Open palates with large interpterygoid vacuities are a diagnostic characteristic of temnospondyl amphibians, the most species-rich group of early tetrapods. Aside from their functional roles, several other aspects of such vacuities, such as their variation and spatial relationships relative to the orbits, have received only scarce attention. The present work examines patterns of shape and size changes in the orbits and vacuities of temnospondyls using a time-calibrated phylogeny of 69 temnospondyl taxa and 13 additional early tetrapod ‘outgroups' (colosteids, an embolomere, ‘microsaurs' and nectrideans). Orbit and vacuity outlines are quantified in a comparative framework using standard eigenshape analyses. In addition, we employ a series of ratios of linear measurements of both orbits and vacuities, and subject them to a phylogenetic principal component analysis in order to evaluate their proportional changes relative to the skull and to one another. Finally, we examine rates of evolutionary change and their associated shifts for shape and size for both structures, and assess the strength and significance of the correlations between these two variables using phylogenetic generalised least squares analyses. Although orbits and vacuities have fairly simple outlines, they both reveal complex models of proportional change across the temnospondyl phylogeny. These changes exhibit strong phylogenetic signal, that is, trait covariance among taxa is predicted by tree topology. We discuss the hypothesis that, early in tetrapod evolution, the functional role of the vacuities was related to the accommodation of the anterior jaw muscles. Only later in evolution did such vacuities serve to accommodate the eye muscles only.

The present study explores the macroevolutionary dynamics of shape changes in the humeri of all major grades and clades of early tetrapods and their fish-like forerunners. Coordinate point eigenshape analysis applied to humeral outlines in extensor view reveals that fish humeri are more disparate than those of most early tetrapod groups and significantly separate from the latter. Our findings indicate sustained changes in humeral shape in the deepest portions of the tetrapod stem group and certain portions of the crown. In the first half of sampled tetrapod history, subclades show larger than expected humeral disparity, suggesting rapid diffusion into morphospace. Later in tetrapod evolution, subclades occupy smaller and non-overlapping morphospace regions. This pattern may reflect in part increasing specialisations in later tetrapod lineages. Bayesian shifts in rates of evolutionary change are distributed discontinuously across the phylogeny, and most of them occur within rather than between major groups. Most shifts with the highest Bayesian posterior probabilities are observed in lepospondyls. Similarly, maximum likelihood analyses of shifts support marked rate accelerations in lepospondyls and in various subclades within that group. In other tetrapod groups, rates either tend to slow down or experience only small increases. Somewhat surprisingly, no shifts are concurrent with structural, functional, or ecological innovations in tetrapod evolution, including the origin of digits, the water–land transition and increasing terrestrialisation. Although counterintuitive, these results are consistent with a model of continual phenotypic innovation that, although decoupled from key evolutionary changes, is possibly triggered by niche segregation in divergent clades and grades of early tetrapods.