, bracts
adnate to a strongly ribbed and plicate calyx, stamen filaments that are decurrent to the
base of the corolla tube (forming a channel surrounding the style), ligulate stigma with
marginal papillae, and small pallid or gray colliculate seeds (Hilliard 1994).
Although several species of Glumicalyx were originally placed within
Zaluzianskya, Hilliard (1994) suggested in her detailed monograph of the tribe
Manuleeae that these two genera are not particularly closely related. Instead, she
proposed that a close relationship exists between Reyemia and Zaluzianskya. These conclusions were later supported by phylogenetic analyses of ndhF and trnL for tribes
Manuleeae and Selagineae (Kornhall et al. 2001), which placed Reyemia and
Zaluzianskya together in a clade. In fact, the ndhF data set resolved Reyemia within
Zaluzianskya. This data set included four species of Zaluzianskya and one of Reyemia while the combined data set included only one species from each genus (see discussion).
39 Hilliard (1994) recognized 55 species of Zaluzianskya placed within four sections
and two subsections: sections Nycterinia, Macrocalyx, Holomeria, and Zaluzianskya subsections Zaluzianskya and Noctiflora (Fig. 3.1, Table 3.1). Relationships within and between the sections of Zaluzianskya are largely unknown, and the monophyly of the sections themselves has not yet been tested. Johnson et al. (2002) produced a morphological phylogeny for section Nycterinia, but thus far no broad phylogenetic analyses have been conducted across the genus. This study had three main goals: (1) to produce a molecular phylogeny for Zaluzianskya, and use this phylogeny to (2) test the monophyly of Zaluzianskya and its sections and (3) obtain a preliminary look at the taxonomic status of species within this genus. This phylogeny also sets the scene for subsequent evolutionary studies such as studies of floral evolution and biogeography (see
Chapter 4).
METHODS
Taxon collection and sequencing
Fieldwork was conducted to collect as many species of Zaluzianskya as possible.
Twenty-three of the 55 species of Zaluzianskya have been sampled for this survey of the genus, including representatives from each infrageneric taxon. Multiple populations of individual species were collected when possible, resulting in a total of 64 populations.
Initial analyses included all populations; redundant populations were then removed, leaving 28 in the final analyses (Table 3.2). In addition, eleven outgroup species were included to root the phylogeny and test the monophyly of the genus. These outgroups
40 were chosen based on their proximity to Zaluzianskya in broader phylogenetic analyses
within Scrophulariaceae (Wolfe et al. unpub., Kornhall et al. 2001), or due to their
placement in the same tribe (Hilliard 1994). Leaf material (~20 mg) was collected on
silica gel from each population. Voucher specimens were deposited in the Ohio State
University Herbarium (OS). DNA was extracted from leaf samples using standard
CTAB methods (Doyle and Doyle 1987).
Both the nuclear and chloroplast genomes were used as sources of sequence data
for reconstructing a phylogeny of Zaluzianskya. The internal transcribed spacer regions
(ITS-1 and ITS-2) and 5.8S ribosomal DNA were used from the nuclear genome.
Numerous studies have documented the utility of ITS for resolving relationships between
closely related species (reviewed by Baldwin et al. 1995, Soltis and Soltis 1998). The
rpl16 intron and the trnL intron plus the trnL-trnF spacer region (hereafter referred to together as the trnL-F region) were sequenced from the chloroplast genome. The general
effectiveness of these regions is reviewed by Soltis and Soltis (1998). Furthermore, the
utility of combining cpDNA regions for resolving lower level phylogenetic relationships
has been documented by many studies (e.g., Haufler et al. 2003, Mort et al. 2002).
Standard primers were employed for PCR: N-nc18S10 and C26A for ITS (Wen
and Zimmer 1996), L16 exon1 and L16 exon2 for rpl16 (Downie et al. 2000), and
B49317 (“c”) and A50272 (“f”) for trnL-F (Taberlet et al. 1991). PCR reactions (50 µl)
consisted of 1 µl DNA, 1X buffer (0.02 M Tris, 0.05 M KCl, and 0.001% Tween-20), 1.5
or 2.5 mM MgCl2 (for ITS or chloroplast reactions, respectively), 5% DMSO (only in
ITS reactions), 0.2 mM dNTPs, 0.24 or 0.64 µM of each primer, and 2.5 or 5.0 U Taq
DNA polymerase (Invitrogen). A GeneAmp System 9700 (Applied Biosystems)
41 thermocycler was used with the following settings for ITS reactions: 2 min at 94˚C; 35 cycles of 1 min at 94˚C, 1 min at 50˚C, and 2 min at 72˚C; 10 min at 72˚C. Chloroplast reactions used the same settings except that the start at 94˚C lasted for 10 min rather than
2 min. Each thermocycler run included a negative control reaction of 50 µl with all of the reagents except for the DNA. PCR products were purified using a PEG clean-up protocol (Soltis and Soltis 1997). Cycle sequencing reactions (5 µl) used the amplification primers noted above and consisted of 0.5 µl PCR amplification product,
2.5% DMSO, 2 µl Big Dye Terminator (version 2.0, ABI Prism), and 0.2 µM primer.
Thermocycler settings incorporated 2 min at 96˚C; 25 cycles of 30 s at 96˚C, 15 s at
50˚C, and 4 min at 60˚C; 6 min at 60˚C. Sequencing products were cleaned via ethanol precipitation and visualized on an ABI 3100 automated sequencer. Individual “contigs” were edited and assembled using Sequencher (GeneCodes Inc.).
Phylogenetic analyses
Sequences were aligned in Clustal X (Thompson et al. 1997) and adjusted by eye in Se-Al (Rambaut 1996). Gaps were coded as characters using the complex indel coding method of Simmons and Ochoterena (2000). Parsimony analyses were conducted in
PAUP* v. 4.0b10 (Swofford 2003) with the heuristic search option, MulTrees, no
MaxTrees limit, swapping on all trees, and parsimony options set to collapse branches if minimum length is zero (“amb-”). Uninformative characters were excluded before all analyses and all remaining characters were equally weighted and unordered. Ten thousand quick searches with random taxon addition and NNI branch swapping were used to locate multiple islands of most-parsimonious trees, forming a starting pool of
42 trees for more thorough searches employing TBR (Maddison 1991). Relative branch support was assessed using bootstrap (Felsenstein 1985) and decay analyses (Bremer
1988). Bootstrap analyses were conducted with 500 replicates, each with 10 random addition replicates and TBR branch swapping (1,000 MaxTrees limit). Decay values were found using AutoDecay (Eriksson 1999) with 5,000 random addition replicates and
NNI branch swapping (no MaxTrees limit).
All chloroplast sequences were combined into one data matrix a priori, and the chloroplast and ITS data sets were analyzed both separately and combined. Topological and data set incongruence were examined to determine if any observed differences in the phylogenies were due to “hard” incongruence rather than “soft.” The latter occurs when topological differences are due simply to weakly supported clades (Seelanan et al. 1997); the supposed disagreement can often be resolved by combining the data and thus providing enough signal to support one clade over the other (e.g., Olmstead and Sweere
1994). Alternatively, “hard” incongruence is found when strongly supported differences occur due to different underlying evolutionary histories.
Topological incongruence was examined by visually comparing the trees from each data set. Only those nodes with at least 70% bootstrap support were considered in this comparison, an arbitrary but reasonable cutoff as discussed by Kellogg et al. (1996) and Mason-Gamer and Kellogg (1996). This provides information on specific problematic taxa or regions within trees. However, topological heterogeneity does not express the underlying levels of character support for conflicting topologies. Several metrics have been used to assess data set heterogeneity, although each appears to have some problems and no single metric has gained complete support from researchers
43 (reviewed in Johnson and Soltis 1998, acronyms used here follow their terminology).
One frequently used metric, IMF, compares the incongruence within both separate data sets to that found within the combined data set (Mickevich and Farris 1981). This metric is also referred to as the ILD (Incongruence Length Difference), and has been used in
several different forms. For this study, IMF was calculated and tested for significance
using HTF (Farris et al. 1995; partition homogeneity test of PAUP* with 1,000 replicates,
TBR branch swapping, 10 replicates of random taxon addition and MaxTrees set to
1,000).
RESULTS
Initially, broad analyses were conducted for each data set, including all 64 collected populations of Zaluzianskya. The monophyletic status of most of the species with multiple sampled populations was either supported or not contradicted by the data.
Thus, duplicate populations for these species were removed in order to simplify the presentation of phylogenies, resulting in final analyses with 28 populations of
Zaluzianskya and 11 outgroup species. The numbers and types of characters for each data set are summarized in Table 3.3, along with the attributes of the resulting most parsimonious trees (MPTs). One of the MPTs for each data set (ITS, chloroplast, and combined) is shown in Fig. 3.2. Analyses of all data sets resulted in three main clades:
(1) a clade consisting of several members of section Zaluzianskya subsect. Zaluzianskya
(Z. villosa, Z. gracilis, Z. affinis, Z. violacea, Z. bella, and Z. minima) in addition to Z. benthamiana (sect. Holomeria; bootstrap = 100, decay = 19 for the combined data set), 44 (2) a clade consisting of the rest of section Zaluzianskya (Z. cohabitans from subsect.
Zaluzianskya and Z. peduncularis from subsect. Noctiflora) and sect. Holomeria (Z.
pusilla and Z. divaricata), along with Reyemia nemesioides (bootstrap = 99, decay = 7),
and (3) a clade containing all sampled members of section Nycterinia (Z. microsiphon, Z.
natalensis, Z. pulvinata, Z. angustifolia, Z. elongata, Z. pachyrrhiza, Z. maritima, Z.
capensis, Z. glareosa, Z. katharinae, Z. ovata) plus the single sampled species from
section Macrocalyx (Z. mirabilis, bootstrap = 100, decay = 17).
Analyses of the ITS data set
The final ITS data set consisted of 677 characters (666 nucleotide characters and
11 indel characters), 185 of which were informative. Parsimony analyses resulted in six
MPTs of 422 steps (Fig. 3.2A, CI = 0.661, RI = 0.837). The MPTs differ in the
relationships among Z. villosa, Z. gracilis, and Z. affinis and within clade 3. In three of the trees, Z. villosa is sister to Z. gracilis (bootstrap < 50). This clade is subtended by a grade of the two Z. affinis populations (Fig. 3.2A). In the other three trees, Z. affinis-2 is
sister to Z. gracilis (bootstrap = 54), subtended by Z. villosa with Z. affinis-1 in a
polytomy with this clade (Z. affinis-2 + Z. gracilis + Z. villosa) and with a clade
comprising Z. bella, Z. minima, and Z. benthamiana. Within clade 3 in all of the MPTs,
Z. microsiphon-3 is sister to a clade containing Z. pulvinata, Z. ovata, Z. glareosa, Z.
maritima, Z. capensis, and Z. angustifolia-2. In two of the trees, Z. angustifolia-1 is also
included within this clade (Fig. 3.2A), in two of the other trees Z. angustifolia-1 is in a
polytomy with this clade and Z. microsiphon-3, and in the final two trees
45 Z. angustifolia-1 is in the basal polytomy within section Nycterinia. Four accessions, Z. microsiphon-1, Z. elongata, Z. pachyrrhiza, and Z. katharinae are placed within this polytomy in all of the MPTs.
Analyses of the chloroplast data set
The chloroplast data set consisted of 2041 characters (2010 nucleotide characters and 31 indel characters), 313 of which were informative. Parsimony analyses resulted in
36 MPTs of 458 steps (Fig. 3.2B, CI = 0.810, RI = 0.919). The MPTs differ in the basal relationships among the outgroups, the placement of Reyemia nemesioides, and the relationships within clade 3. Within clade 2, Reyemia nemesioides is placed in a polytomy with two other clades, Z. cohabitans + Z. peduncularis and Z. pusilla + Z. divaricata, in 24 of the 36 trees, and as sister to those two clades in the other 12 trees
(bootstrap < 50, Fig 3.2B). Clade 3 is divided into two clades, one with Z. microsiphon-
1, Z. microsiphon-2, Z. glareosa, Z. pulvinata, Z. ovata, Z. natalensis, Z. katharinae, Z. elongata, Z. pachyrrhiza, and Z. capensis (bootstrap = 63, decay = 1) and the other with
Z. microsiphon-3, Z. angustifolia, Z. maritima, and Z. mirabilis (bootstrap = 99, decay =
5). Zaluzianskya maritima and Z. mirabilis are alternately placed as sister to the other taxa within the smaller clade in half of the MPTs. The larger clade is resolved in six different arrangements. Several relationships are consistent between all arrangements: the sister relationship between Z. capensis and the other taxa within this clade (bootstrap
= 86, decay = 2), the placement of Z. elongata and Z. pachyrrhiza in a polytomy basal to
46 most or all of the rest of the populations, the sister relationship of Z. microsiphon-1 and
Z. glareosa (bootstrap = 65, decay = 1), and the close relationship of Z. microsiphon-2
and Z. pulvinata.
Analyses of the combined data set
The combined ITS, rpl16, and trnL-F data set consisted of 2718 characters (2676
nucleotide characters and 42 indel characters), 498 of which were informative.
Parsimony analyses resulted in six MPTs of 893 steps (Fig. 3.2C, CI = 0.728, RI =
0.875). As when the ITS data set was analyzed alone, the resolution of Z. villosa, Z.
gracilis, and Z. affinis varied among the MPTs; however, these taxa were supported as a
clade in all MPTs (bootstrap = 86, decay = 3), as found when the chloroplast data was
analyzed alone. In half of the trees, Z. villosa and Z. gracilis formed a clade sister to Z.
affinis-2, whereas Z. affinis-2 and Z. gracilis forming a clade sister to Z. villosa in the other MPTs. The resolution of clade 3 was similar to the chloroplast phylogenies in that the same two main clades were resolved and Z. maritima and Z. mirabilis were each placed as sister to the other taxa of the smaller clade in half of the trees. The larger clade comprised two weakly-supported clades, one with Z. microsiphon-1, Z. microsiphon-2, Z. natalensis, Z. elongata, Z. pachyrrhiza, and Z. katharinae (bootstrap = 52, decay = 1) and the other with Z. pulvinata, Z. ovata, and Z. glareosa (bootstrap <50, decay = 1).
Zaluzianskya capensis was sister to these two clades in four of the trees (bootstrap = 58,
Fig. 3.2C) and in a polytomy with them in the other two trees. The remainder of the relationships shown in Fig. 3.2C were consistent among all MPTs.
47 Congruence testing
Topologically, the ITS and chloroplast data sets appear to be largely congruent.
