Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Calycadenia : a model system for investigating chromosomal speciation Olson, Teika E. 2002

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_2002-0196.pdf [ 5.15MB ]
JSON: 831-1.0090262.json
JSON-LD: 831-1.0090262-ld.json
RDF/XML (Pretty): 831-1.0090262-rdf.xml
RDF/JSON: 831-1.0090262-rdf.json
Turtle: 831-1.0090262-turtle.txt
N-Triples: 831-1.0090262-rdf-ntriples.txt
Original Record: 831-1.0090262-source.json
Full Text

Full Text

Calycadenia: A Model System for Investigating Chromosomal Speciation by Teika E. Olson B.Sc, The University of British Columbia, 1998 A Thesis Submitted in Partial Fulfilment of the Requirements for the Degree of Master of Science in The Faculty of Graduate Studies (Department of Botany) We accept this thesis as conforming to the required standard The University of British Columbia April 2002 (c) Teika E. Olson, 2002 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of E>Q T l R k P f The University of British Columbia Vancouver, Canada Date A P R I L , \%, 2DO^ DE-6 (2/88) Abstract The theory of chromosomal speciation proposes that cytological differences such as reciprocal translocations and pericentric inversions may present the primary barrier to gene flow among diverging groups. While theory supports the possbil-ity of chromosomal speciation, conclusive evidence from nature is lacking and few natural systems have been developed for investigating the phenomenon. The California tarweed species Calycadenia fremontii A. Gray and C. paucijlora A. Gray provide an ideal system in which to examine some of the predictions of chromo-somal speciation. These species comprise numerous cytological races distin-guished primarily by reciprocal translocations. Cytological differences among the races correlate with reduced fertility among hybrid Fl progeny, suggesting a pos-sible isolating role for chromosomal rearrangements in this group. To test whether cytological differences among populations have played a primary role in driving divergence, genetic distances among 460 individuals collected from 23 populations were calculated from allele frequencies based on 88 polymorphic RAPD markers shared among populations. Genetic distances were used to con-struct Neighbor-Joining and Fitch-Margoliash trees from which past patterns of gene flow among populations and cytological races could be inferred. As an alter-native method for estimating genetic divergence among populations and cyto-types, allele frequencies estimated from RAPD markers were also used to measure centroid F S T for each population. Both the genetic distance trees and centroid F$T statistics indicate that populations are genetically differentiated. Futhermore, cytological races are not resolved into monophyletic clades in the genetic distance trees suggesting that cytological differences among populations do not account for the primary barrier to gene flow among them. Mantel tests confirm that cytolog-ical differences among populations do not explain the variation in genetic dis-tances among populations, although geography does. I conclude that cytological differences among populations in Calycadenia fremontii and C. pauciflora have not presented the primary barrier to gene flow, and find that the theory of chromo-somal speciation is not supported. u Table of Contents Chapter & Section Page Abstract ii Table of Contents iii List of Tables v List of Figures vi Acknowledgements viii 1. Introduction 1 1.1 The Paradoxical Problem of Chromosomal Speciation 1 1.2 Anecdotal Evidence in Support of Chromosomal Speciation 2 1.3 Theory in Support of Chromosomal Speciation 4 1.4 Difficulties With Determining a Primary Reproductive Isolating Effect of Chromosomal Rearrangements in Natural Populations 6 1.5 Testing for Chromosomal Speciation in Calycadenia 10 1.6 The Study System 11 1.6.1 The Genus Calycadenia 11 1.6.2 The Calycadenia fremontii - C. pauciflora Complex 14 1.7 Using Molecular Markers to Describe Natural Populations 19 2. Methods 21 2.1 Sources of Material 21 2.1.1 Field Collections 21 2.1.2 Herbarium and Greenhouse Collections 22 2.2 D N A Isolation and Quantification 26 2.2.1 Qiagen Extractions 27 2.2.2 CTAB Extractions 28 2.2.3 Qualitative and Quantitative Verification of D N A Product 29 2.3 RAPD-PCR and Gel Electrophoresis 30 2.3.1 PCR Conditions 30 2.3.2 Scoring Gels and Testing for Homology of Marker Bands 31 2.4 Data Analysis 33 2.4.1 Relationships Among Individuals 33 2.4.2 Verification of Population Racial Identity 33 2.4.3 Relationships Among Populations 34 2.4.4 Summary of Genetic Variation Within and Among Populations: Percent Polymorphism And F S T 35 iii 2.4.5 Mantel Tests for Correlations Among Data Matrices 36 2.4.6 Distribution of Per Locus F S T 38 3. Results 39 3.1 RAPD Markers 39 3.2 Relationships Among Individuals 40 3.3 Verification of Population Racial Identity 44 3.4 Relationships Among Populations 47 3.4.1 Genetic Distance Trees 47 3.4.2 Using Mantel Tests to Investigate Relationships Between Clades 50 3.5 Summary of Genetic Variation Within and Among Populations: Percent Polymorphism and F S T 52 3.6 Mantel Tests for Correlations Among Data Matrices 52 3.7 Distribution of Per Locus F S T 55 4. Discussion 57 4.1 Relationships among populations 57 4.1.1 Patterns of Relationships From Genetic Distance Trees 57 4.2 Features That Distinguish the Two Clades 59 4.2.1 Flower Color Polymorphism and Flowering-Time Differences 59 4.2.2 Geographic Distinctions Between the Clades 59 4.2.3 Taxonomic Affinities of the Clades 61 4.3 Evidence From Other Sources 63 4.3.1 F S T Analyses 63 4.3.2 Comparison with ITS and ETS Trees 63 4.3.3 Evidence from Mantel Tests 64 i) The Effect of Geography on Genetic Distances 64 ii) The Effect of Cytological Differences on Genetic Distances 66 iii) The Effect of Geography on Cytological Differences 67 4.4 Distribution of Per Locus F S T 59 4.5 Summary and Future Directions 70 4.5.1 Summary 70 4.5.2 Future Directions 71 5. References 73 i v List of Tables Table Page 1.1. Cytological races of Calycadenia fremontii and C. pauciflora. 14 1.2. Crossing data for cytological races of Calycadenia pauciflora A. Gray (G.D. Carr, 1975) and for cytological races of C. fremontii A. Gray (R.L. Carr & G.D. Carr, 1983). 17 2.1. Collection data. 23 2.2. RAPD primers used to generate D N A fingerprints for 460 individuals in the Calycadenia fremontii - C. pauciflora complex collected from 23 sites throughout Northern California. 30 3.1 Results of Mantel tests for investigating relationships between clades. 51 3.2. Descriptive statistics from TFPGA. 53 3.3. Results of Mantel tests from grouped populations 54 List of Figures Figure Page 1.1 Inflorescence detail of Calycadenia. 11 1.2 Ray and disk cypselae. 12 1.3 Range and diversity of the genus Calycadenia in California. 12 1.4 Minimum-length parsimony tree based on ETS & ITS sequence data. 13 1.5 Inflorescences of Calycadenia. 15 1.6 Range of Calycadenia Jremontii A. Gray by county in California. 15 1.7 Range of Calycadenia paucijlorai A. Gray by county in California. 15 1.8 A typical collection site near Redding, CA., June 1, 2000. 16 1.9 Hypothetical diagram of chromosomes repatterned in the formation of known structural races of Calycadenia jremontii. 18 1.10 (a) Proposed relationships of chromosomal races of Calycadenia pauciflora based on maximum meiotic chromosome association from every possible hybrid combination between the cytological races, (b) Summary of cytological relationships in Calycadenia Jremontii based on meiotic pairing of interracial hybrids. 18 2.1 T.E. Olson collection sites for Calycadenia jremontii and C. pauciflora. 21 2.2 A typical roadside collection site for Calycadenia jremontii, east of Redding, CA., June 1, 2000. 22 2.3 Sample data gels for each of the six U B C RAPD primers used in this project. 31 3.1 Test of band homology for four populations for U B C RAPD primer #63. 39 3.2 Complete Neighbor-Joining distance tree from PAUP* (Swofford, 2002) based on RAPD marker fingerprints for 460 individuals collected from 23 sites. 41 3.3 Condensed Neighbor-Toining tree from PAUP* (Swofford, 2002) based on RAPD marker fingerprints for 460 individuals collected from 23 sites. 42 3.4 Condensed weighted-least squares (inverse weighting, power = 2) distance tree from PAUP* (Swofford, 2002) based on RAPD marker fingerprints for 460 individuals collected from 23 sites. 43 3.5 Matching Can's populations to Olson's populations using RAPDs. 44 3.6 Neighbor-Joining distance tree from PAUP* (Swofford, 2002) showing relationships among field-collected samples of R.L. Carr and T.E. Olson based on RAPD marker profiles. 45 3.7 Neighbor-Joining distance tree from PHYLIP (Felsenstein, 1995) based on estimated allele frequencies for 88 RAPD marker loci. 48 v i Figure Page 3.8 Fitch-Margoliash unweighted least-squares distance tree from P H Y L I P (Felsenstein, 1995) based on estimated allele frequencies for 88 P v A P D marker loci. 49 3.9 Distribution of per locus F S T estimated for 88 P v A P D markers scored across 23 populations. 56 4.1 Topographical differences among collection sites in the Calycadenia fremontii - C. pauciflora complex. 60 vn Acknowledgements The past three years of study could not have been possible without the participa-tion, encouragement and support of many people to whom I am greatly indebt-ed. Firstly, I must thank my supervisor, Jeannette Whitton, and the other mem-bers of my academic committee, Mary Berbee, Fred Ganders, Sally Otto, and Michael Whitlock for their advice, their criticisms, and their patience throughout this project. Bruce Baldwin and Bob Carr have been co-operative, generous, helpful collaborators who have never tired of sharing with me their data, their plant material, and their advice. Bruce shared his every discovery from D N A sequencing of Calycadenia with me, and offered much technical support along the way. Bob supplied me with seeds, herbarium specimens, and collection data without which this project could never have been started. I am also honored to have spent several years working alongside the other mem-bers of the Whitton Lab: Stacey Thompson, Nishanta Rajakaruna, Linda Jennings, and Kim Ryall who was an indespensible, consciencious and entertain-ing field and lab assistant. Much thanks also to my family for their love and encouragement. Finally, my partner, Mike Newton, deserves endless praises for his pep talks, his advice, and his many hours spent coaching me on the finer elements of graphic design, typography, and all those non-science aspects of thesis-writing. viii 1. Introduction 1.1 The Paradoxical Problem of Chromosomal Speciation The theory of chromosomal speciation proposes that spontaneous (or induced) changes in chromosomal structure or arrangement may cause reproductive isola-tion between sister taxa and lead to subsequent speciation (White, 1968). In the past, the notion of chromosomal rearrangements as potential agents of sponta-neous, saltational evolution was considered likely to be an important means of achieving reproductive isolation. Numerous examples of closely related plant species differentiated by chromosomal interchanges exist (Singh, 1993). These include species of Oenothera differentiated by reciprocal translocations (Cleland, 1962) and wild perennial Glycine species that are genomically similar, but differ by chromosomal inversions (Singh, 1993). Historically, even noted zoologists have stated that there could be "occasional significance to chromosome aberration, and to hybridization, as direct species forming agencies" (Wright, 1930). In the late 20th Century however, the trend, particularly in the zoological litera-ture, was toward favoring a genie basis to explain the evolution of reproductive barriers among diverging taxa (Coyne & Orr, 1998). In contrast to saltational chromosomal evolution, genie forms of evolution rely on the accumulation of changes among the genes that underlie complex traits, leading to genie incompat-ibilities and therefore reproductive isolation among groups. Given this emphasis, the role of chromosomal rearrangements in reproductive isolation is considered secondary, perhaps acting to protect divergent gene combinations from recombin-ing (Kyhos & Carr, 1994). Although changes in chromosome number and arrangement occur frequently in many groups of organisms including plants, data that unambiguously describe the role that chromosomal changes play in speciation are lacking, and the importance of chromosomal speciation remains the subject of debate (Coyne & Orr, 1998; King, 1993; Kyhos & Carr, 1994; Rieseberg, 2001; White, 1968). The debate concerns the role and timing of chromosomal changes in the evolu-tion of reproductive isolation. Opponents of chromosomal speciation argue that chromosomal rearrangements occur subsequent to genie, ecological or behavioral changes that initiate reproductive isolation and may or may not reinforce other isolating mechanisms (Coyne & Orr, 1998). For example, genie factors may be the primary barrier to gene flow among species, but differences in chromosomal arrangement may reduce or prevent the introgression of genes within the rearrangement by reducing the rates of recombination. Such is the case for Drosophila pseudoobscuras and D. persimilis, two hybrid-forming species that share many genomic similarities but are differentiated by inversions on two chromo-somes (Noor et al, 2001a; Noor, Grams, Bertucci, & Reiland, 2001b). In 1 hybrids between the two species, inversions drastically reduce recombination dur-ing meiotic cross-over throughout their length. This reduction in recombination suggests that chromosomal inversions present a significant barrier to gene flow between these species. In fact, however, experimental evidence suggests that gene flow is unimpeded for regions of the genome not spanned by inversions, and it is clearly polygenes, not chromosomal differences, that result in the primary barrier to hybrid male fertility and female mating preference in both these species (Noor et al, 2001a; Noor et al, 2001b). Chromosomal inversions provide a sec-ondary reinforcement of the genie barrier to gene flow among species by protect-ing sterility loci within rearranged segments from recombination, but rearrange-ments do not confer hybrid sterility or dysfunction themselves. Conversely, proponents of chromosomal speciation suggest that chromosomal rearrangements, including reciprocal translocations, inversions, and fusions can form primary barriers to gene flow, and that chromosomal differences among par-ents result in reduced fertility or viability of their hybrid progeny. Despite the lack of conclusive experimental evidence for the primary role of chromosomal rearrangements in driving divergence among closely related taxa (King, 1993; Lewis, 1966; White, 1968), several lines of anecdotal and preHminary experimen-tal evidence (e.g. Pialek, Hauffe, Rodriguez-Clark, & Searle, 2001), coupled with a large body of theory (Bengtsson & Bodmer, 1976; King, 1993; Lande, 1979; Sites & Moritz, 1987; Spirito, 1998; Walsh, 1982; Wright, 1941) suggest that chromosomal speciation is possible. 1.2 Anecdotal Evidence in Support of Chromosomal Speciation It has been documented that in humans, as in many other animals and plants, individuals that are heterozygous for chromosomal rearrangements like reciprocal translocations produce unbalanced gametes, which are themselves inviable or lead to progeny that have reduced fertility or are infertile (Therman & Susman, 1993). Heterozygous carriers of single reciprocal translocations are expected to produce 50% unbalanced gametes in the absence of recombination, but the reduction in fertility may be as much as 100% depending on the level of recombination between interchanged segments. Reduced fertility for carriers of reciprocal translocations has been documented in numerous animal species including cattle (Ducos et al., 2000), sheep (Anamthawat-Jonsson, Long, Basrur, & Adalsteinsson, 1992), pigs (Makinen, Pitkanen, & Andersson, 1997), mice (Beechey & Evans, 2000), and humans (Daniel, Hook, & Wulf, 1989), as well as in agricultural crops like cotton (Stelly, Kautz, & Rooney, 1990). Multiple rearrangements may reduce fertility of heterozygous carriers and the viability of their offspring in an additive way, but rearrangements may also interact epistatically to produce a vastly greater reduction in fitness than predicted based on the product of their individual fitness 2 reductions (Bengtsson & Bodmer, 1976; Walsh, 1982). One type of chromosomal-mediated instantaneous speciation, amphiploidy (allopolyploidy) (Clausen, Keck, & Hiesey, 1945), involves the whole genome. Reproductively isolated allopolyploid species result from the fusion of two diploid genomes of karyotypically dissimilar species. Numerous genera of plants produce interspecific hybrids that give rise to reproductively isolated, allopolyploid neospecies. These include Madia (e.g. M . nutans X M. rammii gives M. nutrammii), and Layia (e.g. L. pentachaeta X L. platyglossa yields L. pentaglossa), both tarweeds in the family Asteraceae (Heliantheae: Madiinae) (Clausen et al, 1945). While saltational speciation by means of amphiploidy is a virtually uncontested pathway, critical data supporting the role of other types of saltational chromosomal change in driving divergence are lacking (King, 1993; Lewis, 1966; Spirito, 1998; White, 1968). While the role of saltational chromosomal differences in driving divergence may not be clear, there is ample evidence that novel chromosomal arrangements have been fixed in natural plant populations many times (Lewis, 1966). Many closely-related plant species in the genera Lasthenia, Holocarpha, and Allophyllum are known to exhibit multiple chromosomal rearrangement differences that are corre-lated with reproductive isolation among taxa (Lewis, 1966). Whether these cor-relations are due entirely to structural differences among chromosomes or whether genie differences also differentiate these groups is unclear. While it is impossible to generalize about the specific effects and behaviors of chromosomal rearrangements across all biological groups, certain cytological dif-ferences that mark the division between plants and animals may make plants the more likely candidates for exhibiting evolutionarily important chromosomal differ-ences. Firstly, when X-irradiated, sister chromatids of animal chromosomes break and usually do not rejoin. Plant chromatids, however, are capable of repairing radiation-induced breaks, although this can lead to fusion of previously unat-tached parts of chomosomes (Therman & Susman, 1993). Irradiation, therefore, can induce chromosomal rearrangement in plants, whereas such changes to animal chromosomes are generally not immediately repairable and are deleterious. Additionally, the chromosomes of many plants can be broken with lower doses of ionizing radiation than the chromosomes of animals such as Drosophila (Therman & Susman, 1993). Tolerance of chromsomal rearrangements also differs between plants and animals. While some chromosomal polymorphisms such as those that involve pericentric inversions or increases in chromosomal length are benign and often cause no phe-notypic effect in animals (e.g. humans), those that involve translocations, dele-tions, partial monosomies, and formation of multibranched chromosomes usually 3 have serious, deleterious phenotypic effects (Therrnan & Susman, 1993). In con-trast, many plants, though more susceptible to chromosomal mutation, are far more capable of tolerating unusual chromosomal rearrangements. For example, high levels of structural heterozygosity occur in the genus Oenothera (Onagraceae), whose members regularly display numerous complex, multivalent chromosomal structures at meiosis (Cleland, 1972). In light of these behavioral differences among plant and animal chromosomes, one finds that historically botanists have shown much greater willingness to accept cytogenetic events as contributors to speciation than have zoologists. This is clearly evidenced in the disciplines' alternative interpretations of hybrid sterility: in botany, hybrid sterility is often interpreted as "cryptic structural differentiation", whereas zoologists overwhelmingly cite genie factors as contributing to hybrid sterility (Rieseberg, 2001). For instance, hybrid inviability, hybrid male sterility, male and female sexual isolation, male courtship song, and genital morphology are among the numerous genie traits conferring reproductive isolation between species of Drosophila (Coyne & Orr, 1998). 