Pollination and geographical divergence in flower colour of the drosera cistiflora species complex, or 'how the snotrosie got its colours'.

Abstract

The question of floral colour diversity among congeneric plants has intrigued evolutionary biologists since Lamarck—yet remains a hot topic amid the varied groundbreaking ecological insights that emerge to this day. I investigated potential causes of floral colour divergence in Drosera cistiflora sensu lato (Droseraceae), an insectivorous plant species complex which exhibits considerable variation over its geographical range in the Greater Cape Floristic Region of South Africa. Although several studies suggest that the foraging strategies of biotic pollination vectors can generate selective forces for floral trait diversification, no study has demonstrated unequivocally that pollinator-mediated selection is the core driver of shifts in flower colour. Indeed, selection by pollinating agents is not the sole possible explanation for floral colour disparity among populations, and other hypotheses, such as a role for edaphic factors, have been proposed as mechanisms modulating trends in flower colour. D. cistiflora s.l. is an exemplary study species complex for addressing these hypotheses as it displays remarkable heterogeneity in corolla colour—pink, purple,

red, white and yellow—both between and within populations, which occur across a range of soil types. My primary aim was to establish whether pollinators can explain spatial patterns of flower colour in the complex. The thesis is divided into the following chapters: Chapter 1 is a general overview of the theory of pollinator-driven geographical divergence in floral traits, with a focus on flower colour, and includes a detailed account of the study species complex. Chapter 2 is an investigation into the breeding systems of D. cistiflora s.l., to assess whether floral attributes may reflect adaptations for allogamy. I discovered the complex to be highly pollinator-dependent for seed production with variable low autonomous selfing ability among floral colour forms and evidence for pollen limitation of fecundity. Chapter 3 examines associations between floral colour variation and the pollinating fauna and abiotic factors that may have played a role in the evolution of sympatric and allopatric floral colour forms. I show that respective forms are associated with geographically variable pollinator communities dominated by hopliine beetles (Coleoptera: Scarabaeidae: Hopliini), and that floral colour shifts appear to be largely independent of abiotic factors. These findings present a case for pollinator-mediated floral colour divergence in D. cistiflora s.l. In Chapter 4 I present the results of experiments using arrays (of models matched to D. cistiflora s.l. petal spectra and of reciprocally translocated flowering plants) that tested whether local pollinator discrimination among colours can explain geographical variation in flower colour. Analyses show that flower visitors discriminated significantly among models varying in colour and also among translocated flowers of D. cistiflora s.l. forms, and that the flower colour of the local D. cistiflora s.l. phenotype was generally favoured by insect visitors over introduced colours. Differential floral colour preferences and selection exhibited by polylectic pollinator communities across the range of study populations led me to conclude that the floral colour forms represent geographically divergent ecotypes adapted to broad pollinator assemblages. Chapter 5 specifically assesses the level of pollinator isolation between sympatric purple- and red-flowered D. cistiflora s.l. forms, to determine whether pollinators can maintain flower colour in the absence of macrogeographical barriers. Here, morphological observations and reproductive isolation indices demonstrate that strong pre-F1 barriers to gene flow between the two forms are governed largely by floral isolation, through differences in flower colour and beetle pollinator preferences.

Evidence of early-acting postzygotic isolation is revealed by the significantly lower seed set from inter-colour hand cross-pollinations than from crosses within forms. Results support the taxonomic classification of endangered purple- and red-flowered D. cistflora s.l. populations above the rank of form and thereby signal their unprecedented conservation need. The thesis thus provides new evidence for adaptive floral divergence driven by a generalist pollinating fauna and demonstrates how the study of pollination ecotypes may benefit red-listing and conservation of threatened plant populations with poorly understood taxonomic limits. Lastly, in Chapter 6, I offer a summary of my main findings and their evolutionary, taxonomic and conservation significance, and outline key areas for further research.