|USFDC Home | USF Electronic Theses and Dissertations||| RSS|
This item is only available as the following downloads:
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001790596
007 cr mnu|||uuuuu
008 070621s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001513
Do, Kimberly Fearn.
A determination of phylogeny and hybridization history within Clematis L. (Ranunculaceae) using actin and nitrate reductase intron sequences
h [electronic resource] /
by Kimberly Fearn Do.
[Tampa, Fla] :
b University of South Florida,
ABSTRACT: The phylogeny of Clematis, section Viorna, was characterized in this study using molecular data. Two nuclear introns were sequenced for a variety of taxa: actin and nitrate reductase. Actin intron sequence data yielded very little phylogenetic information. Some basal clades were resolved, but there were very few well supported relationships between species of the Viorna section in both the neighbor joining and maximum parsimony analyses. Nitrate reductase intron sequence data was slightly more variable. The number of well supported relationships in both the neighbor joining and maximum parsimony analyses for nitrate reductase was greater, but still not sufficient to yield an informative tree. Two possible explanations for the lack of variation are that these species have not evolved many differences in these intron sequences or that common alleles are flowing between the species. Hybrid analysis using the actin intron was inconclusive because the experimentally generated hybrid possessed an allele that neither parents tested had. More sampling from multiple individuals from both parent and multiple hybrid individuals is necessary to answer this question. The hybrid specimen tested was homozygous for the nitrate reductase intron marker, and both parents also possessed the allele. This did not directly support or refute the use of these markers for tracking the hybridization history within Clematis.
Thesis (M.A.)--University of South Florida, 2006.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 45 pages.
Adviser: Frederick B. Essig, Ph.D.
t USF Electronic Theses and Dissertations.
4 0 856
A Determination of Phylogeny and Hybridization History Within Clematis L. (Ranunculaceae) Using Actin and N itrate Reductase Intron Sequences by Kimberly Fearn Do A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science Department of Biology College of Arts and Sciences University of South Florida Co-Major Professor: Frederick B. Essig, Ph.D. Co-Major Professor: James R. Garey, Ph.D. Richard P. Wunderlin, Ph.D. Date of Approval: April 10, 2006 Keywords: plant, molecular, DNA, clade, allele, Copyright 2006, Kimberly Fearn Do
i Table of Contents List of Tables ii List of Figures iii Abstract iv Chapter One: Introduction 1 The Genus Clematis 1 Previous Classifications 3 Actin Intron Marker 4 Nitrate Reductase Intron Marker 5 Characterizing Hybrid History 6 Objective of Current Study 8 Chapter Two: Materials and Methods 9 Specimen Collection and Preservation 9 Genomic DNA Extraction 9 Polymerase Chain Reaction (PCR) 9 PCR Purification 10 Cloning and Plasmid Extraction 11 Restriction Digestion of Plasmid DNA 12 Cycle Sequencing of Intron Markers 12 Analysis 13 Chapter Three: Results 14 Actin Data Analysis 14 Unsuccessful Marker Attempts 18 Nitrate Reductase Data Analysis 19 Combined Actin and Nitrate Reductase Data Analysis 24 Chapter Four: Discussion 29 Actin Intron Topics 29 Nitrate Reductase Intron Topics 36 Combined Actin and Nitrat e Reductase Intron Topics 40 Conclusions 41 Literature Cited 43
ii List of Tables Table 1. List of described sp ecies included in the study. 2 Table 2. List of substitutions and indels for the hybrid and parent alleles of actin intron sequences. 18 Table 3. List of substitutions and indels for the hybrid and parent alleles of nitrate reductase sequences. 24
iii List of Figures Figure 1. Neighbor joining, e volutionary distance tree for the actin intron sequence data. 16 Figure 2. Maximum parsimony tree for the actin intron sequence data. 17 Figure 3. Neighbor joining, e volutionary distance tree for the nitrate reductase intron sequence data. 21 Figure 4. Maximum parsimony tree for the nitrate reductase intron sequence data. 22 Figure 5. Neighbor joining, evol utionary distance tree for the concatenated sequence da ta set of both actin and nitrate reductase intron sequences. 26 Figure 6. Maximum parsimony tree for the concatenated sequence data set of both actin and nitrat e reductase intron sequences. 27
iv A Determination of Phylogeny and Hybridization History Within Clematis L. Section Viorna (Ranunculaceae) Using Actin and N itrate Reductase Intron Sequences Kimberly Fearn Do ABSTRACT The phylogeny of Clematis section Viorna was characterized in this study using molecular data. Two nuclear introns were se quenced for a variety of taxa: actin and nitrate reductase. Actin intron sequence data yielded very little phylogenetic information. Some basal clades were resolved, but ther e were very few well supported relationships between species of the Viorna section in both the nei ghbor joining and maximum parsimony analyses. Nitrate reductase intron se quence data was slightly more variable. The number of well supported relationships in both the neighbor joining and maximum parsimony analyses for nitrate reductase was gr eater, but still not sufficient to yield an informative tree. Two possible explanations for the lack of variation are that these species have not evolved many differences in these in tron sequences or that common alleles are flowing between the species. Hybrid analysis using the actin intron was inconclusive because the experimentally generated hybrid possessed an allele that neither parents
v tested had. More sampling from multiple i ndividuals from both parent and multiple hybrid individuals is necessary to answer this question. The hybrid specimen tested was homozygous for the nitrate reductase intron marker, and both parents also possessed the allele. This did not directly support or refu te the use of these ma rkers for tracking the hybridization history within Clematis
1 Chapter One INTRODUCTION The Genus Clematis Clematis is a major member of the family Ranunculaceae, known for its basal placement within the group of eudicots in the tree of life (Zomlefer, 1995). The Ranunculaceae still retain some of the characteristics of the an cestral eudicots such as an apocarpous gynoecium. Although it is know n that the Ranunculaceae form a basal relationship within the eudi cots, phylogenetic relationshi ps have not been clearly resolved within the family. This study attempts to clarify some of the relationships within the genus Clematis Clematis distributed world wide, is comprised of about 320 species in 19 sections (Johnson, 2001). The section of inte rest in the current study is Viorna Prantl, which consists of six subsections and 23 species. Nine of the species in this section have been included in this study representing all but one of the subsections (Table 1).
2 Table 1 List of described species included in the study. The au thor of the species name and the section of the species as classified by Tamura (1968) and modified by Johnson (2001) as well as voucher numbers included here. No vouchers were available for C. ochroleuca C. fremontii C. fusca. These species were purchased from nurseries and never flowered before dying. Identity of these can not be irrefutably verified. C. socialis samples were taken from two populations both of which are represented here. The identity of C. jackmanii, even with a vouc her specimen, can also not be irrefutably verified. Species Name Author Tamura's Section Voucher Number (USF) Clematis baldwinii Torr. & A. Gray Viorna Essig 990406-1 Clematis glaucophylla Small Viorna Essig 011001-6 Clematis reticulata Walter Viorna Arias 71 Clematis crispa L. Viorna Essig 981207-2 Clematis integrifolia L. Viorna Essig 011001-2 Clematis socialis Kral Viorna Gasden CS-58 Clematis socialis Kral Viorna Laney CS-97 Clematis ochroleuca Aiton Viorna Clematis fremontii S. Watson Viorna Clematis fusca Turcz. Viorna Clematis campaniflora Lodd. ex Steud Viticella Essig 040824-2 Clematis jackmanii x ? T. Moore Viticella Essig 040824-1 Clematis lasiandra Maxim. Connatae Essig 040824-3 Clematis drummondii Torr. & A. Gray Clematis Gonzalez 7008 Clematis catesbyana Pursh Clematis Hart 10-18-90 Clematis brachyura Maxim. Pterocarpa Essig 020305-2 Clematis terniflora DC. Flammula Essig 890904-1
3 Previous Classifications There have been several propos ed classification schemes for Clematis The most widely accepted of these was proposed by Ta mura (1968, 1989) and was based solely on morphological characters. Tamuras morphologica l classification was then revised in by Johnson (1997, 2001). The division of Clematis into sections and subsections tested in this study is based on this classification. E ssig (1991) proposed a different classification based on seedling morphology. In this system Essig proposed two divisions: Type I seedlings characterized by a lternate phyllotaxy, absence of cataphylls, toothed eophyll margins, and an elongate hypocotyl, and T ype II seedlings characterized by opposite phyllotaxy, presence of cataphylls, entire eophyll margins, and a compact hypocotyl. Essigs system is incongruent with that of Tamura. In more recent years classificat ion schemes have shifted from using morphological data to molecular data. Nucleo tide sequence divergence has been a proven method for resolving phylogenetic questions resulting from morphological data analysis (Hoot and Palmer 1994; Cros et al. 1998; So ltis and Kuzoff 1995; Hardig et al. 2000). Both chloroplast and nuclear genomes have been utilized for these molecular analyses. These genomes are useful for different appli cations depending on th e divergence level of the taxa in question. Chloroplast markers are useful for resolving deeper level (generic and familial) relationships. The chloroplast genome has a slower rate of nucleotide substitution than the nuclear genome of plants because of the nature of the genes carried on that genome (Hillis et al. 1996). For exam ple, there are 37 tRNA coding genes carried on the chloroplast genome; most mutations to any of these genes would be deleterious to their function (Li 1997).
