Floral Morphology and Development in Houstonia Procumbens (Rubiaceae), a Uniquely Distylous, Cleistogamous Species

Among flowering plants, floral form is inherently linked to reproductive success and is therefore a key element in the evolution of angiosperm lineages. Variations in floral form and associated mating systems have demonstrated that flowers are well adapted in many cases to promote predominantly either outcrossing or self-fertilization among members of a given lineage. The focus of this work is to examine the developmental interactions and consequent effects on the morphology of two such flower types within individuals of Houstonia procumbens (J.F. Gmel.) Standl. (Rubiaceae). The two flower types in question are distylous and cleistogamous. Distyly is a form of heterostyly, wherein the reproductive organs of flowers in a given population are spatially separated (herkogamy) and arranged reciprocally to one another (reciprocal herkogamy) among individuals that represent two alternate morphotypes. Cleistogamy refers to the production of both flowers that open (chasmogamous) to interact with pollinators and flowers that self-pollinate precociously in bud (cleistogamous) on individuals of a given species. While both of these pollination syndromes, distyly and cleistogamy, appear to have arisen convergently, multiple times, across angiosperms both pollination syndromes apparently co-occur in only one species H. procumbens, an annual, herbaceous member of the coffee family (Rubiaceae). The goal of this work is to examine the interactions of these two pollination syndromes in H. procumbens by comparing the form and development of the various flower morphs (distylous morphs and cleistogamous flowers) of this species to those of closely related taxa that produce either distylous or cleistogamous flowers, but not both. The first chapter of my dissertation establishes the degree to which discoveries about the heterostylous, cleistogamous flowering plant species H. procumbens are applicable to other species. I use literature searches to establish that other species that are both heterostylous and cleistogamous are currently unknown, and I use phylogenetic inference and ancestral state reconstruction for the Rubiaceae to establish that heterostyly arose 27-36 independent times in the family and that the origin of heterostyly for H. procumbens occurred in an ancestor that produced at least 820 extant, descendent species. I also show that the joint probability of heterostyly and cleistogamy in angiosperm families, genera, and species is quite low, and the observed numbers of taxa in which the two co-occur at the ranks of family and genus (but not species) is greater than expected, though not significantly so. The second chapter addresses the effect of cleistogamy on reproductive organ reciprocity for H. procumbens. I use comparisons among close relatives to show that the overall reciprocity of H. procumbens changed little with the evolution of cleistogamy but that reciprocity in a close distylous relative, H. caerulea, was sufficiently low to qualify it as "style dimorphic" by some standards--a situation that I suggest calls into question the use of the reciprocity index for such distinctions. I also demonstrate that H. procumbens does not have a pollen size or staining dimorphism, as is seen in other heterostylous members of Houstonia, making assessments of disassortative pollen transfer difficult. The third chapter examines the chasmogamous and cleistogamous flower development of H. procumbens and chasmogamous flower development of H. caerulea. I demonstrate that differences in the rate of elongation of styles and corollas lead to the difference in stigma and anther heights seen in the two chasmogamous morphs of H. procumbens and H. caerulea, and that this dimorphism in growth rates is established early in development. Additionally, I show that there is a relatively small anther-stigma distance in the early stages of floral development of H. caerulea, which is unexpected, since one might expect this to be a developmental pattern that makes cleistogamy (e.g., as in H. procumbens but not H. caerulea) evolutionarily simpler (by reaching sexual maturity in those earliest stages). Instead, I show that H. procumbens reduces the anther-stigma distance in its cleistogamous flowers by reaching sexual maturity when the corolla is short (and the anthers low) and by producing helical styles.