Abstract
Many fossil specimens exhibit deformations caused by taphonomic processes. Due to these deformations, even important specimens have to be excluded from morphometric analyses, impoverishing an already poor paleontological record. Techniques to retrodeform and virtually restore damaged (i.e. deformed) specimens are available, but these methods genenerally imply the use of a sparse set of bilateral landmarks, ignoring the fact that the distribution and amount of control points directly affects the result of the retrodeformation. We propose a method developed in the R environment and available in the R-package "Morpho" that, in addition to the landmark configurations, also allows using a set of semi-landmarks homogeneously distributed along curves and on surfaces. We evaluated the outcome of the retrodeformation, regarding the number of semi-landmarks used and its robustness against asymmetric noise, based on simulations using a virtually deformed gorilla cranium. Finally, we applied the method to a well-known Neanderthal cranium that exhibits signs of taphonomically induced asymmetry.