4 Miikeda et al. (1999) published a pre liminary molecular classification of Clematis using a variety of coding and non-c oding chloroplast genome markers. Although over 4000 base pairs were sequenced, the analysis only produced a poorly resolved molecular evidence tree. For speci es level phylogenetic questions, a nuclear marker is more appropriate. Specifically, a nuclear non-coding marker, such as an intron, would have a relatively fast substitution rate yielding sufficient variability to resolve some relationships (Li 1997). Recently, there have been many advances in the development of primers that amplify these hi ghly variable nuclear introns. Strand et al. (1997) developed a methodological approach to finding informative nuclear characters. Using this procedure, Slomba et al. (2004) developed primers that amplify an intron of the nuclear gene actin. Actin Intron Marker The actin protein is an important elem ent of the plant cy toskeleton. Actin is important in cellular expansi on during growth, as well as m ovement of organelles and vesicles around the cell (Mathur and Hlskamp 2002; Mathur 2004). There is also some indication that actin is involved in the gravitropic res ponse that plants display. Data suggests that statolith position is also affect ed by an actin myosin complex (Friedman et al. 2003). Actin may be a useful marker because preliminary analysis shows that the nonsynonymous rate of substitution is up to 19 tim es higher in angiosperm actin genes than mammalian or fungal actin genes (Moniz de S and Drouin 1996). This increased rate of
5 substitution may yield sufficient phylogene tically informative sites to resolve relationships between species hypothesized to be classi fied in the same section. Using taxa chosen from across the genus Slomba (2004), was able to develop a working hypothesis of the phylogeny of Clematis using the actin intron. He found two distinct clades that correspond to the two s eedling morphology types of Essig (Slomba et al. 2004). It is hoped that expanded taxon sampli ng that encompasses a representation of the Viorna section may improve what is known about these relationships. Nitrate Reductase Intron Marker Although one marker may result in a phylogenetic hypothesis, multiple markers yielding the same phylogeny further support th e hypothesis. Therefore, an additional nuclear intron was utilized in this stu dy. Howarth and Baum (2002) developed primers for the second and third introns of nitrate redu ctase. They were successful in amplifying this intron in multiple species of Scaevola (Goodeniaceae). For Howarth and Baum, sequences of this intron were particularly us eful because, in multiple genera studied, the size of the intron was large enough to yield much variability and many phylogenetically informative characters (from 85 to 1600 bp). Nitrate reductase is a low copy number, nuclear en coded gene essential to nitrogen metabolism in plants. Nitrate reduc tase catalyzes the first step of nitrate assimilation where nitrate is reduced to ni trite (Sharma and Dube y 2005). Nitrate is the principal form of nitrogen available to plant r oots in fertilizers and the action of nitrogen fixing bacteria. Nitrate must then be reduced into forms that can be used to make amino acids. Nitrate assimilation takes place in the leaves where light controls the activity of
6 nitrate reductase. This is because the reduction of nitrate to nitrite is closely coupled to both photosynthetic and respiratory reactions (Foyer et al. 2003). Combined data from both actin and nitr ate reductase intron sequences may yield more insights into the phylogeny of this subset of the genus Clematis These intron sequences may be more informative than simply to resolve phylogenetic questions; it may be also possible to track the histor y of hybrid species using these markers. Clematis is a common garden ornamental, which gi ves this genus scientific and economic importance. One of the reasons that Clematis is such a popular garden plant is the relative ease of hybridization within the genus, esp ecially between closel y related species. Hybridization allows plant en thusiasts to manipulate morphol ogical characters to their liking. For example, there are 14 different cultivars of Clematis that originated from hybridization between C. texensis and other large flowered species (Johnson, 2001). This ease of artificial hybridiza tion leads to the hypothesis th at many naturally occurring species may be the result of na tural hybridization events. Characterizing Hybrid History According to Rieseberg (1997), hybrid s are organisms formed by cross fertilization between two indi viduals of two populations dist inguishable from one another by one or more heritable characters. Often, popul ations that interbreed to form a hybrid originate from two different species. The evolutionary influen ce of hybridization to speciation has likely been underestimated b ecause of difficulties in detecting hybrid origins (Rossetto 2005). Difficulties arise in detecting hybrids, as well as in including hybrid species in phylogenetic analyses, because these species result from a reticulation
7 event rather than a bifurcation (Koontz et al. 2004). Before molecular techniques were available, intermediacy in morphological characters was the primary method for detecting a hybrid. Morphologica l evidence of hybrid origins can be very compelling, but it can also be misleading. Intraspecific varia tion may falsely indicate the presence of a hybrid or it may mask the ex istence of a true hybrid popul ation or species. Recently, molecular tools have been utilized for dete rmining the origins of putative hybrid species (Aguilar et al. 1999; Ferguson and Sang 2001; Franzke and Mu mmenhoff 1999; Koontz et al. 2004; Rossetto 20 05; Zhou et al 2005). Popular molecular markers for botani cal studies have originated on the chloroplast genome. The paucity of variable sites on this genome makes it inadequate for differentiating haplotypes at very fine ta xonomic levels like th e species and population level (Rossetto 2005). Chloroplast markers ar e also not sufficient for hybrid studies because of the maternal inhe ritance of this ge nome. Koontz et al. (2004) used the cpDNA marker trnl-F to try to resolve the hybrid origins of Delphinium gypsophilum Ewan (Ranunculaceae) in conjunction with an ITS rDNA nuclear marker. These two markers were used because they represented genes with different modes of inheritance. The chloroplast marker is maternally inherited, while the nuclear marker is biparentally inherited. The hope of the study was that th e chloroplast marker would identify the maternal parent of the putative hybrid and group in a clade with that maternal parent. However, even using the chloroplast marker in association with a nuclear marker could not completely resolve the hybrid relationshi ps due to lack of variation. Since intron regions have a faster substitution rate than the chloroplast genome, using multiple nuclear
8 markers may make it possible to differentiate haplotypes of even recently diverged species (Li 1997). Objectives of Current Study The current study strives to reso lve the phylogeny of the section Viorna of the genus Clematis using molecular intron markers. Anot her objective, also using molecular data, is to characterize the hybr id history of species within Viorna focusing on an experimentally generated hybrid.