1.3 Theory in Support of Chromosomal Speciation Evolutionary theory that explains the mechanisms by which chromosomal specia-tion can occur arose out of an inherent paradox in the theory of chromosomal speciation. The theory of chromosomal speciation predicts that chromosomal rearrangements act as primary barriers to gene flow by reducing the fertility or fitness of rearrangement heterozygotes. Yet in order to see divergence driven by cytological differences, these underdominant chromosomal rearrangements must become established and fixed in diverging lineages. The establishment and fixa-tion of chromosomal rearrangements that, according to the theory, must be nega-tively heterotic seems a troubling paradox. Numerous authors have tackled this apparent paradox of chromosomal evolution for the three most likely types of population dynamics (Bengtsson & Bodmer, 1976; Lande, 1979; Spritio, 1998; Walsh, 1982). These are strict underdomi-nance, underdominance with a selective advantage for the rearrangement homozygote, and underdominance with meiotic drive. For the stricdy underdominant chromosomal mutation, it has been shown that its fixation can occur only if the population size is extremely small (on the order of N £ = 20 individuals) and no immigration occurs (Bengtsson & Bodmer, 1976; Lande, 1979; Spirito, 1998; Wright, 1941). For plants, immigrants could include not only seeds or pollen introduced from neighboring populations, but also indi-viduals from past generations of the local population, represented through the 4 seed bank (Kyhos & Carr 1994; Levin, 1990). In very small populations any immigration to the population would tend to overwhelm the stochastic shift toward fixation for a novel, underdominant mutation. Wright (1941) has shown that fixation of a novel chromosomal mutation like a reciprocal translocation, that results in semisterility (e.g. loss of 50% of the gametes) occurs with a probability (2e/Ne)(3/4)2N, where N e is the effective pop-ulation size (Bengtsson & Bodmer, 1976). In other words, for chromosomal mutations that lead to severe reductions in gamete number or fitness in heterozy-gotes, fixation by drift alone is extremely unlikely even for very modestly sized populations. If mutations are only weakly underdominant (e.g. relative fitness reduction of about 2% for heterozygotes), fixation can occur in larger populations and can occur at faster rates (Walsh, 1982), but such mutations individually would be weak barriers to gene flow and would not be of primary importance in the evolution of new species (Spirito, 1998). However, it has been suggested that weakly underdominant rearrangements may accumulate, so that by acting in con-cert they result in a more drastic reduction to hybrid fitness than any single rearrangement (Dobzhansky, 1970, p. 338, cited in Walsh, 1982). Still, each of these weakly underdominant mutations must become fixed in turn. If the novel homozygote is given a slight selective advantage over the ancestral homozygote, fixation can occur for a strongly underdominant structural mutation in a reasonably large population (Bengtsson & Bodmer, 1976; Spirito, 1998; Walsh, 1982). One scenario whereby novel homozygotes would have an advan-tage over ancestral forms is in the case where the chromosomal rearrangement brings two rare, previously unlinked, advantageous alleles into linkage disequilibri-um (Bengtsson & Bodmer, 1976). Each time selection acts to increase the fre-quency of these alleles, it also increases the frequency of the novel structural mutation. If selection on alleles linked through the rearrangment is strong enough, it may drive an otherwise negatively heterotic structural mutation to fix-ation or near-fixation. Similarly, since rates of recombination are reduced within or adjacent to large rearrangements (Rieseberg, 2001; White, 1973), any genes or gene complexes in these regions would tend to be preserved; if these genes or complexes were of great advantage, any individual carrying a chromosomal arrangement that prevented their disruption would increase in frequency in the population. Fixation of underdominant structural mutations becomes even more likely given moderately high levels of meiotic drive (Bengtsson & Bodmer, 1976; Spirito, 1998; Walsh, 1982). Meiotic drive, or segregation distortion, is the passing on of genomic factors in a non-Mendelian, unequal fashion during meiosis in a het-erokaryotype. This preferential passing on of certain genomic components is often only exhibited by one sex in animals (Lande & Wilkinson, 1999). For 5 example, male stalk-eyed fruit flies (Cyrtodiopsis dalmanni) that exhibit X-linked meiotic drive produce markedly fewer male gametes (Lande & Wilkinson, 1999). Thus, meiotic drive is a mechanism generating female-biased sex ratios in stalk-eyed flies. In maize (Zea mays ssp. parviglumis and Z. m. ssp. mexicand), meiotic drive may be responsible for the evolution of large repetitive sequences of D N A on all chromosomes in mutants carrying an abnormal chromosome 10 (Buckler et al, 1999). In underdominant systems where meiotic drive preferentially produces gametes containing novel structural rearrangements, fixation can occur rapidly and effec-tively (Bengtsson & Bodmer, 1976; Spirito, 1998; Walsh, 1982). While such a scenario is theoretically attractive, it remains unclear whether meiotic drive is actually the driving force behind the establishment of any novel chromosomal rearrangements in nature (Spirito, 1998). To summarize, although perhaps not highly likely, the establishment or fixation of a novel, underdominant chromosomal arrangement is theoretically achievable, given certain conditions. These are small population size, with low immigration, a selective advantage for the rearrangement homozygote, and / or strong meiotic drive operating in favor of the novel rearrangement in heterozygous carriers. Close inbreeding, especially selfmg, can further facilitate establishment of strongly underdominant novel chromosomal arrangements (Lewis, 1966). In the next section, I will examine some cases for which chromosomal rearrange-ments have been implicated as the primary barriers to gene flow among species, and I will discuss the strengths and pitfalls of these interpretations. 1.4 Difficulties With Determining a Primary Reproductive Isolating Effect of Chromosomal Rearrangements in Natural Populations While theory predicts that strongly negatively heterotic novel chromosomal rearrangements can be fixed, the role of chromosomal change as a primary agent of evolutionary divergence in nature remains unclear. Even proponents of chro-mosomal speciation acknowledge that the behavior and consequences of chromo-somal rearrangements differ from case to case and that the effects of chromosomal rearrangements can be moderated by environmental factors (Spirito, 1998). Differences in responses to rearrangements may exist at the level of groups of organisms, among organisms within populations, and between the sexes. Furthermore, the severity of the negative effect of chromosomal mutations varies among the classes of mutations: reciprocal translocations typically result in semi-sterility in heterozygous carriers (Lande, 1979), while the sterility effects of para-centric inversions depends largely on the size of the inversion (Spirito, 1998). In 6 general, chromosomal rearrangements, including both translocations and inver-sions, tend to reduce rates of recombination for loci within or adjacent to the rearrangement (Rieseberg et al, 1995; White, 1973) but may cause no further noticeable phenotypic or fitness effects (Lande, 1979). Recently, Coyne and Orr (1998) have argued against chromosomal speciation, stating that "direct genetic analyses over the last decade have shown conclusively that postzygotic isolation in animals is typically caused by genes, not by large chromosome rearrangements." These authors argue that there are numerous cases where hybrid sterility occurs without chromosomal differences and just as many examples of high fertility among chromosomally divergent groups. In the former scenario, reproductive barriers among taxa may be attributed to isolating genes, as in the case of Drosophila pseudoobscura and D. persimilis (Noor et al., 2001b). In the latter case, where chromosomal differences are not the primary mechanism of reproductive isolation, barriers to gene flow may occur pre- or post-mating, and include ecological, geographic, temporal and, for plants, pollinator differences. It would seem that chromosomal rearrangements have not had a primary role in driving divergence in Drosophila, despite the fact that chromosomal differences exist among species (Coyne, 1993; Coyne & Orr, 1998). Nevertheless, in other animal groups in nature, such as the mole rats (Nevo, 1991), there would appear to be reason to suspect that species may be distinguished primarily by chromoso-mal differences, while genie differences may play a secondary or minor role. As illustrated in the following example involving Nevo's (1991) studies on mole rats, the difficulty of isolating the effects of cytological differences from geriic effects can lead to problems in interpreting the primacy of either mechanism. The subterranean mole rat superspecies Spalax ehrenbergi comprises four chromo-somal sibling species (2n = 52, 54, 58, 60) that differ in basic chromosome num-ber due to Robertsonian (whole arm) translocations and exhibit other minor cytological differences (Nevo, 1991). Based on the study of 25 allozyme loci, Nevo (1991) reported a recent origin for the species inferred from a high genetic identity (1= 0.966). Although species can hybridize, the hybrid zone for any species pair is narrow (0.32 to 2.80 km wide) and often decays, and restriction fragment length polymorphism (RFLP) analyses indicate that little or no gene flow occurs among species (Nevo, 1991). Additionally, hybrid fitness is reduced relative to parental fitness, particularly as one of the chromosomal polymorphisms involves the Y-chromosome and results in partial male sterility in hybrids (Nevo, 1991). While differences in geoclimatic range may represent the most significant present-day barrier to reproduction between the species, Nevo (1991) suggests that the original reproductive isolating mechanism may have involved chromosomal 7 rearrangements. As the most recently diverged species pair (2n=58, 2n=60) exhib-it postmating isolation, but not premating isolation, while the earliest diverging pair of species (2n=54, 2n=52) exhibit both pre- and postmating isolation, Nevo (1991) postulates that the evolution of postmating isolation preceded that of pre-mating isolation in Spalax. According to Nevo (1991), chromosomal differences between the species represent the most likely mechanism of postmating isolation as these differences lead to lower hybrid fitness and hybrid breakdown. According to this rationale, chromosomal differences between the species play a causal, rather than an incidental role in speciation in this complex (Nevo, 1991). Despite Nevo's interpretation of chromosomes playing a causal role in speciation, these results do not preclude genie effects. As long as there is measurable genet-ic distance among chromosomally distinct groups, it is possible that genie differ-ences and not chromosomal differences are the primary factor contributing to reproductive isolation. Despite the high genetic identity calculated from allozyme data, barriers to gene flow in Spalax could still be the result of genie differences. Evolution of reproductive isolation in Spalax could be analogous to the situation in Drosophila pseudoobscura and D. persimilis, in which reduced hybrid fitness and hybrid male sterility are controlled by genes. In Spalax as in Drosophila these genes may map to sites within chromosomal rearrangements, and while the rearrangements may protect the genes from recombination and prevent introgres-sion, the genes still fundamentally are the barrier to successful reproduction between groups. For example, the Y-chromosomal polymorphisms in Spalax are associated with partial male sterility in hybrids, suggesting that a gene contributing to male sterility could lie within the rearranged Y-chromosomal segment. For groups that exhibit any level of genetic variation among diverging, chromosomal-ly distinct lineages, genie effects may always confound any perceived effect of chromosomal rearrangements on gene flow. While chromosomal differences may or may not form the primary reproductive barrier in Spalax, Nevo's extensive studies on mole rats demonstrate that popula-tions of animals that differ karyotypically may exhibit variable abilities to inter-breed. Similarly, genie effects may be important in driving divergence in plants, despite the fact that reduced fitness and reproductive success in plant species hybrids is often associated with karyotypic polymorphism (e.g. Calycadenia (G.D. Carr, 1975; R.L. Carr, & G.D. Carr, 1983), Chaenactis (Kyhos, 1965), Clarkia (Lewis, 1966, 1962; Lewis & Roberts, 1956)). Although hybrid sterility in some animals has been mapped to genes (e.g. male hybrid sterility in Drosophila (Coyne & Orr, 1998)), plant systematists have tradi-tionally relied upon cytological differences to explain reductions in hybrid fertility between closely-related species (Rieseberg, 2001), while ignoring potential genie effects on reproduction. For example, many plant hybrids with reduced fertility 8 recover full fertility by chromosornal doubling, suggesting that cytological factors prevent diploid hybrids from being fully fertile. In many of the genera of California tarweeds (Asteraceae: Heliantheae-Madiinae), including Holocarpha, Haplopappus, Layia, Madia, Hemizonia and Calycadenia, lineages with unique chro-mosomal arrangements exhibit variable reproductive effects, ranging from com-plete isolation to negligible reproductive effects (G.D. Carr, 1980,1977, 1975; R.L. Carr, & G.D. Carr, 1999, 1983; Clausen, 1951; Clausen et al, 1945). Among the tarweed species, Holocarpha virgata, H. obconica, and H. heermannii are three close relatives that have undergone much chromosomal repatterning in their evolutionary history (Clausen, 1951). The species are readily distinguishable by numerous cytological differences. Within each species, populations are found that differ by diagnostic chromosomal rearrangements. Clausen (1951) found that it was almost impossible to intercross cytologically dissimilar individuals, even from different populations within the same species, and when the rare interpopulational hybrid was formed, it was always completely sterile. Similarly, the two maritime species of Layia, L. platyghssa and L. chrysanthemoides, have partially homologous chromosomes based on patterns of chromosomal pair-ing at meiosis and exhibit parallel patterns of genetic linkage, suggesting a similar evolutionary history for these two species (Clausen, 1951). Nevertheless, they are completely unable to intercross either in nature or in experimental studies. In the genus Chaenactis chromosomal repatterning is, once again, associated with spe-ciation (Kyhos, 1965). The extant species Chaenactis glabriuscula has independently given rise to two new species, C. Jremontii and C. stevioides, by aneuploid reduc-tion in chromosome number, followed by reciprocal translocations and inversions. While the patterns of correlations among chromosomal differences and reproduc-tive isolation are striking, these examples by no means eliminate genie differences as potential barriers to reproductive isolation. One tarweed genus, Calycadenia, has been the focus of much biosystematic study in recent years (Baldwin, 1993; Baldwin & Markos, 1998; Bohm, Fong, Hiebert, Jamal & Crins, 1992; G.D. Carr, 1980, 1977, 1975; R.L. Carr, & G.D. Carr, 2000, 1983). In Calycadenia, the situation is much like that described for Holocarpha, where frequent chromosomal repatterning differentiates populations within and among species. For two species, Calycadenia jremontii and C. pauciflora, single chromosomal translocations or pericentric inversions among populations within each species correlate with reduced fertility of hybrid Fl progeny (G.D. Carr, 1980, 1977, 1975; R.L. Carr, & G.D. Carr, 2000, 1983), suggesting a possi-ble isolating role for chromosomal rearrangements in this group. Although cur-rent species designations may be incorrect (B. Baldwin & R.L. Carr, personal communications, January 2000), several cytological races, each corresponding to a novel chromosomal arrangement, have been described in each species (G.D. Carr, 9 1975; R.L. Carr, & G.D. Carr, 2000, 1983). Exhaustive biosystematic study, involving hundreds of greenhouse crosses, has enabled the Carrs to identify cyto-logical racial identities for each population they have collected. Aside from four unusual sites, only one cytological race is present in each population sampled by the Carrs. Few phenotypic differences differentiate the species, and there are no diagnostic phenotypic differences among the races within species. Since most of the karyotypic polymorphism in this group is not associated with phenotypic change, levels of genetic divergence are perhaps likely to be low, providing an ideal scenario for testing the notion that chromosomal rearrangements may play a primary isolating role in this group. Needless to say, many phenotypic effects of chromosomal aberrations, including their role in reproductive isolation, remain to be studied in plant systems. For the purposes of this project, I am interested in examining evolutionary consequences of the reciprocal translocations in Calycadenia. More specifically, I am interesting in developing molecular tools and using population genetic techniques to begin to determine the role that chromosomal rearrangements may play in driving divergence among cytotypes in Calycadenia fremontii and in C. pauciflora and between the two species. 1.5 Testing for Chromosomal Speciation in Calycadenia Chromosomal differences could be inferred to be of primary importance in driv-ing differentiation if it could be shown that there are very few genetic and eco-logical differences among reproductively isolated, cytologically distinct lineages in Calycadenia. Under the chromosomal speciation hypothesis, the genetic distance among reproductively isolated groups that share a common ancestor should be very low, while cytological divergence is great. Also, reproductively isolated groups should be separated by more cytological differences than groups that hybridize. However, if one assumes that taxa that are most recently diverged from a common ancestor are most genetically similar, and reproductive isolation is a by-product of overall genetic divergence involving numerous loci rather than a result of cytological differences, then recently genetically diverged taxa would be expected to form hybrids more readily than distantly diverged taxa irrespective of cytological differences. If hybridization could be shown to occur more readily between genetically diver-gent and geographically distant populations that share a common chromosomal arrangement, than between genetically similar and geographically proximal popu-lations that differ in chromosomal arrangement, this would be strong evidence that chromosomal rearrangements play the primary role in conferring reproduc-tive isolation between species. Clearly, analyses of genetically and cytologically 10 distinct populations can eas-ily be confounded by geo-graphic effects such as isola-tion-by-distance and by population substructuring. In order to gain a better understanding o f the direct role o f chromosomal diver-gence in reproductive isola-t ion, it is important to attempt to isolate the effects o f chromosomal differences from patterns that result from genie and ecological differences that may also affect levels and patterns o f divergence in a plant group. Calycadenia provides an ideal system i n w h i c h to test the theory o f chromosomal speciation i n plants. In this genus, chromosomal repatterning, w h i c h correlates w i th reduced interfertility between taxa, occurs frequently across a w ide geo-graphic range. Calycadenia also has the attributes o f a good biological model sys-tem: it is present i n good supply in nature, w i th easily accessible populations; as an annual, its generation time is short relative to many plants; it is fairly easily manipulated i n the lab, greenhouse, or field; and it lends itself we l l to numerous avenues o f study (biosystematic, molecular, ecological, etc). 1.6 The Study System 1.6.1 The Genus Calycadenia The genus Calycadenia (Asteraceae), a California endemic, comprises approximate-ly nine species o f annual, self-incompatible herbs, ranging in height from 10 to 100 c m (Hickman, 1993). Calycadenia (Greek for "cup gland") plants have unique T-shaped to tack- or saucer-shaped tar glands on the upper leaves, on the tips o f the phyllaries subtending the inflorescence, and at the tips o f the chaffy bracts i n some species (Figure 1.1). The i r generally resinous and sticky leaves and inflo-rescences presumably earned them, and other members o f the subtribe Madiinae, their much-deserved c o m m o n name o f "tarweed". In addition to being a useful taxonomic character, tar glands, as we l l as coarse hairs, are thought to confer pro-tection against herbivory (Baldwin, 1993). F lower ing begins as early as M a r c h for some species and continues through November ; March- f lower ing species are l i ke -ly winter annuals ( R . L . Carr, personal communicat ion, January 4, 2000). Figure 1.1 (a) Inflorescence detail of Calycadenia adapted from Hickman et al. (1993). (b) Dried specimen of Calycadenia Jremontii A . Gray. Note T-shaped tar glands at leaf tips. 11 The fruit, a cypsela, is produced by both ray and disk florets w h i c h are present i n all species of Calycadenia. R a y fruits o f Calycadenia are small, smooth and ovate w i t h hard seed coats and lack a pappus, whi le disk fruits are angled and taper to the base and usually have a pappus consisting o f a c rown o f thin, lanceolate scales (Figure 1.2). A l though a dormancy factor may be present in the seed coat, it appears that the availability o f water in the spring determines whether seeds germinate. D r . R . L . Carr has suggested that disk cypselae remain viable for only a couple o f years, whereas ray cypselae i n some species may persist for several years. A substantial seed bank may exist and may serve to reestablish popula-tions fo l lowing drought years ( R . L . Carr, per-sonal communicat ion, January 4, 2000). Figure 1.2 (a) Ray cypsela. (b) Disk cypsela. (Adapted from Hickman et al, 1993.) Figure 1.3 Range and diversity of the genus Calycadenia in California. Intensity of shading indicates the number of species in each county: darker shading corresponds to more species (1 to 5 species per county). The greatest species diversity occurs in the Great Central Valley, the center of origin for the genus. (Map from CalFlora Database Web Site: The central valley o f California is c o n -sidered the centre o f or igin for the genus and remains the site o f greatest diversity o f Calycadenia species. The genus extends throughout the California Floristic Province, w i t h the greatest density o f species and individuals in the northern counties o f California and into southern Oregon (Figure 1.3). Species i n the genus Calycadenia share the same pollinators, inc luding bees (Hymenoptera) in the genera Melissodes, Ashmeadiella, and Halictis and bee flies (Bombyliidae) i n the genera Exoprosopa and Villa ( G . D . Carr, 1975). These p o l -linators also visit plants in several genera that are closely related to, and sym-patric w i t h , Calycadenia. I have also observed small beetles crawling through the inflorescences o f C. fre -montii, although these may be seed predators rather than pollinators. Species i n Calycadenia are difficult to 12 Arnica mollis Hemizonia perennis Osmadenia tenelis 100 97 9t 100 Calycadenia mollis Calycadenia mollis C. truncata subsp. scabrella hoo 100 100 C. truncata subsp. truncata • Calycadenia hooveri I— Calycadenia villosa '—Calycadenia villosa Calycadenia apicata C. multi. subsp. bi. 66 100 if 86 C. hispida 54 100 ' C. multi. subsp. c. 1 0 0 |- C. oppositifolia C. oppositifolia -C. pauciflora "Elegans" 99 711 p C. pauciflora "Pauciflora'* C. fremontii "Lewiston" C. fremontii "Ciliosa" C. fremontii Figure 1.4 Minimum-length parsimony tree based on ETS & ITS sequence data. Redrawn from Baldwin & Markos (1998). Bootstrap values appear above branches. Names of cytological races appear in quotes after species names. While relationships among many species are clear, relationships of the cytological races within the Calycadenia jremontii- C. pauciflora complex remain unresolved. define. Plants that are morphologically and ecologically distinct often retain the ability to interbreed, thereby making morphological or ecological recognition of a species unreliable. Differential tolerance to water limitation or serpentine soil sub-strates may be ecological factors that play minor roles in distinguishing some species (R. L. Carr, personal communication, January 4, 2000). Also, variation in flowering time may provide some temporal isolation among species. However, the taxonomy of the genus relies heavily on karyotypic features, including chromosome number, and the structural configura-13 Races of Races of C. Jremontii C. pauciflora Ciliosa Elegans Corning Healdsburg Dry Creek Pauciflora (n=5) Fremontii Ramulosa Lewiston Tehama Pillsbury Wurlitzer Table 1.1. Cytological races of Calycadenia fre -montii and C. pauciflora. Al l races except "Pauciflora" are n=6. Most races are named for landmarks near their original collection sites, or for their collectors. Races "Elegans" and "Healdsburg", while traditionally allied with C. pauciflora, have most recendy been reconsidered as "large-headed" races distinct from other races in C. pauciflora (B.G. Baldwin, personal communication, January, 2002). Their appro-priate taxonomic status is currendy under review, and they may well be redefined as subspecies of C. pauciflora or as a new species. tion of chromosomes, as inferred from meiotic pairing in hybrids. Recent phylogenetic studies using variation in nucleotide sequences in the nuclear ribosomal internal transcribed spacer (ITS) (Baldwin, 1993) and external tran-scribed spacer (ETS) regions (Baldwin & Markos, 1998) have provided some insights into chromosomal evolution in Calycadenia (Figure 1.4). The basal chro-mosome number for the genus is n=7, although aneuploid species with n=6, 5, and 4 are known (G.D. Carr, 1977). Variation in chromosome number is a poor characteristic for distinguishing among species as species with different chromo-some numbers (e.g. some Calycadenia pauciflora, n=5 and C. fremontii, n=6) can often produce viable hybrids, whereas some species with the same chromosome number (e.g. C. oppositijolia and C. mollis, n=7) produce very few or no hybrids (G.D. Carr, 1977). Detailed relationships among aneuploid lineages remain large-ly unresolved, despite recent inroads made through analysis of D N A sequence variation (Baldwin, 1993; Baldwin & Markos, 1998). Genomic regions that have been used for sequence comparison (ITS and ETS regions) lack sufficient varia-tion to definitively resolve relationships (Baldwin, 1993; Baldwin & Markos, 1998). 