INFORMATION TO USERS

This was produced from a copy of a document sent to us for microfilming. While the most advanced technological means to photograph and reproduce this document have been used, the quality is heavily dependent upon the quality of the material submitted.

The following explanation of techniques is provided to help you understand markings or notations which may appear on this reproduction.

1.The sign or "target” for pages apparently lacking from the document photographed is "Missing Page(s)”. If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting through an image and duplicating adjacent pages to assure you of complete continuity.

2. When an image on the film is obliterated with a round black mark it is an indication that the film inspector noticed either blurred copy because of movement during exposure, or duplicate copy. Unless we meaht to delete copyrighted materials that should not have been filmed, you will find a good image of the page in the adjacent frame. If copyrighted materials were deleted you will find a target note listing the pages in the adjacent frame.

3. When a map, drawing or chart, etc., is part of the material being photo­ graphed the photographer has followed a definite method in "sectioning” the material. It is customary to begin filming at the upper left hand corner of a large sheet and to continue from left to right in equal sections with small overlaps. If necessary, sectioning is continued again—beginning below the first row and continuing on until complete.

4. For any illustrations that cannot be reproduced satisfactorily by xerography, photographic prints can be purchased at additional cost and tipped into your xerographic copy. Requests can be made to our Dissertations Customer Services Department.

5. Some pages in any document may have indistinct print. In J l cases we have filmed the best available copy.

University Microfilms International 300 N. ZEEB RD„ ANN ARBOR, Ml 48106 8214135

Roberts, Marvin Lee

SYSTEMATIC STUDIES OF NORTH AMERICAN BIDENS SECTION BIDENS (COMPOSITAE)

The Ohio State University Ph.D. 1982

University Microfilms International 300 N. Zeeb Road, Ann Arbor, M I 48106

Copyright 1982 by Roberts, Marvin Lee All Rights Reserved PLEASE NOTE:

In all cases this material has been filmed in the best possible way from the available copy. Problems encountered with this document have been identified here with a check markV .

1. Glossy photographs or pages

2. Colored illustrations, paper or print,_____

3. Photographs with dark background J A. Illustrations are poor copy_____

5. Pages with black marks, not original copy_____

6. Print shows through as there is text on both sides of page_____

7. Indistinct, broken or small print on several pages_____

8. Print exceeds margin requirements_____

9. Tightly bound copy with print lost in spine_____

10. Computer printout pages with indistinct print_____

11. Page(s)______lacking when material received, and not available from school or author.

12. Page(s)______seem to be missing in numbering only as text follows.

13. Two pages numbered______. Text follows.

14. Curling and wrinkled pages_____

15. Other______

University Microfilms International SYSTEMATIC STUDIES OF NORTH AMERICAN BIDENS SECTION BIDENS (COMPOSITAE)

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of the Ohio State University

B y Marvin L. Roberts• B.S., M.S.

The Ohio State University

1982

Reading Committeei Approved By

Dr. Daniel J. Crawford

Dr. Tod F. Stuessy

Dr. Ralph E. Boerner otany ACKNOWLEDGMENTS t

I wish to thank the numerous people who have assisted in the completion of this dissertation. Par­ ticular gratitude is extended to Dr. Daniel J . Crawford for continuous support and advice since the hegining of the project. Dr s. Tod F. Stuessy and Ralph E. Boemer, of my reading committee*deserve thanks for their patience and many criticisms which improved this dissertation. I also thank Dr. Ronald L. Stuckey for advice, information, and criticism regarding many aspects of this project. The line drawings were provided "by Ms. Cheryl Kimerline. Field studies were supported by National Science Foundation Doctoral Dissertation Improvement Grant DEB-78-20212. I wish to thank the curators of the herbaria who loaned specimens for this study (CU, DAO, F, GH, MTJB, NEBC, PH) and to others who have provided material. Special thanks are extended to Dr. A. E. Schuyler of PH for materials and information. VITA

August 7. 1948 Born- Fostoria, Ohio

1971 B.S., The Ohio State University, Columbus,

1972-75 Assistant Curator, Department of Botany, The Ohio State University

1976-77 Graduate Research Associate, Department of Botany, The University of Wyoming, Laramie.

1977 M. S. in Botany, The University of Wyomings Laramie

1977 Graduate Teaching Assistant, Department of Botany, The Ohio State University, Columbus 1981 Visiting Lecturer, Department of Botany, Rutgers University, New Brunswick

PUBLICATIONS

Roberts, M. L. 1972. Butomus umbellatus in the Mississippi Watershed. Castanea 37« 83 -85”.

Roberts, M. L. 1972. Wolffia in the bladders of Utriculariai an "herbivorous" plant? Mich. Bot. 11« 67-69. Roberts, M. L. & J. Youger. 1974. Aquatic and terrestrial plants of the Paint Creek Lake Project Area, pp. 14-8-193, 466-516. In C. E. Herdendorf and D. H. Stansberry, eds. Final Report Environmental Analysis of the Paint Creek Lak§ Project, Ohio. Department of Army, Corps of Engineers, Huntington District.

• • • 111 Stuckey, R. L. & M. L. Roberts. 1974. Additions to the Bibliotheca Rafinesquiana. Taxon 23 * 365-372 .

Stuckey, R. L. & M. L. Roberts. 1974. Bibliography of Theses and Dissertations on Ohio Floristics and Vegetation in Ohio Colleges and Universities.- Ohio Biol. Surv. Inform. Circ. No. 7. 92 pp.

Roberts, M. L. 1976. IOPB Chromosome Number Reports LIII. Taxon 25* 484.

Roberts, M. L. & D. J. Crawford. 1977* An elution chamber for the extraction of flavonoid compounds from paper chromatograms. Phytochem. Bull. 101 6-7. Stuckey, R. L. & M. L. Roberts. 1977- Rare and endangered aquatic vascular plants of Ohio* an annotated list of the imperiled species. Sida 7* 21-41.

Roberts, M. L. 1980. The flavonoids of Meealodonta beckii (Compositae) and isolation of 2',3-dihydroxy-4- methoxy-4'-glucosyl chalcone. Biochem. Syst. & Ecol. 81 115-118. Roberts, M. L., R. L. Stuckey, and R. S. Mitchell, 1981. Hvdrocharis morsus-ranae (Hydrocharitaceae) 1 New to the United States. Rhodora 83 * 147-148.

FIELD OF STUDY

Major Field* Systematic Botany Studies in the systematics and phytochemistry of the genus Bidens (Compositae). Professor Daniel J. Crawford. TABLE OF CONTENTS

Page ACKNOWLEDGMENTS...... ii

VITA...... ill

LIST OF TABLES...... V LIST OF FIGURES...... vii

CHAPTER

I. THE IDENTITY OF BIDENS HETERODOXA AND B. INFIRMA. AND STATUS OF BIDENS SECTIONS PLATYCARPAEA AND HETERODONTA...... 1

Introduction...... 2 Materials and Methods...... 4 Morphology...... 5 Taxonomic History of Bidens heterodoxa... 11 Disposition of Bidens infirma and the varieties of Bidens heterodoxa...... 14 Status of Bidens sections Heterodonta and Platvcaroaea...... 22 Conclusions...... 24 Literature Cited...... 31 II. FLAVONOIDS OF AMERICAN BIDENS SECTION BIDENS (COMPOSITAE)...... 33 Introduction...... 34 Materials and Methods...... 38 Results« Compound Identities...... 41 Results: Compound Destribution...... 48 Discussion and Conclusions ...... 53 Literature Cited...... 84

v CHAPTER Page

III. ALLOZYME VARIATION IN BIDENS DISCPIPEA. 87 Introduction...... 88 Materials and Methods...... 90 Resultsi Morphology and Reproduction.... Results! Genetic Variation...... Discussion...... 100 Conclusions...... 103 Literature Cited...... 108 IV. VARIATION AMONG THE POPULATIONS OF BIDENS BIDENTPIPES (COMPOSITAEi COREOPSIDINAE).... 111

Introduction...... 112 Methods and Materials...... 114 Results...... 117 Discussion...... 127 Literature Cited...... 140 V. CYTOLOGY, BIOLOGY, AND SYSTEMATICS OF MEGALODONTA BECKII (COMPOSITAE)..... 142 Introduction...... 143 Materials and Methods. 144 Results and Discussion 145 Summary...... References...... m APPENDIX A. POPULATIONS OF BIDENS EXAMINED FOR FLAVONOIDS...... 167 LIST OF TABLES Le CHAPTER I Page

1. Varieties of species of Bidens sections Platvcarpaea and Heterodonta (sensu Sherff, 1937) exhibiting extensive parallel awn variation...... 25 2. Distribution of disc corolla and correlated fruit characters in selected species of Bidens section Bidens...... CHAPTER II

1. Taxonomy of Bidens section Bidens according to Sherff (1937, 1955. 1965). with explanatory notes...... 57 2. Morphological characteristics of species groups in Bidens section Bidens. Cytolo'gical data compiled from published sources and my own determinations...... 58 3. Tentative identities of flavonoid glycosides detected in Bidens sect. Bidens*...... 59 Distribution of flavonoids in Bidens sect. Bidens...... 60

5. Color reactions, Rf values, and UV spectral data for selected flavonoids in Bidens sect. Bid6ns« ...... bx CHAPTER III

1. Chromosome counts and reports of Bidens discoidea, B. frondosa. and B. vulgata..#.... 10^ 2. Populations of Bidens discoidea examined for allozymes >.. 105 3. Allele frequencies for seven variable enzyme' loci in Bidens discoidea...... 106

Genetic identities among populations of Bidens discoidea calculated by the method of Nei...... 1°7 vii CHAPTER IV

Page Populations of Bidens bidentoides examined' for enzymes, flavonoids, chromosome number, or morphology ...... 133 Morphological variation in populations of Bidens bidentoides...... 13^ Chromosome numbers of Bidens bidentoides 135 Enzyme phenotype frequencies for polymorphic loci in populations of Bidens bidentoides.... 13° Variability in enzyme phenotypes in populations of Bidens bidentoides...... 13>7

v• m • • LIST OF FIGURES

Figure CHAPTER I Page 1. Achenes of Bidens section Bidens...... 28 2. Disc florets and achenes of Bidens section Bidens...... 30 CHAPTER II

1. UV absorption spectra of compound 4, luteolin-^',7-0-diglucoside...... 65 2. UV absorption spectra of compound 10, butin-7-0-glucoside...... °7

3. UV absorption spectra of compound 10 aglycone, butin ...... 69

4. UV absorption spectra pf compound B, ^-methoxycoreopsin...... 71

5. UV absorption spectra of compound Dc, acylated mare in...... 73 6. UV absorption spectra of compound Da, acylated maritime in...... 75

7. UV absorption spectra of compound E, okanin-V ,^-diglucoside...... 77 8. UV absorption spectra of compound F, 4-methoxymarein...... 79

9. UV absorption spectra of compound G, C-glycosyl xanthone...... 81

CHAPTER IV

1. Diagramatic interpretation of Got-1 and Got-2 electromorphs...... 1^9 CHAPTER V 1. Stem of Megalodonta beckii with segmented leaves, emersed leaves, and a flower at anthesis...... ^ ix Figure Page 2. Turions of Megalodonta beckii...... 161

3. Root tip chromosomes of Megalodonta beckii.. 2n = 26m 163

x CHAPTER I

THE IDENTITY OF BIDENS HETERODOXA AND B. INFIRMA. AND STATUS OF BIDENS SECTIONS PLATYCARPAEA AND HETERODONTA

0

1 INTRODUCTION

Variation in northeastern North American Bidens was extensively investigated by M. L. Femald (1903, 1908, 1913)

1917» 1924, 1932, 1938) and his students (Femald & St. John, 19155 Fassett, 1924, 1925a, 1925b, 1928) in the early part of this century. E. E. Sherff (1937, 1955)» in his monograph of the genus, essentially accepted taxonomic treatments proposed by these workers at the specific and varietal levels to the extent of incorporating Fassett's (1925b) key in his 1955 revision. Sherff place two species of the North American group (sect. Platvcaroaea DC. as delimited by Wiegand, I899) in sect. Heterodonta (Nutt.) Sherff. To B. bidentoides (Nutt.) Britton , the type of section Heterodonta. Sherff added B. eatoni Fern. As thus circumscribed the section was composed entirely of plants confined to freshwater intertidal zones of northeastern North American rivers. Section Platvcaroaea then consisted of ten North American and two primarily Eurasian species (Sherff, 1937)* The taxa involved in understanding the identity of B. heterodoxa Fern, include B. comosa (A. Gray) Wiegand, B. connata Willd., B. frondosa L., and B. tripartita L. of sect. Platvcaroaea. and B. eatoni Fern, of sect. Heterodonta. These same taxa are central to defining species groups and the problem of sectional delimitation and represent most of the morphological 3 diversity within the section. The B. (Platvcamaea) cemua species group (B. c e m u a . B. laevis. B. hyperborea) forms a natural assemblage which is set apart from the other species morpholgically. It is not involved directly in the problems to be discussed. The purpose of this paper is to investigate the taxonomy of B. heterodoxa (including B. infirma). the relationship between sections Platvcamaea and Heterodonta. and to discuss the distribution of several morphological features important to understanding their relationships. MATERIAIS AND METHODS Herbarium specimens of Bidens sections Platycarpaea and Heterodonta were "borrowed from CU, DAO, GH, MTJB, NEBC, and PH. Additional specimens were studied at F, MO, and OS. Additional specimens were collected over a period of three years in field studies in the Great Lakes region and coastal areas form Virginia to Quebec. Many sites were visited repeatedly during different parts of the growing season. An attempt was made to visit type localities or nearby sites of collections mentioned in the protologues. Specimens of representative collections and greenhouse progeny are on deposit at OS. Seeds of individuals of all eastern North American species except B. heterdoxa were obtained and cultivated in the greenhouse. The cultures represented most of the varieties recognized by Sherff (1937)* Special emphasis was placed on raising selfs of progeny from individuals in polymorphic populations. In some cases these progeny have been grown through several generations. All mention of progeny refers to self-pollinated plants. Specimens of progeny of B. eatoni populations grown by H. K. Svenson (deposited at PH) were also examined. These collections from Maine,. Massachusetts, Connecticut, and New York provided an opportunity to examine morphotypes which may no longer be present in the individual natural populations and several series represent progeny of type collections of B. eatoni. varieties described by N. C. Fassett. MORPHOLOGY A discussion of the taxonomic disposition of B. heterodoxa and sections Platycarpaea and Heterodonta requires a discussion of the morphological features which distinguish, or fail to distinguish, the related species and varieties. The taxonomy of the genus Bidens has understandably been based largely on achene morphology.

Particularily in North American species, the most obvious features of the achenes, the awns, are subject to several parallel variations. Similar patterns of variation are found within species of sections Platycarpaea. Heterodonta. and Medusae. Most of this variation has been the basis for taxonomic recognition at the varietal level (Sherff, 1937,

1955)• In sections Platycarpaea and Heterodonta. B. frondosa. B. connata. and B. eatoni exhibit parallel variation recognized at the varietal level as indicated in Table 1. The other species of the two sections are less polymorphic at the population level for these characters and their subspecific taxonomy is not so extensive (Sherff,

1937)• 6 Greenhouse progeny studies indicate that the awn features of individual plants form polymorphic populations are heritable through several generations. For example, individuals of B. connata from Indian Lake, Ohio (Roberts 5506) could be identified as B. connata varieties netiolata (retrorse barbs) anomala (antrorse barbs), and ambiversa (barb direction mixed) on the basis of the direction of awn barbing (Table 1). The plants having different awn forms are indistinguishable on the basis of any other known morphological feature. Progeny of 28 self-pollinated plants (20 antrorse, 2 "mixed", 6 retrorse) breed true through two generations (4-12 progeny per individual). F^ progeny of the two individuals having "mixed" barbs (var. ambiversa) still had a phenotype identical to the original'collection, with a few retrorse barbs interspersed with the predominantly antrorse barbs (as in Figure 3)» Fewer numbers of individuals from other populations of B. connata polymorphic for the direction of awn barbing also showed true-breeding phenotypes.^ These populations originated from Ottawa Co., Ohio (Roberts

5515), Cayuga Lake, New York (Roberts 5487), and the Miramichi River, New Brunswick (Roberts 4956). Wiegand noted the persistence of the polymorphic population in Cayuga Lake, New York (Wiegand & Eames, 1926), where mixed populations of all three awn forms can still be found. Similar progeny studies with individuals of B. frondosa awn variants (antrorse and retrorse) from the St. Lawrence

River, Quebec (Roberts 4938). the Hudson River, New York (Roberts 4-991), and the Delaware River, New Jersey (Roberts 5010), showed no segregation of phenotypes through the Fg generation. Greenhouse studies of S. eatoni from the St. Lawrence River (Roberts 4-931. 4-936. & 54-50) and the Merrimack River, Massachusetts (Roberts 5221) also show that progeny from individuals in polymorphic populations breed true for direction of awn barbing and awn length. The results suggest that each of the polymorphic populations are composed of several inbreed lines which persist for long periods. The individuals varying in awn morphology are otherwise similar in allozyme phenotypes and identical in flavonoid profile within populations (Roberts, unpub.). Features other than awn barbing are also highly heritable in progeny,, but are subject to developmental differences and phenotypic plasticity. Some species of Bidens show pronounced dimorphism between outer and inner

achenes of the same head (Sherff, 1937; Burtt, 1976; Hart, 1979)• In the species considered here the outer achenes differ from the inner in being shorter and having 2 rather than 3* or 3 rather than 4- awns per achene. This type of achene polymorphism is exhibited by each of the species to the extent that Sherff (1937) provides ranges of variation for inner versus outer achenes for most species. A similar type of variation is found associated with smaller heads formed on axillary branches late in the season or on plants which are depauperate due to poor growing

conditions. 8 These smaller heads have fewer flowers which produce smaller achenes having fewer awns. They resemble the achenes of outer whorls of well developed heads on robust plants. This phefiotypic plasticity can be observed within natural populations and in progeny of inbred plants in the greenhouse. Floral features have been used only infrequently to determine systematic relationships within Bidens. Greene (1901) and Wiegand (1899) recognized that differences in the color of the head could separate B. frondosa ftom the vegetatively similar B. vulgata Greene. Similar differences in corolla features distinguish B. comosa from B. connata (Wiegand, 1899) and from other species. The major corolla features and their correlated characters are summarized in Table 2 and Figure 2. The species listed in Table 2 were selected to be representative of the combinations of character states found in the North American species. They sire also the species most often misidentified and which are

involved in the resolution of taxonomic problems associated with B. heterodoxa. Species of the B. cemua group share the corolla features of B. connata and associated species but can easily be distinguished by vegetative morphology and achene features. The corollas of the North American species B. connata. B. frondosa. and B. eatoni have a campanulate throat rather sharply constricted to a short narrow tube as in Figure 2A. The short tube is herbaceous in texture and color. The teeth and upper portion of the corolla are delicate in texture and more heavily colored with yellow pigments. The lower one-third to one-half of the throat is of nearly the same herbaceous texture and color as the tube. After anthesis the teeth and upper portipn of the tube become flaccid, acquire an orange tint, and collapse around the normally exserted anthers. The corolla becomes loosely attatched after anthesis and is readily dislodged. The species of the Platycarpaea with this syndrome of disc corolla features are all native to the Western Hemisphere and share a number of achene features which do not «occur in the contrasting species. They characteristically have named varieties with antrorsely barbed awns, achenes with a < pustulose surface or slightly thickened margins, and achenes which are not strongly flattened or winged. Additional distinguishing features are presented in Table 2. The corollas of the North American species B. comosa and B. vulgata. and the Eurasian B. tripartita are funnelform in general outline as in Figure 2B. The throat of B. comosa is pale yellow, narrow, and gradually tapers to the long, narrow tube. The tube itself is less herbaceous than in the species with eampanulate throats, and the herbaceous texture and green color does not extend far up the throat. After anthesis the entire corolla dries uniformly and acquires a chartaceous texture. The pale anthers are not exserted from the corolla tube either at anthesis or as a result of 10 collapse of the throat. The corollas are relatively persistent and usually remain firmly attatched and erect as the achene matures. The species with this syndrome of corolla characters includes only two American species and two native European species of section Platvcamaea (sensu Sherff, 1937)- The corolla characters, in conjunction with correlated achene characters, suggest a closer relationship for B. comosa and B. vulgata with the Eurasian B. tripartita than with other sympatric American species. All have achenes which are relatively strongly flattened, smooth or sparsely pubescent, and often somewhat alate (Table 2). TAXONOMIC HISTORY OF BIDENS HETERODOXA The identity of Bidens heterodoxa has been problematical

since its description and is illustrative of the questionable

separation of the North American species into two sections.