9 Chapter Two MATERIALS & METHODS Specimen Collection and Preservation Specimens were obtained from multiple s ources. Fresh leaf material was collected from the yard of Dr. Frederick Essig (cul tivated specimens, natural population unknown), the University of South Florida Botanical Gardens (cultivated specimens, natural population unknown), and Lettuce Lake Park in Tampa (natural population). Leaf material was examined under a dissecting scope to remove parasites and other foreign organic matter. Leaf material was then ho mogenized using liquid nitrogen and a mortar and pestle. Ground samples were stored in microcentrifuge tubes at -80 C. Genomic DNA Extraction Genomic DNA was extracted from gr ound leaf samples using the Plant DNA Isolation Kit (13000-50) from MO BIO (S olana Beach, CA). The kit bound DNA to a filter spin column to purify it from other plant cell contents. DNA was then eluted from the membrane in the spin column with an elution buffer and stored at -20C.
10 Polymerase Chain Reaction (PCR) Polymerase chain reaction (PCR) was used to amplify desired genetic markers for all extracted DNA samples using a Biometra T3 thermal cycler (#1510303). Primers used for the various gene regions were manufact ured by IDT (Integrated DNA Technologies Coalville, IA). Actin intron forward and re verse primers used were published by Slomba (2004). Nitrate reductase third intron forwar d and reverse primers used were published by Howarth and Baum (2002). PCR amplifications were set up in 200 l thin walled, hinge capped microcentrifuge tubes to a final volume of 50 l and consisted of 38.75 l molecular biology grade water, 5 l 10X Reaction Buffer (ID Lab Inc., London, ON, Canada), 1 l dXTP mix (1:1:1:1 of 12.5 mM dATP, dGTP, dCTP, and dTTP), 0.25 l high fidelity ID Proof DNA Polymerase (ID Lab Inc., London, ON, Canada), 1 l forward and reverse primer at 10 M, 2 l of 25mM MgCl2, and 1 l template DNA (0.25 g/ l). Thermal cycling parameters for the actin intron were a 2 min. initial denature at 95 C followed by 35 cycles of 15 sec. at 95 C, then a 48 C annealing temp. for 30 sec. with a 72 C extension for 1 min. 30 sec. followed by a 10 min. final extension at 72 C. The nitrate dehydrogenase third intron used touch down thermal cycling parameters. PCR Purification All PCR prducts were purified to re move excess primers (primer-dimers) and unincorporated dNTPs before cloning. Purifi cation was carried out using the QIAquick Gel Extraction Kit from Qiag en (Valencia, CA). The full volume of sample was electrophoresed onto a 0.9% ag arose gel at 115V. DNA bands representing the desired
11 amplicon were then visualized with a longw ave UV light source and excised from the gel into microcentrifuge tubes. The Qiagen kit dissolved the agarose gel and separated the DNA using a spin filter method. DNA was th en eluted off the membrane with 30 l of molecular biology grade water (MBW). Cloning and Plasmid Extraction In order to test for the presence of mu ltiple alleles of these introns, PCR product was cloned into One Shot Mach1-T1R chemically competent Escherichia coli cells (Invitrogen; Carlsbad,CA) using a Topoisome rase TA Cloning Kit (Invitrogen; Carlsbad, CA). After 18 hours of incubation at 37C, colonies that grew on LB agar with ampicillin (50g/ml) were those that contained an insert in the plasmid vector These colonies were then picked off the agar plates and grown in LB broth on a microtitre plate and incubated overnight at 37 C (170 l LB broth per well with ampicillin (50g/ml)). One drop of glycerol was added to each well prior to being stored at -80 C. Once twelve specimens had viable colonies, a 96 we ll culture plate was inoculated with stored clones; each specimen represented by one row on the plate that was then incuba ted overnight at 37 C. Plasmid DNA containing the amplicon of intere st was then extracted from the bacterial cells using the Perfectprep Plasmid 96 V ac, Direct Bind Kit (Eppendorf, Brinkmann Instruments, Inc; Westbury, NY). Final eluti on of DNA off of the filter plate was done with 50 l of molecular biology grade water rather th an the specified elution buffer. It was also evident that ethanol was not removed sufficiently using the Perfectprep kit, therefore plates were also completely dried us ing a Savant SpeedVac Plus. DNA was then resuspending in 50 l of MBW.
12 Restriction Digestio n of Plasmid DNA To verify presence of proper insert in plasmid DNA preparations, restriction enzymes were used to cut the insert out of the plasmid. Reactions consisted of 14 l of MBW, 2 l plasmid DNA, 2 l EcoRI buffer (Promega; Madison, WI), 1 l EcoRI enzyme (Promega; Madison WI), and 1 l of BSA (bovine serum albumin)(Promega; Madison, WI). Reactions were incubated at 37 C for 2 hours; 10 l of each reaction was then eletrophoresed onto a 0.9% agarose gel run at 115V. The resulting image separated the false positive clones from those with an insert, and also removed those with an incorrectly sized insert. This also allowe d for gel quantification of the amount of amplicon present in the sample. Cycle Sequencing of Intron Markers Sequencing was done on a Beckman Cou lter CEQ8000 Genetic Analysis System using the GenomeLab Dye Terminator Cycle Sequencing with Quic k Start Kit (Beckman Coulter; Fullerton, CA). Reactions were ac tually run at quarter strength of that recommended by Beckman with a final volume of 10 l. For most samples 4 l of template plasmid DNA had the appropriate concentration for sequencing. Template DNA was combined with 2 l of MBW and incubated at 96 C for 1 min. and then transferred to ice to uncoil the plasmid and increase primer binding efficiency. The preheated template DNA and MBW were combined with 2 l of plasmid specific primer at 3.2 M and 2 l of DTCS Quick Start Master Mix (Beckman Coulter; Fullerton, CA). Sequencing parameters for this reaction were 40 cycles of 96C for 20 sec. followed by 50C for 20 sec. and 60C for 4 min. The reaction could then be held indefinitely at 4C. Once the
13 thermal cycling program was finished, dye term inated fragments were then precipitated and purified. A solution was prepared fresh c onsisting of 2 parts 3M Sodium Acetate (pH 5.2), 2 parts 100mM Na2-EDTA(pH 8.0), and 1 part 20 mg/ml glycogen; 5 l of this solution was added to each sequence in addition to 60 l of cold 100% ethanol. The mixture was then vortexed and centrifuged at 4 C for 15 min at 13,000 rpm. This should result in pelleted DNA; supernatant was re moved and discarded. Pellet was rinsed twice with cold 70% ethanol and centrifuged for 2 mi n. at 4C. After each rinse supernatant was removed and discarded. Pellets were then drie d for 4 minutes on low heat in the Savant SpeedVac Plus, and finally resuspended in 40 l of sample loading solution. Analysis Phylogenetic inferences were made from DNA sequence alignments generated by Clustal W (Thompson, et al. 1994; Higgins et al. 1996) and modified by multiedit (American Cybernetics, AZ). Nieghbor join ing evolutionary distance trees were developed using MEGA 3.1 (Kumar et al. 2004).
