A cheek tooth recently unearthed in the Lopar Sandstone unit, of late Eocene age, in the northern part of Rab Island, Croatia, is one of the very few Eocene mammalian remains found in the Adriatic area. Thorough comparison of this tooth with those of Old-World Palaeogene mammalian orders suggests that it is a M3 belonging to an embrithopod afrothere. The specimen is referred to as Palaeoamasia sp. This genus was formerly known only in Eocene deposits of Anatolia but with close relatives in Romania among Palaeoamasiinae. The geographical distribution of this subfamily perfectly matches the recently-named Balkanatolian landmass, which experienced in-situ evolution of endemic mammals prior to the Grande Coupure event that occurred around the Eocene–Oligocene transition. This last event is characterised by massive Asian immigration in Western Europe and the supposed extinction of many endemic Central and Western European mammals, including Palaeoamasiinae.
  

Micro-CT imaging is an increasingly popular method in paleontology giving access to internal structures with a high resolution and without destroying precious specimens. However, its potential for the study of hypoplasia defects has only recently been investigated. Here, we propose a preliminary study to test whether hypoplastic defects can be detected with micro-CT (μCT) scan and we assess the costs and benefits of using this method instead of naked eye. To do so, we studied 13 fossil rhinocerotid teeth bearing hypoplasia from Béon 1 (late early Miocene, Southwestern France) as positive control and 11 teeth of the amynodontid Cadurcotherium (Oligocene, Phosphorites du Quercy, Southwestern France), for which enamel was partly or totally obscured by cement. We showed that all macroscopically-spotted defects were retrieved on 3D reconstructions and selected virtual slices. We also detected additional defects using μCT scan compared to naked eye identification. The number of defects detected using μCT was greater in the Cadurcotherium dataset (paired-sample Wilcoxon test, p-value = 0.02724) but not for our control sample (paired-sample Wilcoxon test, p-value = 0.1171). Moreover, it allowed for measuring width and depth of the defects on virtual slices (sometimes linked to stress duration and severity, respectively), which we could not do macroscopically. As μCT imaging is both expensive and time consuming while not drastically improving the results, we recommend a moderate and thoughtful use of this method for hypoplasia investigations, restricted for instance to teeth for which enamel surface is obscured (presence of cement, uncomplete preparation, or unerupted germs).