The only strongly supported discrepancy is the placement of Z. mirabilis (section
Macrocalyx). In the ITS trees, Z. mirabilis is sister to a clade containing all of the sampled species of section Nycterinia (bootstrap = 70, decay = 2). However, analyses of the chloroplast data set place this taxon within section Nycterinia (bootstrap = 99, decay
= 5). In addition to this topological assessment, ILD tests also indicate a significant degree of incongruence (α = 0.05). While P-values for an ILD comparison of the two chloroplast regions verified their congruence (p = 0.540), incongruence was found between the ITS and chloroplast data sets (p = 0.004). Further testing with subsets of taxa found that the two data sets did not significantly conflict in the region of the tree containing taxa from sections Zaluzianskya and Holomeria (clades 1 and 2, p = 0.509), but rather only in the Nycterinia + Z. mirabilis clade (clade 3, p = 0.002). Focusing in on clade 3, when the ILD analyses were repeated with Z. mirabilis pruned, there was not statistically significant incongruence between these data sets (p = 0.058). However, if Z. mirabilis was allowed to remain and instead the populations of Z. microsiphon were pruned, this again resulted in a statistically significant degree of incongruence between the data sets (p = 0.019). It appears that these two data sets are largely congruent with each other with the exception of the placement of Z. mirabilis.
48 DISCUSSION
Zaluzianskya and Reyemia
The monophyly of Zaluzianskya is disrupted by the inclusion of Reyemia nemesioides within the Zaluzianskya clade. This placement is strongly supported by the
ITS, chloroplast, and combined data sets (bootstrap = 99, decay = 7 for the combined data set, Fig. 3.2). Analyses constraining Zaluzianskya to be monophyletic resulted in trees with an additional seven, five, or 12 steps (for the ITS, chloroplast, and combined data sets, respectively, Table 3.4). As mentioned in the introduction, this is not the first phylogenetic study to resolve Reyemia within Zaluzianskya. Kornhall et al. (2001), for
Bayesian and parsimony analyses of ndhF in tribes Selagineae and Manuleeae, sampled four species of Zaluzianskya (Z. katherinae, Z. glareosa, Z. minima, and Z. benthamiana) as well as Reyemia chasmanthiflora. Despite their low taxon sampling within
Zaluzianskya and the inclusion of a different species of Reyemia, they also found that
Reyemia clustered within the Zaluzianskya clade (jackknife = 81, posterior probability =
0.99). Reyemia was described by Hilliard (1992) as a genus of only two species: R. chasmanthiflora and R. nemesioides. The former species was not discovered until 1986, whereas the latter was first described in 1897 and originally named Zaluzianskya nemesioides (Diels 1897). I propose that these two species be returned to the genus
Zaluzianskya due to the molecular data shown here as well as morphological similarities.
Hilliard noted (1994, p. 522) that “the affinity of Reyemia appears to be with
Zaluzianskya: they have similar calyces, which are distinctive in the Manuleae [sic] by virtue of being plicate, and their seeds are similar.” Zaluzianskya and Reyemia are also
49 distinguished from most other members of the tribe in that their “bracts are usually
sharply differentiated from the leaves and are adnate to the calyx almost to its apex.”
(Hilliard 1994, p. 14). Also, within the Manuleeae sensu Hilliard (1994), the type of
zygomorphy found in Reyemia (with a single isolated anticous lobe) is shared only with
some species of Zaluzianskya and Manulea (Hilliard 1994). Additional similarities that set Zaluzianskya and Reyemia apart from some other members of the tribe include stamen
filaments that are decurrent to the base of the corolla tube and a ligulate stigma with
marginal papillae. Although chromosome counts have been completed for nine species
of Zaluzianskya (2n = 12, Jong 1993, Hedberg 1970), Reyemia has not yet been
examined. Hilliard separated Reyemia from Zaluzianskya based on its loose panicles of
flowers (vs. an elongate or capitate spike or very rarely a raceme in Zaluzianskya),
resupinate corolla-limb (unique to Reyemia within Manuleeae), clavate hairs in the back
of the corolla throat (vs. throat always glabrous in Zaluzianskya, although many species
have clavate hairs in a complete or partial circlet around the mouth), and two posticous
(adaxial) stamens plus two small anticous (abaxial) staminodes. Species of Zaluzianskya
have no staminodes except in rare individuals and have either four stamens (46 species),
only the two posticous stamens (seven species), or only the two anticous stamens (two
species).
The corolla of Reyemia is noticeably different from Zaluzianskya. As stated by
Hilliard (1994, p. 522), “The flowers bear no resemblance to those of any species of
Zaluzianskya,” although this is said within the context of the tribe Manuleeae, in which
all flowers share many similarities. The affinity of Reyemia has been uncertain for many
botanists; Hilliard went on to say that “specimens of R. nemesioides may be found in
50 herbaria under Nemesia, Sutera, and Phyllopodium, or nameless at the end of the family.”
The resupination of the corolla in Reyemia results in a flower with a unique appearance
within the tribe. Nevertheless, the differences seen between flowers of Reyemia and
Zaluzianskya could be considered no more striking than those seen between species
within the latter; there is a considerable amount of floral variation within Zaluzianskya.
The specific placement of Reyemia within clade 2 of Zaluzianskya is also supported by
several morphological traits (discussed below, also see Chapter 4). It appears from the
molecular data that the distinctive morphology of Reyemia simply represents a derived
taxon within Zaluzianskya.
Sectional classification within Zaluzianskya
Thus far, multiple taxa from three of the four sections within Zaluzianskya have
been sampled. One or both of the molecular data sets contradict the monophyly of each
of these sections. As stated above, three main clades are resolved by each data set (Fig.
3.2): (1) a clade containing species of section Zaluzianskya subsection Zaluzianskya
along with Z. benthamiana (from section Holomeria) that is sister to the other two clades,
including a clade (2) with the rest of the sampled taxa from sections Zaluzianskya and
Holomeria (in addition to Reyemia nemesioides) sister to a clade (3) with all of the
species from section Nycterinia as well as Z. mirabilis (section Macrocalyx). The lack of
monophyly of sections Zaluzianskya and Holomeria is strongly supported by these data.
The species from these two sections are intermixed among two clades. Also, merging
these sections into one would still not solve the monophyly problem because the
Nycterinia clade would then be derived from within that new section.
51 Clade 1 is kept from being comprised solely of species from section Zaluzianskya
subsection Zaluzianskya by the inclusion of Z. benthamiana. This is a surprising result,
based on floral morphology (Fig. 3.1). The flowers of all of the other species in this
clade are similar, with bifid petal lobes and white to purple petals (often varying within a
species). In contrast, Z. benthamiana has entire, pale yellow petal lobes. This
morphology fits much better with the rest of section Holomeria in clade 2. In fact, all of
the species in clade 2 have entire petal lobes and all (including Z. peduncularis from
section Zaluzianskya subsect. Noctiflora and Reyemia nemesioides) are yellowish, except for the pink interloper from section Zaluzianskya subsect. Zaluzianskya (Z. cohabitans).
However, the placement of Z. benthamiana (clade 1), and of Z. peduncularis + Z. cohabitans + R. nemesioides (clade 2) is strongly supported by the combined data
(bootstraps = 100 and 99, decay = 19 and 7 for clade 1 and 2, respectively). Three to four
separate populations of each of these species were included in the original analyses and
all were placed in the same clades as the populations represented in Fig. 3.2. Analyses
constraining sections Zaluzianskya and Holomeria to be monophyletic resulted in trees
with 36, 26, or 61 additional steps (for the ITS, chloroplast, and combined data sets,
respectively, Table 3.4). A revised infrageneric taxonomy is needed to address these
issues. It is possible that clades 1 and 2 could each be considered a separate section.
Entire petal lobes may be a morphological synapomorphy supporting clade 2, while
spreading acute stem hairs appear to delineate clade 1(with some exceptions, see Chapter
4). When Zaluzianskya was first divided into sections the separation was based mainly
on entire (sect. Holomeria sensu Bentham 1846) versus notched petal lobes (sect.
Zaluzianskya sensu Bentham). Zaluzianskya peduncularis was originally placed in
52 section Holomeria along with the other entire-lobed species known at the time: Z.
divaricata , Z. pusilla, and Z. benthamiana. These phylogenies reaffirm the affinity of Z.
peduncularis with some members of section Holomeria sensu Hilliard. It is necessary to collect additional species of this genus to address more fully these taxonomic questions.
Section Nycterinia is generally cohesive; the only problem arises with the possible addition of Z. mirabilis from section Macrocalyx. This inclusion was
unexpected based on the generally invariable floral morphology of section Nycterinia.
Zaluzianskya mirabilis does not share the typical appearance of all Nycterinia flowers:
bifid petal lobes that are white on their upper surface (sometimes with orange at the base
of the lobes) and dark red on the lower surface (Fig. 3.1). Instead petal lobes of Z.
mirabilis are doubly bifid and pink on the upper surface. However, this species does
share some characteristics with section Nycterinia, such as red coloring on the lower
surface and larger flowers (also see Chapter 4). As discussed above, the placement of
this species is currently uncertain. Section Macrocalyx encompasses five species.
Potentially, additional sampling of these taxa will help to resolve the position of Z.
mirabilis. The chloroplast and combined data sets support the inclusion of this species
within the Nycterinia clade (bootstrap = 99 and 75, decay = 5 and 1, respectively)
whereas the ITS data place Z. mirabilis sister to a monophyletic Nycterinia (bootstrap =
70, decay = 2). The reason for this incongruence is unknown, but ILD tests indicate that
it is due to true conflict between the data sets. However, some researchers have
concluded that ILD tests may return a false significant result more frequently than
indicated by an alpha level of 0.05 (e.g., Darlu and Lecointre 2002, Barker and Lutzoni
2002, Siddall 1997), thus it is also possible, for example, that the perceived incongruence
53 is simply the result of poor ITS signal in that region of the phylogeny (Wendel and Doyle
1998, Sullivan 1996). Differences in evolutionary rates between data sets may be one
factor that can lead to significant ILD values despite congruent evolutionary histories, but
the relative evolutionary rates of these two data sets appear to be similar (as indicated by
average sequence divergence values, Table 3.3).
If there is true conflict between the nuclear and chloroplast data sets, there are
several possible causes including hybridization, horizontal transfer, gene duplication, and
lineage sorting, all of which can leave very similar patterns in molecular phylogenies
(reviewed by Wendel and Doyle 1998). Several workers have provided insight into some
of the distinguishing characteristics for these different sources of incongruence (e.g.,
Holder et al. 2001, Sang and Zhong 2000, Andersson 1999, Wendel and Doyle 1998,
Maddison 1997). However, these processes are very difficult to distinguish a posteriori
and a quantitative test to do so using molecular data has not yet been firmly established.
Instead, researchers have often used other biological information for their taxa to explore
the probability of different processes (e.g., Kimball et al. 2003, Rokas et al. 2003, Sota et
al. 2001, van Oppen et al. 2001). For example, it is possible that chloroplast capture has
occurred in Zaluzianskya, altering the chloroplast genome of the sampled individuals of
Z. mirabilis. However, hybridization between a species within section Nycterinia and Z.
mirabilis seems unlikely. Zaluzianskya mirabilis is endemic to western areas of South
Africa, where few members of section Nycterinia occur. Only two species from that
section both occur in the same general region of the country as Z. mirabilis and also
might overlap in flowering season: Z. capensis and Z. ovata. The flowering season of the latter (October – March) might only occasionally overlap with that of Z. mirabilis
54 (August – September). Neither of these two species group with Z. mirabilis in any of the
phylogenies. The range of Z. mirabilis could be larger than currently known, because it
has been poorly collected; it is possible that this species overlaps with additional species
within section Nycterinia. However, another potential isolating factor between the western species of this section and Z. mirabilis is that they are all night flowering whereas Z. mirabilis is day flowering (although there is a precedent in this genus for
potential hybridization between day- and night-flowering species, see Chapter 2). In light
of these factors, hybridization seems an unlikely source of incongruence in this case.
Horizontal transfer is considered to be a relatively rare occurrence in plants
(Wendel and Doyle 1998). There is no apparent reason to invoke it here, partially
because such a gene transfer would again face the geographic disjunction between Z.
mirabilis and most of section Nycterinia (including all species of that section that formed
a clade with Z. mirabilis). Gene duplication also seems unlikely; its effects are generally
seen at higher taxonomic levels. Lineage sorting, on the other hand, is more likely to be
seen near the species level (Maddison 1997). The region of the phylogeny in question,
clade 3, appears to be quite recently derived, based on the low resolution in the molecular
phylogenies as well as a low level of morphological divergence within this clade.
Lineage sorting could be a possible explanation for the incongruence seen here (if indeed
it is hard incongruence), but further work including the addition of other unlinked DNA
sequences and increased population sampling is necessary to more vigorously test this
hypothesis.
55 Species-level taxonomy
The status of several species has also been called into question by analyses of
these data. These questions cannot be answered simply with a phylogeny, but rather will
require detailed study of each taxon. However, the phylogeny can point out possible
problematic species, raising questions as well as allowing for preliminary hypotheses.
For example, analyses of all of the data sets show that populations of Z. pusilla are not
monophyletic, because the sole collected population of Z. divaricata is derived from within them (bootstrap = 100, decay = 6 for the combined data set). Although the requirement of monophyly for higher-level taxa has become widely accepted in the systematics community, there are varying opinions as to whether or not species must be monophyletic and whether or not the term is even applicable at this level (e.g., Luckow
1995). If, in this case, Z. pusilla populations form a morphologically and ecologically cohesive group, it might be considered that they should still be considered a species despite their lack of monophyly (e.g., Rieseberg and Brouillet 1994, Wheeler and Nixon
1990, but see de Queiroz and Donoghue 1988, Donoghue 1985). This could be defended by the notion that the alteration of one group within this clade into “Z. divaricata” does not change the attributes or relationships of the remaining populations of Z. pusilla.
However, Z. pusilla is a variable species (for example, it has a “remarkable range in the size of the corolla limb” Hilliard [1994], p. 518), and it may be more appropriate to divide this species into multiple morphologically cohesive species rather than to lump it with Z. divaricata or leave the taxonomy as it is. Determining the most appropriate delimitation of these species will require further study of populations across their ranges.
56 The molecular data also give cause to doubt the status of Z. villosa, Z. affinis, and
Z. gracilis. The combined data show populations of both Z. villosa and Z. gracilis as
derived from within a clade of Z. affinis (bootstrap = 87, decay = 3). The chloroplast data
place these populations together in an unresolved clade whereas the ITS data separate Z.
affinis-1 as sister to the others (bootstrap = 81, decay = 2). Although Zaluzianskya
contains a diversity of characteristics as a whole, many individual groups of species are
quite similar. This is true of these three species. Zaluzianskya villosa and Z. affinis are nearly indistinguishable, differing only slightly in leaf and bract shape and indumentum.