14 Chapter Three RESULTS Actin Data Analysis Actin intron sequence results were analy zed using a neighbor joining tree building algorithm. Each allele recognized for ever y species was included in the analysis. Numbers of alleles per species varied between two and one. Actin intron sequences yielded a tree with very little reso lution of relationships in the selected sampling of species. One of th e only stable relations hips was the grouping together of both alleles of C. terniflora with 99% bootstrap suppor t (Letter A of Figure 1). There was also 99% bootstrap support fo r a relationship where the first allele of C. catesbyana was a sister group to a large clade co mprised of all other taxa except C. terniflora (Letter B). A sister group relati onship between the first allele of C. reticulata and a larger clade comprised of all taxa expect alleles of C. terniflora and the first allele of C. catesbyana is also supported with a bootstrap value of 63 (Letter C, Figure 1). The only other well supported clade is a groupi ng together of the first allele of C. crispa with the only allele of C. fusca for the actin intron sequences (bootstrap value 66, Letter D). Beyond the relationships described above, th ere is no other resolution in the tree of Figure 1. In fact there is a large polyto my including all the members of the section Viorna ; however, this polytomy does not only include member of Viorna Surprisingly,
15 C. drummondii is a part of this polytom y. According to Essig (1991), C. drummondii is very different from members of the section Viorna because it is characterized by the type II seedling. A maximum parsimony analysis was also performed on the actin intron sequence data set (Figure 2). Except for relationships involving C. terniflora and the first allele of C. catesbyana, there is no resolution in this tree. Th is is a direct result of the variation being concentrated in the sequences from C. terniflora and C. catesbyana There is only one parsimony informative site in the actin data set when those taxa are removed, while there are 20 parsimony informative sites wh en all taxa are included. However, the relationships that are formed in the maximu m parsimony analysis also appear as well supported relationships in the neighbor joining analysis. As can be expected with very little support for branches, members of different sections based on the most recent Johnson ( 2001) classification are scattered throughout the tree. The section Flammula represented by C. terniflora is the only section supported by this analysis. Although there was very little support for the neighbor jo ining tree re sulting from the actin intron data, this data was still informative in diagnosing hybrid species. The experimentally generated hybrid was a cross between C. crispa and C. reticulata Therefore, it was hypothesized that the hybrid sequences w ould group into two alleles; and that one allele would correspond with C. crispa and the other allele would correspond with C. reticulata
162catesbyan a fremontii Lsocialis lasiandra 1ochroleuca x 2crispareticulata drummondii 2baldwinii 2ochroleuca 2reticulata 1baldwinii x 1crispareticulata jackmanii x 1crispa fusca 1integrifolia campaniflora Gsocialis glaucophylla 2brachyura 1brachyura 2crispa 2integrifolia 1reticulata 1catesbyana 1terniflora 2terniflora 99 66 63 0.01A99B C D Figure 1. Neighbor joining, evolutiona ry distance tree for the actin intron sequence data. Bootstrap values over 50 (with 500 replicates) are associated with the branches, and these relationships are labeled with letters. Tree is drawn to scale.
171baldwinii 2catesbyana 2integrifolia 2crispa reticulata1 brachyura2 glaucophylla x 1crispareticulata baldwinii2 x 2crispareticulata ochroleuca1 campaniflora 2reticulata jackmanii x 1brachyura integrifolia1 drummondii fremontii 2ochroleuca lasiandra Lsocialis Gsocialis 1crispa fusca catesbyana1 terniflora1 2 terniflor a 100 61 100 Figure 2. Maximum parsimony tree for the actin in tron sequence data. Bootstrap values over 50 (with 100 replicates) ar e associated with the br anches. Letters indicate relationships that appear and are well supported on the neighbor joining tree.
18 It was found that the sequence of the second allele of C. crispa and the second allele of C. reticulata were exactly the same as th e first allele of the hybrid C. crispa x C. reticulata (Table 2). The sequence of the sec ond allele of the h ybrid matches the sequence of the first allele of C. reticulata, but these sequences are not exactly the same. The sequence of the hybrid has an insertion of three thymine bases and a transition from a thymine to a cytosine, while the sequence of the first allele of C. reticulata has a transition from a guanine to an adenine (Table 2). However, these sequences do share other characteristics; they bot h have an insertion of one t hymine base at the same locus and another insertion of two thymine bases. Table 2 List of substitutions and indels for the hybrid and parent alle les of actin intron sequences. Nucleotide Position Species/Allele DesignationFrequency 37426883 84 85 8687149176 C. crispa1 A 2 G C C T C. crispa2 B 4 G T T T C. crispa x C. reticulata 1 B 4 G T T T C. crispa x C. reticulata 2 C 2 T G T T T T T T T C C. reticulata 1 D 2 T A T T T T T C. reticulata 2 B 3 G T T T Unsuccessful Marker Attempts Before a successful second nuclear plan t marker was found, some efforts were undertaken to sequence other markers. In trons of isocitrate dehydrogenase and glyceraldehyde 3-phosphate dehydrogenase were both attempted for collection of sequence data (Weese and Johnson 2005; Strand et al. 1997). In both cases, multiple bands resulted from PCR using published primers. Because size of introns between
19 genera is not conserved, only intron locati on is, the appropriate band (PCR amplicon) was not easily evident. There were also not consistent bands for different species. Determination of appropriate band would have required doing a test sequencing of all possible bands which is cost and labor intensiv e. However, if this was to be accomplished in the future, these two markers ma y yet be phylogenetically useful. Nitrate Reductase Data Analysis Sequences of the third intron of the nitr ate reductase (NR) gene were analyzed using a neighbor joining tree bu ilding method. Associated with the branches are bootstrap values greater than 50 based on 500 replicates (Figure 3). C. terniflora was again used as an outgroup, and each allele present in the se quences for all the species was included in the analysis. Numbers of alleles per speci es varied again between one and two. Like the actin intron data, the NR thir d intron data analysis yields a stable neighbor joining tree with some well supported relationships and othe r relationships with little to no support. C. terniflora was designated the outgroup, a nd the one allele of this species did group outside the rest of the species in th e tree. The two alleles of C. lasiandra group together with 68% bootstrap sup port (Letter A), and the two alleles of C. baldwinii also group together with 82% bootst rap support (Letter F) (Figure 3). According to this analysis, the only allele of C. campaniflora forms a sister group with C. integrifolia (86% bootstrap support, Letter H). Also, the first allele of C. socialis from the Laney population (population L) form s a sister group w ith both alleles of C. baldwinii (57% bootstrap support, Letter E). In addition, the second allele of C. socialis
20 population L forms a sister group w ith the only alle les of both C. glaucophylla and C. socialis the Gasden population (population G) (59% boot strap support, Letter I). Three more relationships in this anal ysis are supported with a 50 or more bootstrap value. Each of these is indicated by letters on Figure 3. Letter B represents a sister group relationship betw een the first allele of C. brachyura and a much larger clade consisting of all alleles except those of C. fusca C. lasiandra C. terniflora supported by a bootstrap value of 86%. Lett er C represents a sister group relationship between the C. jackmanii x allele and a clade consis ting of all allele s except those above and the first allele of C. brachyura This relationship is supported with a 65% bootstrap value. Finally, letter D represents a sister group relationship between a clade consisting of th e first allele of C. socialis population L and both alleles of C. baldwinii and a clade consisting of both alleles of C. ochroleuca C. crispa C. reticulata and the hybrid and th e second alleles of C. brachyura and C. socialis population L, and th e only alleles of C. fremontii C. drummondii C. campanifora C. integrifolia and C. socialis population G. The relationship has bootstrap support of 64%. The clade represented by the letter D is pr imarily composed of alleles of species in the Viorna section (Figure 3), howev er the relationships with in the clade have very little support with the exceptio ns at letters E and G.
21 glaucophyll a Gsocialis socialis L2 drummondii 1reticulata ochroleuca fremontii 2reticulata crispa x crispareticulata brachyura2 integrifolia campaniflora L1socialis 2baldwinii baldwinii1 jackmanii x 1brachyura 1fusca lasiandra2 1lasiandra fusca2 68 86 86 82 65 64 57 59 64 0.02ternifloraA B C D E F G H I Figure 3. Neighbor joining, evolutiona ry distance tree for th e nitrate reductase intron sequence data. Bootstrap values over 50 (with 500 re plicates) are associated with the branches, and these relationships are labele d with letters. Tree is drawn to scale.
22 2 baldwinii baldwinii1 socialis L1 x crispareticulata ochroleuca crispa 2brachyura integrifolia campaniflora reticulata2 reticulata1 L2socialis glaucophylla Gsocialis fremontii drummondii jackmanii x 1brachyura 2fusca 2lasiandra lasiandra1 1fusca terniflor a 74 64 51 70 51 94 74 68 Figure 4. Maximum parsimony tree for the nitrate reductase intron sequence data. Bootstrap values over 50 (with 100 replicates) are associated with the branches, and these relationships are labeled with letters.
