Abstract

Adaptive radiation, the diversification of species from a common ancestor into ecologically distinct forms, can occur following the evolution of key innovations. Such innovations require sufficient genetic variation, yet only a subset of emerging traits will ultimately contribute to diversification. In plants, sexual deception, in which flowers mimic female insects to attract male pollinators that attempt pseudocopulation, is considered a potential key innovation. However, few sexually deceptive systems include multiple co-occurring phenotypes, limiting our ability to link insect-mediated gene flow with underlying genetic modules and their phenotypic consequences. Gorteria diffusa, a South African daisy, is unusual in possessing a spectrum of sexually deceptive morphotypes. Over the past approximately 1.5 million years, the species has radiated into 16 morphotypes, several of which hybridise in narrow secondary contact zones. This diversification is hypothesised to stem from the evolution of sexually deceptive petal spots that mimic female Megapalpus capensis bee flies, although this remains an ongoing area of research. To address this, we: 1) sampled 500 G. diffusa individuals across four hybrid zones with varying levels of sexual deception in 2018; 2) conducted manual pollinator observations in 24 Gorteria populations in 2025; and 3) collected 280 M. capensis pollen loads in 2025 to examine M. capensis foraging patterns. Our Genotyping-by-Sequencing data identified six genomic clusters with hybrids occurring predominantly within these clusters. Admixture was strongest in central regions of the hybrid zones, and cline shapes varied, with the steepest between strongly deceptive and moderately or weakly deceptive morphotypes. Megapalpus visits in 2025 were low, 0.019 ± 0.0025 SE flies per flower, equivalent to one fly per 53 flowers. Megapalpus visited approximately eight species per 750 m2 and was 2.3 times more likely to forage on G. diffusa than on other Asteraceae, although deceptive and non-deceptive morphs received similar visit rates. Together, these results suggest that adaptive radiation in G. diffusa reflects complex interactions among genetic structure, pollinator behaviour, and both pre-zygotic and post-zygotic barriers, contributing to novel phenotypes and potential key innovations.