A TRANSITION TO AUTOMATED PHENOTYPIC METHODS

Phenotypic methods provide a comprehensive way to assess patterns of change in observable characteristics known as traits. However, not all traits are visible externally and can be complex to measure using traditional geometric morphometric approaches. Furthermore, the ability to extract and compare 3D phenotypic data across many individuals remains problematic due to the time-consuming nature of accurate manual processing, which is exacerbated as trait complexity increases. There is, therefore, a need to implement new technologies that allow for the extraction and analysis of large amounts of data in repeatable frameworks. Computed tomography (CT) allows for non-destructive observations of internal structures of organisms and provides the basis for the application of automated segmentation techniques. Here, we use these techniques to generate phenotypic data for four closely related species of planktonic foraminifera and compare these results to those obtained from manually labelled data. The automated phenotypic data yield size and shape signatures that are statistically indistinguishable from their manually generated counterparts, suggesting that automated approaches offer an accurate and time-effective method for rapidly generating phenotypic data with increased repeatability and without the biases of manual processing. Further development of these techniques offers the potential to transition towards methods which are free from the restrictions of manual landmark positioning. As a result, these methods hold great promise for improving data processing and ultimately increasing our ability to understand and reconstruct phenotypic evolution.