SAS Best Oral Presentation at ICAS-EMME - Elissavet Ganiatsou

Application of machine learning on isotopic data from tooth micro sections for reconstructing weaning patterns and physiological stress 

Elissavet Ganiantsoua, Angeliki Georgiadoua, Angelos Soullessa, Asterios Adonisa, Tania Protopsaltib, Stravoula Tzevrenib, Krino Konstantinidoub, Stella Vasileiadoub, Frank Siegmundc, Christina Papageorgopouloua

Laboratory of Physical Anthropology, Department of History and Ethnology, Democritus University of Thrace, 69100, Komotini, Greece.

Ephorate of Antiquities of Thessaloniki City, Ministry of Culture and Sports, 54003, Thessaloniki, Greece 

c Abt. Ur- und Frühgeschichtliche Archäologie, Department of History, University of Münster, Germany.

*corresponding author:

Breastfeeding and weaning are linked to large-scale sociocultural phenomena and have extensive effects on human growth and development. In recent years, bioarchaeologists have taken an interest in investigating their manifestation in the past, as a comparative lens to our understanding of childhood concepts, familial relations, and demographic strategies (Lewis, 2007). The recent development of measuring stable isotope ratios (δ15N and δ13C) in collagen from tooth microsections enables the reconstruction of individual diets and has been widely used for the study of breastfeeding, weaning and, physiological stress in archaeological datasets revealing its temporal and spatial variation (Bourbou et al., 2013; Dupras et al., 2001; Dupras & Tocheri, 2007; Keenleyside et al., 2009; Prowse et al., 2008; Redfern et al., 2018; Beaumont et al., 2013, 2015; Beaumont & Montgomery, 2016; Craig-Atkins et al., 2018; King et al., 2018). However, there are only few studies focusing on diachronic trends in breastfeeding and weaning patterns (Kwok, 2015).


The scope of this study is to elucidate the dynamic variability of breastfeeding and weaning of infants originating from the same site. As a case study, we analyzed individuals from ancient Thessaloniki (4th c. BC-16th c. AD), a multicultural metropolis that underwent a series of transformations in terms of human and material culture. We applied the incremental dentine analysis, measuring the δ15N and δ13in collagen from the first permanent molars of 45 adult individuals to reconstruct the infant diet. To this dataset, we added 20 previously published individuals from the same site (Ganiatsou et al., 2022) increasing the number of individuals to 65. We estimated the weaning duration with WEAN, an automated application that provides robust weaning age estimations based on the δ15N measurements and refined age-at-increment assignment. To discern clusters of different isotopic patterns and potential signals of stress, we examined the covariance of stable nitrogen and carbon ratios before and after the age of two by applying the k-means machine learning algorithm.


Of the 45 molars, 31 produced adequate collagen for analysis corresponding to 589 dentine sections. We selected 381 samples for measurement representing the diet from birth to the age of ca. seven years old. WEAN estimated the weaning age of 45 individuals, most of them dating to the Roman (n=26) and Post-Byzantine periods (n=14), while the ones from the Hellenistic and the Byzantine periods are fewer (three respectively). Although the archaeological periods are not equally represented, the mean weaning age of each period is similar (between 2-2.5 years old) with the exception of the three Byzantine individuals (1.6 years old) (Figure 1). 

Figure 1. Modified z-scores of the estimated weaning ages, color-coded based on chronological period, showing the relationship between the weaning age of each individual and the median weaning age of the population (2.1 years). 
(Hellenistic: 4th c. BC-1st c. BC, Roman:1st c. BC-4th c. AD, Byzantine: 4th c. AD -15th c. AD, individuals dating to the  post-Byzantine period between 15th c.-16th c. AD).


Five individuals, however, were breastfed for an even shorter period or were never breastfed at all. The weaning diet was comprised mostly of animal protein and Cplants, while the consumption of small fish and/or C4 plants intensified from the Roman period onwards. Potential signals of physiological stress as attested by isotopic evidence were identified in the values of nine individuals, which may be linked to maladaptive breastfeeding patterns. We could not identify statistically significant differences in weaning ages between sex groups or chronological periods, but a qualitative observation based on the individual isotope profiles shows variability in weaning duration and diet (Figure 2).  Overall, in the site of Thessaloniki weaning patterns varied; this has been brought forward by the combination of novel laboratory and computational methods highlighting the complexity of decisions on breastfeeding and weaning in ancient societies.


