Adaptive evolution of flowers to optimize pollen transfer by animal pollinators is considered a key driver of plant speciation and the success of flowering plants. Plant-pollinator interactions are commonly embedded in complex abiotic (climatic) and biotic (other plants/pollinators) contexts, and the structure and function of plant-pollinator interaction networks changes across such contexts. To date, however, we lack a broad-scale perspective on how varying a-/biotic contexts affect which flower phenotypes are ‘fit’, and how flowers evolve to adapt to these contexts. This knowledge gap limits our understanding of the processes that generate and maintain biodiversity, critically important in light of current global change.

In MountBuzz, I aim at developing a novel context-dependent ecological perspective on the processes structuring the evolution of flower diversity by linking the commonly separated fields of community ecology and macroevolutionary modelling. First, to determine which flower phenotypes are ‘fit’ (high reproductive success) in different a-/biotic environmental contexts, my team and I will analyze plant-pollinator interactions and flower and pollinator trait data along four elevational gradients across the tropics. We will combine empirical field observations with pollination experiments to pinpoint context-dependent changes in phenotype-fitness relationships. Second, synthesizing across these results, we will test whether patterns of flower macroevolution follow predictable, context-dependent trajectories by employing machine-learning based predictive modelling and phylogenetic comparative methods.

The results of MountBuzz will deliver a new perspective on the relative importance of pollinator-mediated selection and environment-dependent processes in driving flower evolution and plant diversification. The study set-up (cross-continental, cross-environmental, cross-lineages) further allows for identifying generalities in patterns, thereby delivering novel hypotheses for future research.

Approach

MountBuzz is structured into four major work packages:

Work package 1: Diverse teams consisting of researchers of the University of Vienna and researchers from each collaborating institution in the respective four countries will work together in the field to collect first hand natural history data on Melstomataceae flowers and their pollinators. Each team will spend several months in the field at each elevational gradient and document plant-pollinator interactions from lowland rainforests to montane cloud forests. The Melastomataceae plant family is mainly pollinated by bees (ca. 96 % of species), but in cloud forests, shifts to vertebrate pollinators or generalist insect assemblages have been documented in the South American tropics. Our study will, for the first time, investigate pollinator shifts in detail also in the African and Asian tropics, and provide a detailed assessment of the subtler changes occurring in the bee pollinator fauna along elevation. We will not only perform pollinator observations, but also monitor how plant mating systems, flower longevity and plant reproductive success change along elevation. Correctly identifying plants and pollinators will be a key task in work package 1, and we will tackle this task using both classic morphological assessments as well as molecular techniques to build phylogenies for both Melastomataceae and (bee) pollinators.

Work package 2: To understand under which (a-/biotic) contexts certain floral phenotypes are fitter than others, we additionally record a series of floral functional traits relevant for the interaction with pollinators. These traits include flower color, the spatial arrangement of reproductive organs and pollen reward quantity and quality as well as micromorphological differences between stamens captured through computed tomography. In addition, we will record traits of the bee pollinators we catch on Melastomataceae flowers to quantify how flower and bee traits match. We will connect these datasets using network statistics, trait-matching models and phylogenetic comparative methods.

Work package 3: The bee pollination system of Melastomataceae is functionally highly specialized so that only bees capable of producing vibrations between 100-400 Hz can effectively remove pollen from the flowers (“buzz-pollination”). In the third work package, we will expand ongoing efforts to studying the biomechanics of buzz-pollination in the wild by conducting pollination experiments using artificially synthesized vibrations across a wide range of Melastomataceae flowers. These experiments will help understand whether buzz-pollinated flowers are ‘tuned’ to different a-/biotic contexts and deliver new insights on the biomechanical make-up of flowers of the most diverse radiation of pollen flowers.

CT-scans of Melastomataceae stamens to study biomechanics adaptations to buzz-pollination by different bees (Medinilla magnifica, Graffenrieda cucullata, Blakea sp.)

Work package 4: In the final phase of the project, we will synthesize across work packages 1-3 and test for generalities in the patterns of the distinct community-level interaction networks. Specifically, we will use machine-learning-based predictive modelling to test whether flower and bee traits may be used to predict interaction networks across study plots, gradients and continents. Further, we will pull in floral trait data from additional Melastomataceae species and explore whether patterns of floral trait differentiation show a signature of niche expansion into altered a-/biotic contexts when incorporating phylogenetic relationships.