23 A maximum parsimony analysis was also conducted on the nitrate reductase data set (Figure 4). This data set has more varia tion than the actin data set. There were 25 parsimony informative sites (out of 194 overall sites) when all taxa were included. When the most different taxa were excluded, ther e were still 9 parsim ony informative sites. Still, there are many more parsimony informa tive sites attributable to the few most distantly related taxa of the analysis than a ny of the closely related taxa. For this reason, there are well supported relationshi ps between less closely related taxa than between taxa in the Viorna group. Many of these relations hips remain unresolved. The maximum parsimony tree did replicat e all the relationships that were well supported in the neighbor join ing tree except one (Letter D, Figure 3). The bootstrap values on these relationships were also greater than 50. Similarly to the actin intron analysis, the neighbor joini ng tree of the third intron of NR is mostly uninformative. Unfortunately, in the case of the third intron of NR, this marker does not diagnose the parentage of the experimentally generated hybrid. Sequences from the C. crispa x C. reticulata specimen did not sort into two distinct alleles (Table 3). Wh en alleles of both C. crispa and C. reticulata were determined, it was found that they shared an allele. This shared allele was passed to the hybrid making it homozygous for this marker. Although this doe s not diagnose the hybrid parentage, the nitrate reductase marker also does not refute the hybrid hypothesis.
24 Table 3 List of substitutions for the hybrid a nd parent alleles of nitrate reductase sequences. Nucleotide Position Species/Allele DesignationFrequency 90 126 158 C. crispa A 4 A T T C. crispa x C. reticulata A 6 A T T C. reticulata 1 B 1 G C C C. reticulata 2 A 3 A T T Combined Actin and Nitrate Reductase Data Analysis Sequences of both the actin and nitrate re ductase introns were concatenated into one alignment for a final joint analysis. A neighbor joining tree was created with bootstrap values greater than 50 (based on 500 replicated) shown with the branches (Figure 5). Again, C. terniflora was used as an outgroup, and each allele of both markers was included in the analysis. Alleles for both actin and NR intron sequences were combined based on what taxa they originated from. In cases where one marker had two alleles, but the other only ha d one allele, the two unique alle les were concatenated with two sequences from one allele of th e other marker within the same taxa Analysis of the concatenated data set yi elds a tree where some relationships are highly supported while others still have little to no support. Highly supported relationships are labeled A through L on Fi gure 5. Letter A represents the highly supported relationship (bootstrap value 99) that groups the two alleles of C. terniflora together. There is also a 99% bootstrap value (B) supporti ng grouping the two alleles of C. lasiandra together. At letter C there is a boots trap value of 55 that joins the second allele of C. fusca into a sister group relationshi p with the two alleles of C. lasiandra
25 The relationship at letter D (bootst rap value 86) unites the clade of C. lasiandra and the second allele of C. fusca and the first allele of C. fusca with a larger clade comprised of all other alleles except C. terniflora There is also a highly supported relationship at letter E; the first allele of C. brachyura is a sister group to a much larger clade (bootstrap value 86). This clade encomp asses all included alleles from the section Viorna (except C. fusca ) in addition to the second allele of C. brachyura C. campaniflora and C. drummondii Letter H also represents a well supported clade that unites both alleles of C. baldwinii into a one group (bootstrap value 85); letter G shows a relationship where the first allele of C. socialis population L is joined into a sister group relationship with both alleles of C. baldwinii (bootstrap value 57). Letter F joins the C. jackmanii x allele into a sister group relationship with a larger clade including both alleles of C. crispa C. ochroleuca C. reticulata C. integrifolia and the hybrid specimen, as well as the single alleles from C. fremontii C. glaucophylla C. socialis (population G), C. campaniflora and the second alleles of both C. socialis (population L) and C. brachyura Two alleles of C. integrifolia also group together with a high level of support (95% bootstrap; letter K). These two C. integrifolia alleles form a sister group relationship with C. campaniflora (bootstrap value 73; letter J). This clade forms a sister group with the second allele of C. brachyura at letter I, with 64% bootstrap value. Finally, a clade uniting C. glaucophylla with C. socialis (population G) and C. socialis (population L) is supported with a bo otstrap value of 61% (letter L).
261reticulat a 2reticulata x 1crispareticulata fremontii drummondii ochroleuca1 glaucophylla socialis G L2socialis x 2crispareticulata crispa2 ochroleuca2 2brachyura campaniflora integrifolia1 integrifolia2 1crispa socialis L1 baldwinii1 baldwinii2 jackmanii x 1brachyura fusca1 fusca2 lasiandra1 lasiandra2 terniflora1 2terniflora 99 99 55 95 73 86 85 63 64 57 61 57 0.01A B C D E F G H I J K L Figure 5. Neighbor joining, evolutionary distance tree for the c oncatenated sequence data set of both actin and nitrate reductase intr on sequences. Bootstrap values over 50 (with 500 replicates) are associated with th e branches. Tree is drawn to scale.
27x 2 crispa reticulat a fremontii 1ochroleuca reticulata2 2crispa x 1crispareticulata ochroleuca2 reticulata1 L1socialis baldwinii1 baldwinii2 drummondii glaucophylla Gsocialis L2socialis crispa1 brachyura1 brachyura2 campaniflora 1integrifolia integrifolia2 jackmanii x fusca1 fusca2 1lasiandra lasiandra2 terniflora1 terniflor a 2 93 83 58 65 98 100 83 60 Figure 6. Maximum parsimony tree for the concatenated sequence data set of both actin and nitrate reductase intron se quences. Bootstrap values over 50 (with 100 replicates) are associated with the branches.
28 Many relationships in this tree, especial ly those among species within the section Viorna that are hypothesized to be more clos ely related, are still unresolved. Figure 5 illustrates the tree from the concatenated data set of both actin and NR intron sequences where all nodes without at least 50% bootstrap support are collapsed. Figure 5 shows how little information has actually been glean ed from the combined sequences of two intron markers in regards to determining very fine phylogenetic relationships. A maximum parsimony analysis was also conducted on the concatenated data set (Figure 6). Not all the well supported clades in the neighbor joining tr ee were replicated in the maximum parsimony tree, but many were. The most interesting clade that appears on both trees is labeled F on Figure 5 and supported by a bootstrap value of 83 on the Figure 6. This clade creates a large polytomy of mostly species from the section Viorna, but not exclusively those species. Combining the two markers does not give any more information regarding hybrid parentage. Because the hybrid is homozygous fo r the NR marker, the actin marker is the only one informative about this topic in this study. Because both putative parents also share the allele of the hybri d, the homozygous state of the hy brid for the NR marker does not refute the hybrid hypothesis. However, the NR marker also does not support the results of the actin marker.
29 Chapter Four DISCUSSION Actin Intron Topics Phylogenetic relationships were not well resolved using actin intron sequences. Very few clades were well supported. One co mmon relationship in all analyses was a clade formed with the two alleles of C. terniflora and that this clade was always a sister group to rest of the tree (most ba sal). This relationship leads to the belief that this species is the least closely related to the others included in the study. This is reasonable considering that according to Johnsons classification, based on Tamuras original classification, C. terniflora is the only species included in th e analysis that is a member of the section Flammula Flammula is comprised predominantly of Asian species, with a few European, but no species from the New World. Unlike Viorna the flowers have four tepals but are not urn-shaped or campanulate, and they do not have th e leathery texture of flowers of Viorna (Johnson 2001). One major reason that there are so few well supported clades in the actin intron sequence alignment, is that there is not e nough variation in the data to tease apart the relationships. When consideri ng all taxa, 21% of sites were variable, but only 7.5% of sites were parsimony informative leaving 13.5% of overall sites or 64% of variable sites singletons. Overall, this amount of variation seems sufficien t. However, when variable
30 sites are counted again with out C. terniflora or C. catesbyana (first allele) in the data set, there are only 7.9% variable sites, 0.3 % parsimony informative sites, and 7.5% singletons. The sharp decline in variable sites and, especially, parsimony informative sites after removing the most basal taxa shows that those taxa contain most of the variation. Therefore, it will never be possi ble to resolve relati onships among the other taxa without more data. The data would have to either be another marker with more variability or just keep adding more markers to the current data set. In the actin data set the most variab le region is a thymine rich zone where the number of T nucleotides varies widely among the taxa. The T-rich region is the location of much of the variation observed fo r this marker in the taxa sampled. However, this variation is not very suitable for phylogenetic recons tructions because it can change too easily from one individual to another re gardless of which species that individual belongs to. This region is also riddled w ith gaps making it less useful for phylogenetic analysis. Many factors contribute to the paucity of informative variation within the actin data set. Using an intron seque nce increases the amount of po tential variation because the sequence mutations are not constrained by f unction. This gives introns a higher mutation rate than either coding regi ons of nuclear DNA or chloropl ast DNA. Unfortunately, of the approximately 266 base pairs that make up the actin marker, only about 105 bp make up the intron. The rest of the sequence is made up of flanking exon regions that have slower mutations rates due to f unctional constraints. Another factor contributing to the sm all amount of observed variation, is the occurrence of one identical allele in ten different ta xa analyzed. Not surprising, C.
