. ASPECTS OF THE REPRODUCTIVE BIOLOGY, BREEDING ~ ' \ SYSTEM AND HORTICULTURAL IMPROVEMENT OF THE GENUSPANDOREA

submitted by Elizabeth Ann James B.Sc.(Hons) (LaTrobe)

A thesis submitted in total fulfilment of the requirements for the degree of Master of Science

School of Botany

University of Melbourne Parkville, Melbourne, 3052

April, 1992 pandorana (GS) ii

DECLARATION

Except where reference is made in the text of the thesis, this thesis contains no material published elsewhere or extracted in whole or in part from a thesis presented by me for another degree or diploma. No other person's work has been used without due acknowledgement in the main text of the thesis. This thesis has not been submitted for the award of any other degree or diploma in any other tertiary institution.

.. .\.~: ..Lt.92-

E. A. James 111

ACKNOWLEDGEMENTS

The experiments in this thesis were carried out at the Institute of Sciences (IPS) as part of a research program of the Department of Food and Agriculture (DoPA).

I would like to thank the following people for their support and guidance:

My supervisor, Prof. R. Bruce Knox, Botany School, and Dr. David Beardsell, Dennis Richards and Bill Thompson (IPS) for their most welcome advice and comments and open door policy. Dr. Graeme Frith, Director of IPS for permission to pursue these studies in the Botany School at the University of Melbourne. Dr. John Reynolds and Peter Franz, Biometricians (DoPA), for statistical advice far beyond the call of duty. Dr. Robert Bertin, College of the Holy Cross, Massachussetts, USA and Dr. Alwyn Gentry, Missouri Botanic Gardens, Missouri, USA for advice and information on the family . Philip Taylor for sharing his expertise in electron microscopy. Drs Gordon Guymer and Bill McDonald, National Herbarium, , Elizabeth Brown, National Herbarium, NSW and Alexander Buchanan, National Herbarium, Tasmania, for their assistance in the location of plant material. Mr. Col Harman, O'Reilly's Guest House, Green Mountain, Qld for leading me to P. baileyana. Peter and Ann Radke (Yuruga Nursery, Tolga), Bruce Gray (CSIRO, Atherton), David Beardsell and Tony Slater for some additional plant material. Dr. Yvonne Fripp and Harald Hozmec, Department of Genetics and Human Variation, LaTrobe University, for advice and technical assistance with the isozyme analyses. Francha Horlock and Michelle Bankier for technical assistance and the maintenance of my Pandorea collection. Jocelyn von Kanel and Dr. Doris Blasing for translations of French and German manuscripts. In addition to those 1nentioned above, I would like to thank my colleagues at IPS and the Botany School for their friendship and encouragement especially , Trish Grant, Roger Ashburner, Vijay Kaul and Jan1es Hutchinson. My family, especially Gavan, Davis and Arlie, thanks tean1! iv

REASONS FOR THIS STUDY

The ornamental plant industry relies on the introduction of new products with unusual or novel attributes. Conventional plant breeding techniques in conjunction with mutation breeding and the selection of natural mutations have been the major sources of new material for the ornamental plant industry.

The Australian tlora has 1nuch potential in the ornarnental plant industry and is being used increasingly by both the cuttlower industry and the nursery industry. The majority of material sold on local and export markets is unirnproved although selection and plantation production of material is increasing. The Pandorea has a small local market for two , P. jasminoides and P. pandorana, in the landscape industry. The former species has large flowers but very little intraspecific variation. The latter species has smaller flowers but considerable variability in both flower size and colour. The two remaining Australian species, P. baileyana and P. nervosa are not in commercial horticulture. The potential to produce elite 1naterial for the horticultural and landscape industries through a breeding program within the genus is good. Interspecific hybrids have been produced in the family between species in the tribe Tecomeae to which the genus Pandorea belongs.

This study aims to study aspects of the reproductive biology and breeding system of the genus Pandorea, under experimental conditions, and to develop techniques for the production of interspecific hybrids from selected material. v

ABSTRACT

The Australian flora represent a potential genetic resource for the production of new cultivars for the local and international ornamental plant industry. The genus Pando rea contains four species in Australia. All are clitnbers with a range of flower colour and size. This study aimed initially to collect a range of genotypes from the Australian species of Pandorea. It then sought to identify and study aspects of the reproductive biology and breeding system of the genus Pandorea. It also aimed to cross species using the basic techniques of plant breeding to provide unique genetic combinations as a basis for new horticultural varieties of Pandorea.

Floral dimensions were xneasured and con1pared for all species. The reproductive biology was indicative of a genus comprising obligate outcrossing individuals. Strong self-incompatibility was found in three species. The fourth was not tested. The stigrnas were receptive prior to anthesis. Pollen viability deteriorated over a five day period.

Interspecific hybrid seedlings were obtained for some crosses through embryo culture. Embryos aborted if left on the parent plant apparently due to endospenn failure. The success of rescuing interspecific hybrid etnbryos was related to the stage at which embryo develop1nent was arrested. Well-developed cotyledonary einbryos were readily grown in aseptic culture and were acclitnatized with ease fron1 tissue culture. Less mature embryos gern1inated precociously and failed to continue through the normal embryo developrnent.

Isozyme analyses were useful for corroborating the hybrid status of so1ne putative hybrid seedlings produced in this study.

One hundred and seventeen confirmed interspecific hybrids and seven unconfinned hybrids have been acclimatized to standard nursery conditions. They are anticipated to flower for the first time in the 1994 flowering season when their floral characteristics can be assessed for horticultural potential. vi

TABLE OF CONTENTS

Page

CHAPTER 1: REPRODUCTIVE BIOLOGY, FLORAL STRUCTURE AND PHYLOGENY OF THE BIGNONIACEAE ...... 1

1.1 SYSTEMATICS AND EVOLUTION ...... 1 1.1.1 Evolution and phylogeny ...... 1 1.1.2 Plant and Flower Morphology ...... 3 1.1.2.1 Habit ...... 3 1.1.2.2 ...... 4 1.1.2.3 Indument ...... 4 1.1.2.4 ...... 6 1. 1. 2. 5 Calyx ...... 7 1.1.2.6 Corolla ...... 7 1.1.2.7 Stamens ...... 9 1.1.2.8 Pistil ...... 9 1.1.2.9 Disk ...... 10 1. 1 . 2. 10 Fruit ...... 10 1.1. 2.11 Seeds ...... 10 1.1.1.12 Seedlings ...... 11 1.1. 3 Cytology of fatni1 y ...... 11 1.1.4 Chemical constituents ...... 15 1.2 REPRODUCTIVE BIOLOGY ...... 16 1.2.1 Pollen ...... 16 1.2.1.1 Description of pollen types in the family Bignoniaceae ...... 16 1.2.1.2 Evolution of pollen type ...... 19 1.2.2 Pistil ...... 20 1.2. 3 Embryology ...... 21 1.2.4 Conclusions ...... 22 vii

1.3 REPRODUCTIVE BIOLOGY IN THE FAMILY BIGNONIACEAE ...... 24 1.3.1 Pollination biology ...... 24 1.3.2 Influence of pollination on fruit set ...... 26 1.3.3 Self-incompatibility ...... 26 1.3.4 Interspecific hybridization ...... 28 1.3.5 Conclusions ...... 29 1.4 AUSTRALIAN BIGNONIACEAE ...... 29 1.5 GENERAL CONCLUSIONS ...... 31

CHAPTER 2: MATERIALS AND METHODS ...... 32

2.1 PLANT MATERIAL USED IN THIS STUDY ...... 32 2.1.1 Reproductive biology and breeding system ...... 32 2.1.2 Tissue culture ...... 32 2.2 FLORAL DIMENSIONS ...... 35 2.3 BREEDING SYSTEM ...... 36 2.3.1 Reproductive structures ...... 36 2.3.1.1 Pistil ...... 36 2.3.1.2 Pollen ...... 36 2.3.2 Pollen-ovule ratio ...... 38 2.3.3 Pollen-tube growth in selfed and outcrossed flowers ...... 38 2.3.4 Separation of male and female function ...... 38 2.3.5 Self-incon1patibility, genotype and reciprocal effects ...... 39 2.4 INTERSPECIFIC HYBRIDIZATION ...... 39 2.4.1 Pollinations ...... 39 2.4.2 Embryo rescue ...... 39 2.4.3 Fruit 1neasurements and seed germination ...... 41 2.4.4 Isozy1ne analysis of putative hybrids ...... 41 2.5 STATISTICAL ANALYSES ...... 44

CHAPTER 3: THE GENUS PANDOREA ...... 45

3.1 DESCRIPTION OF THE GENUS ...... 45 Vlll

3.2 FLORAL MORPHOLOGY ...... 46 3.2.1 Inflorescence ...... 46 3.2.2 Flower structure ...... 46 3.2.3 Floral dimensions ...... 52 3.3 THE AUSTRALIAN SPECIES OF PANDOREA ...... 52 3.3.1 Key to the Australian species ...... 52 3.3.2 Description of the Australian species of Pan.dorea ...... 60 3.4 COMMERCIAL FORMS OF PANDOREA ...... 62 3.5 SUMMARY ...... 62

CHAPTER 4: SOME ASPECTS OF THE REPRODUCTIVE BIOLOGY OF TI-lE GENUS PANDOREA ...... 64

4.1 INTRODUCTION ...... 64 4.2 RESULTS AND DISCUSSION ...... 64 4.2.1 Reproductive structures - pistil ...... 64 4.2.1.1 Pistil structure ...... 64 4.2.1.2 Stigma receptivity ...... 68 4.2.1.3 Separation of tnale and female function ...... 69 4.2.1.4 Pollen-pistil interactions in selfed and outcrossed flowers . . . . . 71 4.2.2 Reproductive structures - pollen ...... 74 4.2.2.1 Pollen viability ...... 74 4.2.2.2 Pollen structure ...... 77 4.2.3 Pollen-ovule ratios ...... 79 4.3 SUMMARY ...... 81

CHAPTER 5: ASPECTS OF TI-lE BREEDING SYSTEM OF TI-IE GENUS PANDOREA ...... 90

5.1 INTRODUCTION ...... 90 5.2 QUANTITATIVE ANALYSIS OF SELF-INCOMPATIBILITY, GENOTYPE AND RECIPROCAL EFFECTS ...... 90

5 .2.1 Method of Analysis ...... ~ ...... 91 ix

5 .2.2 Results ...... 92

5. 3 DISCUSSION . 0 • • • • • • • • • • 0 • • • • • • • • • • • • • • • • • • • • • • ••••• 98

5.4 SUMMARY ..... 0 ••••••••••••••••••••••••••••••••• 100

CHAPTER 6: INTERSPECIFIC HYBRIDIZATION ...... 101

6.1 INTRODUCTION ...... 0 •••••• 101

6.2 RESULTS AND DISCUSSION 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • 102 6.2.1 Interspecific cross pollinations and pollen tube growth ...... 102 6.2.2 Fruit dimensions and seed gerrnination ...... 108 6.2.3 En1bryo rescue of putative hybrids ...... 115 6.2.4 Isozyme analysis of putative hybrids ...... 120

6.2.4.1 P. baileyan.a x P. nervosa .. 0 ••• 0 • • • • • • • • • • • • • • • • 120

6.2.4.2 P. pandoran.a x P. nervosa ...... 0 • • • • • • • • • • • • • • 121

6.2.5 Tissue culture ...... 0 •• 0 ••• 0 • • • • • • • • • • • • • • • • • • 125 6.3 SUMMARY 127

CHAPTER 7: CONCLUSIONS AND FUTURE PROSPECTS ...... 0 • 128

APPENDIX Selections resulting from the study . 0 0 0 • • • • • • • • • • • • • 132

REFERENCES • 0 • 0 0 •••••••••••••••••••••••••••••••••• 134 X

LIST OF TABLES Page

1.1 Distribution of tribes in the family Bignoniaceae. . ... 2 1.2 Examples of species and genera with particular inflorescence types. . 6 1.3 Chromosome numbers found in each tribe in the family Bignoniaceae. 13 1.4 Classification of pollen types by Gomes ( 1955) based on the exine details of fresh pollen...... 18 1.5 Summary of flowering phenologies and associated pollinators found in Central American Bignoniaceae...... 25 1.6 Records of self-incompatibility in the family Bignoniaceae...... 27 1. 7 Species of the family Bignoniaceae which occur naturally in Australia. 30 2.1 Description and source of genotypes of P. pandorana, P. jasmin.oides, P. n.ervosa and P. baileyana...... 34 2.2 Components and their concentrations in basal medium for the establishment of Pandorea aseptic culture...... 35 2.3 Four media used initially for the culture of immature putative hybrid embryos of Pandorea...... 40 2.4 Components and their concentration in basal n1edium used for embryo rescue...... 40 2.5 Enzyme systen1s initially assayed fro1n parents and their putative hybrids...... 42 2. 6 Enzyme extraction buffer used for Pando rea samples undergoing isozyme analysis...... 43 3.1 Species of Pandorea. 46 3.2 Floral dimensions for P. pandorana, P. jas1ninoides, P. baileyana and P. n.ervosa...... 58 4.1 Proportion of flowers of P. jasmin.oides (genotypes HW, AL, F827) and P. pandorana with separating stigma lobes and dehisced anthers prior to and at anthesis...... 67 4.2 Pollen germination and growth on different aged stigmas of P. pandorana and P. jas1ninoides, genotypes GAS and HW, respectively...... 69 xi

4.3 Mean time to stigma opening, anther dehiscence and corolla abscision for P. pan.dorana on four separate occasions for genotype GAS (field-grown) and single occasions each for genotypes MM and VC

(glasshouse-grown). . 0 0 ••• 0 •••• 0 ••••••••••••••••••••• 71 4.4 Pollen-tube growth in self- and outcrossed flowers of P. jasmin.oides. 72 4.5 Pollen-tube growth in self- and outcrossed flowers of P. pandorana. 73 4.6 Pollen-tube growth in self- and outcrossed flowers of P. baileyana. 73 4.7 Qualitative ability of pollen of P. jasminoides (genotype HW) and P. pandorana (genotype GAS) of varying ages to germinate and grow to the base of the style. ... 77 40 8 Number of pollen grains per flower , number of ovules per flower and the pollen-ovule ratios for four genotypes of P. jasminoides

and two genotypes of P. pan.dorana. . . . 0 • • • • • • • • • • • • • • • • • • • 79 5.1 Proportion of fruit set and fitted values resulting from various numbers of pollinations carried out in the diallel crosses of P. pandorana...... 94 5.2 Residual deviance, residual degrees of freedom and significance of lack-of-fit between fruit set, resulting from diallel pollinations among four genotypes of P. pandoran.a, and fitted values...... 96 5. 3 Proportion of fruit set and fitted values resulting from various numbers of diallel pollinations between three genotypes of P. jasn1inoides...... 97 5.4 Proportion of fruit set and fitted values resulting from various numbers of diallel pollinations between two genotypes of P. bailey ana...... 98 6.1 Proportion of fruit set two weeks after various nu1nbers of interspecific pollinations using P. pandorana as the fen1ale parent. . . . . 106 6.2 Proportion of fruit set two weeks after various numbers of interspecific pollinations using P. jas1ninoides as the female parent. . . . 107 6.3 Proportion of fruit set two weeks after various numbers of interspecific pollinations using P. baileyan.a as the fe1nale parent. . . . . 108 6.4 Proportion of fruit set two weeks after either intra- or interspecific pollinations where P. pandoran.a, P. jasminoides xu

or P. baileyan.a were used as female parents...... 108 6.5 Fruit dimensions and percentage seed germination for intra- and interspecific crosses of P. jasn1inoides...... 110 6.6 Fruit dimensions and percentage seed germination for intra- and interspecific crosses of P. pandorana. .... 111-112 6. 7 Fruit dimensions and percentage seed germination for intra- and interspecific crosses of P. baileyana...... 113 6. 8 Proportion of embryos from P. jasminoides x P. nervosa alive after 14 days in culture (Proportion transferred to 1% sucrose/Gelrite medium)...... 116 6. 9 C01nparison of four different ex of P. ja.. wninoides on the number of shoots, total shoot height, total number of nodes and the number of nodes per meriste1n. . . . . 126 Xlll

LIST OF FIGURES

Page

1.1 Examples of variations in leaf type in Tabebuia and Tecon1a. 5 1.2 Examples of corolla morphology in the family Bignoniaceae. 8 1.3 Possible pathway to the different chromosome nurnbers found in the family Bignoniaceae...... 14 2.1 Diagram of the four different explants of P. jasn1inoides used for tissue multiplication. 33 3.1 The Australian species of Pandorea. 47 3. 2 Distribution maps for the four Australian species of Pando rea...... 48-51 3.3 Mature fruit of P. jasminoides showing the arrangement of winged seeds (s) and the position of the septum perpendicular to the plane of (c) dehiscence. ... 53 3.4 Stigma of P. jasn1in.oides stained with toluidine blue showing the two lobes, the lower one recurved, covered in dark-staining papillae. . .. 53 3.5 Two of the genotypes of P. jasminoides (HW, AL) used for the Jneasuretnent of floral din1ensions. . ' ...... 54 3. 6 Three of the genotypes of P. pandorana (BF, TS 1, TS4) used for the measurement of floral dimensions...... 55 3. 7 Genotype of P. nervosa used for the rneasurement of floral dimensions. . 56 3.8 Side views of P. baileyana (2, 3) compared toP. pandorana(VC) and P. nen,osa...... 56 3.9 Floral dimensions measured for genotypes of each species of Pandorea. 57 3.10 Plot of corolla tube length .against corolla outside width for all genotypes used in floral din1ensions measurements showing the separation of species...... 59 3.11 Key to the Australian species of Pandorea...... 63 4.1 Transverse section through the style half-way between the stigma and the ovary showing the central canal (c) lined with epithelial cells (arrowed), vascular bundles (v), single-layered epidermis (e) and multicellular glands (g)...... 66 xiv

4.2 Flower of P. jasn1in.oidesWS 1 at an thesis with numbered in order of opening...... 66 4. 3 Scanning electron micrograph of an aged stigma of P. pandorana with recurved lobes and papillae (arrowed) on the receptive inner surface...... 66 4.4 Probability of pollen grain fluorescence of pollen from different aged flowers following staining with fluorescein diacetate. a. P. pandorana. b. P. jasminoides...... 75 4.5 Light micrograph of pollen from P. jasmin.oides flowers, three days after anthesis, germinating on a receptive stigma...... 76 4. 6 Light micrograph of an outcross pollen-tube of pollen, from P. jasmin.oides flowers three days after anthesis, entering an ovule...... 76 4. 7 Tricolporate pollen grains of Pandorea species showing finely reticulate exine (e), position of colpi (c), colpus tnembrane (em), and ruptures in the colpus membrane (r)...... 83-84 4. 8 Transmission electron tnicrograph of a pollen grain of P. jasmin.oides(HW) showing an exine structure comprising a tectum (t), columellae (co), foot layer (f) and a double-layered intine (i1,i2), the outer layer (i1) being more electron dense. .... 85 4. 9 Scanning electron micrograph of whole fixed pollen grains showing ruptures of the col pus metnbranes of a. P. pandoran.a(GS) and b. P. n.ervosa...... 86 4.10 Scanning electron micrograph of unfixed, air-dried pollen grains with pollencoat tnaterials visible within the columellae. a. P. baileyC!na b. P. nervosa...... 86 4.11 Scanning electron micrograph of starch grains (arrowed) present in the mature pollen grains of P. ja.s1ninoides...... 87 4.12 Light micrograph of bicellular pollen of P. jasmin.oides(HW), stained with DAPI, showing the two nuclei. Generative cell (arrowed) fluoresces 1nore strongly than the vegetative nucleus (x400). Inset: whole pollen grains stained with toluidine blue with generative cell (gc), in transverse plane, the tnore diffuse vegetative nucleus (vn) and nucleolus (n) (xlOOO). . 87 XV

4.13 Transmission electron micrograph of the vegetative nucleus (vn) and nucleolus (n) with a transverse section through the generative cell (gc) of a pollen grain of P. jasn1inoides...... 87 4.14 Transmission electron micrograph of the generative cell in a pollen grain of P. jasmin.oides(HW) showing the nucleolus (n), nucleus (nu), cytoplasm (c) and generative cell wall (arrowed). .. 88 4.15 Transmission electron micrograph of microtubules (arrowed) lying in axial orientation in the generative cell...... 88 4.16 Transmission electron micrograph of microfilaments (arrowed) in the vegetative cell cytoplasm...... 88 4.17 Transmission electron micrograph of P. jasmin.oides illustrating the close association of the generative cell (gc) in the vegetative cell cytoplasm (cy) with the pollen wall (i=intine, e=exine). Generative cell wall is visible (arrowed)...... 89 5.1 Flow diagram showing alternative model pathways depending on whether the bulk of outcrosses is considered to result in a common outcome or in discrete and varied results. ... 93 6.1 Light micrographs of pollen tube growth in self-pollinated pistil of P. baileyan.a(3). a. pollen germination on the stigma (x400) b. pollen tubes present in midMstyle (x400) c. pollen tubes at base of style and traversing ovules (x 100)...... 103 6.2 Light micrographs of pollen tube growth in intraspecific outcross-pollinated pistil of P. baileyana(3). a. pollen gern1ination on the stigma (x400) b. pollen tubes present in n1id-style (x 100) c. pollen tubes at base of sty Ie and traversing ovu Ies (x 100). . . . . 103 6.3 Light micrographs of pollen tube growth in interspecific outcross-pollinated pistil of P. baileyana(2) x P. pan.dorana(SCH). a. pollen gennination on stigma (x400) b. pollen tubes present in mid-style (x400) c. pollen tubes at base of style and traversing the ovules (x400)...... I 03 6.4 Light micrograph of entry of P. nervosa pollen tube into micropyle (arrowed) of P. baileyana(2). pt=pollen tube, o=ovule, h =hypostase (x400)...... 104 XVl

6.5 Comparison of seed from interspecific (P. jasminoides x P. nervosa) and intraspecific (P. jasminoides) pollinations...... 114 6.6 Success of culturing embryos taken from interspecific fruit from the cross P. baileyan.a x P. nervosa. a. proportion alive after 14 days culture in each of four media. b. proportion of embryos which had grown sufficiently to be transferred to a germination medium. 118 6. 7 Well-developed cotyledonary P. baileyana(2) x P. n.ervosa embryo. Inset: less well-developed cotyledonary embryo. c=cotyledon, r=radicle. . . . . 119 6.8 Immature embryos showing precocious germination. c=cotyledon, rh =root hairs...... 119 6.9 Schematic representation of the zymograms of the phosphoglucomutase loci in P. baileyan.a (Pb), P. nervosa (Pn) and putative hybrid seedlings (numbered). . ... 122 6.10 Schematic representation of the zymograms of the leucine amino peptidase loci in P. baileyana (Pb), P. nervosa (Pn) and putative hybrid seedlings (nuinbered)...... 123 6.11 Schematic representation of the zymogram of glucose phosphate isomerase loci in P. pandorana (Pp), P. nervo.sa (Pn) and putative hybrid seedlings (nu1nbered)...... 124 6.12 Schematic representation of the zytnogram of shiki rnic acid dehydrogenase loci in P. pandorana (Pp), P. nervosa (Pn) and putative hybrid seedlings (numbered). . . . . 124 CHAYfER 1

REPRODUCTIVE BIOLOGY, FLORAL STRUCTURE AND PHYLOGENY OF THE FAMILY BIGNONIACEAE

1.1 SYSTEMATICS AND EVOLUTION OF THE FAMILY BIGNONIACEAE

1.1.1 Evolutionary Concepts and Phylogeny The family Bignoniaceae was first characterized in 1719 by Tournefort using Bignonia cap reo lata as the generic lectotype (Gentry, 1972). The fa1nily comprises 113 genera and approximately 800 species divided into 8 tribes (Gentry, 1979). Five of the tribes are confined to the neotropics (Bignonieae, Crescentieae, Eccromocarpeae, Schlegelieae and Tourrettieae). The tribe Tecomeae, to which the genus Pandorea belongs, occurs in Australia, Asia, India, east Africa and north and south America with the majority of species in the Americas. The tribe Oroxyleae is found in tropical Asia and Coleeae occurs in Africa and Madagascar (Cronquist, 1981) (Table 1.1, p2).

The Bignoniaceae are at least of Tertiary age (van Steenis, 1927) and are probably close to the basal stock of the Tubiflorae (Goldblatt & Gentry, 1979). The vast majority of species are woody, a character considered to indicate pri1nitiveness. The fossil record can be traced with some uncertainty to the upper Cretaceous (Darrah, 1939) and flowers and fruit of a species of Catalpa, one of the most advanced genera of the family, are well-preserved in the Eocene London Clay (Paclt, 1952), whilst the oldest fossil record accepted for the family is for Bignonicapaula formosa from the Eocene of Tennessee (U.S.A.)(Harland, 1967). Fossil pollen of Teco1na and Jacaranda has been identified frotn the Oligocene San Sebastian formation of Puerto Rico (Graham and Jarzen, 1969). Neuy-Stolz (1958)(cited in Buunnann, 1977) claimed that the pollen of Tetradopollenites quatuor from the Upper Miocene of West Gern1any is Catalpa pollen. However, detailed c01nparisons of the fossil pollen with conte1nporary genera are required before they can be assigned to those genera because the fossil pollen samples have characteristics atypical of the extant genera to which they have been assigned (Buurmann, 1977). 2

Table 1.1 Distribution of tribes in the family Bignoniaceae.

TRIBE DISTRIBUTION

Bignonieae North, South America

Tecomeae Australasia, Asia, India, east Africa, North & South America

Crescentieae tropical, central America

Schlegelieae Columbia, southern central America

Tourrertieae Length of Andes in south America, north in central Cordilleras to Mexico

Eccremocarpeae Central Andes

Oroxyleae Tropical Asia

Coleeae Africa, Madagascar

(after Gentry, 1980)

The family probably has a monophyletic origin. .The tribes and species are closely related and too sharply separated from related families for a polyphyletic origin to be seriously considered (van Steenis, 1927). There are, however, a few problem genera such as Paulown.ia which show affinity with the Scrophulariaceae (Armstrong, 1985; Cronquist, 1981; Goldblatt & Gentry, 1979). Mitra (1968) suggested a polyphyletic origin on the basis of pollen diversity and cytological heterogeneity. However, a comprehensive cytological study (Goldblatt & Gentry, 1979) and pollen study (Gentry & Tomb, 1979) provide convincing arguments for the developrnent of the Bignoniaceae from a single ancestral stock. Although the pollen types within the family are heterogeneous, most pollen types have evolved independently in several evolutionary lineages occurring for exan1ple in both major tribes (Bignonieae and Tecomeae) and in unrelated genera. The finely-reticulate tricolpate pollen type has been inferred as the ancestral or basal type from both its broad distribution across all tribes and geographic regions (Gentry and Tomb, 1979) and its prevalence in taxa regarded as less specialized on morphological grounds. In addition, in the family as a whole, it is found most 3 commonly in the tribes Tecomeae and Oroxyleae which are considered morphologically less advanced (Gentry and Tomb, 1979). Cytological examination suggests that despite the different chromosome numbers found in the family, their occurrence is consistent with a single ancestral number of n = 7 with polyploidy and aneuploid addition or loss of chromosomes accounting for the variability currently found in the family (Goldblatt and Gentry, 1979).

The Malayan Bignoniaceae show affinities with the neotropical genera, and have probably originated from them and taken part in a widespread Tertiary flora (van Steenis, 1927). Van Steenis (1927) suggests a migration of the tropical genera in a pattern similar to that shown for many other families. The most likely route was along the Bering Straits Bridge with migration along the Aleutian Islands also possible. Most of the migratory genera belong to the Tecomeae group with one section going east to Malaya, Philippines, and Australia. The other migrated to India and east Africa, perhaps along the north of the Arabian Sea, and directly to east Africa via Sri Lanka, the Mascarenes, Seychelles and Madagascar.

There are many endemic species and genera which often have disjunct distribution. They were probably more widespread in earlier times but have become restricted in distribution with changes in climatic conditions so that the remaining species are relict­ endemics (n1embers of an earlier flora) rather than progressive-ende1nics resulting from recent speciation in their present habHats.

1.1.2 Plant and Flower Mo1phology Most aspects of the plant morphology of the family show a wide range of variation. Many of those variations are useful for delimitation of taxa within the family. A detailed sun1mary of morphological features for the New World taxa of the family Bignoniaceae was made by Gentry (1980).

1.1.2.1 Habit Trees, shrubs, lianes and herbs can all be found in the family Bignoniaceae. All genera of the tribe Bignonieae are exclusively or predominantly lianes. In the tribe Tecomeae, most genera are trees (30 or more metres tall), some are shrubs, a few such 4 as Pandorea and are lianes and two are herbaceous. The tribes Crescentieae and Coleeae are exclusively trees or shrubs. The tribes Tourrettieae and Eccremocarpeae contain herbaceous to wiry whereas species in the tribe Schlegelieae are mainly epiphytic or semi-epiphytic vines. Although differences in habitat are not always taxonomically significant, the dichotomy between lianes 1n Bignonieae and arboreal Tecomeae and Crescentieae is a fun dam en tal one in the neotropics. In general, New World lianes are tendril climbers whereas Old World climbers, lacking tendrils, climb by twining or adventitious roots.

1.1.2.2 Leaves Leaves in the majority of species are compound and opposite, a condition so prevalent that most species of angiosperms with opposite compound leaves belong to the Bignoniaceae. However, there are several genera characterized by simple leaves and a few simple-leaved species or individual variants in normally compound leaved genera. The predominant compound leaves may be 2-3 pinnate, palmately compound or variously hi-compound. A division occurs between Old World and temperate zone Tecomeae which have pinnately compound (or derived simple leaves) and most New World genera with palmately compound leaves (or derived simple leaves) (Fig. 1.1, p5).

Seibert ( 1948) commented that leaves were of limited taxonomic use because of the variation in leaf type even on the same individual, particularly in lianes. Van Steenis ( 1927) recorded a species, Pando rea leptophylla, which was collected in New Guinea and identified by leaf type. However, in his revision of the family for the 'Flora Malesiana', van Steenis (1977), retracts that species as it turned out to be a specimen of P. pandorana with juvenile leaves.

