Abstract
With increasing museum digitization efforts, we are beginning to realize the potential for discovery using big data and biodiversity archives. As 3-dimensional (3D) imaging with micro-computed tomography (micro-CT) becomes more common, workflows to streamline processes and develop novel studies using large-scale digital imaging are needed, especially for soft-tissue botanical specimens in a vertebrate-focused field. I present a case study on the 3D pollination biology of Theobroma cacao and its relatives that highlights discoveries possible through combining cutting-edge imaging modalities and big data, which I call Morphology 3.0. First, we used micro-CT, 3D geometric morphometrics, and scanning electron microscopy to precisely quantify plant-pollinator geometry, functional size limits for pollinators, and floral reward structures in cacao. Next, I applied these methods to study cacao and its relatives, and showcase their strength as a reproducible, precise toolkit to incorporate morphology into taxonomic evaluations of a group whose relationships are still actively being resolved. Finally, I repurposed floral micro-CT data to train the computer vision model, UNETR, to automate 3D image processing and shape prediction using MONAI and pytorch. These projects will contribute to future large scale biodiversity informatics analyses as digital archives and 3D imaging efforts grow, especially for botanical soft-tissue specimens.