In fact, Z. affinis was so named due to its close affinity with Z. villosa. Hilliard (1994) also noted that the distinction between these two species is blurred where their distributions overlap. She hypothesized that this could be due either to hybridization or possibly to “incomplete speciation.” Zaluzianskya affinis occurs in northwestern South
Africa while Z. villosa occurs in the southwest. The samples of these species included in this study were located far from the area where their distributions meet. Each of the two populations of Z. affinis were located at least 120 km north of this region, and Z. villosa was collected near the southeastern edge of its range, over 160 km from Hilliard’s (1994) described range for Z. affinis. If in fact introgression or “incomplete speciation” is responsible for the proximity of these populations in the phylogeny, the genetic effects are spread throughout their ranges, raising the question of whether they should simply be considered one species. The third taxon in this clade, Z. gracilis is also morphologically similar to the other two. The only difference between Z. gracilis and Z. villosa cited by
Hilliard (1994) is that the former species has significantly smaller flowers. Zaluzianskya gracilis appears to be confined to calcareous sands covering a relatively small area east of
57 the range of Z. villosa. Again, the population of this species considered here was not located at the edge of its range or adjacent to the described range of Z. villosa but rather over 100 km to the east. The molecular data and morphological similarity suggest that the prudence of merging two or three of these taxa should be further investigated.
The last species under consideration is Z. microsiphon. This species is the only day-flowering member of section Nycterinia and is one of the more morphologically distinct members of that section. Many of the 20 species within section Nycterinia are difficult to distinguish, but Z. microsiphon differs in several floral features including floral symmetry, petal orientation, indumentum, and scent. Despite these putative apomorphies, separate and combined analyses of the molecular data divide the populations of Z. microsiphon into at least two groups. The chloroplast and combined data supported a basal split in section Nycterinia. Two populations of this species, Z. microsiphon-1 and another not included in the final analyses, grouped with Z. angustifolia, Z. maritima, and Z. mirabilis (chloroplast: bootstrap = 99, decay = 5; combined: bootstrap = 75, decay = 1), while the other populations of Z. microsiphon fell in a clade with the rest of section Nycterinia (chloroplast: bootstrap = 63, decay = 1; combined: bootstrap = 52, decay = 1). Zaluzianskya microsiphon-3 was not only separated from the other two populations of Z. microsiphon by these data, but was also derived from within two populations of Z. angustifolia, albeit with moderate to weak support (chloroplast: bootstrap = 69, decay = 1; combined: bootstrap = 58, decay = 1).
The ITS and combined data supported a clade of Z. microsiphon-2 + Z. natalensis, separating this population from the other two (bootstrap = 79 and 73, decay = 2 and 2).
The chloroplast data weakly support the sister relationship of Z. microsiphon-1 and Z.
58 glareosa (bootstrap = 65, decay = 1). The position of Z. microsiphon-2 is not strongly
resolved by the chloroplast data, but it appears to be closely related to Z. pulvinata and Z.
microsiphon-1. Although the two data sets show some differences in the placement of the Z. microsiphon populations, these conflicts are generally not strongly supported (as corroborated by the insignificant ILD P-value reported above).
There are two main possible explanations for the division of this species. One is
that the morphological unit that has so far been labeled “Z. microsiphon” is actually
several lineages that have converged on this distinctive floral morphology, thus the
molecular phylogeny is correct in dividing these populations. Another possibility is that
hybridization has muddled the genomes of these populations, resulting in their division in
the phylogeny despite the single origin of Z. microsiphon. It is not currently possible to distinguish between these hypotheses. The fact that populations of Z. microsiphon share a series of distinct morphological characteristics makes it seem unlikely that they have all arisen independently. However, the traits that separate Z. microsiphon from its cohorts are all floral. Johnson et al. (2002) determined that this species is pollinated primarily by long-proboscid flies (Nemestrinidae) rather than the hawkmoths (Sphingidae) that pollinate the other Nycterinia species examined. The day-flowering, zygomorphic, scentless flowers of Z. microsiphon are concordant with that pollination syndrome
(Goldblatt and Manning 2000). It is well established that pollinator selection can lead to convergent evolution of floral traits (e.g., Faegri and Pijl 1979); perhaps this suite of characteristics has in fact evolved convergently within section Nycterinia due to selection
from long-proboscid flies.
59 The alternate hypothesis that hybridization influenced the molecular phylogeny is
also feasible. Previous studies have shown the potential for hybridization and
introgression between Z. microsiphon and Z. natalensis (see Chapter 2). Zaluzianskya
microsiphon is also sympatric with several other members of section Nycterinia that
share a floral syndrome with Z. natalensis, raising the question of potential hybridization with those species as well. Crossing studies are needed to determine if Z. microsiphon is
able to hybridize with these species. Many questions remain with Z. microsiphon and
further field studies are underway to investigate its attributes across its range.
Conclusions
Although additional work remains to be done on classification within
Zaluzianskya, the molecular phylogenies brought several issues to light. First, Reyemia
was derived from within Zaluzianskya and thus should be merged with that genus.
Second, the current sectional division of Zaluzianskya appears to be unnatural and should
be reexamined. In particular, sections Zaluzianskya and Holomeria are clearly not
monophyletic, although the status of section Nycterinia is uncertain and the monophyly
of section Macrocalyx has not yet been tested. Furthermore, the placement of
populations of several species (Z. affinis, Z. villosa, Z. gracilis, Z. pusilla, Z. divaricata,
and Z. microsiphon) also raised questions regarding their status. These species should be
reexamined with detailed field studies.
60 Nycterinia Macrocalyx Holomeria Z. subsect. Z. subsect. Zaluzianskya Noctiflora No. species 20 (11) 5 (1) 10 (3) 16 (7) 4 (1) (sampled)
HabitPerennial or Annual Annual Annual Annual Annual
Petal lobes Notched Notched Entire Variable Entire
Corolla: Red Pink to Brown to White to Yellow to color outside magenta maroon to yellow to red purplish red brown
Corolla: White White to Cream to White to Cream to color inside pink or yellow, purple, yellow yellow orange, or yellow, or red orange
Corolla: None or None or None or Various Various throat rarely yellow to yellow to colors of star colors of patterns orange ring red star red bar on or dots star, ring, or lobes or star dots
Stamens 4 4 2 anticous or 2 posticous 4 4 or 4
Anthesis Nocturnal Diurnal Diurnal Diurnal Nocturnal (19) or diurnal (1)
Nectariferous Free from Free from Adnate to Free (if 2 Adnate to gland ovary ovary ovary stamens) or ovary adnate (if 4)
Table 3.1. Several characteristics of the sections and subsections of Zaluzianskya.
61 Section/Subsection Species Author Collection Number of Number Populations Sequenced Nycterinia Z. microsiphon-1 (O. Kuntze) K. SJ 19ii00 s. n. 1 9 Z. microsiphon-2 Schum. JKA 98.a Z. microsiphon-3 SJ 30xii99 s. n. Z. natalensis Krauss JKA 114.a 3 Z. pulvinata Killick A&W 5.a 3 Z. angustifolia-1 Hilliard & Burtt JKA 119 2 Z. angustifolia-2 CP xi00 s. n. Z. elongata Hilliard & Burtt A&W 8.a 3 Z. pachyrrhiza Hilliard & Burtt SJ 01maz s. n. 1 Z. maritima (L. f.) Walp. JKA 8.1 2 Z. capensis (L.) Walp. JKA 2.1 3 Z. glareosa Hilliard & Burtt JKA 90.a 3 Z. katharinae Hiern ADW 729 1 Z. ovata (Benth.) Walp. JKA 111.a 2 Macrocalyx Z. mirabilis Hilliard JKA 68 2 Zaluzianskya subsect. Z. villosa F. W. SchmidtJKA 1.1 1 Zaluzianskya Z. affinis-1 Hilliard JKA 9.1 4 Z. affinis-2 JKA 29.1 Z. gracilis Hilliard JKA 6.2.a 1 Z. bella Hilliard JKA 44.a 2 Z. violacea Schltr. JKA 18.1 4 Z. minima (Hiern) Hilliard JKA 58.a 3 Z. cohabitans Hilliard JKA 46.1 4 subsect. Noctiflora Z. peduncularis (Benth.) Walp. JKA 45.a 3 Holomeria Z. divaricata (Thunb.) Walp. JKA 31.a 1 Z. pusilla-1 (Benth.) Walp. JKA 76.2 4 Z. pusilla-2 JKA 26.1.1 Z. benthamiana Walp. JKA 12.1 3 Tribe Manuleeae Reyemia nemesioides (Diels) Hilliard ADW 986 1 Glumicalyx flanaganii (Hiern) Hilliard T. Edwards 1 & Burtt 1774 Glumicalyx goseloides (Diels) Hilliard Hilliard & Burtt 1 & Burtt 4ii98 Polycarena aurea Benth. ADW 979 1 Manulea altissima L. f. ADW 708 1 Manulea rubra (Berg.) L. f. ADW 718 1 Sutera hispida (Thunb.) Druce ADW 930 1 Lyperia tristis (L. f.) Benth. ADW 879 1 Jamesbrittenia adpressa (Dinter) Shijman 1851 1 Hilliard Tribe Aptosimeae Aptosimum depressum Benth. ADW 685 1 Aptosimum indivisum Benth. ADW 910 1
Table 3.2. The 28 populations of Zaluzianskya and 11 outgroup taxa included in the final phylogenetic analyses.
62 Dataset ITS Chloroplast (rpl16, trnL-trnF) Combined
No. nucleotide characters 666 2010 (958, 1052) 2676
No. indel characters 11 31 (10, 21) 42
Total no. characters 677 2041 (968, 1073) 2718
Total no. informative 185 313 (155, 158) 498 characters
Ave. sequence divergence 0.2698 0.2275 0.2458 in all taxa
Ave. sequence divergence 0.1314 0.1019 0.1139 in Zaluzianskya
Ave. sequence divergence 0.02724 0.03051 0.02920 in Clade 3
CI of MPTs 0.661 0.810 0.728
RI of MPTs 0.837 0.919 0.875
RC of MPTs 0.553 0.744 0.637
Number of MPTs 6 36 6
Length of MPTs 422 458 893
Table 3.3. The number and types of characters in each data set as well as characteristics of the resulting most parsimonious trees (MPTs) for the final analyses including 39 populations.
63 Zaluzianskya Section Nycterinia Sections Zaluzianskya & Holomeria ITS 7 (1.7%) 0 (0.0%) 36 (8.5%)
Chloroplast5 (1.1%) 5 (1.1%) 26 (5.7%)
Combined 12 (1.3%) 1 (0.1%) 61 (6.8%)
Z. microsiphon Z. pusilla Z. affinis
ITS 4 (1.0%) 1 (0.2%) 4 (1.0%)
Chloroplast16 (3.5%) 5 (1.1%) 0 (0.0%)
Combined 17 (1.9%) 6 (0.7%) 4 (0.5%)
Table 3.4. Number of additional steps found for the most parsimonious trees when the designated groups are constrained to be monophyletic for each data set. Percentage of total number of steps given in parentheses.
64 Figure 3.1. Representatives from each infrageneric taxon in Zaluzianskya: (A) section Nycterinia, Z. pulvinata (left) and Z. capensis (right); (B) sect. Macrocalyx, Z. mirabilis; (C) sect. Holomeria, Z. divaricata (top) and Z. benthamiana (bottom); (D) sect. Zaluzianskya subsect. Zaluzianskya, Z. violacea (top left), Z. affinis (top right), and Z. gracilis (bottom); and (E) subsect. Noctiflora, Z. peduncularis. 65 Figure 3.2. One of the MPTs for analyses of (A) ITS, (B) chloroplast, and (C) combined data sets with 39 accessions. Bootstrap support values greater than 50% are indicated above the branches followed by decay values; branch-lengths are provided below. Asterisks mark those nodes that are not recovered in the strict consensus. Major clades are labeled on the combined phylogeny. 66 CHAPTER 4
CHARACTER EVOLUTION WITHIN ZALUZIANSKYA: IMPLICATIONS FOR
CLASSIFICATION, FLORAL EVOLUTION, AND BIOGEOGRAPHY.
INTRODUCTION
One of the problems faced when differentiating groups of organisms in basic taxonomy is determining which traits delimit natural groups and which are changeable or homoplastic. Although detailed morphological examination can sometimes answer this question, particularly if aided by developmental studies (e.g., Tucker 1996), phylogenetic analyses have proven extremely valuable for using the congruent signal of many characters to determine the homology of single characters (e.g., McDade 1992). Even the most careful morphological examination can potentially lead to incorrect conclusions of homology, for example, if a simple trait were to evolve twice in parallel, each with the same developmental pathway. Molecular phylogenies offer the additional benefit of independence from the morphological data, bypassing the potential problem of convergent evolution of suites of characters (Givnish and Sytsma 1997). However,
67 molecular data raise many of their own complications, for example, those arising from gene duplication, lineage sorting, hybridization, and polyploidy (Wendel and Doyle
1998).
In addition to determining which characters are useful for classification, investigations of character evolution using a phylogenetic framework also allow insight into the patterns of morphological evolution and diversification within a group. This has provided access to information regarding the number and timing of character transitions and the direction of evolution for particular traits (e.g., Weller and Sakai 1999), allowing insight into such things as the original evolution of the flower (Endress 2001, Soltis et al.
2000) and the evolution of pollinator interactions with changes in floral morphology
(e.g., Johnson et al. 1998, McDade 1992), among many other subjects. In this study, the evolution of a number of morphological traits within the genus Zaluzianskya F. W.
Schmidt (Scrophulariaceae s.s., tribe Manuleeae) was investigated to identify potential synapomorphies for infrageneric groups, to examine the evolution of floral morphology within this diverse genus, and also to explore potential correlations between life-history traits and biogeography.
Zaluzianskya, a genus of 55 species of annual and perennial herbs or shrublets, is endemic to southern Africa but distributed mainly within South Africa, (Fig 4.1, Hilliard
1994). This botanically distinctive area contains plant diversity comparable to many tropical regions (Goldblatt and Manning 2002, Gibbs Russell 1987). As researchers attempt to explore and document this diversity, the current taxonomy of several groups has been found to be problematic (e.g., Mort et al. 2003, Kornhall et al. 2001, Linder
1996), including that of Zaluzianskya. Recent phylogenetic analyses of this genus (using
68 sequences of ITS, rpl16, and trnL-trnF) demonstrated that it was not monophyletic, due to strong support for the inclusion of one of the purported outgroups: Reyemia nemesioides (Diels) Hilliard (see Chapter 3). Reyemia, a genus of two species (R. nemesioides and R. chasmanthiflora Hilliard), has always been considered to be very closely related to Zaluzianskya (Hilliard 1994, Diels 1897). Based on the molecular phylogenies as well as morphological characteristics, Reyemia will be officially merged into Zaluzianskya within the article manuscripts published from Chapter 3. For simplicity, species of “Reyemia” will be referred to as species Zaluzianskya throughout this chapter.