The taxon was originally described by Femald (1913) from Prince Edward Island, Canada, as var. heterodoxa of the.

polymorphic Eurasian species B. tri-partita. Fernald (1913) based the separation on the presence of upwardly barbed awns in the Prince Edward Island plants as compared to the

uniformly retrorsely barbed awns of the Eurasian plants. Thus began a tradition which was to persist for about thirty years, namely creating names at the varietal level for variants of many North American species which are polymorphic for the direction of awn barbing and other awn features

(Table lj Sherff, 1937). Fernald later elevated his variety heterodoxa to specific rank as B. heterodoxa and in addition described variety orthodoxa to accomodate similar plants with retrorsely barbed awns occuring in coastal marshes of the

Magdalen Islands, Quebec (Fernald & St. John, 1915). Two additional varieties, B. heterodoxa var. monardaefolia and var. agnosti'ca were described from

freshwater lake Pocotopaug, Connecticut (Fernald, 1917i

Boivin, 1967, indicates the spelling should be corrected

to monardifolia). Femald (1917) indicated that they were

similar to the two varieties from eastern Canada in having 11 12 2-awned achenes, but differed in having petiolate, bluntly toothed leaves and short flowering branches. He indicated that var. agnostica differed from var. monardifolia only in the former having nearly smooth achenial awns (i.e. without well developed barbs) (Fernald, 1917). In comparing these two varieties to other species, he seemed doubtful as to where to place the two varieties form Connecticut, stating that they "must, at least for the present, be placed with

B. heterodoxa." (Fernald, 1917 p. 258). Fassett (192*0 described a plant from the freshwater estuaries of the Kennebec River, Maine, as B. heterodoxa var. interstes. He suggested that the leaf morphology of the plant was most similar to B. eatoni var. Itennebecensis Fern, with which it was sympatric, while the other characters were "intermediate" with those of var. monardifolia and var. orthodoxa (Fassett, 192*0. A year later, in his monograph of B. eatoni, Fassett (1925) considered var. interstes one of the seven varieties of that polymorphic species, because of the smaller heads than the other varieties of B. heterodoxa. The last taxon associated with the name B. heterodoxa is from the St. Lawrence River estuary, Quebec. Fernald (1932) named this B. heterodoxa var. atheistica. and- subsequently raised it to specific rank as B. Infirma (Fernald, 1938). He characterized this species as having the awns absent or reduced, with antrorse barbs. Sherff (1937) apparently overlooked Fernald's (1932) description of var. atheistica and cited no specimens from the protologue in his monograph under other names. In his 1955 revision Sherff treated the St. Lawrence material as B. heterodoxa var. atheistica in section Platvcaraaea. DISPOSITION OF BIDENS INFIRM AND THE VARIETIES

OF BIDENS HETERODOXA.

While investigating the variation primarily within what

Sherff (1937) had treated as Bidens section Heterodonta. it was necessary to evaluate the status of the varieties which had been placed in both B. (Heterodonta) eatoni and

B. (Platvcaroaea) heterodoxa. B. heterodoxa var. atheistica

Fern. (= B. infirma Fern.) was particularily intriguing because it had been placed in sect. Platycarnaea by Sherff, was an intertidal zone plant, and was a narrow geographic endemic. Related estuarine taxa had all been placed by Sherff with section Heterodonta. Field and herbarium studies indicated that B. infirma formed large populations of plants in the St. Lawrence in habitats similar to those occupied by B. eatoni in the northeastern United States. Furthermore, it occurred in mixed populations in these localities with what would be identified (sensu Sherff, 1937* 1955) as B. eatoni var. eatoni. var. fallax. var. mutabilis. and var. illicita (Roberts 4926. 4928. 4931. 4936. 6141. 6146. 6460.

The populations of B. infirma along the St. Lawrence are polymorphic for a number of achene features. Most individuals have the awns reduced or entirely absent, leaving a cuneate achene with a rounded summit (Figure 1A).

Others have short awns (o.5-1*0 ram) with either antrorse

(as the original description), mixed, or retrorse (Figure 10)

14 barbs. Occurring with these plants are occasional individuals with well developed awns (to 3.4> mm) barbed in all possible combinations. The holotype (Femald 2952,GH) and specimens cited in the protolgues of B. heterodoxa var. atheistica and B. infirma (Fernald, 1932» 1938) are predonimately awnless or have short antrorsely barbed awns. However, even one of these sheets (Fernald 2960. GH) has a plant with short retrorsely barbed awns (as in Figure IB). Other specimens collected from the same localities by r Fernald were identified as B. eatoni var. fallax (Fernald

2963, GH; 3-4- awns with retrorse barbs) or as B. connata var.

netiolata (Femald 2953. GH; 3 awns with retrorse barbs). Other than the awn differences the plants fall within the range of variation found in other populations of B. eatoni. Some of the plants in the St. Lawrence River have achenes virtually indistinguishable from those of B. eatoni var. mutabilis from the Kennebec River of Maine (Figure 1C) and very similar to those of B. eatoni from the Miramichi River, New Brunswick (Figure ID) . Even the individuals with reduced awns (Figures 1A & IB) bear a remakable resemblance to B. eatoni (Figure 1C) in appearance of the achene body. The St. Lawrence River plants are also similar to B. eatoni in vegetative features. In their natural habitats they are often stunted by constant innundation on the low tidal flats, but similar depauperate forms occur in the Miramichi River, New Brunswick (Roberts 5180), the Kennebec River, Maine (Fassett 2128. GH) and elsewhere. Plants at 16 higher levels with respect to high tide are more robust and typically form tripartite primary leaves and simple petiolate upper leaves as do most B. eatoni plants

i While the achene variation found in the St. Lawrence River populations is extreme compared to southern populations, the patterns of variation are parallel (Table 1). Only the reduction in awn length exceeds the intrapopulational variation found in other localities. The recognition of the St. Lawrence populations at specific rank as B. infirma Fernald is certainly not justified. Recognition of the variants at the varietal level (under B. eatoni) is also questionable, given the intrapopulational polymorphism. Such recognition, to be compatible with the Fassett (1925) and Sherff (1937* 1955) taxonomy, would require varietal epithets for each of the combinations of awn polymorphisms. Such excessive splitting has already occurred in the polymorphic populations of the Merrimack estuary, Massachusetts which contains two "endemic" varieties of B. eatoni (var. eatoni & illicita) and the Kennebec River, Maine which contains three "endemic" varieties (vars. kennebecensis. interstes. mutabilis: Sherff, 1937* 1955)• Little is to be gained by taxonomic recognition of additional combinations of polymorphisms in other populations. Therefore, B. infirma (= B. heterodoxa var. atheistica) should be considered conspecific with the polymorphic species B. eatoni Fern. 17 higher levels with respect to high tide are more robust and typically form tripartite primary leaves and simple petiolate upper leaves as do most B. eatoni plants While the achene variation found in the St. Lawrence

River populations is extreme compared to southern populations, the patterns of variation are parallel (Table 1). Only the reduction in awn length exceeds the intrapopulational variation found in other localities. The recognition of the St. Lawrence populations at specific rank as B. infirma Fernald is certainly not justified. Recognition of the variants at the varietal level (under B. eatoni) is also questionable, given the intrapopulational polymorphism.

Such recognition, to be compatible with the Fassett (1925) and Sherff-(1937» 1955) taxonomy, would require varietal epithets for each of the combinations of awn polymorphisms.

Such excessive splitting has already occurred in the polymorphic populations of the Merrimack estuary, Massachusetts which contains two "endemic*' varieties of B. eatoni (var. eatoni & illicita) and the Kennebec River, Maine which contains three "endemic" varieties (vars. kennebecensis. interstes. mutabilisi Sherff, 1937» 1955)* Little is to be gained by taxonomic recognition of. additional combinations of polymorphisms in other populations. Therefore, B. infirma (= B. heterodoxa var. atheistica) should be considered conspecific with the polymorphic species B. eatoni Fern. The other varieties of Bidens heterodoxa are more problematic because extant populations are unavailable for investigation. Fernald (1913) originally placed the type variety of B. heterodoxa. from a salt marsh on Prince Edward Island, in the Eurasian species B. tripartita on the basis of its variable leaf forms and flattened achenes. He indicated it differed from the Eurasian species proper in having only two awned fruits with antrorse barbs. In a later paper, when they elevated the variety to specific rank as

B. heterodoxa. Fernald & St. John (1915) also expanded the description to include 2-k awned and pubescent fruits. Fernald & St. John's (1915) diagnosis, of B. heterodoxa var. orthodoxa. based on more mature specimens from the Magdalen Islands, Quebec, simply indicates they are retrorsely barbed. Examinations of the types, all other specimens cited in the protologues (Fernald, 1913; Fernald & St. John, 1915), and other specimens form the type localities indicates the two varieties are conspecific with B. connata. The flowers are not of the funnelform type found in B. tripartita, but have campanulate throats as in B. connata (Figure 2A). The achenes, although immature, show the development of a truncate apex, pustulose surface, carinate face, and four awns on some achenes (Figure 2A). The similarities with B. connata have also been suggested by Sherff (1937) and Fassett (1957)» The varities heterodoxa and orthodoxa should be relegated to synonomy with B. connata. 19 It is possible that Fernald's concept of var. orthodoxa was originally based partly on material which included a mixture of B. connata with B. frondosa. On a specimen of B. frondosa (Fernald. Long. & St. John 8199. GH) from the type locality of var. orthodoxa. Fernald indicated in a note on the sheet "B. heterodoxa. var. orthodoxa (material mixed with no. 8199 during mounting)." The two depauperate specimens are clearly B. frondosa L. The possible inix-up would explain the emphasis in the original description of var. heterodoxa of the "2 awned" achenes. The two varieties of B. heterodoxa described from Connecticut differ from each other only in var. agnostica having the upper portions of the awns naked, while var. monardifolia (Figure 2B) has the awns retrorsely barbed. Both have corolla and achene features associated with B. comosa (Table 2), such as included anthers, flattened achenes, and funnelform corollas. Contrary to the original decriptions (Fernald, 1917), some of the achenes are 3-awned rather than 2-awned. The plants also have the distinctive short-racemose branching pattern of B. comosa. with the very short axillary branches diverging at an acute angle with the main stem. They also resemble B. comosa in the pale, somewhat glaucous, appearance of the leaves, the alate petioles, the eccassionally lobed blades, and long {foliose phyllaries. The plants of both varieties differ from the common form of B. comosa in having the leaves with longer with more distinct petioles. Leaf forms similar to those 20 of B. heterodoxa vars. agnostica and monardifolia are found

in plants of B. comosa from southern Ontario and Quebec (Marie-Victorin 9737. GH; Raymond & Cinq-Mars sn. 14 Sep 1932. MTJB). Although I have not encountered normally developed individuals of B. comosa with smooth awns in an examination of hundreds of herbarium specimens, or among numerous populations in the field, such polymorphisms are not unexpected in the group. The pattern of variation among the specimens of B. heterodoxa form Lake Pocatopaug suggests that var. agnostica does differ from var. monardifolia only in having smooth awns, and that the degree of barbing on individual plants and achenes of var. agnostica is variable. Thus, the variation parallels the pattern of intrapopulational awn polymorphism found in other species (Table l). I have been unable to locate additional material of the Pocatopaug morphotypes in two visits to the type locality. The most recent collections known to me are from 1932 (Jansson s.n.. 25 Sep 1932. both vars; CU). It is probable that shoreline development at the type locality has caused the dissapearence of the morphotypes. The varieties monardifolia and agnostica of B. heterodoxa should be considered conspecific with B. comosa to indicate their natural morphological relationships. Bidens heterodoxa var. intersetes Fassett was later combined as a variety of B. eatoni by Fassett (1925)• An examination of the type (Fassett 852. GH) and other specimens from the polymorphic population at the type locality, indicates the variety is most similar to B. eatoni in its floral, achene, and vegetative features, and should he retained there as has heen done hy Sherff (1937, 1955) and Fernald (1950). Its recognition as a variety distinct from the other sympatric polymorphic awn variants in the Kennebec estuary is less certain. STATUS OF BIDENS SECTIONS HETERODONTA AND PLATYCARPAEA • The similarity between Bidens sections Platycarpaea and Heterodonta is indicated by the previous differences in disposition of the elements of B. heterodoxa and B. eatoni.

Sherff (1937» 1955) also indicates the closeness of several varieties of B. (Heterodonta) eatoni to B. (Platycarpaea) connata. Scoggan (1979s following Boivin, 1966) suggests merging B. comosa. B. connata. B. eatoni. B. infirma. and the Canadian elements of B. heterodoxa into a simgle species under the name of the Eurasian species B. tripartita, although this inclusive species concept merges too many distinct and recognizable elements, it does indicate the species which should be included in a natural species group which contains members currently classified in two sections. If an inclusive species concept were to be followed, it would be more natural to include B. comosa as an element of the European B. tripartita, and the other American taxa under the name B. connata. The present arrangement also does not recognize the greater differences between the species discussed here and the species of the B. (Platycarpaea) cernua group (B. cernua. B. laevis, B. hvperborea). A more reasonable course, which better reflects natural relationships, would be to consider all species of both sections as composing a single section in which several natural species complexes occur. 22 Flavonoid chemistry also supports the natural unity of the species presently placed in the two sections. There are no consistent differences "between B. (Heterodonta) eatoni and several species of Sherff's Platycarpaea (Roberts, unpub.). The real chemical dichotomy among all the species of the two sections is with the species in the B. cernua group and remaining species. The common biosynthetic capacity of the species in the two sections is another indication that they are closely related.

A single section containing all of the members of both sections would include B. tripartita L., the type of the genus. The correct sectional name, according to Article 22 of the International Code of Botanical Nomenclature (Stafleu et al, 1978) would be Bidens L. sect. Bidens. CONCLUSIONS

’ Bidens heterodoxa is composed, as currently recognized, of five varieties each of which is more properly placed with other more widespread species. The inclusion of the varieties of B. heterodoxa in B. connata. B. eatoni. and B. comosa does not substantially enlarge the range of variation already described in these polymorphic species. Bidens heterodoxa. B. connata. and B. eatoni illustrate the history of describing variant phenotypes as varieties when these occur as intrapopulational polymorphisms in otherwise indistinguishable plants. The description of variants for awn length and direction of barbing has resulted in naming many subspecific taxa in North American Bidens on the basis of minor differences. The parallels in variation and overall similarity among the species of Bidens sect. Platycarpaea and sect. Heterodonta indicates that subgeneric categories separating the sections are unwarranted. The species of both sections should compose a larger, more inclusive Bidens sect. Bidens.

24 25

Table 1. Varieties of species of Bidens sections. Platycarpaea and Heterodonta (sensu Sherff, 1937) which exhibit extensive parallel awn variation. Only varieties distinguished primarily by awn features are included.

Awn Species • Condition B. frondosa B. connata B. eatoni

Barbs var. frondosa var. connata var. eatoni retrorse

Barbs var. anomala var. anomala var. fallax antrorse

Barb not described var. ambiversa var. illicita direction var. mutabilis mixed

Reduced not described var. submutica var. mutabilis in length as B . infirma*

*see text. Table 2. Distribution of disc corolla and correlated fruit characters in selected species of BidenB section Bidens.

Species

Character B. eatoni B. connata B. frondosa B. vulgata B. comosa B. tripartita leaves simple or simple or compound compound simple or simple to lobed lobed lobed divided disc lower 1/3-J lower l/3-i lower l/3-i uniformly uniformly uniformly corolla herbaceousi herbaceousi herbaceous! stramineous stramineous stramineous upper dark upper dark upper dark funnelform funnelform funnelform yellow to yellow to yellow to orangei orange i orangei * campanulate campanulate campanulate antheri number* (4)5 (4)5 , (4)5 4(5) 4(5) (4)5 color black (brown) black (brown) black pale-brown pale pale-brown achenei cross- plano-convex plano-convex plano-convex strongly flattened plano-convex section to rhomoboidal to rhomoboidal flattened to flattened Burface smooth to papillose and papillose and smoothsi smooth! smooth or slightly carinatei carinatei glabrous or glabrous or carinatei pustulosei hirsute hirsute sparsely sparsely glabrous to glabrous to pubescent pubescent pubescent sparsely pubescent margins smooth to corneous, corneous or membraneous membraneous membraneous slightly some antrorse cartilaginous, 4 fragilei 4 fragilei to corneous, cartilaginous, hairs usually antrorse hairs hairs all hairs mostly hairs retrorse antrorse or present usually retrorse retrorse retrorsely present setose awns i • number 2 - i* i or absent 2-41 or absent 2(-3)i 2| 2-3t-^)i 2-41 barbing antrorse or antrorse or antrorse or retrorse retrorse retrorse retrorse retrorse retrorse natural- ME No. Am. No. Am. No. Am. No, Am. E. No. Am, Eur. range

♦anther number is Btrongly correlated.with the number of corolla lobes. Flowers with 4 anther and 4-lobed corollas can be found on smaller heads of any species. 27

Figure 2. Disc florets and achenes of Bidens section Bidens.

A. B. heterodoxa var. orthodoxa. (Fernald. Long & St. John

8203» GH; HOLOTYFE; Quebec, Grindstone Island, Edtang du Nord). B. B. heterodoxa var. monardifolia. (Jansson s.. n., 25 Sep 1932, CU; Connecticut* Middlesex Co., lake Pocotopaug). >'■ I U 1 U I|.

83 29

Figure 1. Achenes of Bidens section Bidens* A. B. eatoni.

(Roberts 5^50; Canadas Quebec, St. Lawrence River at

St. Michael). B. B. eatoni (Roberts 5*t-50: Canadas Quebec, St. Lawrence River at St. Michael). C. B. eatoni var. mutabilis. (Fassett 2116, GH; HOLOTYFEs Maines Lincoln Co., Kennebec River at Cedar Grove). D. B. eatoni. (Roberts 5180; Canadas New Brunswick, Miramachi River at Red Bank).

LITERATURE CITED

Boivin B. 1966. Enumeration des plantes du Canada. Nat.

‘ Can. 93* 989-IO63 . . 1967. Connecting vowels in epithets of Latin origin.

Rhodora 6 9* 451-455• Fassett, N. C. 1924. A new variety of Bidens heterodoxa.

Rhodora 26« 177-178. . . 1925a» Bidens eatoni and its varieties. Rhodora 27* 142-146. . 1925"b. A key to the northeastern American species

Of Bidens. Rhodora 27* 184-185* . Notes from the herbarium of the University of

Wisconsin. Rhodora 3 O 1 31-35* . 1957* A Manual of Aquatic Plants, 2nd. ed. Univ. Wisconsin Press, Madison. Fernald, M. L. 1903* A new Bidens from the Merrimac Valley.

Rhodora 5* 90-92. . 1908. Bidens connata and some of its American allies.

Rhodora 1 0 1 197-203. . 1913. Some noteworthy varieties of Bidens. Rhodora

15* 74-78. . 1917. A remarkable colony of Bidens in Connecticut.

Rhodora 19* 257-259* . 1932 . Another localized variety of Bidens heterodoxa.

