Comparisons of extant and extinct biodiversity are often dependent on objective morphology-based identifications of fossils and assume a well-established and comparable taxonomy for both fossil and modern taxa. However, since many modern (cryptic) species are delimitated mainly via external morphology and/or molecular data, it is often unclear to what degree fossilized (osteological) remains allow classification to a similar level. When intraspecific morphological variation in extant taxa is poorly known, the definition of extinct species as well as the referral of fossils to extant species can be heavily biased, particularly if fossils are represented by incomplete isolated skeletal elements. This problem is especially pronounced in squamates (lizards and snakes) owing to a lack of osteological comparative knowledge for many lower taxonomic groups, concomitant with a recent increase of molecular studies revealing great cryptic diversity. Here, we apply a quantitative approach using 3D geometric morphometrics on 238 individuals of 14 genera of extant Australian and Papua New Guinean agamid lizards to test the value of two isolated skull bones (frontals and maxillae) for inferring taxonomic and ecological affinities. We further test for the consistence of intra- and interspecific morphological variability of these elements as a proxy for extinct taxonomic richness. We show that both bones are diagnostic at the generic level, and both can infer microhabitat and are of palaeoecological utility. However, species-level diversity is likely underestimated by both elements, with ~40% of species pairs showing no significant differences in shape. Mean intraspecific morphological variability is largely consistent across species and bones and thus a useful proxy for extinct species diversity. Reducing sample size and landmark completeness to approximate fossil specimens led to decreased classification accuracy and increased variance of morphological disparity, raising further doubts on the transferability of modern species borders to the fossil record of agamids. Our results highlight the need to establish appropriate levels of morphology-based taxonomic or ecological groupings prior to comparing extant and extinct biodiversity.

The evolution of limb reduction in squamates is a classic example of convergence, but the internal morphological patterns associated with it are unexplored. To provide insights into the biomechanical and developmental consequences of transitions to limb reduction, we use geometric morphometrics to examine the morphology of pectoral and pelvic girdles in 90 species of limb-reduced skinks and their fully limbed relatives. Clavicle shapes converge towards an acute anterior bend when forelimbs are lost but hindlimbs are retained — a morphology typical of sand swimmers. This may either indicate functional adaptations to locomotion in fine substrates, or a developmental consequence of complete limb loss. The shape of limb-bearing elements of both girdles (coracoid and pelvis) instead closely mirrors limb reduction, becoming more simplified as undulation replaces limbed locomotion. Integration between girdles decreases in taxa lacking elements of the forelimbs but not hindlimbs, indicating differential selection on each girdle in response to distinct locomotory strategies. However, this pattern becomes less clear when considering phylogenetic history, perhaps because it is limited to one specific clade (Lerista). We show how the functional demands of locomotion, as well as developmental mechanisms, can induce changes at different levels of organismal organisation, including both external and internal structures.