The molecular data also revealed problems with the infrageneric classification for
Zaluzianskya. Hilliard (1994) divided this genus into four sections and two subsections:
Nycterinia, Macrocalyx, Holomeria, and Zaluzianskya subsections Zaluzianskya and
Noctiflora. This classification was based primarily on detailed morphological examinations of herbarium specimens. Our analyses have shown that at least two of these sections are clearly not monophyletic (sections Holomeria and Zaluzianskya), while the monophyly of sections Nycterinia and Macrocalyx is ambiguous and untested, respectively (Fig. 4.2, see Chapter 3). These analyses included one of the five species in section Macrocalyx, Z. mirabilis, and the data placed it either within section Nycterinia
(according to analyses of the chloroplast data and the combined chloroplast and nuclear data) or sister to a monophyletic Nycterinia clade (according to separate analyses of ITS).
This conflict between data sets was significant according to results of ILD tests and comparisons of robustly supported nodes (i.e. with at least 70% bootstrap), although the remainder of the nuclear and chloroplast data sets appear to be congruent. The reason for
69 the incongruence involving Z. mirabilis is currently unknown; several possibilities are
discussed in Chapter 3. Until further data are available, the implications of both potential
placements for this taxon will be considered.
Several species of Zaluzianskya were also not resolved as monophyletic, the most
drastic example being Z. microsiphon whose populations were spread between two or
three areas within clade 3 in the phylogeny. This implies that a “Z. microsiphon-like”
morphology has evolved convergently, possibly as a result of pollinator selection.
Pollination studies have determined that Z. microsiphon, the single day-flowering species
within section Nycterinia, is pollinated by long-proboscid flies (Nemestrinidae) whereas
the night-flowering species within this section appear to be pollinated by hawkmoths
(Sphingidae, Johnson et al. 2002, Goldblatt and Manning 2000, McGregor 1989). The
main morphological differences between Z. microsiphon and most of its night-flowering
counterparts appear to correspond to a long-proboscid fly pollination syndrome,
including its zygomorphic flowers with a vertical corolla limb and lack of scent
(Goldblatt and Manning 2000). Several other characteristics of Z. microsiphon also
appear to be adaptive for long-proboscid fly pollination, such as an elongate, cylindrical
floral tube and a nectar reward, but these traits are also considered likely in hawkmoth-
pollinated plants and occur throughout the section. Potentially these preadaptations for
long-proboscid fly pollination eased the way for multiple transitions to this pollination
system (with subsequent selection for other relevant features). However, hybridization
between Z. microsiphon and Z. natalensis (and potentially other members of section
Nycterinia) is also likely (see Chapter 2, Johnson et al. 2002), raising a potential alternative explanation for the phylogenetic pattern. Hybridization could have resulted in
70 genome contamination within Z. microsiphon, artificially dividing the genus into multiple lineages (see Chapter 3). Whether convergent evolution or hybridization (or some combination of the two) is responsible for the phylogenetic pattern seen for Z. microsiphon is currently unknown, thus both hypotheses will be discussed in relation to the evolution of this distinctive species. Several other species, such as Z. pusilla and Z. affinis, were also not resolved as monophyletic due to the derivation of other species from within their clades (Z. divaricata and Z. villosa + Z. gracilis, respectively, discussed in Chapter 3), thus multiple populations for these species will also be considered here.
The molecular phylogeny for this genus clearly demonstrates the need for taxonomic revision. The phylogenetic framework developed in Chapter 3 will be used to explore the relation of morphology to these new hypotheses of relationships, including an investigation of the implications of each potential placement for Z. mirabilis and Z. microsiphon. In particular, the formation of the floral diversity found in this genus will be investigated, along with its relation to both classification and potential pollinator interactions. Although all species of Zaluzianskya share several floral traits such as a narrow corolla tube and five petal lobes, variation occurs in tube length, lobe shape, petal indumentum, color on the upper and lower surfaces, patterning, and androecial characters. Each of these traits is likely to be important in pollinator interactions (Faegri and Pijl 1979), which could have implications in the diversification of the species themselves. The breeding systems of Z. microsiphon, Z. natalensis, Z. glareosa, and Z. spathacea have been explored within this genus, all of which were found to be obligate outcrossers (Johnson et al. 2002, McGregor 1989). Johnson (1996) proposed that ethological isolation and pollinator selection has potentially been one of the main factors
71 responsible for the impressive plant diversity found in South Africa, although certainly other factors such as substrate gradients are also likely to have been important in speciation of groups within this region (Linder 1985). The floral diversity in this genus suggests that it may be one of those taxa that have diversified at least in part due to pollinator interactions.
The evolution of habit (annual vs. perennial) will also be considered, in connection with biogeography. Although both relatively widespread and locally endemic species occur in this genus, Hilliard (1994) noted that the perennial species of
Zaluzianskya occur only in the east whereas annual species occur only in the west. This potential correlation between habit and distribution will be explored via its possible connection to rainfall in these areas. As with the investigations of floral evolution, this will allow a reconstruction of the ancestral character states (habit and region of origin, in this case) as well as potentially reveal the number of migrations between regions within
South Africa.
Thus, the following questions will be explored using a phylogenetic framework for Zaluzianskya: (1) what morphological characters may be suitable for delimiting a revised classification within Zaluzianskya? (2) how has floral evolution progressed within this diverse genus and what implications might this have for pollinator interactions? and (3) is there a relationship in species of Zaluzianskya between the evolution of habit and distribution?
72 METHODS
Each of the questions presented here requires a robust phylogenetic framework.
The phylogenies given in Chapter 3 resolved three main clades with strong support as well as providing good resolution for most of the smaller clades (Fig. 4.2). Although future data collection will aid in resolving some difficult groups such as section
Nycterinia, the current resolution allows for an initial exploration of character evolution.
Concerns have been raised in previous papers about investigating character evolution without complete taxa sampling (e.g., Mc Dade 1992). However, complete taxa sampling is a goal that is rarely reached in natural groups. In many cases, it is nearly impossible due to extinction of some lineages. Twenty-three of the 55 species of
Zaluzianskya, representing all infrageneric groups and much of the diversity within this genus, have been included in this study. This should be sufficient to begin the investigation of character evolution within this group, as long as it is kept in mind that further taxa collection may result in adjustments of my conclusions (cf. Johnson et al.
1998).
Character evolution
Full details on the phylogenetic methods used are given in Chapter 3. The paths of character evolution were traced onto each of the most parsimonious trees from the combined analyses using MacClade (Maddison and Maddison 1992). Alternate resolutions of polytomies and alternate optimizations of character evolution were inspected (via “equivocal cycling”) to find the different possible locations of changes in
73 character states. In addition, the effects, if any, of the alternate placements for Z.
mirabilis (based on the ITS data analyzed alone) and Z. microsiphon were explored. The
outgroups and Z. nemesioides (formerly placed in the ditypic Reyemia along with Z. chasmanthiflora, which was not sampled in this survey) were coded as genera rather than as species, in order to account for unsampled variation within the genera. Although sampling of Z. chasmanthiflora has not yet been possible, based on a number of morphological traits it is very likely that it will form a clade with the other former member of Reyemia, Z. nemesioides. Because these two species vary in several traits of interest, Z. nemesioides was coded with the characteristics of both species (thus appearing as “Reyemia”) to account for this variation. Conclusions regarding this uncollected taxon
are obviously tentative; future work will focus on collecting Z. chasmanthiflora to verify
its location within the phylogeny.
Characters were chosen due to their potential importance in pollinator interactions
or to their proposed taxonomic value in previous or current classification systems for this
genus (Table 4.1, Hilliard 1994). Character coding was determined using the
descriptions provided by Hilliard (1994) and personal observation (discussed further
below). Distributional data were also collected from Hilliard (1994). The two most
distantly related outgroup species (Aptosimum depressum and A. indivisum) were used to root the other taxa. These species were determined to be more distantly related to
Zaluzianskya than the other outgroups based on current taxonomy (placement in a separate tribe, Aptosimeae) and average sequence divergence. As a result, they served as outgroups for both Zaluzianskya and the other original outgroup species. This allows not only for the same rooting of Zaluzianskya as when all 11 “outgroups” were used for
74 rooting, but also provides a more realistic rooting for the other taxa – of potential importance in determining the ancestral character state within Zaluzianskya. The taxon order amongst the other genera using this rooting is congruent with that found in previous analyses at the tribe level (Kornhall et al. 2001). The species of Aptosimum were then not considered further (or shown) in considerations of character evolution because they are too distantly related to be of import for issues of character evolution within Zaluzianskya.
Concentrated-changes tests (Maddison 1990) were conducted in MacClade to test for a correlation between changes in distribution (in relatively arid vs. mesic regions of
South Africa) and habit (annual vs. perennial). These tests are meant to determine whether the presence of a certain character trait (the independent trait) predisposes another trait to evolve (the dependent trait). This is done by testing whether the depending trait evolves in the presence of the independent trait more frequently than expected by chance. Because distribution in more mesic areas was derived basally to changes to perennial habit (see results), the probability that distribution in a mesic rainfall regime facilitates the evolution of perenniality was tested. Concentrated-changes tests cannot be conducted when polytomies are present, thus the single polytomy within the ingroup was arbitrarily resolved. All possible resolutions of the polytomy were tested and found to have little effect on p-values (the polytomy was not variable for either trait).
Probabilities were determined by exact count with either character state set at the ancestral node. Analyses were run both with “distinguished branches” defined only as branches with the derived character state and also as those with the derived state plus equivocal branches. All possible reconstructions were considered on all most parsimonious trees.
75 RESULTS
For each character examined, Table 4.2 gives the ancestral character state for
Zaluzianskya as well as the number of changes within the genus for that character. The combination of the number of most parsimonious trees (six), number of characters (11), alternate placements for Z. microsiphon and Z. mirabilis, as well as multiple possible reconstructions for many characters makes it unfeasible to show the evolution of each character here. However, several reconstructions for flowering time are shown in Fig.
4.3, Fig. 4.4 summarizes the changes in other traits within Zaluzianskya, and Fig. 4.5 shows the reconstruction of habit and distribution.
Flowering time
Within Zaluzianskya, the ancestral state is day flowering with the derivation of night flowering within one species of “Reyemia” (Z. chasmanthiflora), Z. peduncularis
(sect. Zaluzianskya subsect. Noctiflora), and clade 3 (Fig. 4.3). Within clade 3, either day or night flowering could be ancestral, depending on the most parsimonious tree and character reconstruction. This clade involves two to five changes in flowering time. The fewest changes occur when Z. mirabilis is placed sister to the other members of the clade and the lineages of Z. microsiphon are considered to be monophyletic (Fig. 4.3D).
Flowering time within several outgroups is not clearly documented and thus they are coded as uncertain. The flowers of some outgroup species may not open and close daily
76 like Zaluzianskya but instead simply become receptive at certain times of day.
Regardless, the coding of these genera does not affect the reconstructed ancestral state for
Zaluzianskya.
Floral symmetry
Actinomorphic flowers are ancestral within Zaluzianskya with three independent
derivations of slight or occasional zygomorphy (in Z. elongata, Z. violacea, and Z.
minima) and two to four derivations of “strong” zygomorphy (in Z. microsiphon and
“Reyemia”). In the latter case, the larger number of derivations occurs when the
populations of Z. microsiphon are divided as indicated by the phylogeny.
Throat indumentum
A full ring of eglandular hairs surrounding the corolla throat occurs in most
species of Zaluzianskya and is ancestral within the genus. This ring of hairs has been
partially lost twice (in Z. elongata and Z. benthamiana) and completely lost 4-6 times (in
Z. microsiphon, Z. mirabilis, Z. peduncularis, and Z. minima), again depending on the status of Z. microsiphon. The former species of Reyemia are similar to many of the outgroup species in having hair within the throat rather than (or in addition to) around the mouth.
Petal shape
Reconstructions of the evolution of petal shape (notched vs. entire) are consistent amongst all trees and all placements of Z. microsiphon and Z. mirabilis. However, the
77 ancestral state is equivocal. If entire petals are ancestral, there have been two
independent derivations of notched petals, with entire petals retained in clade 2. If
notched petals are ancestral, there have been two derivations of entire petals. Regardless
of the ancestral state, there is definitely a reversal to entire petals within Z. benthamiana
of clade 1.
Stamen number and position
The ancestral androecial state is possession of four stamens, a posticous (adaxial)
pair and an anticous (abaxial) pair. The anticous pair was reduced to staminodes in
“Reyemia” and was lost twice (in Z. cohabitans and Z. affinis + Z. villosa + Z. gracilis).
The posticous pair was only lost once (in Z. benthamiana).
Corolla tube length
Three states were arbitrarily devised to reflect natural gaps in the continuous character distribution for maximum corolla tube length (see Stevens 1991). The shortest length range (5 – 15 mm) was ancestral with five derivations of the intermediate length range (20 – 35 mm; in Z. mirabilis, Z. peduncularis, Z. divaricata, Z. nemesioides and a subclade of clade 1, with a reversal in Z. gracilis) and one derivation of the longest length range (40-60 mm; in clade 3). The ancestral state for clade 3 is either 40-60 mm (if Z. mirabilis is nested within section Nycterinia) or equivocal (if Z. mirabilis is sister to section Nycterinia). Outgroups were coded to represent the majority of species in each genus. In the case of Sutera (coded as 5 – 15 mm), 10% of the species had slightly
78 longer corolla tubes (up to 17 – 20 mm in four species and up to 28 mm in one species), and in the case of Polycarena (coded as 5 – 15 or 20 – 35 mm) one of the 17 species had corolla tubes up to 40 mm.
Nectary
The nectaries of Sutera and Jamesbrittenia are annular, but I was unable to discern from descriptions whether they are adnate or free from the ovary and thus coded these two genera as uncertain. However, the coding of these two genera has no effect on the reconstruction of character evolution within Zaluzianskya because all of the other outgroups have adnate nectaries. This state is ancestral for Zaluzianskya with three independent derivations of free nectaries in the Z. villosa + Z. gracilis + Z. affinis clade,
Z. cohabitans, and clade 3.
Calyx lobing
The calyces within Zaluzianskya are consistently toothed, including the two species formerly placed within Reyemia. Those species immediately basal to
Zaluzianskya have calyces that are lobed halfway (or sometimes to the base in
Glumicalyx) while Jamesbrittenia and Lyperia have calyces that are lobed to their base.