Rhodora 34* 116-117* 3.1 32 . 1938. New species, varieties and transfers. Rhodora 40: 331-358. . 1950. Gray's Manual of Botany. 8th ed. American Book Co., New York. . & H. St. John. 1915* Some anomalous species and varieties of Bidens in eastern North America, Rhodora 17* 20-25. Greene, E. L. 1901. Identity of B. frondosa. L. Pittonia 4: 246-250. Scoggan, H. J. 1979* The Flora of Canada. Pt. 4 — Dicotyledoneae (Loasaceae to Compositae). Natl. Mus. Nat. Sci. Publ. Bot. No. 7. Sherff, E. E. 1932. Studies in the genus Bidens. X. Family Compositae. Bot. Gaz. 93* 213-220. . 1937. The genus Bidens. Field Mus. Nat. Hist. Publ. 388. Bot. Ser. (Fieldiana) vol. 16, pts. 1 &'r2. . 1955* Bidens. In: E. E. Sherff & E. J. Alexander, (eds.). Compositae-Heliantheae-Coreopsidinae, pp. 70-129. North Am. Flora Ser. II, pt. 2. Stafleu, F. A. et al, eds. 1978. International Code of Botanical Nomenclature ( Lenningrad, 1975)• Regn. Veg. 97* 1-457. Wiegand, K. M. 1899* Some species of Bidens found in the United. States and Canada. Bull. Torr. Bot. Club. 26: 399-422. . & A. J. Eames. 1926. The Flora of the Cayuga Lake Basin, New York. Cornell Univ. Agric. Exp. Sta. Mem. 92, Ithaca. CHAPTER II

FLAVONOIDS OF AMERICAN BIDENS SECTION BIDENS (COMPOSITAE)

33 INTRODUCTION

The cosmopolitan genus Bidens was divided by Sherff

(1937# 1955) into fourteen sections. His sections Platvcarpaea and Heterdonta together form a natural group with a center of diversity and probable origin in North America. The two sections are distinguished from others in the genus primarily on the basis of achene morphology. Section Heterodonta, as defined by Sherff, is composed of two estuarine species occurring in freshwater intertidal zones of the eastern United States and Canada (Table l). Each species has a number of named geographically restricted varieties (Fassett, 1925a, 1925b; Blake, 1929; Sherff, 1937» 1955)* Section Platvcaroaea (sensu Sherff) contains ten primarily North American species and two Eurasian species (Table l). Most species of Platvcaroaea are more widespread in range and often occur with several conspecific morphotypes (recognized as varieties by Sherff, 1937) in polymorphic populations. Revisionary studies (Fassett, 1925a, 1925b; Sherff, 1937. 1955)» regional taxonomic treatments (Fernald, 1950; Cronquist, 1952, 1955; Scoggan, 1979)» and systematic studies (Hall, 196^; Weedon, 1973; Koch, 1975) have contributed to an ■understanding of the geography, morphological variation, and chromosome number variation within the group. However, disparate taxonomic treatments 3 ^ resulting from these investigations suggest that systematic relationships within the group are still poorly resolved, particularily at the varietal level.

The Bidens connata complex is representative of the differences in approach. Sherff (1937t 1955» 1965* Table 1) recognized B. connata as consisting of nine varieties and B. comosa as a monotypic species. Cronquist (1952, 1955) and Scoggan (1979) have suggested that both are variants of the polymorphic Eurasian species, B. tripartita. Within the species of both sections, Sherff (1937» 1955) recognized many morphological and geographic variants which most later authors have ignored or rejected. The varietal recognition of morphological diversity has been most pronounced with B. connata (9 varieties,

Table l) and B. eatoni (8 varieties, Table 1). In both species much of the subspecific taxonomy is based on achene morphology, particularily variation in the awns. Many of these variants, as originally described, were thought to occur as allopatric populations. Subsequent herbarium collections and field studies have indicated that most of the achene variants occur sympatrically in polymorphis populations throughout much of the range of the species. Taxonomic recognition of many of these entities is probably unwarranted. Morphological and cytological evidence suggests that the two sections recognized by Sherff (1937» 1955) do not represent the natural discontinuities within the group as a whole. A more complete discussion of species relationships and reasons for considering the species of both sections PIatvcarpaea and Heterodonta as members of one natural group, Bidens sect. Bidens. will be presented elsewhere. A synopsis of the species groups and their distinguishing features within the more inclusive sect. Bidens is presented in Table 2 to facilitate a discussion of flavonoid distribution. The three species of the B. cernua group have large radiate heads, simple leaves, and diploid chromosome numbers. They form the most distinctive morphological group within the section. Bidens connata of Sherff's section Platvcarnaea is most similar to B. eatoni and B. bidentoides of the Heterodontae. Together with several other species, they form a somewhat diverse but natural

group of mostly tetraploid species (Table 2). The species of the Bidens comosa group can be distinguished by their distinctive disc corollas and fruits. The two American species appear to be more similar to the two Eurasian species than to the sympatric species of the B. connata group. Investigations of flavonoid chemistry have contributed to systematic studies of other genera in the 37 Coreopsidinae (Melchert, 19665 Crawford & Smith, 1980,

Crawford et al, 1980; Smith & Crawford, 1981, references therein; Giannasi, 1975? Roberts, 1980) and in other sections of the genus Bidens (Ballard, 1975; Hart, 1973» 1979). The subtribe Coreopsidinae is characterized by

the presence of anthochlor compounds (Bohm, 1975;

Crawford & Stuessy, 1981) which are otherwise of limited distribution in the family Compositae and the plant kingdom in general. European phytochemists have partially characterized some of the flavonoid constituents of several species of section Bidens (Romussi & Pagini, 1970; Serbin et al, 197^b, 1975; Dakshini, 1975; Borisov et al, 1979). Their results have not been maximally useful from a systematic point of view because of the focus only on major compounds. This investigation of the flavonoids of North American Bidens section Bidens (PIatycarpaea and

Heterodonta of Sherff, 1937) was initiated to 1) compare the chemistry of the two sections (sensu Sherff) with each other and to species groups in other sections; 2) investigate chemical relationships within the section at the specific and varietal levels; 3) determine the extent and nature of geographical variation within taxa of the section. MATERIALS AND METHODS

Details of collections and localities of each taxon are presented in Appendix 1. In a number of cases in which the populations contained more than one morphotype (often recognized as varieties by Sherff, 1937» 1955) individuals were examined separately. A number of the populations were examined with material collected at different seasons or years, with the results indicating only minor quantitative variation. Representative material of at least one population of most species was sorted into components containing leaves, phyllaries, ray florets, disc florets, and pales. Each was chromatographed separately. Dried material was ground in a Wiley Mill and eluted in 8$%> methanol. The extract was concentrated and partitioned with chloroform and ethyl acetate. The chloroform fraction contained no compound not present in higher yields in the other fraction and was discarded. The ethyl acetate fraction was used to prepare standard 2-D chromatograms of each population on Whatmann 3MM paper (Mabry et al, 1970) or thin-layer cellulose plates. Larger quantities of some compounds were prepared by additional purification on sephadex LH-20 columns, polyamide columns, MN-polyamide-2 thin-layer plates, and Avicell microcrystal­ line cellulose thin-layer plates. Flavonoid profile variation anong individuals of the same population was 38 39 examined "by means of thin-layer chromatography. Solvents for paper chromatography were tertiary butanoli glacial acetic acidi distilled water (3*1*1} TBA), 15# glacial acetic acid (HOAc), or distilled water (Mabry et al, 1970). A variety of solvent systems from various sources (Anderson et al, 19701 Harborne, 1973; Julian & Crawford, 1972) were used on cellulose plates for purifica­ tion and co-chromatography. Polyamide thin-layer solvents were those of Wilkins and Bohm (1976). Particular use was made of BMAW (butanone* methanol* acetic acid* water?

5 0*1 0 *1 0 *2 .5) for separating methylated compounds. Flavonoids were detected on plates in UV light or after spraying with 0.1%.B-aminoethyl diphenyl borinate (Wilkins and Bohm, 1976). Compounds were identified by chromatographic properties, co-chromatography, color reactions, UV-spectroscopy, enzyme hydrolysis, and partial acid and alkaline hydrolysis. Standard procedures of UV-spectroscopy using diagnostic reagants were followed (Mabry et al, 1970). Glycosides were hydrolyzed with 5*Q?° hydrochloric acid or dilute triflouroacetic acid in 50% isopropanol (Wilkins and Bohm,

1976). The hydrolyzate was dried, taken up in water, extracted with ethyl acetate, and concentrated. Sugars in the water layer were identified by circular co-chromatography on precoated Polygram Cell A00 plates in ethyl acetate* pyridine* distilled water (6*3*2; Becker et al, 1977)* Sugars were visualized by spraying with p-anisidine-pthallic acid in ethanol and heating at 100°C.. Some glycosides were routinely hydrolyzed with J3-glucosidase (Sigma Chemical Co.)* Mild alkaline hydrolysis of acylated pigments was

i carried out under nitrogen, in the dark at room temperature, with either 10# potassium hydroxide or 2N sodium hydroxide for 0.5-2.0 hr (Anderson et al, 1970). Samples were removed at intervals, neutralized, extracted with ethyl acetate, and dried under nitrogen. Both fractions were co-chromatographed with flavonoid, sugar, and phenolic

acid standards in benzenei acetic acid (4j1) on polyamide and with a number of TLC solvents previously mentioned. The identity of the acids in the acylated pigments was also investigated spectroscopically after further purification and by recording difference spectra using purified anthochlor glucosides or commercial phenolic acid standards in the reference cell (Anderson et al, 1970). RESULTSi COMPOUND IDENTITIES

The tentative identities of flavonoids isolated from the members of the section are presented in table 3» A total of 28 compounds were encountered and isolated» including naturally occurring aglycones and both chalcone and aurone forms of anthochlors. Spectral data and chromatographic characteristics are presented in table 5 and figures 1-14 for compounds which are new or have been incompletely characterized in published literature. The flavone diglycosides (compounds 2 & 4) could often not be detected in mature plants from which they were found at an earlier growth stage. The compound tentatively identified as luteolin-4',7-0-diglucoside (compound 4) yields luteolin on acid hydrolysis or with B-glucosidase. Its spectral properties (table 5) and color reactions oji paper indicate it is substituted in both the A- and B-rings. Rf values of 0.16 in TBA and 0.42 in HOAc indfcate it is a diglycoside. Partial acid hydrolysis yields apigenin, apigenin-7-0-glucoside (compound 3) and luteolin aglycone, but not the expected luteolin-4'-glucoside. The compound identified as apigenin-7-0-diglucoside (compound 2 has spectral properties identical to compound lp but chrom­ atographic behaviour indicative of a diglycoside. Partial hydrolysis yields apigenin, apigenin-7-0-glucoside (compound 1), and glucose. 41 42 A single flavanone compound (number 10) and its corresponding aglycone were detected and purified..

This compound was detectable only in younger leaf tissues before anthesis, and then only in low concentrations. The compound was bright purple under UV light, turning to blue-green upon fuming with ammonia. Enzyme hydrolysis and Rf values indicate the compound is a monoglucoside. The lack of a NaOAc Band II shift (table 5) in the glycoside and corresponding large shift in the aglycone suggest the sugar was substituted at the 7-hydroxyl of the A-ring. The lack of a shift with HCL with respect to AlCl^ suggests an absence of A-ring dihydroxyls. Other spectral shifts do not distinguish between a 5-hydroxy or 5-deoxy A-ring structure. Compound 10 may be the same as the butin-7-0-glucoside reported from B. tripartita

(Serbin et al, I97^b). This flavanone is homologous in substitution pattern with the chalcone butein and might be expected in plants producing this glycoside.

Isokanin, the flavanone homologue of marein, has also been reported from B. tripartita (Serbin et al, 19751 but was not encountered in this survey. A number of unusual chalcones and aurones have been isolated from species of section Bidens. Unfortunately, the more unusual are produced in lower quantity relative to the ubiquitous marein-maritimein and coreopsin- sulphurein anthochlor pairs. A new chalcone, 2 ',3-dihdroxyl-4-methoxy-4'-glycosyl chalcone (compound

B), occurring in all members of section Bidens. was recently described from the related genus Megalodonta

(Roberts, 1980). Another new chalcone glucoside (compound

F) with a similar 4-methyl substitution on an okanin type of A-ring has been discovered for the first time as a natural product. Compound F has chromatographic properties similar to those of marein on paper (table 5) * "but does not fume to bright red with ammonia as do chalcones with a free 4-hydroxyl (Giannasi, 1975)• Chromatography on polyamide in non-polar systems suggested the presence of a methyl group (Rf in BMAW = 0.65, for marein = 0.31)*

The major UV absorption peak in MeOH of 373 11111 is characteristic of chalcones with two oxygen functions in the B-ring (Giannasi, 1975* Ballard, 1975)• The small bathychromic shift and collapse with NaOMe is characteristic of chalcones with both the 4- and 4'-hydroxyls substituted (Giannasi, 1975)• The bathychromic shift of 45 nm with AlCl^ and lack of a major

shift with addition of HD1 indicates a free 2 '-hydroxyl and lack of ortho-dihydroxyl groups on the B-ring. The absence of shifts with NaOAc and H^BO^ confirm the lack of free dihydroxyls or a free 4'-hydroxyl. The sugar identification is based on the rapid hydrolysis by B^glucosidase and co-chromatography after acid hydrolysis. The aglycone has Rf values similar to okanin on paper. The lack of conversion of the aglycone to an aurone

(Ballard, 1975) and i^s lack of change with fuming (Giannasi, 1975) is further evidence that the 4-position is substituted. Spectral data of the aglycone showed a methanol spectrum similar to the glycoside. Mass spectral data are needed for confirmation, but the new compound

is tentatively identified as 2 ',3 ', 3-'trihydroxy-4-methoxy- 4'-glucosyl chalcone, or 4-methoxymarein. The aglycone

and a 3 '-glycoside have been reported from species in the B. nilosa complex (Gallard, 1975) and compare to the characteristics of the compound in Bidens. An additional okanin glycoside (compound E) was discovered which has properties different from known compounds. The compound has the chromatographic properties of a diglycoside (Rf in TBA = 0.12, HOAc = 0.18) and does not change color on fuming with ammonia. Enzymatic hydrolysis of the compound yielded okanin aglycone. The methanol UV absorption Band I at 357 nm and decomposition with NaOMe (table 5) is characteristic of chalcones glycosylated at the 4-hydroxyl (Giannasi, 1975) • as is the shift of 43 nm with NaOMe. The AlCl^ shift and lack of a shift with A1C1^/HC1 indicate a free 2-hydroxyl and absence of a B-ring dihydroxyl. On the basis of this evidence, the compound is identified as 2 ',3 '»3 -’fc**ihydroxy

2 ',3 ',3 -trihydroxy-4',4-diglucosyl chalcone, or okanin-4',4-diglucoside. Compound E may be present in all taxa of the section, but appears in detectable quantities only in young vigorous tissues or flowers at anthesis, and not in mature leaves or flowers.

Several compounds with chromatographic properties similar to the aglycones okanin and maritimetin, but having distinctive spectral properties were present in large quantities in many species. Purification proved difficult due to conversion of chalcones to aurones and probable isomerization. Both the chalcone and aurone forms have major absorption peaks (312 -31 ^ nm) indicative of acylation. The acylated chalcone (compound Dc) had spectral absorption roughly equivalent to a mixture of marein and p-coumaric acid, with major peaks at 3^2 nm and 311 nm (table 5)• Additive spectral absorbence is typical of flavonoids which are acylated with hydroxycinnamic acids (Anderson et al,, 1970). Difference spectra using MeOH and marein or p-coumaric acid as standards in the reference beam also indicate the presence of both an okanin type compound and a p-coumaric acid. Mild alkaline hydrolysis of the parent compound yields marein, p-coumaric acid, and several other unidentified compounds. The unknown compounds are presumed to be degradation products of the chalcone maritimetin. The marein and p-coumaric acid products were verified by spectral analysis and co-chromatography in TBA and H2O on cellulose. Acid hydrolysis of the compound yielded okanin, meritimetin, two isomers of p-coumaric acid, and glucose. Attempted hydrolysis of the parent compound with ^-glucosidase was unsuccessful, suggesting the acyl group is esterified to the V-sugar. The acylated aurone (compound Da) occurred naturally and formed spontaneously in methanolic solution from the chalcone. Its spectral properties and chromatographic behavior indicates it is homologous with the chalcone. the two compounds are tentatively identified as okanin- okanin-4' - 0-JB-D- (p- coumarylglucoside) and meritimetin-6-0-J3-D-(p-coumarylglucoside). It is possible that the chromatographic heterogeneity of the compound is due to l) differences in the position of substitution of the p-coumaric acid on the glucose,

2 ) isomerization of the acyl group, or 3 ) the presence of small amounts of an acylated 4-methoxymarein (the acylated form of compound F). Compound G is a dull yellow substance which turns to bright orange-pink with ammonia fuming. Spectral

data show major absorption peaks at 38 ^, 33 ^, 300 , and 2^5 nm (table 5)» similar to many xanthones (Harborne, 1967). Mild acid and alkaline hydrolysis results

in no spectral or chromatographic change in the compound, suggesting the presence of C-glycosyl linkages common to many xanthones. Although this is not a flavonoid compound, it is included here because of its systematic significance with relation to the section. RESULTS* COMPOUND DISTRIBUTION The occurrence of flavonoid compounds in the species of Bidens section Bidens investigated are presented in table 4. Each profile represents an aggregate for the taxon. Compounds found only in floral tissues are also distinguished. Except for cases to be discussed, differences within taxa due to compounds occurring near the limits of detectability are attributed to developmental or environmental differences rather than genetic differences. Glycosides of apigenin, luteolin, and 6-hydroxyluteolin have been reported from several genera of the Coreopsidinae (Giannasi, 1975* crawford, 1970; Smith and Crawford, 1981), and all but 6-hydroxyluteolin have been reported from Bidens (Hart. 1979; Ballard, 1975)- Within section Bidens, 6-hydroxyluteolin (compound 5) and diglycosides of apigenin and luteolin (compounds 2 & 4) are not found in the B. cernua group (table 4). The glycosides of quercetin encountered in section Bidens are widely distributed in the Coreopsidinae and in other sections of Bidens (Hart, 1979; Ballard, 1975;

Roberts, unpub.). Rutin (compound 8) is found only in B. comosa. but has also been found in the related genus Megalodonta (Roberts, 1980). The 6-hydroxyquercetin glucoside (compound 9) has not been reported from Bidens. but is found in the related genus Coreopsis (Crawford, 1 9 8 0;

48 k 9 Smith and Crawford, 1981). The single flavanone compound

(no. 10) is usually found in all species in low concentration.

Both butein and okanin based chalcones and aurones are widespread in the Coreopsidinae (Bohm, 1975; Crawford and Stuessy, 1981) and in the genus Bidens (Hart, 1973, 1979;

Ballard, I98I). Although derivatives based on both occur in all the species of section Bidens, the substitution patterns and organ expression differ between the two major species groups. The three species of the B. cernua group

(table 2 ) do not produce either okanin-meritimetin aglycones or their substituted derivatives in vegetative tissues (table ^). In the other species a complete flavonoid profile' is found in leaves and stems. In B. frondose okanin aglycone and a number of derivatives can be found in young plants by the time primary leaves are expanding. Butein-coreopsin aglycones and glycosides can be found in all organs of all species, indicating the regulatory differences are specific to the okanin products and not anthochlors in general. Within flowering heads the distri­ bution of anthochlor compounds varies only quantitatively among phyllaries, rays, pales, and disc florets. The distribution of the acylated okanin derivatives (compound D) coincides with the differences in organ distribution. The species of the Bidens cernua group do not produce detectable amounts of the acylated glycosides even in the flowers. Some populations of B. connata and B. bidentoides also do not produce detectable quantities of the compound, while other populations produce large amounts. In both cases the differences have been confirmed by repeated sampling. In the morphologically variable and taxonomically complex B. connata group the distribution does not correlate with geography or with varietal differences as delimited by Sherff (1937» 1955)* Individuals representing extreme morphotypes with respect to leaf shape and achene features were selected from several polymorphic populations ob B. connata (R5487. R5506, R5515), each including var. petiolata. var. anomala, and var. ambiversa. The flavonoid profiles of each of the variants in the populations are identical, although they would be recognized taxonomically as three varieties (Sherff, 1937» 1955)• Bidens eatoni exhibits geographic variation in the production of the acylated anthochlors. This chemical variation does not coincide with the taxonomic varieties described by Sherff (1937)* Morphotypes corresponding to B. eatoni var. eatoni differ chemically between northern and southern populations. The Merrimack estuary, Massachussetts, contains three named varieties (B. eatoni var. eatoni. var. fallax. var. illicita) but a collection containing morphotypes corresponding to the three taxa (R5221) contained no detectable variation among the individuals surveyed. ' In Bidens bidentoides the populations of the typical variety in the Hudson, South, Delaware, and Maurice Rivers do not have the acylated okanin derivatives as do populations from several sites in northern Chesapeake Bay. The Chesapeake populations have been taxonomically recognized as var. mariana (Blake) Sherff. They differ from the other populations qualitatively in their pilose disc corolla tubes and quantitatively in several morphological features (Sherff, 1937)- Within the species the difference in flavonoid chemistry corresponds with both geographic and morphological diversity. The okanin diglucoside (compound E) and Jf-methoxy derivatives occur as minor constituents of all the taxa with the okanin derivatives occurring only in capitula of the B. cernua species group. Chalcone V,4-diglucosides have not been reported from Bidens, but several derivatives of 4-methoxy okanin are found in the B. pilosa species group (Ballard, 1975)• The ^-methoxy derivative of butein has not been reported from other sections of Bidens. but

is found in Megalodonta (Roberts, 1980) and Dahlia (Giannasi, 1975)* The unknown xanthone compound (G) is apparently

confined to Bidens section Bidens. Xanthones have not been reported in other phytochemical studies of Bidens and preliminary chromatographic surveys indicate they are absent from Bidens sect. Medusae and several species of of section Psilocarpaea. They are also absent from the genus Megalodonta (Roberts, 1980) and all species of C

Coreopsis which have been studied (Crawford, 1970;

Crawford & Smith, 1980). The only reported occurrence of xanthones from the Coreopsidinae is in one species of the genus Dahlia (Giannasi, 1975)• The three compounds reported from Dahlia appear to be different from compound G on the basis of chromatographic and spectral properties, but are also C-glycosides. DISCUSSION AND CONCLUSIONS

Flavonoid chemistry clearly indicates a close biosynthetic relationship among the species of Bidens section Bidens which had previously been segregated in sections Platvcaroaea and Heterodonta. The species share most of the same flavonoid aglycones and differ primarily in the degree of substitution at several positions. The only exception to this generalization is with respect to the lack of 6-hydroxylation in species of the B. cernua group. This difference is paralleled by differences in organ specificity for the production of okanin derivatives. The unity of the section is also supported by the occurrence of the xanthone compound which- has not been found in other sections of the genus. Further discussion of the utility of flavonoids in sectional definition must await phytochemical surveys of additional species in the primarily tropical sections of the genus and of the other North American section, the Medusae.