1.1.2.3 1ndwnent The indument can be variable both inter- and intraspecifically and with care can be useful taxonomically. In In any species, plants fro1n arid habitats will show greater pubescence. Gentry ( 1980) has even listed "species" which had been assigned on the basis of the pubescence on herbarium specimens but which later were found to be growing on the same plant, for example, Lundia densifolia DC. and L. stricta Bur. & K. Schum. 5

c. o.. b.

e . .::

Figure 1.1 Examples of variations in leaf type in Tabebuia and . a. Tabebuia chrysantha, b. T. rosea, c. T. uliginosa, d. Tecoma castanifolia, e. Tecoma stans. (Bar = Scm) 6

The type of trichotnes found in the indument is of taxonomic significance for some genera and is generally constant for all parts of the plant. Stellate and dendroid trichomes are a constant character for most species. In Tabebuia, differences in pubescence, especially of the calyx, are taxonomically important and for the yellow­ flowered species of the genus, rather subtle variations between simple, forked, stellate, dendroid and thick-stellate trichomes are major specific determinants. For Anemopaegma, species can be differentiated largely by the prescence or absence of glandular-lepidote trichomes on the corolla tube (Gentry, 1980).

1.1. 2. 4 Inflorescence of Bignoniaceae may be terminal intergrading to axillary depending on species. There is inherent intraspecific variability in Bignoniaceae inflorescences. Some species and inflorescence types are listed in Table 1. 2. Inflorescence pubescence tends to be more constant in a species than that of most vegetative plant parts. Inflorescences tend to be many flowered.

Table 1.2 Examples of species/genera with particular inflorescence types (from Gentry, 1980).

GENUS/SPECIES INFLORESCENCE TYPE

Kige!ia Flagelitlorous On>xylum "Pincushion" Hap!ophragma "Pincushion" Amphitecna breedlovei A. Gentry Tenninal A. costara A. Gentry Cau litlorous A. montana L. 0. Williams Axillary or on small branches Parmenriera millspaughiana L. 0. Williams Terminal, single flower P. trunc{flora Standley & Williains Cauliflorous Crescenria linearifolia Miers Axillary, single flower 7

1.1.2.5 Calyx The calyx is usually cup-shaped but truncate, 5-denticulate, bilabiate or spathaceous calyces are common. Variations range from the thin foliaceous calyx of some species of Jacaranda (Gentry, 1980) to an unusual "double" calyx with a thin frilly sub-marginal rim in Amphilobium, Haplolophium and Glaziovia (Gentry, 1980). The value of calyx type for taxonomic purposes depends on the genera and species under study because variation within species and genera and between genera is not constant for the family as a whole.

1.1. 2. 6 Corolla Flowers of the Bignoniaceae show 1nuch morphological differentiation (Fig. 1.2, p8) whilst still sharing the cotntnon theme of the sympetalous Tubiflorae flower (Gentry, 1980). Corolla length varies fron1 0.4 ctn in many Tynanthus species to 35 em in Tanaciwn jaroba Sw. Corolla shape varies from narrowly tubular as in Schlegelia sulfurea Diels to the broadly catnpanulate flowers of Kigelia and Spathodea (Gentry, 1980). Flower size, shape and colour is correlated with pollination syndrotnes and is described in some detail by Gentry (1974a). Most species have variously tubular­ campanulate to tubular-infundibuliform corollas, usually slightly bilabiate with a pair of longitudinal ridges in the t1oor of the tube and the upper two of the five corolla lobes slightly stnaller than the lower three. Most species of Crescentieae differ in having much reduced corolla lobes and frequently a horizontal fold across the floor of the corolla. Although some genera can be differentiated on the basis of subtle differences in corolla shape, in others such as Tabebuia, t1owers of different species vary to include all range of colours and shapes, with subsequent diverse pollination syndromes (Gentry, 1980).

There is a basic dichotomy between purple or pink flowers and yellow flowers. Genera which have purple-t1owered species have no yellow flowered species and vice versa (Gentry, 1980). There are humtning-bird pollinated genera and species related to both yellow-flowered and purple-flowered groups. White flower colour has less significant taxonotnic significance being found in otherwise yellow-flowered and otherwise purple-flowered genera. Whilst some genera such as have white- , yellow- and pink-flowered species, the relationship between flower colour and pollinator is unknown. 8

b.~

Figure 1.2 Examples of corolla morphology in the family Bignoniaceae. a. Tynanthus croatianus, b. Arrabidaea candidans, c. A. verrucosa (calyx removed), d. Pithetoctenium crudgerum e. Pandorea ja.sminoides, f. Pandorea nervosa, g. , h. Pandorea baileyana, i. Deplanchea tetraphylla. (Bar = lcm). 9 Corolla pubescence can be important in differentiating both genera and species, but the presence of glandular trichomes on the corolla at the level of stamen insertion are too constant to be of significance in most groups, although the length of trichomes inside the corolla is useful in separating species of Tabebuia (Gentry, 1980). Corolla texture, whether it is thick or thin, can be used to distinguish some genera and can also be correlated with pollen vector (Gentry, 1980).

1.1.2. 7 Stamens Stamen number is a useful character in defining generic boundaries. Most species have four didynamous stamens. Catalpa, (Tecomeae) and Pseudocatalpa (Bignonieae), however, have only two stamens whereas several Old World genera including Oroxylwn and Rhizogum have five sta1nens (Gentry, 1980). Most Bignoniaceae have bithecate anthers but one theca is lost in the section Monolobos of Jacaranda. The thecae may be parallel, divergent or divaricate. The fifth stamen is usually replaced by a small vestigial and posterior staminode, although in Jacaranda and Digomphia, the staminode is greatly elaborated and larger than the fertile stamens. A few genera and species are characterized by the absence of even a vestigial staminode (Gentry, 1980). Crud en (1977) found that for pollen ovule ratios, the number of pollen grains varies between species depending on whether they are inbreeding or outbreeding, but the number of ovules for individual species remains fairly constant.

1. 1. 2. 8 Pi sri l The bilan1ellate stigma and elongated style are remarkably uniform for most species of Bignoniaceae. The stigma lobes are wet-papillate (Heslop-Harrison and Shivanna, 1977; Heslop-Harrison, 1981) and are sensitive to physical stimulus (Newcombe, 1922; 1924). The stigma will be considered further in section 1.2. In contrast, ovary shape provides useful taxonomic characters. Ovary shape also tends to influence fruit shape and has been suggested as a guide for fruit shape for species where fruit is still not known (Gentry, 1980). The ovaries are usually bilocular with a separating septum and axile placentation. In genera with axile placentation, each locule has two parallel longitudinal placentae. Because of the arrangement of ovules, it is often unclear whether a placenta has two series of widely spaced ovules or one series of closely spaced ones. Number of 10 ovules has been used as a major specific determinant but is highly variable within genera (Gentry, 1980).

1.1.2.9 Disk In most Bignoniaceae, a nectar-producing disk surrounds the ovary. It is usually annular-patelliform in shape but is absent in a few genera including Clydista and Clyanydia (Gentry, 1980). Loss of the disk is correlated with intermittent flowering phenology and pollination by deception (Gentry, 1980). Some genera with poorly developed disks have a dense ring of glandular trichomes surrounding the ovary (Gentry, 1980).

1.1. 2.10 Fruit Fruiting characteristics are fundamental to the subdivision of the family. Tecomeae and Bignonieae have dehiscent, usually 2-valved capsules with the seeds borne on a septum. In Tecomeae dehiscence is perpendicular to the septum and in Bignonieae it is generally parallel to the septum (Gentry, 1980). Indehiscent fruits characterize Crescentieae and Schlegelieae (New World) and Coleeae (Madagascar) plus Kigelia of continental Africa. The fruit of the single species of Tourrettia is a 4-valved capsule which does not dehisce cotnpletely. The only genus of Eccremocarpeae has thin-walled ovoid 2-valved capsules without a septum.

At the generic level, fruits are also of great taxonon1ic importance (Gentry, 1980). They are more poorly collected than flowers and as yet unknown for 1nany species and some genera. Van Steenis (1977) has stated that fruit set in Bignoniaceae is often poor. This has been attributed to high temperatures and low hutnidity for Crescentia cujete, Millingtonia hortensis, Pyrostegia venusta and Sparhodea campanulata (Chauhan et. al., 1987).

1.1. 2.11 Seeds The majority of species in Bignoniaceae have winged seeds. Even in some of the genera with indehiscent fruit, seeds have vestigial wings. Seeds in Bignoniaceae can often be characterized by presence or absence of wings, thickness of seed body, wing texture, wing structure and wing placement (Gentry, 1980). The presence or absence of 11 wings, and the concomitant thickening of the seed body reflect adaptive shifts from wind to water dispersal (Gentry, 1980). Despite the presence of winged seeds in Tecomanthe in the tribe Tecomeae, Cranwell (1962) disagrees with Oliver (1948) that the presence of wings indicates wind dispersal. She cites another tropical family, the Dipterocarpaceae, which despite having massively winged seeds is rarely carried more than 30-40m on air currents from the parent tree. Seeds do not contain starch and are rich in proteins and oils (20 - 30%) which are often largely unsaturated (Hegnauer, in van Steenis, 1977). In genera such as Crescentia, lncarvillea, Paulownia and Stereospermum, the only major fatty acids of the seed oils are oleic, and/or linolenic and/or linoleic acid, whereas in other taxa the common fatty acids are accompanied or replaced by large amounts of unusual fatty acids (Hegnauer, in van Steenis, 1977). Endosperm is not present in mature seeds (Armstrong, 1985).

1.1.1.12 Seedlings Seedlings are generally very uniform in all tribes of the fatnily (van Steenis, 1977). However, within the tribes Bignonieae and Crescentieae there is a fundamental distinction between epigeal and hypogeal germination. Apparently all Tecomeae have epigeal gern1ination. In general, species or genera with thick bodied seeds germinate hypogeally due to the retention of cotyledons within the seed coat whereas species with thin-bodied seeds have photosynthetic, foliaceous cotyledons elevated above ground by an elongating hypocotyl (Gentry, 1980).

The first leaves are almost always simple. Most genera of Tecomeae have conspicuously dentate tirst leaves whereas those in Bignonieae may be entire or serrate.

1.1.3 Cytology of the farnily The first chromosome counts for the family were done on Tecmnaria capensis and Pandorea jasminoides by Nakajima (1936). Various studies over the next 40 years included counts on a limited number of species generally using cultivated material of unknown origin. In the intervening tilnespan, taxonomic revisions of the family have resulted in counts being recorded under combinations that are no longer valid, or perhaps under two or three different natnes. 12

In 1975, Raven suggested that the family Bignoniaceae was one of 16 angiosperm families most in need of additional cytological investigation. This resulted in the detailed work of Goldblatt and Gentry ( 1979) who reviewed the literature and provided chromosome counts for 91 species in 51 genera. They determined which genera or species had been counted under an earlier name and which counts they regarded as suspect due to taxonomic uncertainty or an apparently unlikely chromosome number. They provided an excellent summary of all known chromosome counts for the family and arranged them according to the currently accepted of the family (Table 1.3, p13).

The predominant chroinosome number for the family is n =20 and it is the only number found so far in the advanced tribes Crescentieae (3 genera counted), Tourrettieae (monotypic tribe) and Coleeae (2 genera counted) (Goldblatt and Gentry, 1979). It is the most frequent number in the tribe Bignonieae, being found in 20 of the 23 genera studied. The three counts of n = 19 and n= 18-19 recorded for Bignonieae may indicate aneuploid genera, related through the loss of a single chromosome from the basic n =20 (Goldblatt and Gentry, 1979). The tribes Tecomeae and Oroxyleae are less specialized and show a greater cytological heterogeneity (Goldblatt and Gentry, 1979). In the Oroxyleae, Oroxylun1 has been found to contain n = 14 and Millington.ia, n = 15. These counts, together with the predominance of n =20 in the family as a whole, suggest a possible basic number of x =7 for the family, with all other numbers resulting from an increase in ploidy level and/or an aneuploid loss or addition of chromosomes. For example, the numbers of two spedalized herbaceous genera in the tribe Tecomeae, Argylia, n = 15 and Incarvillea, n = 11 may involve aneuploidy at the tetraploid level whereas Jacaranda, n=18, Pandorea, n=l9, Teco1na, n=18, Tecomaria, n=18 and Tecomanthe, n = 18 and possibly n = 19, might have arisen from aneuploid chromosome losses from paleo hexaploid stock (Goldblatt and Gentry, 1979). Two other genera provide evidence consistent with a basic chromosome number for the family of x =7. Firstly, Goldblatt and Gentry (1979) record n =21 in an isolated Andean genus, Delasroma, and suggest that it may represent the relict of a hexaploid stock with n =21. Second! y, Spathodea with counts of n = 13, n = 18 and n = 19 suggests two cytological lines, one from a tetraploid line with a loss of one chromosotne (n = 14 to n = 13) and one aneuploid arising from a hexaploid line (n=21 to n=18-19) (Fig.l.3, pl4). These 13 counts were, however, nearly all done on clonally propagated material of horticultural value and so may not be representative of the genus (Goldblatt and Gentry, 1979). The most recent count for a species in the family Bignoniaceae, Parmentaria valeri, with 2n =40, is consistent with the number reported previously for three other species in the genus (Goldblatt, 1980).

Goldblatt and Gentry (1979) conclude that most evolutionary radiation occurred subsequent to an initial aneuploid loss of one chromosome from a paleohexaploid stock, with only a few genera such as Oroxylum, Argylia and lncarvillea remaining at the ancestral tetraploid level. The predominance of n =20 within the family supports the view of a fairly close relationship between the tribes and a common ancestry in a single evolutionary line (Fig. 1.3, p14).

Table 1.3 Chromosome numbers found in each tribe in the family Bignoniaceae.

TRIBE CHROMOSOME NUMBER

Bignonieae n = 20,19,18

Tecomeae n = 20, 19, 18, 15, 13, 11

Crescen ti eae n = 20

Shlegelieae n = 20

Tourrettieae n = 20

Eccremocarpeae not known

Oroxyleae n = 15, 14

Coleeae n = 20 14

Diploid X = 7 (2n = 2x)

v

Increase in ploidy level

tetraploid hexaploid (2n = 4x) (2n = 6x)

v v

n = 14 (Oroxylum) n = 21 (Delostoma)

I I I I I I I aneuploid aneuploid aneuploid loss addition loss I (most evolutionary I I I radiation occurred I I I from this point) I I I I I n = 20 + I I n = 15 I I (Mi llingtonia) (Argylia) I I I n = 19 (Pandorea) I I (Spathodea) n = 13 (Spathodea) I I I n = 18 (Jacaranda) ~ (Tecoma) n = 11 (lncarvillea) (Spathodea)

Fig. 1.3 Possible pathway to the different chromosome numbers found in the family Bignoniaceae. Examples of genera at each chromosome number are given in brackets. Genera belonging to the tribe Tecomeae are shown in bold. 15

1.1.4 Chemical Constituents The family Bignoniaceae has not been the subject of intensive chemosystematic studies, but from the work done so far they will provide characteristics which can be utilized for taxonomic purposes. Hegnauer reviewed the chemical constituents in the family Bignoniaceae for the 'Flora Malesiana' (van Steenis, 1977). Gentry (1980) summarized that review and also mentioned phytochemically related features, such as odour, of certain genera and species which are already used taxonomically. Both authors conclude that phytochemically, the family Bignoniaceae forms a cohesive group and that it is also close to several other families within the Tubiflorae. The following is taken from those authors except where indicated.

Iridoid glucosides have been isolated from most members of the family analyzed so far. Iridoid alkaloids are found in some species of Campsis, lncarvillea and Tecon1a (Tribe Tecomeae) and Hegnauer believes that many more will be found as the family is studied further. Naphthaquinones and anthraquinones are prevalent, especially in taxa of the tribes Tecomeae and Coleeae. Woods which have been noted for their resistance to marine borers, white ants and fungi and for causing skin irritations and allergies in humans have also been found to contain large amounts of quinoid compounds. True tannins are replaced by more or less complex esters and glycosides of o-diphenolic cinnamic acid derivatives. Simple esters of caffeic acid and biosynthetically related derivatives of cinnamic and benzoic acid occur in appreciable amounts in many taxa of Bignoniaceae. Orobanchin-type glucoside esters were definitely demonstrated in a number of genera including Pandorea. Jacaranone, isolated from leaves and twigs of Dh Jacaranda caucana, is a quinoid compound exhibiting anti-tumour and cytotoxic activity and is chernically very similar to a glucoside present in Digitalis purpurea. Leaf and flavonoids in the family Bignoniaceae are mainly flavones rather than flavonols, a pattern common in Tubiflorae (Gentry, 1980). Harbome (1967) has shown that the major agents of petal colouration in a wide range of genera are cyanidin-3-rutinoside, a common anthocyanin, and various carotenoids. Gentry (1980) mentioned the basic dichotomy between petal colour in the tribes Tecorneae and Bignonieae which was supported by Harborne's finding that the former have anthocyanins and the latter carotenoids. The only known 3-deoxyanthocyanins in the Bignoniaceae, carajurin and 16 carajurone, are found only in the genus Arrabidaea (Scogin, 1980) and are responsible for the red colouration in the dried leaves of Arrabidaea chica (Gentry, 1980).

Scogin (1980) began a study of t1oral pigments in bat-pollinated species of the family Bignoniaceae which was expanded to determine whether the pigments found in bat­ pollinated plants were typical of the family in general or represented an adaptation to a specialized pollinator. The greatest variation in floral pigments occurs in the tribe Teco1neae where glycosides of cyanidin, delphinidin and pelargonidin have been identified (Scogin, 1980). Harborne and Smith (1978) suggest that bee flowers usually have delphinidin-dominated floral pigments, while cyanidin-based pigments are attractive to lepidopterans. At a generic level, the type of anthocyanin present is of little use systematically but there can be considerable quantitative variation in relative amounts of floral pigments in species within a genus, for example in Catalpa. The floral pigmentation of Paulownia, whilst not conclusive, is consistent with its placement in Bignoniaceae (Scogin, 1980). As a compound class, triterpenoids appear to be rare in the family Bignoniaceae, although /3-amyrin has been found to occur in Bignonia unguiscati and Halophragma adenophyllum (Ngouela et. al., 1988).

1.2 REPRODUCTIVE BIOLOGY OF THE FAMILY

1.2.1 Pollen

1. 2.1.1 Description of pollen types in the family Bignoniaceae The family Bignoniaceae exhibits a wide range of diversity m pollen morphology, from polyporate grains in genera such as An1phitecna and Saritea to inaperturate grains in, for example, Disrictella and Stereospermwn (Buurman, 1977; Gentry & Tomb, 1979). The pollen exine varies from the coarsely structured exine of Chilopsis linearis and Tanaeciun1 nocturnun1 to s1nooth exine found in Macfadyen.a unguis-cati and

Eccremocarpus scaber (Gentry & Tomb, 1979)~ Monads are found in the majority of genera but tetrads occur in Catalpa, Chilopsis, Cuspidaria (Gentry and Tomb, 1979), Anomoctenium, 8/epharitheca, Saldanhaea and Sideropogon (Buurman, 1977). Polyads 17 occur in a single species of Cuspidaria, C. bracteata, (Erdtman, 1952; Gentry & Tomb, 1979) but not in Stereospermun1 as stated by Suryakanta ( 1973).

The most recent contributions are the extensive studies by Buurman (1977) who concentrated on the Malesian genera, and Gentry and Tomb (1979) with their study of the neotropical genera. The most common pollen type in the family and the tribe Tecomeae is tricolpate with finely-reticulate exine (Gentry and Tomb, 1979). For the tribe Tecomeae, the genera are fairly well defined both morphologically and palynologically. Only three pollen types, tricolpate-finely reticulate, psi late- or microperforate-tricolpate and areolate, are found in the New World Tecotneae. Two more, inaperturate (or monocolpate) and stephanocolpate and subspinulose occur in the Old World species (Gentry and Tomb, 1979). Cranwell (1962) studied acetolyzed grains of Tecon1anthe speciosa, T. hillii (Australian species) and other species not named. Suryakanta (1973) commented that the pollen grains of all lianes (climbers with or without tendrils) exhibit a lot of variation regarding both apertures and exine patterns. However, clin1bers belonging to the tribe Tecomeae, such as Pando rea, do not exhibit that wide variation (see Chapter 4, section 4.2).

The study of the pollen of Bignoniaceae began in 1835 with the work of Mohl (1835) and has continued to the present day with the emphasis remaining on the taxonomic value of pollen morphological characters. The pollen morphology of the majority of genera in the family has been described (Buurman, 1977). Most studies have relied on light microscopy where interpretation of pollen morphology is often difficult. Nevertheless important contributions have been made by a number of workers. Schumann (1894) described the tricolpate grains which have been found subsequently to be the most common type in the farnily.

The first detailed study of the pollen morphology was done by Urban (1916) who drew attention to the taxonomic significance of aperture type. Genera within Bignoniaceae are extremely difficult to recognize (Gentry, 1973) and Urban was the first to suggest, as has been done subsequently, that the morphological characters of pollen could provide useful supplementary evidence of generic delimitation within the fan1ily (Buunnan, 1977; Gentry and Tomb, 1979). Van Steenis (1927) also commented on the 18 taxonomic value of pollen morphology although he did not make use of it himself. Urban ( 1916) cautioned that taxonomically unrelated genera in the main subdivisions of Bignoniaceae could be united if pollen characters alone were used in classification. This is because a number of different aperture types are present in each tribe, and tetrads occur in more than one tribe. Gentry and Tomb (1979) also stressed that although pollen morphology provides a valuable tool in the delimitation of genera and in understanding relationships within the family, it should not be given undue emphasis in taxonomic analyses. These authors cited examples (Mitra, 1968; Pichon, 1945) where over-reliance on pollen variation led to misinterpretation of the relatedness of various taxa. They also commented that most of these recent studies are marred by taxonomic errors when dealing with the difficult neotropical genera. Despite such pitfalls, variation in pollen morphology has become a useful adjunct to other plant morphological characters in understanding a taxonomically complex family.

Using fresh pollen, Goines (1955) produced a classification of pollen types although the study of acetolyzed pollen had already begun (Erdtman, 1952). Gomes (1955) used exine details and stressed the importance of the correct staining procedures for his classification (Table 1.4). Table 1.4 shows the four major classes of pollen grains in the Bignoniaceae. The three major classes are further subdivided depending on whether the exine structuring is smooth, reticulate or alveolate (Buurman, 1977). The presence of polyads in the genus Cuspidaria (Gentry & Tomb, 1979) suggests that category 'B' should be termed "polyads" rather than "tetrads".

Table 1.4 Classification of pollen types by Gomes (1955) based on the exine details of fresh pollen (from Buurman, 1977).

Category Grain type ; exine structure

A Single grain I. tricolpate II. more than 3 colpi III. inaperturate

B Tetrads 19 There are also examples of unusual pollen features in the Bignoniaceae. Cranwell (1962) described "rift-like" pores in the aperture membrane of Tecoman.the speciosa from Three Kings Islands, . Rupture patterns in the colpus membranes of tricolpate grains have been described for acetolyzed pollen grains (Buurman, 1977). They may be an artefact of the preparation rather than a structural feature of taxonomic merit. Guinet (1962) described dimorphism in the length of the pollen axis which was related to stamina! dimorphism in Milling toni a hortense. It is possible that such dimorphism may also occur in other genera.

With the emphasis on taxono1nic clarification, pollen studies have concentrated on external features. Ultrastructure and pollen development have received scant attention, and indeed the only published description of pollen internal structure appears to be that of Bignonia ven.usta described by Duggar in 1899. He found that B. venusta pollen contained a larger vegetative nucleus and a smaller generative nucleus which stained more strongly. He considered the generative nucleus to be a "free nucleus" in the mature grain because he found no permanent cell wall separating the two nuclei in a daughter cell. He also noted that the generative nucleus did not divide before germination of the pollen grain. He was in fact describing the bicellular pollen now known to occur in the Bignoniaceae (Cronquist, 1981), li1nitations of the light microscope causing Duggar's n1isinterpretation. The application of scanning electron microscopy techniques has been limited to only a few genera (Ferguson & Santisuk, 1973; Buurman, 1977; Gentry & Tomb, 1979) even though it alleviates the difficulties encountered in the analysis of pollen types using the light microscope. The pollen classification of Gomes ( 1955) should be re-examined using an analysis based on scanning electron microscopy. The use of transmission electron microscopy has been mentioned (Buurman, 1977) as necessary for the confirmation of tentative exine ·• structures based on light microscopy and to further characterize the pollen types found in the family.

1.2.1.2. Evolution ofpollen type The finely reticulate, 3-colpate pollen type is considered to be the ancestral or basal type for the family by both Gentry and Tomb (1979) and Buurman (1977), although they differ on the derivation of the Tecoma pollen type. Support for a relatively 20 unspecialized 3-colpate type from which more specialized types arose comes from the widespread distribution of that type across all tribes and in all geographic regions. In each tribe, the 3-colpate type is prevalent in taxa regarded as less specialized on morphological grounds and it is most common in the tribes Tecomeae and Oroxyleae, the putatively less advanced tribes in the family (Gentry and Tomb, 1979). The proposal that spirotreme pollen is the primitive type from which all other types have arisen (Suryakanta, 1973) is discounted by the work of both Buurman (1977) and Gentry and Tomb (1979).

1.2.2 Pistil The pistil has been described earlier. Of specific interest is the stigma due to its sensitivity to physical stimuli. The two lobes of the stigma separate as the stigma becomes receptive with the upper lobe lying against the dorsal side of the corolla tube and the lower lobe curving towards the ventral side of the corolla tube. It remains as such with the lower lobe recurving as the flower ages unless it is touched in which case the lobes close. In most cases the closure would be stimulated by the activity of pollinators in the corolla tube. It has been variously reported that the stigma will re­ open if the closure is the result of touch but pollination has not occurred (Newcombe, 1922; Newcombe, 1924; Petersen et al., 1982). Primary and secondary closure has been reported (Newcombe, 1922; Newcombe, 1924) but it is not known if this is a common feature of the mechanism. Differing reasons have been given for the mechanism, such as providing a humid environment for pollen germination (although presumably the stigmatic fluid itself would be sufficient) or preventing the loss of pollen from the pollinated stigma. The closure prevents multiple pollen depositions unless the observations of secondary closure imply that the stigma lobes reopen and can subsequently receive additional pollen. Presumably those visitors to the flowers which do not sti1nulate closure of the lobes are not likely to be pollinators (Gentry, 1974a). Ruby-throated honey eaters have been observed to touch the anthers or stigma when feeding by probing down the corolla tube on Camps is radicans, but when probing between calyx and corolla they did not effect pollination (Bertin, 1982). Xylocopid bees can extensively damage the corolla of Ane1nopaegm.a orbicu.latum as they rob nectar without pollinating (Gentry 1974a). 21

1.2.3 Embryology The embryology of a few members of the family Bignoniaceae has been studied in detail including Bignonia megapotamica (Swamy, 1941), Catalpa kaempferi (Soueges, 1940; Mauritzon, 1935) Parmenteria cerifera, Tecoma stans, Kigelia pinnata (Govindu, 1950), Jacaranda mimosaefolia (Mauritzon, 1935; Govindu, 1950), Bignonia tweediana, Phaedranthus sp., Catalpa bignonioides, Incarvillea delavayi, I. compacta, I. grandijlora (Mauritzon, 1935).

The ovules of plants within the Bignoniaceae are tenuinucellate with a massive single integument and an integumentary tapetum (Cronquist, 1981). They are numerous, anatropous or hemitropous, commonly erect with the micropyle directed downwards (Cronquist, 1981). The archesporia! cell is found in the hypodermal layer of the nucellar primordium (Govindu, 1950) and can be distinguished from the surrounding cells prior to the differentiation of the integuments. It functions directly as the megasporocyte and enlarges before undergoing meiosis to form, in most cases, a linear tetrad of megaspores. Occasionally, two linear tetrads were seen lying side by side in Kigelia (Govindu, 1950) and Swamy (1941) recorded their occurrence in Bignonia megapotamica. Two embryos have been observed in developing seeds of Pandorea jasminoides (James, unpublished observations). Like most dicots, the chalaza! megaspore is usually the functional one. At this stage the single integument is 2-4 cell layers thick and can enclose the nucellus almost completely, as in Parmenteria, extend only half its length, as in Jacaranda and Tecon1a or remain at a still lower level, as in Kip,elia (Govindu, 1950).

The embryo sac develops from the chalaza! megaspore via three mitotic divisions resulting in an 8-nucleate embryo sac of the Polygonum type (Maheshwari, 1948) comprising an egg apparatus, three antipodal cells and a binucleate central cell at maturity. For taxa studied by Govindu ( 1950), the mature ernbryo sac was dilated at the micropylar end and tapered towards the chalaza! end, the exception being Kigelia where the embryo sac was approximately the same width at both ends. The synergids are long and hooked and have pointed apices (Govindu, 1950; Mauritzon, 1935; Swatny, 1941) with nuclei situated in the apical region (Govindu, 1950). The primary endospenn nucleus is formed prior to fertilization by the fusion of the polar nuclei which 22 are about equal in size. Basal vacuoles are sometimes seen in the antipodals ( e.g. Tecoma) which are organized into definite cells (Govindu, 1950). The antipodals persist up to the early stages of endosperm formation, except in Kigelia. During the entry of the pollen tube into the embryo sac, one of the synergids degenerates. The tapetum is formed from the innermost layers of the integument following the disorganization of the nucellar epidermis and is in direct contact with the developing embryo sac (Govindu, 1950). In general, the embryo sac of members of the Bignoniaceae is not mature at anthesis (Duggar, 1899), and usually consists of four cells, however there are no other reports on whether mature embryo sacs were found in flowers at anthesis or later.