31 glaucophylla C. crispa C. reticulata C. baldwinii and C. integrifolia share this allele and occur throughout Florida. These species are also all memb ers of the section Viorna as proposed by Tamura and Johnson (1968, 2001). Somewhat surprising, C. brachyura and C. campaniflora also have this allele. These sp ecies are members of the sections Viticella and Pterocarpa respectively, and do not grow in cl ose vicinities either. Very suprising, C. catesbyana also possesses this allele. The other allele of C. catesbyana was one of the most variable taxa in the analysis. This was growing in a botanical garden in the vicinity of other Clematis one of which was C. baldwinii which also possessed this allele. However, Essigs seedling mor phology classification identified C. catesbyana as a having a type II seedling while C. baldwinii has a type I seedling. The ability of a type I seedling plant to share genetic material with a type II seedling pl ant is very unlikely, although technically it is hypothesi zed that the species are inte rfertile. The experimentally generated hybrid and a hybrid from Gig Harbor, Washington, C. jackmanii x, both possess this allele as well. Possession of this allele by the experimentally generated hybrid is not surprising, since both parent sp ecies also possess the allele. Possession of this allele by another hyb rid (parentage unknown) from the opposite coast is more puzzling. Obviously, other species in the Viorna group included in this study did not share this particular allele, so this sequence does not occur in all Viorna or only in Viorna This allele being so common, but inconsistent, may be the result of multiple different processes. The allele may be select ed for in some way th at is not understood, considering it is hypothesized that typical introns are not selected for or against. This also may have been the original allele of the genus Clematis, and not all species that have radiated since the advent of the genus have ha d mutations to change this allele. One more
32 possible explanation lies in the fact that, si nce mutations are random, the occurrence of the same allele in many species is the result of convergent evoluti on, but is not due to common ancestry. This last explan ation is more unlikely due to the fact the allele occurs in ten out of the 17 taxa included in the study. The possibility of the same allele occurring at random two different times seems credible. However, the probability of the same allele occurring at random ten different times out of 17 is so small it is negligible. One last possible explanation is the fact that the different species of Clematis sampled are interfertile enough that there is gene flow between them. This gene flow may cause a homogenization of the alleles present in the di fferent species. This gene flow also may have happened in the past, especially at the conclusion of the last ice age when the range of these plants was much smaller than it is now, and more of the species may have overlapped in geographic distribution. Future studies may sample many individuals of the same taxa to determine how many alleles each one has. Testing many individuals of one taxa for variation may also yield more information about whether all the members of Viorna actually do have the common allele in their gene pool. This anal ysis was conducted with just one individual for most taxa; if more individuals were teste d, the common allele may have been a part of a different individuals genome, even though it was not in the first individuals genome. One possible conclusion that can be made from the lack of variation, even in an intron sequence, is that the time since divergence for this group of taxa is very small. In other words, there has not been sufficient tim e for all the different taxa to accumulate mutations for all unique alleles.
33 Another possible conclusion th at can be made from the lack of variation is that, especially for closely related species, reticu late evolution (hybridi zation) has clouded the signal of the molecular data. If different species are not completely reproductively isolated, as many plants are not, gene flow is possible between species to varying degrees. In this way common alleles can be pa ssed between species almost as if closely related species are one large population. Using the actin intron marker to diagnos e hybrid species may not be as clear cut as once hypothesized. When the expe rimentally generated hybrid ( C. crispa x C. reticulata ) actin intron sequences were first anal yzed, there were clearly two alleles present. Two different alleles first support the use of this marker to diagnose hybrids and determine their origin. However, one of th e two alleles is the very common allele. Unfortunately, C. crispa and C. reticulata both possess the common allele as well. Therefore, it is not immediately clear which pa rent species contributed that allele to the hybrid. When the other hybrid allele is consid ered, it is not found to exactly match any of the other parent alleles. There are many possible explanations as two why the second hybrid allele does not match any other parent allele. Either the second hybrid allele was contributed by C. crispa and one or both underwent further mutations so that the two sequences no longer match, or the second hybrid allele was contributed by C. reticulata and one or both mutated. Another possibility is that an individual of C. crispa or C. reticulata does have an allele that exactly matches that of the hybr id, but this was not sequenced in this study. The experimentally generated hybrid was deve loped more than ten years ago, which does provide for enough time for mutations to occu r in the hybrid or the putative parents
34 which would cause the sequences of this ma rker to no longer match. However, the chance that three separate mutations all accumulate d in such a short time is very small. Another explanation is that, since the hybrid does not exist in a large population of similar hybrid taxa, but ex ists in a small area where many other species of Clematis grow, there may have been further hybridi zation with another taxa at a later unknown time. This unknown taxa may have contributed th e second allele sequenc ed in this study. The most likely explanation is that because the exact same plants that created the hybrid were not sampled, only plants of the sa me species, the exact same allele that was contributed to the hybrid was not sequenced. One more complicating factor is the o ccurrence of the exact same allele, except for one transition, in C. ochroleuca This coincidence must be the result of convergent evolution, not common ancestr y, because at no point was C. ochroleuca growing in the vicinity of the experimentally generated hybrid. Therefore, C. ochroleuca could not have contributed its allele to the hybrid. Howe ver, if this hybrid was truly from unknown origin, sequence data woul d certainly suggest that C. ochroleuca was a parent species. It is impossible at this time to determ ine the history of the hybrid using the available data. Because one of the alleles present is shared by both parent species, determining which parent contributed that mark er is impossible. Because the other allele is not matched exactly by either parent, this also is impossible to trace. Even though, the first allele of C. reticulata only differs from the second hybrid allele in three ways, it is highly unlikely that there were three mutations from the r ecent hybridization event to present. The C. reticulata allele has one transition from a guanine to an adenine which no other allele shares. This substitution may have occurred after C. reticulata contributed the
35 allele to the hybrid. The second hybrid allele also has one transition substitution that is shared by no other allele. This transition fr om a thymine to a cytosine may also have occurred after the inception of the hybrid. The last difference is in the number of thymine nucleotides in a very T-rich ar ea of the intron. In the C. reti culata allele there are nine thymine bases in a row, while in the hybrid there are 12. The most likely mechanism for this disparity in length of T -rich region is replication sli ppage. Replication slippage is common in areas of microsatellite D NA and involved DNA polymerase pausing and momentarily dissociating from the template during replication. When DNA polymerase reforms the bond with the template DNA, it is not in the same place as where it left off (Viguera et al 2001). If the DNA polymerase re attaches further downstream from the site of dissociation, the replicated DNA will have a deletion of the skipped nucleotides. If the DNA polymerase reattaches further upstream from the site of dissociation, there will be an insertion into the new re plicated DNA strand. A preliminar y survey of a ll the alleles reveals a great disparity in the number of t hymine nucleotides in this region across all taxa. The most common number being seven th ymine bases, but there may be as many as twelve in other alleles. This process of insertion is what is hypothesized to have happened in the hybrid second allele to give it three more thymine nucleotides than the parent C. reticulata allele. Any one of the above mutations alone may be conceivable, but having all three in such a short time does not seem feasible. To improve upon this in the future a ne w cross should be made where the hybrid origin is known and the actual parent plants are sampled. Multiple individuals of the hybrid from multiple seeds of one hybrid cross should also be sampled. By just including
36 one hybrid individual not all possible alle lic combinations were represented. One hybridization event would have produced ma ny seeds, providing multiple pollen grains came into contact with multiple stigmas. Some percentage of those seeds would be viable depending on compatibility between the two pa rent species genomes (Judd et al. 1999). Planting all those seeds would yield hybrid pl ants that may differ in which alleles came from which parent. By analyzing sequences from multiple hybrid plants derived from seeds from the same flower, it would be possi ble to find out more about what the actin intron sequences can tell us about hybridization. The taxa from gig harbor, C. jackmanii x was a putative hybrid. Unfortunately, the actin intron sequences for this taxa did not sort into two distinct alleles. All sequences for this taxa were the same allele, specifically the very common allele. Therefore, the actin intron marker can not support or refute the hybrid hypothesis for C. jackmanii x Nitrate Reductase Intron Topics In general, many of the themes of the act in intron sequence analysis hold true for the NR intron sequence analysis as well. Th e neighbor joining tree th at resulted from the NR intron sequences also had few well supported relationships. At the base of the tree are the alleles of C. terniflora (grouping together with a bootstrap value of 99), C. lasiandra (grouping together with a bootstrap value of 66), and C. fusca There is not much resolution in relationships between these spec ies, however, these three species all have Asian geographic distributions. C. terniflora C lasiandra and C. fusca are all classified into different sections by Tamura and later by Johnson as well (1987, 2001). These Asian taxa must be the most different from the other (mostly North American) taxa which is