Stem indumentum
Multiple types of trichomes occur within tribe Manuleeae and both the occurrence and orientation of these hairs have been considered to be taxonomically important
(Hilliard 1994). The evolution of eglandular stem hairs was examined in this study. The
79 ancestral state is equivocal, with clade 1 marked by spreading hairs (with one change to retrorse hairs in Z. violacea) and clade 2 + 3 marked by retrorse hairs (with one loss of hairs in Z. mirabilis and one change to spreading hairs in Z. ovata). Also, four species in clade 2 + 3 had eglandular hairs that were intermediate between spreading and retrorse
(Z. katherinae, Z. glareosa, Z. divaricata, and Z. pusilla). If Z. mirabilis sister to section
Nycterinia, the ancestral state of clade 2 + 3 is equivocal, if Z. mirabilis is placed within section Nycterinia, the ancestral state is retrorse hairs.
Habit and biogeography
An annual habit is ancestral within Zaluzianskya, with one to three derivations of a perennial habit (depending on the reconstruction) within clade 3 (Fig. 4.5). The distributions of these species were explored in two ways, both relating to rainfall. South
Africa is divided generally into two main rainfall regions, a western winter rainfall region and an eastern summer rainfall region (Fig. 4.6). Because annual species of Zaluzianskya are distributed in the west and perennials in the east, the relationship of these distributions to rainfall regions was explored, but no connection was apparent. Broadly speaking, another rainfall trend can be seen from west to east in South Africa. Most of the western part of the country is quite arid while much of the east is comparatively mesic
(Fig 4.6; Linder and Kurzweil 1999, p. 9-10). The distributions of annuals and perennials more closely matched this latter rainfall trend. Using the rainfall maps provided by
Linder and Kurzweil (1999), the distributions of species of Zaluzianskya were coded into two states relating to the relative aridity of the region (annual rainfall of less than 50 cm or greater than 50 cm), regardless of the time of year of maximum precipitation. The
80 division between these two states was chosen arbitrarily by seeking a natural break in rainfall range. When traced onto the phylogenies, the origin for Zaluzianskya is estimated to be an arid region, possibly with several migrations between regions.
Depending on the topology and reconstruction of character states, there could have been one migration to a mesic region with or without one return to arid areas or two migrations with no returns. Additionally, four species were coded as polymorphic and may represent range expansions into either region (Z. ovata [perennial], Z. glareosa
[perennial], Z. capensis [annual], and Z. peduncularis [annual]). Concentrated-changes tests comparing these two variables gave p-values that were not significant, ranging from p = 0.0808 to p = 0.2047.
DISCUSSION
The evolution of several of the traits investigated demonstrated potential synapomorphies for the infrageneric groups resolved by the phylogenetic analyses as well as raising implications regarding pollinator interactions. Both of these will be discussed below, for each trait, followed by a discussion of the evolution of habit and biogeography. As pointed out above, the traits of both of the former species of Reyemia will be considered – however, because Z. chasmanthiflora has not yet been sampled, its placement is inferred from the known placement of Z. nemesioides. Thus all conclusions involving character states of Z. chasmanthiflora are predictions rather than direct conclusions based on the phylogeny.
81 Flowering time
Flowering time is clearly important in relation to pollinator interactions. Beyond
the need to attract and correctly position a compatible pollinator, a plant first must be
located in the same general space and time. Flowers of Zaluzianskya open and close
daily, some species opening during the day and some opening at dusk. Reconstructions
of this trait show that day flowering is ancestral within the genus, and there have been at
least three changes to night flowering (Fig. 4.3). Two occur within clade 2: Z.
peduncularis and Z. chasmanthiflora. The former is the sole collected member of section
Zaluzianskya subsection Noctiflora (comprising four night-flowering species), and the
latter is one of the two species formerly included in Reyemia (the other, Z. nemesioides, is day flowering).
The remaining night-flowering species of Zaluzianskya occur within clade 3
(section Nycterinia plus Z. mirabilis). Within this clade the evolution of flowering time becomes less clear due to the uncertain status of both day-flowering species, Z. mirabilis and Z. microsiphon. Based on the molecular data, it is not possible to determine whether day flowering is ancestral for clade 3 with one to multiple derivations of night flowering
(Fig. 4.3B, D), or if night flowering is ancestral, with subsequent derivations of day flowering (Fig. 4.3C). However, the latter seems more likely because multiple other characteristics have been independently derived within the Z. microsiphon lineages (see below). Phylogenetic analyses of morphological data also support this conclusion
(Johnson et al. 2002).
Although the evolution of flowering time appeared quite simple in the previous classification, with all members of section Nycterinia being night flowering except for Z.
82 microsiphon, the possibility has now been raised that flowering time could be fairly labile
within this clade. First, the potential introduction of Z. mirabilis within the clade would
imply an additional change in flowering time. Second, the division of Z. microsiphon
into multiple lineages indicates that flowering time is fairly changeable in this clade, with
at least three changes in flowering time even if Z. mirabilis is sister to the Nycterinia species. However, as discussed in the introduction, the separation of the Z. microsiphon lineages might be an artifact of hybridization, thus it is possible that only one change in flowering time has occurred within section Nycterinia (Fig. 4.3D). As with several of the other distinctive traits of Z. microsiphon (when compared to other members of section
Nycterinia), clarification of flowering time will require further study of Z. microsiphon itself.
Floral symmetry
Although zygomorphic flowers are considered the norm in North American species of Scrophulariaceae s.s., many South African species within this family are actinomorphic (e.g., Hilliard 1994). This is true of the majority of the species of
Zaluzianskya; actinomorphic flowers are ancestral in this genus (Table 4.2). Only three of the 55 species of Zaluzianskya are strongly and consistently zygomorphic (Z. microsiphon, Z. nemesioides, and Z. chasmanthiflora), two of which were formerly placed in Reyemia. Some individuals of up to 17 other species in the genus (including many not sampled in this study) are very slightly zygomorphic, with the degree of zygomorphy varying between populations, individuals, or sometimes even between flowers within individuals (personal observation, Hilliard 1994). Floral symmetry within
83 Zaluzianskya is homoplastic, even if you exclude the potential convergent evolution of this state in Z. microsiphon. Slight zygomorphy evolved twice on the phylogeny within clade 1 and once in clade 3, and strong zygomorphy was independently derived in clades
2 and 3 (Fig. 4.4). In regards to pollination, a zygomorphic, vertically oriented corolla limb is considered attractive to long-proboscid flies (as well as others, such as some bees). In contrast, moths appear to be attracted to actinomorphic flowers with corolla limbs oriented horizontally to the ground (Faegri and Pijl 1979). This could explain the derivation of zygomorphy within the Z. microsiphon lineages and Z. nemesioides (which is day flowering but whose pollinator-interactions have not been studied), but not within
Z. chasmanthiflora, which is night flowering. Little is known of the pollinator- interactions and breeding systems within Zaluzianskya outside of section Nycterinia
(Johnson et al. 2002). Further studies in this area may aid in understanding floral evolution.
Throat indumentum
Most species of Zaluzianskya have a noticeable ring of relatively long eglandular hairs surrounding the mouth of the corolla tube, but some species have only a partial ring and others completely lack these hairs (Fig. 4.7). Particularly in those species with a partial ring, the number of hairs sometimes varies between individuals in a species.
Many of the outgroup species have hairs in various configurations actually within the throat rather than surrounding the mouth. This is also true of the two former species of
Reyemia, which have hairs at the base of the posticous corolla lip that continue down the back of the throat. Although the different arrangements of hairs within the throat may not
84 be homologous, they were all coded as one character-state because this variation would only be important for outgroup species, which are not the focus of this study. A full ring is ancestral in Zaluzianskya, with a single change to throat-hairs in Z. nemesioides, two independent changes to a partial ring in Z. benthamiana and Z. elongata, and four to six complete losses of the ring (depending on the placement of the Z. microsiphon lineages).
Petal indumentum is potentially important to pollinators as a tactile guide, visual guide, or due to its possible role in scent production (although Johnson et al. 2002 concluded that scent production from these hairs within Zaluzianskya may not be likely).
As discussed by Johnson et al. (2002), the ring of hairs is visually similar to the dissected inner corona of moth-pollinated Silene and may be adaptive for moth pollination as a visual or tactile guide. This may explain why it was lost in the long-proboscid fly- pollinated lineages within section Nycterinia. However, this ring of hairs also occurs in most of the day-flowering species of Zaluzianskya and its function in these species, if it has one, has yet to be determined. These species are basal to the moth-pollinated species and have different throat indumentum than the outgroups, thus the occurrence of this ring is not explained by phylogenetic constraint. Johnson et al. (2002) hypothesized that the ring of hairs may aid in reducing nectar robbing and helping to orient pollinators for maximal pollen dispersal, but these hypotheses remain to be tested.
Petal shape
Two main petal shapes occur in Zaluzianskya; each of the five petal lobes is either entire or bifid (Fig. 4.7). There are also some minor variations such as lobes that are retuse or doubly bifid, however, these often vary within a species and will not be
85 distinguished here. The ancestral state within Zaluzianskya is uncertain. Either there was a basal derivation of notched petals with a subsequent reversal to entire petals in clade 2, or there were two separate derivations of notched petals in clades 1 and 3. Regardless, petal shape is a potentially useful character for infrageneric classification; each major clade is marked by the presence of either bifid or entire petals (Fig. 4.4). The only exception within the currently collected species is that of Z. benthamiana, the single member of section Holomeria that was placed within clade 1. This species has entire lobes, contrasting with the bifid lobes found in other members of clade 1. As discussed in Chapter 3, Z. benthamiana is morphologically distinct from other members of clade 1 in several respects – most conspicuously petal color and shape, but also throat indumentum (see above). However, the placement of this species within clade 1 is well supported by the molecular data (bootstrap = 100, decay = 19). This distinctive species warrants further investigation.
As mentioned above in regards to throat indumentum, a dissected corolla limb is considered to be attractive to hawkmoths (Hilliard 1994); this could explain the bifid petal lobes found in the night-flowering members of clade 3. However, again this explanation would not extend to the day-flowering species of clade 1.
Stamen number and position
Four androecial states occur within Zaluzianskya: four stamens (a posticous and anticous pair), two posticous stamens with the anticous stamens reduced to staminodes
(occurring only within the former species of Reyemia), two posticous stamens only, or two anticous stamens only (occurring only within Z. benthamiana and a species not
86 included in the current data set, Z. diandra). The occurrence of all four stamens is ancestral and widespread within the genus, with two independent losses of the anticous pair: within the Z. affinis + Z. villosa + Z. gracilis clade and within Z. cohabitans (Fig.
4.4). The former group was discussed in Chapter 3 as three very closely related species that potentially should be merged into one or two. These three species are nearly identical morphologically, with some intergradation of those characteristics that have been used to separate them (Hilliard 1994). In addition, Z. villosa and Z. gracilis are derived from within a clade of Z. affinis populations, also supporting the close relationship of these groups. Within clade 1, it appears that this loss of the anticous stamens could be a synapomorphy for the clade of these three species – however, several uncollected species within subsection Zaluzianskya share this character state. It remains to be seen what the status of this trait will be once these species are added to the data set.
The only other sampled member of this genus with only two posticous stamens is Z. cohabitans, which was placed in clade 2.
The unusual nature of Z. benthamiana is also highlighted by this trait. This is one of only two species within the genus to retain the anticous pair of stamens while losing the posticous pair. The stamens of the former Reyemia species are also unusual, with two posticous stamens and the anticous stamens reduced to staminodes. According to
Hilliard (1994, p. 17), the occasional reduction of stamens to staminodes has occurred in other species of Zaluzianskya, but not consistently.
Further variation in stamen orientation occurs within this genus, and could be of interest in future studies. For example, although all species in section Nycterinia have four stamens, those of Z. microsiphon are oriented differently in the mouth versus many
87 other members of the section. This is likely important in interactions with their
pollinators (long-proboscid flies). Goldblatt and Manning (2000) found that flowers with
this pollination syndrome often had different, but specific, anther arrangements. As a
result, sympatric plant species with the same pollinator placed their pollen on different
parts of the fly. This specificity could be important in reducing stigma contamination
because long-proboscid flies do not show floral constancy while foraging (Goldblatt and
Manning 1999).
Corolla tube length
All species of Zaluzianskya have a narrow, fused, corolla tube, but the length of
this tube varies greatly among species (4 – 60 mm). This variation was divided into three
character states that reflected natural gaps (0 = up to 5 – 15 mm, 1 = 20 – 35 mm, 2 = 40-
60 mm). Zaluzianskya chasmanthiflora and Z. nemesioides have contrasting corolla tube lengths (c. 20-30 and c. 7-9, respectively), thus “Reyemia” was coded as polymorphic
(0/1). The outgroup genera comprise species with short corolla tubes or with either short or intermediate tubes. Among those taxa sampled, only Zaluzianskya includes species with long corolla tubes (all placed within section Nycterinia). This trait marks clade 3, with the exception of Z. mirabilis, whose tubes are intermediate in length (character state
2). Short corolla tubes are ancestral in the genus, with five derivations of intermediate tubes and one reversal to short tubes. This reversal occurs in Z. gracilis of clade 1 and is the main difference between this species and the closely related Z. villosa (i.e., smaller flowers). Clade 1 is otherwise largely marked by intermediate tube lengths, except
88 additionally to the Z. minima + Z. bella clade. Clade 2 includes three derivations of
intermediate tube length (Z. chasmanthiflora, Z. divaricata, and Z. peduncularis).
Long and narrow corolla tubes have long been considered important in
specialized pollination systems, particularly for moths and more recently for long-
proboscid flies (Goldblatt and Manning 2000, Darwin 1862). The presence of long tubes
in most of clade 3 is thus not surprising. The impact of this trait on pollinator interactions
within the other two clades of the phylogeny cannot be determined until pollination
studies have been done on these species, although it is possible to surmise that the
shortest tube lengths found in some species of clades 1 and 2 probably do not
significantly restrict visitor access. The direction of evolution in this trait is consistently
towards longer corolla tubes – with the exception of Z. gracilis and potentially Z.
mirabilis (if it is nested within section Nycterinia).
Nectary
All species of Zaluzianskya have a nectariferous gland at the base of the ovary. In
fact this condition is common in a large number of angiosperms, including much of the
Lamiales (Brown 1938). However, this particular type of nectary, a small gland often
adnate to the base of the posticous side of the ovary, appears to be unique to Manuleeae
and is found in 13 of the 17 genera in this tribe. Differing adnation of this gland was
used by Hilliard (1994) in support of her infrageneric classification for Zaluzianskya.
Nectaries of species within sections Nycterinia and Macrocalyx are free from the ovary, those from section Zaluzianskya subsect. Zaluzianskya are either free (in those species with two stamens) or adnate (in those species with four stamens), while those from
89 subsect. Noctiflora and section Holomeria are adnate. Diversity in nectaries has also been used to define groups at low taxonomic levels in other groups. For example, a type of bracteal nectary delimits a monophyletic lineage within Aphelandra (Acanthaceae,
McDade and Turner 1997), and epistaminal nectaries were considered to link Penstemon and Chionophila (Scrophulariaceae, Straw 1966; although later phylogenetic analyses suggested that this nectary type may have evolved independently in these two genera,
Wolfe et al. 2002). Nectary characteristics also delimit clades in the current phylogeny.