1.6.2 The Calycadenia fremontii- C. pauciflora Complex Species within Calycadenia tend to cluster into complexes of a few morphological-ly, ecologically, and karyotypically intergrading species complexes. One such complex comprises two species, Calycadenia Jremontii, and C. pauciflora. These morphologically variable species are subdivided into at least 12 chromosomal races (mostly n=6, one n=5 due to chromosomal fusion), each unique in its karyotype due to novel reciprocal translocations or pericentric inversions (Table 1.1). The Jepson Manual of the Higher Plants of California (Hickman, 1993) describes the leaves of Calycadenia Jremontii (which includes the formerly recognized species C. ciliosa) and C. paudflora as "generally aromatic, alternate, sessile and linear." Flowers may be yellow or white to rose, and each inflorescence may consist of 1 14 Figure 1.5 Inflorescences of Calycadenia. (a) White-flowered morph of C. fremontii A. Gray, (b) Detail of inflorescence of C. fre -montii. Note the six ray florets, each with three deep lobes. Disk florets (2-20) are usually the same color as ray florets. Anthers are dark maroon to black, (c) Yellow-flowered morph of C. fremontii. (d) Inflorescence of Calycadenia pauciflora A. Gray. Note the reduction of ray florets and more slender stems as compared to C. fremontii. cr \ o r Figure 1.6 Range of Calycadenia fremontii A. Gray by county in California. C. fremontii populations in the present study are in (1) Siskyou, (2) Shasta, (3) Tehama, (4) Lake, and (5) Mendocino Counties. v Figure 1.7 Range of Calycadenia pauciflora A. Gray by county in California. C. pauciflora populations in the present study are in (1) Lake County, (2) Colusa County, (3) the border of Lake and Napa Counties, and (4) Sonoma County. 15 to 6 three-lobed ray florets and 2 to 20 disk florets (Hickman, 1993) (Figure 1.5). L ike other species i n the genus, both ray and disk florets produce viable cypselae. Calycadenia fremontii and C . pauciflora are restricted to fourteen counties i n north-ern California and share an overlapping distribution in only three o f these coun-ties (Figure 1.6; Figure 1.7). As is typical for the genus, these species are found in mesic to xeric habitats, and populations o f C . pauciflora are capable o f thriving on serpentine soils ( G . D . Carr, 1975). W h i l e Calycadenia fremontii is not associat-ed w i t h serpentine soils, it tends to be found in hard-packed clay soil in extreme-ly hot, dried out ditches i n fields or along roadsides, where spring rains w o u l d have provided ample moisture for germination and early growth (Figure 1.8). Based on crossing studies, populations o f these two species, as we l l as the remain-der o f the genus, have been found to be mostly monotypic for chromosomal arrangement, although they demonstrate an unusually high frequency o f successful establishment o f novel arrangements ( G . D . Carr, 1980, 1977, 1975; R . L . Carr & G . D . Carr, 2000, 1983). O n l y four polytypic populations in the Calycadenia fre -montii - C. pauciflora complex have been documented, yet even here few struc-turally heterozygous individuals were found ( R . L . Carr & G . D . Carr, 1983). In greenhouse studies, the fertility o f synthetic hybrids between the races is reduced relative to intra-racial crosses (Table 1.2). G iven the paucity o f heterokaryotypes found i n the field i n even the racially polytypic populations, the fitness o f the hybrid heterokaryotype is also inferred to be reduced in nature ( G . D . Carr, 1980,1977, 1975; R . L . Carr & G . D . Carr, 2000, 1983; R . L . Carr, personal corn-Figure 1.8 A typical collection site near Redding, CA. , lune 1, 2000. Such parched, dry soil is characteristic of Calycadenia fremontii habitat. Here, white flowered Calycadenia flourishes in dried up tractor ruts in a roadside ditch. 16 Inter/In tra-racial Cross Pollen Stainability (%) Minimum Mean Maximum Pauciflora X Pauciflora 32 93 100 Ramulosa X Ramulosa 76 86 93 Healdsburg X Healdsburg 68 80 95 Elegans X Elegans 39 87 99 Ciliosa X Ciliosa 93 98 99 Pillsbury X Pillsbury 90 90 90 Dry Creek X Dry Creek 49 84 99 Corning X Corning 33 84 99 Pauciflora X Ramulosa 6 16 19 Pauciflora X Healdsburg 0 10 16 Pauciflora X Elegans 3 8 21 Pauciflora X Ciliosa 1 1 1 Ramulosa X Healdsburg 0 7 11 Ramulosa X Elegans 2 10 25 Ramulosa X Ciliosa 1 2 3 Healdsburg X Elegans 39 53 64 Healdsburg X Ciliosa 9 19 29 Elegans X Ciliosa 23 49 73 Ciliosa X Pillsbury 1 1 1 Ciliosa X Dry Creek 17 31 44 Ciliosa X Corning 27 44 57 Pillsbury X Dry Creek 36 48 56 Pillsbury X Corning 16 35 57 Dry Creek X Corning 9 21 33 Table 1.2. Crossing data for cytological races of Calycadenia pauciflora A . Gray (G.D. Carr, 1975) and for cytological races of C . Jremontii A . Gray (R.L . Carr & G . D . Carr, 1983). Crosses involved 2 to 41 individuals, generating 1 to 30 first generation hybrid progeny. Pollen stainability is a measure of the percentage of pollen grains from individual F l hybrid progeny that stain with cotton blue stain, indicating that pollen is likely viable. This measure is used as an estimate o f hybrid male fertility. % pollen stainability was based on sev-eral hundred pollen grains (G.D. Carr, 1975). 17 Chromosome 1 Chromosome 2 D r y Creek 1 2 3 4 5 6 7 S T U V W X Y Z S T U V W C B A 1 2 3 4 5 6 7 Lewiston 1 2 6 5 4 3 X Y Z S T U V W 7 Ci l iosa • • • • • • • • C Z Z Z D i • mm 1 2 6 5 4 3 X Y Z A B C V W 7 7 W 6 5 4 3 X Y Z S T U V 2 1 Pillsbury H Z Z H M H C Z Z Z ] i • — Chromosome 3 A B C D E F G Z Y X D E F G A B C D E F G S T U D E F G zm:: :.:::zr: • A B C D E F G Figure 1.9 Hypothetical diagram of chromosomes repatterned in the formation of known structural races of Calycadenia fremontii. Only one homolog of each pair is shown. Each race contains three additional pairs that have not undergone detectable structural differentiation. (Adapted from R.L . Carr & G.D. Carr, 1983.) Healdsburg Elegans (a) V I h Ramulosa 1 • Tehama Pauciflora Lewiston 2&3 Cil iosa D r y Creek . 1&2 Cil iosa 2&3 . C o r n i n g (b) |1&2 Pillsbury Figure 1.10 (a) Proposed relationships of chromosomal races of Calycadenia pauciflora and race Ciliosa of C. fremontii based on maximum meiotic chromosome association from every possi-ble hybrid combination between the cytological races. Numbers represent the number of reciprocal translocations or pericentric inversions that separate the races. (Based on G.D. Carr, 1975.) (b) Summary of cytological relationships in Calycadenia fremontii based on mei-otic pairing of interracial hybrids. Each race is separated from each adjacent race by a single reciprocal translocation. The numbers between the races indicate the restructured chromo-somes that differentiate the races. (Based on R.L . Carr & G.D. Carr, 1983.) 18 munication, January 4, 2000). Dr. G.D. Carr (University of Hawaii) and Dr. R.L. Carr (Eastern Washington University) have proposed an evolutionary sequence of chromosomal rearrange-ments for the races of C. pauciflora and C. fremontii (Figure 1.9; Figure 1.10) (G.D. Carr, 1975; R.L. Carr & G.D. Carr, 1983). Nevertheless, relationships among the races remain confusing, and the Carrs, who have performed cytological analy-ses of Calycadenia, suspect the currently recognized species may be improperly defined evolutionary units (R.L. Carr, personal communication, January 4, 2000). At present, the only sure method for discerning the racial identity of any popula-tion in the Calycadenia Jremontii - C. pauciflora complex is through careful biosys-tematic study. Individuals of unknown race must be grown to maturity and intercrossed with individuals of known race to form Fl hybrids. Once the Fls reach maturity, meiotic chromosome preparations can be made from pollen, and pairing anomalies diagnostic of particular interracial chromosomal configurations can be observed. Quite obviously, this is painstaking, slow work. It is hardly surprising, then, that the Carrs have dedicated decades to the careful documenta-tion of all known chromosomal configurations in this group and have typified numerous natural populations according to their meiotic pairing properties when intercrossed. 1.7 Using Molecular Markers to Describe Natural Populations The advent of molecular techniques involving the Polymerase Chain Reaction (PCR) holds great promise for resolving relationships within the Calycadenia jre -montii - C. pauciflora complex. One technique that is particularly useful for gen-erating molecular fingerprints and for describing population genetic differences among closely related, recently diverged taxa is Randomly Amplified Polymorphic D N A (RAPD) PCR (Williams, Kubelik, Livak, Rafalski, & Tingey, 1990). Randomly Amplified Polymorphic D N A relies upon the ability of short (10 bp) D N A primers of arbitrary sequence to bind to homologous regions in the target genome. If primers adhere to complimentary strands in the appropriate orienta-tion (e.g. with the 3' ends toward each other) at a reasonable distance (100 base pairs to 4000 bp apart), the intervening fragment will be amplified via the Polymerase Chain Reaction. The set of fragments amplified in the presence of each primer can be separated by agarose gel electorphoresis and visualized under ultraviolet light after staining with ethidium bromide. In the present investigation, RAPD markers were used to describe genetic differ-ences among populations and among individuals within 23 populations spanning the entire geographic range of Calycadenia fremontii and Calycadenia pauciflora. 19 Dendrograms based on levels of genetic similarity between individuals and popu-lations were used to reconstruct the possible evolutionary history of the popula-tions. The cytological racial identities for each of these populations was mapped onto these trees. In this way, correlations between genetic distances and cytolog-ical differences could readily be visualized. To assess the relative importance that each factor may play in reducing fertility between the karyotypically distinct races in this group, the significance of correla-tions among genetic distance, cytology, and geographic distance was examined. Despite results from biosystematic studies (G.D. Carr, 1980,1977, 1975; R.L. Carr & G.D. Carr, 2000, 1983) and from D N A sequencing (Baldwin, 1993; Baldwin & Markos, 1998) suggesting that chromosomal rearrangements provide a repro-ductive barrier among groups in Calycadenia, the present population genetic and genetic distance studies indicate that genetic distances among populations in the C. fremontii - C. pauciflora complex do not correlate with cytological differences and suggest that chromosomal differences among populations are not important barriers to gene flow. 20 2. Methods 2.1 Sources of Material 2.1.1 Field Collections Field-collected specimens constituted the bulk o f the material used i n this project. 460 plant samples, representing 23 populations (20 individuals per population), and 9 cytological races from sites throughout Nor thern California were collected i n the summer o f 2000 (Table 2.1; Figure 2.1). M y collection sites and times were based upon the collection records o f D r . Rober t L . Car r since each o f his field-collected populations has been cytologically typed through several years o f painstaking biosystematic study ( G . D . Carr, 1980,1977, 1975; R . L . Car r & G . D . Carr, 2000, 1983). Rober t Carr graciously provided me w i t h his field collection records for all populations o f Calycadenia fre -montii and C . pauciflora identified by himself and D r . Gerry D . Carr, so that I w o u l d be able to collect specimens o f k n o w n racial identity, wi thout having to perform cytological identification myself. Based on his extremely detailed and accurate collection records, I was able to precisely locate many o f the populations described by Rober t Carr. Populations tended to be discrete and widely disjunct, often separated by at least tens o f miles. \ C. fremontii "Pillsbury": 22,24, 23? C. fremontii Ciliosa'. 17,19, 25, 21 ? C. fremontii Xorning": 9,11 C. fremontii "Dry Creek": 10.12,13 C. fremontii 'Fremontii": 4, 5, 6 .10 C. pauciftora "Healdsburg": 28, 29 • J t J C. pauciftora "Pauciflora*: 14b C. pauciftora ""Ramulosa": 18, 20 Figure 2.1 T.E. Olson collection sites for Calycadenia Jremontii and C. pauciflora. 21 A few populations were inaccessible as they are n o w located o n private land, whi le others have been destroyed through land development. Ultimately, I was able to collect 22 populations, 20 o f w h i c h had been previ -ously characterized by the Carrs. F r o m each population, 20 plants were sampled. This Figure 2.2 A typical roadside collection site for Calycadenia sample Size was fixed by fremontii, east of Redding, CA. , June 1, 2000. laboratory and computation-al constraints, not by popu-lation size. For some popu -lations, 20 individuals represented nearly every individual in the population, whi le for others, this sample l ikely consisted o f less than 1% o f the plants i n the popula-t ion (Table 2.1). Almost all populations grew along roadsides, in ditches, and throughout adjacent fields (Figure 2.2). Plants were collected haphazardly along a crude transect run -n ing parallel to the road. A t least one whole representative plant from each pop-ulation was collected and pressed to serve as a herbarium voucher specimen. These have been deposited in the herbarium at the Universi ty o f Bri t ish C o l u m b i a . O n l y plants that had set seed were collected. Provided there were at least a few hundred plants in a population, whole plants were harvested. Where populations were extremely small, only a few mature seed heads, some leaves, and flower buds were harvested. Mature seed heads were removed and placed i n paper seed envelopes. These were stored i n the dark at r o o m temperature. T h e remaining plant tissues were tightly packed into 3 m L cryogenic collection vials and stored i n a l iqu id nitrogen-charged cryoshipping canister whi le i n the field, and in a minus 80° C freezer at the Universi ty o f Bri t ish C o l u m b i a for long term storage. 2.1.2 Herbarium and Greenhouse Collections D r i e d whole plant tissues, including stems, roots, leaves, flower buds, and fruits were donated by D r . Robe r t L . Car r (Eastern Washington University, Cheney, W A ) from collections made from specimens in his greenhouse. Robe r t Carr's greenhouse specimens were propagated from seeds o f Calycadenia that were field-22 (NJ JU - Q CO Additional Comments previously uncollected site; morph. identical to pop. 5 &6 Small, much-branched, growing in dry hardpan soil. Mosdy done flowering. Estimated Size of Population >1000 OOOK OOOK 100-500 OOOK OOOK OOOK 001> Elevation 250 m 300 m 200 m 1100 m 300 m 250 m 250 m 250 m Coordinates z P ?3 z P © y z P it z P z P in -?3 z P C N z P r* ~ z P L o S P Location S side Hwy 44, 200 m E of Bear Creek bridge, ca. 3.2 km W of Hwy A-17 W side of Hwy A-17 (Ash Creek Rd); 1.5 km S of jet with Dersch Rd. ~7 km S of jet with Hwy 44. E side Milville Plains Rd at Boscom Rd. 5 km E of Callahan on N and S sides of Gazelle/ Callahan Road. 1.5 km S of Aoudad Dr. on Hwy A-17 (Dersch Rd). 3.4 km S of WUdcat Rd (goes to Black Butte). 4 km E of Balls Ferry Resort on Hwy A-17, E. of Cottonwood In dry irrigation ditch behind the home of Mis. Nell Pope, 1228 Pinon Dr., Anderson, CA. 8 km W of jet with Hwy AS on Hwy 36, W of Red Bluff County Shasta 3 J m Shasta Siskiyou Shasta Shasta Shasta Tehama Flower Color V •3 • Is white white yellow yellow yellow yellow yellow Cyto-logical Race fremontii :a a o S coming dry creek coming dry creek dry creek (A 1) • • 4 fremontii tin := := := := Spec s fremontii fremot fremot 5 a. rt nk T C N o oo 7/06 ago "2" o 0 0 83/0! 78/01 i0/08 78/0' 78/01 78/0 die 2228 2250 2131 2203 2121 2135 2105 TEO Collect. Date 00/06/01 00/06/01 00/06/01 00/08/11 U/80/00 00/08/12 00/08/12 00/08/12 O n - . ui .tS D in O N o - C N C O :5 3 "O T3 O '5 23 Additional Comments Site 14 on S side of Hwy, 15 on N. Treated as 14b in subsequent analyses. No obvious mature disk achenes. No obvious mature disk achenes. No obvious mature disk achenes. Few plants, crowded out by Centauria solstitalis. Disk achenes present. White form like site 17, 18, 19 & at roadside. Yellow form up hillside. Large population; in dried out flat-bottomed creek o lake bed. previously uncollected site; morphologically like 24; no disk achenes Small population at edge of road, in front of construction yard. "3 S » *a .S o CQ « o o o o o o o o o in i o o o CJ o o H 8 3 n fi A 1 A 1 "A A* V oo A* V V a 0 •a A E o o E o H o s o E o E o S o E o £ o tu in •*r cn in cn W Coordinates z P B e s3 z P S3 z P s P * in ? ^ S 3 z P S3 z P s3 z P s3 z P ga S3 z P ^ b s3 z P S3 Location At Colusa & Lake Co line on N side of Hwy 20. On S side Hwy 29, 4.5 km N of 11th St exit from Lakeport, at exit to county jail (Hill Rd @ Park Way) W side of Hwy 29, 4 km S of 11th St exit from Lakeport. Near jet of Hwy 29 & Hwy 175. NW side of Hopland Rd (Hwy 175), 2 km from jet with Hwy 29. 2.7 km SW of Kelsey Creek Dr on NW & SE sides of Wight Way, at entrance to Deer Run Ranch. 3.5 km S of Wight Way on W side Kelsey Creek Dr. Growing in ditch, and up steep embankment. 1.5 km W of Scott Dam on Elk Mountain Rd; at jet to Lake Pillsbury Resort. W side. Mile 26, Elk Mountain Rd, just S of Little Squaw Valley Ranch Rd. East Side Potter Valley Rd @ Pine St. County Colusa u 3 Lake Lake Lake Lake J Lake Mendocin 0 Flower Color white white white white pale yellow? White? yellow & white yellow white white Cyto-logical Race pauciflora ciliosa ramulosa ciliosa ramulosa ciliosa pillsbury pillsbury Species 1 a. fremontii pauciflora fremontii pauciflora fremontii a o E ' fremontii fremontii ~£ A 18/20 18/19 )8/19 )8/19 )8/20 )8/20 2 o Q <—> 0 0 OO" 00 06" oo" oo" H •- S 2 «« M 2152 O OS Q o 2147 CJ -, Q in O ^ 2146 2145 2150 2149 _ <J 0 u » W A •*r •»>• in in in oo" 0 0 oo" oo" g" o" cT oo" O O" O o o o o o" o o B" o o" o o o" o o" o o" o o" o 2 .So - ~ I 00 ON o tN CN CN CN CN 24 (NJ -Q Additional Comments Very large population throughout fields. On gravel runaway lane, under pine canopy. Very • delicate inflorescence. Morph. like pop 27, but inflorescence larger, more robust. Like pop 28. Estimated Size of Population OOOK OOOK <100 OOOK 100-500 Elevation 400 m 400 m 1300 m £ o o 400 m Coordinates kS zP L, S P zP s3 zP . O N ? " zP kS Location 10 km NW of Lakeport near 4550 Scotts Valley Rd. In ditch & up hillside on E side of road. W side of Hwy 29 @ Grange Rd. Napa & Lake Co. line on W side Hwy 29. 3.6 km N (up) Ida Cbyton Rd from Hwy 128. Beautiful view of wine countryl! N side Mill Creek Rd at jet with Mill Creek Lane; 2.9 km SW of Healdsburg. County Lake Lake a rt z Sonoma Sonoma Flower Color yellow white white white white Cyto-logical Race ciliosa elegans u u a - O •3 6 0 TJ u JC hcaldsbur g Species c E pauciflora pauciflora pauciflora f o. RLC Collect. Date 78/08/19 78/08/19 78/08/19 78/08/19 78/08/19 2148 2143 2142 2141 2139 TEO Collect. Date 00/08/15 00/08/15 00/08/15 00/08/15 00/08/15 0 « -W D I N *o I N I N 0 0 C N < N 25 collected by himself and by Dr. Gerry D. Carr (University of Hawaii, Manoa, HI) between 1975 and 1998. These voucher specimens represented 25 disjunct populations comprising 12 cytological races of Calycadenia pauciflora and C. fremon -tit identified by Robert and Gerry Carr. Dried voucher specimens from Robert Carr were stored at room temperature in envelopes in a dry, dark cupboard for the duration of this project. Vouchers of dried material donated by Robert Carr contained several viable seeds. For each of the 25 populations for which he donated vouchers, both ray and disk cypselae were germinated on filter paper soaked with distilled water, after rinsing the cypselae with 2% sodium hypochlorite bleach. Ray cypselae required stratification in order to germinate, but disk cypselae germinated readily without any special treatment. The petri dishes containing the germinating cypselae were covered but not air-sealed, wrapped in wet paper towels, and stored at room temperature in an airtight container in the dark for up to ten days. Etiolated seedlings in petri dishes were exposed to light after cotyledons had fully emerged. After 12 - 24 hours under lamps in a plant growth stand, seedlings were transferred to potting mix in 4" greenhouse pots using sterile for-ceps. Potting mix consisted of 1/3 sterilized sand, 1/3 perlite, 1/3 garden topsoil (not sterilized) plus 1 teaspoon 14-14-14 generic slow release granular fertilizer per 4" pot. Pots were kept in standing tap water in trays in a 25° C growth chamber where they were exposed to 24 hours constant light. This temperature was probably slightly below the optimum for maximum growth and vigor (R.L. Carr, personal communication, January 4, 2000). 