The flavonoid chemistry within Bidens section Bidens is remarkably uniform within the species groups as delimited in table 2. Indeed, the magnitude of differences at the species level are often equaled by those at the varietal and populational level. At the level of species-groups a natural pattern emerges. The B .cernua-B. laevis-B. 53 hvoerborea species complex produce a rather simplified profile lacking acylated-marein, 6-hydroxyluteolin, and

6-hydroxyquercetin. They also differ from all other species in lacking the okanin derivatives in vegetative tissues. The three species are morphologically similar in having simple sessile leaves, relatively large radiate heads, and very similar achenes (Sherff, 1937). Bidens laevis is the only perennial in the genus in temperate North America and the only species of sect. Bidens native to South America. Bidens hvoerborea is presumably a derivative of one of the other two which occurs only in freshwater intertidal zones north of the glacial boundary. Within this limited area, the various populations have been divided into four varieties by Sherff (1955)• The collections inluded in the flavonoid survey include all but var. svensoni Fassett, which is apparently extinct at the type locality due to construction of a dam. The variation ini B. hyperbore a recognized on the basis of morphology is not reflected in the chemical variation, and flavonoid chemistry does not indicate which other species might be a progenitor. Taxonomic problems in the Bidens connata species complex cannot be resolved biochemically due to the lack of consistent variation among the taxa. In the case of possible diploid-tetraploid species pairs such as B. discoidea (2h) and B. frondosa (4n) there are no chemical differences between the two species and they are indistinguishable from several others. The lack of chemical differences between £. discoidea-B. frondosa- B. vulgata does not support the placement of the latter species with B. comosa as proposed on the basis of achene and fruit morphology (table 2). Throughout the B. connata group the chemical differences between species are equaled or exceeded by differences at the varietal and populational level.

As compared to other species groups in the genus Bidens. section Bidens does not show equivalent differences in flavonoid compliments between taxa. Speciation within the group has not been accompanied by substantial changes in flavonoid composition. Section Bidens does not exhibit the diverse array of compounds found in the Bidens

(Psilocarnaea) ferulaefolia complex of three species (Hart, 1979)• Most of the flavonoid diversity in this group is based on the same okanin and butein aglycones, but with a much larger array of sugar substitutions, including mono- and di-glucosides and rutinosides. Differences among taxa are primarily due to differences in glycosylation of anthochlors and other flavonols. In the B. (Psilocaroaea) pilosa complex (Ballard, 1975) a diverse array of flavonoids is produced which distinguishes each of the six diploid

taxa. In addition to okanin and butein glycosides, the anthochlors are represented by numerous derivatives based on position of methylation and glycosylation of the basic okanin ring. Flavonoid variation within species is extensive and equals or exceeds that between some species pairs. In conclusion, flavonoid chemistry indicates a close relationship among the specis of Bidens section Bidens. Morphological differences between species exceed chemical / differences. Flavonoid profiles are useful in delimiting species groups and occasionally at lower taxonomic levels, but will not discriminate between most taxa within species groups or between the varieties of most species. Table 1. Taxonomy of Bidens section Bidens according to

Sherff (1937, 1955* 19&5)* with explanatory comments. Sect. Heterodonta B. bidentoides (Nutt.) Britton (2 varieties) B. eatoni Fern. (8 varieties)

Sect. Platvcarpaea B. discoidea (T. & G.) Britton B. frondosa L. (4 varieties) B. vulgata Greene (3 varieties) B. comosa (A. Gray) Wiegand B. connata Willd. (9 varieties) B. heterodoxa Fern. & St. John (5 varieties; elements of this species belong with B. connata. B. comosa. and B* eatoni. see text). B. tripartita L. (5 varieties; Eurasian) B. radiata Thuill. (Eurasian) B. amplissima Greene (known only from coastal British Columbia) B. cernua L. (2 varieties; also in Europe) B. laevis (L.) B. S. P. B. hvnerborea Greene (*J- varieties) 58

Table 2. Morphological characteristics of specie's groups in Bidens section Bidens. Cytological data are compiled from published sources and my own determinations.

1 cernua groupi (B. cernua. n=12j B. laevis. n=l2 ; B. hvoerborea. n=12) leaves simple, sessile} heads radiate} disc florets with campanul&te throat} anthers black} achenes quadrangulate, 2-4 awned, retrorsely barbed awns} with cartilaginous summit and ribs, surface smooth to pustulose} B. connata group* (B. connata. n=24, 3 6 } B. eatoni. n=24; B. frondpsa, n=24* B. discoidea. n=12} B. bidentoides. n=25) leaves simple, cleft, or compound} heads discoid or with short raysi disc florets with campanulate throat} anthers usually black} achenes plano-convex to quadrangulate} 2-4 awned} margins thin to cartilaginous* surface smooth to pustulose, awns retrorsely or antrorsely barbed* B. comosa groupi (B. comosa. n=24* B. vulgata. n**12) leaves simple, lobed, or compound* heads discoid or with rudimentary rays* disc florets funnelifofm1 achenes flattened to plano-convex, 2-4 awned* margin thin to winged, surface smooth to sparsely pubescent, retrorsely barbed awns* 59

Table 3* Tentative identities of flavonoid glycosides detected in Bidens sect. Bidens.

Compound Class and designation Identity

Flavones 1 apigenin-7-0-glucoside 2 apigenin-7-0-diglucoside 3 luteolin-7-0-glueoside * luteolin-V,7-diglucoside 5 6-hydroxyluteolin-7-0-glucoside Flavonols 6 quercetin-3-0-glucoside 7 quercetin-3-0-rhamnoside 8 quercetin-3-0-rutinoside 9 6-hydroxyquercetin-7-0-glucoside Flavonone 10 butin-7-0-glucoside Anthochlors A coreopsin-sulphurein chalcone-aurone pair B k-methoxycoreopsin C marein-maritimein chalcone-aurone pair D acylated marein-maritimein chalcone— aurone pair E okanin-V,^-diglucoside F 4-me thoxymare in Xanthone G C-glycosyl xanthone Table k. Distribution of flavonoids in Bidens section Bidens. Compound designations are as in table 2. Parentheses indicate the compound was not detected in some populations. An x designates the compound was detected in floral tissues but not in leaves. ? « not reported.

Compound Taxon 1 2 3^'56 7 89 10 ABCDEFG

B. laevis + + + + + + + X (X) X + B. cernua (+) + + + (+) + + X (X) X B. h yperborea (+) + + + ( + ) + + X X + B. frondosa + + + (+) + + + + + + + + +, < + + + B. discoidea + + + (+) (+) + + + ( + ) + + + + + + + B. connata + (+) + (+) (+) + + + ( + ) + + + (+) (+) + + B. connata var. pinnata + + + + + + + ( + ) + + + + + + B. eatoni (+) + + + + + + + + + + + (+) (+) + + B. bidentoides + + + + + 4* (+) + + + + + + B. bidentoides var..mariana + + + + + + (+) + + + + (+) + + B. v u lgata + + + (+) + + + + + + + + + (+) + + B. comosa + + + + + + + + + (+) + + + + (+) + + B. tripartita* •> •> + ? + + ? + + + + ?? ??

•Compiled from data published by Serbin et al (Serbin, et al, 197^a, 197^b, 1975) and Baranska (1963)* Table 5. Color reactions, Rf values, and UV spectral data for selected flavonoids in

Bidens section Bidens. Compound designations as in table 3* c = chalcone, a = aurone,

agl = aglycone, dec = decomposes, b = bright, d = dull, bl = blue, br = brown, g =

green, or = orange, p = purple, pk = pink, y = yellow, sh = shoulder or inflection.

Color Rf UV Spectral Peaks Compound designation UV +NH3 TBA HOAc MeOH NaOMe A1C13 HC1 NaOAc HBO3

Ac pg r 46 28 382 447 495 432 450 500 290 sh 280 360 . 388sh 405 408 262 304 285 2 77 243 sh 2693J6

Aa y or 43 26 401 485 446 399 450 42 5 335 345 356 3o5sh 387 330 270 286 292 322 283 280 253 235 269 255 252sh

B pg Pg 54 19 375 408 435 426 375 375 312 322 318 385sh 292 256 254 276 263 256 222 dec £

Cc g r 24 14 380 443 504 418 390 390 325 sh 340 366 330 sh 325 324 263 282sh 334 267 273 272 256 282sh 245sh ON M* 273 Table 5 (continued).

Ca y or 26 16 414 490 452 412 452 436 330 340 323 323 335 325 272 285sh 286 270 280sh 279 243sh 258 255

Dc g r 42 10 381 435 504 418 378 383 311 36 I 312 312 308 3I6 280 sh 282sh 309 272 268 267 268sh 260sh 246sh

Da y or 4-2 12 410 487 453 413 44 5 435 313 355 350 sh 312 316 312 285sh 292sh 308 297sh 285sh 285sh

E p p 12 18 357 400 411 400 351 348 281 340 325 325 276sh 275sh 273sh 276 269 265 274 274 dec

br 27 14 372 * 405 420 408 36 O 430sh 320 sh 346 363 320 255sh 356 264 272 320 268 257sh dec 268sh 242 245sh

dyg r 63 32 384 500 sh 492 427 407 412 334 402 360 327 350 345 300 278 303 300 sh 240 240 245 257 280sh 245sh ON to Table 5 (continued).

P P 37 30 333 379 380 375 350 sh 3^7 338 270 300 300 275 275

p p 16 kz 333 357 357 355 332 337 280 sh 268 310 310 268 272 270 dec 272 268

10 p bg 6k 6k 327 kz5 37^ 372 329 321 283 357 31^ 331 275 274 287 dec

10 agl P P 89 k3 325 321 367 367 321 325sh 287 zkz 306 306 296sh 285 226sh 219 288 Figure 1. UV absorption spectra of compound 4-, luteolin-4',7-0-diglucoside. n ji O V 500

+BO3 300 NaOAc i i 500 400 400 300 A1C1 &+HC1 500 300 NaOMe HeOH

Figure j-* Figure 2. UV absorption spctra of compound 10, butin-7-0-gluco side. ON 500 400 300 500 400 300 +HC1 AXC1 400 300 500 Mei

Figure PO 68

Figure 3. UV absorption spectra of compound 10 aglycone, butin. H*•*1 s

•»

•A1C1 +BO. 300 500 300 500

CN vo Figure 4-. UV absorption spectra of compound B, ^-methoxycoreopsin. H

Ms OH Ms NaOMe

Figure ^ • Figure 5. UV absorption spectra of compound Do, acylated marein. V j J 500400 300 500 300 +HC1 MCI 300

Figure 5 Figure 6. UV absorption spectra of compound Da, acylated maritimein. NJ 500 300500 300 +HC1 A1C1 400 300

Figure 6 Figure 7. UV absorption spectra of compound okanin-V,4-diglucoside. ^ 3 -v] 500 300 500HOO i 300 + H C 1 1 A1C1 ■ x 300 HOO 500

___ I NaOM rfeOH

Figure' 7 Figure 8. UV absorption spectra of compound F, 4--me thoxymar e in. VO - > 3 400 300 500 400 300 AlCl +HC1 500 400 300 MeOH NaOMe

Figure 8 Figure 9. UV absorption spectra of compound G. C-glycosyl xanthone. oo H 500 400 3 bo + NaOAe 300 500 300400 400 +HC1 AXC1 500 300 NaOMe MeOH

Figure 9 LITERATURE CITED

Anderson, D. W. , E. A. Julian, R. E. Kepner, and A. D. Webb.

1970. Chromatographic investigation of anthocyanin

pigments in Vitis cinerea. Phytochemistry 9* 1569-1578. Ballard, R. E. 1975* A biosystematic and chemosystematic study of the Bidens pilosa complex in North and Central America. Ph. D. Dissertation, Univ. of Iowa, Iowa City. Becker, H. ,. J. Exner, and J. E. Averett. 1977* Circular chromatography, a convenient method for phytochemical

analyses. Phytochem. Bull. 10: 36 -^1 .

Blake, S. F. 1 9 2 9 . A new estuarine Bidens from Chesapeake

Bay. Rhodora 31*87-90. Bohm, B. A. 1975* Chalcones, aurones, and dihydrochalcones. Pp. 4^2-50^ in The Flavonoids, ed. J. B. Harborne,

T. J. Mabry, and H. Mabry. New York: Academic Press.

Borisov, M. I., T. I. Isakova, and A. G. Serbin. 1979-

Flavonoids of Bidens cernua. Chem. Nat. Compounds 15*

197-198. Crawford, D. J. 1970. Systematic studies of Mexican Coreopsis (sect. Anathysana), with special reference to the relationship between £. cyclocarpa and C.. pinnati-

secta. Bull. Torr. Bot. Club. 97*161-167. , and E. B. Smith, 1980. Flavonoid chemistry of Coreopsis grandiflora (Compositae). Brittonia 32:

15^-159. 83 , _____ , and A. M. Mueller. 1980; Leaf flavonoid

chemistry of Coreopsis (Compositae) section Palmatae. Brittonia 32: 452-463. , and T. F. Stuessy. 1981. The taxonomic significance of anthochlors in the subtribe Coreopsidinae

(Compositae, Heliantheaea). Amer. J. Bot. 6 8: 107-117- Cronquist, A. 1952. Compositae, Pp. 323-545 in The New Britton & Brown Illustrated Flora of the Northeastern

United States and Ad.iacent Canada, vol. 3 ., ed. H. A. Gleason. Lancaster, Pa. : Lancaster Press.

, 1955* Vascular Plants of the Pacific Northwest. Part Compositae, ed. C. L. Hitchcock, A. Cronquist, M. Ownbey & J. W. Thompson. Seattle: Univ. Washington

Press. Dakshini, K. M. M., and S. K. Aggarwal. 1975* Phytochemical profile patterns in the identification of Bidens-species.

Indian J. Exp. Biol 13: 317-319- Fassett, N. C. 1925a. Bidens eatoni and its varieties.

Rhodora 27: 142-146. , 1925b. Bidens hyperborea and its varieties. Rhodora 27: 166-171. Fernald, M. L. 1950. Gray's Manual of Botany. 8th ed. New York: American Book Co. Giannasi, D. E. 1975- The flavonoid systematics of the genus Dahlia. Mem. New York Bot. Gard. 26s 1-125- 84 Hall, G. W. 1967* A biosystematic study of the North American complex of the genus Bidens (Compositae). Ph. D. Dissertation, Indiana Univ., Bl&omington. Harborne, J. B. 1967- Comparative biochemistry of the flavonoids. New York: Academic Press.

, 1973* Phytochemical Methods. London: Chapman and Hall. Hart, C. R. 1973* Systematics of the Bidens ferulaefolia complex (Compositae). Syst. Bot. 4: 130-147. Julian, E. A., and D. J. Crawford. 1972. Sulphuretin glycosides of Coreopsis mutica. Phytochemistry

1 1 : 1841-1843. Koch, R. 1975* A taxonomic study of the genus Bidens L. (Compositae) in the western Great Lakes states. Ph. D. Dissertation, Univ. of Nebraska, Lincoln. Mabry, T. J., K. R. Markham, and M. B. Thomas. 1970.

The Systematic Identification of Flavonoids.

New York: Springer-Verlag.

Melchert, T. E. 1966. Chemo-demes of diploid and tetraploid Thelesperma simplicifolium (Heliantheae, Coreopsidinae) Amer. J. Bot. 53* 1015-1020.

Roberts, M. L. I98O. The flavonoids of Megalodonta beckii (Compositae) and isolation of 2*,3-dihydroxy-4— methoxy-4'-glucosyl chalcone. Biochem. Syst.

Ecol. 8 : 115-118. 65 Romussi, G. , and F. Pagini. 1970- Constituents of Bidens

frondosa. Boll. Chim. Farm. 1091 467-475* Scoggan, H. J. 1979* The flora of Canada. part 4- Dicotyledoneae (Loasaceae to Compositae). Natl. Mus. Nat. Sci. Publ. Bot., No. 7* Serhin, A. G. , M. I. Borisov, and V. T. Chernobai. 1974a. Flavonoids of Bidens tripartita. I. Chem. Nat.

Compounds 8 : 119-120. , _____, . 1974b. Flavonoids of Bidens

tripartita. II. Chem. Nat. Compounds 8 : 439-441. , _____, _____ , I.P. Kovalev, and B. G. Gordienko.

1975* Flavonoids of Bidens tripartita. III. Chem. Nat. Compounds 9* 160-162. Sherff, E. E. 1937* The genus Bidens. Field Mus. Nat. Hist.

Publ. 388. Bot. Ser. (Fieldiana), vol. 16, pts. 1 & 2. , 1955- Bidens. Pp. 70-129 in E. E. Sherff and E. J. Alexander, North American Flora, Ser II, Pt 2.

