Digital range of motion analysis is sensitive to subjective steps in joint model construction.

Reconstructing locomotion and behaviour in extinct vertebrates requires a detailed understanding of joint mobility to constrain the range of potential limb orientations to more biologically plausible poses. Joint mobility is typically assessed using range of motion (ROM) analysis, which is increasingly implemented within digital workflows. Digital ROM analysis requires the positioning of bones in three-dimensional space into a starting pose that enables systematic sampling of rotational and translational configurations (e.g. reference pose). However, this initial process involves subjective steps, particularly in selecting articular surfaces and defining a joint centre of rotation. No thorough sensitivity analysis has yet been published. In this study, we have conducted systematic sensitivity analyses of a complete six degree-of-freedom automated ROM analysis workflow to evaluate how variation at each stage of the reference pose assembly, stemming from both input data and user decision, affects the determined viable poses in the ankle and tarsometatarsophalangeal III joints of Guinea fowl (Numida meleagris). Our results reveal that ROM analysis outputs are sensitive to variation in reference pose assembly, especially changes in articular surface selection and the primitive used to define the joint centre. This sensitivity may be unlikely to directly affect the overall conclusions of any given individual study, particularly when using maximum viable rotations to constrain biomechanical models. However, it potentially makes comparison between studies and taxa problematic, and we therefore advocate that future ROM studies should prioritise providing complete joint models as supplemental data to enable replicability.