Recent methodological advances have led to a rapid expansion of evolutionary studies employing three-dimensional landmark-based geometric morphometrics (GM). GM methods generally enable researchers to capture and compare complex shape phenotypes, and to quantify their relationship to environmental gradients. However, some recent studies have shown that the common, inexpensive, and relatively rapid two-dimensional GM methods can distort important information and produce misleading results because they cannot capture variation in the depth (Z) dimension. We use micro-CT scanned threespine stickleback (Gasterosteus aculeatus Linnaeus, 1758) from six parapatric lake-stream populations on Vancouver Island, British Columbia, to test whether the loss of the depth dimension in 2D GM studies results in misleading interpretations of parallel evolution. Using joint locations described with 2D or 3D landmarks, we compare results from separate 2D and 3D shape spaces, from a combined 2D-3D shape space, and from estimates of biomechanical function. We show that, although shape is distorted enough in 2D projections to strongly influence the interpretation of morphological parallelism, estimates of biomechanical function are relatively robust to the loss of the Z dimension.

GM landmark coordinates were obtained from CT scans made using a Zeiss XRadia Versa 520, segmented in Dragonfly, and landmarked in Meshlab.

Diet data was collected from gut contents, and collections were made and sex recorded as described in Hanson et al. 2016. J Evol Biol, 29:47–57.

This dataset is for use with the accompanying code, which is also available on github at https://github.com/ghaines3/2Dv3D-parallel. Diet data is presented as proportion of gut contents per individual, so all rows sum to one. GM coordinates for individuals are all on a common scale.