2.2 DNA Isolation and Quantification Tarweeds have extraordinarily high levels of secondary compounds, pectins and other polysaccharides (Bohm et al, 1992), which made D N A isolation one of the most problematic aspects of this project. In order to maximize both quality and quantity of isolated DNA, numerous protocols were tested using a variety of fresh, frozen, and dried plant tissues, including leaves, stems, roots, flowers, and embryos. Protocols and systems tested included a modified version of the hexa-decyltrimethylammonium bromide (CTAB) D N A isolation technique (J. J. Doyle & J.L. Doyle, 1987) and the Qiagen DNeasy Plant Mini Kit (Qiagen Inc., Mississauga, ON). A modified CTAB (J. J. Doyle & J.L. Doyle, 1987) method seemed promising initially as the volume of D N A precipitated was high. However, when PCR was attempted, I found that the final CTAB product was of poor quality for RAPD-PCR amplification. Qiagen's DNeasy Plant Mini Kit provided a more useful 26 isolated product as quality was adequate for consistent, stable RAPD-PCR ampli-fication, despite yields that were lower than those obtained using the CTAB method. 2.2.1 Qiasen Extractions D N A was isolated from various plant tissues using the Qiagen DNeasy Plant Mini Kit. Some modifications to the recommended protocol were necessary to opti-mize D N A extraction. For dried material, approximately 0.10 g of stem, leaf, or flower bud tissue was ground in liquid nitrogen. For fresh material, a 0.5 cm z leaf disk was harvested from basal rosette leaves when plants were at about the 10 to 15 leaf-stage (about one week old) and ground in liquid nitrogen. 400 uL Buffer API was immedi-ately added to ground tissue before it thawed. Tissue was mixed with buffer to form a slurry, which was carefully transferred from the mortar to a labeled 1.5 mL microfuge tube. 4 uL RNAse was added, contents vortexed vigorously, then incubated at 65° C for 10 minutes to lyse cells. Embryonic tissue was obtained by excising a plump, viable embryo from a single ray cypsela from each plant. Single embryos were ground in 100 uL Buffer API at room temperature in a 1.5 mL centrifuge tube, using a micropestle. Once ground, another 300 \ i L buffer plus 4 uL RNAse was added, and the contents were vortexed and incubated to lyse cells as described above. Following incubation, protocols for the two tissue types were identical. A vol-ume of 130 uL Buffer AP2 was added to precipitate detergent from Buffer API, and to precipitate proteins and polysaccharides released during lysing. A 5 to 10 minute spin of the microfuge tube at 13,000 rpm was usually necessary to make a pellet of the thick precipitate. After spinning, lysate was transferred to the QIAshredder column, and spun two minutes at 13,000 rpm to pass lysate through the QIAshredder membrane. This filtered out remaining precipitate. If the extract became highly viscous upon addition of Buffer AP3/ EtOH, contents were transferred to the DNeasy mini spin column, which was then spun two minutes at 10,000 rpm rather than the recommended one minute at 8,000 rpm. If the extract still did not pass through the column membrane, tubes were spun again for two minutes at 10,000 rpm, and so on until most of the extract had passed into the collection tube. D N A was eluted from the DNeasy mini spin column membrane using two 50 LIL washes of Buffer AE at 65° C, allowing the membrane to soak in Buffer AE at room temperature for 20 minutes prior to spinning. Final elution volume was 100 LIL. D N A was stored at -20° C for the duration of this project. 27 2.2.2 CTAB Extractions D N A was isolated from dried leaf and stem tissue, fresh leaf tissue, and frozen whole plant tissue using slight modifications to the CTAB total D N A isolation protocol for fresh plant tissue (J. J. Doyle & J.L. Doyle, 1987). Unless otherwise noted, all steps were carried out at room temperature in a fume hood. For dried material, approximately 0.25 g of leaves or stems were used from voucher specimens donated by Dr. R.L. Carr. For extractions on fresh material, 1 or 2 basal rosette leaves were removed from greenhouse grown plants at the 1 to 2 week stage. Extractions on frozen leaves, stems, flowers, roots, and seeds were done in both high tissue to buffer volumes ("small volume" extractions) and low tissue to buffer volumes ("large volume" extractions). Small volume extrac-tions typically used one gram frozen tissue in 7 mL 2X CTAB buffer, while large volume extractions used only 0.25 g tissue in 7 mL buffer. Tissue was ground in liquid nitrogen. 7 mL of 60° C 2X CTAB buffer (1 M Tris-HCl, pH 8.0; 5 M NaCl, 0.25 M EDTA, pH 8.0; CTAB; PVP-40; sodium bisulfite, 2-p,-mercaptoethanol) was added to frozen, powdered tissue, and mixed to make a slurry. The slurry was transferred from the mortar to a 15 mL Corex centrifuge tube, and incubated at 60° C for 30 minutes, inverting occasionally to mix contents and ensure even digestion and lysing of cells. Following incubation, 2/3 volume of 24:1 chloroform:isoamyl alcohol was added to wash out organic contaminants. Tubes were sealed and inverted and shaken vigorously to form an emulsion. Emulsified lysate was spun in an IEC clinical centrifuge (Thermo International Equipment Company, Neeham Heights, MA) at 4000 rpm for 10 minutes. After spinning, the aqueous top layer containing D N A was removed with a wide-bore pipette tip and transferred to a clean 15 mL Corex test tube. Again, 2/3 volume of 24:1 chloroform:isoamyl alcohol was added, and the contents were emulsified and spun for 10 minutes. After the second spin, the aqueous layer was transferred to a clean 15 mL Corex test tube, to which one volume ice-cold 100% isopropanol was added to precipi-tate DNA. The tube was sealed and inverted to mix contents thoroughly. Once isopropanol had been added, tubes were left in the -20° C freezer overnight (8-12 hours) to fully precipitate DNA. For many samples, high quantities of excess polysaccharide in extracts from dried, fresh, or frozen tarweed tissue necessitated modifications to the CTAB protocol to reduce polysaccharide. Following the two chloroform washes, and prior to addition of isopropanol, isolates that were particularly polysaccharide-rich were 28 resuspended in 7 mL 60° C CTAB buffer, and incubated at 60° C for 10 min-utes. After incubation, the resuspended D N A and contaminants were washed in 2/3 volume of 24:1 chloroform:isoamyl alcohol. After a 10-minute spin at 4000 rpm, the supernatant was withdrawn, and processed as usual, with the addition of ice-cold isopropanol to precipitate DNA. Following several hours incubation at -20° C, tubes were again spun in the clini-cal centrifuge for 10 minutes to form a D N A pellet. Supernatant was discarded, and the pellet was soaked with 5 mL of 76% EtOH / 0.01 M N H 4 O A c for 10 minutes, after which the supernatant was again discarded. Samples were dried 10 minutes at 40° C to evaporate any remaining ethanol. Finally, pellets were suspended in 400 LIL T E and incubated at 37° C until fully dissolved. Dissolved D N A was purified using the S&S Elu-Quik D N A Purification Kit (Schleicher & Schuell, Keene, NH) to remove residual contami-nants, such as polysaccharides and proteins. Re-precipitating D N A in ethanol was an alternative method of purification that was attempted with little success. 2.2.3 Qualitative and Quantitative Verification of DNA Product Regardless of isolation protocol, all D N A products were quantified using a Hoefer DyNAQuant 200 fluorometer (Amersham Biosciences, Baie d'Urfe, PQ). Two LIL D N A was added to 2 mL of fluorometry solution (90 mL ddH 2 0, 10.0 |xL commercially prepared Hoechst dye H33258, and 10 mL 10X T N E buffer), vortexed, and assayed in a crystal cuvette in the fluorometer. Fluorometry read-ings at or above 10 ng/LiL were considered ideal, although some D N A samples with readings as low as 2 ng/jiL were sufficient to produce a clear, stable, repeat-able RAPD-PCR product. For D N A samples with fluorometry readings above 10 ng/[i.L, 50 LIL aliquots standardized to 10 ng/LiL were prepared by diluting the appropriate volume of D N A with T E . Relative sizes of fragments of D N A were assessed by running 5 \iL of raw, undi-luted D N A extract on a 1.5% agarose gel in 0.5X TBE buffer. Gels were stained with ethidium bromide and visualized under U V light using an Alphalmager 1200 Documentation and Analysis System and Multilmage Light Cabinet (Alpha Innotech Corporation, San Leandro, CA). Fragments of at least several hundred kilobases are necessary to obtain reliable RAPD-PCR products. Thus D N A pro-ducing a gel that showed only smears of small fragments was discarded, whereas a gel with a bright band or blob at the top was ideal. Gels served a secondary purpose, which was to verify fluorometry readings: brighter bands on the gel cor-roborated higher fluorometry values (i.e. greater D N A concentration). This visu-al assay was particularly useful for very low fluorometry readings (1 or 2 ng/uL) to ensure that at least some D N A was actually present in the sample. 29 2.3 RAPD- PCR and Gel Electrophoresis 2.3.1 PCR Conditions The RAPD-PCR cocktail consisted of 10 ng template DNA, 48 mM Tris-HCl, 120 mM KC1, 4.8 mM MgC12, 0.1 mM of each dNTP, 0.2 mM 10-mer primer, and 1.5 units Taq polymerase in a 25 \ i L volume. Amplification was carried out in a 60-well PT-100 thermal cycler (MJ Research, Inc., Waltham, MA), using an initial denaturing period of 4 minutes at 94° C, followed by 45 cycles of 1 minute denaturing at 94° C, 30 sec annealing at 37° C, and 90 seconds exten-sion at 72° C, and a final extension step of 5 minutes at 72° C. Following amplification, samples were incubated at 4 ° C for at least 10 minutes or until electrophoresed. RAPD-PCR products were electrophoresed on 1.5% agarose gels in 0.5X TBE buffer, pH 8.0, run in 0.5X TBE buffer (pH 8.0) for about two hours at 180 V. Gels were dry loaded so that wells were filled with PCR product only and no loading dye. Ladder lanes, containing loading dye for tracking the progress of the electrophoretic front, were loaded in the first and last wells of each gel. "Dry" gels were electrophoresed for 5 tolO minutes, so that D N A product migrated into the gel, then were flooded with TBE buffer. Electrophoresed gels were stained with ethidium bromide solution for 15 minutes and then floated in tap water for 15 minutes to remove residual stain. Gels were visualized and photo-captured under U V light using an Alphalmager 1200 Documentation & Analysis System and Multilmage Light Cabinet. A single representative D N A sample from each of the 25 populations for which Dr. Robert Can-had provided voucher specimens was used to screen RAPD primers for usability in subsequent analyses. Approximately 80 10-mer primers from the University of British Columbia's NAPS unit were screened. Of these, six primers that produced U B C RAPD Sequence Number of Primer Scorable Bands 5 C C T G G G T T C C 12 20 T C C G G G T T T G 15 21 A C C G G G T T T C 16 44 T T A C C C C G G C 16 47 T T C C C C A A G C 14 63 T T C C C C G C C C 15 Table 2.2. R A P D primers used to generate D N A fingerprints for 460 individuals in the Calycadenia Jremontii - C. pauciflora complex collected from 23 sites throughout Northern California. 30 POPULATION 21, UBC PRIMER #5 P O P U L A T I O N 1 4 , U B C P R I M E R # 2 0 POPULATION 28, UBC PRIMER #21 IMJ P O P U L A T I O N 14. U B C P R I M E R #44 P O P U L A T I O N 15, U B C P R I M E R #47 Figure 2.3 Sample data gels for each of the six U B C R A P D primers used in this proj-ect. Negative control lanes on some gels are marked " N O DNA". Each data lane represents one field-collected individual. consistent, bright, easily-scorable bands were selected for final analysis. Resultant bands were also highly polymorphic both wi th in and across populations; across all six primers, only two bands were produced that were monomorphic across all populations. Each o f the six primers yielded about 15 reliable bands (Table 2.2). 2.3.2 Scoring Gels and Testing for Homology of Marker Bands Gels were scored manually, w i t h presence o f bands being scored as 1 's and absence o f bands as 0's. Bands were scored only i f they were relatively sharp (Figure 2.3), reproducible, occurring i n at least two individuals per population, 31 and homologous within the population. To reduce bias attributed to the domi-nant nature of RAPD markers in subsequent analyses, band loci were retained as useful markers only if the frequency of the null allele (no band) exceeded 3/n, where n is the sample size of the population (Lynch & Milligan, 1994). Homology of scored bands within and across all populations must be established if one aims to measure genetic distance or similarity across populations. One of the methods for testing band homology involves the analysis of patterns of restriction enzyme digestion of PCR products (Rieseberg, 1996). Restriction enzyme analysis involves performing complete digests on PCR products using frequent cutting enzymes (4-, 5-, or frequent 6-cutters). If PCR products are of a reason-ably large size (several hundred base pairs), there is a good chance that at least one restriction site will be present in the fragments. Thus, digestion will produce a distinctive pattern of cleaved fragments for each PCR amplified band. Identical restriction fragment patterns are evidence for homology among bands (Rieseberg, 1996). From each of my 23 populations, I selected two individuals representative of as much of the total polymorphism in the population as possible. I repeated ampli-fication of these D N A samples using each of the six RAPD primers. Each PCR product was divided into four aliquots: an uncut "control" sample and one for digestion with each of three restriction enzymes, BSTU-I (a 4-cutter), Taqa-I (another 4-cutter), and Hinf-I (a 6-cutter) (New England Biolabs, Mississauga, ON). Each sample was incubated at its optimal digestion temperature for one hour, then electrophoresed on an agarose gel alongside the control for easy com-parison of patterns of fragment digestion. Bands that were not clearly homologous between populations based on patterns of restriction enzyme digestion were discarded. If restriction patterns revealed that bands in two populations represented two similarly sized fragments with dis-tinct digestion patterns these bands were retained. In this case, each of these was converted to a new marker. For example, a 1 kb fragment may have appeared in several populations, and in some populations the fragment cleaved into two frag-ments following digestion with BSTU-I, two fragments with Taqa-I , and three fragments with Hinf-I, while the fragment did not cleave with any of the enzymes in the other populations. Thus, the 1 kb fragment that cleaved repre-sented one marker locus, while the non-cleaving fragment represented a different locus. Homologous bands that appeared at very low frequency (e.g. one or two individuals in one or two populations) or whose homology was ambiguous were discarded from the data set. 32 2.4 Data Analysis 2.4.1 Relationships Among Individuals The first step in analyzing the entire molecular data set derived from my field-collected samples was to establish the relationships among individuals within and among populations. Relationships of individuals within and among populations were assessed for evidence of population cohesion. One method for estimating the relatedness of individuals within or across popula-tions is to calculate the genetic distance between them. Genetic distance can be determined using a number of different methods, ranging from a simple count of the absolute differences among pairs of taxa (Euclidean distance) to slightly more sophisticated models that make assumptions about rates of evolution of loci, rela-tionships among loci, and the evolutionary dynamics of the population under investigation. The phylogenetic analysis software package, PAUP* 4.0b8a (Swofford, 2002), measures genetic distance by simple counts of either total char-acter differences or mean character differences between pairs of taxa. The binary data matrix resulting from the RAPD profiles of 460 individuals col-lected from 23 populations was converted to a NEXUS-format file using MacClade 3.07 (Maddison & Maddison, 1997) then analyzed in PAUP* 4.0b8a (Swofford, 2002) with the optimality criterion set to distance. PAUP* 4.0b8a (Swofford, 2002) was used to compute pairwise genetic distances from a pres-ence-absence matrix of RAPD bands for 460 individuals collected from 23 popu-lations. Total character differences between pairs of taxa were used to calculate genetic distances, and these were then used to derive Neighbor-Joining trees in PAUP*. 2.4.2 Verification of Population Racial Identity Before proceeding with analyses of genetic and cytological relationships among populations based on scored RAPD bands from my field-collected specimens, it was necessary to verify the racial identity of each population. My collections had been based on the collection records of Dr. R.L. Carr, and while in most cases there was little ambiguity regarding the locality, I tested that my collections were drawn from the same populations as his using RAPD markers. I compared RAPD marker profiles for two randomly chosen D N A samples from each of my field-collected populations against two D N A samples from the corre-sponding R.L. Carr collection. For each set of four DNAs, I ran RAPD-PCRs for all six primers and scored the resultant gels. The relationships of my field-collected samples to Dr. R.L. Carr's samples were 33 visualized by using the RAPD data to generate a tree based on genetic distances among individuals using PAUP* 4.0b8a (Swofford, 2002). The binary data matrix resulting from the RAPD profiles of Dr. Carr's greenhouse specimens and my field samples was converted to a NEXUS-format file and analysed in PAUP* as described above. Relationships among the D N A samples were inferred from a PAUP* Neighbor-Joining tree, in which the most similar D N A samples shared the shortest branches. Adjacent clustering of Dr. R.L. Carr's populations to my populations was interpreted as a match between the two samples. 2.4.3 Relationships Among Populations In a similar way that relationships among individuals were assessed in a phyloge-netic context, patterns of populations' relationships were also revealed through systematic analysis. Relationships among populations were examined for evidence of clustering by race or by other obvious inter-group differences. Although PAUP* was able to generate trees to show relationships among all the individuals in all populations, PAUP* could not generate trees based on popula-tion summaries (e.g. allele frequencies) alone. Testing relationships among popu-lations without first summarizing variation among individuals complicates the der-ivation of statistical support for any resultant groupings. The phylogenetic soft-ware package PHYLIP (Felsenstein, 1995), however, can handle population-only data based on summaries of data from their component individuals. PHYLIP can also make estimates of statistical support for the trees it creates. So while PAUP* was used to assess relationships among individuals within and among populations, PHYLIP was used to assess relationships among populations. First, the binary data matrix based on scored RAPD bands was converted to a matrix of estimated allele frequencies. Allele frequencies were estimated by assuming all markers were in Hardy-Weinberg equilibrium and then calculating the frequency of the null allele from the frequency of the null RAPD genotype (lack of band). The programs GENDIST, FITCH, and the neighbor-joining option of NEIGHBOR in PHYLIP were then used to generate, respectively, a matrix of genetic distances between populations based on allele frequency esti-mates, Fitch-Margoliash trees, and Neighbor-Joining trees. Trees were unrooted, and a constant rate of evolution was not assumed (Felsenstein, 1995). Genetic distance was calculated in GENDIST according to the methods of both Nei (1972) and Reynolds, Weir, and Cockerham (1983). Nei's (1972) distance is based on an infinite isoalleles model that assumes mutation is neutral, occurs at the same rate across all loci, and each mutation is to a new allele. Like many other models for genetic distance, Nei's assumes that genetic differences between populations arise through mutation and fix by drift. Nei's distance requires that 34 population size has remained constant through time, and it is in part due to this assumption that Nei's model predicts a linear rise in genetic distance over time (Felsenstein, 1995). In contrast, the distance model of Reynolds et al. (1983) does not assume mutation but relies solely on drift to alter allele frequencies among populations. This model has the added advantage of accounting for fluctuations in population size through time, since larger populations are likely to experience drift more slowly than smaller populations. Reynolds et aVs model may be more appropriate than Nei's for evolution in the short term, where the incidence of mutation is expected to be low. The distance trees provide an aesthetically pleasing, simple way of visualizing genetic distances. FITCH and the neighbor-joining option in NEIGHBOR make equivalent assumptions regarding the nature and distribution of the genetic distance estimates that are used to generate trees, and neither assumes an evolu-tionary clock, although they do differ in the way they build trees (Felsenstein, 1995). Nevertheless, for well-behaved data, both methods usually produce fairly similar trees (Felsenstein, 1995). To assess the strength of support for branches in the Fitch and Neighbor-Joining trees, bootstrap analyses were performed. The program SEQBOOT in PHYLIP was used to create a data set of 200 replicate matrices of allele frequencies, by randomly sampling (with replacement) values from the original matrix of estimat-ed allele frequencies. These bootstrap allele frequency matrices were converted to genetic distance matrices (Reynolds et al., 1983) in GENDIST, then distance trees for each genetic distance matrix were created in either FITCH or NEIGHBOR. Finally, a consensus tree with bootstrap support values for each branch was gener-ated from the 200 bootstrap distance trees. Values for bootstrap support repre-sented the percentage of times out of 200 that each branch appeared in distance trees created through the bootstrapping process. 2.4.4 Summary of Genetic Variation Within and Among Populations: Percent Polymorphism and F S T Given that populations represent distinct, cohesive genetic units, I described vari-ation within and among populations using population genetic methods. Descriptive summary statistics such as percent polymorphic loci (%P) and popula-tion differentiation (centroid F S T ) were generated for each population using the data analysis software package Tools For Population Genetic Analysis (TFPGA) (Miller, 2000). TFPGA calculates percent polymorphism based on direct counts of presence-absence variation at each locus. As a step in the Genetic Distance option, TFPGA calculates Reynolds, Weir, and Cockerham's (1983) coancestry coefficient, 6. The coancestry coefficient is 35 equivalent to Wright's F S T (Weir, 1990), and it is used to estimate the divergence among populations. 