Figure 2. Violin charts showing the distribution of nitrogen ratios (A) and carbon ratios (B) between males, females and undetermined individuals from Thessaloniki, northern Greece per chronological period (Hellenistic: 4th c. BC-1st c. BC, Roman :1st c. BC-4th c. AD, Byzantine: 4th c. AD -15th c. AD, individuals dating to the Post-Byzantine period between 15th c.-16th c. AD).


The authors would like to thank the Ephorate of Antiquities of the city of Thessaloniki for entrusting the human skeletal material, Efrossini Vika for valuable help in stable isotope analysis, Panagiota Xanthopoulou, Christina Kakasa, Maria Christodoulou, Eugenia Mantatzi from the Laboratory of Physical Anthropology DUTh for providing anthropological data. The authors extend their gratitude to Natalie Wing for helpful advice in proof-editing and Steven Brookes from Iso-Analytical Ltd for analytical measurements.


This research has been cofinanced by the European Regional Development Fund of the European Union and Greek national funds through the Operational Program Competitiveness, Entrepreneurship and Innovation, under the call RESEARCH – CREATE – INNOVATE (project title: ECHOES-Development of a methodology for the digital reconstruction of ancient human biographies through the study of archaeo-anthropological material, project code:T2EDK-00152). 



Beaumont, J., Geber, J., Powers, N., Wilson, A., Lee-Thorp, J., & Montgomery, J. (2013). Victims and survivors: Stable isotopes used to identify migrants from the Great Irish Famine to 19th century London. American Journal of Physical Anthropology150(1), 87–98.

Beaumont, J., & Montgomery, J. (2016). The Great Irish Famine: Identifying Starvation in the Tissues of Victims Using Stable Isotope Analysis of Bone and Incremental Dentine Collagen. PLOS ONE11(8), e0160065.

Beaumont, J., Montgomery, J., Buckberry, J., & Jay, M. (2015). Infant mortality and isotopic complexity: New approaches to stress, maternal health, and weaning. American Journal of Physical Anthropology157(3), 441–457.

Bourbou, C., Fuller, B. T., Garvie-Lok, S. J., & Richards, M. P. (2013). Nursing mothers and feeding bottles: reconstructing breastfeeding and weaning patterns in Greek Byzantine populations (6th–15th centuries AD) using carbon and nitrogen stable isotope ratios. Journal of Archaeological Science40(11), 3903–3913.

Craig-Atkins, E., Towers, J., & Beaumont, J. (2018). The role of infant life histories in the construction of identities in death: An incremental isotope study of dietary and physiological status among children afforded differential burial. American Journal of Physical Anthropology167(3), 644–655.

Dupras, T. L., Schwarcz, H. P., & Fairgrieve, S. I. (2001). Infant feeding and weaning practices in Roman Egypt. American Journal of Physical Anthropology115(3), 204–212.

Dupras, T. L., & Tocheri, M. W. (2007). Reconstructing infant weaning histories at Roman period Kellis, Egypt using stable isotope analysis of dentition. American Journal of Physical Anthropology134(1), 63–74. 

Keenleyside, A., Schwarcz, H., Stirling, L., & ben Lazreg, N. (2009). Stable isotopic evidence for diet in a Roman and Late Roman population from Leptiminus, Tunisia. Journal of Archaeological Science36(1), 51–63.

King, C. L., Halcrow, S. E., Millard, A. R., Gröcke, D. R., Standen, V. G., Portilla, M., & Arriaza, B. T. (2018). Let’s talk about stress, baby! Infant-feeding practices and stress in the ancient Atacama desert, Northern Chile. American Journal of Physical Anthropology166(1), 139–155.

Lewis, M. E. (2007). The bioarchaeology of children: perspectives from biological and forensic anthropology (Vol. 50). Cambridge University Press.

Prowse, T. L., Saunders, S. R., Schwarcz, H. P., Garnsey, P., Macchiarelli, R., & Bondioli, L. (2008). Isotopic and dental evidence for infant and young child feeding practices in an imperial Roman skeletal sample. American Journal of Physical Anthropology137(3), 294–308.

Redfern, R., Gowland, R., Millard, A., Powell, L., & Gröcke, D. (2018). ‘From the mouths of babes’: A subadult dietary stable isotope perspective on Roman London (Londinium). Journal of Archaeological Science: Reports19, 1030–1040.