One of the fundamental characteristics of a species is the amount of genetic variation segregating in its population, which can impact several aspects of a species' biology, including the phenotypic variability it exhibits and its response to selection. Predicting how much variation we should see in a given species is relatively simple under neutral and equilibrium conditions; genetic diversity should scale positively and linearly as a function of census population size and a species' mutation rate. Indeed, many species appear to have levels of variation that correspond with ecological approximations of their census population sizes. That said, despite this being a simple prediction, many studies have found no correlation between aspects of species ecology and geography and their genetic diversity. And, those studies that do find a correlation all show a puzzling pattern: species exhibit a much narrower range of genetic diversity (i.e., a narrower range of effective population sizes) than one would expect given their range of census population sizes. We revisit this puzzle by investigating patterns of genetic diversity across ~80 species in a diverse group of Australian lizards. Applying a phylogenetically-informed approach to genomic data, we find that occcurrence data from museum databases is the best predictor of levels of genetic diversity. However, our proxy for census population size shows a thousand-fold difference across our sampled species, yet, our corresponding estimate for effective population size shows nearly two orders of magnitude less variation across the same species. We explore some of the factors that might help explain this discrepancy.