As the embryo sac enlarges, the nucellus is destroyed at the micropylar end but persists towards the chalazal end up to the early stages of endosperm development (Govindu, 1950). Endosperm development is cellular with a chalaza! and sometimes also a micropylar haustorium (Cronquist, 1981). The chalaza! haustorium is usually aggressive. The primary endosperm nucleus divides earlier than the egg (Govindu, 1950). Two types of endosperm formation were described by Mauritzon (1935), the Catalpa type in which the micropylar and chalazal haustoria are each 4-celled and the lncarvillea type in which both haustoria are single-celled. Govindu (1950) found that with only minor variations, the early development of Jacaranda, Pannentiera and Tecoma follows the Catalpa type. After the formation of the 4-celled endosperm, the development differs. Both the integumentary tapetum and endosperm haustoria are involved in the nutrition of the seed. Nutritive tissue is also organized at the chalazal region of the ovule. Embryo development follows that of the Onagrad type (Davis, 1966) that is, the zygote divides by a transverse wall and the apical cell divides longitudinally in the second cell division. The basal cell has little or no part in the formation of the en1bryo (Johansen, 1950).

1.2.4 Conclusions

Pollen The external features of the pollen grains of species in the family Bignoniaceae have been well studied with the light microscope. However, more extensive use of scanning 23 electron microscopy (SEM) would resolve some discrepancies in earlier studies of exine structure and may reveal errors in the interpretation of observations with the light microscope. For example, the tricolpate grains of Crescen.ria alata has been compared to those of C. cujete which have variously been described as inapterturate (Urban, 1916) or multiporate (Mitra, 1968) whereas the SEM reveals that pollen of the two species is similar in both the dimensions and the exine sculpturing with the main difference being the number and size of the colpi (Gentry and Tomb, 1979).

Transmission electron microscopy (TEM) has not been used to study the pollen of Bignoniace-.ae at all. Buurn1an (1977) has given tentative exine structures derived from SEM of thin sections prepared with a freezing microtome then coated with gold. These can only be considered preliminary and should be followed up with TEM studies to confirm Buurman' s descriptions. In addition, TEM studies would provide information currently not available, on the ultrastructure of the pollen of the family Bignoniaceae and could be used to describe pollen development.

Pistil The general anatomy of the ovary, and position and arrangement of ovules has been adequately described for the family. The mechanism of stigma closure whilst observed has not been studied. No information is available on the biochemistry or physiology of the mechanism. Stigma lobes of Pando rea species remain open if frozen in liquid nitrogen but close on thawing or if chemically fixed (James, unpublished observations) suggesting that the closure mechanism may involve a proton pUinp similar to that found controlling column movement in species of Stylidium (Findlay and Pallaghy, 1978). It appears that the phenomenon is common in the family but its extent has not been recorded. Further studies into its presence or absence in various taxa may highlight areas of the breeding systems within the family which would profit from an approach incorporating both physiological and ecological aspects.

Embryology Descriptive studies of the embryology of the family have enabled a good general understanding of post-fertilization events. However, this area of research would benefit from a reappraisal using modem microscopy and histology techniques. In addition, 24 expansion of the field to include embryo development following interspecific hybridization would provide valuable information not available for the fan1ily.

1.3 REPRODUCTIVE BIOLOGY IN THE FAMILY BIGNONIACEAE

1.3.1 Pollination Biology The pollination biology of the family has been studied in most detail in the species found in Central America. Floral morphology can be correlated with five distinct pollinator groups. These are hummingbirds, bats, large and medium-sized bees, butterflies and small bees (Gentry, 1974a,b).

Gentry (1974a) has described five types of flowering phenology found in Central American Bignoniaceae and has related each to the floral morphology and pollinator type found in the species. The range of floral morphology in the family has enabled species to make use of nearly all the potential pollinators in a tropical community, and presumably had some input into the evolution of animal specialists such as flower bats, hummingbirds, hawkrnoths and long-tongued bees. The flowering phenologies and associated pollinators can be summarized as shown (Table 1.5, p25).

Catalpa speciosa is a tree native to North America. It flowers for only 8 to 10 days with individual flowers lasting a maximum of two days (Stephenson & Thomas, 1977). It has an unusual pollination biology in that flowers are pollinated during the day by bumble bees and carpenter bees and diurnally by various moths. This is unusual because the floral adaptations normally associated with each pollinator type (Faegri & van der Pijl, 1979) are in conflict. Although flowers of Catalpa speciosa have characteristics of each pollinator type there is no character limiting them to one type of pollinator. Nectar production is higher at night, 3.23J.d compared to 2.56J.d, but sugar concentration, measured as sucrose equivalents, is lower, 23.7% compared to 39.0% (Stephenson and Thomas, 1977). This is not due simply to the evaporation of water from the nectar during the day and shows a diurnal pattern consistent with bee pollination during the day and moth pollination at night. The fragrance also changes diurnally suggesting that the occurrence of two distinct pollinator groups is not due to a 25

Table 1.5 Summary of flowering phenologies and associated pollinators found in Central American Bignoniaceae.

Phenological Type Description Pollinators Type 1 Flowers produced almost every day bats ("steady state") throughout plants reproductive life Type 2 Differs from type 1 in having shorter bees ("modified steady state") flowering periods. Separation of both types not always clear-cut Type 3 Generalised Bignoniaceae flowering bees, ("cornucopia") strategy shown by most temperate plants. hummingbirds, Species bloom seasonally for several hawkmoth, weeks. Individual plants synchronized bee/butterfly only according to season. Type 4 Single, brief, conspicuous burst of mass- general, ("big bang") flowering during the dry season only. mostly bees Blooming highly synchronized between individuals. Type 5 Numerous short flowering periods naive bees ("multiple bang") throughout the year. Good synchronization between individuals. Only occurs in vines. Nectiferous disc and nectar production often absent.

(Gentry, 1974b) breakdown in a co-evolved pollinator-flower syndrome but rather that there has been an adaptation to 24-hour pollination which results in an increased fruit set in a plant which flowers over a very short time span (Stephenson & Thomas, 1977). The only study of the pollination syndrome of an Australian member of the family Bignoniaceae has been on Deplanchea tetraphylla which has a highly developed syndrome for bird pollination probably by lorikeets (Weber and Vogel, 1986). The inflorescence forms a platform at the top of the tree on which the birds could perch whilst feeding, bending downwards to feed on dark brown nectar exposed in a spoon-shaped depression on the lowermost corolla lobe. The flowers are yellow, short-tubed and strongly zygomorphic with the mouth closed through lateral compression (Fig. 1.2, p8). The anthers and stamens are exserted and brush against the throat or breast of the feeding lorikeet. 26

1.3.2 Influence of pollination on fruit set The pollination history of an inflorescence has been shown to influence the resultant fruit set for both Catalpa speciosa (Stephenson, 1982) and Campsis radicans (Bertin, 1985). In C. speciosa, each flower in an inflorescence is capable of setting a fruit if it is the only outcrossed flower on that inflorescence, but as the number of outcrossed flowers per inflorescence increases, the probability that a particular flower will initiate a fruit decreases (Stephenson, 1979). Fruit set was more successful in the final phase of flowering when the number of flowers was low and decreasing compared to the initial phase in which number of flowers was low but increasing, and the middle phase of peak flower production. It is thought that the peak period is important in attracting pollinators which then continue to forage during the third phase. The rewards are less frequent as the proportion of pollinators to flowers increases and so the pollinators change trees more frequently in search of flowers which contain nectar, and as a result more flowers are outcrossed (Stephenson, 1982).

For C. radicans, the likelihood of fruit set differed with the identity of the pollen parent and also the number of prior (compatible) pollinations (Bertin, 1985). As the number of fruit on an inflorescence increased, the selectivity increased slightly so that the success of certain "non-favoured" pollen donors was reduced compared to the favoured donors (Bertin, 1982b). The pollen donors favoured tended to give rise to fruit with many large seeds (Bertin 1982b) which may have made them better resource sinks than fruit containing few or small seeds. Whilst fruit abortion seems to have been more important in donor selectivity than pre-zygotic phenomena, there are parallels with multifactorial gametophytic incompatibility (Bertin, 1982b). Bertin suggests that the selectivity of a pollen recipient with respect to the pollen donors is an integrated response increasing the average genetic quality of the zygotes. Also, as a plant's success as a male was inversely proportional to its success as a female, there may be gender specialization even though individuals are hermaphroditic.

1.3.3 Self-incompatibility The presence of a self-incompatibility mechanism was first mentioned for Bignoniaceae by Muller (1868) who found that flowers of Tecoma sp. were infertile to their own pollen. Duggar (1899) also found that self-pollinated flowers of Bignonia 27 venusta never set fruit and Elrod (1904) described a self-incompatibility response in Tecoma radicans. Since then, the following species have been found to be self­ incompatible : Campsis radicans (Bertin, 1982), Chilopsis lin.earis (Petersen et al., 1982), Catalpa speciosa (Stephenson and Thomas, 1977), Tabebuia rosea (Bawa and Webb, 1984), T. rosea, T. palmeri and T. neochrysantha (Bawa, 1974) (Table 1.6). The phenomenon is probably widespread throughout the family (Gentry, 1987) making the majority of species obligate outcrossers. There is a single record of a species which is apparently self-compatible. Cultivated plants of in New Zealand originated from one plant in which fertilization could be effected by artificial self­ pollination (Hunter, 1967). Bees were able to pollinate the species although it was far from its native habitat on Three Kings Island and far from its natural pollinators. There are no reports of post-zygotic failure after fertilization.

Table 1.6 Records of self-incompatibility in the family Bignoniaceae.

I SPECIES I NATURE OF EVIDENCE I REFERENCE I

Bignonia Observation that "artificial pollination with pollen from the same Duggar (1899) venusta plant was ineffective" during study of embryo-sac development suggesting that self-pollen did not result in fertilization.

Camps is (As Tecorna radicans) Controlled hand pollinations of plants outside Elrod ( 1904) radicans native habitat. Controlled hand pollinations in field with bagged and unbagged Bertin (1982) controls.

Cara.lpa Controlled self- and outcross-pollinations in field with comparisons Stephenson & .,peciosa between bagged and unbagged flowers. Thomas ( 1977)

Chilopsis Controlled self- and outcross-pollinations in field with comparisons Petersen et al. linearis between bagged and unbagged controls. (1982)

Tahl'huia Flowers bagged for controlled self- and outcross-pollinations in field Bawa (1974) neochrysantha and compared to open pollinated fruit set.

Tabebuia Controlled hand pollinations in field Bawa (1974) rose a Bawa & Webb (1984)

Tecoma Observation that flowers of Tecoma species were infertile to their Muller (1868) spp. own pollen. 28 Gametophytic self-incompatibility systems are generally correlated with wet stigmas (Heslop-Harrison and Shivanna, 1977) and binucleate pollen (Brewbaker, 1957, 1967), with the inhibition of the pollen tube occurring in the style or even the ovary. Species of Bignoniaceae have binucleate pollen (Cronquist, 1981), wet stigmas with low to medium papillae on the receptive surface (Heslop-Harrison & Shivanna, 1977), and can be designated a wet-papillate type according to the revised classification of Heslop­ Hamson (1981). It is likely, that if present, the self-incompatibility system of PanLiorea is a gametophytic one.

1.3.4 Interspecific Hybridization All records of interspecific hybridization are between species within the tribe Tecomeae. Gentry (personal communication) considers that interspecific hybridization rarely occurs naturally in Bignoniaceae. A putative hybrid between Tabebuia chrysantha and T. palmeri was found in Rio Mayo, Sonora (Mexico), but this is the only known natural hybrid (Gentry, 1942). Two species of Crescentia, C. alata and C. cujete, apparently hybridize in nature but there is morphological and palynological evidence to suggest that their segregation into two species is not valid (Gentry and Tomb, 1979). Both interspecific and intergeneric hybrids have been produced artificially in the Bignoniaceae but it is not known whether or not such hybrids are the results of exhaustive attempts at hybridization or were achieved readily. Gentry (personal communication, 1987) considers that artificial hybridization is achieved artificially with comparative ease.

Records of hybridization within the genus Catalpa include a cross between different Caralpa species (Smith, 1941; Palet, 1952) and a cross between Catalpa ovata Don and C. bignonioides Walt which has been known as C. syringijolia Sims (van Steenis, 1977). A cross between Ca1npsis grandijlora and C. radicans is known as Campsis x ragliabuana 'Madame Galen' (van Steenis, 1977). Tecon1a smithii W. Watson is reputedly a hybrid made by E. Smith at Adelaide in 1882 between T. velutina (a hairy form of T. stans) and T. capensis (now Tecomaria capensis) (Watson, 1894). It has been known also as T. alara (van Steenis, 1977). An intergeneric hybrid has been produced recently in Russia between species of Catalpa and Chilopsis and is becoming widespread in horticulture (Gentry, personal communication, 1987). 29 1.3.5 Conclusions The pollination biology of some North and South American species within the family Bignoniaceae have been studied in some detail and the pollination mechanisms are known for many more, but considering the size of the family it warrants further study. In addition, the reproductive biology has been overlooked. In particular, the existence of self-incompatibility within the family is recognised in a number of taxa but the site of incompatibility has not been determined for any of them. In the light of the 'late-acting' self-incompatibility discussed by Seavey and Bawa (1986) and its predominance in woody plants, the family with its range of plant habits ranging from herbs to woody vines to large forest trees and its wide ecological and geographical distribution provides a broad base from which to study the phenomenon at the microscope level.

1.4 THE AUSTRALIAN SPECIES OF THE FAMILY BIGNONIACEAE

Twelve species, belonging to five genera of the tribe Tecomeae, occur naturally in Australia (van Steenis, 1977) (Table 1. 7, p30). The number may be reduced depending on the status of species of Dolichandrone and Neosepicaea.

Van Steenis (1977) considers that Dolichandrone hererophylla and D. altern{folia should be regarded as a single variable species as suggested by Seeman ( 1870) who accepted the epithet hererophylla. Van Steenis (1977) states that another species, D. sparhacea, does not occur in Australia. Thomas and McDonald (1989) however, have not only listed it amongst the rare species of Queensland but have also given locations. Its natural occurrence is restricted to the Olive River and Cowal Creek (one plant) near the tip of Cape York (N. Duke, pers. comm.) The status of Ne({sepicaea in Australia is also unclear. Van Steen is ( 1977) lists a single endemic species, N. jucunda, which does not appear to be listed elsewhere. Thomas and McDonald (1989) mention a second species, N. viricoides, as a poorly known species in Australia whose conservation status in unknown. It was formerly considered to be endemic to New Guinea (van Steenis, 1977). It was presented by Williarns (1987) as Pandorea sp. nov. but has since been identified from a cultivated plant of N. viticoides which had been collected as a seedling from the Iron Range in North Queensland (McDonald, pers. comm.). It is possible that 30 the present distribution of N. viticoides in northern Australia reflects recent naturalization of the species rather than a restricted distribution of a naturally occurring species.

The leaf and stein anatomy of P. jasminoides have been recorded (Hovelaque, 1888). Differences in the anatomy of internodal zones of one year old stems of P. jasminoides and P. pandorana (as P. australis) have been compared. The two species differ markedly in the disposition of the xylem and phloem around the median axis. Whereas P. pandorana shows a radial symmetry throughout the various cross-sections of stem and has a strongly developed central xylem cylinder, stems of P. jasmin.oides are bifacial in their symmetry and have a continuous central xylem cylinder. The fact that the phloe1n layer is as equally well-developed as the xylem in the latter species and also the inclusion of phloem in the secondary xylem, tends to indicate that P. jasminoides has more flexibility in torsion than P. pandorana.

Table 1.7 Species of the family Bignoniaceae which occur naturally in Australia.

SPECIES HABIT

Pandorea pandorana Iiane P. jasminoides Iiane P. nervosa liane P. baileyana liane Tecomanthe hillii Iiane Neosepicaea jucunda Iiane N. viricoides Iiane Doficlwndrone heterophylla tree to 20m D. alrern{fblia tree to 20m D. .\pathacea mangrove 5 -20m D. jil{formis tree to 20111 Deplanchea tetraphylla tree to 25zn 31 1.5 GENERAL CONCLUSIONS

The family Bignoniaceae is large with a wide geographical distribution. Whilst it probably has a monophyletic origin, the family is at least of Tertiary age and has been subject to periods of speciation and migration so that specific and generic limits are not always readily defined. The Australian species make up approximately 1% of the family Bignoniaceae with Pandorea pandorana probably the most temperate species in the family as a whole. Certain aspects of the family, such as floral and vegetative characteristics and pollen morphology have received detailed attention in an attempt to clarify the taxonomy of the family. Other areas, in particular ultrastructure, reproductive biology and physiology have barely been considered. Horticulturally valuable material has already been produced from interspecific hybridization within the tribe Tecomeae providing a sound basis for the current breeding programme. 32 CHAPTER 2

MATERIALS AND METHODS

2.1 PLANT MATERIAL USED IN THIS STUDY

2.1.1 Reproductive biology and breeding system The origin of all plant material, flowering characteristics and the coding system used are shown in Table 2.1 (p34). Clonal material of three genotypes each of P. jasminoides (HW, LD, AL) and P. pandorana (SCH, AF and GS) were obtained from commercial nurseries as rooted cuttings. The remaining genotypes were selected from wild populations or arboreta in New South Wales, Victoria and Queensland.

Genotypes other than those bought from nurseries were propagated from 10 em cuttings with the basal leaves removed. Fresh cutting material was firstly dipped in a fungicide Rovral, then the base of each cutting was dipped in 2000mgL-1 IBA for 5 sec, then inserted into individual propagation tubes containing 1 peat : 2 sand (v/v) and placed under intermittent mist until rooting occurred after about 6 weeks.

Rooted cuttings were potted into growing media comprising 2 pinebark : 2 coarse washed river sand: 1 brown coal (v/v) and fertilised with 3 kg N:P:K: Osmocote, 1 kg Micromax, 0.5 kg coated Fe and 2 kg dolomite per m3 at the tin1e of media preparation. Plants were grown in a glasshouse and transferred to an insect-proof house for experimentation. Pots were watered automatically once each day with a drip-feed system which ensured that flowers were kept dry.

2.1.2 Tissue Culture Pando rea species are easily propagated from nodal cuttings, however, as hybrid material resulting from a breeding program may be in short supply, a suitable tissue culture medium was sought. The medium could then be used as a basis for developing the optimum medium for the multiplication of elite hybrid clones. 33

Plant material was collected from container-grown stock plants which had been kept in an insectproof, temperature-controlled glasshouse and watered with a dripper system for six weeks prior to tissue culturing to minimnize fungal and bacterial contamination on the explants.

Shoot tip explants were sterilized in 1% available chlorine (8ml sodium hypochlorite in lOOml distilled H20) for 15min and rinsed 3x in sterile distilled H20. Explants were put into individual tubes containing basal media (Table 2.2, p35) and allowed to establish.

Four different explants (Fig 2.1) were then taken from the established cultures and put into tubs containing basal medium amended with 0.5p.M each of benzyl amino purine (BAP) and kinetin. After 4 weeks, measuretnents were taken of shoot height, new shoot number and number of nodes. Each treatment consisted of 5 tubs which were considered replicates.

n 1 meristem

1]'" 2 meristems l_n:~~.l ~~ I ::od: I 4 meristems ·~ Root - _ _..,.~g

Fig. 2.1 Diagram showing the source of the four different explants of P. jasminoides used for tissue multiplication. 34

Table 2.1. Description and source of genotypes of P. pandora1Ul (P.p.), P. jasminoides (P.j.), P. nervosa (P.n.) and P. baileyano. (P.n.) used in study.

Species Genotype Description of Flowers Source

P.p. TS3 Cream flowers, maroon centres, Single population TS4 unscented near Merimbula, TS6 New South Wales

vc Cream flowers, maroon centres, Valencia Creek, Gippsland, unscented Victoria.

TH Cream flowers, maroon centres, Thorpdale, Gippsland, unscented Victoria.

SCH Pale yellow, scented Schubert's Nursery, Victoria.

AF Pale yellow, scented Austraflora Nursery, Victoria.

GS Narrow tubed, yellow, Austratlora Nursery, unscented Victoria.

MM Small, narrow tubed, Mt Mee, largely maroon, unscented Southern Queensland

BF Yellow, broad corolla tube, Mt Bartle Frere, scented Central Nth Qld

AL Cream, no corolla markings, Humphris Nursery, unscented Victoria.

P.j. HW Pink flowers, faintly scented Hardwick's Nursery, Victoria.

LD White flowers, faintly scented Austraflora Nursery, Victoria.

AL White flowers, faintly scented Banksia Nursery, Victoria.

ATH 1 Pink, faintly scented CSIRO Arboretum, Atherton, Qld

F827 White, faintly scented F 1 between AL & LD (current study)

DKP Dark pink, faintly scented F 1 between unknown parents

LGF Pale pink, large flower, F 1 between unknown parents faintly scented

P.n. Single White, mauve markings at front of corolia CSIRO Arboretum Atherton genotype tube, unscented

P.b. 2 Pale yellow/green, unscented Green Mountain, Sthn Qld

3 Pale yellow, unscented Green Mountain, Sthn Qld 35

Table 2.2 Components and their concentrations in the basal medium used for the establishment of Pandorea tissue cultures.

Component Concentration Component Concentration

(mM) (JJM)

NH 4N03 10.0 myo-lnositol 300.0

KN01 10.0 Nicotinic acid 20.0 NaH~P0 4 l.O Pyricloxine.HCI 3.0 CaCI 2 2.0 Thiamine.HCI 2.0

MgS04 1.5 Biotin 0.2 D .Ca. Pantothenate 1.0 (JJM) Ribotlavin 0.1 Ascorbic acid 0.1

H3 B04 50.0 Choline chloride 0.1

MnS04 50.0 Glycine 0.1

ZnS04 20.0 L. Cysteine.HCI 60.0 KI 2.5

Na2Mo04 0.1 (g/'1)

CuS04 0.1

CoCI 2 0.5 Sucrose 20.5 FeSO./Na~EDT A 50.0 Gel rite 2.5 pH 5.7

2.2 FLORAL DIMENSIONS

Nine nora! di1nensions were measured for 20 tlowers each for 10 genotypes of P. pandorana, six genotypes of P. jasminoides and a single genotype each of P. baileyana and P. nervosa. The dimensions measured were calyx length, corolla tube length, corolla outside length, corolla width-inner, corolla width-outer, corolla depth-inner, corolla depth-outer, pistil length and stamen length. They are described diagrammatically in Fig. 3.9 (p57). 36

2.3 BREEDING SYSTEMS

2.3.1 Reproductive structures

2.3.1.1 Pistil

Stigmatic receptivity Stigmatic receptivity was determined by pollinating stigmas of different ages with pollen known to be viable and observing whether or not pollen germinated and pollen tubes were present in the style. Pistils were harvested 48h after pollination, fixed in 3:1 ethanol:glacial acetic acid and prepared for aniline blue fluorescence to observe pollen tube growth. Esterase activity could not be used as an indication of receptivity due to closure of the stigma lobes during treatment. Excising the lobes resulted in a staining reaction from injured cells and so was not considered to be a viable method to test for receptivity.

Stylar anatomy To determine whether the style was hollow, styles were fixed in 2. 8% glutaraldehyde in O.lM phosphate buffer at pH 7.4. After fixation, samples were washed in O.lM phosphate buffer, dehydrated through an ethanol series and embedded in LR white (London Resin Company). Blocks were sectioned with a Reichert-Jung Microtome using glass knives.

2.3.1.2 Pollen

Microscopy For trans1nission electron Inicroscopy (TEM), pollen was collected from dehiscing anthers and fixed for 2h in 2.8% glutaraldehyde in O.lM phosphate buffer at pH 7.4. After fixation, samples were washed in O.lM phosphate buffer, post-fixed for 2h in 1%

Os04 and washed in distilled H20. Pollen was stained en bloc with uranyl acetate for lh before washing in distilled H20, dehydrated in an acetone series and embedded in Spurr's resin. Sections were cut with a Reichert-Jung Ultracut microtome, using a diamond knife, to 80nm, picked up on copper grids and stained with uranyl acetate and 37 lead. Sections were viewed with a JEM1200 transmission electron microscope and photographed.

Pollen samples were prepared in two ways for scanning electron microscopy (SEM). One set of anthers was placed directly into a dessicator for 48h before mounting on SEM stubs. A second set of anthers had pollen removed, mounted onto SEM stubs with double-sided sticky tape and fixed in 3.5% glutaraldehyde in O.lM phosphate buffer at pH 7.4 for 24h at room temperature. Pollen was dehydrated in an ethanol series and critical-point dried with C02 • Samples were coated with platinum/palladium using a sputtercoater Model SC150 and observed with a JEOL JSM-840 Scanning Electron Microscope.

For light microscopy, pollen was fixed for 24h in 3:1 ethanol:glacial acetic acid and rehydrated in distilled H20 before staining with 4' ,6' diamidino-2-phenylindole (DAPI) (Coleman and Goff, 1985). After 1.5h, satnples were observed using an Olympus Vanox microscope fitted with a fluorescence unit.

Pollen viability Pollen viability was determined for P. jas1ninoides and P. pandorana in two ways. Firstly, pollen from flowers ranging from -ld to 5d post-anthesis was stained with fluorescein diacetate and the number of fluorescing and non-fluorescing grains recorded. Six fields of view were counted per flower. Second I y, pollen from the same pollen sample was used to pollinate ld old stign1as which had been shown previously to be receptive. The presence or absence of germinating grains on the stig1na and of pollen tubes at the base of the style was recorded 48h after pollination. Flowers for the second experiment were removed from the plant prior to pollination and placed in individual vials. They remained in a constant temperature room (20°C) with 12h/12h day/night cycle. After 48h they were fixed in 3:1 ethanol:glacial acetic acid for 2-24h and stored in 70% ethanol until required. Pistils were softened irt 5% sodium sulphite by pressure cooking in a domestic pressure cooker for 5min at 121 oc and 1OOkPa pressure. After rinsing in distilled H20, pistils were placed individually on glass slides and gently squashed in decolorized aniline blue. Slides were kept at 4°C for 12-36h prior to 38 observation of pollen-tube growth ustng an Olympus Vanox microscope fitted with a fluorescence unit.

2.3.2 Pollen-ovule ratio Ovule number was determined for P. jasminoides and P. pandorana by counting all ovules per ovary (one per flower). Pollen grains were counted from the upper lobe of the upper right-hand anther for each flower and multiplied by eight to give total number of grains per flower. Preliminary counts had not highlighted any difference in the numbers of pollen grains between the upper and lower anthers.

2.3.3 Pollen-tube growth in selfed and outcrossed flowers Selfed and outcrossed pollinations were performed on all species except P. nervosa where only a single genotype was available. Flowers were pollinated in an insect-proof glasshouse and pistils were collected for assessment 48h after pollination. Material was fixed and prepared for observation using decolorized aniline blue as described previously. The very large numbers of overlapping pollen tubes in the pistil precluded accurate quantitative data collection.

2.3.4 Separation of male and female function Flowers of P. jasmin.oides and P. pandorana were studied to determine whether distinct male and female phases occurred in the species. For P. jasn1in.oides, plants were maintained in a temperature controlled glasshouse (min. l6°C, max. 27°C). For P. pandorana, plants were growing in the ground outside and were subject to ambient weather conditions. Flowers were tagged at an thesis on day 0 and observed at 9 .OOam and 4.00pm daily for stigma opening, anther dehiscence and corolla abscission. Additional flowers of P. jasminoides were studied to pinpoint the time of stigma opening relative to anther dehiscence.

Measurements were made of the length of the stamen and pistil and the likelihood of contact was inferred from those results. 39 2.3.5 Self-Incompatibility, Genotype and Reciprocal Effects Pando rea jasminoides, P. pandoran.a and P. bailey ana were used in breeding experiments consisting of intraspecific diallel crosses in order to test for the presence of a self-incompatibility system and for genotype and reciprocal effects. Pandorea nervosa was not included because only one genotype was available.

Experimental pollinations involved removing pollen from anthers of the desired male parent with the tip of forceps, then applying it by gentle brushing onto the open stigma lobes of the female parent either on the day of an thesis or the following day.

Fruit set was recorded two weeks after pollination. At this stage, the corolla tube had fallen off and the swollen ovary was clearly visible above the level of the calyx. Each flower has only one ovary giving rise to a single fruit.

2.4 INTERSPECIFIC HYBRIDIZATION

2.4.1 Pollinations All four species of Pandorea were used in interspecific pollinations to provide interspecific hybrid material. Experimental pollinations were as described in section 2.3.5.

2. 4.2 Embryo rescue Mature fruit of interspecific crosses between Pandorea species was obtained on a number of occasions. The fruit was of the size expected from intraspecific pollinations. Seed was usually smaller and failed to genninate. In vitro culture of immature embryos, after their removal from the parent plant, has been successful in obtaining interspecific hybrids of a number of plants including Ornithogalun1, (Niederwieser et a/., 1990), Actinidia (Kin et. al., 1990) and Nicotiana (Reed and Collins, 1978).

For interspecific fruit of Pandorea, the developmental stage and size of embryos ranged widely within each fruit. Not all embryos in a fruit were cultured. Those 40 largest embryos or most swollen embryo sacs were preferred. Intact embryo sacs were only cultured if the embryos within could not be readily excised without damage.

Initially four media were tested (Table 2.3) based on a modified basal medium (Table 2.4).

Table 2.3 Four media used initially for the culture of immature putative hybrid embryos of Pandorea.

Medium I -Basal (Table 2.3) + 6g agar + 70g sucrose Medium 2 - Basal + 6g agar + 50g sucrose Medium 3 -Basal + 2.5g Gelrite + 70g sucrose Medium 4 - Basal + 2.5g Gelrite + 50g sucrose

Table 2.4 Components and their concentration in basal n1edium used for embryo rescue.

Component Concentration (mg.l-J)

M & S Inorganics (Murashige and Skoog, 1962) FeNaEDTA 25.0 glutamine 400.0 cysteine 20.0 glycine 10.0 arginine 10.0 myoinositol 100.0

Media were poured into petri dishes under sterile conditions, wrapped in gladwrap and stored at < 4°C and used as required. Fruit was surface-sterilized for 15min in 2% available Cl (16ml x 12.5% available Cl from Sodium hypochlorite in 1OOml distilled

H20) followed by three rinses in sterile distilled H 20 under sterile conditions. 41

When excised embryos had grown to the stage where cotyledons were a minimum of 1-2mm in diameter and the radicle was > 2mm long, they were transferred to a germination medium. Germination medium comprised the basal medium with 1% sucrose and 2.5g Gelrite per litre. It was put into individual polycarbamate tubes, sterilized for 15min at 121°C and stored < 4°C until required.