8 is related to genetic distance (D) by the following relation-ship: D = -In (1-0 ) (Reynolds et al, 1983). This measure of distance is most accurate for small pop-ulations over relatively short periods of time, as it relies on drift alone and not mutation to account for differences in allele frequencies among populations (Reynolds et al, 1983). Here, 9 itself, as an estimator of F S T , was used to estimate the degree of differen-tiation of each population, race, and species from the pooled total of all genetic variation. First, all populations except the population for which F S T was being tested were pooled to form one large, relatively variable population. Then, each of the 23 actual populations were, in turn, designated as the second population for which centroid F S T was then calculated. The measure of F S T obtained in this manner reflected the degree of differentiation of each "actual" population from a centroid rooting of all populations (Dr. K. Ritland, personal communication, January 29, 2002), hence the term "centroid F S T". The same procedure was con-ducted to estimate the degree of differentiation of each race and each species of Calycadenia. All populations of a given race, or all populations of races of a species, were pooled into one population, which was then tested against all other pooled populations of remaining races and species. By this method, larger values of centroid F S T reflect greater genetic differentiation of the local population than do smaller values. To verify and account for the patterns that emerged from phylogenetic analyses in PHYLIP (namely two well-differentiated clades with no internal resolution in either clade), TFPGA was used to estimate F S T between and within the two clades. Populations within each clade were pooled and F S T was calculated across all 88 RAPD loci for the two pooled populations. The average of these F S T esti-mates describes the degree to which the two clades are differentiated from each other. TFPGA within-clade F S T was calculated across 88 loci for all the (non-pooled) populations in each clade. The values generated in this manner indicate the average degree to which populations in the clade are differentiated from one another. 2.4.5 Mantel Tests for Correlations Among Data Matrices Alternatives to the hypothesis of chromosomal speciation were tested for their goodness of fit in explaining the observed distribution of genetic variation among populations, races, and species in the Calycadenia fremontii - Calycadenia pauciflora 36 complex. Factors that could be relevant to the evolution of reproductive barri-ers among the cytological races include the influence of geographic distance among populations or groups of populations and the actual number of chromoso-mal rearrangements (cytological distance) among cytotypes. Correlations between genetic distances between populations and differences in flower color and species identity were also tested. Therefore, I formed distance matrices for flower color, species identity, geographic distance, cytological distance, and genetic distance (Reynolds et al., 1983), and I tested for the strength and direction of correlations among all combinations of these matrices using simple and partial Mantel tests. Cytological differences between populations were inferred from counts of the number of translocations separating cytological races (Figure 1.10), as described by G. D. Carr (1975) and R.L. Carr & G. D. Carr (1983). Populations having the same flower color or belonging to the same species were given a distance of 0 and populations that differed for either factor were given a distance of 1. Tests were performed on the data set as a whole for species, flower color, genet-ic, geographic, and cytological distance matrices. After examining the results of the distance analyses from PHYLIP, two clades emerged (see Results: Figures 3.7 and 3.8). It then became necessary to perform the Mantel tests on each of these groups separately, to test for the influence of geography and cytological differ-ences on the pattern of genetic variation within each of the two clades. To statistically determine which of these factors, if any, contributed to the divi-sion of populations into the two groups that emerged in the distance trees, a test similar to a sign test but involving a phylogenetic correction to account for the split of the populations into two groups would have had to be undertaken. Unfortunately, I was not able to learn of such a test nor devise one myself. Instead I used Mantel tests to estimate the correlation between each of these fac-tors and the genetic distances between populations, irrespective of the clades defined in the distance trees. Given that genetic distances between populations ultimately led to the recognition of two clades, these clades are inferred to repre-sent the most genetically distant groups in my data set. If Mantel test results indicate a correlation between a given factor and genetic distance, then popula-tions that are most distant should differ the most in that factor. Therefore, any factor that is strongly correlated with genetic distance should (but does not neces-sarily have to) differ between clades, if the distance between the clades is the greatest genetic distance in the whole species complex. Obviously, this method only very indirectly estimates of the contribution of flower color, species differ-ences, and other factors in distinguishing the two clades that emerged in the dis-tance trees. Mantel tests were performed using the statistical software package Le Progiciel R 4.0 (Casgrain, Legendre, & Vaudor, 2001). This program performs simple Mantel 37 tests to estimate the correlation between a pair of matrices and partial Mantel tests to measure the degree of association between two matrices when controlling for a third matrix. The test statistic is the standardized Mantel statistic r M , a form of the original Mantel statistic z M , for which distance values have been standardized within each matrix (Casgrain et al, 2001). The statistic r M is equivalent to Pearson's r, the coefficient of linear correlation, such that negative values of r M that are significant for a= 0.05 reflect a negative correlation among the two matrices, and positive significant values reflect positive correlations. Although partial Mantel tests have been criticized for underestimating true Type I error (Raufaste & Rousset, 2001), two types of matrix permutations have been shown to reflect correct Type I error and have good power (Casgrain et al, 2001) in the partial Mantel test. The appropriate use of either of these permutation tests is discussed in the documentation for the software package. Initially, I ran partial Mantel tests using both types of permutation, but I observed no effective difference between their results over several partial Mantel tests. Thereafter, I used only the Type 1 permutation (permutation of the elements of the first matrix prior to computation). 2.4.6 Distribution of Per Locus F S T The last analysis involved a comparison of levels of divergence for each of the 88 RAPD marker loci. Again, TFPGA (Miller, 2000) was used, this time to calcu-late total F S T for each locus, based on counts of RAPD marker bands at each locus for all 460 individuals. The purpose of examining the distribution of F S T per locus was to determine whether certain loci were more genetically diverged than others. A low level of genetic divergence for a locus suggests that alleles at that locus are freely recom-bining among all individuals. In contrast, a high level of genetic divergence for a locus suggests that the locus is in a region of the genome that is not free to recombine. Non-recombining regions may be associated with chromosomal rearrangements such as inversions or translocations. Thus, an unusually high F S T for any RAPD marker locus could indicate that the locus lies within or adjacent to a chromosomal rearrangement. 38 3. Results 3.1 RAPD Markers A total of 123 P v A P D fragments that were polymorphic within or among popula tions were scored. Of these, homology testing revealed 88 reliable, replicable marker bands that were inferred to be homologous across populations (Figure 3.1). The rest of the bands were shown either to be non-homologous or to occur at too low a frequency to have any statistical power, or they were unreli-able, appearing in some gels, but not in replicate gels. Therefore these markers were discarded. CC Ul Q : I 1 POP 22 BSTU 1 UNCUT r - Mtfbik aim POP 23 BSTU 1 UNCUT — & J_ • m A* POP 24 BSTU 1 UNCUT — •*» mm mm 1— Jk < POP 25 BSTU 1 UNCUT ***" •_. mk — ai* m & m% DC UJ a a < jjk m t • $ j \ — S i Pr imer U B C #63 Figure 3.1 Test of band homology for four populations for U B C PvAPD primer #63. For bands of similar size in different populations, matching banding profiles for bands cut with restriction enzymes (here, BSTU I, a 4-cutter) is interpreted as band homology. For example, bands in populations 23 and 24 indicated by the double-headed arrows are the same size, and when cut with BSTU I, they produce the same banding pattern for cut fragments, as indicated by the single-headed arrows. The bands marked by the double-headed arrows are scored as homologous between populations 23 and 24. 39 3.2 Relationships Among Individuals Three trees were generated in PAUP using the PvAPD band presence-absence data matrix of 460 individuals, collected from 23 populations. These trees were: (1) a Neighbor-Joining tree with distance calculated from total character differ-ences (Figure 3.2); (2) a tree with distances based on total character differences, but the objective function set to weighted least-squares with inverse-square weighting (power=2). This is analogous to the Fitch method for calculating distance trees in the software package PHYLIP. Although two different tree-generating algorithms were used, the patterns of rela-tionships for individuals within geographic collection sites remained largely the same in both trees. The results of this analysis, shown in Figure 3.2, indicate that out of 460 individuals, only eight failed to group with other individuals collected from the same geographic site, instead usually clustering with other genetically close populations. This result indicates that genetic differences within populations are low relative to the genetic differences among populations, and that popula-tions truly do represent distinct demographic units. For simplicity's sake, once cohesion among individuals within populations was confirmed, within-population branches were collapsed to condense the trees for easier comparison of overall topology (Figures 3.3, 3.4). Two major anomalies stand out in the pattern of clustering of individuals from the same site. Firstly, at collection site 21, large genetic distances separate indi-viduals of different flower color. Individuals with yellow inflorescences (3, 4, 5, 7, 9, 12, 14, 16, 19) cluster together, as do white-flowered plants (individuals 1, 2, 6, 8, 10, 11, 13, 15, 17, 18, 20), but the two groups are only distantly related based on their PvAPD band profiles indicating either the evolution of a strong barrier to gene flow among alternate flower-color morphs in this population, or a very recent introduction of one of the color-morphs to the population such that introgression cannot yet be detected. Given that both flower color morphs are present at high frequencies at the site, and that the plants are annual outcrossers, it seems most likely that the two flower color morphs are reproductively isolated at this site. For the remainder of the analyses these two subpopulations were treated as 21 (yellow) and 21 (white). The second notable clustering of individuals is seen at sites 14 and 15. Individuals from these two sites are genetically indistinguishable on the tree (Figure 3.2). The result is very simply explained as these two "populations" actu-ally represent two collections from the same population. Site 14 is on the south side of a major highway, while site 15 is a few hundred feet north, on the oppo-40 28:1 • 20 27: 1 - 20 25: 1 -20; 11:13 22:1-20 9:1 - 20 6:3 12:1 -20 4: 1,2, 4,6-11.13-20 5: 1 -20; 6: 1 -20; 4: 3, 5,12; 12: 11 11:1 -20 10:1 -20 13:1-20 21: 3,4, 5, 7, 9, 12,14,16,19 21:8,18:12 24:1 • 20 23:1 • 20 21:1,2, 6,10,11,13,15,17,18, 20 26:1 -20 14:1-20; 15: 1 -11,13-20 19:15; 15:12; 13: 16 18:1-11,13-20 19:1,2,4-9,11-14,16-20 17:1-20 20:1 - 20 19: 3,10 29: 5, 9,15, 20 29: 3, 4,6-8,12-14,16,18,19 29:10 29:11,17 29:1 Figure 3.3 Relationships among individuals within populations. Condensed Neighbor-Joining tree from P A U P * (Swofford, 2002) based on R A P D marker fingerprints for 460 individuals collected from 23 sites. For ease of viewing, branches have been collapsed for individuals from single pop-ulations that cluster together as terminal, monophyletic groups. Tree is rooted by the longest branch. Genetic distance calculated from total character differences between pain of individuals. Taxon names correspond to numbered collection sites followed by numbered individuals at each site. For all nodes bootstrap values were less than 50%. 42 27: 1-20 , 25: 2,17.19, 20 25: 1F 3-16, 18; 11:13 • 22: 1-20 • 9:1-20 6:3 12:1-10,12-20 ,4:1,2,4,6-11, 13-20 • 4: 3, S, 12; S: 1-20; 6: 1, 2, 4-20; 12:11 • 11:1-12,14-20 .10:1-20 .13:1-15,17-20 -21:3, 4, 5, 7, 9,12,14,16,19 • 28:1-20 • 21:8; 18:12 • 24: 1-20 • 23: 1-20 •21: 1,2, 6,10,11, 13,15, 17, 18,20 • 26: 1-20 • 14:1-20; 15:1-11,13-20 • 19:15; 15:12; 13:16 • 18: 1-11,13-20 19:1, 2, 4-9,11-14,16-20; 20:15 17:1-20 20: 1-14,16-20 19: 3,10 29:5,9,15,20 29:3,4,6-8,12-14,16,18,19 29: 10 29: 11,17 29: 2 29:1 Figure 3.4 Relationships among individuals within populations. Condensed weighted-least squares (inverse weighting, power = 2) distance tree from P A U P * (Swofford, 2002) based on R A P D marker fingerprints for 460 individuals collected from 23 sites. For ease of viewing, branches have been collapsed for individuals from single populations that cluster together as terminal, monophyletic groups. Tree is rooted by the longest branch. Genetic distance is calculated from total character differences between pairs o f individuals. Taxon names corre-spond to numbered collection sites followed by numbered individuals at each site. For all nodes bootstrap values were less than 50%. 43 site side o f the highway. These specimens were labeled as two separate collec-tions as a test o f the efficacy o f a highway to act as a barrier to gene flow. Clearly, the highway has not been a significant barrier to gene f low wi th in this large population. For the remainder o f the analyses i n this project, ten individuals from each o f population 14 and 15 were pooled together to form population 14b. G i v e n the overwhelming tendency for individuals from the same collect ion site to cluster as nearest genetic neighbors, plants from a collection site were treated as distinct, cohesive demographic units for the remaining analyses. Variat ion wi th in populations was summarized using population genetic methods, w h i c h facilitated comparisons among populations. 3.3 Verification of Population Racial Identity A sample o f the R A P D fingerprint o f two individuals from each site that Rober t Carr had collected, compared w i t h two individuals from each site where I had collected is shown i n Figure 3.5. B y performing the same type o f distance analysis as described for comparisons among m y 460 field-collected individuals, I was able to generate the Neighbor-Jo in ing distance tree i n Figure 3.6. Figure 3.5 Matching R.L . Carr's populations to T. E. Olson's popula-tions using R A P D markers. Here, bands for U B C R A P D primer #63 are shown for two individuals from each of four populations from each collection. Similarity of banding patterns between R .L . Carr's samples and T.E. Olson's samples is interpreted as a match between the two collections. 44 Figure 3.6 Neighbor-Joining distance tree from PAUP* (Swofford, 2002) showing rela-tionships among field-collected sam-ples of R .L . Carr and T.E. Olson based on R A P D marker profiles. A genetic match for the two collections is interpreted as a match also for col-lection sites. Each site collected by Carr has been cytotyped and given a racial designation (R.L. Carr & G.D. Carr, 1983), so for Olson populations that match with Can-population, cytotypes can be inferred from Carr's data. O L S D K 2B 4 5 The only mismatches for our collection sites were for my sites 20 and 21, both near Lakeport, CA. In the case of population 20, my specimens grouped apart from Carr's specimens on a nearby branch, suggesting either that we had collect-ed different populations, or that allele frequencies have changed noticeably in this one population in the 22 years between when Robert Carr collected (August, 1978) and when I collected (August, 2000). For two reasons, I am more inclined to accept the latter scenario, despite the fact that twenty years does not seem like a significant length of time in an evolutionary context. Firstly, this roadside col-lection site was very clearly described in Robert Carr's collection notes and despite its small size was not difficult for me to locate, as it was the only popula-tion of Calycadenia for several miles along either side of that road. I suspect, therefore, that this was the correct collection site. Secondly, allele frequencies in this population could have been substantially altered by inbreeding or drift, or a combination of the two between the time when Robert Carr collected, and when I collected. By August 2000, this population was reduced to fewer than thirty individuals. The development of adjacent fields for equestrian activities prevented Calycadenia from growing anywhere but in the narrow ditch alongside the gravel road, and even here the population was obviously losing a competitive battle against star-thistle (Centauria solstitalis). If the population had not been larg-er twenty years ago, it would have been extremely unlikely that the Carrs would have noticed a mere thirty specimens hidden amongst the roadside grasses and weeds since they usually identified collection sites from the car as they drove (R.L. Carr, personal communication, summer, 2000). The other mismatched population was population 21. In this case, the popula-tion was represented by two different flower color morphs: yellow and white. The Carrs collection record gave no indication of flower color polymorphism for this population, so when collecting I was uncertain as to which flower color morph may have correctly been labeled Race Ciliosa. However, when I later grew plants from seed donated by Robert Carr from his collection, I found that his collection site 2145, whose geographic description corresponds to my popula-tion 21, generated yellow-flowered plants. I therefore suspect that the yellow-flower morph is the Carrs' Race Ciliosa, while the racial identity of the white form is unknown. However, the relationship between my population 21 and Carr's 2145 is not straightforward. My yellow-flowered specimens and my white-flowered specimens from site 21 fall into completely different branches from Robert Carr's samples (Figure 3.6). Carr's samples from the sites correspon-ding to my sites 20 and 21 cluster together, and my samples 20 and 21 (yellow) cluster together but our collections from the same sites are genetically quite dif-ferent. Even when all my individuals from this site are included in these analyses, Robert Carr's samples still fail to group with either my white- or yellow-flow-ered specimens (results not shown), but remain closely allied with samples from his "population 20". Failure of Carr's samples to pair with either of my flower-46 color groups could indicate that I collected at the wrong site, or that allele fre-quencies have changed significantly in this population as well. I suspect the latter is true for both populations 20 and 21 (yellow) the following reasons: sites 20 and 21 are geographically close (1 km or less separates them). They are also both small populations (fewer than 100 individuals each; Table 2.1). Carr's samples from these two locations, collected in 1978, show the same pattern of genetic similarity as do my collections (of yellow-flowered plants) from the same locations 22 years later (Figure 3.6). If substantial changes in allele frequencies over the 22 years between our two col-lection dates really does explain the poor genetic match between Carr's samples and my samples for populations 20 and 21 (yellow), then the changes have occurred in parallel in both populations. Given this scenario, coupled with their geographic proximity and their small populations sizes, I suspect these two popu-lations may be actively exchanging migrants and evolving in concert. Within my population 21, the fact that the two flower-color morphs remain genetically distinct could indicate a very recent introduction of either flower color to the site, which may explain why the Carrs did not describe the flower-color polymorphism. Even if one of the color morphs represents a very recent introduction to the site, this does not account for the failure of Carr's specimens to match with whichever flower-color morph was originally present, unless allele frequencies in that group have changed dramatically in the 22 years between our collection dates. 3.4 Relationships Among Populations 3.4.1 Genetic Distance Trees The Neighbor-Joining tree from PHYLIP (Figure 3.7) is shown with bootstrap values for well-supported branches (i.e. bootstrap > 50%). The Fitch-Margoliash tree from PHYLIP is shown with bootstrap support in Figure 3.8. These genetic distance trees clearly depict the resolution of two major, well-supported clades, which, henceforth, will be referred to as the "top clade" and the "lower" or "bottom clade." Although relationships among the populations within each clade are not resolved, my data suggest the cytological races do not form the distinct monophyletic clades that would be predicted if chromosomal differences have presented the greatest barrier to gene flow among these populations, particularly for those races represented in both clades of the tree (e.g. race Elegans 26 and 27, and race Ciliosa 17, 19, 21, and 25). Overall, both trees reveal similar topologies and similar levels of support. Bootstrap support for both trees is low (<50%) for nearly all branches. One 47 — C. pauciflora "Healdsburg" 28 • C. pauciflora "Healdsburg" 29 — C. pauciflora "Ramulosa" 20 C. pauciflora "Ramulosa" 18 —^— C. pauciflora "Elegans" 26 C . pauciflora "Pauciflora" 14b • C . fremontii " C i l i o s a " 17 •C. Jremontii " C i l i o s a " 19 C. fremontii " U n k n o w n " 21 (white) C . fremontii "Pil lsbury (?)" 