The occurrence of a free nectary is ancestral, with three independent derivations of the adnate state, one marking clade 3, one for Z. cohabitans within clade 2 (again distinguishing it from other members of that clade), and one marking the Z. affinis + Z. villosa + Z. gracilis clade within clade 1 (again supporting the cohesiveness of these three species). Thus this character is congruent with several of the characters previously discussed.
Calyx lobing
Within tribe Manuleeae, calyces vary from being lobed nearly to the base, lobed halfway down the calyx, to merely toothed. The latter condition occurs in all species of
Zaluzianskya, including those formerly placed within Reyemia. This is one of the morphological traits supporting the latter’s inclusion within Zaluzianskya. The only other genus in tribe Manuleeae (sensu Hilliard 1994) with toothed calyces is Tetraselago (not included in this study). It appears from the sampled taxa that the evolution of this trait within the tribe may have progressed from fully-lobed calyces, through half-lobed calyces (with some variation within Glumicalyx), to toothed calyces. However, Kornhall
90 et al. (2001) examined the evolution of this trait with increased taxon sampling at the tribal level and found additional changes between fully-lobed and half-lobed calyces as well as an independent derivation of toothed calyces in Tetraselago.
Stem indumentum
Hilliard (1994) considered both orientation and type of indumentum to be very useful for the systematics of both genera and species within tribe Manuleeae. For example, Phyllopodium and Polycarena are “readily distinguished” by differences in vegetative indumentum, one of the few traits differentiating Z. villosa and Z. affinis is leaf and bract indumentum, and a similar distinguishing trait for Z. bella and Z. violacea is hair orientation on the stem (Hilliard 1994). The latter trait was examined here, i.e., the orientation of acute hairs on the stem (spreading vs. retrorse). The ancestral state for
Zaluzianskya is uncertain. However, with the exception of Z. violacea, clade 1 is marked by spreading stem hairs while clades 2 and 3 largely contain species with retrorse stem hairs. Exceptions include Z. mirabilis, which does not have eglandular hairs on the stem
(but does have spreading glandular hairs) and Z. ovata with an independent derivation of spreading hairs. The Z. pusilla + Z. divaricata clade, Z. glareosa, and Z. katharinae all have stem hairs described as intermediate between retrorse and spreading (Hilliard 1994), although the collections of these species used for sequencing largely had retrorse stem hairs (corresponding to the state found in most other members of their clades). This characteristic appears to mark clades fairly well, although its utility for identification may be somewhat limited by the fact that it can be difficult to distinguish the orientation of stem hairs, particularly in herbarium specimens.
91 Habit and biogeography
An annual habit is ancestral within Zaluzianskya (Fig. 4.5). All perennials within
the genus are placed in section Nycterinia, along with five annual species (two of which
are included in this data set: Z. capensis and Z. maritima). The placement of these two annuals, in addition to Z. mirabilis, within clade 3 divides the evolution of perennials into one to three groups, possibly with one or two reversals to an annual habit. As mentioned above, annual species of Zaluzianskya generally occur in the western part of southern
Africa whereas perennial species occur in the east. Hilliard and Burtt (1983) noted a similar pattern in several other South African genera, including Diascia, Nemesia, and
Hebenstretia within Scrophulariaceae, and Felicia and Cotula within Asteraceae. There are many possible explanations for this trend, including potential differences in climate, disturbance regimes, soils, or simply phylogenetic history (e.g., Ehrlén and Lehtilä 2002,
Dean and Milton 1995, Garnier and Laurent 1994, Gross et al. 1992). I investigated the potential relation of the evolution of habit to two climatic factors: rainfall season and minimum annual rainfall. The distributions of each species of Zaluzianskya were compared to both winter vs. summer rainfall regions and relatively arid vs. mesic rainfall regions, each of which also displays a general west to east trend (Fig. 4.6; Linder and
Kurzweil 1999). The winter and summer rainfall regions are considered to be floristically distinct, with the distributions of some plant groups focused in one region or the other (e.g., Goldblatt et al. 2001, Esler et al. 1999, Gibbs Russell 1987). However, seasonality of rainfall showed no relationship to habit within Zaluzianskya, possibly because many of the species occur near the boundary between these two regions. The quantity of rainfall appeared to be more relevant; it was possible to code most species as
92 being distributed either within relatively arid regions (annual rainfall of less than 50 cm) or more mesic regions (greater than 50 cm). Evolutionary reconstruction of distribution showed a similar pattern to that seen with habit, but not identical. The region of origin was reconstructed to be arid, with one to two migrations to relatively mesic regions and no returns or a single return to arid regions (again, all within clade 3). Additionally, there were up to four other movements between regions because four species occur in both regions (Fig. 4.5). Unfortunately, it is not possible to more specifically narrow down the number and type of movements, due to equivocal branches and differences between most parsimonious trees.
Although the reconstructed patterns for the evolution of habit and distribution were similar, concentrated changes tests did not show a significant correlation between these variables. These tests can be useful in determining whether changes in one trait predispose a taxon for changes in another. However, a non-significant result does not necessarily establish the lack of an association between traits. In fact, if two traits are very tightly linked the test may give a non-significant p-value because the traits will appear to change at the same location in the phylogeny rather than change in one trait leading to change in another (Weller and Sakai 1999). The topological similarities between the reconstructions of these two traits (habit and distribution) are suggestive of an association, but this hypothesis cannot be definitively tested with these data. Certainly the amount of rainfall is important in relation to life-history traits. However, the proposed relationship is not perfect. One of the sampled species of Zaluzianskya does not follow the general trend, and four others occur in both rainfall regimes. The former,
Z. maritima (an annual distributed within relatively mesic regions), occurs on the
93 southern coast of the Eastern Cape province, extending into the southeastern coast of the
Western Cape province. This narrow coastal area generally receives more rain than nearby inland regions. Although Hilliard (1994) considered this species an annual, she also noted that they were “sometimes persisting for more than one season” (similarly noted for Z. capensis). The single population of Z. maritima seen in the field included individuals that appeared to be perennial, but further field studies are necessary to more closely examine the life-history of this species. The four relatively widespread species that occur in both rainfall regimes (Z. capensis, Z. glareosa, Z. ovata, and Z. peduncularis) include two annual species and two perennial species. As mentioned, at least one of these species, Z. capensis, was described as annual but sometimes survives for more than one year. Population-level sampling of these species may help to reveal whether lifespan varies predictably across their ranges.
These reconstructions suggest that Zaluzianskya originally evolved in an arid region, followed by dispersal to more mesic areas. However, they contain the implicit assumptions that the distributions of current populations and the distributions of current rainfall regimes reflect those found in the past. Obviously, these assumptions may not be correct. Current distributions may be smaller, larger, or relocated from those in even the recent past. For example, Axelrod and Raven (1978) discuss the recent (in the past 500 years) expansion of many fynbos taxa from the southwest coast of Africa eastward into the summer rainfall region. Certainly, the distributions of most plant species are in a state of flux, historically due to the continuously changing climate and other environmental factors and even more so now with the escalating influence of humans on other organisms. Unfortunately, detailed information on historic distributions for many
94 species is often unavailable because it relies on unearthing, identifying, and dating a significant amount of fossil material. As a result, phylogenetic reconstructions of ancestral areas can be a useful first look at this history.
The distribution of ‘relatively arid’ regions is also changeable. Aridity began to spread in South Africa in the mid-Tertiary, potentially favoring at that time “the evolution of herbaceous forms with adaptations to survive annual periods of aridity” (Goldblatt et al. 2002 p. 357). Axelrod and Raven (1978) proposed that xerophytic species, which would in the future be preadapted for radiation in the winter rainfall region, began to be established in this time period, if not earlier. Southern Africa continued to become drier and it was not until approximately 5 mya (Goldblatt et al. 2002, Axelrod and Raven
1978) that the winter rainfall region was truly established.
As noted by Richardson et al. (2001), continental species do not have the benefit of using island age to calibrate node ages in phylogenies. Richardson et al. also noted the scarcity of fossils in the Cape; I am not aware of fossil evidence for this group.
Additionally, molecular clock estimates, always made questionable based on the number of assumptions involved, are particularly untrustworthy when used in groups with both annuals and perennials (Andreasen and Baldwin 2001). As a result of these factors, I do not currently have time estimates for the evolution of Zaluzianskya but it appears likely that this species diversified at some point after the establishment of arid regions within
South Africa, with subsequent dispersal and diversification into more mesic regions.
This is in contrast to the results found for Thamnochortus (Restionaceae; Linder and
Mann 1998), where the arid-adapted species appeared to be derived from mesic-adapted species.
95 Summary
Beyond those traits examined here there are obviously many other characteristics that could be explored that may or may not support the same conclusions. Traits were chosen if they appeared relevant based on previous and current classifications, potential pollinator interactions, and personal observation. Certainly many other characteristics of these taxa also merit future investigation.
Several of the traits examined support major or minor clades within the molecular phylogeny. For example, night flowering may be a good synapomorphy for most of clade 3. However, this depends on whether or not Z. microsiphon is a single lineage, if it is not then flowering time is somewhat labile within this clade. Petal shape, tube length, nectary adnation, and stem indumentum all mark major clades within the genus. Clades 1
(except Z. benthamiana) and 3 comprise species with notched petal lobes while clade 2 includes species with entire petal lobes. With the exception of Z. mirabilis, clade 3 is also distinguished by relatively long corolla tubes. Clade 2 is variable for this characteristic, while a subclade within clade 1 possesses tubes of largely intermediate length (except Z. gracilis, which have short tubes). Similarly, a free nectary marks all of clade 3 and a smaller clade within clade 1. Clade 2 is marked by adnate nectaries, except for Z. cohabitans. Clade 2 + 3 comprises species with retrorse stem hairs, with a few exceptions, while almost all species in clade 1 have spreading hairs.
The placement of “Reyemia” within clade 2 of Zaluzianskya is supported by petal shape. The inclusion of the former species of Reyemia within Zaluzianskya in general is supported by calyx lobing (among other traits, see Chapter 3). However, these two species’ unusual nature is emphasized by several of the other characters examined here.
96 For example, they are unique within Zaluzianskya in having eglandular hairs inside the
corolla throat (a common, if variable, condition in other members of tribe Manuleeae).
They are also unusual because of their strong zygomorphy (occurring only in one other
species of the genus, Z. microsiphon) and reduction of the anticous pair of stamens to
staminodes.
Zaluzianskya benthamiana is also unusual in its androecial characters, having
only the two anticous stamens. This species is strongly placed in clade 1 according to the
molecular data but is different from other members of that clade in having entire petal
lobes (and yellow flowers, although the evolution of flower color was not explored here)
and only a partial ring of hairs around the corolla throat. Of those species sampled, the
latter condition was only found in this species and Z. elongata. However, tube length
corroborates the phylogenetic placement for Z. benthamiana, as does stem indumentum.
This species from section Holomeria was placed in clade 1 with members of section
Zaluzianskya, whereas two other species (Z. peduncularis and Z. cohabitans) from the latter section were placed in clade 2 with several other species of section Holomeria. As discussed above, Z. peduncularis (sect. Zaluzianskya subsect. Noctiflora) is similar to other members of clade 2 in both petal shape and color. This species is unusual within that clade in being night flowering and without a ring of hairs, but those traits distinguish it from most members of clade 1 as well. On the other hand, Z. cohabitans, although it also has entire petal lobes, differs in having a free nectary and only the two posticous stamens (as opposed to having all four). This species shares each of these traits only with the subclade of clade 1 comprising Z. affinis, Z. villosa, and Z. gracilis.
97 Many of the characters examined here are likely to be important in pollinator interactions. Included among these are some of the characteristics distinguishing Z. microsiphon from night-flowering members of section Nycterinia, such as floral symmetry and throat indumentum. As discussed above, the lability of evolution for these traits is uncertain as long as the status of Z. microsiphon is uncertain. Although the direction of evolution of flowering time is unclear within clade 3, it is possible to determine that both zygomorphy and lack of a circlet of hairs are derived within this clade. It appears that long-proboscid fly evolution has evolved from hawkmoth pollination in this clade. This is contrary to the results of Goldblatt and Manning (2000) who found that long-proboscid fly pollination often evolved from long-tongued bee pollination. However, Johnson et al. (2002), using a morphological phylogeny, also found that Z. microsiphon was derived within a hawkmoth-pollinated clade. They discussed the likelihood of this result given that each pollinator syndrome shares several characteristics (e.g., long, narrow corolla tubes and a nectar reward) that could be preadaptations to the derived pollinator syndrome (Stebbins 1970). Goldblatt and
Manning (2000) also found that although the long-proboscid fly pollination system was rare, it was “highly labile in several genera of the Iridaceae and in Disa (Orchidaceae),” evolving repeatedly in different lineages (Goldblatt and Manning 2000, p. 169).
Potentially, the phylogeny produced in Chapter 3 is correct in dividing the Z. microsiphon lineages and this “species” is another example of lability in evolution of these suites of characteristics.
The ancestor of Zaluzianskya appears to be an annual, adapted to relatively arid conditions, with subsequent dispersal to more mesic areas and evolution of perennial
98 species. These changes occurred in clade 3 of the phylogeny, with at least two derivations of perenniality and several potential movements between arid and mesic regions.
99 0123
Flowering time Diurnal Nocturnal
Floral symmetry Actinomorphic Slightly Strongly zygomorphic zygomorphic
Throat Hairs inside Full ring of hairs Partial ring Eglandular indumentum throat around mouth of hairs hairs (eglandular) absent
Petal shape Entire Bifid
Stamen number 4 stamens 2 posticous 2 posticous 2 anticous and position (2 posticous, 2 stamens, 2 anticous stamens only stamens anticous) staminodes only
Corolla tube Up to 5-15 mm 20-35 mm 40-60 mm length
Nectary Adnate to ovary Free from ovary
Calyx lobing Lobed to base Lobed halfway Toothed to base
Stem indumentum Acute hairs Retrorse hairs Spreading (acute) absent hairs
Habit Annual Perennial
Annual rainfall < 50 cm > 50 cm across distribution
Table 4.1. Characters and character states examined within Zaluzianskya and the outgroup taxa.