2.4.3 Fruit measurements and seed gennination Fruit was collected from plants when the capsules were beginning to dehisce and allowed to dry at room temperature (18-22°C) for a minimum of 7d. Whole fruit, capsule, septum, seed weights, numbers of seed per fruit and days to maturity were recorded.

Seeds were germinated by placing on Whatman No.3 filter paper in 9cm petri dishes

and kept moist with distilled H20 as required. Germination experiments were carried out in a controlled temperature room (200C) under a light regime of 12h day/ 12h night.

2.4.4 Isozym.e analysis of putative hybrids The isozyme assays were performed on leaf tissue from the youngest fully expanded leaves in the seedlings and the equivalent material where possible from the parent plants using the Titan III cellulose acetate gel system of Helena Laboratories. Initially 13 enzyme systems were tested for differences between parents and thus their suitability for identifying hybrid offspring (Table 2.5, p42).

Approximately 25mm2 of each sample was ground in a Cooke Microtitre tray in 50J.il of extraction buffer. Initially, material was extracted in either Coates buffer (Coates, 1988) or a more complex buffer (Table 2.6, p43). Coates buffer proved inferior and was not used again. The prepared leaf extracts were transferred from the microtitre tray by pipette to a sample plate containing 12 sample wells. An applicator with 12 independently suspended platinUJn wires was used to apply a small a1nount of each extract was to the cellulose acetate gel (Titan III, 94 x 76mm) which had been soaking for 15-20 min in the running buffer. The loaded gels were placed with the gel surface down in a Titan Gel Chamber. Each end of the gel rested on a disposable paper wick and the end of the gel containing the samples was placed at the cathode. Electrophoresis 42 was carried out at room temperature for 20-30 1nin using a Heathkit Regulated H. V. power supply.

Table 2.5 Enzyme systems initially assayed from parents and their putative hybrids.

Enzyme System Abbreviation E.C. Number

alcohol dehydrogenase ADH 1.1.1.1 aspartate amino transferase AAT (=GOT) 2.6.1.1 isocitrate dehydrogenase IDH 1.1.1.42 glycerate dehydrogenase GLY 1.1.1.29 fluorescent esterase FE 3.1.1.1 esterase EST 3.1.1.1 glutamate dehydrogenase GDH 1.4.1.2 glucose phosphate isomerase GPI 5.3.1.9 malate dehydrogenase MDH 1.1.1.37 leucine amino peptidase LAP 3.4.11.1 phosphoglucomutase PGM 2.7.5.1 shikimic acid dehydrogenase SDH 1.1.1.25 phosphogluconate dehydrogenase 6PGD 1.1.1.49

For all enzymes other than leucine amino peptidase (LAP), the buffer (TEM) comprised 80mM Tris and 3mM Na2EDTA with the pH adjusted to 8.2 using a saturated solution of maleic acid at pH 7. 8 (Richardson et. al., 1986). The gels were stained after electrophoresis by applying the staining solution as an agar overlay made up of 6ml of 1% agar at 4Y'C added to 21nl of buffer containing the staining ingredients.

For EST, the stain used comprised 0.2ml a-naphthyl acetate (lOrng.rnl-1 in 1:1

1 acetone: H20), 0.2 ml ,B-naphthyl acetate (10mg.rni· in 1:1 acetone: H 20) and 4mg Fast Garnet GBC salt dissolved in 2rnl 0.2M phosphate buffer (pH 7. 0) to which was added 6ml of 1% agar. The stain for GLY was made up of 1.2ml O.lM Tris buffer pH 9.5; 6mg DL-glyceric acid (hemicalcium salt), 0.2ml nicotinamide adenine dinucleotide 43

(NAD, lOmg/ml); 0.2ml methyl thiazoyl blue (MTT, 2rng/ml); 0.2ml phenazine methosulfate (PMS, lmg/ml) and 6ml of I% agar. For the remaining enzymes, the staining protocols were obtained from the following sources: LAP (Adams & Joly, 1980), FE (Mitton et.al., 1979), ADH, AAT, GPI, MDH, 6PGD, PGM (Fripp, 1982) IDH, SDH (Leeton & Fripp, 1991) and GDH (Richardson et. a/., 1979).

Table 2.6 Enzyme extraction buffer used for Pandorea leaf samples undergoing isozyme analysis.

Component Weight per 10 ml

PVP-40T (Polyvinylpyrrolidone) (av.mol. wt. 40,000) 0.7g PVP-360 (av.mol. wt. 360,000) O.lg L-ascorbic acid 5.0mg Diethyldithiocarbamic acid (sodium salt) 4.5mg Sodium metabisulphite 38.0mg DIT (DL-Dithiothreitol, "Cleland's Reagent"; DL-DIT) lO.Omg

Na2-EDTA 17.0mg Bovine Serum Albumin 10.0mg D-Sorbitol l.Og NAD (free acid, crystallized) 5.0mg NADP (disodium salt) 3.0mg

In IOml O.lM Tris-HCI buffer at pH 7.0

For crosses involving P. nervosa and P. baileyana, three of the enzyme systems evaluated, PGM, MDH and LAP, looked promising for differentiating the hybrid seedlings. For crosses involving P. pandorana and P. nervosa only GPI and SDH showed differences between the parents. As the actual parent plants were known in each case, the gels were run with samples of each parent and a sample of the putative hybrid offspring. 44 A random sample of 16 was taken from the P. baileyana x P. n.ervosa seedlings. The three enzymes which seemed promising were tested with both parents and the 16 seedlings. As there were only seven putative P. pan.dorana x P. nervosa hybrids, all were tested.

2.5 STATISTICAL ANALYSES

Data were processed using a Genstat 5 statistical package (Lawes Agricultural Trust, Rothamsted, UK). Results with a binoxnial distribution including all pollen viability results from the FCR test, pollen-tube growth, diallel crosses and interspecific pollinations were analysed by fitting one or more generalised linear models using a legit­ link function (McCullagh and Neider, 1989) and assuming a binotnial error distribution. Details of the analytical methods for the diallel crosses are given with the results in Chapter 5. Non-binomial data such as floral dimensions, pollen-ovule ratios and fruit dimensions were analysed by analysis of variance. 45 CHAYfER 3

THE GENUS PANDOREA (Endl.) Spach1

3.1 DESCRIPTION OF THE GENUS

The genus Pando rea was first described by Endlicher (1839). It belongs to the tribe Tecomeae which is the most widespread of the tribes in the family Bignoniaceae and has an apparent centre of distribution in Brazil (Gentry, 1980). The tribe Tecomeae occurs in Australasia, Asia, India, east Africa and North and South America and is perhaps the oldest extant group of Bignoniaceae (Gentry 1976b). The tribe is characterized by fruits (capsules) which dehisce perpendicular to the septum except in a few Old World genera (Gentry, 1980). They are mainly trees or shrubs with a few erect herbs or non­ tendrillate woody lianes with compound leaves. The ovary is bilocular with axile placentation.

Pandorea is one of 5 genera of Bignoniaceae which occur naturally in Australia (Table 1. 7, p31). All species are woody vines which cli1nb without tendrils. The most recent review of the genus was done by van Steenis (1977). He considers the genus to contain only 6 species (Table 3.1, p46) of which 4 occur in Australia. Three of these are endemic with restricted distribution. Probably the most well-known species is Pandorea jasn1inoides (Bower of Beauty), a popular garden climber with a showy floral display typical of the fatnily (Fig. 3.la, p47). This species occurs from northern Queensland to southern New South Wales. Although it has been found in Victoria, the plant was thought to be beyond its natural range and was quite likely to be a 'garden escape' (Willis, 1972). A seco_nd species, P. pandorana (Fig. 3.1b, p47) which includes P. doraroxylon (van Steenis, 1977), is the only Australian species whose range extends beyond the Australian continent. It is found in habitats ranging from the wet, tropical rainforests of northern Queensland and New Guinea to the arid centre of Australia. It is found throughout the continent and is also the most variable of the Australian species.

1Material from this chapter has been used as the basis of an article: James E. (1990) The Australian species of Pandorea. Aust Plants 15(124):361-368. The following article has been submitted to Australian Plants: James E. (199-) Australian representatives of the family Bignoniaceae. 46

A third species, P. nervosa (Fig. 3.lc, p47) occurs in the Atherton region of Queensland, whilst the fourth species, P. baileyana (Fig. 3.ld, p47) has been collected from a small area around the south-eastern border of Queensland and New South Wales (Thomas and McDonald, 1989) and from southern New South Wales (Fig. 3.2, pp48- 51). All species produce fruit capsules typical for the family (Fig. 3.3, p53).

Table 3.1 - Species of Pandorea

Species Location

Pando rea jasminoides Australia, endemic P. nervosa Australia, endemic P. baileyana Australia, endemic P. pandorana Australia, , Indonesia P. montana Papua New Guinea, Indonesia P. stenantha - Papua New Guinea, Indonesia

3.2 FLORAL MORPHOLOGY

3.2.1 Inflorescence All Pandorea species are lianes with flowers consisting of terminal inflorescences (van Steenis, 1977). Pandorea baileyana has flowers borne on a long or spike (Maiden & Baker, 1895). The remaining species have branched inflorescences with determinate main axes and indeterminant axillary axes forming a pyramidal cluster.

3.2.2 Flower structure The individual flowers of Pando rea species vary in size and shape depending on the species. All, however, are tubular and attached to the by an obvious pedicel. The flower consists of a cup-shaped fused calyx which usually splits into 5 lobes. The corolla is also fused and inserted within the calyx; the length of the fused section varying with species. The corolla divides into 5 lobes with the upper 2 usually slightly smaller than the lower three. The lobes recurve to varying degrees as the flower ages. The four stamens are arranged in 2 pairs with a staminode (vestigial stamen) between them. The stamen filaments are fused to the base of the corolla and the anthers each consist of 2 lobes containing a single locule which dehisces longitudinally from slits. Figure 3.1 Examples of the Australian species of Pandorea. a. P. jasminoides(HW) b. P. pandorana(AL) c. P. nervosa d. P. baileyana (2) • 47 48

11-Har-9 Old Herbarium Distribution Mep ror Species: Pendoree jesm1no1des

--- r r

I I r-- t I --I iS I I I I I I I I ,----~--- I 1 - r-- I ~ I --I ::o I I I I I I I I I I ~-r-----.----L-_-_r_- 1 - - - r ____ I I -125 \ I I I I I I I r -, 30 I I

1 I I I --~-----~----~-----~ -- a_l ,' ------~-- ~ w ~ r - r--~- 1 U, ~ ~ o (/1 I tJ ;---- I 40 0 SIMPLE CONIC PRO~ECTION I I I SCALE 1:25 000 000 I I o Accurate to 1 deg. ____ ~ _ _ 1 ;~~~ic~ated :; .. ______1., 5 • Ul ~ i; C/)

Figure 3.2 Distribution maps for the four Australian species of Pandorea from collections held at the National Herbarium QLD (unless otherwise indicated). a. P. jasminoides ( • collection from Nat. Herb. NSW) b. P. pandorana <• locations from Nat. Herb. TAS) c. P. nervosa d. P. baileyana. 49

11-Mar-9 Old Herbarium Distribution Hap for Species: Pandorea pandorena _ ----~----r----r----r-- - - 2.Ql_S.£8Clmena 1 1 --- I - r------r 1 1~1 1 I ._I lO r I I I I I I I I I I I r----r ----r I 1 I --- r Ir-- ._- ~- I 1 zs I I I I I I I I I I I r----r-- I I I I -- 1 r-- 1 11---- r- I --I ~0 I I I I I I I I I ----~----r--- 1 I I r.------r--~--~ I -I 25 I I I I I ----~----~--- 1

------'j;. SIMPLE CONIC PROJECTION SCALE 1:25 000 000 I o Accurate to 1 dag. I + Cui ttvated ---- - ... - I 45 JK Na t1 va • 8 ii

Figure 3.2 b. P. pandorana <• locations from Nat. Herb. TAS) 50

11-Mar-9 Dld Herbarium Distribution Map for Species: Pendorea nervose - r - - - - r - - - - r l;J .§P!' c2 men a I I I - - r r--- -Ito r 1 I I I I I I I --~- ___ I I ---~-- I -r- r I 15 I I I' I I I I I I I --~--- r----r-- I 1-- I 1 I I - - co I 1 I I 1 I I I '----~----r----r-- I ----r I r 1 - ,

I I I r -I ~o I I I I

I I .... ---... N ... 0 .... Ul 0 SIMPLE CONIC PROJECTION SCALE 1: 25 000 000 o Accurate to 1 dag. + CuI tJ va tad ---- I lN Native :- --- <15 01"' ~ m -----··------· ·------· Figure 3.2 c. P. nervosa 51

11-Mar-9 Old Herbarium Distribution Hap for Species: Pendoree balleyena ----~----r-- - - r - - -- r -- - - r -· --- r5_SJ?8C i mens -- ( I I I~ I I --r-- r 1 I I I - -,zo I I

I r-- I I --I J5 ( I I I I

I I I I -,25 I

I I I r- I -,:30 I I I I I I I 1 I I 1 \ - ---- r - - - - r - - - - r -c:::» 1 ',- --- r I I I 1 I I I I I I 1 I I 1 I I I ------~~~P~: ~:N~~ ~:o~~c~:o~---- r-- -N ~f--- -/ •o SCALE 1:25 000 000 :v I I o Accurate to 1 deg. _____ I I 1 ·~Cultivated --- 1 )( Net 1 v B b (j, --- 45 g: u, U1

Figure 3.2 d. P. baileyana 52

The lower part of the pistil consists of a superior ovary surrounded by an annular nectary. The style is elongated and the wet stigma has two lobes which are covered in papillae (Fig. 3.4, p53).

3.2.3 Floral dimensions Flowers representative of the species P. jasminoides, P. pandorana, P. nervosa and P. baileyana are illustrated in Figs 3.5 - 3.8 (pp54-56). Their floral dimensions (Fig. 3.9, p57) have been estimated for 10 genotypes of P. pandorana, six genotypes of P. jasminoides and a single genotype each of P. nervosa and P. baileyana (Table 3.2, p58).

Calyx and corolla tube length were similar for P. pandorana and P. baileyana. Stamen length was similar for P. jasminoides and P. nervosa. All other dimensions measured were quite distinct between the species. Within species, individual genotypes were distinctive when floral dimensions were considered in total, although the dimensions of single parameters were not always significantly different. A scatterplot of the means of corolla tube length against outer corolla width plotted for each genotype for which floral dimensions were measured, shows no overlap between species (Fig. 3.10, p59).

3.3 THE AUSTRALIAN SPECIES OF PANDOREA

3. 3.1 Key to tlze Australian species The species can be separated readily on floral characteristics but leaf differences also occur. Pando rea pandorana is the only species in which floral differences can be quite marked but the characteristics do not overlap with those of other species. A key to the Australian species is included (Fig. 3.11, p63). A key for all species of Pandorea can be found in the revision of van Steenis (1977). Genotypes of each species used in this study are listed (Table 2.1, p34). Figure 3.3 iY1ature fruit of P. jasminoides showing the arrangement of winged

seeds (s) and the position of the septum (arrowed) perpendicular l: the plane of capsule (c) dehiscence. Bar == lcn1.

Figure 3.4 Stigma of P. jasminoides stained with toluidine blue showing the two lobes, the lower one recurved, covered in dark-staining papillae, Bar ::: 0.5mm. 53

3.3

- Figure 3.5 Two of the genotypes of P. jasminoides used for the measurement of floral dimensions. a-c. P. jasminoides(HW) a. front b. side c. top d-e. P. jasminoides(AL) d. front e. side f. top. Bar = lcm. 54 Figure 3.6 Three of the genotypes of P. pandorana used for the measurement of floral dimensions. a-c. P. pandorana(BF) a. front b. side c. top d-f. P. pandorana(TSl) d. front e. side f. top g-i P. pandorana(Ts4) g. front h. side i. top. Bar = 5mm. 55 Figure 3. 7 Genotype of P. nervosa used for the measurement of floral dimensions. a. front b. side c. top Bar = lcm ..

Figure 3.8 Side views of two genotypes of P. baileyana(Pb2, Pb3) con1parcd to P. nervosa and P. pandorana(VC). Bar = lcm. 56 ------~~--~£D~---~f.U.mlBE~-~--IImB&#•m--mWM~

57

G

c D

E

Figure 3.9 Floral dimensions measured for genotypes of each species of Pandorea. A. Calyx length B. Corolla tube length C. Corolla outside length D. Corolla width • inner E. Corolla width .. outer F. Corolla depth . inner G. Corolla depth - outer H. Pistil length I. Stamen length. (Bar = lcm) Calyx Corolla lube Corolla Corolla width Corolla width Corolla depth Corolla depth Pistil Stamen length length outs ids -mner -outer -mner -outer length length (mm) (mm) length (mm) (mm) (mm) (mm) (mm) (mm) (mm)

P.pandorana X X X X X X X X X vc 2.0 16.0 21.5 4.2 15.8 4.0 15.0 15.6 11.9 GS 2.4 15.0 20.9 3.9 13.3 4.0 12.8 14.5 11.6 MM 2.1 11.1 15.1 3.2 12.6 3.1 11.7 12.7 10.8 BF 2.0 15.2 21.1 4.7 18.8 4.5 16.0 16.6 12.0 TS3 2.0 10.6 15.6 4.9 16.3 4.9 15.0 11.5 8.9 TS6 2.0 13.3 16.8 4.7 13.9 4.3 13.2 14.2 11.2 TH 3.0 14.1 19.5 4.0 13.4 3.8 12.1 15.5 11.2 SCH 2.0 10.1 13.4 4.0 12.7 3.7 11.7 11.1 9.0 AF 3.0 12.7 17.3 3.6 13.6 3.5 12.2 13.0 9.3 AL 2.9 13.3 18.2 6.0 16.0 5.4 14.6 14.0 10.9 L.S.D. (0.2) (0.4) (0.6) (0.3) (0.7) (0.3) (0.6) (0.6} (0.4}

P.jasminoides ATHl 7.1 32.8 54.8 19.3 51.3 20.6 49.4 21.3 18.7 LD 6.9 35.6 61.6 28.6 63.9 31.2 60.0 22.5 18.8 H\V 5.7 32.1 56.5 24.9 58.3 26.7 55.8 25.5 16.0 AL 6.3 31.5 60.1 30.3 65.5 31.7 61.2 24.1 19.6 F827 6.4 31.0 56.2 28.1 59.5 29.6 56.8 23.6 18.8 DKP 8.3 40.3 70.2 31.3 72.8 31.8 70.4 30.8 20.0 L.S.D. (0.3) (1.2) (2.2) (1.4) (2.9) (1.4) (2.9) (0.6) (0.5)

P.baileyana (s.e.m) (s.e.m) (s.e.m) (s.e.m) (s.e.m) (s.e.m) (s.e.m) s.e.m) (s.e.m) 2 2.6 (0.1) 12.8 (0.2) 13.6 (0.3) 1.5 (0.1) 5.9 (0.1) 1.2 (0.1) 5.9 (0.1) 8.5 (0.1) 7.2 (0.1)

P.nervosa 6.5 (0.1) 28.9 (0.2) 38.5 (0.2) 9.2 (0.2} 24.8 (0.4} 9.1 {0.2} 24.1 (0.4) 17.5 (0.2) 20.9 (0.2)

..... " WH ...... J n • ~- ~ .1 -- n L ~ ~ ...... --- - ..l n ( ~ ~ ~~~~..t~-..t ~f - - --·· ..... -- ---r..- ...... mean. x = mean value for 20 flowers per genotype. ~ I -p------~--~~~~--~-=-m==--,~---~~---~~·t· ,,

59

~ r------~

40

35 _....., E ~ ...:::::.. 30 .c...... CJ) • c CD - 25 ..aCD ...... ::J ctl = 20 a...... ua 15 • 10 *X *

5~------~ 0 20 40 60 80 Corolla width outer (mm).

P.p P.j P.b P.n >..: A • •

Figure 3.10 Plot of corolla tube length against corolla outside width for each genotype used in floral dimensions measurements showing the separation of species. (Data points = mean of 20 flowers. Data from Table 3.2) .,~------.--.ama•••---•am-.awo·wa-~

60

3.3.2 Description of the Australian species

P. jasminoides(Lindl.)K.Sch. (Fig. 3.1a, p47) This species was described as Tecoma jasminoides in 1838, firstly with an illustration by John Lindley and then a botanical description by G. Don. It remained unchanged until 1894 when Schumann assigned it to the genus Pandorea.

Description: Tall, glabrous woody climber. Leaves 2-jugate. Leaflets entire, ovate to lanceolate, blunt at apex, nerves and veins not prominent above, dark glossy green. Calyx campanulate, truncate or obscurely toothed, 6 mm long. Corolla 4-5 em long, white to pink with crimson throat. Corolla tube > 2.5 em, much more dilated upwards than in P. pandoran.a, the lobes very broad, more than half as long as the tube, the throat scarcely bearded inside. Fruit as for P. pandorana, may be larger, but seeds broader, the wing either entirely surrounding them or chiefly on the 2 sides.

P. pandorana(Andrews)v .St. (Fig. 3.1 b, p47) This species was first described as Bign.onia pandoran.a by Andrews in 1801, then as Tecoma australis in 1810 by Robert Brown. In 1839, Endlicher described Pan.dorea as a section of the genus Tecoma. Spach renamed the species Pandorea australis raising the section to generic rank in 1840. It was finally renamed P. pandoran.a by van Steenis in 1929, the specific epithet of Andrews having precedence over that of Robert Brown.

Descrip6on: Tall glabrous, woody climber. Leaves usually 3 to 9-jugate. Leaflets ovate to ovate-lanceolate, or almost linear in arid zone variants, entire or less often coarsely crenate, 2.5 - 7.5 em long, glabrous, venation not prominent. Flowers 1-2 em long in loose, terminal panicles. Calyx smooth, 2-3 mm long. Corolla colour variable, white, yellow, orange, often streaked inside with purple or red. Corolla tube 1-2 em long, slightly curved and dilated upwards, lobes not one-third as long as the tube, the throat bearded inside on the lower lip, no hair-ring inside near ~p------.-,9-JBa-wuaum.-.m&ca~-~

61 insertion of stamens. Capsule 3-7 em long, usually acute at both ends, the valves hard and very concave. Seeds very flat, obovate, surrounded by a broad, papery wing.

P. ne!1'osav.St. (Fig. 3.lc, p47) First described by van Steenis in 1931 , it is allied to P. jasminoides.

Description: with slender, ribbed, purple stems. Leaves with 5 leaflets, dark green, ovate oblong, 2.5 - 6 em long and 1.5 - 4 em broad. Flowers odour less, showy. Calyx glabrous, campanulate, 6-7 mm long, splitting irregularly into lobes up to 2.5 mm. Corolla white, the tube yellow inside, funnel-shaped, slightly curved, 3.5 em (excluding lobes). Fruit and seeds not recorded.

P. baileyana(Maiden and Baker)v. St. (Fig. 3.1 d, p4 7) Maiden and Baker (1895) first described Tecoma baileyana in honour of Mr. Fred Manson Bailey, the contemporary government botanist. Van Steenis renamed it P. baileyana following his revision of the Queensland Bignoniaceae in 1929.

Description: Tall, glabrous, woody climber. Leaves 3-jugate. Leaflets usually 7, elliptic - oblong or obovate, 6 x 12.5 em, coriaceous, pale beneath, reticulations very prominent below but less marked above, margins slightly recurved. Flowers in a long raceme or spike. Calyx glabrous, 5- toothed. Corolla tubular, 1.5 em long with 5 very short lobes, not dilated upwards, cream with pink throat. Fruit and seeds not recorded.

The family is being reviewed for the ''. Whilst no major changes are anticipated for the Australian members of the family, it is thought that the P. pandorana complex may include a taxon from southern Queensland which should be given species status, tentatively P. jloribunda, (G. Guymer, personal communication). 62

3.4 COMMERCIAL FORMS OF PANDOREA

The popularity of P. jasminoides and P. pandorana has led to the selection of horticulturally desirable forms which are floriferous with aesthetically appealing flowers. The horticulturally desirable features include a heavy, massed floral display which continues for a number of weeks, a vigorous growth form, ease of propagation from vegetative material and a low susceptibility to pests and diseases.

Pandorca jasmin.oides is available in the standard pink form ("Bower of Beauty"), a white form ("alba" or "Lady Di ") and a variegated leaf form ("Charisma"). Pandorea pandorana is available in many forms often depending on the characteristics of the local plants. Two commercially available forms are "alba" or "Snow Bells" with white flowers and "Golden Showers" with small bright yellow flowers. Another variant with highly dissected leaves can be obtained but is often sold under its old name of P. doratoxylon. Pandorea pan.dorana is occasionally found as an upright shrub in arid habitats but no selection appears to have been done on plants of upright habit. The very limited ranges of P. n.ervosa and P. baileyana have restricted their availability and so they have yet to be tested under a range of environmental conditions. They may well prove to be as adaptable as the other two species and certainly have features of horticultural merit.

3.5 SU.MMARY

The Australian species of the genus Pandorea are well-defined except for the P. pandorana complex which is currently undergoing revision (G. Guymer, perso.nal communication). It is thought that a fifth species, tentatively P. jloribunda, warrants separation from the species, P. pandoran.a as defined by van Steenis (1977). The small ranges of P. nervosa and P. baileyana make them potentially vulnerable but at this stage their habitats are not under threat. jiLP '' .§W

63

Fig. 3.11 Key to the Australian species of Pandorea

1. Corolla tube glabrous outside, usually rather wide, without hair ring inside near the insertion of the stamens. Base of filaments glandular, ovary with similar sessile glands. Calyx cupular, c. 2-3mm. Venation not prominent above. Filaments inserted close to the base of the tube which is not narrowed ...... P. pandorana

Corolla tube (Except at very base) outside pubescent with fine, soft hairs. Base of filaments without sessile glands. Venation mostly prominent ...... 2

2. Flowers large, 4-Scm long (incl. lobes), white with crimson throat, the tube c. twice as long as the rounded lobes. Leaves 2-jugate, the entire leaflets blunt at the apex, nerves and veins not distinctly prominent above. Queensland and New South Wales ...... ••.. . P.jasminoi'.des

Flowers much smaller, at most 3 .Scm, the tube 4-6 times as long as the lobes; lobes less than c. 1em diameter. Leaflets acute, nerves and veins usually distinctly prominent above, mostly toothed towards the apex ...... 3

3. Corolla 12-15mm long, narrow cylindrical, the tube c. 6 times as long as the lobes. No hair ring at stamina] base. Calyx 3-5mm. above at base with one or few large glands; leaf rachis narrowly winged. Leaflets entire, 3-4- jugate, 5-12 by 2-Scm. Corolla c. 12-13mm, tube ± curved, c. 2-3mm wide, no beard in mouth and tube, lobes c. 2mm. Ovary orbicular, with sessile glands. Venation on upper surface of leaflets raised but not fine-tessellate. Flowers

cream coloured, lobes and throat may or may not be pink-s~_aded. Queensland ...... P. baile yana

Cqrolla c. 30-35mm long, funnel-shaped, white with pale yellow mouth, ± mauve markings, with a fairly narrow lower part of the tube, widened apically, inside at the insertion of the stamens with a few hairs. Ovary ellipsoid, with sessile glands. Pedicel c. Smm. Queensland .... . P. nervosa. ------·~··------~~-~

64 CHAPTER 4

SOME ASPECTS OF THE REPRODUCTIVE BIOLOGY OF THE GENUSPANDOREA

4.1 INTRODUCTION

Selection of genotypes with varied floral and vegetative characteristics such as flower size, colour and leaf variegations followed by intra- and interspecific hybridization has been an i1nportant component in the horticultural development of elite ornamental crops. However, few Australian native species used in ornamental horticulture have been intensively selected or bred for the horticultural industry and in general, little is known of the breeding systems. More recently, identification of the potential of the Australian flora for both the local and export markets has resulted in an increased interest in the selection, breeding and management of endemic Australian plants including Thryptomen.e calycina (Beardsell et. al., 1988), Banksia (Elphick, 1985; Fuss and Sedgley, 1990, 1991b), Anigozanthos (Watkins, 1985), Blandfordia (Zimmerman and Pyke, 1988) and Telopea (Faragher, 1986a, 1986b, 1989).

The ain1 of this series of experiments was to define the reproductive structures and breeding system of Pandorea species so that the periods of female receptivity and male viability were understood prior to interspecific hybridization of selected genotypes to produce elite material for the horticultural industry.

4.2 RESULTS AND DISCUSSION

4.2.1 Reproductive structures -pistil

4. 2.1.1 Pistil structure The ovary of Pandorea is superior and divided into two locules containing numerous ovules. The ovules are tenuinucellate and attached anatropously to the placenta. The 65 style is hollow with a central canal (Fig. 4.1, p66) which appears to be filled with mucilage in fresh sections.

The receptive stigma of Pandorea conforms to the wet-papillate type described by Heslop-Harrison and Shivanna (1977) and Heslop-Harrison (1981). The papillae are unicellular and are only present on the inner surface of the stigma lobes. For the purposes of this study, the stigma lobes were considered to be separating if there was a visible divergence of the tips of the lobes. At this stage, the lobes could be separated further by the use of forceps and pollinated artificially. An thesis was operationally defined to be when the corolla lobes had begun to unfurl although the corolla did not open fully for some hours (Fig. 4.2, p66).