23 C . /remonrii"Pillsbury" 24 99 C . Jremontii " C i l i o s a " 21 (yellow) 63 C. Jremontii "C i l i o sa" 25 — — — C . pauciflora "Elegans" 27 C . Jremontii "Pi l l sbury" 22 — — — — C . Jremontii " C o r n i n g " 9 C. Jremontii " C o r n i n g " 11 - C. Jremontii " D r y Creek" 10 C. fremontii " D r y C r e e k " 12 • C. Jremontii "F remont i i " 5 C. fremontii "Fremont i i " 4 C. fremontii "Fremont i i " 6 — C. Jremontii " D r y Creek" 13 3.7 Neighbor-Joining distance tree from PHYLIP (Felsenstein, 1995) based on estimated allele fre-quencies for 88 R A P D marker loci. Taxon names represent the putative species (Calycadenia Jre -montii or C. pauciflora), followed by the cytological race (G.D. Carr, 1975; R .L . Carr & G.D. Carr, 1983) in quotes and the population collection number. Population 14b contains 20 individ-uals, with 10 collected from each side of a major highway. Population 21 contains two different flower color morphs: individuals 1, 2, 6, 8 ,10 ,11 ,13 ,15, 17, 18, 20 are white-flowered and individuals 3, 4, 5, 7, 9, 12, 14, 16, 19 are yellow-flowered and fall into separate clades. Yellow color bars denote yellow-flowered populations. Al l other populations are white-flowered. Bootstrap values greater than 50% are shown above branches. Rooting is arbitrary, based on the longest branch between taxa. 48 C. pauciflora "Healdsburg" 29 C. pauciflora "Healdsburg" 28 — C. pauciflora "Ramulosa" 20 « C. pauciflora "Ramulosa" 18 ——— C. pauciflora "Elegans" 26 C. pauciflora "Pauciflora" 14b • C. fremontii " C i l i o s a " 17 C . fremontii " C i l i o s a " 19 C. fremontii " U n k n o w n " 21 (white) — — — — C . fremontii "Pi l lsbury (?)" 23 99 C . Jremonrii'"Pillsbury" 24 —————— C. fremontii " C i l i o s a " 21 (yellow) C. fremontii " C i l i o s a " 25 ———————— C . pauciflora "Elegans" 27 C . fremontii "Pi l l sbury" 22 H : J 65 — C . fremontii " C o r n i n g " 9 —— C. fremontii " C o r n i n g " 11 —— C. fremontii " D r y Creek" 10 — C . fremontii " D r y Creek" 12 C . fremontii "F remont i i " 5 ~ C. fremontii "Fremont i i " 4 C. fremontii "F remont i i " 6 — C. fremontii " D r y Creek" 13 3.8 Fitch-Margoliash unweighted least-squares distance tree from PHYLIP (Felsenstein, 1995) based on estimated allele frequencies for 88 R A P D marker loci. Taxon names represent the puta-tive species (Calycadenia fremontii or C. pauciflora), followed by the cytological race (G.D. Carr, 1975; R .L . Carr & G.D. Carr, 1983) in quotes and the population collection number. Population 14b contains 20 individuals, with 10 collected from each side of a major highway. Population 21 contains two different flower color morphs: individuals 1, 2, 6, 8 ,10 ,11 ,13 ,15, 17, 18, 20 are white-flowered and individuals 3, 4, 5, 7, 9, 12, 14, 16, 19 are yellow-flowered and fall into sepa-rate clades. Yellow color bars denote yellow-flowered populations. Al l other populations are white-flowered. Bootstrap values greater than 50% are shown above branches. Rooting is arbi-trary, based on the longest branch between taxa. 49 exception is the branch leading to population 10 (race Dry Creek) and population 11 (race Corning). This branch receives moderate support (63% to 65%), indicat-ing that the relationship between these two populations may be weakly resolved. A notably small geographic distance (4 km) separates these two populations. The only other strongly supported node in these trees marks the division between two distinct clades, which are recognized in both trees. In the lower clade, most populations have yellow flowers, while in the upper clade all populations have white flowers. The lower clade contains the populations of races of Calycadenia fremontii, and one population of C. pauciflora, while the upper clade contains sev-eral populations of both C. fremontii and C. pauciflora. The populations of the upper clade are centered on Lakeport, California, in Lake County, while the pop-ulations in the lower clade are centered on Redding, California, in Shasta County. Population 21, from which I collected both white and yellow flowered individuals, was split into its two component flower color groups. In PHYLIP analyses, these two subgroups are divided amongst the two clades, with the white-flowered section falling into the top clade and the yellow-flowered section in the lower clade. Relationships of populations are not resolved within either clade. 3.4.2 Using Mantel Tests To Investigate Relationships Between Clades The results of the Mantel tests with which I assessed the contribution that flower color, species differences, geography, and cytological differences make to deliniat-ing two clades in the distance trees is shown in Table 3.1. Mantel tests indicate that there is a strong correlation between flower color differences and genetic dis-tances between populations. Regardless of whether other factors such as geo-graphic distance between populations, cytological differences, or species differ-ences are accounted for, differences in flower color are always strongly associated (P < 0.004, r M = at least 0.2725) with genetic distances between populations (Table 3.1). Geographic distances between populations are similarly strongly cor-related with genetic distances. Species differences between populations do not exhibit such a strong association with genetic distances. When geographic dis-tances, cytological distances, or flower color differences between populations are accounted for, species differences are insignificandy correlated with genetic dis-tances between populations (P > 0.05, r M = only as much as 0.1305). Cytological differences between populations correlate with genetic distances only when accounting for flower color or species identity (P< 0.026, r M = 0.1465) but not when accounting for geographic distances (P > 0.262, r M = -0.0605). 50 MATRIX A permuted against M A T R D C B While All Populations M A T R I X C constant tM 2 tM P-VALUE Genetic Distance Geographic Distance 0.5560 0.3091 0.001 Genetic Distance Geographic Distance Cytological Distance 0.5534 0.3063 0.001 Genetic Distance Geographic Distance Flower Color 0.5204 0.2708 0.001 Genetic Distance Geographic Distance Species 0.5439 0.2958 0.001 Genetic Distance Cytological Distance 0.0887 0.0079 0.196 Genetic Distance Cytological Distance Flower Color 0.1465 0.0215 0.026 Genetic Distance Cytological Distance Species 0.1481 0.0219 0.022 Genetic Distance Cytological Distance Geographic Distance -0.0605 0.0037 0.262 Genetic Distance Flower Color 0.3506 0.1229 0.004 Genetic Distance Flower Color Species 0.3459 0.1196 0.004 Genetic Distance Flower Color Geographic Distance 0.2725 0.0743 0.006 Genetic Distance Flower Color Cytological Distance 0.3363 0.1131 0.004 Genetic Distance Species 0.1439 0.0207 0.045 Genetic Distance Species Geographic Distance -0.0425 0.0018 0.305 Genetic Distance Species Cytological Distance 0.1009 0.0006 0.085 Genetic Distance Species Flower Color 0.1305 0.0170 0.053 Table 3.1. Results of Mantel tests between matrices of genetic distances (Reynolds et al , 1983), flower color differences, geographic distances, cytological differences, and species differences among populations. Tests were performed on all populations of both species to determine which of these four factors, if any, significantly correlate with genetic distances and therefore may explain the pattern of genetic distances among populations that resulted in the genetic distances trees produced in PHYLIP. The Mantel test statistic, rM, measures the correlation between data matrices. rM' indicates the proportion of variation in data in Matrix A that can be explained by variation in Matrix B. Significant results (P<0.05) are shown in bold-face type. 51 3.5 Summary of Genetic Variation Within and Among Populations: Percent Polymorphism and F g T Descriptive statistics and estimates of genetic divergence (centroid F S T ) for each population, race, species, and clade calculated in TFPGA are shown in Table 3.2. Statistics are presented for Population 21 as a whole and also for each of its two different flower-color groups. The range of values for population-level centroid F S T is 0.1170 to 0.4014, with an average of 0.1898 (Table 3.2). Population 27 (Calycadenia pauciflora "Elegans", F S T = 0.4014) is more highly differentiated than all other populations. Aside from population 27, other populations exhibit relatively low F S T values, indicating that each population in the Calycadenia Jremontii- C. pauciflora complex, is poorly dif-ferentiated from all other populations, a result that supports the lack of resolution of relationships among populations in the genetic distance trees. Between- and within-clade F S T analysis supports the result of genetic distance analyses in PHYLIP, in which two clades were resolved in the Calycadenia fremon -Hi - C. pauciflora complex. Jackknifing 1000 times over 88 RAPD loci results in F S T between the two clades of 0.2156 +/- 0.0554, which is slightly greater than the average of centroid F S T for all the populations (0.1898). This moderately high value of F S T suggests that the two clades represent moderately well-differen-tiated groups. F s x was also calculated within each clade. F S T within the top clade is 0.2481 +/- 0.0225 (jackknifed 1000 times), and F S T within the bottom clade is 0.3193 +/- 0.0253 (jackknifed 1000 times). 3.6 Mantel Tests for Correlations Among Data Matrices Mantel tests were undertaken to search for correlations between genetic distances and other factors that could potentially explain the relationships of populations in the Calycadenia fremontii - C. pauciflora complex. The factors tested were geo-graphic distance between populations and the number of chromosomal rearrange-ment differences between populations. Tests were performed for all populations of both species, for populations within each geographic region (Lakeport and Redding) and within each of the top and bottom clades of the distance trees. Results for all simple pairwise and partial Mantel tests are shown in Table 3.3. There is a notable geographic component to the partitioning of genetic distance among all populations in this species complex, implying that the closer popula-tions are, the more likely they are to share genetic similarities. In other words, 52 Population Race Polymorphic Pairwise Loci (%) FST 29 Healdsburg 65.9 0.1510 28 Healdsburg 61.4 0.1867 All Healdsburg 63.7 0.1008 27 Elegans 34.1 0.4014 26 Elegans 58.0 0.1871 All Elegans 46.1 0.0684 14b Pauciflora 58.0 0.1587 All Pauciflora 58.0 0.1587 20 Ramulosa 54.5 0.1473 18 Ramulosa 55.7 0.1263 All Ramulosa 55.1 0.0968 All Calycadenia pauciflora 55.0 0.0280 25 Ciliosa 44.3 0.1494 21 (all) Ciliosa?? 60.2 0.0638 21 (white) Ciliosa?? 46.6 0.1145 21 (yellow) Ciliosa?? 45.4 0.2373 19 Ciliosa 48.9 0.1517 17 Ciliosa 51.1 0.1887 All Ciliosa 47.3 0.0278 24 Pillsbury 36.4 0.1274 23 Pillsbury?? 38.6 0.2237 22 Pillsbury 44.3 0.2383 All Pillsbury 39.8 0.0442 13 Dry Creek 54.5 0.1553 10 Dry Creek 59.1 0.2138 12 Dry Creek 56.8 0.2087 All Dry Creek 56.8 0.0729 11 Corning 58.0 0.2460 9 Corning 46.6 0.2553 All Corning 52.3 0.1249 6 Fremontii 64.8 0.1170 5 Fremontii 67.0 0.1784 4 Fremontii 62.5 0.1729 All Fremontii 64.8 0.0984 All Calycadenia fremontii 52.2 0.0085 Top Clade vs Bottom Clade 0.2156* Both Species 53.4 Table 3.2. Descriptive statistics from TFPGA. Centroid F^ measures the degree of differen-tiation of each group from a centroid root-ing of all groups. For each group, centroid F s x was calculated by comparing allele fre-quencies across all loci within that group to the pooled total allele frequencies across all loci of all other popu-lations. Total Fgr, pre-sented for all popula-tions in each of the top and bottom clades of the genetic distance trees (Figures 3.7, 3.8), was estimated by pool-ing all populations within each clade and averaging the total FJT from 88 PvAPD loci. Centroid F C T for Race Ciliosa, and for all populations of Calycadenia fremontii was calculated exclud-ing population 21, whose racial designa-tion is unclear. * Total F s x . Standard deviation is 0.0554. 53 (c) Redding Region P-VALUEJ| 0.044 0.041 0.385 0.313 0.064 0.060 (c) Redding Region "a 0.3480 0.3508 0.0009 0.0053 0.0224 j 0.0267 (c) Redding Region 3 0.5899 0.5923 0.0301 -0.0728 0.1496 0.1633 (b) Lakeport Region P-VALUEJ| 0.064 0.010 0.301 0.033 0.001 0.001 (e) Bottom Clade P-VALUE 0.001 0.004 0.036 0.229 0.039 0.129 (b) Lakeport Region "2 0.0494 0.1341 0.0154 0.1031 0.3416 0.4002 (e) Bottom Clade 0.3869 0.3192 0.1199 0.0227 0.1453 0.0509 (b) Lakeport Region 0.2223 0.3662 -0.1241 -0.3211 0.5845 0.6326 (e) Bottom Clade 3 0.6220 0.5650 0.3463 0.1508 0.3812 0.2257 (a) Both species || P-VALUEJ| 0.001 0.001 0.196 0.262 0.026 0.028 (d) Top Clade P-VALUE 0.039 0.081 0.211 0.470 0.001 0.001 (a) Both species || 0.3091 0.3063 0.0079 0.0037 0.0611 0.0571 (d) Top Clade 0.0939 0.0646 0.0319 0.0006 0.4140 0.3950 (a) Both species || 0.5560 0.5534 0.0887 -0.0605 0.2472 0.2390 (d) Top Clade 0.3064 0.2542 0.1787 -0.0254 0.6434 0.6285 While MATRIX C constant Cytological Distance Geographic Distance Genetic Distance While MATRIX C constant Cytological Distance Geographic Distance Genetic Distance MATRIX B Geographic Distance Geographic Distance Cytological Distance Cytological Distance Cytological Distance Cytological Distance MATRIX B Geographic Distance Geographic 1 Distance Cytological Distance Cytological Distance Cytological Distance Cytological Distance MATRIX A permuted against Genetic Distance Genetic Distance Genetic Distance Genetic Distance Geographic Distance Geographic Distance MATRIX A permuted against Genetic Distance Genetic Distance Genetic Distance Genetic Distance Geographic Distance Geographic Distance TEST < CQ U Q W TEST < pq O Q W l l , •6 a « « u S e <— ' 2 * ? •a u y « JO fc 3 C! i/i 4 S3 5b _e u o « •c -3 o 4-* W U-i U O g o a . I-I o ~0 <U - f l . P C -3 5 S 3 . ifi a o § a & al (Li 5 4 populations in the Calycadenia fremontii - C. pauciflora complex are at least partially genetically isolated by geographic distance. This can be seen in the results of the simple Mantel tests for genetic distance by geographic distance (Table 3.3, Test A) or the partial Mantel tests for genetic distance by geographic distance, accounting for the effect of cytological differences among populations (Table 3.3, Test B). The results of most of these tests are significant at the 5% level across all popula-tions, within geographic regions (Lakeport and Redding), and within each of the two clades (top and bottom) from genetic distance trees (Figures 3.7, 3.8). Across the entire region, there is a very small effect of geography on the variation in cytology among populations (Table 3.3 (a), Tests E , F). Geography accounts for about 5% of the variation in cytology seen among populations in the entire complex. Interestingly, this pattern becomes much stronger for populations in the Lakeport region (Table 3.3 (b)) or the top clade (Table 3.3 (d)), where geog-raphy accounts for about 40% of the cytological variation among populations, while the opposite is true for populations near Redding (Table 3.3 (c)) or in the bottom clade (Table 3.3 (e)). 3.7 Distribution of Per Locus F c T The results of this test, shown in Figure 3.9, indicate that few loci have very low F S T and most loci have an F S T between 0.10 and 0.50 suggesting they are rela-tively well differentiated. A few loci have extremely high F S T , with two loci having F S T greater than 0.80. The high degree of differentiation for these two loci may be explained if these two loci could be mapped to sites within or near rearranged segments of chromosomes, for in such a case the rearrangements may contribute to the differentiation of these loci by protecting them from gene flow. 55 Distribution of per locus FST 16 0.0 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 to to to to to to to to to to 0.05 0.15 0.25 0.35 0.45 0.55 0.65 0.75 0.85 0.95 Perlocus FST Figure 3.9 Distribution of per locus FST estimated for 88 R A P D markers scored across 23 populations. This test measures the level of differentiation of each of the R A P D loci. Loci with low FSTs are inferred to be recombining across aE populations. Loci with high FSTs are inferred to be non-recombining. 56 4. Discussion Based on biosystematic studies undertaken by R.L. Carr and G.D. Carr over the past two decades (G.D. Carr, 1980, 1977, 1975; R.L. Carr, & G.D. Carr, 2000, 1983), I hypothesized that chromosomal differences among populations in the Calycadenia jremontii - C. pauciflora complex may be directly responsible for the observed reduction in hybrid fertility among the species and their component cytological races. If this were so, chromosomal-based barriers to gene flow in Calycadenia could provide evidence in support of the much-debated theory of chromosomal speciation. Initially, certain predictions were made concerning the possible outcomes for rela-tionships of genetic distance between populations to their cytological identities. The most straightforward prediction was that if cytological differences among populations act as the primary barrier to gene flow among them, genetic distance among populations of the same race should be zero or very low while genetic distances among populations of differing races should be high. In other words, if populations of the same race were separated by the shortest branches and clus-tered as nearest neighbors on a tree of genetic distances, this would support a pri-mary role for chromosomal rearrangements in driving divergence or evolution in this group. As we have seen, the most straightforward result was not to be. 4.1 Relationships Among Populations 4.1.1 Patterns of Relationships from Genetic Distance Trees While the genetic distance trees (Figure 3.7, Figure 3.8) resolve two major, well-supported clades, these clearly do not correspond to the two species or to distinct cytological races as most of the races span both clades. None of the races form the distinct monophyletic clades that would be predicted if chromosomal differ-ences have presented the greatest barrier to gene flow among these populations. Instead, some races have representative populations in both clades of the tree while other races, although confined to just a single clade, do not have popula-tions resolved into monophyletic terminal clusters within the clade. In the present analysis, very low values of centroid F S T for races relative to the centroid F S T of individual populations suggest that races may not describe mono-phyletic groups. For example, excluding population 21 whose racial identiy is uncertain, the populations of race Ciliosa have centroid F S T ranging from 0.1440 to 0.1786 (Table 3.2). However, the centroid F S T for the race is only 0.0278. This extremely low value for centroid F S T indicates that the distance between the node that unites populations of race Ciliosa and the root of all populations in the 57 whole species complex is very small. This is exactly the result suggested by the genetic distance trees (Figures 3.7 and 3.8), for race Ciliosa spans both the top and bottom clade in these trees. According to the distance trees, the node that unites populations of this race also unites nearly all the other populations in the tree. Had analyses shown monophyly for the races, two inferences could have resulted: 1) the rearrangements have had a measurable effect on gene flow, and the races are old enough that this effect can be seen in the genetic distance data, and 2) races need have arisen only once in the evolutionary history of this group. Instead, given that some races span both clades, one can infer either a) the cyto-logical races predate the genie divergence between the two clades, or b) those races represented in both clades have arisen at least twice and evolved in parallel in both clades. If (a) is correct, then cytological differences clearly have not been a strong barrier to gene flow among these populations. If (b) is correct, cytologi-cal differences may currently be a good barrier to gene flow, but this barrier has not existed long enough to have altered allele frequencies detectably. Note that under this scenario, gene flow would have a homogenizing effect, and assuming isolation by distance, geographically proximal populations of the same race would be expected to be most genetically similar, even though the chromosomal arrangements may have evolved in parallel. The issue of whether cytological differences among populations currently act as a good barrier to gene flow could be resolved by thorough, controlled crossing studies. While the Carrs conducted numerous crosses to assess the racial identites of each of their populations, they were not interested in measuring levels of crossability within races or in assessing other factors that may contribute to repro-ductive isolation between populations or races. Individuals used in their crosses were often pooled from several populations of the same race, thereby obscuring any reproductive isolating effects of geographic distance or other factors differenti-ating populations within races. To determine whether cytological differences alone account for barriers to gene flow among populations of different races, con-trolled crossing studies will have to be conducted to eliminate the possibility that other factors may potentially contribute to reproductive isolation. For example, to assess the effect of geographic isolation as a barrier to reproduction, one would have to cross individuals from geographically distant populations of the same race and also cross individuals from geographically close populations of different races. Regardless of whether cytological differences arose early in the evolutionary his-tory of this group or quite recently, their effects of gene flow need not be homo-geneous throughout the genome. As discussed by other workers (Rieseberg et al, 1995; Noor et al, 2001a,b), chromosomal rearrangements may limit gene flow in regions of the genome that are structurally heterozygous in inter-racial hybrids, 58 while allowing gene flow in non-rearranged portions of the genome. The effects of chromosomal rearrangements on gene flow could only be assessed if mapping studies were to identify loci of interest within, adjacent to, and external to rearrangements. By crossing individuals and tracking these marker loci, one could compare levels of gene flow for loci within and external to rearranged seg-ments. 4.2 Features that Distinguish the Two Clades As we look at features that distinguish the two clades, I will attempt to infer which of these factors, if any, may potentially drive the divergence of the two clades. 4.2.1 Flower Color Polymorphism & Flowering-Time Differences The results of the Mantel test that shows a strong; correlation between flower color differences and genetic distances between populations (Table 3.1) suggests that populations that are genetically similar tend to share the same flower color. This is certainly the pattern evident from the genetic distance tress, in which the clades represent the greatest genetic distance between any two groups, and the top clade is exclusively white-flowered, while the bottom clade is mostly yellow-flowered but has some white-flowered populations. Four populations in the lower clade do not share the yellow flower color that would otherwise be diag-nostic of this clade. These include the three populations of Calycadenia fremontii "Fremontii" and the only representative of C. pauciflora in this clade, race "Elegans" population 27. Based on flower color, population 27 remains an unex-plained anomaly in the lower clade. The white color of race Fremontii, which could be explained most parsimonious-ly by a either a single reversion to an ancestral white flower color or parallel evo-lution of white flower color in both clades if yellow is ancestral, is not the only difference that distinguishes these populations from all others in this clade; these populations are also the only three to blossom and mature in the early spring (Table 2.1). Flowering-time may effectively isolate populations of race Fremontii, and this may be reflected in the presence of flower-color differences. All other populations in both clades flowered in late summer, and so were clearly not iso-lated from each other by flowering-time differences. Flower color has not previ-ously been used to separate groups in the Calycadenia fremontii - C. pauciflora com-plex (B. Baldwin, personal communication, January 9, 2002). 