Laser-Induced breakdown spectroscopy (LIBS) for the analyses of faunal bones: Assembling of individuals and elemental quantification

Among the innumerable remains that appear in the archaeological sites, the faunal fossil remains are one of the most numerous. Different specialities study these fossil remains of fauna from different perspectives. Among them, the quantification of the remains is some of the most important information, since they allow us to estimate what species there are and what representativeness they can have. The problem is that the archaeological sites usually consist of several bones from different species and individuals. The minimum number of individuals (MNI) is usually used to estimate the representativeness of the species (e.g. there are 4 individuals of deer and 3 individuals of horses). The problem with this technique is that it does not allow to relate the remains that appear with each individual in a concrete way. Just the DNA analysis is necessary to reassemble the remains. However, the high cost and difficulty of implementation of DNA analysis do not allow performing a genetic study. In this paper, the combination of Laser-Induced Breakdown Spectroscopy (LIBS) and Neural Networks (NN) has been successfully implemented to discriminate and reassemble wild deer bones from different individuals. This is a step further from human bones classification and one step behind the archaeological bones classification. The elemental composition of bones provides enough information to get the correct discrimination of different individuals. The spectral correlation has exceeded 93 %. There have been no false positives and negatives. Also, the quantification of the different inorganic elements present in all bone samples has been achieved by the Calibration-Free LIBS (CF-LIBS) approach. Both approaches are different, NN is more qualitative and CF-LIBS, quantitative. Usually, chemometric methods are employed for the discrimination of samples, but it is an approach that requires a lot of LIBS spectra in order to perform the training, validation and test. In the case of CF-LIBS, only one spectrum with a certain degree of accumulation is necessary to obtain the elemental concentration. Thus, both approaches have been explored in this work. This fast analysis could allow archaeologist to perform a correct classification of individuals without performing several DNA ' s analysis. This would save time and resources.