The stigma lobes of P. jasminoides are usually closed when the corolla opens but separate soon after, exposing the receptive surface. In P. pandorana, the stigma lobes have begun to separate before an thesis (Table 4.1, p67). The lower lobe, in particular, of all species recurves as the flower ages (Fig. 4.3, p66). The stigma lobes are sensitive to mechanical stimulation and if a flower is pollinated, the lobes close permanently, usually within 5 - 30 seconds. The mechanism consists of two parts, a relatively quick response to the stimulation, followed by a slower resetting (2-5 min) period during which response to stimulation is either n1inimal or does not occur. The presence of stigmas sensitive to mechanical stimulation has been recorded for other species of the family Bignoniaceae (Newcombe, 1922, 1924; Stephenson, 1979; Bertin, 1982; Petersen et. al., 1982) and for other families including Martyniaceae, Scrophulariaceae, Acanthaceae and Lentibulariaceae (Newcombe, 1924).

Several reasons have been given for the stigma closure. The most commonly cited is that closure prevents the deposition of self-pollen on the stigma lobes as the pollinating insects backs out of the flower because the receptive inner surfaces of the stigma are no longer exposed (Darwin 1876: Elrod, 1904; Newcombe, 1922: Thieret, 1976). Alternatively, that it provides a humid environment for pollen germination (Newcombe, 1924) and prevents removal of pollen by insect foragers as opposed to pollinators. Figure 4.1 Transverse section (2 JtiD) through the style half-way between the stigma and the ovary showing central canal (c) lined with epithelial cells (arrowed) vascular bundles (v), single-layered epidermis (e) and multicellular glands (g), stained with toluidine blue. Bar = O.lmm

Figure 4.2 Flower of P. jasrninoides(WS 1) at an thesis with petals numbered in order of opening. Bar = 5mm.

Figure 4.3 Scanning electron micrograph of an aged stigma . of P. pandorana with recurved lobes and papillae (arrowed) on the receptive inner surfaces. Bar = lOOp.m...... i C£ 424Zt

66 ~p------·-"·--~==------~----·~-~~=mw==-

67

Table 4.1 Proportion of flowers of P. jasminoides (genotypes HW, AL, F827) and P. pandorana (genotypes VC, GAS) with separating stigma lobes and dehisced anthers prior to and at anthesis. Observations made on plants grown in a glasshouse unless otherwise indicated. (n =25)

Bud (12-24h before anthesis) Flower at anthesis Species and Genotype Proportion of Proportion of Proportion of Proportion of buds with buds with flowers with flowers with separating dehisced anthers separating dehisced stigma lobes stigma lobes anthers

P. jasn1inoides

HW 0 0 0.04 0.08 AL 0 0 0.00 0.12 F827 0 0 0.04 0

P. pandorana

vc 0.84 0 0.88 0 VC (Field) 0.76 0 1.00 0.08 GAS (Field) 0.88 0 0.92 0.04

The basis of the mechanism has been studied in Stylidium (Stylidiaceae) in which the fused style and filaments, the column, undergo a rapid firing movement when mechanically stimulated, usually by the action of pollinators (Findlay and Findlay, 1989). The n1ovement of the column following mechanical stimulation has been attributed to specialized anatmny of the cells and tissues in a particular region, the 'bend', of the column. The physiological basis to the movement is that the stimulation causes changes in the distribution of chloride and potassium ions in the moving region of the colun1n. Those changes result, in turn, in changes in turgor of the motor cells (Findlay and Pallaghy, 1978; Findlay and Findlay, 1984 ). The rate of firing of the column in Stylidiwn is limited by passive physical processes (Findlay, 1978). The resetting process, on the other hand, is closely linked to the metabolic energy supply as it does not occur under anaerobic conditions when respiration is inhibited (Findlay and Findlay, 1981). The time taken for stigma closure in Pandorea species, Proboscidea louisian.ica (Thieret, 1976), Chilopsis lin.earis (Petersen, 1982) and Catalpa speciosa jiiiS HIIOODII!liilit.i n 'liillllll

68 (Stephenson, 1979) is similar (2-60s) but much slower than Stylidium (15-30ms). The resetting process takes approximately 5min in each case and so may involve the same active ion pump process.

The resetting process is also influenced by the amount of pollen applied to the stigma. The stigma lobes of Catalpa speciosa (Stephenson, 1979) and Proboscidea louisianica (Thieret, 1976) remain closed if the pollen load is heavy but will otherwise re-open. Another species, Hibiscus trionium (Malvaceae), has a style split into five branches which, whilst not sensitive to mechanical stimulation, recurve progressively downwards towards the dehisced anthers as the flower ages (Buttrose et. al., 1977). The curvature can be counteracted, however, by a readily translocated signal produced as a result of pollination. A similar signal may operate in C. speciosa when the pollen load on stigmas is high. If sufficient pollen germinates on the stigma lobes the curvature in H. trionium is stopped and self-pollination is prevented. The mechanism promotes outcrossing but allows for self-pollination if pollen transfer from another flower is low. Comparable sty lar curvature in Pavonia zeylanica has been interpreted as a mechanism to ensure self-pollination (Vasil and Johri, 1964) but can also be interpreted to allow autogamy if external pollen transfer is low. Spathodea campanulata (Bignoniaceae) requires both pressure and pollen for stigma closure to be complete. The rate of stigma closure for this species is too slow to prevent self-pollination but the distant position of the stigma to the open anthers and the protogynous nature of the flowers may minimize self-pollination (Newcmnbe, 1924). The stigma lobes of Pandorea rarely re-opened after experi1nental pollination. As only large amounts of pollen were applied it is not known whether pennanent closure is dependent on pollen load.

4. 2.1. 2 Sti!!,nUJ receptivity The stigmas of both P. pandoran.a and P. jasminoides are receptive prior to anthesis. Stign1as were scored qualitatively for pollen germination and growth of tubes to the base of the style (Table 4.2, p69). For P. pandorana, stigma receptivity was low one day before anthesis, complete for four days then started to decline. There was no significant decline in stigma receptivity in P. jasminoides over the time period studied. For the purposes of artificial pollinations, the stig1na of both species can be considered receptive for at least three days after anthesis. ~------~------~--~5'~--

69

Table 4.2 Pollen germination and growth on different aged stigmas of P. pandorana and P. jasminoides, genotypes GAS and HW, respectively. (n=3)

No. stigmas with germinating pollen grains Stigma age (tubes at base of style) (Days from anthesis) P. jasmin.oides HW P. pandoran.a GAS

-1 3 (3) 2 (1) 0 3 (3) 3 (3) 1 3 (2) 3 (3) 2 3 (3) 3 (3) 3 3 (3) 3 (3) 4 3 (3) 2 (2) 5 3 (3) 1 (1)

4. 2.1. 3 Separation of male and female function

Spatial Separation Mechanical self-pollination has not been observed in any of the four species of Pando rea studied. A comparison of pistil and stamen lengths of genotypes of all four species in chapter 3 (Table 3.2, p58) shows that it is unlikely that self-pollen could be transferred to the stigma of a flower in the absence of a pollen vector. Rarely, the lower stigma lobe n1ay recurve sufficiently to make contact with the upper anthers but there was no evidence of its occurrence in this study. Mechanical self-pollination of Catalpa speciosa (as C. bign.onioides) has been observed in a few instances (Newcombe, 1922) where pollen was deposited onto the stigma from open anthers pressed against the stigma lobes. Spathodea campanulata (Bignoniaceae), on the other hand, has widely separated anthers and stigma lobes making mechanical self-pollination very unlikely.

Temporal Separation Pandorea pandorana is protogynous (Table 4.3, p71) but there is no clear temporal separation of male and female function in P. jasntinoides under glasshouse conditions. These results are in contrast to those for Campsis radicans (Bertin, 1982) which is 70 markedly protandrous. In the latter species, anthers dehisce about 12h prior to the separation of the stigma lobes. Due to the minimal spatial separation observed by Bertin (1982), of the stigma and anthers in Campsis radicans, large amounts of self-pollen would be deposited on the stigma if the lobes were open as the anthers dehisced. Temporal separation of the male and female phases promotes outcrossing in C. radicans because insects remove all self-pollen from flowers before the stigma lobes separate (Bertin, 1982). Considerable amounts of self-pollen may still be deposited as a result of geitonogamous pollinations (between flowers on the same plant). This is also likely to be true for P. jasnzinoides, P. pandorana and P. baileyana because many flowers are open on each plant during their peak flowering phases. This may be a partial explanation for the low fruit set observed for the family as a whole (van Steenis, 1977) if self-incompatibility is found to be prevalent. Outcrossing in mass-flowering plants like Pandorea is greatest at the start or finish of the flowering season when interplant visits by foraging pollinators are more frequent (Stephenson, 1982).

There was no significant difference in the time to abscision of the corolla for the five genotypes of P. jas1ninoides tested. The time to corolla drop was 4.1d, 4. 7d, 4.5d, 5 .Od and 4.7d (L.S.D. = 0.9, p=0.05) for the genotypes PjLGF, PjF827, PjDKP, PjAL and PjLD, respectively.

The timing of stigma lobe separation has been compared between P. pandoran.a and P. jasminoides. When buds, estimated to be 12-24h prior to anthesis, and flowers at an thesis were compared, the time of stigma lobe separation was significantly different between the two species (residual deviance of 276.78, X/ =6.63, p<0.01). Pandorea pandorana stigmas had already begun to separate in the bud whereas those of P. jasminoides were still not separating at an thesis. Within species, however, there was no significant difference in the number of stigmas with separating lobes between buds and flowers at anthesis. Nor were there any significant differences between genotypes within species.

For both species, there was a small but significant difference in the number of flowers with dehisced anthers at anthesis compared to in bud (residual deviance of 11.47, X/= 9.49, p < 0. 05). There was no difference between or within species for the ~~------·~------~--~------£----~

71 genotypes studied. For practical purposes, whilst the time difference was statistically significant, few flowers had dehiscing anthers at anthesis.

Table 4.3 Mean time to stigma opening, anther dehiscence and corolla abscision for P. pandorana on four separate occassions for genotype GAS (field­ grown) and single occassions each for genotypes MM and VC (glasshouse-grown). (n =number of replications).

Genotype Mean time to Mean time to Mean time to Min/max stigma opening anther dehiscence corolla drop (d) temperature (d) (d) ec)

GAS3 n=20 0 2.7 6.9 11.1/18.5 GAS4 n=20 0 3.6 5.3 12.3/18.1 GAS6 n=20 0 3.2 5.8 12.4/18.7 GAS13 n=10 0 1.3 3.9 10.5/21.5

MM n= 16 0 *nd 4.4 18.0/27.0

vc n=19 0 *nd 4.5 18.0/27.0

(L.S.D., (0.4)** (0.5)*** p=0.05)

*nd = not done. ** L.S.D. = 0.5 for comparing genotype GAS on occassions 3, 4 and 6 with occassion 13. ***L.S.D. = 0.6 for comparison between GAS13 and all others.

4. 2. 1. 4 Pollen-Pistil Interactions in selfed and outcrossed flowers Pollen tube growth patterns are shown for P. jasmin.oides (Table 4.4, p72), P. pandoran.a (Table 4.5, p73) and P. baileyan.a (Table 4.6, p73). An F-test for each species showed no overall significant differences between pollen tube behaviour for any cross and in any species. Presence of ppllen tubes in ovules could not be scored from pistil squashes. In almost all pistils for all treatments, pollen grains had germinated on the stigma. Quantitative comparisons were not possible due to the closure of the stigma following pollination but on general observation there appeared to be no difference in the ability of self- or outcross pollen grains to germinate on the stigmas. Pollen tubes grew to the base of the style indicating that there is no incompatibility barrier in the style. There was no detectable difference in the number of pollen tubes reaching the ovary --~·uro.ar~-"""'______..... ______

72 compared to the number at the base of the style for any treatment but as there were many pollen tubes, an accurate count was not possible so quantitative results could not be presented. There was also no difference between treatments on the presence of pollen tubes in the ovary. Again, it was only possible to record the presence or absence of pollen tubes in the ovary rather than to quantify the number of tubes present.

Late-acting self-incompatibility, where pollen tube arrest does not occur on the stigma or style, is common in woody plants (Seavey and Bawa, 1986). It also occurs in Pando rea. The self-pollen tubes and outcross-pollen tubes in P. jasminoides and P. pandorana both grow into the ovary but self-pollinations do not result in fruit set. The style of Pandorea species is hollow (Fig. 4.1, p66). Brewbaker (1957) postulated that self-incompatible plants with hollow styles commonly showed a delayed arrest in self­ pollen tubes because of the lack of intimate contact between the style and the pollen tubes. The observation of late-acting self-incompatibility in Pandorea is not in disagreernent with the small amount of literature available on the subject. Ovule clearing may be a useful technique to determine the site of the incompatibility barrier. It has been used successfully to identify a post-zygotic mechanism for Rhododendron (Williams et. a!., 1984).

Table 4.4 Pollen-tube growth in self- and outcrossed flowers of P. jasminoides (n=20).

Cross Observed proportion Observed proportion Observed proportion of stigmas with of styles with pollen of ovaries with germinating pollen tubes at base pollen tubes present

HW (self) 1.0 1.0 0.85 HW x LD 1.0 1.0 J.O HW x AL 1.0 1.0 0.90

LD (self) 1.0 0.95 0.85 LD x HW 1.0 1.0 0.85 LD x AL 1.0 1.0 1.0

AL (self) 0.95 0.90 0.80 ALxHW 1.0 1.0 0.95 AL x LD 1.0 1.0 0.85 73

Table 4.5 Pollen tube growth in self- and outcrossed flowers of P. pandorana (n=20).

Cross Observed proportion Observed proportion Observed proportion of stigmas with of styles with pollen of ovaries with pollen germinating pollen tubes at base tubes

BF (self) 1.0 1.0 0.95 BF x GS 1.0 1.0 1.0 BF x AL 1.0 1.0 1.0

GS (self) 1.0 0.95 0.95 GS x BF 1.0 0.90 0.90 GS x AL 1.0 1.0 0.95

AL (self) 0.95 0.95 0.95 AL x BF 1.0 1.0 1.0 AL X GS 1.0 0.95 0.95 AL x SCH 1.0 1.0 1.0

Table 4.6 Pollen-tube growth in self- and out-crossed flowers of P. baileyana (genotypes 2, 3). (n=20).

Observed proportion Observed proportion Observed proportion Cross of stigmas with of styles with pollen of ovaries with pollen genninating pollen tubes at base tubes present

2 (self) 1.0 0.95 ! 0.95 2 X 3 1.0 1.0 0.95 3 (self) 0.95 0.95 0.95 3 X 2 1.0 1.0 1.0 ......

74

4.2.2 Reproductive structures - Pollen

4. 2. 2.1 Pollen viability Pollen viability was estimated by two different methods for both P. jasminoides and P. pandorana. Firstly, the FCR score was highest either at or just prior to an thesis and declined over a five day time period to 8-10% (Fig 4.4, p75). At the end of this time, the corollas, to which the stamen filaments are attached, shed in both species. Secondly, pollen from the flowers used in the FCR test was also used to pollinate stigmas from flowers at an thesis. For P. jasmin.oides, pollen germinated and pollen tubes grew down to the base of the style irrespective of the ages of the flowers from which the pollen was collected. Pollen germinated on the style (Fig. 4.5, p76) and pollen-tubes could be seen to enter ovules in some instances (Fig. 4.6, p76). For P. pandoran.a, the ability of pollen to germinate deteriorated from 3d after an thesis (Table 4. 7, p77) and correlated with an FCR score of 15%. The difference between the fit of the models to the raw data for the two species is due to the higher variability in the data for P. pandoran.a as a result of the plants being field-grown whereas P. jasminoides was grown under controlled glasshouse conditions. The length of time pollen retains its viability is likely to be related to the average time taken for the complete removal of pollen by insect visitors (Bertin, 1988). The pollen of Campsis radicans (Bertin, 1982) was collected by pollinators within 2h of anthesis and presumably transferred to stigmas shortly after. Its longevity has not been recorded. For plants growing where pollinator activity is unreliable, pollen longevity may assume greater importance in terms of reproductive success. Banksh1 spinulosa (Proteaceae) has a pollen viability of 50% 8 days after anthesis. For that species, pollen removal is incomplete and takes place over several days (Ramsey and Vaughton, 1991). In contrast, B. n1enziesii pollen is removed within a few hours of anthesis and has few viable grains remaining after 24h (Ramsey and Vaughton, 1991). Pando rea pandoran.a frequently flowers when temperatures are low (15°C) and rain is common (Jaxnes, unpublished observations). From this study, the pollen longevity suggests that pollen remains in the flowers for some time. Pan.dorea pandorana is protogynous but the female phase overlaps with the male phase for some days. The absence of ten1poral separation in P. jasmin.oides suggests that self-pollination is minimized by the spatial separation on anthers and stigma. ADDENDUM

Examiner #2

3. The FCR test showed diminishing pollen viability with increased time after anthesis, but this was not reflected in the in vivo test. This apparent anomaly can be explained by considering the sensitivities of the two tests.

The FCR test allowed a quantitative analysis of pollen viability based on the integrity of the membrane within each pollen grain. The in vivo test, however, was qualitative. Closure of.the stigma lobes after pollination prevented a comparison of the number of germinating and non-germinating pollen grains. Therefore, if pollen tubes were found in the style, the pollen was considered to be viable. As the pollinations were done with large amounts of pollen, a decrease in the number of viable grains with increasing time would not be apparent until the number of pollen tubes in the style was very low (say < 30). For practical purposes it may mean that despite a decrease in pollen viability with increased time after anthesis, pollinations with pollen from older flowers may be effective provided that the amount of pollen transferred is high. .. _,

:~~ " 75

0.8 ...... ~ ....._, 0.7· a (1) r::C) (1) C.) 0.6 (1.) (1) ~ 0 0.5 0 ...... ~ ·ar:: 0.4 S-4 ~ ...... 0.3 0 ~ ...... ,.> ...... ~ 0.2 ..0

1.0 . ..--...... _,~ 0.9 b (1) (.) ~ 0.8 (1) C) Cll QJ 0.7 r... 0 ~ ~ 0.6 ~ Q -~ 0.5 r... 0.0

~ 0.4 0 t:>-...... ,J 0.3 .....~ ..c ([j 0.2 ..c 0 r... 0.1 ~ 0.0 "' -2 -1 0 2 3 4 5 6 Flower age (days from anthesis)

Figure 4.4 Decline in pollen grain FCR fluorescence (pollen viability) over time. a. P. pandorana eqn: Iog(p/1-p) = 0.29 - 0.49x b. P. jasminoides

eqn: log(p/1-p) = 0.23 - 0.95x , where x = days from ~nthesis. Figure 4.5 Light micrograph of pollen from P. jasminoides flowers, 3 days after anthesis, germinating on a receptive stigma. g = pollen grains, pt = pollen tubes. (x 100).

Figure 4.6 Light Jnicrograph of an outcross pollen tube (pt.) of pollen, from P, jasminoides flowers 3 days after anthesis, entering an ovule (o) of P. jasminoides. Micropyle arrowed. (x 400). 76 ~------~-~·--•~m--~~~~--·a~~

77

Any chance self-pollinations through geitonogamy may be prevented from forming fruit by the presence of self-incompatibility which forces outcrossing (see Chapter 5).

Table 4.7 Qualitative ability of pollen of P. jasminoides (genotype HW) and P. pandorana (genotype GAS) of varying ages to germinate and grow down to the base of the style. (n=3)

Age of flowers used as No. styles with germinating pollen grains pollen source (no. styles with tubes at base of style) (days from anthesis) P. jasminoides HW P. pandoran.a GAS

-1 3 (3) 3 (3) 0 3 (3) 3 (3) 1 *nd 3 (3) 2 3 (3) 3 (3) 3 3 (3) 3 (3) 4 3 (3) 2 (2) 5 3 (3) 1 (1)

;end = not done

4.2.2.2 Pollen structure Pollen grains of Pando rea species (Fig. 4. 7, p83-84) are tricolporate with finely

reticulate exine and are approximately 35 ~-tm in diameter. This type of grain is considered to be one of the less specialized types from which have developed the more advanced types in the family Bignoniaceae (Gentry and Tomb, 1979). The exine of P. jasminoides consists of a foot layer, columellae and partial tectum and can be distinguised from the intine which comprises two layers, the outer layer being more electron dense (Fig. 4. 8, p85).

The pollen of P. jasmin.oides and P. pandoran.a have been previously described by light microscopy and limited SEM (Buurman, 1977; Suryakanta, 1973). The tentative exine structure described by Buurman (1977) from light microscope observations, consisted of an endexine (sensu lato), a layer of columellae and a tectum. TEM 78 observations of P. jasminoides support those observations (Fig. 4.8, p85). Rifts apparent in the colpus membrane (Fig. 4.9, p86) are similar to rupture patterns described by Buurman (1977) from both light microscope and SEM observations. Although Buurman ( 1977) mentions the possibility of rupture patterns in the col pus membrane being artefacts, that author considers that they originate during the ontogeny of the pollen grains and are probably caused by the mechanical forces acting on the developing pollen grain. Whilst mechanical forces during ontogeny may predispose the col pus membrane to rupturing in a certain pattern, the actual rupturing is considered, in this study, to be an artefact of preparation. Cranwell (1962) also mentions "rift-like" apertures in the colpus membrane of Tecomanthe speciosa.

In unfixed samples, pollencoat materials within the columellae can be seen (Fig. 4.10, p86). The air-dried, non-hydrated pollen grains of P. jasm.inoides are contorted (Fig. 4.7b, p83) whilst for the other species they are not (Fig. 4.10, p86). The colpus membrane is not clearly evident due to the closure of the colpi in the non-hydrated state. The difference in shape between hydrated (Fig. 4.9b, p86) and non-hydrated grains (Fig. 4.1 Ob, p86) is typical of harmomegathic responses reported for other species.

The grains contain large amounts of starch at maturity and so belong to the starch-rich type (Fig 4.11, p87). This is typical of wind-pollination but the flowers of Pando rea require a biotic vector, probably a bee or hawkmoth (G. Sankowsky, personal communication). The presence or absence of starch at pollen shedding has been reported for a number of species (Baker & Baker, 1979; Franchi & Pacini, 1988). It has also been associated with developmental stage such that its presence is only transitory in some species, being correlated with differentiation and dedifferentiation of plastids (Pacini & Franchi, 1988).

The pollen grains of P.jasmin.oides are bicellular at anthesis as demonstrated by DAPI fluorescence of the nuclei (Fig. 4.12, p87) and toluidine blue-stained whole grains (Fig 12 inset). The vegetative nucleus with nucleolus (Fig 4.13, p87) and the generative cell (Fig. 4.14, p88) are often closely associated. Some TEM images were suggestive of a formal connection between the vegetative nucleus and the generative cell as has been found in Rhododendron (Kaul et. al., 1987). Microtubules were present in axial 79 orientation lying in an array immediately beneath the spindle-shaped generative cell (Fig. 4.15, p88). Microfilaments can be seen in the cytoplasm of the vegetative cell (Fig. 4.16, p88). There appears to be a connection between the generative cell and the pollen wall (Fig. 4.17, p89).

4.2.3 Pollen-ovule ratio The pollen-ovule ratios varied from 322-434 for P. jasmin.oides and from 220-461 for P. panfiorana. Numbers of pollen grains and ovules per flower and the pollen-ovule ratio were determined for four genotypes of P. jasminoides and two genotypes of P. pandorana (Table 4.8). The number of pollen grains per flower varied from 90 000-113 000 in P. jas1ninoides compared to 23 000-35 000 for P. pandoran.a with ovule number ranging from 250-312 and 76-97 for the two species, respectively. For P. jasmin.oides, pollen grain number, ovule number and pollen-ovule ratios were variable and showed statistically significant differences between some genotypes. However, the genotypes involved differed with the parameter being measured. For P. pan.dorana, the number of pollen grains, the number of ovules and the pollen-ovule ratios were all significantly different for the two genotypes measured.

Table 4.8 Number of pollen grains per flower, number of ovules per flower and the pollen-ovule ratio for four genotypes of P. jasminoides (LGF, HW, DKP, LD) and two genotypes of P. pandorana (GAS, GS).

Species/ genotype No. pollen grains No. ovules per Pollen-ovule ratio per flower flower

P. jasn1inoides

LGF 106 184 250.6 434 HW 90 760 254.2 356 DKP 113 261 287.4 396 LD 99 765 312.8 322 (L.S.D) (20 376) (41.2) (89)

P. pandorana

GAS 35 142 76.2 461 GS 23 109 97.0 220 (L.S.D.) (5 840) (10.0) (42) 80 Pollen-ovule ratios have been found to be highly correlated with the breeding systems of plants, with a general trend towards a lower pollen-ovule ratio as the degree of outcrossing decreases (Cruden, 1977; Preston, 1986; Lloyd, 1965)). They should, however, be viewed cautiously as indicators of breeding systems (Webb, 1984). The pollen-ovule ratios for Pandorea are much lower than expected for the type of breeding system suggested from other breeding system indicators such as floral morphology. Using the parameters for determining the outcrossing index (OCI)(Cruden, 1977), Pandorea species can be categorized as facultatively xenogamous. The pollen-ovule ratios for the two species of Pandorea examined, P. jasminoides and P. pandorana, fell between 434 and 220 which places those species at the lower end of the range for the facultative xenogamous species listed by Cruden (1977). A study of 66 Cruciferae taxa found that allogamous crucifers had pollen-ovule ratios typically greater than 3500 whereas those of autogamous crucifers were generally below 1000 (Preston, 1986). The genus Pandorea, however, has a strong self-incompatibility system (see Chapter 5). Presumably the species contain obligate outbreeders and must be pollinated by pollen of a different genotype from the female parent to effect fertilization of ovules.

The pollen-ovule ratios for obligate xenogamous species in the families Asclepiadaceae (3. 8-1 0.8) and Mimosaceae (246-863) are also low. Both families are exceptions in the xenogamous group because pollen occurs in aggregates (pollinia or polyads). The result in terms of breeding success for those species with aggregated pollen grains is that the reproductive return on each pollinium or polyad is low but if a pollination is successful, the reproductive return is high (Cruden, 1977). Polyads are present in only a few species in the family Bignoniaceae (Buurman, 1977; Gentry and Tomb, 1979) but they do not occur in the genus Pando rea. Therefore, the low pollen­ ovule ratio in these apparently obligate xenogamous species cannot be explained by a packaging of pollen which ensures that large quantities are transferred by each pollination. In Camps is radicans, a species in which a strong self-incompatibility system has been demonstrated (Bertin, 1982), the assumption of the species as an obligate outbreeder may be incorrect. Whilst self-pollinations rarely set fruit, pollinations comprising a mixture of self- and out-cross pollen resulted in 2-30% of selfed seeds in individual fruits (Bertin and Sullivan, 1988; Bertin et a.l., 1989). The comparative success of selfed and outcross progeny in unknown but the presence of significant 81 numbers of selfed progeny in a species which was considered highly self-incompatible provides support for the notion that the low pollen-ovule ratios for Pa.ndorea may reflect a level of self-fertility in the genus which has not been exposed because all experimental pollinations were limited to either self-pollen or outcross-pollen. Pollen-ovule ratios also reflect the efficiency of pollen transfer so that cleistogamous flowers which do not require a vector have the lowest pollen-ovule ratios. High pollen-ovule ratios may be required where the efficiency of pollen transfer is low, for example, for plants where the area of the stigma is small compared to the pollen-carrying area of the pollen vector (Cruden and Miller-Ward, 1981) so that the chance of the stigma touching an area of the vector which actually contains pollen, is limited. For vertebrate-pollinated plants such as Banksia the pollen-carrying area of the pollinators is large relative to the stigma and the pollen-ovule ratios are also large at about 10 000 (Ramsey and Vaughton, 1991). The low pollen-ovule ratios for Pandorea may also mean that pollen transfer is efficient.

In addition, the majority of plants where the pollen-ovule ratios were compared between autogamous and allogamous taxa have typically been herbaceous with few perennial taxa included. Thus, the comparison of the pollen-ovule ratios of Pandorea with those of plants such as crucifers (Preston, 1986) may be misleading in terms of a standard for each breeding system.

4.3 SUMMARY

The reproductive biology of the genus Pandorea allows artificial pollinations to be carried out from the time the anthers and stigma are accessible. Despite the difference in the temporal separation of male and female phases for P. pandoran.a (protogynous) and P. jasminoides under these experimental conditions, the stigmas of both species were receptive and could be pollinated prior to anthesis, provided that the stigma lobes could be prised apart to expose the receptive inner surfaces. Pollen from P. jasminoides deteriorated as the flower matured, with a viability of about 8% determined by fluorochromatic reaction, five days after anthesis. It germinated on receptive stigmas and grew to the ovules suggesting that those pollen grains which germinated were also capable of fertilization. The response of P. pandoran.a pollen was similar but its ability ------~-·"'!

82 to genninate on the stigma deteriorated after three days. Pollen structure was similar for each species and conforms to the tricolpate type in the family Bignoniaceae.

Pollen-ovule ratios were lower than expected for species which contain floral characteristics, such as thigmotropic stigma lobes and separation of male and female function in addition to a self-incompatibility system (chapter 5), which point to a reproductive system based on outcrossing.

The reproductive biology of the species does not pose problems for artificial breeding purposes before the level of the ovary. p

83

Figure 4.7 Tricolporate pollen grains of Pandorea species showing finely reticulate exine (e), position of col pi (c), col pus men1brane (em), and ruptures in the colpus membrane (r) a. P. jasminoides(liW) (fixed) b. P. jasminoides(HW) (air-dried) c, d. P. pandorana(GS) (fiXed) e, f. P. baileyana(2) (fl.Xed) g, h. P. nervo'sa (fiXed). Bar = lO,um. H4

Figure 4.7 e, f. P. baileyana(2) (fiXed) g, h. P. nenosa (fixed). Bar = lOJlm. 85

4.8

Figure 4.8 Tr·ansmission electron micrograph of a pollen grain of l'. jastninoides(HW) showing an exine structure comprising a tecturn (t), columellae (co), foot layer (0 and a double-layered intine (it, i2), the outer layer (il) being more electron dense. Figure 4.9 Scanning electron micrograph of whole, fiXed pollen grains showing ruptures of the colpus membrane. a. P. pandorana(GS) b. P. nervosa. Bar = lOJLm.

