Evaluation of Shape Complexity Metrics
Our first task was to assess what component of shape was driving increased complexity scores across the dataset. We define ‘complexity’ as the difference between the 4th and 20th harmonic model. On average across all taxa, the 4th harmonic model describes 87.9% of the complexity of the original femur shape and 85.5% in the tibia, while the 20th harmonic model describes around 99.7% of the complexity of the original segment shape both segments. Thus, gross complexity comprises approximately 14% of the total shape complexity.
On a qualitative level, the main visual difference between the 4th and the 20th harmonic outlines appears to be the presence of large spines in the 20thharmonic outline. This would suggest that gross complexity is heavily influenced by these ‘major’ spines. Indeed, initial inspection of boxplots suggests that species that are characterised by larger spines relative to the central pedipalp shaft have high gross complexity scores (see Figure 2). Observationally, the total number of large spines also appears to influence gross complexity. This is best illustrated in the tibial segment, in which species with relatively long but few large spines, such as H.longicornis and P.whitei , have lower gross complexity scores than other species with a larger number of ‘major’ spines. The tibia segment was found to have a statically higher complexity (Wilcoxon, p = <0.001, v = 2849) than the femur.