NERC/NSF Mind the Gap: Tackling the Molecule–Fossil Divide in Angiosperm Evolution Using Fossil Pollen, Superresolution Microscopy, and Deep Learning

Reconstructing the evolutionary history of angiosperms is vital to explaining their spectacular rise to ecological dominance. However, the timing of angiosperm diversification and the evolutionary relationships between major lineages are currently unresolved, with molecular dates for their origin typically tens of millions years older than the dates of their first appearance in the fossil record. This has created an impasse, with successive molecular phylogenetic studies consistently providing older origination dates than fossils, and the fossil record yielding no convincing angiosperm fossils of a comparable age. We propose to bridge the gap between molecular and fossil origination dates using pollen grains, a category of terrestrial microfossils that have been largely neglected in research on the early evolution and diversification of angiosperms. Pollen grains are abundant in Early Cretaceous sediments worldwide, but methodological limitations and the ambiguous taxonomic status of key taxa mean that the evolutionary information that pollen grains can provide has not yet been fully harnessed in interpreting early angiosperm evolution. We will overcome these barriers by employing methodological innovations developed and pioneered by our team. Specifically, we will use superresolution microscopy to increase the amount of morphological information that can be captured from fossil pollen and make more precise assessments of botanical affinity than is possible using traditional microscopy techniques such as transmitted light. This morphological detail will be quantified and interpreted within a phylogenetic framework through the use of machine learning. This will allow us to produce more accurate and fine-scale assessments than would be possible by a human analyst alone. Our research is focused on the Chloranthaceae, an early-diverging lineage of angiosperms that occupy a critical position in angiosperm evolution. We will build a new phylogenomic tree of this group and will use novel deep learning methods to place Early Cretaceous (c.125 million years ago [Ma]) fossil pollen grains with putative affinities to the Chloranthaceae in the phylogeny, determining whether the fossils are early- or late-diverging members of the Chloranthaceae or fall outside the family. By establishing the timing of Chloranthaceae diversification, we will provide a new window into the tempo and timing of early angiosperm evolution. Recent phylogenomic work indicates modern Chloranthaceae diversified as part of an explosive evolutionary radiation of angiosperms in the Late Jurassic and Early Cretaceous (154–c.125 Ma), and our approach will use fossil pollen to date the diversification of Chloranthaceae and allow us to evaluate this hypothesis empirically. Our results will also help resolve the timing of diversification of early angiosperms more widely, by constraining the age of lineages at the base of the angiosperm phylogeny, and will provide new insights into the diversification of angiosperms at a critical period in their evolution, transforming knowledge of the Early Cretaceous terrestrial biosphere. Our research will provide unique training in cutting-edge methods for two postdoctoral researchers, and will support the pursuit of international biodiversity targets through the development and application of quantitative methods to characterise biodiversity across scales.