Figure 4.10 Scanning electron micrograph of unfixed, air-dried pollen grains with pollencoat materials visible within the columellae. a. P. baileyana (Bar = lp.m). b. P. nervosa (Bar = lJLm). 86 Figure 4.11 Transmission electron micrograph of starch granules (arrowed) present in the mature pollen of P. jasminoides(HW).

Figure 4.12 Light micrograph of bicellular pollen of P. jasnzinoides(HW) stained with DAPI. Generative cell (arrowed) fluoresces n1ore strongly than vegetative cell nucleus. (x 400). Inset: Whole pollen gr·ains stained with toluidine blue with generative cell (gc) in transverse plane, the more diffuse vegetative nucleus (vn) and nucleolus (n). (x 1 000).

Figure 4.13 Transmission electron n1icrograph of the vegetative nucleus (vn) and nucleolus (n) with a transverse section through the generative cell (gc) of a pollen grain of P. jasminoides. 87 4.11 -----·-·------·------illlllllll

Figure 4.14 Transmission electron micrograph of the generative cell in pollen from P. jasminoides(H.W) showing the nucleolus (n), nucleus (nu), cytoplasm (c) and generative cell wall (arrowed).

Figure 4.15 Transn1ission electron n1icrograph of microtubules (arrowed) lying in axial orientation in the generative cell.

Figure 4.16 Transn1ission electron rnicr·ograph of rnicrofihunents (arrowed) in the vegetative cell cytoplas1n. , ______------··---~------.. ·-··------·-·----,

88 ·------·-·~··

89

Figure 4.17 Transmission electron micrograph of the close association of the generative cell (gc) in the vegetative cell cytoplasm (cy) with the pollen wall (i = intine, e = exine). Generative cell wall is visible (arrowed). 90 CHAYTER5

ASPECTS OF THE BREEDING SYSTEM OF THE GENUS PANDOREA1

5.1 INTRODUCTION

Little is known of the breeding systems of Pandorea species. The presence of a self­ incompatibility mechanism in related genera was first mentioned for Bignoniaceae by Muller (1868) who found that flowers of Tecoma sp. failed to set seed when self­ pollinated. Exa1nples of species within the family Bignoniaceae in which self­ incompatibility has been demonstrated are shown in Table 1.6 (p28). Self­ incompatibility IS a genetically controlled mechanisrn which promotes outcrossing. Whilst it is usually considered to be a pre-fertilization barrier (de Nettancourt, 1977), it can also affect post-zygotic development in some taxa (Seavey and Bawa, 1986). This phenomenon is probably widespread throughout the family (Gentry, 1990) but there is a single record of a species, TecOJnan.the speciosa, which is apparently self-compatible (Hunter, 1967) and one case of self-fertilization in Catalpa speciosa (as C. bignonioides, Newcombe, 1922).

The aun of this series of experiments was to test for the presence of a self­ incompatibility system in P. pandorana, P. jasmin.oides and P. baileyana.

5.2 QUANTITATIVE ANALYSIS OF SELF-INCOMPATIBILITY, GENOTYPE AND RECIPROCAL EFFECTS

Breeding experiments consisting of intraspecific pollination diallels were used to determine the presence of self-incompatibility in P. jasnu'noides, P. pandorana and P. baileyan.a.

1 Material from this chapter has been used as the basis of an article: James EA, Thompson WK, Richards D, Knox RB (199-) Quantitative analysis of pollination diallels in two Australian species of Pandorea. (Accepted for publication in Theoretical and Applied Genetics). ,.,------·~----··--··-···--

91

5.2.1 Method of Analysis For the diallel crosses, a series of hierarchical models was devised to test, in tum, for the influence of self-pollinations, genotype combinations and reciprocal effects. Figure 5.1 represents two pathways of sequential models that can be used to fit the data from a 4 x 4 diallel. A full explanation of this figure is given below when the different models are described. The suitability of any model to describe and fit the experitnental data was judged on the significance of the residual deviance ("lack-of-fit") and, where appropriate, on the significance of the change in residual deviance when two models were compared.

The simplest model that can be used to explain any diallel data is one where the probability of fruit set is the same for all crosses (Null Model, Fig 5.1). In this case, the logit-link model can be expressed as: (Model 0)

log P LJ.. J=k ( 1-Pij 0

where P iJ is the probability of fruit set fr01n a cross of a female plant of genotype (i) and

a male plant of genotype (J) and ~' a constant, is the log-odds ratio of fruit set.

All subsequent models were derived to allow for different probabilities of fruit set between self-pollinations (geitonogamous) and outcross-pollinations (xenogamous). If it is assumed that within these pollination categories the probabilities of fruit set are constant then the appropriate n1odel has two parameters and is of the form:

(Modella)

where S, a constant, is the log-odds ratio of fruit set for self-pollinations (i.e. i = J) and

k1, another constant, is the log-odds ratio for outcross-pollinations (i.e. i ~ J). 92

If the probabilities of fruit set in the various outcross-pollinations are allowed to have independent values then the log-odds ratios for these outcrosses can be represented as pairs of parameters (aa to ff in Model 1b, Fig 5.1, p93).

Models 1a and 1b are symmetrical, in that no terms are introduced to allow for any reciprocal effects which depend on the direction in which the cross was performed. Terms to allow for these reciprocal effects can be added to the models, if justified by the "lack-of-fit'' test. In Fig 5.1, Model 2a is an extension of Model 1a where one genotype displays certain reciprocal effects. In this example, the majority of outcross-pollinations produce the same probability for fruit set but the reciprocal pairs for genotype number 2 have different log-odds ratios (i.e. K 2k1, K3k 11 K 4k1). Similarly, Model 2b is an extension of Model 1b with genotype number 2 displaying reciprocal effects (i.e. Aa, Dd, Ee) and all other outcross-pollinations having independent but paired values (i.e. bb, cc, ff).

5.2.2 Results

Pandorea pandoran.a pollinations Table 5. 1 a-c (p94) shows the proportion of fruit set resulting frotn various numbers of pollinations carried out in three diallel crosses for P. pandorana.

Combining the data for all three diallels, self-pollination produced extremely low fruit set (1 fruit frorn 353 pollinations). As a consequence, the Null Model, which predicts that all crosses will result in a constant fruit set, is not applicable to describe the data and is, therefore, not considered further.

For the 2 x 2 diallel (Table 5. 1a, p94), no fruit fanned from self-pollinations but a sirnilar nun1ber of fruit developed in both crosses between the two genotypes . When the data were fitted to the simplest extension of the Null Model, Model 1a, the resultant residual deviance was 0.0006 and the probabilities of fruit set for self- and outcross­ pollinations were zero and 0.3, respectively (Table 5.1a). ~~------

93

Model 0 (Null model)

o genotype (f)

1 2 3 4

? 1 ~ leo ~ ~ genotype 2 ~ leo ~ ~ (i) 3 ~ leo ~ ko

4 ~ leo ~ ko

Model 1a (Simple self and outcross model) Model 1 b (General symmetrical model)

o genotype (J) o genotype (J')

1 2 3 4 2 3 4

? 1 s:: k, k, k, ? !:'$.::: a b c genotype 2 k, :s:: k, k, genotype 2 a %$il: d e (i) (i) 3 k, k, '$.) k, 3 b d ::.&:::: f 4 k, k, k, Y$:n 4 c e f V$!.:

Model 2a (Simple self and outcross model with Model 2b (General symmetrical model with selected reciprocal crosses) selected reciprocal crosses)

d genotype (J) o genotype (;')

2 3 4 2 3 4

k a b c 9 s. I k, k, ? ::.$::: genotype 2 K2 :'S\ k, k, genotype 2 A ::::s::: d e (i) (i) ·:·:·:·:.:·: 3 k, K3 .$)! k, 3 b D :·:$-: f 4 k, K4 k, :::e.:::: 4 c E f :::§J

Fig.S.l Flow diagram showing alternative model pathways depending on whether the bulk of outcrosses are considered to result in common outcome or in discrete and varied results. 94

Table 5.1 Proportion of fruit set and fitted values [] resulting from various numbers of pollinations ( ) carried out in the diallel crosses for P. pandorana. Shading refers to self-pollinations. S.la. 2 genotypes. 5.1 b. 3 genotypes. S.lc. 4 genotypes.

5 .la. 2 x 2 diallel

6 genotype (J)

ATH TS4

Q ATH 0 (13) 0.30 (10) genotype /[OJ [0.30] (i) TS4 0.30 (10) 0 (10) [0.30] :19).::

5.1 b. 3 x 3 diallel

0 genotype (J)

GS TS6 TS3

GS 0.03 (40) 0.40 (5) 0.30 (30) ~ genotype ··m~·()JJ\ [0.38] [0.38] (i) TS6 * (*) 0 (10) 0.60 (5) [0.38] {:[QiP::U/ [0.38] TS3 0.90 (30) 0.60 (5) 0 (40) [0.90] [0.38] ::x~&g~:J\~:

5.lc. 4 x 4 diallel

o genotype (j) SCH GS TH vc

9 SCH 0 (30) 0.87 (IS) 0. 97 (30) 0. 77 (30) (OJ': genotype [0.81] [0.81] [0. 81] (i) GS 0.47 (15) 0 (30) 0.70 (30) 0. 73 (30) [0.47] :T9l:: [0.81] [0.81] TH 0.90 (30) 0.47 (30) 0 (90) 0. 80 (30) ~ [0.81] [0.47] :X9li!i [0.81] vc 0.90 (30) 0.47 (30) 0. 75 (60) 0.01 (90) [0.81] [0.47] [0. 81] :::;rqn 95 For the 3 x 3 diallel (Table 5.lb, p94), a high proportion of fruit set occurred in outcrosses but only a single fruit formed from the 90 self-pollinations. Consequently, when Model la was fitted rather than Model 0, the change in residual deviance was 80.67 using one degree of freedom (dt) which represented a highly significant (P < 0.001) difference in fruit set between self-pollinations compared with outcrosses. However, Model 1a did not fit the observed data closely and the residual deviance of 27.00 with 6 df was significant at P < 0. 001. The "lack-of-fit" appeared largely to be due to the cross of TS3 ~ x GS o which had a much higher percentage fruit set than the other outcrosses (Table 5.1b). Using a model akin to Model 2a (Fig. 5.1, p93), that allowed for a reciprocal effect for that particular cross, the fit improved substantially and the residual deviance of 4.46 with 5 df was no longer significant (0. 25 < P < 0.5). The fitted model was, therefore, of the fonn:

p J S for i = j log i j = K2 for, i = 3, j = 1 (i.e.TS3~xGS\f') ( K -:Fj = 1 1-Pij 1 fori except i 3,j =

where the estin1ated values of the log-odds parameters, S, k1, and K2 were significantly different from zero (P < 0.001) and corresponded to estimated probabilities for fruit set of 0.01, 0.38 and 0.90, respectively (Table 5.lb, p94).

For the 4 x 4 diallel (Table 5.1 c, p94), the average effect of outcross-pollinations again resulted in a much higher percentage of fruit set (68%) con1pared to self­ pollinations (0. 7%). In addition, reciprocal effects were evident in some of the primary data presented in Table 5.1c. The two pathways of sequential models (Fig 5.1, p93) that allow for reciprocal effects, depend on whether the outcross-pollinations generally have a similar outcome (Model 2a) or are independent (Model 2b). Table 5.2 (p96) shows the cmnparison of the residual deviance, df and significance when the data were sequentially fitted to the two alternative model pathways. Significant "lack-of-fit" was eliminated for both Model 2a and 2b when reciprocal terms accounting for the crosses GS'? x SCHo, TH'? x GSa and VCS? x GSo were included. 96

In this study, no statistical analysis to compare the models across the two pathways has been devised. Thus, for the present 4 x 4 diallel, either Model 2a or 2b could be used to describe the data. The predicted values from Model 2a, the simpler model, have been chosen for presentation in Table 5.1c.

Table 5.2 Residual deviance, residual degrees of freedom and significance of lack-of-fit between fruit set, resulting from pollinations between four genotypes of P. pandoralla (Table S.lc), and fitted values.

Model Residual Residual degrees of Significance of deviance freedom "Jack-of-fit"

Null 0 436.95 15 P< 0.001

Simple la 51.23 14 P< 0.001

Simple + reciprocals 2a 17.59 11 ns

Null 0 436.95 15 P

General lb 19.07 9 0.05 > P>O.Ol

General + reciprocals 2b 5.48 6 ns

Pando rea jasminoides pollinations Table 5. 3 (p97) shows the proportion of fruit set resulting from various numbers of pollinations carried out in a 3 x 3 diallel for P. jasmin.oides.

As for P. pandorana a higher proportion of fruit set occurred in outcross- compared with self-pollinations, which only yielded 4 fruit from 235 pollinations. The model best suited to explain the experimental data was akin to Model 1b which accounted for both the differences between self- and outcross-pollinations and for the observed genotype differences. The resultant model had a residual deviance of 10.49 with 5 df and a "lack­ of-fit" that was not significant (P > 0.05). The predicted probability for fruit set in self-pollinations was close to zero. No reciprocal effects were apparent but when genotype AL was crossed with LD the probability of fruit set was more than halved when compared to AL x HW or LD x HW crosses (Table 5.3, p97). 97

Table 5.3 Proportion of fruit set and fitted values [ ] resulting from various numbers of pollinations () between three genotypes of P. jasminoides. Shading refers to self-pollinations.

0 genotype (J)

HW LD AL

9 HW 0.00 (150) 0. 71 (104) 0.57 (49) [0.69] [0.62] genotype m:q~J LD 0.66 (67) 0.06 (64) 0.25 (36) (i) [0.69] r9Y9.?.1 [0.24] AL 0.73 (22) 0.22 (18) 0.05 (21) [0.62] [0.24] r~~Ptl

Pandorea baileyana pollinations Table 5.4 (p98) shows the proportion of fruit set from various numbers of pollinations carried out in a 2x2 diallel for P. baileyana.

Again, the proportion of fruit set was higher for outcrossed compared to self­ pollinated flowers although the difference was not as great as for the other two species. Outcrosses yielded 16 fruit from 25 pollinations whereas seven fruit resulted from 40 self-pollinations. The null hypothesis was a poor fit. When the data were fitted to Model 1a, there was an improvement in the fit of the model, the residual deviance being 14.08 with 2 degrees-of-freedmn, but the "lack-of-fit" was still significant.

It was not considered appropriate to fit the next hierarchical model which allowed for reciprocal effects because the two outcross cells would be fitted exactly leaving only one degree-of-freedom. However, the significant "lack-of-fit remaining after fitting Model 1a suggests that reciprocal effects may occur. An extension of the diallel to include a greater number of genotypes and pollinations per cell is required to resolve the "lack-of­ fit" between the data and the model but was beyond the scope of this study. 98

Table 5.4 Proportion of fruit set and fitted values [ ] resulting from various numbers of pollinations ( ) between two genotypes of P. baileyana. Shading refers to self-pollinations.

o genotype (j)

2 3

2 0.25 (20) 1.00 (10) genotype ri;::::'1':7'} .vt::.})l( [0.64] (i) 3 0.40 (15) 0.10 (20) [0.64] t9ii~'()

5.3 DISCUSSION

Pollination effects Self-pollination in P. pandorana, P. baileyan.a and P. jasminoides always resulted in a highly significant reduction in, or absence of, fruit set. This observation is consistent with the responses found for other members of the Bignoniaceae, including: Campsis radican.s (Elrod, 1904; Bertin, 1982); Catalpa speciosa (Stephenson and Thomas, 1977); Chilopsis lin.earis (Petersen et al., 1982); Tabebuia n.eochrysantha (Bawa, 1974); Tabebuia rosea (Bawa, 1974; Bawa and Webb, 1984). The demonstration of this self­ incompatibility, however, depended on comparisons of fruit set resulting from pollinations involving only self-pollen with those involving only outcross-pollen. Bertin and Sullivan (1988) and Bertin et al. (1989) whilst confirming self incompatibility in Campsis radican.s when using only self-pollen, also showed a degree of self-fertility when using a mixture of self- and outcross-pollen. Selfed-progeny ranging from 2-30% of seeds in individual fruits were recorded after pollination with mixed pollen. This percentage was found to vary with the genotype of the outcross-pollen donor.

Whilst no data were obtained in the present study on the site of the barrier to fruit set in P. pandoran.a, P. jasmin.oides or P. baileyan.a, self- and outcross-pollen tubes were observed in the style tissue and ovary. Similar observations in Campsis radican.s led to the conclusion by Bertin and Sullivan ( 1988) that there was no self-incompatibility mechanism operating in the stigma or style. Rather, a late-acting self-incompatibility mechanism, found also in other woody plants (Seavey & Bawa, 1986) is indicated. Further studies with that species (Bertin et al., 1989) show sporadic abortion of embryos ADDENDUM

Examiner #2

4. The possibility exists that the reciprocal effects detected in the diallel experiments with genotype GS are due to dominance effects within the self-incompatibility system itself. The genotype OS is quite distinct from all other genotypes used but as its origin is not known neither is its relationship to the other genotypes in this study. Additional pollinations are required to clarify any doininance effects within the self-incompatibility systen1 as shown here for the species, P. pandorana, as a whole. 99 and reduced seedling vigour following self-pollinations suggesting the occurrence of inbreeding depression. As a result, the self-sterility observed in C. radicans has been interpreted as a partial self-incompatibility with inbreeding depression reducing embryo and seedling viability of selfed progeny which bypass the self-incompatibility barrier (Bertin et al, 1989). Elucidation of the exact site of self-pollen tube arrest or the premature tennination of zygotic development, which could explain the very low fruit set observed, awaits further microscopy studies for both Pan.dorea and Campsis.

In some of the P. pandoran.a crosses, fruit set for a pair of genotypes varied according to the direction in which the cross was performed. Notably, these reciprocal effects always involved the GS genotype but there was no consistent pattern between either the magnitude or direction of the reciprocal effect and whether GS was used as

the r3 or ~ parent. Reciprocal effects have been reported for Macadamia (Sedgley et al., 1990) and for Ban.ksia coccin.ea (Fuss & Sedgley, 1991). Within Bignoniaceae, Campsis radicans also appeared to exhibit reciprocal effects (Bertin 1985). The present reciprocal effects with GS are difficult to explain without further study and a knowledge of the geographical origin and distribution of this popular nursery selection. Whilst no reciprocal effects were observed in P. jasminoides, crosses involving genotypes LD and AL showed a reduced fruit set compared to outcrosses genotype HW with either genotype LD or AL. Both AL and LD are nursery selections with white flowers and may be more closely related to each other than to the pink-flowered genotype HW. The differences in fruit set for outcross pollinations in P. baileyana may be due to reciprocal effects but the species requires further study to detennine whether the observed values are representative of the genotypes involved or are an artefact of the small sample size.

Suitability of models The developtnent of the generalised linear models proved a useful approach in fitting diallel data that had a binomial distribution. The progression of models allowed for factors such as the influence of self-pollination (as distinct from outcrosses), the effect of different genotypes and the effect of reciprocal crosses to be added and evaluated in tum. By fitting several n1odels of increasing complexity it was possible to determine when the residual deviance ("lack-of-fit") first became insignificant (P > 0.05). It is 100 possible to extend the number of models further to allow for any, or all, differences in the diallel cells to be fitted. However, continued extension of the model will lead to a point where data are overfitted. Such overfitted complex models may not be helpful in describing the underlying general trends in breeding behaviour. Thus, the use of the models requires a certain subjective balance between oversimplification of the trends and overfitting of the actual data.

Another shortco1ning of the present statistical approach is the need to subjectively decide between the model pathways 1a/2a and 1b/2b (p93). This decision depends on whether the bulk of the outcrosses are considered to result in a common outcome or in discrete and varied results. Fruit set for outcross pollinations where no reciprocal effects are evident, varies from 0.3 (Table 5.1a, p94), 0.38 (Table 5.lb, p94) to 0.81 (Table 5.1c, p94) for P. pandorana and 0.24 to 0.62 to 0.69 for P. jasminoides making pathway 1b/2b the appropriate pathway for the data presented here. No statistical basis for this choice has been devised here. Moreover, no statistical method to compare models across these two pathways was developed.

5.4 SUMMARY Despite the constraints mentioned, the present approach allowed an adequate description of the breeding behaviour of P. pandorana, P. baileyan.a and P. jasminoides. That is, that these species all exhibit strong self-incompatibility systems. However, improvements on the model approach used here are warranted. In particular, a statistically-based algorithm that allows comparisons between and within paired diallel cells and addresses the concerns of overfitting data will be needed if this approach is to be widely adopted as an analytical tool. 101 CHAPrER6

INTERSPECIFIC HYBRIDIZATION

6.1 INTRODUCTION

The ornamental plant industry relies on the introduction of new plant varieties to perform competitively in the Australian and international marketplace. Interspecific hybridization provides the potential to obtain new genetic combinations for assessment as elite clones for the ornamental plant industry. The aim of this series of experiments was to produce interspecific hybrid material from selected Pandorea genotypes.

Embryo culture techniques were developed to rescue interspecific hybrids as the embryos aborted if left to mature in the seeds on the parent plant. Hybrid embryos cultured in vitro have extended the range of material available for plant breeding. Interspecific hybrids of Lilium were the first incompatible interspecific crosses to be rescued through an embryo culture technique (Laibach, 1925, 1929). Embryo culture has been used frequently since then to rescue novel hybrid plants which would otherwise be unobtainable. Incompatible hybrid crosses rescued through embryo culture include Trifolium (Williams, 1980), Actinidia (Kin et. al., 1990) and Bras sica (Agnihotri et. al., 1990). Commercial cultivars of Prunus persica such as 'Summerglo' and 'Culenborg' originated via embryo culture (Ram1ning, 1983). Nectarine cultivars have also been released as a result of embryo culture (Torroba and Frangi, 1979; Torroba et. al., 1980).

Isozymes have been used for the verification of hybrid status in a number of woody plants including intergeneric plum x apricot hybrids (Byrne & Littlejohn, 1989) and interspecific hybrids between three cultivars of Cam.ellia reticulata and C. chrysantha (Hwang et. al., 1991). To see whether there were differences in the enzyme systems of the parents which could be utilized for establishing the hybrid identity of the seedlings, 'f . Vr:L-1 an Isozyme analysis undertaken of putative hybrids of P. baileyan.a x P. nervosa and P. pandorana x P. ne~osa obtained from 1990 pollinations. It was not possible to test the 102 test the hybrid identity of seedlings by measuring morphological characters because seedlings of the parental species, P. nervosa and P. baileyana, were not available. As the seedlings exhibit juvenile foliage for at least one year, it was not possible to compare them to the adult foliage of the parents. Flowering was not anticipated until the third year of growth where P. pandorana was involved and the juvenility phase for P. nervosa and P. baileyana is unknown. An indication of the hybrid status of the seedlings was required to minimize the number of plants which were to be maintained until floral characteristics can be assessed.

Tissue culture techniques were developed to ensure that hybrid material could be multiplied quickly to provide sufficient material, firstly, for assessment of horticultural traits and secondly, to facilitate release to industry of elite clones (Chu and Kurta, 1990).

6.2 RESULTS AND DISCUSSION

6.2.1 Interspecific cross-pollinations and pollen-tube growth

Pollen-rube growth in inter~pecific pollinations Pollen used for interspecific pollinations germinated on the stigtnas of the female plants. Pollen-tubes grew down the style and into the ovary in both intra- and interspecific pollinations. Thus any incompatibility or incongruity encountered is likely to be either at the ovular level or post-zygotic. Due to the large numbers of overlapping pollen-tubes, in the order of several hundred, quantitative measurements were not possible. Pollen-tube growth for intra- and inter-specific crosses of P. baileyana are shown m Figs 6. 1-6.3 (p 103). Occassionally, a pollen tube could be seen entering an ovule. A pollen tube of P. nervosa can be seen entering an ovule of P. baileyana (Fig. 6.4, pl04) with fluorescing hypostases clearly visible in the ovules.

Figures 6.1 to 6.4 following ... Figure 6.1 Light micrograph of pollen tube growth in self-pollinated pistil of P. baileyana(3). a. pollen germination on the stigma (x400) b. pollen tubes present in mid-style (x400) c. pollen tubes at base of style and traversing ovules (xlOO).

Figure 6.2 Light micrograph of pollen tube growth in intraspecific cross­ pollinated pistil of P. baileyana. a. pollen germination on the stigma (x400) b. pollen tubes present in mid-style (xlOO) c. pollen tubes at base of style and traversing ovules (xlOO).

Figure 6.3 Light micrograph of pollen tube growth in interspecific cross­ pollinated pistil of P. baileyana(2) x P. pandorana(SCH) a. pollen germination on stigma (x400) b. pollen tubes present in mid-style (x400) c. pollen tubes at base of style and traversing the ovules (x400). 103 104

Figure 6.4 Light micrograph of entry of P. nervosa pollen tube into micropyle (arrowed) of P. baileyana(2). pt = pollen tube, o = ovule, h = hypostase. (x400). 105 Interspecific cross pollinations Interspecific cross pollinations were performed on P. jasminoides, P. pandoran.a and P. baileyana. Results of fruit set two weeks after pollination are presented in Tables 6.1 (p106), 6.2 (p107) and 6.3 (p108), respectively. Fruit set was obtained between all species but not necessarily between all genotypes pollinated. A summary table based on species is also presented (Table 6.4, p108).

Due to the small numbers of interspecific pollinations in most cells (Tables 6.1-6.3), it was not appropriate to fit generalised linear n1odels. The raw data are presented as an indication of trends but it is stressed that where fewer than 25 pollinations were performed, the trends may not be reliable and comparisons between treatments is not possible. Individual genotypes within species may perform differently in each interspecific combination but this conjecture can only be verified by further work increasing the numbers of pollinations per cell.

For the summary table (Table 6.4, p108), analysed by fitting a generalised linear model, fruit set was reduced for a particular species when used as a female, if the pollen source was interspecific rather than intraspecific. The success of a particular species as a male was also greater if the female was of the same species.

In addition, male success differed between species for a particular female species. More fruit was set using P. pandorana as the pollen source if P. jasminoides was the female rather than P. baileyana. When the success of P. nervosa as a pollen source is compared for interspecific cornbinations, the highest fruit set is obtained again when the female was P. jasminoides. There was no significant difference between fruit set obtained when P. pandorana was used as a female compared to that when P. baifeyana was the female. Interestingly, it was the P. baileyan.a x P. nervosa putative hybrid embryos which were successfully cultured.

Tables 6.1 to 6.4 following ... ADDENDUM Examiner #2

8. The possibility exists that small amounts of self-pollen may have result in selfed progeny in interspecific pollinations. However, great care was taken during interspecific pollinations to prevent contamination from self-pollen and I consider it extremely unlikely that the putative hybrid progeny are, in fact, selfs. P. jasminoides P. baileyana P. nervosa Species/ d' genotype PjHW PjF827 PjATHl PjF917 PjAL Pb2 Pb3 Pn ~

P. pandorana - 0.18(56) PpBF 0 (7) 0.26 (35) 0 (15) 0.13 (15) - - - 0.40 (5) PpTS4 - - - - - 0 (5) 0 (10) 0.46(43) PpGS - - 0.22 (23) - - 0.10 (10) 0.18 (40) - PpAL - - - - - 0 (5) - 0.05 (20) PpSCH ------0.60 (10) PpATH 0 (18) - - - 0 (6) - 0.10 (10) PpMM ------

Table 6.1 Proportion of fruit set two weeks after various numbers of interspecific pollinations ( ) where P. pandorana was used as the female parent.

1-4 0 0\ Species/ 6 P. pandorana P. nervosa Genotype ~ PpBF PpGAS PpGS PpMM PpATH Pn

P. jasminoides PjHW - - - - 0.29 (28) 0.55 (20) PjF827 0.42 (19) 0.10 (10) 0.36 (11) - - 0.10 (10) PjATHl 0.14 (14) 0.60 (5) - 0 (2) - - PjWS1 0.62 (12) - - - 0.67 (3) 0.78 (18) PjAL 0.33 (3) - - - 0 (4) 0.65 (13) PjLD - - - - 0 (2) 0.44 (18) 0 (6) Pj351 - ... - - - 0.75 (8) PjLGF - - - - 0.52 (25) 0.20 (20)

Table 6.2 Proportion of fruit set at two weeks after various numbers of interspedfie poUinations using P. jasmiJWides as the fanate pamtL

8- ,...-.------

108

Table 6.3 Proportion of fruit set two weeks after various numbers of interspecific pollinations ( ) lL.~ing P. baileyaiUl as the female parent.

Species/ P. pandorana P. nervosa Genotype PpGS PpAL PpTS4 Pn

P. baileyana Pb2 0.05 (20) 0 (10) 0 (10) 0.80 (10)

Pb3 0.10 (50) 0 (15) ~ 0 (15)

Table 6.4 Summary of fruit set two weeks after various numbers of intra- and inter­ specific pollinations ( ). Shading refers to intraspecific pollinations.

Species P. pan.dorana P. jasminoides P. baileyana P. nervosa

P. pandorana .·:<·.· .. <<·:·>:·:·/. P. jasminoides 0.36 (146) 0.45 (112) P. baileyana 0.14 (105) Q:it,~;;:~??) 0.32 (25)

6. 2.2 Fruit dimensions and seed ge1mination Results of fruit measurements and seed germination tests are shown in tables organized for the female parent P. jasm.inoides, (Table 6.5, pllO), P. pandorana (Table 6.6, pl11-112) and P. baileyan.a (Table 6. 7, pll3).

Seed resulting from intraspecific crosses of P. jasmin.oides has a percentage germination of between 88.3 and 97.5%. No germination was obtained from seed resulting from interspecific crosses when the male parent was either P. pandoran.a or P. n.ervosa. Total fruit weight and seed weight was also significantly lower for interspecific crosses (compared using individual s.e.ds for each pair of genotypes). Incompatibility is often encountered in interspecific crosses and results in shrivelled seeds containing aborted embryos (Hu and Wang, 1987). The failure of interspecific 109

Pandorea embryos is probably due to the degeneration of the endosperm, where the first signs of abnormal developn1ent in interspecific hybrids typically occur (Wardlaw, 1965).

