Ricerc@Sapienza

The study of the fossil record is fundamental to understand the evolution of morphological traits. Fossil remains often appear fragmented and/or deformed by taphonomic processes, which result in cracks, missing portions and deformation of the original morphology. In paleoanthropology, cranial remains are at the same time the most informative and sometimes the most badly deformed fossil portions, often presenting themselves broken, partially incomplete, and/or distorted. It is nowadays possible to overcome some of these issues, thanks to three-dimensional imaging techniques which allow to operate on digital models of the specimens. It is possible, in particular, to virtually restore damaged remains and recover or estimate the missing information by relying on the one that is present. In this communication, we present a new semi-automatic, landmark-based alignment software, the Digital Tool for Alignment (DTA), embedded in the R package ‘Arothron’. DTA uses the shape information contained in a reference sample to find the best alignment solution for the disarticulated portions of a specimen. We showed DTA performance on two different case-studies: a modern human skull, in which a casual disarticulation in two fragments has been simulated and a real case of a disarticulated human fossil specimen, the Amud 1 cranium (Homo neanderthalensis).The first case study consists of an artificially disarticulated model of a skull from a female modern human (“Bol-2548”), on which the DTA was applied using a comparative sample of 50 modern human specimens and a landmark configuration of 32 landmarks (17 on the first module and 15 on the second one). The DTA performance on “Bol-2548” was both calculated using the original complete skull and compared to 11 manual alignments. The second case-study consists on the application of DTA on the Amud 1 cranium, in which part of the facial skeleton is missing, with no recovered points of contact between the calvarium (Amud 1a) and the maxillary region (Amud1b).We acquired two landmark sets of 16 (Amud 1a) and 17 (Amud 1b) landmarks; in addition, we defined two semi-landmark sets of 104 and 42 points respectively on Amud 1a and Amud 1b. We used separately the Neanderthal specimens Shanidar 1 and La Ferrassie 1 as reference models. The result of the DTA was compared to the original restoration of Amud 1 by Suzuki. The application to the modern human case study returned a Procrustes distance between the Aligned Model (AM, built on “Bol-2546”) and the Starting Model (SM, unbroken “Bol-2548”) equal to 0.0046.The average Procrustes distance of the manual alignments group from the SM was twice that much, reaching 0.0110 on average (95% CI = 0.0079 - 0.0142).The range of the distances between the manual alignments and the SM was bracketed from 0.0032 to 0.0175. The mean Procrustes distances between the SM and the AMs performed on the comparative sample was 0.006. For what concerns the case study of Amud 1, the average distances between the surfaces of the SM of Amud 1 and the AM Shanidar 1 amounts to 3.44 or 5.20 mm, using only landmarks or both landmarks and semi-landmarks, respectively. Using La Ferrassie 1 as the reference model, the average distances are equal to 5.87 (landmarks) and 7.06 (landmark + semi-landmarks), respectively. The results of DTA performed on the artificial case-study (Homo sapiens) show that DTA works almost as well as expert-based reconstruction, which is fundamental in situations where the handling of fossil items is difficult, dangerous (because the specimen is delicate), or simply unavailable. The alignment via DTA performed on the Amud 1 cranium differs by only 3 millimetres from the original reconstruction made by Suzuki. This result confirms the usefulness of DTA as a technique for aligning broken fossil specimens.