, 1965* Notes on varieties of Bidens connata and a

hybrid with B. cernua. Rhodora 67 s 59-62. Smith, E. B. , and D. J. Crawford. 1981 Comparative leaf flavonoid chemistry of Coreopsis nuecensoides and C. nuecensis (Compositae), a progenitor-derivative species-pair. Brittonia 108: 7-12. Weedon, R. 1973* Taxonomy and distribution of the genus Bidens (Compositae) in the north-central plains states. Ph.D. Dissertation, Univ. of Kansas, Lawrence. Wilkins, C. K. , and B. A. Bohm. 1976. Chemotaxonomic

~ studies in the Saxifragaceae s. 1. 4. The flavonoids of Heuchera micrantha var. diversifolia. Can. J. Bot. 54: 2133-2140. CHAPTER III

ALLOZYME VARIATION IN BIDENS DISC03DEA (COMPOSITAE)

87 INTRODUCTION

Bidens discoidea (T. & G.) is an annual herbaceous plant which typically occurs in swamp forests of Eastern North America. The geographic range of the species extends eastward from the Great Plains and northward to southern Canada, but it is generally absent from the unglaciated

Applachian Plateau. Sherff (1937) considers it a member of section Platycarnaea DC., a group of predominately North American diploid and polyploid species. Bidens discoidea is unusual among the North American taxa of the section in that intraspecific morphological variation has not been taxonomically recognized, or at least formally described (an average of four varieties per species is recognized for eastern members of the section in Sherff*s 1955 monograph and subsequent publications). Much of the taxonomic proliferation in the group may be related to the extreme phenotypic plasticity characteristic of amphibious plants

(Sculthorpe, 1967). Comparatively, B. discoidea seems morphologically homogeneous. This study utilizes allozymes to investigate patterns of variation within and between populations of this widespread, morphologically uniform, diploid species of Bidens. The patterns of genetic variation are related to observations on morphological characteristics and reproductive biology of the species. It will later serve as a basis of comparison for variation in geographically 89 restricted species of the group, and for comparison with the tetraploid members of sections Platycarpaea and Heterodonta. Morphological features of Bidens discoidea suggest that it is most closely related to B. vulgata Greene and B. frondosa L. These two species also occur in Eastern North America but their geographic ranges extend into the Great Plains and they often occur in more xeric habitats. Vegetatively the three species can be difficult to distinguish, but characters of the capitulum, flowers, and mature achenes provide several distinguishing features (Sherff, 1937)•. Bidens discoidea is a diploid (Table 1) which differs from the more robust tetraploid B. frondosa in having fewer phyllaries and flowers per head, smaller achenes with shorter awns, sparse pubescence, and leaves with fewer, more acuminate leaflets. Bidens discoidea and B. frondosa differ from the diploid B. vulgata in the texture of the flowers, characters of the capitulum, shape of the achene, and nature of the achene surface. The tetraploid B. frondosa is regularly found with one or the other of the two diploid species, but the two diploids are rarely found together. 90 MATERIAIS AND METHODS

Specimens or seed from individual plants were collected over a three year period from eight populations (Table 2) occurring in a single woodland pond or the shoreline of an individual lake. The small number of individuals from several localities reflects the inavailability of mature viable fruit or collecting time rather than relative size‘of the population. Fruits from individual plants were germinated on filter paper in petri dishes under Zk hr light. The embryos were excised from the fruit wall and testa to induce germination. Young seedlings (ca. O.Olg) were ground in chilled trays in approximately 0.2 .-ml of a buffer composed of 0.1'M tris-HCl, 1.0 ' inM EDTA, 10 mM KC1, 10 mM MgC^i 1° roM mercaptoethanol, k0% sucrose, pH 7 .5 (Gottlieb, 1981b). The extract was centrifuged and the supernatant pipetted into the wells of vertical acrylamide gels or taken up on wicks and inserted into slots of horizontal starch gels. Starch gels were composed of 12.5$ Sigma starch using two different electrode-gel buffer systems. System 1 used a gel buffer of 9 parts tris-citrate (pH 8.3; 0.2 M) and one part lithium-borate (pH 8.3; 0.2 M) with lithium-borate alone employed as the electrode buffer (Gottieb, 1973)• System 2 employed a 0.^ M sodium citrate electrode buffer adjusted to pH 7.0 with HC1 and a 0.02 M histidine-HCl gel buffer adjusted to pH 7.0 with NaOH, and was run at 100 ma. 91 for approximately 8 hours (Gottlieb, 1981b). Acrylamide gels

were composed of a 5-5$ separating gel with a tris-HCl (pH 8.9; 0.4 M) buffer, a 2.0$ spacer gel with a tris-HCl (pH 6.9; 0.06 M), and an electrode buffer of tris-glycine (pH 8.3; 0.05 M). Enzyme staining solutions were modified only slightly from those of Gottlieb (1973)* Satisfactory resolution and activity were found for the following enzymes: alcohol dehydrogenase (Adh), phosphoglucomutase (Pgm), phosphoglucose isomerase (Pgi), leucine aminopeptidase (Lap), glutamate dehydrogenase (Gdh), glutamate-oxaloacetate transaminase (Got), 6-phosphogluconate dehydrogenase (6-Pgd), esterase (Est), and acid phosphatase (. Acph) Est, Lap, and Adh were run on starch systems 1, 6-Pgd on starch system 2, Got on acrylamide and system. 1 and the other enzymes on system 2 with occassional comparisons on system 1. Extracts of plants from various populations were run side by side in various combinations. For all enzyme systems a control was run without substrate to assure enzyme identity. Adh would occassionally stain weakly without added ethanolbut not without the addition of NAD. Adh and Gdh were generally stained on the same gel slice. Additional loci for Adh, Est, and Got were sometimes present but were not scored. Individuals of B. discoidea were run with other species of the genus which have allelic variants for most of the loci scored as an internal control of resolution. In Table 3 the locus with the most anodal migration is designated as "1" with the most anodal allele at each locus designated

"a". Root tips for chromosome counts were pretreated with

0 .1# colchicine and fixed in 3 parts ethanol: 1 part acetic acid. Buds were fixed in U parts ethanol: 3 parts chloroform: 1 part acetic acid. Chjromosomes were squashed and stained by the method of Snow (1963).

Voucher specimens for chromosome counts and populations examined electrophoretically are deposited at OS. 93

RESULTSi MORPHOLOGY AND REPRODUCTION The chromosome number of B. discoidea was known from only one count prior to the present study and that of B. vulgata has not been reported. A brief summary of previous counts with my own additional determinations are presented in Table 1. The counts are consistent with a presumed base number of 2n = 2^ for Bidens section Platvcarnaea. The determinations also verify that the two most similar species, B. frondosa and B. discoidea, have different ploidy levels. Bidens discoidea has small inconspicuous heads with only 2-5 outer involucral bracts and no ray flowers. The inner involucral bracts nearly cover the 8-20 short disc florets with their inflexed tips at anthesis. The stamens, style, and stigmas barely extend beyond the incurved tips of the corolla lobes. Very low numbers of pollen grains are produced relative to related species. The plants self-fertilize with nearly 100% seed set when isolated from other individuals or if the heads are bagged. Flowers do not set fruit when the styles and anthers are excised, thus they do not appear to be apomictic. The styles have a small nectary at the base but the flowers produce no detectable odor or nectar. No visitation by potential pollinators was observed in natural populations during repeated visits to several populations at various times of the day. At localities with mixed populations of several Bidens species, insect visitors were confined to the larger-headed species of more open habitats. Natural 9^ reproduction is highly autogamous, especially for small heads produced late in the season. These lateral heads often set seed without the phyllaries and flowers opening and are functionally cleistogamous. The fruits of B. discoidea differ from B. frondosa and B. vulgata in "being smaller with the two awns reduced in length relative to achene size. The outer achene surface is pustulose with antrorsely oriented epidermal hairs and the awns are always antrorsely "barbed. Bidens discoidea is the only diploid of section Platycarpaea to have consistently antrorsely barbed awns. The awns of the diploid B. vulgata are always retrorsely barbed while those of the tetraploid B. frondosa are polymorphic within populations. North American Bidens species are usually considered to be animal dispersed, but the antrorse barbing and reduced awn length in several taxa suggest a reduction of the dispersal function parallelling that found in the Pacific Island Bidens species discussed by Carliquist (1966). Although the fruits of B. discoidea are poorly suited for long distance terrestrial dispersal as compared to the retrorsely barbed species, they probably function well in local dispersal and seedling establishment. Viable fruits float for long periods of time. Most achenes were still afloat and ungerminated after six months in beakers subjected to periodic shaking. In contrast, B. vulgata fruits sank within a few days or germinated immediately under identical conditions. The fruit walls of B. discoidea are composed of several layers of elongate water resistant thick-walled cells and covered by a dense layer of hairs. In nature fruits apparently float for 95 long periods of time until adhering to a substrate, or they may float until germination when the seedling becomes attached. In mid-summer the seedlings are mostly found at a consistent height with relation to water level on the bark of emergent trees, old stumps, horizontal logs, or occasionally in a line of debris marking a previous high water level. As the water recedes in late summer the plants produce long aerial roots extending to the water or substrate. They may superficially appear to be epiphytes occurring as a ring of individuals on tree trunks two or three feet above ground level. This form of seedling establishment differs from that of other species of Bidens which occur in similar habitats (including B. frondosa, B. connata. and B. cernua). These species normally occur at a variety of heights in relation to water level, suggesting that seed placement is determined by where they settled after sinking to the substrate or where they adhered by their retrorsely barbed awns. In mixed populations B. discoidea occupies the most shaded habitats while related species are found in more open marshy areas. In comparison to Bidens frondosa and B. vulgata. B. discoidea is a. habitat specialist. It's morphological adaptations suggest that gene exchange within populations through outcrossing is limited and that fruit dispersal between population is limited. 96

RESULTS: GENETIC VARIATION

The I83 individuals examined electrophoretically represent the progeny of 112 parents collected in the field. For population 2 (Table 2) self-pollinated Fg greenhouse progenies of 9 plants from 3 parents, and F^ progenies of 8 plants from 2 parents were examined. They were run with F^ progeny of the original field-collected plants with no detectable differences in the banding patterns. No variability was found in progeny collected in different years from populations 2 and 4. The nine enzymes were interpreted as being coded by 16 gene loci. Allele frequencies for the polymorphic loci Adh, Pgm-1, Pgm-2, Pgi-1, Pgi-3, Got-1, and 6-Pgd are reported in Table 3* The loci coding for Lap, Gdh, Pgi-2, Got-2, 6-Pgd-2, 6-Pgd-3, Est, and Acph were interpreted as being monomorphic. The genetic interpretation for enzymes in which heterozygotes were found is based on segregation of bands in progeny of field collections and on the known subunit structure of the enzymes in otheir plants species (Gottlieb, 1981a). The explanations for invariant loci are based on observations of variation in related diploid and tetraploid species and the number of isozymes present in other diploid plants (Gottlieb, 1981c). The enzymes Pgi and 6-PGD are normally expressed as one cytoplasmic isozyme and one plastid enzyme in most diploid plants (Gottlieb, 1981c). In all B. discoidea populations both Pgi and 6-Pgd have 97 consistently four-banded phenotypes. For Pgi only the slowest most strongly staining band is expressed in uncrushed soaked pollen extract and ungerminated seeds. This suggests that the slowest locus is cytoplasmic while the three more anodal bands expressed only in green tissue are localized in the chloroplasts (Weeden & Gottlieb, 1979» 1980a, b). As the three faster bands do not segregate in progeny or populations and are located in a single organelle, they are best interpreted as the two homodimers and the interlocus heterodimer coded by two duplicated gene loci (Weeden & Gottlieb, 19801,b; Gottlieb, 1981c). The slower mobility of the presumed interlocus heterodimer between Pgi-lb and Pgi-2 in population 4 as compared to the heterodimer produced in populations with the fast allele Pgi-la is consistent with this hypothesis. The interpretation of the four-banded pattern of 6-Pgd as a gene duplication is based upon similar patterns of slower mobility in the interlocus heterodimer of population 4, lack of segregation in progeny and populations, the known dimeric subunit structure of the enzyme, and the known number of isozymes in other plants (Weeden & Gottlieb, 1980b; Gottlieb, 1981c). The two-banded Got pattern of population 4, as opposed to the 3-bande.d pattern of other populations, can be explained by presuming either a null allele at the Got-1 locus or an allele with a mobility identical to Got-2. The alternate hypothesis that the majority of populations have a duplicated locus can be rejected because no heterodimer is formed between any of the loci reported for Got. The Got-1 interpretation makes little difference for the purposes of this 98 analysis, but the isozymes are best regarded as being coded by three loci with a probable null allele at Got-1 in population 4-.

An extremely low amount of genetic variation as measured by electrophoresis occurs within any population (Table J>). Each population is essentially fixed for a single allele at each locus. No variation was detected at localities such as

Lake Podotopaug (^) and Calamus Swamp (2 ) in 27 and 26 individuals, respectively. These populations contain thousands of individuals and samples were collected from a large area of the habitat. The intrapopulational variation detected for Pgm-1 and Pgm-2 consists of alleles present in low frequency but which are fixed in a number of other populations. The single individual variant at Pgi-3 found in population 8 is a rare allele not detected in other populations. The five other allelic variants coding for Adh-a,

Pgm-2, Pgi-lb, Got-lb, and 6-Pgd-lb are unique to single sampled populations and fixed in them. The highest mean number of alleles per locus in a population is 1.13 with a mean across all populations of only I.03 (Table 3 ). Considering only the six loci polymorphic fer the species as a whole (by the 99% criterion, Gottlieb, 198la), the mean proportion of loci polymorphic for the species as a whole is 0 .3 8 . The mean number of loci polymorphic per population for the species is only 0 .0 3 .

The genetic identity between plant populations and species is a widely used measure of their relative genetic 99 similarity (Nei, 1972). The genetic identity I, ranges from 1 .0 when two population have identical allele frequencies at each locus, to 0 .0 when two populations do not have any alleles in common at any locus. The average genetic identity between populations of B. discoidea is 0.865 (Table 40 , but ranges from 0.688 between the Lake Pocotopaug population (40 and several others to 1 .0 between a North Carolina (7) and Virginia (6) population. There is a low correlation between genetic identity and geographic distance. The two Connecticut populations (4 and 5) are separated by only 25 km but have only a 0.688 identity, while the

identical Virginia (6) and North Carolina (7) populations are separated by 160 km. The three southeastern populations 6, 7» and 8 have an average genetic identity

of 0.973 with each other but also show an average identity of 0.956 with population 2 from Ohio. 100

DISCUSSION Comparably low levels of heterozygosity and withi^n ‘ population variation are mostly confined to self-pollinating plants (Gottlieb, 1977, 1981a; Nevo, 1978 Hamrick et al, 1979)* In Bidens discoidea only three heterozygotes were found over all loci in the original sample, or only 0.2% heterozygosity for the 1792 loci tested. Rick and Fobes *(1975) found equivalent patterns of low within population variation and low heterozygosity (0.2$) in populations of autogamous Galapagos tomatoes. Moran and Marshall (1978) found that most populations of the annual selfing weed Xanthium were monomorphic at all loci in Australia where it has been introduced. Annual self-pollinating plants such as Chenooodium (Crawford & Wilson, 1977) and Hordeum (Nevo e$ al. 1979) often have few heterozygotes despite the fact that populations may occasionally be polymorphic with two or three alleles at some loci. Within the range of patterns in variation reported for selfers (Brown, 1979; Gottlieb, 1981a) B. discoidea is typical in the low number of alleles per polymorphic locus but has a high percentage of polymorphic loci for the species. The range of mean genetic identities between pairs of conspecific self-pollinating plant populations is 0.890-1.00, with a mean of 0 .975i for 13 species tabulated by Gottlieb

(1981a). The average genetic identity of O .865 between all B. discoidea populations is unusually low. Even if the divergent population 4 from Connecticut is excluded, the 101 average interpopulation identity is only 0.885. Mean genetic identities of only 0.846 have been reported between all populations of Chenonodium incanum (Crawford, 1979)» "but within any one of the varieties of the species the lowest level is 0.871. Brown (1979) has pointed out that inbreeding plant species generally have greater differences between populations than outbreeders relative to levels of variation within population. Bidens discoidea typifies this pattern. The mean of interpopulational genetic identities for B. discoidea falls, within the broad range of interspecific genetic dientities reported for congeneric plants, but is well above the mean of O.67 for such comparisons (Gottlieb, 1981a). The Lake Pocotopaug population (4) consistently shows low genetic identities with all other populations owing to fixation of several unique alleles. The plants appear typical morphologically, but were growing in rather open habitats for the species. It was anticipated that the more northern populations might show a reduction in genetic diversity due to the founder effect. All occur in lowland habitats resulting from Wisconsin Glaciation. Although the sample of southern populations is small it is clear that differences exist between populations rather than within them. Most potential colonizers migrants would be very homozygous and result in monomorphic populations with a composition dependent on that of the founder individual. If multiple founder events had occurred several distinguishable lineages might occur within a population. An alternative hypothesis which cannot he excluded is the possibility of genetic drift owing to repeated fluctuations in population size. The low levels of identity between B. discoidea populations may simply reflect the randomness of dispersal and lack of gene flow between discontinuous habitats. The extreme homozygosity of B. discoidea contrasts sharply with that of B. frondosa and other tetraploids. Most populations of these species have higher functional heterozygosity due to the additive expression of the duplicated genome. Preliminary studies indicate the presence of a number of alleles not found in B. discoidea and strongly suggest that B. frondosa is not an autopioid derivative. 103

CONCLUSIONS

Bidens discoidea exhibits low levels of intrapopulation electrophoretic variation coupled with substantial differences between populations. This apportionment of variation is consistent with, but extreme, in comparison to patterns typically found in inbreeding plants. The allozyme variation is not congruent with any geographic pattern or the relative morphological uniformity of the species. The low levels of heterozygosity observed are consistent with the rather high rate of selfing suggested by morphological evidence and greenhouse investigations. Near fixation of a single allele in each population at loci polymorphic for the species as a whole may largely be due to the founder effect and low rates of migration into established populations. 104-

Table 1. Chromosome counts and reports of Bidens discoidea(

B. frondosa* and B. vulgata. Collection numbers are those of the author. A number of identical counts have been nublished for B. frondosa.

Taxon, Citation Chromosome or report Number

B. discoidea n=12, 2n=24- Ohioi Pickaway Co., No. 4-84-2 Maryland* Mangaly et al, 1967 n=12 B. frondosa Ohio* Delaware Co., No. 4-905. 5377 n=24- Quebec* Drummond Co., No. 4-920 n=24- British Columbia* Coquitlam, No. 53^7 n=24- B. vulgata Ohio* Pickaway Co., No. 5029 n=12, 2n=24- Ontario* Thunder Bay District, No,. 5^07 2n=24- 105 Table 2. Populations of Bidens discoidea examined for allozymes. Each population is assigned a reference number and information is presented fori a) collection number(s) of the author; b) geographic locality; c) number of parental plants examined; d) actual number of individual seedlings examined.

1« 5391» Wisconsin, Barron Co., Veterans Park. 14 plants, 22 progeny. 2. 4842, 55°8, Ohio, Pickaway Co., Calamus Swamp.

36 plants, 57 progeny. 3. 5505» Ohio, Logan Co., Indian Lake.

12 plants, 16 progeny. 4. 4980, 5249* Connecticut, East Hampton Co., Lake Pocotopaug. 2? plants, 34 individuals. 5« 5245, Connecticut, Middlesex Co., Northwest River. 9 plants, 19 progeny. 6. 5563, Virginia, Norfolk Co., Deep River. 6 plants, 8 progeny. 7. 5548, North Carolina, Pitt Co,, Farmville. 4 plants, 9 progeny. 8. 5561, North Carolina, Washington Co., Plymouth. 4 plants, 18 progeny ZABLE 3. Allele frequencies for seven variable entyae loci In Bldena discoidea. Locus abbreviations are

are explained In Materials and Methods. « H . ■ nuaber o f observed heterozygotea of In i t ia l seedling froa each parent; P“ proportion of ODS lo c i polymorphic per population; A * aeon nuaber of a lle le a per locus*

* Locus

Population H . P A Adh Pga-1 Pgm-2 P81-1 Pgi-3 Got-1 6Pgd Numbers ob# a b a b a b c a b a b a b a b

1 0.013 0.117 1.13 1.0 0.07 0.93 ' 0.96 0.04 1.0 1.0 1.0 1.0 2 0.0 0.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 3 0.0 0.0 * 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 4 0.0 0.0* 1.0 1.0 1.0 1.0 1.0 1.0 1.0 5 0.0 0.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 6 0.0 0.0 1.0 1.0 l.o' 1.0 1.0 1.0 1.0 1.0 7 0.0 0.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 8 0.0 0.117 1.13 1.0 1.0 0.75 0.25 1.0 0.75 0.25 1.0 1.0

Mean -0.002 0.03 1.03 90T Table 4. Genetic Identities among populations of Bidens discoidea calculated by the method of Nei (19V2).

Mean Population Znterpopulational Humber Id e n tity 1 2 3 4 5 6 7 1 0.835

2 0.881 0.902

3 0.935 0.938 0.882

4 0.6880.6880.750 0.717

5 0.821 0.938 0.875 0.688 0.877

6 0.823 0.938 0.875 0.750 0.938 0.898

7 0.823 0.938 0.875 0.750 0.938 1.00 0.898

8 0.872 0.992 0.928 0.704 0.944 0.960 0.960 0.909 Overall 0.865 Mean 107 LITERATURE CITED

Brown, A. H. D. 1979* Enzyme polymorphism in plant populations. Theor. Pop. Biol. 15* 1-42. Carlquist, S. 1966. The biota of long-distance dispersal. II. Loss of dispersibility in Pacific Compositae.

Evolution 2 0 : 3°-48. Crawford, D. J. 1979* Allozyme studies in Chenopodium incanum: intraspecific variation and comparison with

Chenopodium fremontii. Bull. Torr. Club 106: 257-261. Crawford, D. J. & H. D. Wilson. 1979* Allozyme variation in several closely related diploid species of Chenopodium of the western United States. Amer. J. Bot. 6 6: 237-244.

Gottlieb, L. D. 1973* Genetic differentiation, sympatric speciation and the origin of a diploid species of

Stephanomeria. Amer. J. Bot. 6 0: 545-553* Gottlieb, L. D. 1977* Electrophoretic evidence and plant systematics. Ann. Mo. Bot. Gard. 64: 161-180. Gottlieb, L. D. 1981a. Electrophoretic evidence and plant populations. Progress in Phytochemistry 7* 1-46.

Gottlieb, L. D. 1981b. Gene number in species of Astereae with different chromosome numbers. Proc. Natl. Acad. Sci. U.S.A., in press. Gottlieb, L. D. 1981c. Conservation in and duplication of isozymes in plants. Science: in press. Hamrick, J. L . , Y. B. Linhart & J. B. Mitton. 1979- Relationships between life history characteristics and electrophoretically detectable genetic variation in plants. Ann. Rev. Ecol. Syst. 10s 173-200. Mangaly, J. K . , R. A. Davidson & R. A. Dunn. 1967- I0PB chromosome number reports IX. Taxon 1 6: 62-6 6. Moran, G. F. & D. R. Marshall. 1978. Allozyme uniformity within and variation between races of the colonizing species Xanthium strumarium L. Aust. J. Biol. Sci. 31

283-291. Nei, M. 1972. Genetic distance between populations. Amer. Nat. 106: 283-292. Nevo, E. 1978. Genetic variation in natural populations: patterns and theory. Theor. Pop. Biol. 13s 121-177- Nevo, E. , D. Zohary, A. H. D. Brown & M. Haber. 1979- Genetic diversity and environmental associations of wild barley, Hordeum snontaneum. in Israel. Evolution 33* 815-833- Rick, C. M. & J. F. Fobes. 1975- Allozymes of Galapagos tomatoes: polymorphism, geographic distribution, and affinities. Evolution 29s • Sculthorpe, C. D. K. 1967. Biology of Aquatic Vascular Plants, Academic Press, New York.

