Modularity and evolution of flower shape: the role of efficiency, development, and spandrels inErica

Summary Three hypotheses can explain floral modularity: the attraction-reproduction, the efficiency, and the developmental hypotheses. In order to test these hypotheses and understand if pollination specialisation and pollination syndrome influence floral modularity, we focussed on the genus Erica : we gathered 3D data from flowers of species with diverse pollination syndromes via Computed Tomography, and analysed their shape via geometric morphometrics. In order to provide an evolutionary framework for our results we tested the evolutionary mode of floral shape, size, and integration under pollination syndrome regimes, and - for the first time-reconstructed the high-dimensional floral shape of their most recent common ancestor. We demonstrate, for the first time, that the modularity of generalist flowers depends on development and that of specialists is linked to efficiency: in bird syndrome flower, efficiency modules were associated with pollen deposition and receipt, whereas in long-proboscid fly syndrome, they were associated with restricting the access to the floral reward. Only shape PC1 showed selection towards multiple optima, suggesting that PC1 was co-opted by evolution to adapt flowers to novel pollinators. Whole floral shape followed an OU model of evolution, and demonstrated relatively late differentiation. Flower shape modularity thus crucially depends on pollinator specialisation and class.