Evolutionary reversals, including re-evolution of lost structures, are commonly found in phylogenetic studies. However, we lack an understanding of how these reversals happen mechanistically. A snake-like body form has evolved many times in vertebrates, and occasionally, a quadrupedal form has re-evolved, including in Brachymeles lizards. We use body form and locomotion data for species ranging from snake-like to quadrupedal to address how a quadrupedal form could re-evolve. We show that large, quadrupedal species are faster at burying and surface locomotion than snake-like species, indicating a lack of expected performance trade-off. Species with limbs use them while burying, suggesting that limbs are useful for burying in wet, packed substrates. Paleoclimatological data suggest that Brachymeles originally evolved a snake-like form under a drier climate likely with soil in which it was easier to dig. The quadrupedal clade evolved as the climate became humid, where limbs and large size facilitated fossorial locomotion in packed soils.

The dataset was primarily collected in the field, from high-speed videos of animals moving on the surface and burying into a coarse and a fine substrate. Force readings were also recorded in the field, as were soil load-bearing capacities and soil moisture. Morphometrics were collected form the preserved specimens. Dataset processing consisted of taking species averages of individual-level data for comparative analyses.

The dataset contains several spreadsheets. It contains species means for 14 species of lizards (13 Brachymeles skinks, plus Lygosoma bowringii). These data are morphometric (body proportions and measurements), ecological (soil moisture and load-bearing capacity), surface locomotion (running or crawling on coarse and fine substrates), buring locomotion (in coarse and fine substrates), and burrowing force data. There is also a dataset of head length for 40 species of Brachymeles, individual morphometrics for the animals used in locomotion work (161 individuals distributed across the species), individual burrowing force data for 268 trails, individual-level surface locomotion data for 307 trials, and individual-level burying data for 297 trials. The individual-level data were used to calculate the species mean data.

Additionally, there is a text file containing the two phylogenies that we used to analyze the data. One phylogeny contains 40 species of Brachymeles and was used with the head length data for the same number of species. The other phylogeny contains the 13 species of Brachymeles, plus Lygosoma bowringii).

Finally, there is a text file containing R code used to conduct the analyses and construct the figures for the manuscript.

Everything needed to use the data is included in a single Excel file. There are several data spreadsheets, corresponding to the above description. Each spreadsheet also has a readme sheet included that explains each variable with a description and units of measurement.