100 Character Ancestral State No. derivations of new No. reversals to character states ancestral state Flowering time Diurnal (0) 3-6 (of 1) 1-4 Floral symmetry Actinomorphic (0) 3 (of 1) 0-1 2-4 (of 2) Throat Full ring (1) 1 (of 0) 0-1 indumentum 2 (of 2) 4-6 (of 3) Petal shape Entire or Bifid (?) If entire (0) ancestral: 2 (of 1) 1 If bifid (1) ancestral: 2 (of 0) 0 Stamen number 4 stamens (0) 1 (of 1) 0 and position 2 (of 2) 1 (of 3) Corolla tube length Up to 5-15 mm (0) 5 (of 1) 1 1 (of 2) Nectary Adnate (0) 3 (of 1) 0 Calyx lobing Toothed (2) 0 0 Stem indumentum Absent, retrorse, or If absent (0) ancestral: spreading (?) 2 (of 1) 1 2 (of 2) If retrorse (1) ancestral: 1 (of 0) 1 2 (of 2) If spreading (2) ancestral: 1 (of 0) 1 2 (of 1) Habit Annual (0) 1-3 (of 1) 0-2 Annual rainfall < 50 cm (0) 1-2 (of 1) 0-1 across distribution
Table 4.2. Ancestral character states (coding indicated in parentheses, see Table 4.1) for Zaluzianskya and number of changes within the genus for each character examined. Ancestral state refers to the basal condition for Zaluzianskya, not for the outgroups (e.g., for throat indumentum, a full ring of hairs around the corolla throat [1] is basal for Zaluzianskya, but the ancestral state for the entire phylogeny is having hairs inside the throat [0]). 101 Figure 4.1. The distribution of Zaluzianskya (shaded) within southern Africa (Hilliard 1994).
102 Figure 4.2. One of six most parsimonious trees for combined analyses of ITS, rpl16, and trnL-F with 39 accessions. Bootstrap support values greater than 50% are indicated above the branches and decay values are provided below. Asterisks mark those nodes that are not recovered in the strict consensus. Three major clades within Zaluzianskya are marked.
103 Figure 4.3. Evolution of flowering time within Zaluzianskya, (A) shown on the combined strict consensus tree. Several possible resolutions of this trait were possible, depending on the topology within clade 3. Arrows mark the locations of night-flowering taxa. Three of the potential resolutions within clade 3 are shown (bars indicate synapomorphies): (B) four derivations of night flowering and one reversal to day flowering, (C) a single derivation of night flowering but four reversals to day flowering, and (D) a single derivation of night flowering with a one reversal to day flowering.
104 Figure 4.4. Evolution of eight characters within Zaluzianskya. Character type is indicated by letter; character state is indicated by shade (see Table 4.1). Potential synapomorphies are marked on the combined strict consensus tree; coding is shown on the right. Some of these traits are only synapomorphies in some reconstructions. The only such traits affecting major clades are “c” and “h,” e.g. for “c” if “1” is basal in Zaluzianskya, “2” is a synapomorphy for clades 1 and 3; if “2” is basal, “1” is a synapomorphy for clade 2. The precise location of synapomorphies for “a” and “b” may be shifted slightly depending on the placement of Z. microsiphon. The ITS placement for Z. mirabilis may shift basal reconstructions in clade 3 for “e” and “h” (see text). 105 Figure 4.5. Evolution of habit, i.e. annual (0) or perennial (1), and distribution within relatively arid (0) or mesic (1) regions of South Africa, shown on the combined strict consensus tree. Character coding is indicated next to the taxon names and the three major clades are marked by boxes.
106 Figure 4.6. Diagram of rainfall regions within South Africa, as described by both seasonality and amount of rainfall. The area to the left of the gray line is the winter rainfall region (WRR); the area to the right is the summer rainfall region (SRR). Shaded areas indicate relatively mesic regions, unshaded areas are relatively arid (modified from Linder and Kurzweil 1999).
107 Figure 4.7. Photographs showing some of the character variation within Zaluzianskya for throat indumentum (A, B) and petal shape (C, D). (A) full ring of eglandular hairs around mouth of Z. villosa; (B) glabrous mouth of Z. microsiphon; (C) entire petal lobes of Z. peduncularis; (D) bifid petal lobes of Z. capensis.
108 LIST OF REFERENCES
Allan, G. J., C. Clark, and L. H. Rieseberg. 1997. Distribution of parental DNA markers in Encelia virginensis (Asteraceae: Heliantheae), a diploid species of putative hybrid origin. Plant Systematics and Evolution 205: 205-221.
Andersson, M. 1999. Hybridization and skua phylogeny. Proceedings of the Royal Society of London Series B, Biological Sciences 266: 1579-1585.
Andreasen, K. and B. G. Baldwin. 2001. Unequal evolutionary rates between annual and perennial lineages of checker mallows (Sidalcea, Malvaceae): evidence from 18S- 26S rDNA internal and external transcribed spacers. Molecular Biology and Evolution: 936-944.
Arnold, M. L. 1992. Natural hybridization as an evolutionary process. Annual Review of Ecology and Systematics 23: 237-261.
Axelrod, D. I. and P. H. Raven. 1978. Late cretaceous and tertiary vegetation history of Africa. In M. J. A. Werger, ed. Biogeography and Ecology of Southern Africa Part 1. 77-130. W. Junk, The Hague, Netherlands.
Backeljau, T., L. De Bruyn, H. De Wolf, K. Jordaens, S. Van Dongen, and B. Winnepenninckx. 1996. Multiple UPGMA and neighbor-joining trees and the performance of some computer packages. Molecular Biology and Evolution 13: 309- 313.
Baldwin, B. G., M. J. Sanderson, J. M. Porter, M. F. Wojciechowski, C. S. Campbell, and M. J. Donoghue. 1995. The ITS region of nuclear ribosomal DNA: a valuable source of evidence on angiosperm phylogeny. Annals of the Missouri Botanical Garden 82: 247-277.
Barker, F. K. and F. M. Lutzoni. 2002. The utility of the incongruence length difference test. Systematic Biology 51: 625-637.
Barton, N. H. 1979. The dynamics of hybrid zones. Heredity 43: 341-359.
109 Barton, N. H., and G. M. Hewitt. 1985. Analysis of hybrid zones. Annual Review of Ecology and Systematics 16: 113-148.
Bell, T. L. and F. Ojeda. 1999. Underground starch storage in Erica species of the Cape Floristic Region – differences between seeders and resprouters. New Phytologist 144: 143-152.
Bentham, G. 1836. Synopsis of the Buchnerae, a tribe of Scrophulariaceae. In W. J. Hooker, Companion to the Botanical Magazine 1: 356-384.
Bentham, G. 1846. Scrophulariaceae. In A. P. de Candolle, ed. Prodromus Systematis Naturalis Regni Vegetabilis pars 10. Treuttel et Würtz, Paris, France.
Bremer, K. 1988. The limits of amino acid sequence data in angiosperm phylogenic reconstruction. Evolution 42: 795-803.
Brown, W. H. 1938. The bearing of nectaries on the phylogeny of flowering plants. Proceedings of the American Philosophical Society 79: 549-594.
Capula, M. 2002. Genetic evidence of natural hybridization between Podarcis sicula and Podarcis tiliguerta (Reptilia: Lacertidae). Amphibia-Reptilia 23: 313-321.
Christian, C. E. 2001. Consequences of a biological invasion reveal the importance of mutualism for plant communities. Nature 413: 635-639.
Cruzan, M. B., and M. L. Arnold. 1993. Ecological and genetic associations in an Iris hybrid zone. Evolution 47: 1432-1445.
Darlu, P. and G. Lecointre. 2002. When does the incongruence length difference test fail? Molecular Biology and Evolution 19: 432-437.
Darwin, C. 1862. On the Various Contrivances by which British and Foreign Orchids are Fertilised by Insects. Murray, London, UK.
Datwyler, S. L. 2001. Evolution and dynamics of hybridization in Penstemon subgenus Dasanthera (Scrophulariaceae s.l.). Ph.D. dissertation, Ohio State University. Columbus, OH, USA. 114 p. de Queiroz, K. and M. J. Donoghue. 1988. Phylogenetic systematics and the species problem. Cladistics 4: 317-338.
Dean, W. R. J. and S. J. Milton. 1995. Plant and invertebrate assemblages on old fields in the arid southern Karoo, South Africa. African Journal of Ecology 33: 1-13.
110 dePamphilis, C. W. and R. Wyatt. 1989. Hybridization and introgression in buckeyes (Aesculus: Hippocastanaceae): a review of the evidence and a hypothesis to explain long-distance gene flow. Systematic Botany 14: 593-611. dePamphilis, C. W. and R. Wyatt. 1990. Electrophoretic confirmation of interspecific hybridization in Aesculus (Hippocastanaceae) and the genetic structure of a broad hybrid zone. Evolution 44: 1295-1317.
Diels, L. 1897. Beiträge zur kenntnis der Scrophulariaceen Afrikas. Botanische Jahrbuch 23: 471-517.
Donoghue, M. J. 1985. A critique of the biological species concept and recommendations for a phylogenetic alternative. The Bryologist 88: 172-181.
Downie, S. R., D. S. Katz-Downie, and M. F. Watson. 2000. A phylogeny of the flowering plant family Apiaceae based on chloroplast DNA rpl16 and rpoC1 intron sequences: towards a suprageneric classification of subfamily Apioideae. American Journal of Botany 87: 273-292.
Doyle, J. J., and J. L. Doyle. 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemistry Bulletin 19: 11-15.
Ehrlén, J. and K. Lehtilä. 2002. How perennial are perennial plants? Oikos 98: 308-322.
Ellis, A. G. and S. D. Johnson. 1999. Do pollinators determine hybridization patterns in sympatric Satyrium (Orchidaceae) species? Plant Systematics and Evolution 219: 137-150.
Endress, P. K. 2001. The flowers in extant basal angiosperms and inferences on ancestral flowers. International Journal of Plant Sciences 162: 1111-1140.
Eriksson, T. 1999. AutoDecay ver. 4.0 (program distributed by the author). Bergius Foundation, Royal Swedish Academy of Sciences, Stockholm.
Esler, K. J., P. W. Rundel, and P. Vorster. 1999. Biogeography of prostrate-leaved geophytes in semi-arid South Africa: hypotheses on functionality. Plant Ecology 142: 105-120.
Esselman, E. J., L. Jianqiang, D. J. Crawford, J. L. Windus, and A. D. Wolfe. 1999. Clonal diversity in the rare Calamagrostis porteri ssp. Insperata (Poaceae): comparative results for allozymes and random amplified polymorphic DNA (RAPD) and intersimple sequence repeat (ISSR) markers. Molecular Ecology 8: 443-451.
Faegri, K. and van der Pijl. 1979. The Principles of Pollination Ecology, 3rd ed. Pergamon Press, New York, NY, USA.
111 Farris, J. S., V. A. Albert, M. Källersjö, D. Lipscomb, and A. G. Kluge. 1996. Parsimony jackknifing outperforms neighbor-joining. Cladistics 12: 99-124.
Farris, J. S., M. Källersjö, A. G. Kluge, and C. Bult. 1995. Testing significance of incongruence. Cladistics 10: 315-319.
Feliner, G. N., J. F. Aguilar, and J. A. Rosselló. 2002. Reticulation or divergence: the origin of a rare serpentine endemic assessed with chloroplast, nuclear, and RAPD markers. Plant Systematics and Evolution 231: 19-38.
Felsenstein, J. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783-791.
Felsenstein, J. 1993. PHYLIP (Phylogeny Inference Package) version 3.5c. Distributed by the author. Department of Genetics, University of Washington, Seattle.
Francisco-Ortega, J., A. Santos-Guerra, S.-C. Kim, and D. J. Crawford. 2000. Plant genetic diversity in the Canary Islands: a conservation perspective. American Journal of Botany 87: 909-919.
Frontier, S. 1976. Étude de la decroissance des valeurs propres dans une analyze en composantes principales: comparison avec le modèle de baton brisé. Journal of Experimental Marine Biology and Ecology 25: 67-75.
Garnier, E. and G. Laurent. 1994. Leaf anatomy, specific mass and water content in congeneric annual and perennial grass species. New Phytologist 128: 725-736.
Gibbs Russell, G. E. 1987. Preliminary floristic analysis of the major biomes in southern Africa. Bothalia 17: 213-227.
Givnish, T. J. and K. J. Sytsma, eds. 1997. Molecular Evolution and Adaptive Radiation. Cambridge Univ. Press, New York, NY, USA.
Goldblatt, P. and J. C. Manning. 2002. Plant diversity of the Cape region of southern Africa. Annals of the Missouri Botanical Garden 89: 281-302.
Goldblatt, P. and J. C. Manning. 2000. The long-proboscid fly pollination system in southern Africa. Annals of the Missouri Botanical Garden 87: 146-170.
Goldblatt, P. and J. C. Manning. 1999. The long-proboscid fly pollination system in Gladiolus (Iridaceae). Annals of the Missouri Botanical Garden 86: 758-774.
Goldblatt, P., J. C. Manning, and P. Bernhardt. 2001. Radiation of pollination systems in Gladiolus (Iridaceae: Crocoideae) in southern Africa. Annals of the Missouri Botanical Garden 88: 713-734.
112 Good, R. 1974. The Geography of Flowering Plants, 4th edition. Longman, London, UK.
Goulson, D. and K. Jerrim. 1997. Maintenance of the species boundary between Silene dioica and S. latifolia (red and white campion). Oikos 79: 115-126.
Gross, K L., D. Maruca, and K. Pregitzer. 1992. Seedling growth and root morphology of plants with different life-histories. New Phytologist 120: 535-542.
Haufler, C. H. W. A. Grammer, E. Hennipman, T. A. Ranker, A. R. Smith, and H. Schneider. 2003. Systematics of the ant-fern genus Lecanopteris (Polypodiaceae): testing phylogenetic hypotheses with DNA sequences. Systematic Botany 28: 217- 227.
Hedberg, O. 1970. The genus Zaluzianskya F. W. Schmidt (Scrophulariaceae) found in Tropical East Africa. Botaniska Notiser 123: 512-517.
Hiern, W. P. 1904. Scrophulariaceae. In Thiselton-Dyer, Flora Capensis 4: 121-420.
Hilliard, O. M. 1992. Miscellaneous new names and combinations in Scrophulariaceae - Manuleae. Edingburgh Journal of Botany 49: 297.
Hilliard, O. M. 1994. The Manuleae. A Tribe of Scrophulariaceae. Edinburgh University Press, Edinburgh, UK.
Hilliard, O. M. and B. L. Burtt. 1983. Zaluzianskya (Scrophulariaceae) in south eastern Africa and the correct application of the names Z. capensis and Z. maritima. Notes from the Royal Botanic Garden Edinburgh 41: 1-43.
Hodges, S. A., and M. L. Arnold. 1994. Floral and ecological isolation between Aquilegia formosa and Aquilegia pubescens. Proceedings of the National Academy of Sciences of the United States of America 91: 2493-2496.