37 why they are grouping outside of the major clad e of the tree at letter C (Figure 3). Not all the Asian taxa fall unresolved at the ba se of the tree. The two alleles of C. brachyura fall within the clade marked at letter C, but th ey do not fall together. The first allele of C. brachyura falls at the base of the clade, forming a sister group with th e rest of the North American and European species whic h form a clade at letter D. One other well supported clade, at letter F of Figure 3, also groups together taxa of similar geographic distribution. C. campaniflora and C. integrifolia are the two European species included in the study. These two species are also not in the same section according to Tamura (1987). However, th e clade at letter F is better evidence for possibly modifying the existing classification than the unresolved relationship of the Asian species. Among North American species there is ve ry little resolution. Like the actin intron analysis, the NR intron analysis was al so lacking in sufficient variation to tease apart the relationships of the closest related species. In the nitrate reductase data set 31% of sites were variable, but 58% of variable sites were singletons. Singletons do not contribute much information to a neighbor join ing tree. In the data set, 12.9% of sites were parsimony informative. As with the actin data set, the variati on is concentrated in the sequences of the few least closely related taxa. When the sequences of C. terniflora C. lasiandra and C. fusca were removed from the data set, the number of variable sites was reduced to 14.9%, 69% of these variable sites being singletons Removing those taxa also reduced the parsimony informative sites to 4.6%. This again explains why the most basal groups are fairly well suppo rted, but why there is very lit tle resolution in the rest of the tree in both the neighbor joining and maximum parsimony analyses.
38 Again there were many taxa that shared one most common allele. In this case C. crispa C. fremontii C. reticulata C. ochroleuca and C. crispa x C. reticulata all shared one allele and are all members of the proposed Viorna section. However, C. drummondii also shared this allele, but is considered a member of the Viorna section. Obviously, there was no resolution obtained in the relationships of the alleles with the same sequence. In two separate analyses, usi ng two different intron markers, C. drummondii grouped in a clade with other members of the Viorna section. Judging by morphological data, C. drummondii has quite disparate characters than those of the Viorna section. For example, C. drummondii (as well as the other members of its subsection) are dioecious, and have white flowers, while members of section Viorna are all monoecious with more showy flowers typically urn shape d. Admittedly, the clade that united C. drummondii with members of the Viorna section has almost no further resolution, and C. drummondii may be grouping with the other taxa ba sed on their North American connection. Another interesting relati onship from the NR intron sequence analysis is the clade, letter H, uniting the first allele of C. socialis population L into a sister group relationship with the two alleles of C. baldwinii This is interesting because the other allele of C. socialis population L falls into a different polytomic clade with C. socialis population G and C. glaucophylla This may support a hypothesis that C. socialis population L is a hybrid between C. socialis population G and C. baldwinii This may also just be a coincidence that the two alleles of C. socialis population L are very different from one another and one allele ha ppens to have a sequence very similar to C. baldwinii The coincidence hypothesis is supported by the fact that C. socialis is an endangered species with all four known populations occurr ing in Alabama (Kral 1982;
39 Boyd and Hilton 1994). C. baldwinii however, has only been documented to occur in Florida (Johnson 2001). Anothe r possibility is that C. socialis population L is a hybrid, but not between C. socialis population G and C. baldwinii but between C. socialis population G and another species not include d in this study, but with an NR intron sequence related to that of C. baldwinii Unfortunately, the experimentally generated hybrid was homozygous for the NR intron marker. This makes sense because both C. crispa and C. reticulata possess the most common allele for this marker. Since one al lele is so prevalent for this marker, it is not a good hybrid diagnostic t ool. The chance of the same allele coming together from two different taxa is too great and when that process occurs detecting the hybrid state is impossible. In general, there are fewer alleles in this marker than were found for the actin intron marker. Only six of the 17 taxa anal yzed had two alleles rather than only one. However, the differences that did exist we re more pronounced than those for the actin intron marker, in general. Disregarding the si x taxa that shared th e most common allele, the other alleles present in the analysis were very different from one another, with many substitutions. Even though the marker wa s only about 200 base pairs long, the whole length of it was made up of intron. Therefore, there were many substitutions throughout the sequences for different alleles. Unlike the actin intron sequences, there was no region dominated by one base or any short repeats so replication slippage was not an issue. Ultimately, this means that the phylogenetic utility of this marker is greater than the actin intron marker, even though overall the NR intron is shorter.
40 This analysis was specifically conducte d using the third intron of the nitrate reductase gene. Primers were also available for the second intron of the nitrate reductase gene. Preliminary analysis s uggests that this intron may be longer than the third, and therefore have more phylogenetic informative characters. Future analyses should include exploring the phylogenetic utility of this in tron in conjunction w ith the third nitrate reductase intron and the actin intron. PCR need s to be optimized to amplify the second intron preferentially, before sequenc ing efforts can be efficient. Combined Actin and Nitrate Reductase Intron Topics When data sets are combined, the goal is to get more information from the sum of the parts than is available by the parts indivi dually. Molecular data sets are combined by concatenating the sequences. Very little info rmation was added to the analyses from the concatenated data set. The major relationships delineated in the nitrate reductase analysis were reiterated in the analysis of the concatenated data se t, but with lower bootstrap values. The neighbor joining tree resulting fr om the concatenated data set further supports the conclusion that the actin data set does not yield informative phylogenetic relationships because the tree had no furthe r resolution with the added actin sequence data. Previous work on this genus using the actin intron resulted in a tree with better resolution and higher bootstrap values, but the taxa included were from many different sections within Clematis When the taxa sampling is limited to mostly one section, Viorna there is little to no resolution within the section. In the previous study, C. brachyra showed anomalous relati onships within the trees developed (Slomba 2002);
41 anomalous results were c onsistent in this study. C. brachyura had two alleles for both markers tested; when the markers where concat enated based on allele the two sequences grouped into very different places on the tree (Figure 3). The first allele of C. brachyura groups at the base of a larger clad e composed mostly of members of Viorna (which C. brachyura is not). Placement here would be hypot hesized. However, the second allele of C. brachyura groups well within the clade of Viorna species. In fact, the second allele groups in a sister relationship to a clade of C. campaniflora and C. integrifolia It may be that C. brachyura has a faster rate of evol ution than other taxa in Clematis Some alleles are becoming less closely related to each othe r than they are to other alleles from different taxa. This would make C. brachyura appear more closely related to other taxa, but this relationship would act ually be due to random mutation and not close common ancestry. One way to test the relationships of C. brachyura would be to sample from many individuals to see what alleles are presen t in the population, rather than in just one individual. This may refute th e results shown here and in pr evious analyses, or it may verify an interesting evolutionary phenomenon in C. brachyura Conclusions Overall conclusions that come from this study include the fact that intron sequence data will not provide a well reso lved tree for closely related species of Clematis in the Viorna section using typical phylogenetic met hods. The probability of interfertility allowing gene flow between species (hybridization) is so great that re ticulate evolution is confounding the image of a tree that assume s bifurcation. Because of this problem, adding more intron sequences may not pr ovide any more resolution for these
42 phylogenies. Instead, a population level approach should be take n in the future that would allow for frequencies of alleles to be cal culated and different methods of phylogeny can be employed like population analyses that make minimum spanning networks. In order to accomplish this, more sampling should be done to multiple individuals within the current species included. Sampling more individuals would also help overcome the problem encountered when diagnosing hybrid hist ory. Many, if not all, possible allelic combinations must be sampled to determine the hybrid parentage molecularly.