Sherff, E. E. 1937- The genus Bidens. Field Mus. Nat. Hist. Publ. 388. Bot. Ser. (Fieldiana) pts. 1 & 2. 110

Sherff, E. E. 1955* Bidens. pp. 70-129* Ini E. E. Sherff E. E. Sherff & E. J. Alexander, (editors). Composita Compositae— Heliantheae— Coreopsidinae. North Am. Flora. Ser. II, Pt. 2. Snow, R. 1963. Alcoholic hydrochloric acid-carmine as a stain for chromosomes in squash preparations.

Stain Tech. 3 8 * 9-13- Weeden, N. F. & L. D. Gottlieb. 1979* Distinguishing allozymes and isozymes of phosphoglucoisomerase by electrophoretic comparisons of pollen and somatic tissues. Biochem. Genet. 17* 287-296. Weeden, N. F. & L. D. Gottlieb. 1980a. The identification and isolation of cytoplasmic enzymes form pollen. Plant Physiol. 66s 4-00-^03. Weeden, N. F. & L. D. Gottlieb. 1980b. The genetics of chloroplast enzymes. J. Hered. 71: 372-396. CHAPTER IV

VARIATION AMONG THE POPULATIONS OF BIDENS BLDENTOIDES (COMPOSITAEt COREOPSIDINAE)

111 112

INTRODUCTION The freshwater estuaries of major northeastern North American rivers contain a number of endemic plant species (Fassett, 1928). Many of these species have been considered highly endemic, with a particular taxon confined to a single estuarine system or disjunct to several nearby estuaries. Fassett (1928) said that "Estuarine plants are of ancient distribution, as evidenced by their development of different varieties in each esturary." He suggested that the distributions of the plants dated to an early postglacial migration and that they had subsequently differentiated into a myriad of taxonomically recognizable forms' "because of the extreme conditions under which they grew" (Fassett, 1928). The majority of the estuarine endemics occur in river systems from the Chesapeake northward to the St. Lawrence. This general distribution also coincides with the distribution of the majority of estuarine endemics described in the genus Bidens. Of the three species of Bidens which are restricted to estuarine habitats, B. bidentoides (Nutt.) Britton is the least similar to any of the widespread inland species of the genus, and is geographically less widespread than the other estuarine species. It was originally described by Nuttall from the intertidal shores of the Delaware River. It has since been discovered in the Hudson River of New York, the Maurice, South, and Raritan Rivers of New Jersey, and the tributaries of northern Chesapeake Bay. Habitats and the distribution of B. bidentoides in the Delaware River area have been described by Ferren & Schuyler (1980). 113

On the basis of morpholgy and flavonoid chemistry, B. bidentoides belongs to the B. connata Willd. species complex (Roberts unpub.). B. eatoni Fern., an estuarine species occuring with B. bidentoides and B. connata in the Hudson River estuary, is also a member of this group of species. B. bidentoides is similar to B. eatoni in having the ability to produce a variety of leaf forms on a single plant, ranging from simple and lanceolate to deeply lobed. Both species are also tolerant of the twice-daily flooding by fresh water which occurs in the environments in which they grow. B. bidentoides is easily distinguished from B. eatoni and B. connata on the basis of achene and capitulum capitulum features (Sherff, 1937)* Blake (1929) segrated the Chesapeake Bay populations of B. bidentoides as a distinct species, B. mariana Blake.

Sherff (1930 ) suggested that if this were done, the many variants of other Bidens species would have to be recognized at an equivalent rank for the sake of consistency. He argued that it should be considered a variety of B. bidentoides

(Sherff,, 1930 ), and followed this opinion in subsequent monographic treatments of the genus (Sherff 1937» 1955)* This paper investigates the variation among the populations of B. bidentoides and the var. mariana using enzyme electrophoresis. The results are briefly compared to geographical variation in morphology, flavonoid chemistry and cytological data. 114 MATERIALS AND METHODS

Population samples consisting of entire specimens or fruiting heads from individual plants of Bidens bidentoides were collected over two seasons of field studies from each known extant population of the species (Table l). Fruits from individual plants were soaked on filter paper in petri dishes under 24 hr. light. The embryos were excised from the fruit wall and the testa to induce germination. Plants from the Chesapeake, Delaware, Maurice, and Hudson estuaries were grown to maturity in the greenhouse.

Young seedlings (ca. O.Olg) were ground in chilled trays in approximately 0.2 ml of a buffer composed of 0.1'iM tris-HCl, 1.0 mM EDTA, 10 mM K C ;, 10 mM MgClg, 10 mM mercaptoethanol, 40$ sucrose, pH 7*5 (Gottlieb 1981b). The extract was centrifuged and the supernatant pipetted into the wells of vertical acrylamide gels or taken up on wicks and inserted into slots of horizontal starch gels. Starch gels were composed of 12.5$ Sigma starch using two different electrode-gel buffer systwms. System 1 used a gel buffer of 9 parts tris-citrate (pH 8.3; 0.2 M) and one part lithium-borate (pH 8.3; 0.2 M), with lithium-borate alone employed as the electrode buffer (Gottlieb 1973)*

System 2 employed a 0.4 M sodium citrate electrode buffer adjusted to pH 7*0 with HC1 and 0.02 M histidine-HCl gel buffer adjusted to pH 7*0 with NaOH, and was run at 100 ma.for approximately 8 hrs. (Gottlieb 1981b). Acrylamide gels were composed of a 5*5$ separating gel with a tris-HCl 115 (pH 8.9? 0*^ M) buffer, a Z.Ofo spacer gel with a tris-HCl (pH 6.9 ; 0.06 M), and an electrode buffer of tris-glycine

(pH 8.3; 0.05 M). Enzyme staining solutions were modified only slightly from those of Gottlieb (1973)* Satisfactory resolution and activity were found for the following enzymes* alcohol dehydrogenase (Adh), phosphoglucomutase (Pgm), phosphoglucose isomerase (Pgi), leucine aminopeptidase (Lap), glutamate dehydrogenase (Gdh), glutamate-oxaloacetate transaminase (Got), 6-phosphogluconate dehydrogenase (6-Pgd), esterase (Est), and acid phosphatase (Acph). Est, Lap, and Acph were run' on starch system 1, Got on acrylamide, and the other enzymes on system 2 with occasional comparisons on system 1. Extracts of plants from various populations were run side by side in various combinations. For all enzyme systems a control was run without substrate to assure enzyme identity. Adh would stain weakly without x added ethanol but not without the addition of NAD. Adh and Gdh were generally stained on the same gel slice. Additional loci for Acph, Est, and Got were sometimes present but were not scored. Individuals of B. bidentoides were run with other species of the genus having allelic variants for most of the loci scored as an internal control of resolution.

In Table 4, the locus with the most anodal migration is designated as "I", with the most anodal allele at each locus designated ''a'*. The scoring system was designed to accommodate plants from several related species, therefore 116 all alleles at a particular locus do not necessarily occur in B. bidentoides. Morphological measurements were made from vigourous mature plants collected in the field. Large primary leaves and ray florets are often absent on such plants at maturity and could not be compared on the basis of available collections. Vouchers of collections of the author (Table 1) are deposited at OS with duplicates at PH. Other specimens were examined from the following herbaria* DAO, F, GH, MTJB, NEBC, OS, and PH. 11?

RESULTS MORPHOLOGY— The Chesapeake populations were segregated by Blake (1929) primarily on the basis of the pubescent corolla tubes of the disc florets. He also indicated they differ from the Delaware River populations in the length of the achenes, awns, and phyllaries, and in leaf shape. An examination of several hundred specimens from each of the major populations indicates that there are consistent differences between the Chesapeake and more northern populations. Chesapeake plants differ from all the other populations in consistently having pilose disc corollas. However, within this population the density of the pubescence varies considerably from plant to plant. In the other populations only the ray florets have pubescent corollas. The differences in pubescence were maintained in selfed progeny of plants from the respective populations cultivated in the greenhouse. In the Chesapeake populations the phyllaries are, on the average, shorter than those of the more northern populations (Table 2). The achenes and awns in the Chesapeake populations are also, on the average, shorter than those of the other poulations (Table 2). The ratio of awn length to achene body length is also lower for the Chesapeake populations (x = 0 .67). The Chesapeake populations also differ in having longer ray 118 florets which are conspicuous when the plants are in flower. They far exceed the shorter phyllaries (rays to 20 mm in length, or 1.5 times the length of the inner phyllaries). The populations of the Delaware and Hudson are discoid or have rays barely exceeding the inner phyllaries. Measurements of ray floret length are not presented because collections from most northern populations were made after anthesis of the primary heads, and the fragile rays do not persist in the intertidal environment. Although the morphological measurements indicate that the populations from each watershed can be separated by quantitative characters, the features do not differ as discretely as indicated by Blake (1929) and Sherff (1937» 1955)• Blake (1929) also suggested that the leaves of the Chesapeake populations produce laciniate-lobed leaves with "sharper" teeth. Although collections from the Delaware and Hudson produce deeply lobed leaves less frequently, this may be due to habitat differences. Lobed leaves are generally produced only on the lower nodes of vigourous plants. Late in the season such leaves are usually absent from plants in intertidal environments. Several of the Chesapeake populations are from rather "high" marshes where only the lower portions of the plants are innundated by tides and large primary leaves are likely to be more persistent. CYTOLOGY— Meiotic and mitotic preparations from B. bidentoides and B. bidentoides var. mariana indicate the chromosome nuber to be n = 24, 2n = 48 (Table 3). The base chromosome number of the genus Bidens is considered to be x = 12 (Solbrig et al. 1972; Stuessy 1977)* and diploids with this number have been reported from most of the North American species groups. A tetraploid condition is not unexpected for B. bidentoides. as counts • of n = 24, 2n = 48 have been reported from the related B. connata (Weedon and Butler 1976), and have been found in other populations of that species and other species. This represents the first report for B. bidentoides and the populations examined are listed in Table 3* The polyploid nature of the other populations of B. bidentoides is indicated by the banding patterns observed with enzyme electrophoresis.

In the Hudson River populations (Table 3 ) various numbers of multivalents and univalents were observed in several cells from one of the two plants studied. It has not been determined whether the meiotic irregularities are normal in the population. I have observed that Bidens in stressful conditions (primarily heat) often show meiotic irregularities which are not found in other progeny of the same plant in less stressful environments. Cytological preparations were made from a larger numer of Chesapeake Bay plants and all showed uniform bivalent pairing at meiosis. FLAVONOID CHEMISTRY — A survey of the populations of Bidens bidentoides was included in a study of the flavonoid chemistry of North American Bidens species (Roberts unpub.). The species is very similar chemically to related tetraploid species in the B. connata species group. The only detectable difference among the populations of B. bidentoides is the presence of a single pair of compounds

in the Chesapeake Bay populations. These compounds appear to be acylated derivatives of the anthochlors marein and maritimein. They occur in high concentration in the Chesapeake Bay populations, in trace amounts in the Maurice River populations, but were not detected in the other populations. In B. bidentoides the production of the acylated compounds coincides geographically with the populations segregated as B. mariana by Blake (1929). 121 ELECTROPHORESIS— The 213 plants examined represented the progeny of 122 parents collected in the field. The additional individuals are progeny of field pollinated plants which were examined for inheritance or as internal standards in comparisons. Each plant was examined for 8 enzyme systems putatively coded by 11 loci. The term locus is used in the sense of a diploid plant, when in reality two gene loci presumably code for each isozyme in the tetraploid plants, i.e. they are duplicated loci. Electromorph frequencies at the polymorphic loci Adh, Pgi, Pgm-2, Got-1, and Got-2 are presented for each population in Table 4-. All plants examined were identical at loci coding for Lap, Gdh, Pgm-1, Est, and Acph. A consistently one-banded pattern was produced by Lap and Gdh. A consistently two-banded pattern was produced by Pgm-1, Est, and Acph. Related diploid species produce one-banded patterns at all of these loci in homozygours plants. Pgi-1 was always three-banded and monomorphic except for 2 individuals from the Hudson River estuary which had a faster allele for the more anodal band. This rare allele is not listed in Table 4, as it barely meets the criterion for polymorphism (Gottlieb I98I; mean frequency of most common allele O.997S). Results of progeny tests indicate that most individuals are "fixed heterozygotes" at many loci. The multiple-banded electrophoretic patterns indicative of heterozygotes in related diploid species (B. discoidea. B. cernua. B. laevis) do not segregate in progeny of B. bidentoides. No indication 122 of true genic heterozygosity was found. In some cases the true-breeding phenotypes indicating fixed heterozygosity characterize all of the populations of B. bidentoides ( for Pgm-1, Est, and Acph). No "homozygous" phenotypes were observed at loci coding for Pgm-1, Est, Acph, or Pgi-1. For Adh, Pgi-2, Pgm-2, Got-1, and Got-2, individuals or populations were found which were electrophoretically "homozygous" (Table A). The genetic interpretations for enzyme phenotypes of B. bidentoides are based on the known subunit structure of the enzymes in other plant taxa (Gottlieb 1981a), and my own observation of allozyme variation in related diploid Bidens species. The absence of an allele in an individual of B. bidentoides is interpreted as indicating the plant is a fixed homozygote at the locus. It cannot be demonstrated from the available data that such a pattern is not due to a "null" allele being fixed at a locus in one of the genomes. The true-breeding 3-^anded patterns exhibited by most B. bidentoides populations for Adh and Pgi-2 are indicative of plants "fixed" for two alleles at the loci (the three bands representing the two homodimers and heterodimer). The one-banded patterns found for both loci in some populations indicate that both genomes of the tetraploid produce the same electrophoreticall indistinguishable allele. Pgm is a fixed heterozygote in all populations and always showed a two-banded pattern. Pgm-2 is homozygous in all populations except the Hudson and South River populations, where a high 123 percentage of plants are heterozygous and two-banded. Banding patterns of Got were difficult to interpret due to the large numer of bands and the probable overlap of alleles produced at two loci. As the progeny of individual plants breed true, segregation patterns could not be observed, and no plants were found which were homozygous at both duplicated gene loci (two-banded). The plants from the Hudson River populations were particularily variable at Got-1 and Got-2 (Fig. l). Plants producing 4-, 5-» and 6-banded patterns were observed. The several ^—banded patterns (Fig.1) are interpreted as being heterozygous at one locus and homozygous at the other. The 6-banded patterns are undoubtedly due to heterozygosity at both loci of the dimeric enzyme. The 5-banded patterns are believed to be due do overlap of the fastest allozyme homodimer produced at locus Go.t-1 (Got-lcc) with the slowest allozyme homodimer produced at locus Got-2 (Got-2ad). Such an overlap would be expected on the basis of the relative mobilities of the alleles observed in plants homozygous at the two loci. Variation in gene dosage may exist at several loci. This effect was noted for Adh, Pgi-2* Pgm-1, and Pgm-2. For Adh, seedlings which had not been well watered prior to examination often did not exhibit the expected hterozygous condition. Examination of additional progeny from the same plant indicated the bb homodimer and ab heterodimer produced by the Adh-ab genotype were induced differentially or were less active. When consistent differences in staining 124 intensity were observed in the other loci, the less common and less widespread allele was always the one which stained less heavily. No attempts were made to evaluate differences in staining intensity between populations due to difficulty in scoring them consistently. For each of the polymorphic loci, one 'lalleld' was nearly * ubiquitous. For Adh, the Adh-a allele was present in every population. Polymorphism was present only due to the absence of the Adh-b allele from most plants in the Chesapeake population (Table 4). Other alleles which were ubiquitousin all populations, even at polymorphic loci, include Pgi-2d, Pgm-2b, and Got-2d. Got-la was present in all plants other than a small number from the Hudson River population. Polymorphism within and differences between populations are almost entirely due to the absence of the less common allele at the variable loci. The alleles which are ubiquitous in B. bidentoides are those which are also ubiquitous or most common in other tetraploid species such as B. connata. B. eatoni, and B. frondosa (Roberts unpub.). The allelic differences between the species are primarily due to the alleles whose presence is variable. The amount of variation within populations can be measured by the number of loci which are polymorphic within the population and the number of "alleles" within the population (Table 5)* By both measures the Hudson River populations contain more genetic diversity than the others, with 23 alleles in the population. Populations of the smaller Delaware and Maurice Rivers are monomorphic at every locus and the South River population has only one variable locus in the small sample. The Hudson River population has more true biochemical variability due to the presence of the higher number of alleles and higher number of loci with unique allelic combinations (Table 5)- The Chesapeake Bay population has many plants with less biochemical variability because many of the unique allelic combinations are due to the "homozygous'' condition at loci which are fixed heterozygotes in other populations. The Delaware and Maurice populations are monomorphic but lose little biochemical diversity because they are fixed in the heterozugous condition at 8 of the 11 loci. The difference between the populations cannot be analyzed in terms of standard measures of genetic similarity due to the tetraploid condition (Gottlieb 1981a). Differences between them can be summarized on the basis of how many unique allelic combinations they contain. The Chesapeake . population contains unique allelic combinations at ADH and

PGI-2 (Table 5)» The Hudson River populations contain five allelic combinations not found in any other watershed. They are at Pgi-2, 2. at Got-1, and 2 at Got-2, for a total of 5. However, two of the Got combinations are present in low frequency (Table and would not contribute heavily to most measures of genetic distance (Gottlieb 1981a). The Chesapeake Bay and Delaware River populations differ in certain allelic combinations, but both contain the same set of alleles (Table 4). The Chesapeake population differs only in that it contains individuals which have a subset of the alleles of the entire population. The Chesapeake population differs from the Hudson in lacking 4 alleles present in the latter; the Hudson and Delaware populations differ in the absence of the same four alleles. Selected plants from the Hudson and Chesapeake populations would, however, vary by greater numbers of alleles, and the populations as a whole would differ in the relative frequency of alleles and allelic combinations. 127

DISCUSSION

The populations of Bidens bidentoides are geographically discontinuous. Each is confined to a short stretch of freshwater intertidal zone which is separated from other such areas by saltwater or terrestrial habitats. Blake (1929) suggested that the Chesapeake and Delaware populations have been separated only since the late Pleistocene when sea level changes resulted in their isolation. The Maurice River population can probably be considered a subpopulation of the Delaware River population on the basis of geography and the similar floristic composition of the two watersheds (Ferren & Schuyler, 1980). Morphological and allozyme data also indicate the two are highly similar. The two watersheds may have been contiguous in the late Pleistocene during channel changes which occurred in the Delaware River and the greater flow of freshwater due to glacial meltwater (Twichell et al. 1977). The Hudson River population is probably a derivative one which has arisen since the retreat of Pleistocene glaciers which covered the watershed. The population could have been derived by immigration from the Delaware populations of from populations which may have survived in the Raritan River at the edge of the Wisconsin Terminal Moraine. If the Hudson River population is considered to be a derivative of one of the southern populations, it should be 1 2 8

expected to be more similar to one of them in morphological and biochemical features, and be less variable than either of them due to the founder effect. If the current systematic treatment is correct, the Hudson population should be more similar to the Delaware populations of the typical variety than those of the Chesapeake separated as variety mariana. Morphologically, the Hudson River plants resemble the Chesapeake Bay populations more closely than do the Delaware and Maurice populations for phyllary length, achene length, and awn length (Table 2). The Hudson populations are essentially intermediate between the other populations in all three features, and are slightly more similar to the Chesapeake populations than to the Delaware-Maurice plants for phyllary length and achene length (Table 2). With respect to disc floret pubescence and absence of the acylated anthochlor compound, the Hudson populations are identical to the Delaware population. For ray floret length and leaf morphology, Hudson populations also seem to be more similar to the Delaware populations. The allozyme phenotypes of the Hudson River population suggest that it is about equally similar to the Delaware and Chesapeake populations. The Chesapeake population contains unique phenotypes of Adh and Pgi-2 in high frequencies (Table ^)« The Delaware and Maurice populations contain no unique alleles or allelic combinations. The Hudson population (including the South River population) contains a number of unique alleles (Pgi-2a, Pgm-2a, Got-lc, 129 and Got-2a) and 6 unique allelic combinations (Table 5)* The Hudson River shares one allelic combination with the Chesapeake Bay populations which was not found in the intermediate Delaware-Maurice population. This is the Got-laa genotype present in the Hudson in low frequency (Table 4). The systematic implications of such shared homozygous allelic combinations is probably small due to the possibility of parallel recombinational events. The unusually high variability at both Got loci (Table 4) is largely responsible for the distinctiveness of the Hudson River population. The Hudson River population is variable at more loci (4) and contains more alleles (23) than any of the other populations. It has been suggested that the number of alleles in a population relative to the number in the species as a whole may be the best indicator of the evolutionary potential of the population (Gottlieb 1975)• In this repect the Hudson River population does not indicate a loss of variability consistent with expectations of the founder effect. The Chesapeake population is only slightly less variable, having more than one allelic combination at 3 isozyme loci and containing 19 alleles. However, the alleles Adh-b and Pgi-2b are present in very low frequencies (Table 4). The geographically intermediate populations of the Delaware and Maurice are remarkable only in their uniformity. They are monomorphic and lack unique alleles. The alleles in the population are the same as those of the Chesapeake, but they lack several allelic

\ combinations found in that population. The high level of biochemical diversity due to fixed heterozygosity has been one explanation for the success of polyploidy in plants (Roose & Gottlieb 1976). In Bidens, species such as B. discoidea have extremely low levels of heterozygosity and polymorphism within populations (Roberts unpub.). Alloplyplids which incorporate divergent genomes are potentially able to produce a greater diversity of polypeptides, including novel intergenomic heteromers of polymeric enzymes (Mitra & Bhatia 1971; Gottlieb 1981a). In Bidens bidentoides each population shows a fixed heterozygous phenotype at 5 5f°) to 9 (82$) of the 11 enzyme loci scored. This is a minimum estimate limited by the resolution of electrophoretic techniques. The level of fixed heterozygosity is high compared with that observed in other plant genera in which tetraploids have been observed (Mitra & Bhatia 1971; Roose & Gottlieb 1976). In recently derived allotetraploid species of the genus Tragouogon. Roose and Gottlieb (1976) reported fixed heterozygous phenotypes at and 33 ^ of the gene loci for two species. The level of fixed heterozygosity was proportional to the divergence between the genomes of the progenitor species.