4.2.2 Geographic Distinctions Between the Clades Like flower color differences, the strong correlation between geographic distances and genetic distances between populations (Table 3.1) supports the pattern of geographic separation between the two clades if clades represent the greatest 59 -124 -123 -122 -121 -120 Longitude Figure 4.1 Topographical differences among collection sites in the Calycadenia Jremontii - C. pauciflora complex. Collection sites are marked by white squares. genetic distances between two groups i n the Calycadenia fremontii - C. pauciflora complex. The pattern o f geographic separation between the two clades is striking i n that populations in the top clade were all collected i n the Coast Ranges o f northern California, mostly near Lakeport in Lake C o u n t y (Figure 2.1; Figure 4.1). Populations collected at the northern end o f the Sacramento Valley near R e d d i n g in Shasta County , over 200 miles northeast o f Lakeport, all fall into the bot tom clade along wi th a number o f populations collected i n the Coast Ranges. These include populations 9, 22, 25, and 27 in the lower clade, as w e l l as the ye l low half o f population 21. Population 9 was the northernmost collection site, beyond the northern tip o f the Sacramento Valley, i n the Klamath Mountains o f 60 the Coast Ranges. Populations 21, 22, 25, and 27 were all collected closer to Lakeport. Although geography likely contributes a great deal to the pattern of disjunction of genetic distance between the two clades, it does not fully account for the pattern of relationships between all populations. 4.2.3 Taxonomic Affinities of the Clades Taxonomic affinities of the clades, namely the association of a particular species with each clade, are more poorly defined than flower color differences, and this observation in substantiated by the results of the Mantel tests which find no sig-nificant correlation between genetic distances and species differences. Although the top clade contains all but one (population 27) of the populations of Calycadenia pauciflora, it also contains numerous populations known to be, or thought to be Calycadenia fremontii (populations 17, 19, and 24, and perhaps 21 and 23). Considering these anomalous populations, they could represent populations that have been improperly assigned to taxonomic species. An indication that this may have occurred for populations 17 and 19 is found in a 1975 paper by Gerry Carr. In this paper, Carr explains that the behavior of populations of race Ciliosa differ between geographic locales. In Tehama County, hybrids between sympatric race Ciliosa and race Ramulosa occur with moderate frequency, and both occupy sim-ilar habitats. Near Lakeport, populations of race Ciliosa are contiguous with pop-ulations of race Ramulosa, although individuals of race Ciliosa generally occupy the more mesic habitats, while individuals of race Ramulosa occur in more xeric habitats (G. Carr, 1975). Hybrids occur infrequendy between the two races near Lakeport, perhaps because these ecological differences act as barriers to hybridiza-tion (G. Carr, 1975). Given these facts, several explanations for the placement of race Ciliosa are possi-ble. First, the collections of race Ciliosa from the Lakeport area, represented by populations 17 and 19 could be not race Ciliosa at all, but rather they may be race Ramulosa since the two races are sympatric in this area. Unlike all my other collection sites, these sites were based on records from Gerry Carr not Robert Carr. As I had no corresponding D N A sample from Gerry Carr's collec-tion, I could not verify the racial identity of these populations in the same man-ner I had for all the other populations (Figure 3.6). Instead, I had to rely on detailed collection records to precisely locate these populations and then assume that these samples truly were Gerry Carr's race Ciliosa. If, in fact, these popula-tions are race Ramulosa, much of the confusion regarding the placement of race Ciliosa could be resolved. Alternatively, if populations 17 and 19 truly are race Ciliosa, the same or a very similar chromosomal rearrangements could have evolved separately in the 61 Lakeport area and in Tehama and Shasta Counties, defining two origins for race Ciliosa. This could account for the large genetic distance and flower-color differ-ence between populations of race Ciliosa in the top clade (white-flowered) and populations supposedly of the same cytological configuration in the bottom clade (yellow-flowered). For example, race Ciliosa near Lakeport could have evolved from a white-flowered Calycadenia pauciflora "Ramulosa" by two simple chromo-somal interchanges (Figure 1.10). Further east, in Tehama and Shasta Counties, yellow-flowered race Ciliosa could have again evolved independently from a Calycadenia fremontii ancestor. Again, under either of these hypotheses, populations 23 and 24 (C. fremontii race Pillsbury) fall into the clade that otherwise includes only races of C. pauciflora. If the race Pillsbury populations are assumed to be correctly identified, it is again possible that they could have evolved in parallel in each of the clades. Alternatively, these populations may be misidentified. Under any of the above hypotheses, population 27 (Calycadenia pauciflora race Elegans) is an outlier in the bottom clade. This population does not fall where expected if the clades are distinguished by geography, morphology, or species (Figures 3.7, 3.8). It is a very small population growing at moderately high ele-vation at the edge of a thick conifer forest, an unusual habitat for Calycadenia (Table 2.1). Centroid F§ T tests indicate that it is highly genetically differentiated from other populations in this species complex (F S T = 0.4014) (Table 3.2). Population 27 also exhibits a lower level of polymorphic loci (34.1%) than most populations, suggesting that its small size may have facilitated rapid genie, mor-phological, and perhaps physiological evolution through drift or severe inbreed-ing, or a combination of the two. Of the 88 RAPD markers used in this study, none is unique to population 27, suggesting inbreeding or drift could have result-ed in a shift of allele frequencies from something resembling a Lakeport-area, white-flowered, C. pauciflora, "top clade" type of population toward something more closely resembling a Redding-area, yellow-flowered, C. fremontii, "bottom clade" type, of population without requiring the evolution of novel mutations. In other words, rapid divergent evolution in this population could also account for its problematic placement in a tree of genetic relationships among populations. Overall, it is difficult to argue that each of the populations in this study that do not fall into the clade that contains the majority of the populations of that tax-nomic species represent errors. Therefore, it seems that while the taxonomic species previously identified are reflected to some degree in my analyses, taxo-nomic revision will be required in order to reflect the phylogenetic affinities identified in my work. However, before this can be done, additional sampling should be undertaken to include all the races of the two species, including those not sampled in this study (i.e. races Tehama, Wurlitzer, and Lewiston). 62 4.3 Evidence From Other Sources 4.3.1 F S T analyses A moderately high F S T between the two clades (0.2165) supports the results of phylogenetic analyses in which the two clades received high bootstrap support (99%). Since relationships among populations within either clade remain unre-solved, populations within each clade are either not genetically differentiated, or they are well-differenatiated, but the magnitude of the genetic differences among populations is similar. The results of the within-clade F S T analyses support the latter scenario, as populations within each clade are quite well-differentiated from one another (F S T ( t o p c l a d e ) = 0.2481, F S X ( b o t t o m c l a d e ) = 0.3193). The high F s x coupled with the lack of phylogenetic resolution within clades suggests that populations within each clade are about equally highly-differentiated. 4.3.2 Comparison with ITS and ETS trees Support for the topology of my distance trees can be sought in the work of Bruce Baldwin, who has generated phylogenetic trees based on ITS and ETS nrDNA sequences for many of the races in the Calycadenia fremontii - C. pauciflora complex. In his most recent tree, Baldwin finds strong support (100% bootstrap support) for the major branches grouping together the species C. fremontii, C. oppositifolia, and C. pauciflora, but fails to resolve relationships among races within the species (B. Baldwin, personal communication, January 2002). The lowermost clade in his tree, which receives 100% support, clusters the races of Calycadenia fremontii and a single population of C. pauciflora race Pauciflora. Baldwin's tree, like my distance trees, clusters the races of C. fremontii together. As with my clades based on genetic distances, there is no support for the pattern of branching within this group, suggesting that the races and their representative populations are not much-diverged. While the Calycadenia Jremontii and C. pauciflora "Pauciflora" group forms the lower branch of a major bifurcation in the ITS/ETS tree, the upper branch of this bifurcation leads to all the other races of C. pauciflora and to C. oppositifolia. Within this upper clade, Baldwin resolved relationships between the group of the large-headed C. pauciflora races, Elegans and Healdsburg, and the group of the small-headed races, Ramulosa, Tehama, and Wurlitzer. Although I was unable to mirror Baldwin's resolution for these races of C. pauciflora, the upper clade in my tree does in some ways correspond to this upper clade in Baldwin's tree in that my upper clade also mostly contains populations of C. pauciflora. However, in my trees, I have populations of C. fremontii "Pillsbury" in the top, supposed 'C . pauciflora' clade, while a single population of C. pauciflora "Elegans" falls into my lower 'C. fremontii' clade. 63 The discrepancies between Baldwin's tree and my trees may result from our hav-ing sampled different populations of the same races. For example, Baldwin finds support for grouping Calycadenia pauciflora "Pauciflora" closer to the races of C. fremontii than to the other races of C. pauciflora, a result that is in contrast to mine. However, we sampled different populations of race Pauciflora. If ITS and ETS sequence data were to be generated for my population 14b (C. pauciflora "Pauciflora") such data may, for example, place my population of race Pauciflora closer to Baldwin's populations of race Ramulosa, Elegans, or Healdsburg, there-by corroborating the placement of this race in my distance trees. Similarly, RAPD data for Baldwin's "Pauciflora" population may confirm that race Pauciflora, like races Elegans, Pillsbury, and perhaps Ciliosa, may comprise several genetically divergent populations. If, as I have shown in my distance trees, popu-lations of the same cytological race can appear in quite different parts of the trees, it is possible that with wider sampling, Baldwin may also lose resolution of racial relationships in his tree. O f course, this is all purely speculative pending sequenc-ing of my samples, a project which I aim to undertake in collaboration with Baldwin at a later date. Again, if wider population sampling contributed to ITS and ETS data suggesting that cytological races could not be resolved into monophyletic groups, this would support the notion that either cytological differences have not been a significant barrier to gene flow among populations or that they have evolved too recently to have had any noticeable effect yet. In this case, the labeling of populations on Baldwin's tree and on my trees perhaps shall have to be amended so that instead of corresponding to evolutionarily meaningless cytological races, they represent the agent that actually correlates with reproductive isolation among groups (e.g. morphological differences like flower color, or geographic distance). 4.3.3 Evidence from Mantel Tests i) The Effect of Geography on Genetic Distance In tests involving all populations across the entire range of the species complex (Table 3.3 (a)), both the simple and partial Mantel tests indicate that the correla-tion between genetic distance and geographic distance is highly significant (P < 0.001 for both tests). According to these tests, for the entire Calycadenia fremontii - C. pauciflora complex, nearly 31% of genetic distances among populations can be explained by geography. Given that the two clades resolved in the distance trees (Figures 3.7, 3.8) can be partially distinguished by geography (e.g. one clade centered near Lakeport, the other near Redding), it is not surprising that Mantel test results indicate that genetic distances between populations can be partially explained by geographic separation between them. To test whether the geographic effect on genetic distance seen in the entire 64 species complex merely reflects strong geographic effects on genetic distance between the two regions where samples were collected or whether the geograph-ic effect extends to populations within each region, Mantel tests were repeated for matrices of genetic distance and geographic distance (with or without accounting for cytological differences) within each geographic region. The results of these tests (Table 3.3 (b),(c)) indicate that the effect of geography on genetic distances among populations is not just a characteristic of the main groupings as a whole, but that geography also explains some of the genetic distance among pop-ulations within each region. For example, in the Lakeport region (white-flow-ered clade) geographic distance explains 13.4% of the variation in genetic distance among populations, after accounting for cytological differences. If cytological dif-ferences are not accounted for, the effect of geography on genetic distance becomes statistically insignificant in the Lakeport region. In the Redding region (mostly yellow-flowered clade), geography has had a significant effect on genetic distance irrespective of cytological differences among populations. Near Redding, geography explains 35% of the genetic distance among populations. To test for the effect of geography on genetic distance within the clades, Mantel tests were again performed, this time on populations clustered according to clade. In this case, I was specifically interested in how well geography explains genetic differences within the bottom clade, given the rather large geographic distance that separates Lakeport populations 27, 25, 22, and 21 from Redding populations 13, 12, 11, 10, 9, 6, 5, and 4. The highly significant results of these tests (Table 3.3 (e)) indicate that within the lower clade, geography accounts for nearly 32% of the genetic distance among populations when cytological differences are con-sidered (Test B). Despite that phylogenetic reconstruction based on genetic dis-tances among populations failed to resolve relationships among populations within the lower clade, Mantel test results indicate that there is a pattern of relationship among populations in the lower clade and the pattern is defined by geographic distances among populations. This result suggests that populations 27, 25, 22, and 21, because they are geographically close to each other and are distant from other populations in the Redding area, are more likely to share genetic similarities with each other than with the rest of the populations in the lower clade. These four populations could have been founded by long-distance migrants from the Redding area to the Lakeport area. This would account for their overall genetic similarity and their morphological similarity to populations near Redding (namely, yellow flowers). Since these populations would have been founded by a few individuals, they would likely be more inbred than other populations in the Redding area. Given that they would contain only a subset of the genetic varia-tion present in Redding populations, this could be enough to differentiate them somewhat from the rest of the Redding group. Interestingly, Mantel test results indicate that geography does not have a particu-65 larly strong effect on genetic distance among populations in the top clade (the white-flowered populations in the Coast Ranges, mostly near Lakeport). The simple Mantel test of genetic distance by geographic distance indicates that geog-raphy accounts for a mere 9% of the variation in genetic distance, but even this weak effect becomes statistically insignificant when cytological differences among populations are accounted for (P> 0.081). In the distance trees, no relationships among populations in this group could be resolved with any certainty, suggesting that genetic divergence among these populations was very low. ii) The Effect of Cytological Differences on Genetic Distance The next set of Mantel tests that were undertaken compared genetic distances among populations to cytological differences among them. First, the effect of cytological differences on genetic distances was tested for all populations across the entire range of the species complex (Table 3.3(a)). Irrespective of whether geography was accounted for or not, cytological differences do not explain the genetic distances among populations within the entire species complex. Similarly, accounting for geographic effects (Table 3.3, Test D), cytological differences do not explain genetic distances for populations clustered by clade (Table 3.3 (d), (e)) or within the Redding region (Table 3.3 (c)). Surprisingly, there is a weak (r M 2 = 10%) negative correlation between genetic distance and cytological differences for populations in the Lakeport region (Table 3.3 (b), Test D), meaning that the most genetically similar populations are gener-ally the most cytologically distant. While this peculiar result could reflect biolog-ical reality, I suspect it is actually due to uncertainties in the matrix of cytological distance for this group. A negative correlation such as this could result if the racial designation of some of the populations is not correct. Although most populations I collected were genetically matched with Robert Carr's collections to determine the race for my populations, the racial designation for two of my populations from the Lakeport area (17 and 19) could not be verified as discussed above. I have designated these two populations as race Ciliosa based on the Carrs' collection notes, but they might be race Ramulosa as these two races are contigu-ous and often grow sympatrically near Lakeport (G. D. Carr, 1975). If the par-tial Mantel test for genetic distance by cytological distance, accounting for the effects of geography, is repeated considering populations 17 and 19 as race Ramulosa, the resulting correlation is insignificant (rM= -0.2265; P > 0.097). If the weak negative correlation between genetic distance and cytological differ-ences in the Lakeport region is the result of biological reality, and the current racial designation of all populations in that region holds (i.e. populations 17 and 19 are race Ciliosa), then genetically similar populations really are cytologically distant. As discussed in section 4.1, there are two ways of interpreting this result. Either cytological races predated genetic divergence of the populations and cyto-66 logical differences have not acted as a barrier to gene flow, or cytological races evolved after most of the genetic distance between populations had been estab-lished and reinforce pre-existing genie restrictions to gene flow but their effects may not yet be detectable using RAPD markers. Under this scenario, cytological evolution must not have happened in a stepwise fashion, for there is not a linear pattern to cytological relationships among the populations most closely related by genie factors. In other words, populations of a single ancestral race must have diverged genetically, then each population evolved a novel cytotype. Depending on the geographic relationships among multiple populations sharing the same cytotype, cytotypes may have evolved in parallel in each population or may have spread between populations by gene flow. In any case, all the results from Mantel tests between matrices of genetic distances and cytological distances among populations indicate that cytological differences among populations are not positively correlated with increasing genetic distance. There is no evidence to support the hypothesis that cytological differences among populations have acted as a primary barrier to gene flow, at least for the 88 RAPD marker loci used in this study. iii) The Effect of Geography on Cytological Differences Finally, the effect of geographic distance on cytological differences among popula-tions was tested to determine whether there is a geographic pattern to the cyto-logical races in the Calycadenia fremontii- C. pauciflora complex. The simplest explanation for the strong geographic pattern to the different cyto-logical rearrangements in populations in the Lakeport area (or top clade) is that each novel cytotype has evolved only one time in this region, and whether recently evolved or ancient, has not spread much beyond its center of origin. The spread of novel chromosomal rearrangements by diffusion is expected to be exceedingly slow (Spirito, 1998). So even if the chromosomal differences among populations in this area are ancient, if they do not confer a strong fitness advan-tage or their frequency is not increasing in a population through meiotic drive (Spirito, 1998; Walsh, 1982), it is entirely conceivable that they may not have spread to many populations or beyond a few miles from their centers of origin. This does not imply that cytological rearrangements are acting as barriers to gene flow among geographically distant populations, for cytological differences have been demonstrated not to correlate with genetic distances among populations (see section 4.3.2.ii). Rather, the results of the present Mantel test merely indicate that any chromosomal differences that may have arisen in one population may be passed to neighboring populations with which they exchange genetic material. These populations are also likely to share genie similarities, not a surprising result given that genes are components of the chromosomes that may be exchanged between hybridizing populations. 