43 LITERATURE CITED Aguilar, J.F., J.A. Rossell, G.N. Feliner. 1999. Nuclear ribosomal DNA (nrDNA) concerted evolution in natura l and artificial hybrids of Armeria (Plumbaginaceae). Molecular Ecology. 8:1341-1346. Boyd, R.S. and C.D. Hilton. 1994. Ecological studies of the endangered species Clematis socialis Kral. Castanea. 59(1):31-40. Cros, J., M.C. Combes, P. Trouslot, F. Anthony, S. Hamon, A. Charrier, and P. Lashermes. 1998. Phylogenetic analysis of chloroplast DNA variation in Coffea L. Molecular Phylogenetics and Evolution. 9(1): 109-117. Essig, F.B. 1991. Seedling morphology in Clematis (Ranunculaceae) and its taxonomic implications. Sida. 14(3): 377-390. Ferguson, D. and T. Sang. 2001. Speciation th rough homoploid hybridization between allotetraploids in peonies ( Paeonia ). PNAS. 98(7):3915-3919. Franzke, A. and K. Mummenhoff. 1999. Recent hybrid speciation in Cardemine (Brassicaceae)conversion of nuclear ribosomal ITS seque nces in situ nascendi. Theor. Appl. Genet. 98:831-834. Friedman, H., J.W. Vos, P.K. Hepler, S. Meir, A.H. Halevy, and S. Philosoph-Hadas. 2003. The role of actin filaments in the gravitropic response of snapdragon flowering shoots. Planta. 216:1034-1042. Foyer, C.H., M. Parry and G. Noctor. 2003. Mark ers and signals associated with nitrogen assimilation in higher plants. Journal of Experimental Botany. 54(382): 585-593. Hardig, T.M., P.S. Soltis, and D.E. Soltis. 2000. Diversification of the North American shrub genus Ceanothus (Rhamnaceae): c onflicting phylogenies from nuclear ribosomal DNA and chloroplast DNA. Am erican Journal of Botany. 87(1): 108123. Higgins, D.G., J.D. Thompson, and T.J. Gibson. 1996. Using CLUSTAL for multiple sequence alignments. Meth. Enzymol. 266:383-402.
44 Hillis, D.M., C. Moritz, and B.K. Mable. 1996. Molecular Systematics 2nd ed. Sinauer Assoc. Inc. Sunderland, MA. Hoot, S.B. and J.D. Palmer. 1994. Structur al rearrangements, including parallel inversions, within the chloroplast genome of Anemone and related genera. Journal of Molecular Evolution. 38:274-281. Howarth, D.G. and D.A. Baum. 2002. Phylogeneti c utility of a nuclear intron from nitrate reductase for the study of closely relate d plant species. Mole cular Phylogeny and Evolution. 23:525-528. Johnson, Magnus. 2001. The Genus Clematis. Magnus Johnson Plantskola AB. Sdertlje, Sweden. Judd, W.S., C.S. Campbell, E.A. Kellogg, and P.F. Stevens. 1999. Plant Systematics: A Phylogenetic Approach. Sinauer Associat es Inc. Sunderland, Massachusetts. Kral, R. 1982. A new Clematis from northeastern Alabam a. Rhodora. 84(838):285-291. Koontz, J.A., P.S. Soltis, and D.E. Soltis. 2004. Using phylogeny reconstruction to test hypotheses of hybrid origin in Delphinium section Diedropetata (Ranunculaceae). Systematic Botany. 29(2): 345-357. Kumar, S., K. Tamura, and M. Nei. 2004. ME GA3: Integrated software for molecular evolutionary genetics analysis a nd sequence alignment. Briefings in Bioinformatics. 5:150-163. Li, W.H. 1997. Molecular Evolution. Si nauer Assoc. Inc. Sunderland, MA. Mathur, J. and M. Hlskamp. 2002. Micr otubules and microfilaments in cell morphogenesis in higher plants. Current Biology. 12: R669-R676. Mathur, J. 2004. Cell shape developments in plants. Trends in Plant Science. 9(12):583590. Miikeda, O., S. Koga, T. Handa, and T. Yukawa. 1999. Subgeneric Relationships in Clematis (Ranunculaceae) by DNA sequences. In : S. Andrews, A.C. Leslie and G. Alexander (Editors). Taxonomy of Cu ltivated Plants: Third International Symposium, pp. 355-358. Royal Botanic Garderns, Kew. Moniz de Sa, M. and G. Drouin. 1996. Phyloge ny and substitution rates of angiosperm actin genes. Molecular Biology and Evolution. 13(9): 1198-1212. Rieseberg, L.H. 1997. Hybrid Origins of Pl ant Species. Annu. Rev. Ecol. Syst. 28:35989.
45 Rossetto, M. 2005. A simple molecular approach for identifying a rare Achronychia (Rutaceae) provides new insights on its multiple hybrid origins. Biological Conservation. 121:35-43. Sharma, P., R.S. Dubey. 2005. Modulation of n itrate reductase activity in rice seedlings under aluminium toxicity and water stress: role of osmolytes as enzyme protectant. Journal of Plant Physiology. 162: 854-864. Slomba, J.M., J.R. Garey, and F.B. Essig. 2004. The Actin I intron A phylogenetically informative DNA region in Clematis (Ranunculaceae). Sida. 21(2):879-886. Soltis. D.E. and R.K. Kuzoff. 1995. Disc ordance between nuclear chloroplast phylogenies in the Heuchera group (Saxifragaceae). E volution. 49(4):727-742. Strand, A.E., J. Leebens-Mack, and B.G. Milligan. 1997. Nuclear DNA-based markers for plant evolutionary biology. Molecular Ecology. 6:113-118. Tamura, M. 1968. Morphology, ecology and phylogeny of the Ranunculaceae. Sci. Rep. S. Coll. N. Coll. Osaka Univ. 16:21-43. Tamura, M. 1987. A classification of the genus Clematis Acta Phytotax. Geobot. 38:3344. Thompson, J.D., D.G. Higgins, and T.J. Gibson. 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22:4673-80. Viguera, E., D. Canceill, and S. Dusko Eh rlich. 2001. Replication slippage involves DNA polymerase pausing and dissociation. The EMBO Journal. 20(10):2587-2595. Weese, T.L., L.A. Johnson. 2005. Utility of NADP-dependent isocitrate dehydrogenase for species-level evolutionary inferen ce in angiosperm phylogeny: A case study in Saltugilia Molecular Phylogenetics and Evolution. 36:24-41. Zhou, R., S. Shi, and C. Wu. 2005. Molecula r criteria for determining new hybrid speciesAn application to the Sonneratia hybrids. Molecular Phylogenetics and Evolution. 35:595-601. Zomlefer, W.B. 1994. Guide to Flowering Plan t Families. University of North Carolina Press. Chapel Hill, North Carolina. Pg. 42-45.