The three progenitor species had very low levels of actual heterozygosity.

The high levels of fixed heterozygosity suggest that the diploid progenitors of B. bidentoides had divergent genomes cding for different allelic products at many gene loci. The presence of "homozygous" phenotypes in some plants 131 at each of the 5 variable isozyme loci (Table 4) indicates that mechanisms of recombination are present at least in low frequency. The apparent mechanism of generation of homozygous phenotypes is occasional recombination through multivalent formation between homeologous chromosomes. The observation of multivalent formation in the Hudson

River population (Table 3) indicates that the potential for recombination events exists. Recurrent recombination events in inbreeding plants will eventually result in loss of enzyme multiplicity in allotetraploids (Roose & Gottlieb 1976). The extent to which this multiplicity persists in spite of recombination is a measure of selection for heterozygosity at a particular locus (Roose & Gottlieb 1976)• The presence of identical pairs of homozygous-heterozygous phenotypes for several loci in more than one population, suggests that independent parallel recombination events are responsible for generation of the identical phenotypes. For example, the electrophoretically identical Got-laa phenotypes in the Chesapeake and Hudson populations may have been generated by similar recombinational events in the Got-lab plants in each population. 132

CONCLUSIONS The tetraploid freshwater intertidal zone plant Bidens bidentoides is found in three disjunct populations. The Chesapeake Bay population can be distinguished by quantitative and qualitative morphological and chemical characters, and has been recognized taxonomically as variety mariana. The other two populations can be distinguished from each other by minor morphological differences. Populations from all three’ estuarine systems have very similar electrophoretic phenotypes which have high levels of fixed heterozygosity. The allozyme differences between populations are mostly due to different allelic combinations which were probably gererated through occasional recombinational events between the two genomes and promoted by inbreeding. The northernmost population in the Hudson River is the most variable both in having more allelic combinations and in containing more alleles. Geographic information suggests the Hudson River population should be of more recent origin than the others. Morphological evidence suggests the population is phenotypically more similar to the Chesapeake Bay population than populations of the Delaware system. Enzyme electrophoresis indicates that the Hudson River population is genetically about equally similar to the Chesapeake and monomorphic Delaware populations in its allelic composition. 133

Table 1. Populations of Bidens bidentoides examined for enzymes, flavonoids, chromosome number, or morphology. All collection numbers are those of M. L. Roberts, with vouchers deposited at OS.

Taxon and Collection Number Locality var. bidentoides ^995» 5263 NYi Greene Co., Hudson R. at Cosackie. ^990, 5265 NY« Greene Co., Hudson R. at Athens. 5268 NY« Rensselaer Co., Hudson R. at Coeymans. 5002, 5253 NY» Putnam Co., Hudson R. at Cold Spring. 5004, 5276 NJi Middlesex Co., South R. at Old Bridge. 5280 NJi Burlington Co., Delaware R. at Burlington. 5008 NJi Burlington Co., Delaware R. at Beverly. 5012, 5281 NJi Burlington Co., Delaware R. at Delanco. 5020, 5282 NJi Cumberland Co., Maurice R. at Milleville. var. mariana 5022, 5290 MDi Cecil Co., Bohemia R. at Cayote. 5024, 5288 MDi Cecil Co., Sassafras R. at Fredericktown. 5025, 5296 MDi Cecil Co., Chesapeake Bay at Charlestown. 5293 MDi Cecil Co., Elk R. at Elkton. 134

Table 2. Morphological variation in populations of Bidens bidentoides. Values are means and standard error measured in millimeters. Sample size is indicated

in parentheses. Achene measurements are based on 5-20 fruits per individual. Those values followed by the same lower case letter are not significantly

differrent at p=.05.

Character Phyllary Achene Awn Locality length • length Length

Chesapeake x=19-3 a x=8.0 a x=4.6 a (51) Sx=0.60 Sx=0.11 Sx=0. 068

Maurice x =34.5 b x=8.5 a x=5.8 b (12) Sx=l.53 Sx=0.13 Sx=0.O83 Delaware x=31.8 b x=9.1 b x=6.0 b (29) Sx=l.71 Sx=0.11 Sx=0.081

Hudson x=24.4 c X=8.1 a x=5-2 c (28) Sx=0.65 Sx=0.12 Sx=0.080 135

Table 3* Chromosme numbers of Bidens bidentoides. Collection numbers are those of the author as listed in Table 3-

Variety and Gametic Sprorphytic Collection number number Number var. bidentoides 5265 n = 23 II + 2 1 n = ca. 1 IV + 18 11+7 I 5268 2n =48 var. mariana II 5288 n =24 00 ,rv> 13 5024 n =24 Table 4. Enzyme phenotype frequencies for polymorphic loci in populations of Bidens bidentoides.

Locus/ Locality allele Chesapeake Maurice Delaware So. River Hudson

Adh-aa O .83 -ab 0.17 1.00 1.00 1.00 1.00 Pgi-2-bd 0.03 1.00 1.00 1.00 0.26 -cd 0.74 -dd 0.97 Pgm-2-ab O.83 0.59 -bb 1.00 1.00 1.00 0.17 o.4l Got-l-aa 0.66 0.15 -ab 0.34 1.00 1.00 1.00 0.35 -ac 0.48 -cc 0.02 Got-2-ad 0.28 -bd 1.00 1.00 1.00 1.00 0.57 -dd 0.15 Number of 35 13 22 6 46 families examined Number of individuals 87 24 28 6 68 examined 137

Table 5* Variability in enzyme phenotypes in populations of Bidens bidentoides.

Locaility Characteristic Chesapeake Maurice Delaware South Hudson

Number of loci polymorphic in 3 population Number of alleles in 19 19 19 20 23 population Number of loci with fixed 5-9 8 8 8-9 6-9 heterozygous phenotype Number of loci with unique allelic combinations Loci with Pgi-2 unique allelic Adh Got-1 combinations Pgi-2 Got-2 Figure 1. Diagramatic interpretation of Got-1 and

Got-2 electromorphs. See text for further discussion.

I38 Subunit Composition

2dd

2bd 2a d 2bb 1cc,2aa

1bb 1ac 1a b

1aa T Genotype 1aa/2bd 1ab/2bd lac/2bd 1ac/2dd 1aa/2ad 1cc/2bd 1ac/2ad Figure 1.

VO 140 LITERATURE CITED Blake, S. P. 1929. A new estuarine Bidens from Chesapeake

Bay. Rhodora Jit 87-100. Fassett, N. C. 1928. Vegetation of the estuaries of northeastern North America. Proc. Boston Soc. Nat.

Hist. 39* 73-130. Ferren, W. R., Jr. and A. E. Schuyler. 1980. Intertidal vascular plants of river sysetems near Philadelphia.

Proc. Acad. Nat. Sci. Phila. 132s 86-120. Gottlieb, L. D. 1973. Genetic differentiation, sympatric speciation and the origin of a diploid species of Stephanomeria. Amer. J. Bot. 60s 545-553* . 1975* Allelic diversity in the outcrossing annual plant Stephanomeria exigua ssp. carotifera (Compositae). Evolution 29s 213-225* . 1981a. Electrophoretic evidence and plant populations. Progr. in Phytochem. 7s 1-46. . 1981b. Gene number in species of Astereae with different chromosome numbers. Proc. Natl. Acad. Sci. U. S. A., in press. . 1981c. Conservation in and duplication of isozymes in plants. Sciences in press. Mitra, R. and C. R. Bhatia. 1971* Isoenzymes and polyploidy I. Qualitative and quantitative isoenzyme studies in the Triticinae. Genet. Res., Cambr.

18s 57-69* Roose, M. L. and L. D. Gottlieb. 1976. Genetic and

biochemical consequences of polyploidy in Tragonogon.

Evolution 30« 8I8-830 .

Solbrig, 0. T., D. W. Kyhos, M. Powell and P. H. Raven.

1972. Chromosome numbers in Compositae VIII.

Heliantheae. Amer. J. Bot. 59* 869-878.

Stuessy, T. F. 1977- Heliantheae-systematic review. In

The Biology and Chemistry of the Compositae. V. H.

Heywood, J. B. Harborne, and B. L. Turner, eds.

Academic Press, New York, pp. 62I-67I.

Twichell, D. C., H. J. Knebel and D. W. Folger. 1977-

Delaware River: Evidence for its former externsion

to Wilmington Submarine Canyon. Science 195* 483-485.

Weedon, R. R. and M. G. Butler. 1976. In IOPB chromosome

number reports LII. Taxon 25* 346. CHAPTER V CYTOLOGY, BIOLOGY, AND SYSTEMATICS OF MEGALODONTA BECKII (COMPOSITAE)

1 4 2 143

INTRODUCTION

Megaolodonta beckii is a perennial aquatic plant with dimorphic leaves which grows in lakes and streams of glaciated areas, primarily of eastern North America. It is one of the most hydrophytic species of the predominantly terrestrial family Compositae. The submersed leaves of the plant are highly segmented and filiform, while the simple emersed leaves are normally formed prior to anthesis and subtend one to several flowering heads. As with many groups of aquatic plants which are closely related to extant terrestrial groups, generic segregation of Megalodonta has been based largely on morphological adaptations to the aquatic environment. Despite its treatment as a distinct genus in recent monographs (Sherff, 1937* 1955)* many authors of local and regional floras have disputed the segregation and continue to treat the taxon as a member of the genus

Bidens (Cronquist, 1955* Steyermark, 1963* Scoggan, 1979)* The purpose of this paper is to present new information on the cytology, morphology, and reproductive biology of Megalodonta beckii, and to provide information pertinent to its taxonomy and segregation as a genus distinct from Bidens. Information on the flavonoid chemistry of Megalodonta has already been, presented (Roberts, 1980). 144

METHODS AND MATERIAIS

Field observations and collections were made over a four year period (many of the localities * have been cited in Roberts, 1980). The chromosome counts and desription of the turions are based on a collection from Meilleurs Bay of the Ottawa River, Renfrew District, Ontario, Canada, (Roberts ft433. 20 Sep 1980, OS), Observations on pollinators were made on a population in the Dead Stream River, Missauke County, Michigan, (Roberts 4894-, 20-24- Aug 1978, OS). Plants have been grown continuously in aquaria in a muck substrate overlain by 5 cm of sand. Plants were grown under flourescent lights with l6-hr * days. .Chromosome preparations were made from adventitous roots pretreated in 0,5$ colchicine and saturated 8-hydroxyquinoline at room temperature for 3 hr. Root tips v/eire fixed in 3 parts ethanols 1 part acetic acid, stained the method of Snow (1963)» and squashed in dilute Hoyer's solution. Seedlings were germinated by soaking on filter paper in petri dishes and removing the fruit wall and testa. 1^5

RESULTS AND DISCUSSION

Vegetative Morphology Individuals of Megalodonta beckii are heterophyllous and the leaves therefore vary on an individual stem and between plants of the same population. The internal vegetative anatomy of the plant has been described by Hoeck (191^)• Only observations on aspects of variability and phenotypic plasticity relating to taxonomic problems are discussed here. Much of the morphological variation is due to the phenotypic plasticity characteristic of aquatic plants which grow at a variety of depths (Sculthorpe, 1967). Differences in size and rigidity of submersed leaves which are apparent under natural conditions quickly disappear when new leaves are formed on ramets transplanted into aquaria. This is particularily true of the rather short rigid leaf forms which develop in lakes with calcareous water. Stems bearing leaves of this type produce larger, flaccid, highly divided leaves under cultivation in softer water. The opposite submersed leaves are borne on flexuous stems at depths of up to 7 m (Ogden et al, 1976). Stems in shallow water often have a few pairs of submersed leaves transitional to the emersed form (Fig. 1). These may be underwater leaves which are only divided a few times and are somewhat laminar, or they may be aerial leaves which are 146 deeply pinnately divided. The intermediate leaf forms are particularly common on plants which have been stranded

on shorelines or in sites where the water level has'receded during the growing season.

The opposite emersed leaves vary considerably in size, shape,, and degree of division. They may be short (1-4 cm), ovate, and nearly entire or much longer (to 10 cm) with a lanceolate shape and more laciniate margin. Occasionally plants in full flower occur without emersed leaves or with only a single emersed leaf (not an opposite pair). The form of the emersed leaves seems to be determined partly, by the depth at which the plant is growing and partly by the lateness of the season at which the emersed portion develops Populations which produce conspicuous emersed leaves in mid-summer may produce only small bracts below the heads later in the flowering season (Hanes & Hanes, 1947). Aerial stems with the emersed leaf form but lacking flowers or flower buds are regularly found in the field and similar plants can be produced in cultivation. The production of the emersed leaf form is apparently developmentally independent of floral induction. 147 Turions

Many perennial aquatic plants of temperate climates produce overwintering structures or turions which function as perennating structures and in dispersal after fragmentation (Sculthorpe, 1967). Such structures have not been described for Megalodonta beckii and it was presumed that overwintering was primarily by the horizontal underwater rhizomes. Rhizomes sire often vigorous and well rooted in shallower areas of lakes and rivers, but in deeper water are often poorly developed and may disintegrate late in the growing season. Plants in deep water are usually anchored by adventitous roots (Hoeck, 1914) which may extend downward up to a meter from the stems onto the substrate. Numerous turions of Megalodonta beckii were collected

in late September from a population in a bay of the Ottawa River, Ontario, Canada (fig. 2.). The turions were arcuate, O.5-O.8 cm in diameter, and 2-5 cm long. They were borne on both terminal and lateral shoots, often with 6-10 on a submersed stem segment less than a meter long. The turions were composed of a dense cluster of appressed, imbricated leaves.on stem tips having very shortened internodes. The leaves were stiff and highly cuticularized. The nodes immediately below the turions bore normal, expanded, spreading leaves with highly dichotomous, flexuous leaf segments. Normal stem segments of Megalodonta beckii shoots are highly aerenchymous and float when detached. The turions 148 observed in the field were attached to floating dislodged stems. No abscission layer is formed between the turion and the stem to which it is attached. It is likely that the turions attached to normal stem segments eventually sink after decomposition of the bouyant portions. Under some conditions stems of Megalodonta beckii apparently do not deteriorate in autumn. Boyden and Sheldon (1976) have reported that the plant remains photsynthetically active during winter under ice cover. Turions collected in the field were exposed to various conditions in the laboratory to investigate their germination response. Turions transplated immediately to aquaria with 16 hour light at room temperature (ca. 21° C) immediately began elongation and produced expanded leaves at the stem tip and adventitious roots from nodes of the turion itself. Turions transplanted with 10 hours of light at room temperature did not resume growth and deteriorated. Turions stored in a dark refrigerator at 4° C for 108 days retained their viability and resumed active growth when transplanted to 16 hour light at room temperature. They also deteriorated in aquaria with short days and warm temperatures. Turions frozen at - 6° C in a dark freezer for 108 days were apparently killed. The deteriorated on thawing and did not resume growth in either long or short day conditions. These limited observations suggest that day length may be a controlling factor in releasing turions from dormancy. The observations also indicate that the turions are capable 14-9 of functioning as perennating organs during winter conditions underwater. The development of turions was not observed under any of the conditions of cultivation. The turions of Megalodonta beckii are remakably similar to those produced by Myriophyllum exalbescens (Weber, 1972; Aiken & Walz, 1979)* The similarities were conspicuous at the Ottawa River site where the two species occurred together and both bore turions. The turions of the two species were similar in size, shape, texture, density, and general morphology. The only superficial difference is the form of leaf division in the reduced appressed leaves. The striking convergence in turion form between species of unrelated families is remarkable despite the fact that the two species often grow together and share common ecological preferences throughout their range.

Reproductive Morphology and Biology

The radiate heads of Megalodonta are normally solitary, but additional ones may be produced from axils of emersed nodes. The sterile ray florets vary in number from 6-8 and are conspicuous against the background of the water surface. The disc florets are deeply lobed and spreading, the style tips recurved, the anthers exserted, and pollen production copious. These features characterize outcrossing species of Bidens, while autogamous species do not have this syndrome of characters. Isolated heads in cultivation do not produce 150 fruits, and seed set in natural populations is often low, with only a few achenes maturing on each head. During three days of extensive observation in clear weather in late August, numerous flower visitors were observed in a population in Missauke Lake, Michigan. The visitors consisted primarily of Diptera, Lepidoptera, Odonata, Coleoptera, and Hymenoptera, in decreasing order of frequency. The visitors, with the exception of the Lepidopterans, did not exhibit flower constancy. Most simply used the emersed heads as landing platforms above the water surface. The Hymenoptera were restricted to near-shore areas and did not visit heads in the extensive area of deeper water. It is possible that a paucity of effective pollinators may limit fruit production in offshore areas.

The linear fruits are up to 1.5 cm long and bear J-6 long, retrorsely barbed awns up to b cm long. At anthesis the awns normally equal the flowers in length and are retrorsely barbed except at the very base. During maturation the basal portion of the awns elongate and at maturity they appear to be barbed only near the widely

spreading tips. After anthesis the achenes are denser than water and sink slowly when separated from the head. 151 Cytology

Megalodonta beckii is one of only three of the sixteen North American genera of the subtribe Coreopsidinae for which the chromosome number has not been reported (Stuessy,

1977)* Stuessy (1977) and Solbrig et ,al (1972) have shown n = 12 to be the prevalent number for the tribe. Smith (1975) has argued for a higher base number for the subtribe, with aneuploid reduction series occurring in a number of independent lineages. A somatic chromosome number of 2n = 26 has been determined for Megalodonta beckii from root tip cells Cfig«3). This chromosome number is unusual in the Coreopsidinae, being previously reported only in Heterosperma (Solbrig et al.

1972) and in Coreopsis (Smith, 1975)• In the latter genus the primitive shrubby members have a base number of n = 14, whereas the advance North American herbaceous, species have

numbers of n = 1 3 . The genus Bidens contains several base numbers and ploidy levels, but has no species with a

reported number based on n = 131 The predominantly North American sections Platvcarpaea. Heterodonta. and Meduseae all have a chromosome number based on n = 12 (Butler & Weedon, 1978j Roberts, unpub.) Although cytological information does not suggest alternative generic relationships, it does provide an additional feature supporting the taxonomic separation of Megalodonta from Bidens. A chromosome number of A = 13 for

an additional genus in the Coreopsidinae also supports the 152 proposition advanced by Smith (1975) that a pattern of descending aneuploidy has occurred during the evolution of the group.

