Beyond simple identification : advanced statistical approaches
to archaeological lipid  analysis using PCAmix and Bayesian
mixing models

k 2025

Beyond simple identification : advanced statistical approaches to archaeological lipid analysis using PCAmix and Bayesian mixing models

TÓTH, Peter; Veronika BRYCHOVÁ; Miriam NÝVLTOVÁ FIŠÁKOVÁ a Johana MALÍŠKOVÁ

Základní údaje

Originální název

Beyond simple identification : advanced statistical approaches to archaeological lipid analysis using PCAmix and Bayesian mixing models

Autoři

TÓTH, Peter ; Veronika BRYCHOVÁ; Miriam NÝVLTOVÁ FIŠÁKOVÁ a Johana MALÍŠKOVÁ

Vydání

Archaeometric Talks, Brno, 10.-12.11.2025, 2025

Další údaje

Jazyk

angličtina

Typ výsledku

Prezentace na konferencích

Obor

60102 Archaeology

Stát vydavatele

Česká republika

Utajení

není předmětem státního či obchodního tajemství

Odkazy

URL

Označené pro přenos do RIV

Organizační jednotka

Filozofická fakulta

Klíčová slova anglicky

Neolithic; Lipids; statistics; vessel use

Štítky

rivok

Příznaky

Mezinárodní význam

Změněno: 28. 1. 2026 11:37, Mgr. et Mgr. Stanislav Hasil, Ph.D.

Anotace

V originále

Organic residue analysis has become a cornerstone of archaeological research, providing direct evidence of past food processing and vessel use. However, traditional approaches to lipid analysis often rely on qualitative identification methods that can struggle with ambiguous cases and fail to quantify the relative contributions of different food sources. This presentation introduces advanced statistical methodologies that move beyond simple identification to provide quantitative insights into prehistoric culinary practices. We demonstrate the application of PCAmix and Bayesian mixing models using lipid data from Linear Pottery Culture (LBK) pottery vessels excavated at Těšetice-Kyjovice, Czech Republic. PCAmix, a principal component analysis method designed for mixed-type data, enables comprehensive exploration of complex lipid datasets by simultaneously analysing fatty acid concentrations, isotopic compositions, and vessel characteristics. This approach reveals underlying patterns in food processing practices that traditional methods might overlook. Bayesian mixing models, implemented through FRUITS software, provide probabilistic quantification of food source contributions to vessel contents. By incorporating isotopic data from both C16:0 and C18:0 fatty acids, along with appropriate uncertainty estimates, these models can differentiate between ruminant and non-ruminant fat sources with statistical confidence. Crucially, we establish isotopic distinctness criteria to ensure model validity and prevent conflation of similar sources. Our analysis of 62 pottery samples from four LBK longhouses of varying sizes reveals significant household-level differences in food processing strategies. Mid-size houses show specialization in ruminant fat processing, while larger households demonstrate more diverse practices incorporating both ruminant and non-ruminant products. These patterns suggest that vessel use and dietary practices were linked to household social complexity, challenging assumptions about uniform LBK subsistence strategies. The methodological framework presented here offers several advantages over traditional approaches: (1) objective quantification of source contributions, (2) proper treatment of analytical uncertainties, (3) statistical validation of interpretations, and (4) integration of multiple data types. These techniques are broadly applicable to organic residue studies and represent an advancement in archaeological lipid analysis, enabling more nuanced interpretations of past human behaviour and social organization.

Návaznosti

CZ.02.01.01/00/22_008/0004593, interní kód MU

Název: Připraveni na budoucnost: porozumění dlouhodobé odolnosti lidské kultury (RES-HUM)

Investor: Ministerstvo školství, mládeže a tělovýchovy ČR, Připraveni na budoucnost: porozumění dlouhodobé odolnosti lidské kultury (RES-HUM), Priorita 1 - Výzkum a vývoj

EH22_008/0004593, projekt VaV

Název: Připraveni na budoucnost: porozumění dlouhodobé odolnosti lidské kultury (RES-HUM)

Citovat

TÓTH, Peter; Veronika BRYCHOVÁ; Miriam NÝVLTOVÁ FIŠÁKOVÁ a Johana MALÍŠKOVÁ. Beyond simple identification : advanced statistical approaches to archaeological lipid analysis using PCAmix and Bayesian mixing models. In Archaeometric Talks, Brno, 10.-12.11.2025. 2025.

@proceedings{2535520,
   author = {Tóth, Peter and Brychová, Veronika and Nývltová Fišáková, Miriam and Malíšková, Johana},
   booktitle = {Archaeometric Talks, Brno, 10.-12.11.2025},
   keywords = {Neolithic; Lipids; statistics; vessel use},
   language = {eng},
   title = {Beyond simple identification : advanced statistical approaches to archaeological lipid analysis using PCAmix and Bayesian mixing models},
   url = {https://www.arup.cas.cz/akce/konference-archaeometric-talks/},
   year = {2025}
}

TY  - CONF
ID  - 2535520
AU  - Tóth, Peter - Brychová, Veronika - Nývltová Fišáková, Miriam - Malíšková, Johana
PY  - 2025
TI  - Beyond simple identification : advanced statistical approaches to archaeological lipid analysis using PCAmix and Bayesian mixing models
KW  - Neolithic
KW  - Lipids
KW  - statistics
KW  - vessel use
UR  - https://www.arup.cas.cz/akce/konference-archaeometric-talks/
N2  - Organic residue analysis has become a cornerstone of archaeological research, providing direct evidence of past food processing and vessel use. However, traditional approaches to lipid analysis often rely on qualitative identification methods that can struggle with ambiguous cases and fail to quantify the relative contributions of different food sources. This presentation introduces advanced statistical methodologies that move beyond simple identification to provide quantitative insights into prehistoric culinary practices. We demonstrate the application of PCAmix and Bayesian mixing models using lipid data from Linear Pottery Culture (LBK) pottery vessels excavated at Těšetice-Kyjovice, Czech Republic. PCAmix, a principal component analysis method designed for mixed-type data, enables comprehensive exploration of complex lipid datasets by simultaneously analysing fatty acid concentrations, isotopic compositions, and vessel characteristics. This approach reveals underlying patterns in food processing practices that traditional methods might overlook. Bayesian mixing models, implemented through FRUITS software, provide probabilistic quantification of food source contributions to vessel contents. By incorporating isotopic data from both C16:0 and C18:0 fatty acids, along with appropriate uncertainty estimates, these models can differentiate between ruminant and non-ruminant fat sources with statistical confidence. Crucially, we establish isotopic distinctness criteria to ensure model validity and prevent conflation of similar sources. Our analysis of 62 pottery samples from four LBK longhouses of varying sizes reveals significant household-level differences in food processing strategies. Mid-size houses show specialization in ruminant fat processing, while larger households demonstrate more diverse practices incorporating both ruminant and non-ruminant products. These patterns suggest that vessel use and dietary practices were linked to household social complexity, challenging assumptions about uniform LBK subsistence strategies. The methodological framework presented here offers several advantages over traditional approaches: (1) objective quantification of source contributions, (2) proper treatment of analytical uncertainties, (3) statistical validation of interpretations, and (4) integration of multiple data types. These techniques are broadly applicable to organic residue studies and represent an advancement in archaeological lipid analysis, enabling more nuanced interpretations of past human behaviour and social organization.
ER  -

Přehled o publikaci