67 In the Redding area as in the bottom clade from the genetic distance trees, the differences among cytotypes do not correlate with geographic distance, although there is a strong geographic pattern to genetic distance among populations. This means that populations in the Redding area that are geographically closest are most likely to share genie similarities, yet do not necessarily share similar cyto-types. There is also no correlation between cytological differences and genetic distances in this group (Table 3.3 (c), Test D), suggesting that cytological differ-ences have not acted as a barrier to gene flow among populations. Explaining the essentially random pattern of cytological races across geographic space in the Redding area is a two-part problem. First, how can fixed cytological differences among adjacent populations be maintained despite gene flow among them? Second, how can geographically distant populations share the same chromosomal arrangements? The answer to the first question may be resolved using the same argument as was used to explain the geographic pattern for the cytological races in the Lakeport area. Since chromosomal differences have been demonstrated not to correlate with significant barriers to gene flow among these populations, gene flow can occur between adjacent, cytologically different populations. The pattern of fixed cytological differences between adjacent populations could simply reflect the slow rate of diffusion of novel rearrangements for two cytologically different, hybridiz-ing populations. To answer the second question, at least three difference scenarios could be invoked. Firstly, each cytotype may have had multiple independent origins in this area, or very similar-looking cytotypes may have evolved, resulting in several widely spaced populations with the same (or nearly the same) chromosomal con-figuration. Although this may seem an unlikely explanation, consider that only three chromosomes in Calycadenia Jremontii or C. pauciflora are known to exhibit rearrangement. If major rearrangements are restricted to a relatively small portion of the genome, perhaps because arrangements elsewhere are lethal, then it may not be so unlikely that very nearly identical cytological rearrangements can evolve in more than population. Alternatively, if cytological differences have had a single evolutionary origin, their spatial arrangement could be randomized through long-distance migration. In other words, if new populations were most often founded by one or a few indi-viduals from a distant population rather than from a neighboring population, a random arrangement of cytotypes in geographic space would result. In the Redding area, the range of distances separating populations is 118 miles to 2 miles, with an average of 18 miles, so "long-distance" dispersal in this case does not involve unreasonably large distances. 68 Finally, the random pattern of cytotypes in geographic space in the Redding area could reflect that cytotypes are not fixed in populations or that racial designations for my populations are incorrect. I have designated populations as belonging to particular races based on their matching D N A samples from the Carrs' cytologi-cally typified collections. In doing so, I have assumed, as the Carrs have suggest-ed (G.D. Carr, 1975; R.L. Carr & G.D. Carr, 1983), that populations they col-lected are fixed for a given cytological arrangement. Nevertheless, the Carrs have also described a few populations in the Redding area that are known to contain at least two, and sometimes more, cytotypes (R.L. Carr & G.D. Carr, 1983). If too few samples were taken from populations for which the Carrs described fixed cytotypes, cytological polymorphism within these populations may have been overlooked. Further investigation is required to test which of these three hypotheses may be correct. For example, more exhaustive biosystematic studies of cytological varia-tion within populations need to be conducted. These would involve growing numerous plants from each population to maturity and performing crosses between many individuals from many different populations. Patterns of chromo-somal pairing at meiosis in hybrids between populations would indicate whether populations contain just one cytotype or many. 4.4 Distribution of Per Locus F c T The final test that was performed involved comparing the Fs T of each individual RAPD marker locus. In this test, the allele frequency of each locus in each pop-ulation was used to assess the degree of divergence of each locus relative to all other loci. Noor et al (2001a,b) and Rieseberg et al (1995) have demonstrated that loci found within chromosomal rearrangements are often protected from gene flow, and therefore exhibit higher F S T than for loci external to rearrange-ments. My purpose in performing this test was to identify loci with particularly high F S T . Such loci may become the focus of a future mapping study in the Whitton lab which would seek to identify loci within and external to chromoso-mal rearrangements and would compare levels of gene flow for loci within and outside of rearranged segments. I have identified at least two, and perhaps as many as ten RAPD markers that have very high F S T relative to all other markers, suggesting these may represent loci within chromosomal rearrangements. Since most loci used in this study had a low F S T , either very few of the loci used in this study lie within chromosomal rearrangements, or if loci are equally distrib-uted within and outside of rearranged segments, then chromosomal rearrange-69 merits have no effect on inhibiting the introgression of loci within their length. Given the large number of loci used in this study (88) and the relatively small size of the Calycadenia genome (six chromosomes, three of which do not rearrange), it is likely that some loci fell within rearranged segments. If the loci with high F S T are truly the only loci used in this study that are within inter-changed segments, their high level of differentiation relative to other loci suggests that chromosomal rearrangements may currendy play an important role in inhibit-ing introgression of loci housed within rearraged segments. To verify whether this is so, additional tests will have to be performed, involving methods that can identify the chromosomal regions to which these markers map. Techniques such as Fluorescence In-Situ Hybridization (FISH) may be useful to this end. 4.5 Summary & Future Directions 4.5.1 Summary The present study was undertaken to test whether chromosomal rearrangements that distinguish populations in the Calycadenia Jremontii- C. pauciflora complex have played a primary role in driving divergence among them. To do this, genetic distances among populations were calculated from allele frequencies based on polymorphic PvAPD markers shared among populations. Genetic distances were used to construct Neighbor-Joining and Fitch-Margoliash trees from which past patterns of gene flow among populations and cytological races could be inferred. As an alternative way of inferring relationships among populations, centroid F S T was also calculated for each population and for groups of populations that emerged from genetic distance analyses. Mantel tests were used to examine corre-lations among genetic distances geographic distances, and cytological differences among populations, in an effort to explain the patterns of relationships among populations that resulted from genetic distance analyses. Finally, the distribution of F S T for each locus was plotted to judge whether any of the 88 PvAPD loci used in this study may provide a useful starting point for future genetic map-based studies that will investigate the role of chromosomal rearrangements as par-tial barriers to introgression in Calycadenia. Based on the genetic distance trees, races clearly do not form the monophyletic clades that would be expected if cytological differences among groups have pre-sented a strong, primary barrier to gene flow. Mantel test results substantiate that cytological differences among populations do not account for the genetic differ-ences detected among them. Instead, most of the genetic variation among popu-lations can be explained by geography, suggesting that populations are isolated by distance, rather than by chromosomal barriers. 70 Regardless of which factor represents the primary barrier to gene flow among populations, evidence that populations are not well differentiated can be found in both the failure to resolve relationships among populations in distance analyses (Figures 3.7, 3.8) and in the low values of F S T for most populations. While the results of all the tests I have performed in this study suggest that some gene flow has occurred among populations in the Calycadenia jremontii - Cpauci -flora complex despite cytological differences, it is also possible that cytological dif-ferences have evolved too recently to be detectable using RAPD markers. If this is the case, the best way to test whether cytological rearrangements may inhibit gene flow at present would be to perform numerous hybrid crosses between pairs of populations of the same and of different races across the entire range of the two currently recognized species. 4.5.2 Recommendations for Future Directions I have several recommendations for future avenues of research that would assist in resolving the questionable evolutionary significance of cytological differences among the populations in the Calycadenia fremontii - C. pauciflora complex. i) To compare the genetic distance trees based on RAPD markers and trees generated using ITS and ETS sequence data, both techniques must be performed on the same populations and the same individuals. Given the extraordinarily high cost of sequencing relative to RAPD-PCR, my recommendation would be that for every one of my populations,one or two individuals be sequenced and be included in Baldwin's tree, and for every one of Baldwin's populations, his sam-ples be used to generate RAPD fingerprints to include in my distance trees. ii) Numerous hybrid crosses, namely those between races from different species, remain uncompleted. To date, the Carrs have presented data for all intraracial crosses, but not for all possible interracial crosses. Only Calycadenia fremontii "Ciliosa" has been crossed with all races in this complex; all the other races of C. fremontii must still be crossed with all the races of C. pauciflora. Furthermore, these crosses must be performed between pairs of populations, not between pooled populations of the same race. In this way, crossing data will more accu-rately reflect interracial crossing differences and will clarify whether the races truly are evolutionarily significant, reproductively isolated units. Crosses made in this way will also highlight the standard deviation in hybrid fer-tility or crossability within races. At present, we do not know with certainty whether the deviation in intraracial fertility is greater or less than the difference between the mean of intraracial fertility and the mean of interracial fertility. If the standard deviation were greater than the difference of the means, one could not infer a significant difference between rates of intraracial fertility and rates of 71 interracial fertility. Finally, statistics on the rate of crossability - that is, the ease with which hybrids can be formed between any two populations - as opposed to the hybrid fertility would also be valuable. It is entirely possible that two populations of a given race may produce hybrid progeny that are nearly perfectly fertile, but if only one such progeny is formed out of 10,000 crosses, the populations would be effective-ly isolated despite their ability to form fertile hybrids. Crosses made among numerous populations of all the races could also potentially reveal actual differences in crossability or fertility among groups other that cyto-logical races. For example, patterns of barriers to crossability may emerge for groups based on flower color or other morphological differences, on geography, or on ecology. iii) If regions of the genome within chromosomal interchange segments are pro-tected from recombination and also from introgression as Rieseberg et ai (1995) and Noor et al (2001a,b) have suggested, physical mapping of the Calycadenia genome may identify marker loci within and external to interchanged regions of chromosomes in C. fremontii and C. pauciflora. Rates of gene flow for markers within interchanged regions can then be compared to rates of gene flow for markers external to interchanged regions to determine whether chromosomal rearrangements in C. fremontii - C. pauciflora inhibit gene flow for any portion of the genome. 72 5. References Anamthawat-Jonsson, K., Long, S., Basrur, P., & Adalsteinsson, S. (1992). Reciprocal translocation 12 20ql2 q22 in an Icelandic sheep. Research in Veterinary Science, 52, 367-370. Baldwin, B. (1993). Molecular phylogenetics of Calycadenia (Compositae) based on ITS sequences of nuclear ribosomal DNA: Chromosomal and morphological evolution reexamined. American Journal of Botany, 80, 222-238. Baldwin, B., & Markos, S. (1998). Phylogenetic utility of the external transcribed spacer (ETS) of 18S-26S rDNA: Congruence of ETS and ITS trees of Calycadenia (Compositae). Molecular Phylogenetics and Evolution, 10, 449-463. Beechey, C , & Evans, E. (2000). Standard ideogram / anomaly breakpoints of the mouse. Retrieved March 15, 2002, from M R C Mammalian Genetics Unit, Harwell, Oxfordshire Web site: Bengtsson, B., & Bodmer, W. (1976). On the increase of chromosome mutations under random mating. Theoretical Population Biology, 9, 260-281. Bohm, B., Fong, C , Hiebert, M . , Jamal, A., & Crins, W. (1992). Non-polar flavonoids of Calycadenia, Lagophylla, and Madia. Phytochemistry, 31, 1261-1263. Buckler, E. , Phelps-Durr, T., Buckler, C , Dawe, R., Doebley, J., & Holtsford, T. (1999). Meiotic drive of chromosomal knobs reshaped the maize genome. Genetics, L53, 415-426. Carr, G. (1975). Chromosome evolution and aneuploid reduction in Calycadenia pauciflora (Asteraceae). Evolution 29, 681-699. Carr, G. (1977). A cytological conspectus of the genus Calycadenia (Asteraceae): An example of contrasting modes of evolution. American Journal of Botany, 64, 694-703. Carr, G. (1980). Experimental evidence for saltational chromosome evolution in Calycadenia pauciflora Gray (Asteraceae). Heredity, 45, 107-112. Carr, G., & Carr, R. (1983). Chromosome races and structural heterozygosity in Calycadenia ciliosa Greene (Asteraceae). American Journal of Botany, 70, 744-755. 73 Carr, G., & Carr, R. (2000). A new chromosome race of Calycadenia pauciflora Gray (Asteraceae: Heliantheae - Madiinae) from Butte County, California. American Journal of Botany, 87, 1459-1465. Casgrain, P., Legendre, P., & Vaudor, A. (2001). Le progiciel R / The R package for multidimensional and spatial analysis Version 4.0 [Computer program]. Montreal, PQ: Universite de Montreal. Cavalli-Svorza, L. & Bodmer, W. (1971). The genetics of human populations. San Francisco: W H Freeman. Clausen, J. (1951). Stages in the evolution of plant species. Ithaca, NY: Cornell University Press. Clausen, j . , Keck, D., & Hiesey, W., (1945). Experimental studies on the nature of species (Vol. 2): Plant evolution through amphiploidy and autoploidy, with examples from the Madiinae. Washington, D C : Carnegie Institution of Washington Publication 564. Cleland, R. (1972). Oenothera: Cytogenetics and evolution. London: Academic Press. Cleland, R. (1962). Cytogenetics of Oenothera. Advances in Genetics, i i , 147-237. Cockerham, C , & Weir, B. (1993). Estimation of gene flow from F-statistics. Evolution, 47, 855-863. Coyne, J. (1993). The genetics of an isolating mechanism between two sibling species of Drosophila: Evolution, 47, 778-788. Coyne, J., & Orr, H . (1998). The evolutionary genetics of speciation. Philosophical Transactions of the Royal Society of London Series B - Biology, 353. 287-305. Daniel, A., Hook, E., & Wulf, G. (1989). Risks of unbalances progeny at amniocentesis to carriers of chromosome rearrangements: data from USA and Canadian laboratories. American Journal of Medical Genetics, 33, 14-53. Dennis, A. (2000). CalFlora database - distributional data: Diversity oj Calycadenia. Retrieved January 16, 2000, from CalFlora Database Web site: http://v^\^es&gen=Calycadenia. 74 Doyle, J.J. and Doyle, J.L. (1987). A rapid D N A isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin, 19, 11-15. Ducos, A., Dumont, P., Seguela, A., Pinton, A., Berland, H. , Brun-Baronnat, C , Darre, A., Marquant-Le Guienne, B., Humblot, P., Boichard, D., & Darre, R. (2000). A new reciprocal translocation in a subfertile bull. Genetics Selection Evolution, 32, 589-598. Ellstrand, N. , & Levin, D. (1982). Genotypic diversity in Oenothera laciniata (Onagraceae), a permanent translocation heterozygote. Evolution, 36, 63-69. Felsenstein, J. (1995). PHYLIP: Phylogeny inference package Version 3.573c [Computer program]. Seattle, WA: University of Washington. Hickman, J.C., ed. (1993). The Jepson Manual: Higher Plants of California. Berkeley: University of California Press. King, M . (1993). Species evolution: the role of chromosome change. Cambridge: Cambridge University Press. Kyhos, D. (1965). The independent aneuploid origin of two species of Chaenactis (Compositae) from a common ancestor. Evolution, 19, 26-43. Kyhos, D., & Carr, G. (1994). Chromosome stability and lability in plants. Evolutionary Theory, 10, 227-248. Lande, R. (1979). Effective deme sizes during long-term evolution estimated from rates of chromosomal rearrangement. Evolution, 33, 234-251. Lande, R., & Wilkinson, G. (1999). Models of sex-ratio meiotic drive and sexual selection in stalk-eyed flies. Genetical Research, 74, 245-253. Levin, D. (1990). The seed bank as a source of genetic novelty in plants. The American Naturalist, 135. 563-572. Lewis, H . (1962). Catastrophic selection as a factor in speciation. Evolution, 16, 257-271. Lewis, H . (1966). Speciation in flowering plants. Science, 152, 167-172. Lewis, H. , & Roberts, M . (1956). The origin of Clarkia lingulata. Evolution, 10, 126-138. 75 Lynch, M . , & Milligan, B. (1994). Analysis of population structure with RAPD markers. Molecular Ecology, 3, 91-99. Maddison, W., & Maddison, D. (1997). MacClade Version 3.07 [Computer program]. Sunderland, MA: Sinauer Associates. Makinen, A., Pitkanen, T., & Andersson, M . (1997). Two cases of reciprocal translocations in domestic pigs producing small litters. Journal oj Animal Breeding & Genetics, 114, 377-384. McNeilly, T., & Antonovics, J. (1967). Evolution in closely adjacent plant populations. Heredity, 23, 205-218. Miller, M . (2000). TFPGA: Tools jor population genetic analysis [Computer program]. Logan, UT: Utah State University. Nei, M . (1972). Genetic distance between populations. The American Naturalist, 106. 283-292. Nevo, E. (1991). Evolutionary theory and processes of active speciation and adaptive radiation in subterranean mole rats, Spalax ehrenbergi superspecies, in Israel. Evolutionary Biology, 25, 1-125. Noor, M . , Grams, K., Bertucci, L., Almendarez, Y. , Reiland, J., & Smith, K. (2001a). The genetics of reproductive isolation and the potential for gene exchange between Drosophila pseudoobscura and D. persimilis via backcross hybrid males. Evolution. 55. 512-521. Noor, M . , Grams, K., Bertucci, L., & Reiland, J. (2001b). Chromosomal inversions and the reproductive isolation of species. Proceedings oj the National Academy oj Sciences oj the United States oj America, 98, 12084-12088. Pialek, M . , Hauffe, H. , Rodriguez-Clark, K., & Searle, J. (2001). Raciation and speciation in house mice from the Alps: the role of chromosomes. Molecular Ecology, 10, 613-625. Reynolds, J., Weir, B., & Cockerham, C. (1983). Estimation of the coancestry coefficient: Basis for a short-term genetic distance. Genetics, 105, 767-779. Rieseberg, L. (1996). Homology among RAPD fragments in interspecific comparisons. Molecular Ecology, 5, 99-105. 76 Rieseberg, L. (2001). Chromosomal rearrangements and speciation. Trends in Ecology and Evolution, 16, 351-358. Rieseberg, L., Linder, C , & Seller, G. (1995). Chromosomal and genie barriers to introgression in Helianthus. Genetics, 141. 1163-1171. Sites, J., & Moritz, C. (1987). Chromosome evolution and speciation revisited. Systematic Zoology, 36, 153-174. Spirito, F. (1998). The role of chromosomal change in speciation. In Howard, D., & Berlocher, S. (Ed.), Endless forms: species and speciation (pp. 320-329). New York: Oxford University Press. Stebbins, G. (1952). Aridity as a stimulus to plant evolution. The American Naturalist, 86, 33-44. Stelly, D., Kautz, K., & Rooney, W. (1990). Pollen fertility of some simple and compound translocations of cotton. Crop Science, 30, 952-955. Swofford, D. (2002). PAUP*: Phylogenetic analysis using parsimony (* and other methods) Version 4.0b8a for the Macintosh Power PC [Computer program]. Sunderland, MA: Sinauer Associates. Therman, E. , & Susman, M . (1993). Human Chromosomes: Structure, Behavior, and Effects(3rd ed.). New York: Springer-Verlag. Ungerer, M . , Baird, S., Pan, J., & Rieseberg, L. (1998). Rapid hybrid speciation in wild sunflowers. Proceedings of the National Academy of Sciences of the United States of America, 95, 11757-11762. Van Buren, R., Harper, K., Andersen, W., Stanton, D., Seyoum, S., & England, J. (1994). Evaluating the relationship of autumn buttercup (Ranunculus acriformis var. aestivalis) to some close congeners using randomly amplified polymorphic DNA. American Journal of Botany, 81, 514-519. Walsh, J. (1982). Rate of accumulation of reproductive isolation by chromosome rearrangements. The American Naturalist, 120, 510-532. Weir, B. (1990). Genetic data analysis. Sunderland, MA: Sinauer Associates. Weir, B., & Cockerham, C. (1984). Estimating F-statistics for the analysis of population structure. Evolution, 38, 1358-1370. 77 White, M . (1973). Animal cytology and evolution (3rd ed.). Cambridge: Cambridge University Press. White, M . (1968). Models of Speciation. Science, 159, 1065-1070. Williams, J., Kubelik, A., Livak, J., Rafalski, J., & Tingey, S. (1990). D N A polymorphisms amplified by arbitrary primers are useful as genetic markers. Nucleic Acids Research, 18, 6531-6535. Wolfe, A., & Liston, A. (1998). Contributions of PCR-based methods to plant systematics and evolutionary biology. In Soltis, D.E., Soltis, P.S., & Doyle, J.J. (Eds.). (1998). Molecular systematics of plants II: DNA seauencing. Norwell, MA: Kluwer Academic. Wright, S. (1930). The genetical theory of natural selection: A review. Journal oj Heredity, 21, 349-356. Wright, S. (1941). On the probability of fixation of reciprocal translocations. The American Naturalist, 75, 513-522. 78 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items