Taxonomy Bidens beckii originally was described by Torrey in 1821 with a brief description in Sprengel (1821) on the basis of material from eastern New York. Later, Torrey (18^3) provided a more complete description and illustration in the Flora of the State of New York. The species later provided the basis for a new section, Hydrocarpaea described by Gray (188*0. In distinguishing the section, Gray cited the terete achenes, peculiar awns, aquatic habit, and submersed filiform dissectted leaves. For many years generic placement of the species remained unquestioned despite the misplacement of many species in genera of the subtribe Coreopsidinae (Sherff, 1937). E. L. Greene (1901) segregated the genus Megalodonta from Bidens. and included B. beckii along with two additional newly described species. In the protologue, he mentioned the same essential features as Gray, but also emphasized the retuse cuspidate ray florets and long stout awns. Wiegan, who had monographed the North

American Bidens section Platvcarpaea (Wiegand, 1899)» questioned the erection of the new genus saying "This plant is distinctly a Bidens" (Wiegand & Eames, 1926). The most recent monographer of the genus Bidens. E. E. Sherff, excluded B. beckii from the genus Bidens and described several additional varieties in Megalodonta (Sherff, 1936, 1937, 1938, 1955). 153

The overall appearance of the heads of Megalodonta beckii is similar to Bidens (Sherff, 1955) * "but the outer involucre.differs from most Bidens in being more clearly differentiated from the inner involcre. In Megalodonta the outer involucre is composed of 3-5 ovate to obovate, herbaceous bracts which are shorter than the hyaline inner involucre. In Bidens the outer involucral bracts are often more numerous, longer and not as well differentiated from the inner in color and texture. However, features of the involucre vary considerably within the large cosmopolitan genus Bidens, and this feature does not clearly distinguish

Megalodonta. Most authors (Gray, 1884; Greene, 1901; Sherff, 1937) » have emphasized the unusual awns on the achenes of Megalodonta beckii as a character distinguishing the genus. No single feature distinguishes the achenes other than relative size. The achenes are most similar to some species of the diverse typical section Psilocaruaea in having spreading awns at maturity with only the terminal portions barbed, and being nearly terete with longitudinal grooves. The achenes differ from all Bidens in the length of the awns relative to the achene body. The segregation of Megalodonta as a distinct genus cannot be maintained on the basis of unique qualitative characters, but rather requires a combination of characters. 154 This is typical in the subtribe Coreopsidinae, where delimitation of genera and placement of species has often been problematical (Sherff, 1937)• Floral and fruit morphology, and flavonoid chemistry (Roberts, 1980) clearly indicate the close affinities of Megalodonta are with Bidens. The distinctive chromosome number, unique aquatic habitat and vegetative morphology, and fruit morphology combine to justify generic recognition. Although recognition of Megnl odonta as a genus distinct from Bidens seems justified, the taxonomic entities within the genus are of questionable merit. Sherff (1955) provided a complete list of synonomy for the genus. Sheldon (1894) described the first variant within the genus from a Minnesota collection as forma scissa. The taxon was diagnosed as lacking emergent leaves. Greene's (1901) M. nudata and M. remota are also based primarily on leaf characters. He distinguished M. nudata from eastern New York as having "almost" naked peduncles, with the emersed leaves alternate and partly dissected when present. He described M. remota from Green Lake, Washington, as having collapsing submersed foliage with more numerous emersed leaves and smaller heads. Sherff (1936) later suggested that M... remota appeared conspecific with M. beckii. but described as new M„ beckii var. oregonensis from Klamath County, Oregon. He based this variety on the shorter leaves on the lower portion of the submersed stems. Sherff (1938 ) later described M. beckii var. hendersonii from Lane County, Oregon, based on the absence of emersed entire leaves. In his most recent 155 revision of Megalodonta. Sherff (1955) stated that it was "an essentially monotypic genus" and recognized only the typical variety and his varieties from the Pacific Northwest.

Three of the taxa described within the genus Megalodonta are based essentially on the absence of emersed leaves. The other taxa are based mostly on the nature of the submersed foliage. As has been mentioned, the expression of these features is dependent on a variety of environmental conditions. The variation in foliage morphology in heterophyllous aquatic plants frequently contributes to unwarranted taxonomic recognition (McCulley & Dale, 1961). An examination of the types of all proposed taxa and observation of variation in the field and greenhouse indicate that taxonomic recognition of variation, even at the intraspecific level, is not justified. Although it might be suspected that the disjunct populations from the Northwestern United States could be taxonomically distinct, other authors have suggested they are not native to the area. The origin of this idea apparently dates from Howell's (1903) statement in his Flora of Northwestern America that Megalodonta beckii is "perhaps introduced" in Green Lake, Washington. Later authors have modified and strengthened the statement to "probably introduced" (Cronquist, 1955) and "perhaps introduced by aquatic birds" (Ferris, i960). I have been unable to locate information (on herbarium labels, floristic literature, or personal inquiry) substantiating the introduction of the species into the Northwest. The non-indigenous explanation seems to be a suggestion amplified by repetition. The distribution of Megalodonta is within the area of Pleistocene glaciation in the eastern portion of its range, with the possible exception of a questionable older record labeled "St.

Louis" discussed by Steyermark (1963). The Oregon localities are outside the limits of probable Pleistocene

Glaciation (Detling, 1968), and this area may have served as a survivium during the Pleistocene epoch. The northwestern populations may represent the survivors of a more continuous transcontinental postglacial distribution which has since been disrupted by the development of the Great Plains. The presence of suitable habitats, the discovery of additional populations in southern British Columbia, and the similarity with the geographic ranges of other aquatic plants (Ceska & Warrington, 1976; Ceska

& Ceska, 1980), make it more likely the species is indigenous in the Pacific Northwest. 157

Summary Megalodonta beckii is a perennial aquatic which has submersed capillary foliage and may produce lanceolate emersed leaves. Submersed stems may produce turions composed of a dense cluster of modified leaves. Sexual reproduction is by showy emersed heads adapted for outcrossing. The chromosome number of 2n = 2 6, the distinctive long-awned achenes, and unusual submersed foliage distinguish the species from the closely related genus Bidens. The variation described within the genus is largely due to phenotypic plasticity and heterophylly. Most of the variants can be seen within single populations and do not warrant taxonomic recognition. 158

Figure i. Stem of Megalodonta beckii with segmented leaves, emersed leaves, and a flower at anthesis. j

Figure 1. Figure 2. Turions of Megnlodonta beckii. 16l

lllljllllilllip

HKKUUU n It 01 u tz u i I tK II K U II 11 k«»ICI « U K .ggm**n n n n 61 II II II SI II M It It II *1 It «1 M :. II (1II or i l l II U 01 I 1 I I .... II I t 11 J

Figure 2. 162

Figure 3 . Root tip chromosomes of Megalodonta beckii. 2n = 2 6. The bar equals IOji. Figure 3 . REFERENCES

Aiken, S. G. and K. F. Walz. 1979» Turions of Myriophyllum

exalbescens. Aquatic Bot. 6: 357-363* Boyden, C. W. and R. B. Sheldon. 1976. Submergent macrophytes: growth under winter ice cover. Science 19/+: 841-842.

Butler, M. G. and R. R. Weedon. 1978. Chromosome analysis of Bidens coronata (Compositae). Can. J. Genet. Cytol. 2 0 : 147-149. Ceska, A. and 0. Ceska. 1980. Additions to the flora of British Columbia. Can. Field-Nat. 94:69-74. Ceska, A. and P. D. Warrington. 1976. Myriophyllum farwellii (Halorogaceae) in British Columbia. Rhodora 78:75-78

Cronquist, A. 1955* Compositae. In:C. L. Hitchcock, A. Crnnquist, M. Ownbey, and J. W. Thompson. Vascular Plants of the Pacific Northwest. Vol. 5* Univ. of Washington Press, Seattle, Wash., 343 PP*

Detling, L. E. 1968. Historical background of the flora of the Pacific Northwest. Univ. Oregon Mus. Nat. Hist.

Bull. No. 13, 52 pp.

Ferris, R. i960. Compositae. In: An Illustrated Flora of Pacific States, Washington, Oregon and California. Vol. 4. Bignoniaceae to Compositae. Stanford Univ. PreSs,

Stanford, Calif. 732 pp. Gray, A. 1884. Synoptical Flora of North America. Vol. I

pt. 2 . 474 pp. Greene, E. &. 1901. Generic rank for Bidens beckii, Torr.

Pittonia 4: 270-272. Hanes, C. R. and F. N. Hanes. 1947i Flora of Kalamazoo County, Michigan. Vascular Plants. (Authors)

Schoolcraft, Mich. 295 pp. Hoeck, A. V. 1914. The anatomy of Megalodonta beckii. Am. Midi. Nat. 3:336-342.

Howell, T. 1987. A Flora of Northwest America, I. Phanerogamae. Portland, Oregon. 792 pp.

MeCully, M. C. and H. M. Dale. 1961 Heterophylly in Hinnuris: a problem in identification. Canad. J. Bot. 391099-1116. Ogden, E. C., J. K. Dean, C. W. Boylen and R. B. Sheldon.

1976. Field guide to the aquatic plants of Lake George, New York. New York State Museum Bull. No. 426, Albany, N.Y.

Roberts, M. I98O. The flavonoids of Megalodonta beckii

and isolation of 2 '-3-dihydroxy-4-methoxy-4’-glucosyl chalcone. Biochem. Syst. Ecol. 8:115-118. Scoggan, H. J. 1979* The Flora of Canada. Pt. 4— Dicotyledoneae. (Loasaceae to Compositae). Natl. Mus. Nat. Sci. Publ. Bot., No. 7» Ottawa, 1711 pp.

Snow, R. 1963. Alcoholic hydrochloric acid-carmine as a stain for chromosomes in squash preparations. Stain

Tech. 3 8 : 9-13. Solbrig, 0. T., D. W. Kyhos, M. Powell and P. H. Raven. 1972. Chromosome numbers in Compositae VIIIi

Heliantheae. Am. J. Bot. 59* 868-878, i Sprengel, K. P. J. 1821. Species plantarum minus cognitae. Neue Entdeck 2t 95-175*

Steyermark, J. A. 1963* Flora of Missouri. Iowa State

Univ. Press, Ames. 1725 pp. Stuessy, T. F. 1977. Heliantheae-systematic review. Ini V. H. Heywood, J. B. Harborne, and B. L. Turner (eds.). The Biology and Chemistry of the Compositae,

pp. 62I-67I. Academic Press, New York. Torrey, J. 1843. A Flora of the State of New York. Vol. 1. Albany, N. Y. 4-84- pp.

Weber, J. A. 1972. The importance of turions in the propagation of Mvrionhvllum exalbescens tHaloragaceae)

in Douglas Lake, Michigan. Mich. Bot. 111115-121. Wiegand, K. M. and A. J. Eames. 1926. The Flora of the Cayuga Lake Basin, New York. Cornell Univ. Agric.

Exp. Sta. Mem. 92, Ithaca, N. Y. 491 pp. APPENDIX A

POPULATIONS ' BIDENS EXAMINED FOR FLAVONOIDS

16? 168

Appendix A. Populations of Bidens examined for flavonoids. Voucher specimens preceded by "R" are those of the author.

Taxon & Collection Number Locality Data B. laevis R4878 AZi Pima Co, 0.2 mi s of Arivaca on rd to Ruby. R4880 AZ: Cochise Co., 2 mi s of town along US Rte. 80. R4-884- NM: Mora Co. , Pond along Mora River at Int. 25* R4978 CTi New London Co., Connecticut River at Hadlyme. R^99^ NY: Greene Co., Hudson River at Cold Spring, R5001 NY: Putnam Co., Hudson River at Cold Spring. R5003 NJi Middlesex Co., Old Bridge. R5016 NJ: Cumberland Co., Maurice River at Old Bridge. R55*KL NC: Carteret Co., between Carteret and Bogue, pond. R5551 NCi Hyde Co., Lake Matamusket along St. Rte 9^. B. cernua R4838 OH: Stark Co., Jackson Twp., Lake Cable. R48*K) OH: Delaware Co. , Hoover Reservoir at Oxbow Rd. R4-873 ID: Kankakee Co. , Kankakee River at Momence. RJj-886 CO: Weld Co. , Lone Tree Creek n of Nunn. R^888 WY: Albany Co. , pond near Vedavoo, Laramie Range. R^890 MI: Bay Co. , Lake Huron shore at Pinconning. R^900 MI: Lelanau Co., Glen Lake. Rij-906 OH: Delaware Co. , Hoover Reservoir at Oxbow Rd. R^908 ONT: Lake Ontario beach at Outlet Park near Picton. R4911 ONT: St. Lawrence River, at Brockville Cemetery. R4955 NBR: Turtle Creek sw of Moncton. R5028 OH: Pickaway Co. , Darby Creek at Mill Rd. B. hvnerborea R4-923 QUE: St. Lawrence River at St.-Antoinne de-Tilly. R4929 QUE: St. Lawrence River at St. Michel. R/j.937 QUE: St. Lawrence River at L'Islet sur Mer R W O QUE: Dartmouth River, nw of St. Majorique. Appendix A (continued). 169 R49^2 QUEt Cascapedia River south of St. Jules. R*J-9^3 QUE 1 East Kempt River at Rte 132. R^9^9 NBRi Southwest Miramichi River, sw of Chatam Head. R^967 ME: Sagadahoc Co., Androscoggin River, Topsham. RZj-970 ME: Lincoln Co. , Kennebec River. F&998 NY: Greene Co., Hudson River at Cosackie. R5150 QUE: Saguenay River at St. Fulgence. R5175 QUE: Restigouche River e of Athoville. R5207 ME: Penobscot Co., Penobscot River s of Bangor. R5210 ME: Sagadahoc Co. , Kennebec River, Merrymeeting Bay. B. frondosa R4868 OH: Franklin Co., weedlot along Neil Ave. R4871 OH: Franklin Co. , Olentangy River, OSU campus. R487^ IL: Kankakee Co. , Kankakee River at Momence. R4885 NM: Mora Co., pond along Mora River at Int. 25* R4887 CO: Weld Co., Lone Tree Creek n of Nunn. R4901 MI: Lelanau Co. , Glen Lake. R4903 MI: Grand Co. , east arm of Traverse Bay. R4905 OH: Delaware Co., Hoover Reservoir at Oxbow Rd. R4907 ONT: Lake Ontario beach at Outlet Park near Picton. R4913 ONT: St. Lawrence River, Brockville Cemetary. R4916 ONT: St. Lawrence River at South Lancaster. R4920 QTJE: St.-Francois River at Drummondville. R4927 QUE: St. Lawrence River at St. Michel. R4938 QUE: St. Lawrence River at St.-Roch-des-Aulnaies. R4951 NBR: Southwest Miramichi River, sw of Chatam Head. R4953 NBR: Turtle Creek, sw of Moncton. R^991 NY: Greene Co. , Hudson River near Athens. R5017 NJ: Cumberland Co., Maurice River at Millville. R50^9 OH: Erie Co., Lake Erie at mouth of Old Woman Creek. R5070 OH: Ross Co., Ohio & Erie Canal near Yellowbud. R5073 WI: Milwauke, Lake Michigan Shore. R5079 WI: Lincoln Co., Wisonsin River at Tomahawk. R5086 WI: Oneida Co., Minocqua, shore of causeway. R5096 WI: Douglas Co., shore of Lake Superior. Appendix A (continued) 170 R5321 BCi Vancouver area, ditch along Lougheed Highway. R5325 BCj Coquitlam, quarry on e side of Pipeline Rd. R5522 NCj Alleghany Co., New River at Fish Camp Rd.

B. discoidea R4842 OH* Pickaway Co. , Calamus Swamp near Circleville. Rij-980 CT: East Hampton Co. , Lake Pocotopaug. R5245 CT: Middlesex Co., pond near Deep River. R5391 WI. Barron Co., Veteran's Park near Cameron.

B. vulgata R4922 QUE: Drummondville, St.-Francois River. R4981 CTs East Hampton Co., Lake Pocotopaug. R5029 OH. Pickaway Co., Darby Creek. R5100 MN. St. Louis Co., Lake Superior at Two Harbors. R5125 NY: Chataqua Co., Lake Erie shore at Barcelona. R5^2 ONT: Ottawa River at Wendover. R5462 QUE: St. Lawrence River at St.-Antoinne de-Tilly.

B. connata R4867 IL: Kankakee Co., Kankakee River at Momence. R4889 OH: Stark Co., Ohio & Erie Canal at Crystal Springs. R4891 MI: Bay Co., shore of Lake Huron at Pinconning. R^902 MI: Grand Co., se shore of Traverse Bay. RU909 ONT: Outlet Beach Provincial Park near Picton. R^975 CT: New London Co., Connecticut River at Hadlyme. R4983 CT: East Hampton Co., Lake Pocotopaug. R5013 NJ: Gloucester Co., Mantua River at Paulsboro. R5014- NJ: Cumberland Co., Maurice River s of Millville. R5052 OH: Ottawa Co., East Harbor, shore of Lake Erie. R5069 OH: Ross Co., Ohio & Erie Canal near Yellowbud. R5072 OH: Pickaway Co., Calamus Swamp near Circleville. R5085 WI: Oneida Co., Minocqua, rocky shore of causeway. R5095 WI: Douglas Co., Lake Superior, Barker's Island. R511^ WI: Waxhburn Co., Shell Lake, s of Spooner. R5129 NY: Jefferson Co., Lake Ontario, Sackets Harbor R5208 ME: Penobscot Co., Dorthea Dix Park on Rte 1A. Appendix A (continued) 171 R5353 OH: Ottawa Co., Marblehead, Quarry Rd. R5361 OH: Hocking Co., Hocking Valley Canal, Nelsonville. R5381 WI: LaCrosse Co., Mississippi River at French Island. R5^29 ONT: Lake Nipissing, shore near Callander.- R5^87 NY: Cayuga Co., Cayuga Lake, Hibiscus Harbor. R5489 OH: Ashtabula Co., Lake Erie near Conneaut. R5506 OH: Logan Co. , Indian Lake, Long Island. R5521 NC: Alleghany Co., New River at Fish Camp Rd. B. connata var. pinnata R5102 WI: Douglass Co., Halfway Lake s of Gordon. R5106 WI: Washburn Co. , Silver Lake near Lampson. R5398 WI: Washburn Co., Shell Lake, s of Spooner. R5*K)0 WI: Washburn Co., Silver Lake near Lampsoon.

B. comosa R^876a IL: Kankakee Co., Kankakee River at Momence. R4904 OH: Delaware Co., Hoover Reservoir, Oxbow Rd. R4921 QUE: St.-Francois River at Drummondville. R5027 OH: Pickaway Co., Darby Creek. R5044 OH: Delaware Co. , Hoover Resevoir at Oxbow Rd. R5298 IN: Marion Co., Indianapolis, Post Rd. R5302 MO: St. Charles Co., Missouri River. R5*<46 ONT: Ottawa River, Carillon Provincial Park. R5520 NC: Alleghany Co. , New River at Fish Camp Rd.

B. heterodoxa var. agnostica Jausson s. n. CT: Middlesex Co., Lake Pocotopaug. B. heterodoxa var. monardifolia Jausson s. n. CT: Middlesex Co., Lake Pocotopaug. B. eatoni R4926 QUE: St. Lawrence River at St. Antoinne de-Tilly. R^-928 QUE: St. Lawrence River at St. Michel. R4936 QUE: St. Lawrence River at L'Islet sur Mer. R4-950 NBR: Southwest Miramichi River near Quarryville. Appendix A (continued) 172 RA971 ME« Lincoln Co., Kennebec River e of* Richmond. RA977 CTi New London Co., Connecticut River near Hadlyme. RA999 NY1 Greene Co., Hudson River at Cosackie. R5005 NJi Middlesex Co., Old Bridge. R5221 MAt Essex Co., Merrimack River w of Amesbury. B. bidentoides var. bidentoides

R4990 NY 1 Greene Co., Hudson River at Athens. RA995 NYt Greene Co., Hudson River at Cosackie. R5002 NYi Putnam Co., Hudson River at Cold Spring. R5008 NJi Burlington Co., Delaware River at Beverly. R5012 NJ« Burlington Co., Delaware River at Delanco. R5020 NJi Cumberland Co., Maurice River s of Milleville. B. bidentoides var. mariana R5022 MDi Cecil Co., Chesapeake Bay, Bohemia River. R502A MDi Cecil Co., Chesapeake Bay, Sassafras River. R5125 MDi Cecil Co., Chesapeake Bay, Charlestown.