Since prehistoric times, the island of Sardinia—in the western Mediterranean—has played a leading role in the dynamics of human population and mobility, in the circulation of raw materials and artefacts, idioms and customs, of technologies and ideas that have enriched the biological, linguistic and cultural heritage of local groups. For the Phoenician and Punic periods (from the 9th to the 3rd centuries BCE), the ancient site of Nora—in southern Sardinia—represents an emblematic case in the study of migratory phenomena that occurred on the Island from the Iron Age until the Roman conquest. Despite the importance of exploring (and characterising) such movements from a wider bio-cultural perspective, the application of bio-geochemical tools for geographical provenance to the ancient skeletal populations of Sardinia is yet scarce. The present work is the first step towards filling this gap with the development of the first isoscape of southern Sardinia using new bioavailable Sr isotope data and a machine-learning approach. From a geolithological point of view, Sardinia is rather heterogeneous and requires detailed studies to correctly assess the distribution of the isotopic signature of bioavailable Sr. The random forest model employed here to construct the Sr isoscape uses several external environmental and geological variables. The most important predictors are related to age and bedrock type, with additional input from local soil properties. A 10-fold cross-validation gives a mean square error of 0.0008 and an R-squared of 0.81, so the model correctly predicts the 87Sr/86Sr ratio of unknown areas. By using a Bayesian provenance assignment workflow, we tested the isoscape here produced to determine the geographic origin and the mobility of archaeological and modern fauna collected from the Phoenician-Punic site of Nora and the surrounding Pula Plain. Our results indicate that archaeological sheep and goats (87Sr/86Sr < 0.7090) are compatible with areas close to Nora and Pula Plain, in agreement with archaeological evidence of pastoralism in those areas. Modern wild and domesticated fauna (87Sr/86Sr > 0.7090) show compatibility with several natural and anthropogenic locations in southern Sardinia, as expected based on modern species distribution data. Finally, we discuss the large Sr isotopic variability of the Nora baseline, where human mobility studies of human cremated and inhumed individuals are currently underway.

Machine learning-based Sr isoscape of southern Sardinia: A tool for bio-geographic studies at the Phoenician-Punic site of Nora / Gigante, Melania; Mazzariol, Alessandro; Bonetto, Jacopo; Armaroli, Elena; Cipriani, Anna; Lugli, Federico. - In: PLOS ONE. - ISSN 1932-6203. - 18:7(2023), pp. 1-19. [10.1371/journal.pone.0287787]

Machine learning-based Sr isoscape of southern Sardinia: A tool for bio-geographic studies at the Phoenician-Punic site of Nora

Armaroli, Elena;Cipriani, Anna;Lugli, Federico
2023

Abstract

Since prehistoric times, the island of Sardinia—in the western Mediterranean—has played a leading role in the dynamics of human population and mobility, in the circulation of raw materials and artefacts, idioms and customs, of technologies and ideas that have enriched the biological, linguistic and cultural heritage of local groups. For the Phoenician and Punic periods (from the 9th to the 3rd centuries BCE), the ancient site of Nora—in southern Sardinia—represents an emblematic case in the study of migratory phenomena that occurred on the Island from the Iron Age until the Roman conquest. Despite the importance of exploring (and characterising) such movements from a wider bio-cultural perspective, the application of bio-geochemical tools for geographical provenance to the ancient skeletal populations of Sardinia is yet scarce. The present work is the first step towards filling this gap with the development of the first isoscape of southern Sardinia using new bioavailable Sr isotope data and a machine-learning approach. From a geolithological point of view, Sardinia is rather heterogeneous and requires detailed studies to correctly assess the distribution of the isotopic signature of bioavailable Sr. The random forest model employed here to construct the Sr isoscape uses several external environmental and geological variables. The most important predictors are related to age and bedrock type, with additional input from local soil properties. A 10-fold cross-validation gives a mean square error of 0.0008 and an R-squared of 0.81, so the model correctly predicts the 87Sr/86Sr ratio of unknown areas. By using a Bayesian provenance assignment workflow, we tested the isoscape here produced to determine the geographic origin and the mobility of archaeological and modern fauna collected from the Phoenician-Punic site of Nora and the surrounding Pula Plain. Our results indicate that archaeological sheep and goats (87Sr/86Sr < 0.7090) are compatible with areas close to Nora and Pula Plain, in agreement with archaeological evidence of pastoralism in those areas. Modern wild and domesticated fauna (87Sr/86Sr > 0.7090) show compatibility with several natural and anthropogenic locations in southern Sardinia, as expected based on modern species distribution data. Finally, we discuss the large Sr isotopic variability of the Nora baseline, where human mobility studies of human cremated and inhumed individuals are currently underway.
2023
19-lug-2023
18
7
1
19
Machine learning-based Sr isoscape of southern Sardinia: A tool for bio-geographic studies at the Phoenician-Punic site of Nora / Gigante, Melania; Mazzariol, Alessandro; Bonetto, Jacopo; Armaroli, Elena; Cipriani, Anna; Lugli, Federico. - In: PLOS ONE. - ISSN 1932-6203. - 18:7(2023), pp. 1-19. [10.1371/journal.pone.0287787]
Gigante, Melania; Mazzariol, Alessandro; Bonetto, Jacopo; Armaroli, Elena; Cipriani, Anna; Lugli, Federico
File in questo prodotto:
File Dimensione Formato  
Gigante et al. 2023.pdf

Open access

Tipologia: Versione pubblicata dall'editore
Dimensione 2.87 MB
Formato Adobe PDF
2.87 MB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

Licenza Creative Commons
I metadati presenti in IRIS UNIMORE sono rilasciati con licenza Creative Commons CC0 1.0 Universal, mentre i file delle pubblicazioni sono rilasciati con licenza Attribuzione 4.0 Internazionale (CC BY 4.0), salvo diversa indicazione.
In caso di violazione di copyright, contattare Supporto Iris

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11380/1311886
Citazioni
  • ???jsp.display-item.citation.pmc??? 0
  • Scopus 0
  • ???jsp.display-item.citation.isi??? 0
social impact