Seed number was also lower for particular interspecific combinations. Seed collected from interspecific crosses just prior to fruit maturation contained a shrivelled embryo sac containing an immature embryo. Seed from intraspecific fruit at the same stage of maturity contained fully mature embryos (Fig. 6.5, p114).

For the interspecific crosses where P. pandorana was the female parent, fruit weight and seed number were reduced for some combinations but not for others (Table 6.6, plll-112). The percentage germination for outcrossed intraspecific seed ranged from 64.3% to 97.2%. Small numbers of interspecific seed germinated after some interspecific crosses, especially for the crosses P. pandorana(GS) x P. nervosa, P. pandorana(AL) x P. nervosa and P. pandorana(AF) x P. nervosa where germination percentages were 0.4%, 2.9% and 12.0%, respectively. No seedlings were raised from the germinated seed because they were weak and succumbed to damping off before they could be transferred to a seed-raising medium.

For P. baileyana, the greatest seed weight resulted from crosses with P. baileyana as the male parent. Seed weight was reduced in fruit resulting from self- and interspecific pollinations. There was also a reduction in the numbers of seed per fruit of inter­ compared to intra-specific crosses. Fewer seed developed from the self-pollinations of P. baileyana(2) but not P. baileyan.a(3). Percentage seed germination for P. bailey ana outcrosses was low (40.6% and 27.3%) compared to the other two species tested. Selfed seed had a very low level of germination (6.7% and 7.8%). No seed from interspecific crosses germinated.

As interspecific hybrid seed was not fertile due to the apparent abortion of the embryo in the mature seed, an embryo culture technique was developed to rescue the hybrid embryos before they aborted. Cross fruit wt capsule septum wt seed wt seed no. days to % (g) wt (g) (g) (g) maturity germ .. PjAL x PjHW 3.41 * * 1.389 220 119 96.7 PjAL X PjLD 3.54 * * 0.990 165 99 95.4 PjAL X PjAL 5.52 * * 1.003 124 73 96.7 PjAL x Pn 1.40 1.11 0.15 0.129 86 162 0.0

PjATH x Pn 2.38 1.78 0.35 0.249 175 133 0.0

PjHW X PjLD 4.20 2.95 0.36 1.433 208 111 96.8 PjHWx PjAL 3.92 * * 1.409 207 118 97.1 PjHW X PjLGF 4.07 2.32 0.22 1.527 235 97 88.3 PjHW X PpATH 1.78 1.43 0.12 0.224 82 102 0.0 PjHW x Pn 2.44 1.89 0.17 0.369 157 94 0.0

PjLD x PjAL 2.77 * * 1.042 188 130 97.5 PjLD X PjHW 3.77 * * 1.731 228 127 97.5 PjLD x Pn 1.82 1.50 0.18 0.144 56 108 0.0

PjLGF X PpATH 1.60 1.24 0.15 0.216 77 86 0.0 PjLGF x Pn 2.32 1.75 0.29 0.276 101 130 0.0

PjWS X PpATH 1.79 1.42 0.19 0.178 42 108 0.0 PjWS x Pn 1.84 1.39 0.23 0.219 136 93 0.0

Pj351 X PpATH 1.04 0.81 0.08 0.150 54 93 0.0

-

Table 6.5 Fruit dimensions and percentage seed gennination for intra- and interspecific crosses of P. jasminoides...... 0 Cross fruit wt capsule wt septum wt seed wt seed no days to % (g) (g) (g) (g) mature germ

PpAF X PpMM 1.13 0.52 0.21 0.392 108 54 96.1 PpAF X PpSCH 1.21 0.57 0.24 0.399 116 56 64.3 PpAF X PpTH 0.76 0.36 0.14 0.261 98 53 92.8 PpAF x Pn 0.56 0.29 0.13 0.128 67 53 12.0

PpAL X PpGS 1.00 0.45 0.13 0.148 127 55 85.3 PpAL X PpSCH 1.18 0.58 0.16 0.442 113 55 83.7 PpAL X Pb3 0.74 0.44 0.12 0.192 89 48 2.9

PpATH X PpGS 1.70 0.79 0.22 0.687 150 57 85.0

PpBF x Pn 1.93 1.45 0.24 0.229 116 63 0.0 PpBF X PjATH 1.09 0.88 0.12 0.086 107 63 *

i PpGS X PpMM 0.92 0.41 0.13 0.394 82 62 95.0 PpGS X PpSCH 0.78 0.32 0.11 0.359 87 54 95.3 PpGS X PpTH 0.56 0.27 0.09 0.216 79 52 91.1 PpGS X PpTS2 0.59 0.28 0.08 0.213 90 76 82.5 PpGS X PpTS3 0.59 0.27 0.08 0.228 104 71 89.2 PpGS x PpTS4 1.17 0.53 0.15 0.523 114 59 95.0 PpGS X PpTS6 0.66 0.32 0.10 0.242 106 69 88.3 PpGS X PpVC 0.66 0.29 0.09 0.282 91 52 95.4 PpGS x Pn 0.99 0.61 0.23 0.132 93 52 0.4 PpGS X PjATH 0.17 0.12 0.03 0.025 29 66 0.0

97.2 PpMM X PpAF 0.78 0.30 0.12 0.357 85 61

50 95.0 I PpMM X PpGS 0.90 0.40 0.14 0.392 71 95.6 PpMM X PpSCH 0.66 0.27 0.09 0.300 74 64 66 95.0 PpMM X PpTS4 1.06 0.45 0.03 0.514 80 PpMM x PpVC 0.41 0.17 0.06 0.188 80 61 94.2

I I I I II,__. ,__...... Table 6.6 Fruit dimensions and percentage seed germination for intra- and interspecific crosses of P. pandorana. ------Cross fruit wt capsule wt septum wt seed wt seed no days to % (g) (g) (g) (g) maturity germ

PpSCH X PpAF 0.80 0.38 0.13 0.288 114 56 73.7 PpSCH X PpGS 0.77 0.33 0.15 0.284 82 60 96.1 PpSCH X PpTH 0.60 0.28 0.10 0.220 90 57 92.3 PpSCH X PpVC 0.69 0.31 0.12 0.266 103 58 98.6

PpTS3 X PpATH 0.72 0.23 0.10 0.387 96 82 100.0 PpTS3 X PpGS 0.82 0.30 0.13 0.388 89 81 90.7 PpTS3 X PpTS2 0.75 0.27 0.13 0.350 94 74 96.7 PpTS3 X PpTS6 0.57 0.19 0.10 0.275 87 77 95.6

PpTS4 X PpAF 1.46 0.77 0.24 0.450 118 44 90.0 PpTS4 X PpGS 1.18 0.54 0.23 0.415 118 61 95.7 PpTS4 X PpVC 0.96 0.46 0.18 0.323 117 56 93.3 PpTS4 X PpMM 1.01 0.49 0.15 0.379 Ill 63 98.3

PpTS6 X PpTS2 0.84 0.44 0.16 0.234 93 52 90.6 PpTS6 X PpTS3 0.86 0.41 0.18 0.248 94 49 75.0

PpTH X PpGS 1.02 0.51 0.20 0.314 130 63 92.3 PpTH X PpMM 1.29 0.70 0.22 0.377 169 72 82.8 PpTH x PpSCH 0.79 0.36 0.15 0.258 138 58 88.9 PpTH x (self) 0.96 0.33 0.20 0.441 161 61 95.0 PpTH X PpTS4 1.28 0.69 0.25 0.388 158 63 79.7 PpTH x PpVC 0.72 0.36 0.13 0.235 153 57 86.9

PpVC x PpGS 1.02 0.42 0.24 0.370 117 56 87.5 PpVC X PpMM 0.97 0.41 0.22 0.346 121 56 93.3 PpVC X PpSCH 0.98 0.37 0.20 0.421 156 54 91.5 PpVC X PpTH 0.64 0.27 0.13 0.242 119 56 88.1 PpVC X PpTS4 1.16 0.51 0.21 0.444 152 60 52.5

Table 6.6 (cont.) -tv ------Cross fruit wt capsule wt septum wt seed wt seed no days to mature % (g) (g) (g) (g) germination

Pb2 (self) 0.11 0.06 0.02 0.025 84 35 7.8 Pb2 X Pb3 0.22 0.11 0.05 0.058 220 32 .. 27.3 Pb2 X PpGS 0.094 0.05 0.02 0.016 51 32 0.0 Pb2 x Pn 0.23 0.15 0.08 0.001 11 32 *

Pb3 X Pb2 0.33 0.16 0.08 0.091 167 32 40.6 Pb3 x PpGS 0.15 0.08 0.04 0.031 76 28 0.0 Pb3 (self) 0.28 0.11 0.10 0.068 190 31 6.7

Table 6.7 Fruit dimensions and percentage seed gennination for intra- and interspecific crosses of P. baileyana.

~ ~ w

J J ------~-----~·~,------

Figure 6.5 Comparison of seed from interspecific (P. jasminoides x P. nervosa)and intraspecific (P. jasminoides) pollinations. a. shrivelled embryo sac with immature embryo from interspecific pollination b. diagrammatic representation of Fig.. 6.5a, m = micropylar end, es = embryo sac, e = position of immature embryo, p = partly removed seed coat, w = membranous wing of seed. c. Embryo resulting from intraspecific cross-pollination, in embryo sac (left) and excised (right), m = micropylar end, c = cotyledons, r = radicle. (Bar = lmm). 114

b --.

115

6.2.3 Embryo rescue of putative hybrids This work on en1bryo rescue is preliminary to more precise experimentation backed up by histological procedures to determine the stage at which embryo development ceases and embryo abortion occurs.

Putative interspecific hybrid seedlings were obtained for the crosses P. baileyana x P. n.ervosa (117 seedlings), P. pandoran.a x P. n.ervosa (7 seedlings) and P. baileyana x P. pandorana (26 seedlings) by removing immature embryos from the fruit, either within the intact embryo sac or excised, and culturing them in vitro. The proportion of live embryos from P. baileyan.a x P. nervosa fruit was recorded after 14 days of culture (Fig. 6.6a, p118) on each of four tnedia (Ch. 2, section 2.4.2). As a general trend, the older the fruit at harvest, the greater the nun1ber of embryos still alive after culturing for 14 days until a point was reached where the success rate declined.

After 4-6 weeks in culture, embryos which had grown so that cotyledons were at least 1-2mm in diameter and the radicle was > 2mm long, were transferred to a germination medium with 1% sucrose and Gelrite as the gelling agent (Fig. 6. 6b, p118). Some en1bryos had not grown at all and were not tranferred. Again, a higher proportion of embryos excised fro1n older fruit were transferred. Whilst the majority of embryos were well-developed at the cotyledonary stage, individual en1bryos from a fruit differed in their degree of maturity. Well-developed cotyledonary embryos and less mature embryos were present and also cultured (Fig. 6.7, p119). Those grown on 7% sucrose/ Agar mediun1 showed a slight advantage in the raw data but the stnall sample size was insufficient to enable conclusions to be drawn on the suitability of one mediurn over another for the initial in vitro culturing.

Embryos were also present in the imn1ature seeds of a number of other interspecific fruits but could not be successfully cultured using the system found to be successful for the P. baileyana x P. nervosa putative hybrids. Fruit from the cross P. jas1ninoides x P. nervosa were harvested at 24d, 28d, 33d, 41d and 52d after pollination. Embryo sacs were cultured. However, the embryos which were at an early heart stage failed to develop further although the majority remained alive on the media for a minimum of 14 days culture (Table 6.8, p116). Fruits resulting from the crosses of P. pandorana x P. 116 n.ervosa, P. baileyana x P. pandorana and P. jasminoides x P. pan.dorana were harvested between 25 and 80 days after pollination. The embryos of these crosses were arrested at an earlier stage of development, usually at the globular or early heart-shaped stage, when endosperm was mainly liquid, and all attempts to culture them in embryo sacs were unsuccessful. Whilst they could often be maintained in culture for some months, they did not increase greatly in size, frequently germinated prematurely and transfer to other n1edia was not successful.

Table 6.8 Proportion of embryos fron1 P. jasminoides x P. nervosa alive after 14 days in culture (Proportion transferred to 1% sucrose/Gelrite medium)

Medium used for Proportion of en1bryos alive after 14d in culture initial cui ture (Proportion transferred to 1% sucrose/Gelrite medium Age of fruit at harvest [ ] [24d] [28d] [33d] [41d] [52d]

7%sucrose/ Agar 0.6 (0) 1. 0 (0) 0.8 (0) 1.0 (0) 1.0 (0) 5% sucrose/ Agar 0.6 (0) 1.0 (0) 1.0 (0) 1.0 (0) 1.0 (0) 7%sucrose/Gelrite 1.0 (0) 1.0 (0) 1.0 (0) 1.0 (0) 1.0 (0) 5% sucrose/Gelrite 1.0 (0) 1.0 (0) 1.0 (0) 1.0 (0) 1.0 (0)

Precocious gern1ination of imn1ature etnbryos is common (Hu and Wang, 1987) and was first n1entioned by Hannig (1904). En1bryos less than full-term, if they grew at all in vitro were found to germinate and grow as small seedlings instead of completing the normal en1bryonic development. Such germinated seedlings of Pan.dorea are weak and display only those structures which were present at the time of embryo excision or embryo sac culture.

Precocious in vitro germination and its control have been reviewed by Norstog (1979).

Factors in the c~:~r:,~ri..,c ~gciium which are capable of suppressing precocious germination of some species 1\ They-·were high osmotic pressure, elevated potassium level, ample amounts of nitrogen in the n1edium in the fonn of the a1nmonium salt of an organic acid, abscisic acid (ABA), and possibly low oxygen tension. Hormones such as gibberellins and cytokinins promote precocious germination whereas ABA suppresses it (Hu and Wang, 1987). As a general rule, the younger the excised embryo, the higher the 117 medium osmolarity required. High osmotic values in the natural sap of immature ovules have been found for heart-shaped cotton embryos (Kerr and Anderson, 1944), heart­ shaped and late cotyledonary stage bean embryos (Smith, 1971), and torpedo stage of Capsella (Rijven, 1952). The high sucrose requirement (8-12%) for Datura embryos at the pre-heart stage was proven to be osmotically related rather than a metabolic need (Rietsema et. al., 1953). The sucrose level of 7% used as the upper level for the culture of Pan.dorea embryos may not have been high enough for the early heart-shaped stage and may be the reason for the precocious germination of embryos at the early cotyledonary stage (Fig. 6. 8, p 119). Lower sucrose rates have been successful for other plant species. Four percent sucrose was suitable for normal embryo development in species of llex (Hu, 1975).

The low success rate in culturing en1bryos at an early stage of development has been interpreted as a failure to fully provide for the horn1onal and nutritional requirements. The continued development of the embryo fro1n extreme reliance on the maternal tissue to autotrophy depends on the progressive activation of critical enzyme systems or biochemical pathways especially concerned with the synthesis of proteins and/or growth substances (Raghavan. 1965). Where the culture medium does not allow for the activation of such systems, embryo development in culture will not proceed. Culture of immature embryos with "nurse" endosperm, in some cases, has enabled the continued development of embryos by circumventing the nutritional and hormonal problems associated with endosperm failure of otherwise viable interspecific hybrid embryos. Interspecific and intergeneric incompatible crosses where embryo culture has been used to overcone hybrid sterility have been listed by Raghavan (1977), Collins and Grosser (1984) and Hu and Wang (1987). Pasture legumes (Williams and De Lautour, 1980; Williams, 1987)) and the kiwifruit, Actinidia (Kin et. a!., 1990) a woody vine of similar habit to Pandorea have been successfully raised via embryo culture.

A second explanation for the difficulty in culturing very immature embryos is that the delicate manipulation required to excise the small embryos results in damage to the suspensor and loss of important suspensor functions such as endogenous hormone biosynthesis (Ceccarelli et. al., 1981) and nutrient transport. The suspensor is the site of nutrient uptake in the early stages of embryo development (Yeung, 1980; Hu and Wang, 1987). 118

1.-4 il) ...,_) '+-< 0.4 aj 0.3 0.2

0.1

0.0 0 10 20 30 Age of fruit at harvest (days)

1 .0

0.9

0.8

0.7

0.6

0.5

0.4 0.3

0.2

0.1

0.0 0 1 0 20 30 Age of fruit at harvest (days)

Figure 6.6 Success of culturing embryos taken fron1 interspecific fruit froru the cross P. baileyana(2) x P. nervosa. a. proportioll of live after 14 days culture in each of four media. b. Proportion of en1bryos which had grown sufficiently to be transferred to a germination n1ediurn. Figure 6. 7 Well-developed cotyledonary P. haileyana(2) x P. nervosa embryo. Inset: less well-developed cotyledonary embryo. c = cotyledon, r = radicle. Bar = lmm.

Figure 6.8 Im1nature ernbryos showing precocious germination. c = cotyledon, rh = root hairs. Bar = lmm. 119 120 The decline in the vigour of P. baileyana x P. n.ervosa embryos taken just prior to fruit maturation (Fig. 6.6a, b, after day 28, p118) may be due to inadequate nutrition. Factors such as endosperm failure have not been investigated.

Further work is required to determine the factors which inhibit embryo development and maturation in order to optimize the media requirements for the hybrid embryos of Pandorea. The development of a medium suitable for the culture of embryos at an earlier stage than the cotyledonary embryos cultured in this study and a protocol for the aseptic culture of immature embryos would provide scope to increase the number of different interspecific Pandorea hybrids.

6.2.4 Isozyme analysis of putative hybrids Enzyme systems have not been studied in the family Bignoniaceae. Bertin and Sullivan (1988), however, did assay pollen from Campsis radicans at the 6- phosphogluconate dehydrogenase locus to distinguish progeny resulting from self- fertilization from outcross progeny.

6. 2. 4.1 P. bailey ana x P. n.ervosa Only three of the enzymes assayed initially showed differences between the two parents.

Phosphoglucomutase (PGM) Pandorea n.ervosa and P. baileyan.a both produced a single band with P. nervosa having the slower allele. In the first gel (Fig. 6.9a, p122), all seedlings showed a double band. In the second gel (Fig. 6.9b, p122), the first 4 seedling samples had two bands. The second group of four seedlings (Lanes 9-12) were not as clear. However, they appeared to contain two bands, one from each parent. My interpretation is that the ~ parent (P. baileyan.a) is homozygous for a fast allele, the o parent (P. n.ervosa) is homozygous for a slow allele and the offspring are heterozygous, thus indicating their hybrid identity.

Malate dehydrogenase (MDH) Pandorea nervosa gave 3 bands. Pandorea baileyan.a gave 4 bands or 5 bands and 2 samples in which bands could not be distinguished but which extended from the lowest to the highest band of the other P. baileyana sample on each gel. This enzyme was 121 assayed again with diluted samples of the parents. Even with the diluted parents, ~ P. baileyana (lane 2 in all gels) was not clear and definite bands were difficult to identify. Whilst there are indications that seedlings show bands from both parental patterns, they are not conclusive and will not be discussed further.

Leucine amino peptidase (LAP) Both gels (Fig. 6.10a, b, p123) showed a single band for P. baileyana and P. n.ervosa which were different with P. nervosa having the faster allele. In most cases, the seedlings show both parental bands supporting the likelihood of them being hybrids between the two species. On the first gel (Fig. 6.10a, pl23), for seedlings 47R and 45R (lanes 3 &9), the single band is identical to that for the Q parent, P. baileyana. All other seedlings had two bands corresponding to one from either parent. For the second gel (Fig. 6. lOb, pl23), seedlings 40R and 61R again had a single band the same as the Q parent. Seedlings 39R, 34R and 51R (lanes 9,10 & 12) had double bands. Seedlings 42R,62R and 41R (lanes 3,4 & 6) had less discrete bands but appeared to contain two corresponding to each parental band. All double banded seedlings are interpreted to be hybrids. The reason for seedlings with a single band corresponding to that of the 9 parent is unknown. Apomixis has not been reported for the family Bignoniaceae. The unavoidable differences in the age of the leaf tissue being used may have contributed to the unexpected banding patterns. Differential gene expression has been demonstrated in Picea glauco (Misra and Green, 1990) in tissue of differing age and for different genetic backgrounds of Brassica napus hybrids (Schenk and Wolf, 1986). Other explanations include the inheritance of null alleles from one parent (Coulthart and Denford, 1982) or the selective loss of chromosomes so that progeny may exhibit a hybrid banding pattern for one enzyme but not for another (Sjodin and Glimelius, 1989).

6. 2. 4. 2 P. pandor ana x P. nervosa Only two of the enzymes assayed initially showed any differences between the two parent species.

Glucose phosphate isom,erase (GPI) Lanes 5 and 10 show a pattern comparable to o parent, P. nervosa. None correspond exactly to the Q parent, P. pandoranaAF. Lanes 6, 9 and 11 have the male banding pattern plus an extra band. Lane 12 appears to have only the two bands from the male parent (Fig. 6.11, p124). 122

Phosphoglucomutase (PGM)

Gel 1

------

Gel 2

-----

Figure 6.9 Schematic representation of the zymograms of the PGM loci in P. baileyana (Pb), P. nervosa (Pn) and putative hybrid seedlings(numbered). a. L-R: Pn, Ph, 47R, 46R, 54R, 57R, Pn, Pb, 45R, 60R, 59R. b. L-R: Pn, Ph, 42R, 62R, 40R, 41R, Pn, Pb, 39R, 34R, 61R, 51R. 123

Leucine amino peptidase (LAP)

Gel 1

Gel2

------

Figure 6.10 Schematic representation of the zymogram of the LAP loci in P. haileyana (Ph), P. nervosa (Pn) and putative hybrid seedlings (numbered). a. L-R: Pn, Ph, 47R, 46R, 54R, 57R, Pn, Ph, 45R, 60R, 59R, 56R. b. L-R: Pn, Ph, 42R, 62R, 40R, 41R, Pn, Ph, 39R, 34R, 61R, 51R. ,·------~

124

Glucose phosphate isomerase (GPI)

------

---_---.J _____------_

Figure 6.11 Schematic representation of the zymogram of the GPI loci in P. pandorana (Pp), P. nervosa (Pn) and putative hybrid seedlings (numbered). L-R: 4,5,6,Pp, Pn, 9, 10, 11, 12.

Shikimic acid dehydrogenase (SDH)

------

Figure 6.12 Schematic representation of the zymogram of the SDH loci in P. pandorana (Pp), P. nervosa (Pn) and putative hybrid seedlings (numbered). L-R: 4, 5, 6, Pp, Pn, 9, 10, 11, 12. 125

Shikimic acid dehydrogenase (SDH) Parental bands differed, with P. pandoran.a (one band) showing a slower allele than P. nervosa (3 bands). All siblings appeared to have a single pattern which may have

been the same as that of the <;( parent, P. pandorana, although lanes 4, 6, 9 and 11 had bands which were slightly lower (Fig. 6.12, p124). Again as apomixis has not bee:n reported for the family Bignoniaceae, these seedlings are unlikely to be genetically identical to the female parent. The absence of any bands from the male parent is more likely to be accounted for by factors such as different aged tissues as discussed for LAP in the P. baileyana x P. nervosa seedlings.

Results for both putative hybrids, P. baileyana x P. n.ervosa and P. pandorana x P. nervosa, indicate that differences exist between some enzyme systems of the parental genotypes and offspring. For the former offspring, the hybrid status is corroborated by the isozyme analysis. However, for the latter offspring, their hybrid nature was not a.s clearly defined. Whilst further work would be needed to confirm their status as hybridsl the results obtained justify maintaining the seedlings until they develop mature foliage and flowers and can be compared morphologically with the parents. Seedlings of P. pandorana have taken three years to flower from seed under the plant growing conditions used for this study. The juvenility period of P. baileyana and P. nervosa is unknown. The seedlings produced as a result of interspecific crosses in 1990/1991 are not expected to flower until the 1993/1994 flowering season. The oldest are currently 18 months old and still exhibit all juvenile foliage.

6.2.5 Tissue Culture All four species of Pandorea were successfully established in aseptic culture using the basal medi_um (Table 2.2, p35). Results for the tissue culture of four different explants of P. jasmin.oides are presented (Table 6.9, p126).

Root explants did not produce any shoots at all and will not be discusssed further. Shoot tip explants produced significantly fewer shoots than horizontally laid nodal explants. Total shoot length and the total number of new nodes was also significantly lower for shoot tip explants. The differences in those three measurements were not significant between nodal ex plants inserted horizontally and those inserted vertically. I

126 Despite these differences, the most important measurement to consider when bulking material is the number of new shoots resulting from each meristem. When the explant types are compared there is no significant difference between them.

The use of horizontally laid nodes was an attempt to overcome the very strong apical dominance which exists in Pandorea. This was not successful. All explants showed more growth at the node which had originally been closer to the shoot apex.

Table 6.9 Comparison of four different expJants of P. jasminoides on the number of shoots, total shoot height, total number of nodes and the number of nodes per meristem.

Explant Meristems Total shoot Total shoot Total Number of per rep. number length number shoots per (mm) of nodes meristem

root 0.00 0.0 0.0 0.00

shoot tip 10 5.00 11.5 12.0 1.00

nodes 20 11.60 18.5 21.6 1.16 (vertical) nodes 40 17.20 25.1 29.8 0.86 (horizontal)

(L.S.D., p=0.05) (5.50) (9.0) (10.3) (0.33) 127

6.3 SUMMARY

Interspecific pollinations between species of Pandorea result in the formation of hybrid embryos. Embryos abort if the fruit is left to mature on the parent plant. Endosperm failure is the likely cause of embryo abortion but requires verification. Embryo culture was most successful if the embryo was large and well-developed at the cotyledonary stage. Precocious germination of immature embryos was common. Isozyme assays of some of some enzyme systems corroborated the hybrid status of P. baileyana x P. nervosa seedlings and provided justification for maintaining those seedlings until their horticultural potential can be assessed. Pandorea species are readily established and maintained in vitro. Both the species and the hybrid seedlings are readily acclimatized following in vitro culture. 128 CHAPTER 7

CONCLUSIONS AND FUTURE PROSPECTS

The genus Pandorea is a native climber with considerable horticultural potential. Three species, P. jasminoides, P. pandorana and P. nervosa have major horticultural potential, the two former species have small established markets in Australia as landscape plants but the latter species is not readily available to the horticultural industry. Pando rea jasminoides has white to pink tubular flowers Scm long and 5cm broad with a maroon centre. Pandorea pandorana has smaller flowers averaging 2cm by 1.5cm but colour variation from cream through pale to bright yellow to a brown and often with maroon markings on both the inside and outside of the corolla. The third species, P. n.ervosa, has thin, tubular white flowers the length of P. jasminoides and the width of P. pandoran.a.

Pollen of Pandorea species is bicellular, 35~-tm in diameter, and is tricolporate with finely reticulate exine. The pistil features a superior ovary containing 76 - 312 anatropous and tenuinucellate ovules. The style is hollow, containing a mucilage-filled canal bounded by epithelial cells. The stigma is bilobed, the lobes opening when receptive and closing rapidly after pollination. The s6gma is covered with a liquid exudate when receptive and conforms to the wet stigma type (Heslop-Harrison, 1981).

Studies of the reproductive biology of the genus Pandorea show that artificial pollinations can be carried out from before anthesis. The stigmas were receptive and could be pollinated prior to anthesis provided that the stigma lobes were separated to expose the receptive inner surfaces.

The genus Pandorea has floral features which promote outcrossing and minimize self­ pollination. In fact, several factors point to obligate outcrossing in the genus. Fir~~ species have stigmas which are sensitive to mechanical stimulation except P. be The stigma lobes show physical movement, closing within sec 129 pollinator activity. The response prevents the dislodgement of pollen from the stigma as the pollinating insect emerges from the flower. Deposition of self-pollen on the stigma is minimized because the receptive surfaces of the stigma are no longer exposed. Second, temporal separation of male and female phases in P. pandorana indicates that this species is protogynous not prota.ndrous like some other members of the family Bignoniaceae (Bertin, 1982). During periods of high pollinator activity, it is normal for a stigma to be pollinated before the anthers of the flower dehisced. Third, the presence of a self-incompatibility mechanism precludes any type of breeding system other than obligate outcrossing.

The genus has a gametophytic self-incompatibility sytem with the incompatibility barrier occurring at the level of the ovule. It may occur at the nucellus as for Acacia (Kenrick et. al. , 1986) or occur post-zygotically as for Theobrom.a cacao (Cope, 1962) and in some species of Rhododendron (Williams et. al., 1984). The late-acting self­ incompatibility discussed by Seavey and Bawa (1986) has many features which apply to Pandorea. Histological studies are now required to pinpoint the actual site of the incompatibility barrier.

Pandorea pandorana and P.jasm.inoides have an almost complete self-incompatibility system. Pandorea baileyana is also self-incompatible but the self-incompatibility is less efficient as judged from the small numbers of pollinations performed in this study. Cryptic self-fertility, where selfed progeny are produced if the pollen deposited on the stigma is a mixture of both self and outcross pollen has been found for Campsis radicans (Bertin and Sullivan, 1988; Bertin et. al., 1989). It is probable that it occurs in Pandorea. For these plants, pollinators may carry a mixture of self- and outcross pollen and so deposit a mixture on any one stigma. The self-pollinations would be more likely to be geitonogamous rather than autogamous because the insect would be leaving the flower as it picked up pollen from that flower. It could then move to a flower on the same plant or one on a different plant. During peak flowering the pollinator may move less frequently between plants and so would carry an increasingly higher percentage of self-pollen. Therefore, the most efficient pollen transfer in terms of reproductive success is likely to be at the beginning and end of the flowering peak when pollinators are forced to change plants frequently (Stephenson, 1982). 130 The pollen-ovule ratios for P. pandorana and P. jasminoides are low for a plant which appears to contain obligate outcrossers. While the pollen-ovule ratio cannot be considered a definitive indicator of the nature of the breeding system, the low ratio suggests that the genus may not be as totally self-incompatible as it would appear from experimental pollinations using only self- or outcross pollen and from the extent of floral adaptations. There is a possibility that self-incompatibility could be overcome through pioneer or mentor pollen techniques to allow for production of inbred lines for breeding purposes (Knox et. al., 1972a; Knox et. al., 1972b; Stettler and Ager, 1984).