Holder, M. T., J. A. Anderson, and A. K. Holloway. 2001. Difficulties in detecting hybridization. Systematic Biology 50: 978-982.
Huang, S. C., C. C. Tsai, and C. S. Sheu. 2000. Genetic analysis of Chrysanthemum hybrids based on RAPD molecular markers. Botanical Bulletin of Academia Sinica 41: 257-262.
Huchett, B. I., and F. C. Botha. 1995. Stability and potential use of RAPD markers in a sugarcane genealogy. Euphytica 86: 117-125.
Jackson, D. A. 1993. Stopping rules in principal components analysis: a comparison of heuristical and statistical approaches. Ecology 74: 2204-2214. 113 Johnson, L. A. and D. E. Soltis. 1998. Assessing congruence: empirical examples from molecular data. In D. E. Soltis, P. S. Soltis, and J. J. Doyle, eds. Molecular Systematics of Plants II. 297-348. Kluwer Academic Publishers, Amsterdam, Netherlands.
Johnson, S. D. 1996. Pollination, adaptation and speciation models in the Cape flora of South Africa. Taxon 45: 59-66.
Johnson, S. D., T. J. Edwards, C. Carbutt, and C. Potgieter. 2002. Specialization for hawkmoth and long-proboscid fly pollination in Zaluzianskya section Nycterinia (Scrophulariaceae). Botanical Journal of the Linnean Society 138: 17-27.
Johnson, S. D., H. P. Linder, and K. E. Steiner. 1998. Phylogeny and radiation of pollination systems in Disa (Orchidaceae). American Journal of Botany 85: 402-411.
Joliffe, I. T. 1986. Principal Component Analysis. Springer-Verlag, New York, NY, USA.
Jong, K. 1993. Variation in chromosome number in the Manuleae (Scrophulariaceae) and its cytotaxonomic implications. Edinburgh Journal of Botany 50: 365-379.
Kellogg, E. A., R. Appels, and R. J. Mason-Gamer. 1996. When genes tell different stories: the diploid genera of Triticeae (Gramineae). Systematic Botany 21: 321-347.
Kimball, R. T., D. J. Crawford, and E. B. Smith. 2003. Evolutionary processes in the genus Coreopsis: insights from molecular phylogenetics. Evolution 57: 52-61.
Kornhall, P., N. Heidari, and B. Bremer. 2001. Selagineae and Manuleeae, two tribes or one? Phylogenetic studies in the Scrophulariaceae. Plant Systematics and Evolution 228: 199-218.
Levin, D. A., J. Francisco-Ortega, and R. K. Jansen. 1996. Hybridization and the extinction of rare plant species. Conservation Biology 10: 10-16.
Linder, H. P. 1985. Gene flow, speciation, and species diversity patterns in a species- rich area: the Cape flora. In E. S. Vrba, ed. Species and Speciation. 53-57. Transvaal Museum Monograph No. 4. Transvall Museum, Pretoria, South Africa.
Linder, H. P. 1996. Priorities for taxonomic research on the Cape flora. South African Journal of Science 92: 323-326.
Linder, H. P. and H. Kurzweil. 1999. Orchids of Southern Africa. A. A. Balkema, Rotterdam, Netherlands.
114 Linder, H. P. and D. M. Mann. 1998. The phylogeny and biogeography of Thamnochortus (Restionaceae). Botanical Journal of the Linnean Society 128: 319- 357.
Luckow, M. 1995. Species concepts: assumptions, methods, and applications. Systematic Botany 20: 589-605.
Maddison, D. R. 1991. The discovery and importance of multiple islands of most- parsimonious trees. Systematic Zoology 40: 315-328.
Maddison, D. and W. P. Maddison. 1992. MacClade: analysis of phylogeny and character evolution. Sinauer Associates Inc., Sunderland, MA, USA.
Maddison, W. P. 1990. A method for testing the correlated evolution of two binary characters: Are gains or losses concentrated on certain branches of a phylogenetic tree? Evolution 44: 539-557.
Maddison, W. P. 1997. Gene trees in species trees. Systematic Biology 46: 523-536.
Mantel, N. A. 1967. The detection of disease clustering and a generalized regression approach. Cancer Research 27: 209-220.
Mason-Gamer, R. J. and E. A. Kellogg. 1996. Testing for phylogenetic conflict among molecular data sets in the tribe Triticeae. Systematic Biology 45: 524-545.
McDade, L. 1992. Pollinator relationships, biogeography, and phylogenetics. BioScience 42: 21-26.
McDade, L. A. and M. D. Turner. 1997. Structure and development of bracteal nectary glands in Aphelandra (Acanthaceae). American Journal of Botany 84: 1-15.
McGregor, T. E. 1989. Pollination syndromes in the genus Zaluzianskya: Z. microsiphon (Kuntze) K. Schum, Z. glareosa Hilliard and Burtt, Z. ovata Walp. and Z. spathacea Benth. B.S. honours thesis, University of Natal, Pietermaritzburg, South Africa. p. 59.
Mickevich, M. F. and J. S. Farris. 1981. The implications of congruence in Menidia. Systematic Zoology 30: 351-370.
Miller, M. P. 1997. Tools for population genetic analyses (TFPGA) 1.3: A windows program for the analysis of allozyme and molecular population genetic data. Computer software distributed by author.
115 Mort, M. E., J. K. Archibald, and K. N. Fairfield. 2003. Phylogenetics of Crassula (Crassulaceae) inferred from DNA sequence data. Presentation at Evolution, Chico, CA.
Mort, M. E., D. E. Soltis, P. S. Soltis, J. Francisco-Ortega, and A. Santos-Guerra. 2002. Phylogenetics and evolution of the Macaronesian clade of Crassulaceae inferred from nuclear and chloroplast sequence data. Systematic Botany 27: 271-288.
Nei M. and W-H Li. 1979. Mathematical model for studying genetic variation in terms of restriction endonucleases. Proceedings of the National Academy of Sciences of the United States of America 76: 5269-5273.
Neuffer, B., H. Auge, H. Mesch, U. Amarell, and R. Brandl. 1999. Spread of violets in polluted pine forests: morphological and molecular evidence for the ecological importance of interspecific hybridization. Molecular Ecology 8: 365-377.
Olmstead, R. G. and J. A. Sweere. 1994. Combining data in phylogenetic systematics: an empirical approach using three molecular data sets in the Solanaceae. Systematic Biology 43: 467-481.
Page, R. D. M. 1996. TREEVIEW: an application to display phylogenetic trees on personal computers. Computer Applications in the Biosciences 12: 357-358.
Rambaut, A. 1996. Se-Al sequence alignment editor, version 1.0a1. Department of Zoology, University of Oxford, Oxford, UK.
Raymond, M. and F. Rousset. 1995. An exact test for population differentiation. Evolution 49: 1280-1283.
Richardson, J. E., F. M. Weitz, M. F. Fay, Q. C. B. Cronk, H. P. Linder, G. Reeves, and M. W. Chase. 2001. Rapid and recent origin of species richness in the Cape flora of South Africa. Nature 412: 181-183.
Rieseberg, L. H. 1995. The role of hybridization in evolution: old wine in new skins. American Journal of Botany 82: 944-953.
Rieseberg, L. H. and L. Brouillet. 1994. Are many plant species paraphyletic? Taxon 43: 21-32.
Rieseberg, L. H. and J. F. Wendel. 1993. Introgression and its consequences in plants. In R. G. Harrison, ed. Hybrid zones and the evolutionary process. 70-109. Oxford University Press, Oxford, UK.
Rohlf, F. J. 1998. NTSYS-pc. Numerical taxonomy and multivariate analysis system, Version 2.0. Setauket, New York: Exeter Software.
116 Rokas, A., G. Melika, Y. Abe, J.-L. Nieves-Aldrey, J. M. Cook, and G. N. Stone. 2003. Lifecycle closure, lineage sorting, and hybridization revealed in a phylogenetic analysis of European oak gallwasps (Hymenoptera: Cynipidae: Cynipini) using mitochondrial sequence data. Molecular Phylogenetics and Evolution 26: 36-45.
Runyeon-Lager, H. and H. C. Prentice. 2000. Morphometric variation in a hybrid zone between the weed, Silene vulgaris, and the endemic, Silene uniflora ssp. petraea (Caryophyllaceae), on the Baltic island of Öland. Canadian Journal of Botany 78: 1384-1397.
Sang, T. and Y. Zhong. 2000. Testing hybridization hypotheses based on incongruent gene trees. Systematic Biology 49: 422-434.
Seelanan, T., A. Schnabel, and J. F. Wendel. 1997. Congruence and consensus in the cotton tribe (Malvaceae). Systematic Botany 22: 259-290.
Siddall, M. E. 1997. Prior agreement: arbitration or arbitrary? Systematic Biology 46: 765-769.
Simmons, M. P. and H. Ochoterena. 2000. Gaps as characters in sequence-based phylogenetic analyses. Systematic Biology 49: 369-381.
Smith, G. F., A. E. van Wyk, L. A. S. Johnson, and B.-E. van Wyk. 1996. Southern African plant systematics: needs, priorities and actions. South African Journal of Science 92: 314-320.
Smith, J. F., C. C. Burke, and W. L. Wagner. 1996. Interspecific hybridization in natural populations of Cyrtandra (Gesneriaceae) on the Hawaiian Islands: evidence from RAPD markers. Plant Systematics and Evolution 200: 61-77.
Soltis, D. E. and P. S. Soltis. 1997. Phylogenetic relationships in Saxifragaceae sensu lato: a comparison of topologies based on 18S rDNA and rbcL sequences. American Journal of Botany 84: 504-522.
Soltis, D. E. and P. S. Soltis. 1998. Choosing an approach and an appropriate gene for phylogenetic analysis. In D. E. Soltis, P. S. Soltis, and J. J. Doyle, eds. Molecular Systematics of Plants II. 1-42. Kluwer Academic Publishers, Amsterdam, Netherlands.
Soltis, P. S., D. E. Soltis, M. J. Zanis, S. Kim. 2000. Basal lineages of angiosperms: relationships and implications for floral evolution. International Journal of Plant Sciences 161: S97-S107.
117 Sota, T., R. Ishikawa, M. Ujiie, F. Kusumoto, and A. P. Vogler. 2001. Extensive trans- species mitochondrial polymorphisms in the carabid beetles Carabus subgenus Ohomopterus caused by repeated introgressive hybridization. Molecular Ecology 10: 2833-2847.
Stebbins, G. L. 1950. Variation and Evolution in Plants. Columbia University Press, New York, NY, USA.
Stebbins, G. L. 1970. Adaptive radiation of reproductive characteristics in angiosperms, I: pollination mechanisms. Annual Review of Ecology and Systematics 1: 307-326.
Steen, S. W., L. Gielly, P. Taberlet, and C. Brochmann. 2000. Same parental species, but different taxa: molecular evidence for hybrid origins of the rare endemics Saxifraga opdalensis and S. svalbardensis (Saxifragaceae). Botanical Journal of the Linnean Society 132: 153-164.
Stevens, P. F. 1991. Character states, morphological variation, and phylogenetic analysis: a review. Systematic Botany 16: 553-583.
Straw, R. M. 1966. A redefinition of Penstemon (Scrophulariaceae). Brittonia 18: 80- 95.
Sullivan, J. 1996. Combining data with different distributions of among-site rate variation. Systematic Biology 45: 375-380.
Swofford, D. L. 2003. PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods). Version 4. Sinauer Associates, Sunderland, MA.
Taberlet, P., L. Gielly, G. Pautou, and J. Bouvet. 1991. Universal primers for amplification of three non-coding regions of chloroplast DNA. Plant Molecular Biology 17: 1105-1109.
Takhtajan, A. 1986. Floristic Regions of the World. University of California Press, Berkeley, CA, USA.
Thompson, J. D., T. J. Gibson, F. Plewniak, F. Jeanmougin, and D. G. Higgins. 1997. The Clustal X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Research 25: 4876-4882.
Tiffin, P., M. S. Olson, and L. C. Moyle. 2001. Asymmetrical crossing barriers in angiosperms. Proceedings of the Royal Society of London Series B, Biological Sciences 268: 861-867.
118 Tucker, S. C. 1996. Trends in evolution of floral ontogeny in Cassia sensu stricto, Senna, and Chamaecrista (Leguminosae: Caesalpinioideae: Cassieae: Cassiinae); a study in convergence. American Journal of Botany 83: 687-711. van Oppen, M. J. H., B. J. McDonald, B. Willis, and D. J. Miller. 2001. The evolutionary history of the coral genus Acropora (Scleractinia, Cnidaria) based on a mitochondrial and a nuclear marker: reticulation, incomplete lineage sorting, or morphological convergence? Molecular Biology and Evolution 18: 1315-1329.
Wen, J. and E. A. Zimmer. 1996. Phylogeny and biogeography of Panax L. (the ginseng genus, Araliaceae): inferences from ITS sequences of the nuclear ribosomal DNA. Molecular Phylogenetics and Evolution 6: 167-177.
Wendel, J. F. and J. J. Doyle. 1998. Phylogenetic incongruence: window into genome history and molecular evolution. In D. E. Soltis, P. S. Soltis, and J. J. Doyle, eds. Molecular Systematics of Plants II. 265-296. Kluwer Academic Publishers, Boston, MA, USA.
Weller, S. G. and A. K. Sakai. 1999. Using phylogenetic approaches for the analysis of plant breeding system evolution. Annual Review of Ecology and Systematics 30: 167-199.
Wheeler, Q. D. and K. C. Nixon. 1990. Another way of looking at the species problem: a reply to de Queiroz and Donoghue. Cladistics 6: 77-81.
Wolf, D. E., N. Takebayashi, and L. H. Rieseberg. 2001. Predicting the risk of extinction through hybridization. Conservation Biology 15: 1039-1053.
Wolfe, A. D. and W. J. Elisens. 1995. Evidence of chloroplast capture and pollen- mediated gene flow in Penstemon sect. Peltanthera (Scrophulariaceae). Systematic Botany 20: 395-412.
Wolfe, A. D., S. L. Datwyler, and C. P. Randle. 2002. A phylogenetic and biogeographic analysis of the Cheloneae (Scrophulariaceae) based on ITS and matK sequence data. Systematic Botany 27: 138-148.
Wolfe, A. D. and A. Liston. 1998. Contributions of PCR-based methods to plant systematics and evolutionary biology. In D. E. Soltis, P. S. Soltis, and J. J. Doyle, eds. Molecular Systematics of Plants II. 43-86. Kluwer Academic Publishers, Boston, MA, USA.
Wolfe, A. D., Q.-Y. Xiang, and S. R. Kephart. 1998. Assessing hybridization in natural populations of Penstemon (Scrophulariaceae) using hypervariable intersimple sequence repeat (ISSR) bands. Molecular Ecology 7: 1107-1125.
119