Interspecific hybrid embryos were formed in all interspecific crosses attempted. Embryo culture has successfully rescued some interspecific hybrid embryos which were otherwise viable but which aborted if left to mature on the parent plant. The developmental stage at which the embryos are arrested or abort depends on the species pair and also on the direction in which the cross was performed. Bertin (1985, 1988, 1990) has found that even among outcross pollen donors, some are more successful than others in producing progeny. Embryos which were arrested prior to the cotyledonary stage could not be successfully cultured but as they represent crosses between the species P. pan.dorana and P. jasmin.oides which are already successful in commercial horticulture, further development of media is warranted to enable the production of those immature interspecific hybrid embryos. Histological examinations are required to confirm that embryo abortion occurs as a result of endosperm failure. Knowledge of the chromosomal behaviour at the interspecific level both at the time of embryo development and in terms of the fertility of the hybrid plants would be useful.

Isozyme studies have not been performed for the family except for limited analysis on Campsis radicans (Bertin and Sullivan, 1988) and were useful for corroborating the hybrid status of some of the seedlings resulting from this study. Further work on methodology would provide definitive answers on hybrid identity. Isozyme studies may also be valuably employed in taxonomic studies of the family Bignoniaceae to provide an independent line of evidence on the biosystematic questions on the origins of taxonomic groups and on the validity of taxa which are difficult to define morphologically. 131 The basic techniques of conventional plant breeding used here provide unique genetic combinations as the basis for new horticultural varieties of Pandorea. The successful manipulation of the hybrids embryos is dependent on an improved understanding of the nature of reproductive barriers. The refinement of embryo culture techniques, especially the prevention of premature germination of immature embryos, would increase the numbers of interspecific genetic combinations available for screening. For the ornamental plant industry, the possible infertility of hybrid plants is of little importance as material would be vegetatively propagated. Species and hybrids of Pandorea can be established and maintained with ease in aseptic culture and subsequently acclimatized. The new interspecific hybrids developed in this study can form a basis in the future for the production of elite clones with horticulturally desirable features. 132 APPENDIX

SELECTION AND SEEDLINGS RESULTING FROM THIS STUDY

Selections The following selections have been made for P. jasminoides: LGF - large-flowered, pale pink. DKP - medium-flowered, dark pink. The following selections have been made for P. pandorana: MM - sn1all-flowered, cream with extensive maroon markings, small leaves, less vigorous than other forms. BF - Broad flowers, deep yellow. TS 1 - Medium-flowered, rounded corolla lobes, maroon centre.

Intraspecific Seedlings Seedlings from the following intraspecific pollinations will be 1naintained until their floral characteristics can be assessed: P. jasm.inoides - DKP x LGF, HW x LGF. P. pandorana - AF x MM, AF x SCH, AL x GS, GS x SCH, MM x GS, MM x VC, TH x GS, TH x MM, TH x SCH, VC x GS, VC x MM.

Interspecific Seedlings Seedlings from the following interspecific pollinations, obtained by embryo culture, are being maintained until their floral characteristics can be assessed: P. baileyana x P. n.ervosa P. pandorana x P. nervosa P. pandorana x P. baileyana 133

P. jasminoides (DKP)

P. pandorana (MM) ' 134 REFERENCES

Adams WT, J ol y RJ ( 1980) Genetics of allozyme variants in loblolly pine. J Hered 71:33-40.

Agnihotri A, Lakshmikumaran M, Shivanna KR, Jagannathan V (1990) Embryo rescue of interspecific hybrids of Brassica spinescens x Brassica campestris and DNA analysis. Progress in Plant Cellular and Molecular Biology. Eel. HJJ Nijkamp, LHW van der Plas, J van Aartrijk). Kluwer.

Andrews ( 1801) Botanical Register, t. 81 *.

Armstrong JE ( 1985) The delimitation of Bignoniaceae and Scrophulariaceae based on floral anatomy and the placement of problem genera. Am J Bot 72(5):755-766.

Baker HG, Baker I (1979) Starch in Angiosperm pollen grains and its evolutionary

significance. Am J Bot 66:59'1-600.

Bawa KS ( 197 4) Breeding systems of tree species of a lowland tropical community.

Evolution 28:85-92.

Bawa KS, Webb CJ (1984) Flower, fruit and seed abortion in tropical forest trees: implications for the evolution of paternal and maternal reproductive patterns. Am J

Bot 71 (5): 736-751.

Beard sell DV, Knox RB, Williatns, EG (1988) Reproductive biology of Thryptomene calycina (Lind1.) Stapf. Pollination '88. Ed. RB Knox, MB Singh, LF Trioani.

Univ Melb. ISSN 0814.

Bertin RI ( 1982) Floral biology, humming bird pollination and fruit production of trumpet creeper (Cmnpsis radicans, Bignoniaceae). Am J Bot 69:122-134.

* not sighted 135

Bertin RI ( 1985) Nonrandom fruit production in Campsis radicans: between-year consistency and effects of prior pollination. Am Nat 126:750-759.

Bertin RI (1988) Paternity in Plants. Plant Reproductive Ecology, Patterns and Strategies. 30-39. Ed. JL Lovett-Doust, LL Lovett-Doust. Oxford Univ Press, NY.

Bertin RI, Sullivan M ( 1988) Pollen interference and cryptic self-fertility in Camps is radicans. Am J Bot 75:1140-1147.

Bertin RI, Barnes C, Guttman SI ( 1989) Self-sterility and cryptic self-fertility in Campsis radicans (Bignoniaceae). Bot Gaz 150:397-403.

Bertin RI ( 1990) Paternal success following 1nixed pollinations of Camps is radicans. Am Midl Nat 124: 153-163.

Brewbaker JL (1957) Pollen cytology and self-incompatibility systems in plants. 1 Hered 48:271-277.

Brewbaker JL ( 1967) The distribution and phylogenetic significance of binucleate and trinucleate pollen grains in the angiosperms. Am J Bot 54:1069-1083.

Brown R ( 181 0) Prod Flor Novae Hollandiae Ins Van Diem en, 471.

Buttrose, MS, Grant WJR, Lott JNA (1977) Reversible curvature of style branches of Hibiscus trioniumL., a pollination mechanism. Aust J Bot 25:567-570.

Buurman J ( 1977) Contribution to the pollen morphology in the Bignoniaceae with special reference to the tricolpate type. Pollen et Spores 19:447-519.

Byrne DH, Littlejohn TG (1989) Interspecific hybrid verification of plum x apricot hybrids via isozy1ne analyses. HortSci 24:132-134. 136

Ceccarelli N, Lorenzi R, Alpi A (1981) Gibberellin biosynthesis in Phaseolus coccineus suspensor. Z. Ptlanzenphysiol 102:37-44.

Chauhan SVS, Yadav V, Yadav DK (1987) Studies into causes of seedlessness in some Bignoniaceae. J Exp Bot 38(186):173-177.

Chu lYE, Kurtz SL (1990)Commercialization of plant micropropagation. Handbook of Plant Celli Culture 5. Ornamental Species. MacGraw-Hill NY.

Coleman A W, Goff LJ ( 1985) Applications of t1uorochromes to pollen biology I. Mithra1nycin and 4' ,6-diainidino-2-phenylindole (DAPI) as vital stains and for quantitation of nuclear DNA. Stain Techno! 60:145-154.

Collins GB, Grosser JW ( 1984) Culture of en1bryos. Cell culture and Somatic Cell Genetics of Plants 1:241-257. Ed. IK Vasil. Acadetnic Press NY.

Cope FW ( 1962) The mechanism of pollen incompatibility in Theobro1na cacao. Heredity 17:157-182.

Coulthart M, Denford KE (1982) Isozyme studies in Brassica I. Electrophoretic techniques for leaf enzymes and comparison of B. napus, B. can1pestris and B. oleracea using phosphoglucomutase. Can J Plant Sci 62:621-630.

Crane MB, Lewis D (1942) Genetical studies in pears III. Incompatibility and sterility. J Genet 43:31-43.

Cranwell L ( 1962) Endemism and isolation in the Three Kings Islands NZ with notes on pollen and spore types of the endemics. Rec Auckl Inst Mus 5:215-232.

Cronquist A ( 1981) An integrated systen1 . of classification of flowering plants. Columbia Univ Press. 137

Crud en RW ( 1977) Pollen-ovule ratios: a conservative indicator of breeding systems in flowering plants. Evolution 31:32-46.

Cruden RW, Miller-Ward S (1981) Pollen-ovule ratio, pollen size and the ratio of stigmatic area to the pollen-bearing area of the pollinator: an hypothesis. Evolution 35:964-974.

Darrah WC (1939) Textbook of Paleobotany. Appleton Century NY.

Darwin C ( 1877) The effects of cross and self fertilization in the vegetable kingdom. Appleton, NY.

Davis GL ( 1966) Systematic Embryology of the Angiosperms. Wiley NY.

Don G (1838) Tecoma jasminoides. Gen Syst 4:224.

Duggar BM ( 1899) On the development of the pollen grain and the embryo sac in Bignonia venusta. Bull Torrey Bot Club 26:89-105, Plates 352-354.

Elphick P (1985) Comtnercial production of Banksias. Horticulture of Australian Plants. Ed. BB Lamont, PA Watkins. WA Dept Ag, South Perth, WA.

Elrod MN (1904) Botanical notes. Proc Indian Acad Sci 1903:911.

Endlicher ( 1839) Genre Pandorea. Genera Plantarum, 177.

Erdtman G ( 1952) Pollen morphology and plant taxonomy of angiosperms. Almquist & Wiksell, Stockholm.

Fabris HA (1965) Bignoniaceae. Flora Argenti~a. Revista Mus La Plata, Secc Bot nov ser 9:273-419. 138 Faegri K, van der Pijl L (1979) The principles of pollination ecology. 3rd edn. Pergan1on Press.

Faragher JD (1986a) Postharvest physiology of waratah inflorescences (Telopea speciosissima, Proteaceae). Sci Hortic 28:271-279.

Faragher JD ( 1986b) Effects of cold storage methods on vase life and physiology of cut waratah in florescences (Telopea speciosissima, Proteaceae). Sci Hortic 29: 163-171.

Faragher JD ( 1989) Telopea. CRC Handbook of Flowering VI. Ed. AH Halevy. CRC Press, Florida.

Ferguson IK, Santisuk T (1973) Notes on the pollen morphology of some Asiatic Bignoniaceae. Kew Bull 28(2): 187-194.

Findlay GP (1978) Moven1ent of the column of Stylidium crassifolium as a function of tetnperature. Aust J Plant Physiol 5:477-484.

Findlay GP, Findlay N ( 1981) Respiration dependent movement of the column of Stylidium graminifolium. Aust J Plant Physiol 8:45-56.

Findlay GP, Pallaghy CK (1978) Potassium chloride in the motor tissue of Stylidium. Aust J Plant Physiol 8:45-56.

Findlay N, Findlay GP ( 1984) Movement of potassium ions in the motor tissue of Stylidiwn. Aust J Plant Physiol 11:451-457.

Findlay N, Findlay GP (1989) The structure of the column in Stylidium. Aust J Bot 37:81-101.

Franchi GG, Pacini E (1988) Pollen polysaccharide reserves in some plants of economic interest. Sexual Reproduction in Higher Plants, 473. Ed. M Cresti, P Pacini. Springer-Verlag, Berlin. 139

Fripp YJ ( 1982) Allozyrne variation and mating system in two populations of Eucalyptus kitsoniana (Luehm.) Maiden. Aust For Res 13:1-10.

Fuss AM, Sedgley M ( 1990) Floral initiation in relation to the time of t1owering in Banksia coccinnea R.Br. (Proteaceae). Aust J Bot 38:487-500.

Fuss AM, Sedgley M (199la) Pollen tube growth and seed set of Banksia coccinea R.Br. (Proteaceae). Ann Bot 68:377-384.

Fuss AM, Sedgley M (199lb) The developrnent of hybridization techniques for Banksia menziesii for cut flower production. J Hart Sci 66(3):357-365.

Gentry AH (1972) The type species of Bignonia L. Taxon 215/6):659-664

Gentry AH (1973) Flora of Panama. Part IX: Bignoniaceae. Ann Missouri Bot Gard 60(3) :781-977.

Gentry AH ( 1973a) Generic limitations of Central American Bignoniaceae. Brittonia 25:226-242.

Gentry AH ( 1973b) Studies in Bignoniaceae VII. Dendrosicus, Enallagma and Amphirecna. Taxon 22:637-640.

Gentry AH (1974a) Coevolutionary patterns in Central American Bignoniaceae. Ann Missouri Bot. Gard 61:728-769.

Gentry AH ( l974b) Flowering phenology and diversity in Central American Bignoniaceae. Biotropica 6:64-68.

Gentry AH ( 1976) Bignoniaceae of southern Central America: Distribution and ecological specificity. Biotropica 8(2):117-131. 140

Gentry AH (1980) Flora Neotropica Monograph No. 25. Bignoniaceae - Part I (Crescentieae and Tourrettieae). The New York Bot Gard, NY.

Gentry AH (1990) Evolutionary patterns in Neotropical Bignoniaceae. Mem New York Bot Gard 55:118-129.

Gentry AH, Tomb AS (1979) Taxonomic implications of Bignoniaceae palynology. Ann Missouri Bot Gard 66:756-777.

Gentry HS ( 1942) Rio Mayo Plants: A study of the flora and vegetation of the valley of the Rio Mayo, Sonora. Carnegie Inst Washington Publ 527:239-240.

Gleisner G, Ricardi M ( 1969) Revision del Genera Argylia (Bignoniaceae). Gayana 19:17-20.

Goldblatt P ( 1989) Miscellaneous chromosome counts in the Asteraceae, Bignoniaceae, Proteaceae and Fabaceae. Ann Missouri Bot Gard 76:1186-1188.

Goldblatt P, Gentry AH (1979) Cytology of Bignoniaceae. Bot Not 132:475-482.

Gotnes JC J r ( 1955) Contribuicao a sistetnatica das Bignoniaceae Brasileiras. Arq Serv Florest 9:261-296*.

Govindu HC ( 1950) Studies in the etnbryology of some rnembers of the Bignoniaceae. Proc Indian Acad Sci 32: 164-178.

Graham A, Jarzen DM (1969) Studies in Neotropical paleobotany I. The Oligocene com1nunities of Puerto Rico. Ann Missouri Bot Gard 56:308-357.

Guinet Ph (1962) Pollen d' Asie tropica1e. Inst Francais Pondichery. Fascicule I. Travaux de Ia section scientifique et technique. Tome V. I 141

Harborne JB ( 1967) Comparative biochemistry of the flavonoids VI. Flavonoid patterns in the Bignoniaceae and Gesneriaceae. Phytochem 1966: 1643-1651.

Harborne JB, Smith DM (1978) Correlations between anthocyanin chemistry and pollination ecology in the Polemoniaceae. Biochem Syst Ecol 6: 127-130.

Heslop-Harrison Y (1981) Stigma characteristics and angiosperm taxonomy. Nord J Bot 1(3):401-420.

Heslop-Harrison Y, Shivanna K (1977) The receptive surface of the angiosperm stigma. Ann Bot 41:1233-1258.

Hu CY (1975) In vitro culture of rudimentary embryos of eleven flex species. J Arn Soc Hort Sci 100:221-225.

Hu CY and Wang P (1981) Embryo culture: techniques and applications. Handbook of Plant Cell Culture 4. Ed. DA Evans, WR Sharp, PV Atnmirato. Macmillan.

Hunter JA (1967) A further note on Tecomanthe speciosa WRB Oliver (Bignoniaceae). Rec Auckland Inst Mus 6:169-170.

Hwang Y -J, Yoshikawa K, Okubo H ( 1991) Identification of interspecific hybrids between Cmnellia rericulata and C. chrysantha by isozyme analysis. Scientia Hortic 48:269-275.

Johansen DA (1950) Plant En1bryology: Embryogeny of the Spern1atophyta. 4th edn, Waltham, Mass.

Kaul V, Theunis c, Palser b, Knox RB, Willia1ns EG (1987) Association of the

generative cell and vegetative nucleus in th~ pollen tubes of Rhododendron. Annals Bot 59:227-235.

Kenrick J, Kaul V, Williams EG (1986) Self-incompatibility in Acacia retinodes: the site of pollen tube arrest is the nucellus. Planta 169:245-250. 142

Kerr T, Anderson DE ( 1944) Osmotic quantities in growing cotton bolls. Pl Physiol 19:338-349.

Kin MS, Fraser LG, Harvey CF ( 1990) Rescue of hybrid embryos of Actinidia species. Scientia Hortic 44:97-106.

Knox RB, Willing RR, Ashford AE ( 1972a) Pole of pollen-wall proteins as recognition substances in interspecific incompatibility in poplars. Nature 237:381-383.

Knox RB, Willing RR, Pryor LD (1972b) Interspecific hybridization in poplars using recognition pollen. Silvae Genet 21:65-69.

Laibach F ( 1925) Das Taubwerden von Bastardsamen und die kunstliche Aufzent fruh absterbender Barstarde1nbryonen. Z Bot 17:417-459.

Laibach F ( 1929) Ectogenesis in plants. Methods and genetic possibilities of propagating en1bryos otherwise dying in the seed. J Hered 20:201-208.

Leeton P, Fripp YJ ( 1991) Breeding systems , karyotype and variation within and between populations of Rutidosis leptorrhynchoides F. Muell. (Asteraceae: Inulae).

Aust J Bot 39:85-96.

Lindley J ( 183 8) Tecotna jasminoides. Edward's Botanical Register, New Ser X:2002.

Lloyd DG ( 1965) Evolution of self-incompatibility and racial differentiation in Leavenworrhia (Cruciferae). Contr Gray Herb 195:3-134.

Maiden JH, Baker RT (1895) New species of plants from New South Wales. Proc Linn Soc NSW (2nd ser) X:592-595, plate LI.

Maheshwari P (1948) The angiosperm embryo sac. Bot Rev 14:1-56. 143

Martin FW (1959) Staining and observing pollen tubes 1n the style by means of fluorescence. Stain Technol 34:125-128.

Mauritzon J ( 1935) Etwas uber die Embryologie der Bignoniaceen. Bot Not 1935:60- 77.

McCullagh P, N elder J A ( 1989) Generalized linear models. Chapn1an Hall, London.

Misra S, Green M ( 1990) Developmental gene expression in conifer embryogenesis and gennination I. Seed proteins and protein body composition of the n1ature embryo and the megagametophyte of white spruce (Picea glauca [Moench](Voss)). PI Sci 68:163-173.

Mitra K (1968) Pollen morphology in Bignoniaceae in relation to taxonotny. Bull Bot Surv India 10(384):319-324.

Mitton JB, Linhart YB, Sturgeon KB, Hamrick JL (1979) Allozyme polytnorphisms detected in mature needle tissue of ponderosa pine. J He red 70:86-89.

Mohl H (1835) Sur la structure et les fonnes de grains de pollen. Ann des Sci Naturelles, 2 ser T. III, 2:304-346.

Monnier M (1976) Culture in vitro de le'embryon immature de Capsella bursa-pastoris Moench(l). Rev Cytol Bioi Veg 39:1-120.

Monnier M (1982) Culture of mature ectocotyledonous embryos of Phaseolus vulgaris and the nutritional role of the cotyledons. Am J Bot 69:896-903.

Muller F ( 1868) Befruchtungsversuche an Cip6 alho (Bignonia) Bot Zeit 39:625-629.

Murashige T, Skoog F (1962) A revised medium for the rapid growth and bioassays with tobacco tissue culture. Physiol Plant 15:473-497. 144

Nakajima G (1936) Chromosome numbers in some crops and wild angiosperms. Japan J Genet 12:211-218.

Nettancourt D de (1977) Incompatibility in Angiosperms. Springer Verlag Berlin.

New cOin be FC ( 1922) Significance of the behaviour of sensitive stigmas. Amer J Bot 9:99-119.

NewcOinbe FC ( 1924) Significance of the behaviour of sensitive stigmas II. Amer J Bot 11:85-93.

Ngouela S, Tsamo E, Sondengam BL (1988) Extractives from Bignoniaceae: Constituents of the stem bark of Spathodea campanulata. Planta Medica 5. Oct:476.

Niederwieser SG, van der Venter HA, Robbertse PJ (1990) Embryo rescue m Ornithogalum. HortSci 25:565-566.

Norstog K ( 1979) Embryo culture as a tool in the study of comparative and developmental morphology. Plant Cell and Tissue Culture 179-202. Ed. WR Sharp, PO Larsen, EF Paddock, V Raghavan. Ohio State Univ Press.

Oliver WRB (1948) The flora of the Three Kings Islands. Rec Auckl Inst Mus 3:211- 238.

Paclt J ( 1952) Synopsis of the genus Catalpa (Bignoniaceae). III. Candollea 13:241- 285.

Pacini E, Franchi GG (1988) Amylogenesis and amylolysis during pollen grain development. Sexual Reproduction in Higher Plants, 181-186. Ed. M Cresti, : Gori, E Pacini. Springer-Verlag, Berlin. 145 Petersen C, Brown JH, Kodric-Brown SA (1982) An experimental study of floral display and fruitset in Chilo psis linearis (Bignoniaceae). Oecol 55:7-11.

Pichon M (1945) Notes sur les Bignoniacees. Bull Soc Bot France 92:222-229.

Preston RE ( 1986) Pollen-ovule ratios in the Cruciferae. Am J Bot 73:1732-1740.

Raghavan V ( 1965) Hormonal control of embryogenesis in Caps ella. Proc International Conference on Plant Tissue Culture 357-369. Ed. PR White, AR Grove. McCutcheon, Berkerley CA.

Raghavan V (1977) Applied aspects of embryo culture. Applied and Fundamental Aspects of Plant Cell, Tissue and Organ Culture 375-397. Ed. J Reinert, YPS Bajaj. Springer-Verlag Berlin.

Ramsey M, Vaughton G (1991) Self-incompatibility, protandry, pollen production and pollen longevity in Banksia menziesii. Aust 1 Bot 39:497-504.

Raven P ( 1975) The bases of angiosperm phylogeny: cytology. Ann Missouri Bot Gard 76: 1186-1188.

Reed SM, Collins GB ( 1978) Interspecific hybrids of Nicotiana through in vitro culture of fertilized ovules. J Hered 69:311-315.

Richardson BJ, Baverstock PR, Adams M ( 1986) Allozyme Electrophoresis. A Handbook for Animal Systematics and Population Studies. Academic Press, Sydney, 410pp.

Rietsema J, Satina S, Blakeslee AF (1953) The effect of sucrose of the growt Datura stramonium embryos in vitro. Am J Bot 40:538-545.

Rivjen AHGC (1952) In vitro studies on the embryo of Capsella bursa-pasr- .. : Bot Neerl 1:157-200. 146

Sandwith NY (1968) Kew Bull 22(3):403-420. ~

Sargent CS (1899) A hybrid Catalpa. Garden and Forest 2:303-305.

Sax K (1933) Species hybrids in Plantanus and Catalpa. J Arnold Arboretum 14:274- 278.

Schenck HR, Wolf G ( 1986) Characterization of somatic Brassica napus hybrids by polyacrilmnide gel electrophoresis. Plant Breeding 97:72-74.

Schutnann K (1894) Bignoniaceae. Engler und Prantl, Naturlichen Pflanzen Familien 4(36): 189-252.

Scogin R (1980) Anthocyanins of the Bignoniaceae. Biochem Syst Ecol 8:273-276.

Sears ER ( 1937) Cytological phenomena connected with self-sterility in the flowering plants. Genetics 22: 130-181.

Seavey SR, Bawa KS (1986) Late-acting self-incompatibility in Angiosperms. Bot Rev 52:195-219.

Sedgley M, Bell FDH, Bell D, Winks CW, Pattison SJ, Hancock TW (1990) Self- and cross-compatibility of macadamia cultivars. J Hortic Sci 65:205-213.

Seen1an (1870) J Bot 8:382:

Seibert RJ (1948) The use of glands in a taxonomic consideration of the family Bignoniaceae. Ann Missouri Bot Gard 35: 123-136.

Sjodin C, Glimelius K (1989) Transfer of resistance against Phoma lingan1 to Brassica napus by assymmetric so1natic hybridization combined with toxin selection. Theor Appl Genet 78:513-520. 147

Smith E (1894) Tecoma smithii. Gardner's Chron 16:64.

Smith EC ( 1941) Chromosome behaviour m Catalpa speciosa Spaeth. J Arnold Arboretum 22:219-221.

Smith JG (1971) An analytical approach to the culture of globular bean e1nbryos. Ph.D thesis. Univ Michigan, Ann Abor. ..

Soueges R (1940) Embryogenie des Bignoniacees. Developpement de l'embryon chez le Catalpa kaenzpferi Sieb, et Zucc. Acadetnie de Sciences 210: 116-118.

Spach E (1840) Genre Pandorea. Hist Veg Phan IX: 136-137.

Stearn WT (1953) Campsis x Tagliabuana. Curtis's Bot Mag 169: t.198 + 6pp.

Steenis CGGJ van ( 1927) Malayan Bignoniaceae, their taxonomy, origin and geographical distribution. Rec des Travaux Bot Neerland 24:787-1049.

Steen is CGGJ van ( 1929) A revision of the Queensland Bignoniaceae. Proc R Soc Qld 41:43-46.

Steen is CGGJ van (1931) A previously undescribed Pando rea fro1n north-east Queensland, Australia. J Arnold Abor 12:149-151, Plate 35.

Steenis CGGJ van (1977) Flora Malesiana Ser I Spermatophyta. 8:114-186.

Stephenson AG (1979) An evolutionary exarnination of the floral display of Catalpa speciosa (Bignoniaceae). Evolution 33:1200-1209.

Stephenson AG (1982) When does outcrossing occur in a mass-? Evolution 36:762-767. j

.I I 148 r Stephenson AG, Thomas WW (1977) Diurnal and nocturnal pollination of Catalpa !Jpeciosa (Bignoniaceae). Syst Bot 2: 191-195.

Stettler RF, Ager AA (1984) Mentor effects in pollen interactions. Cellular Interactions, 609-623. (Encycl Plant Physiol; new ser 17). Ed. HF Linskens, J Heslop-Harrison. Springer-Verlag.

Suryakanta ( 1973) Pollen morphological studies in the Bignoniaceae. J Palynol 9(1):45- 82.

Swamy (1942) Contribution to the life-history of Bignonia megapaton1ica. J Indian Bot Soc 20:299-305.

Thieret JW ( 1976) Floral biology of Proboscidea louisianica (Martyniaceae). Rhodora 78:169-179.

Thomas MB, McDonald WJF ( 1989) Rare and threatened plants of Queensland. Qld Dept Primary Ind. Inform Ser QI 87003.

Urban I (1916) Uber Ranken und Pollen der Bignoniaceen. Ber Deutsch Bot Ges Band 34(9):728-758.

Vasil IK, Johri MM (1964) The style, stigma and pollen tube - I. Phytomorphology 14:352-369.

Visser T (1983) Pollen and pollination experiments IV. "Mentor pollen" and "pioneer

pollen~~ techniques regarding incompatibility and incongruity in apple and pear. Euphytica 30:363-369.

Wardlaw CW ( 1965) Physiology of emb_ryonic development in Cormophytes. Encyclopedia of Plant Physiology 15/1:844-965. Springer-Verlag Berlin.

Watkins PA ( 1985) Commercial production of kangaroo paws. Horticulture of Australian Plants. Ed. BB Lamont, PA Watkins. W A Dept Ag. South Perth WA. 149

Watson W (1893) Tecoma smithii. Gardner's Chron 14:649-669.

Webb CJ (1984) Constraints on the evolution of plant breeding systems. Plant Biosystematics, 249-270. Ed. WF Grant. Academic Press, Toronto.

Weber A, Vogel S (1986) The pollination syndrome of Deplanchea tetraphylla. Pl Syst Evol 154:237-250.

Williams EG (1980) Hybrids between Trifolium ambiguum and T. hybridum obtained with the aid of embryo culture. NZ J Bot 18:215-200.

Williams EG ( 1987) Interspecific hybridization in pasture legumes. Plant Breed Rev 5:237-305.

Williams EG, De Lautour G (1980) Production of tetraploid hybrids of Ornithopus pinnatus and Ornithopus sativus using embryo culture. NZ J Bot 19:23-30.

Williams EG, Kaul V, Rouse J, Knox RB ( 1984) Apparent self-incompatibility in Rhododendron. ellipticum, R. championae and R. amamiense: a post-zygotic mechanism. Incompat Newsl 16:10-11.

Williams EG, Maheswaran G, Hutchinson JF (1987) Embryo and ovule culture in crop improvement. Plant Breed Rev 5:181-236.

Williams KA W (1987) Native Plants. Queensland. Vol 3. KA W Williams, Nth Ipswich,Qld

Willis JH (1972) A handbook to Plants in Victoria Vol 2. . Melb Uni Press, Melb.

Yeung EC (1980) Embryology of Phaseolus. The role of the suspensor. Z Pflanzenphysiol 96: 17-28.

Zimmerman M, Pyke GH (1988) Pollination ecology of Christmas Bells (Blandfordia n.obilis): Effects of pollen quantity and source on seed set. Aust 1 Ecol 13:93-99.

Minerva Access is the Institutional Repository of The University of Melbourne

Author/s: James, Elizabeth Ann

Title: Aspects of the reproductive biology, breeding system and horticultural improvement of the genus pandorea

Date: 1992

Citation: James, E. A. (1992). Aspects of the reproductive biology, breeding system and horticultural improvement of the genus pandorea. Masters Research thesis, School of Botany, The University of Melbourne.

Publication Status: Unpublished

Persistent Link: http://hdl.handle.net/11343/37251

File Description: Aspects of the reproductive biology, breeding system and horticultural improvement of the genus pandorea

Terms and Conditions: Terms and Conditions: Copyright in works deposited in Minerva Access is retained by the copyright owner. The work may not be altered without permission from the copyright owner. Readers may only download, print and save electronic copies of whole works for their own personal non-commercial use. Any use that exceeds these limits requires permission from the copyright owner. Attribution is essential when quoting or paraphrasing from these works.