Directly modelling population dynamics in the South American Arid Diagonal using ¹⁴C dates.

Area of Science:

• Archaeological Science
• Computational Archaeology
• Bayesian Statistics

Background:

• Radiocarbon (¹⁴C) datasets are frequently used to infer past human population sizes via summed probability distributions (SPDs).
• SPDs can be unreliable proxies for population dynamics, especially with small datasets, potentially leading to inaccurate demographic reconstructions.
• Directly modeling population dynamics and assessing model likelihood against ¹⁴C data offers a more robust inferential approach.

Observation:

• The R package ADMUR is introduced, employing a continuous piecewise linear (CPL) model for population change.
• ADMUR calculates model likelihood with ¹⁴C data, estimates credible intervals via Markov chain Monte Carlo, and includes goodness-of-fit and model comparison criteria (Schwarz Criterion).
• Simulations with toy data demonstrate ADMUR's ability to avoid spurious dynamics in small sample sizes and accurately recover population trends with increasing data.

Findings:

• Application to an improved ¹⁴C dataset from the South American Arid Diagonal reveals distinct Holocene population phases.
• Population trajectory estimates indicate an initial rapid growth (4.15% per generation), followed by a decline (0.05% per generation) between 10,821 and 7,055 years BP.
• Subsequent population growth occurred at a slower rate (0.58% per generation) until 2,500 years BP.

Implications:

• ADMUR offers a more reliable method for reconstructing prehistoric population histories than traditional SPD analyses.
• The findings highlight significant shifts in population dynamics within the South American Arid Diagonal during the Holocene.
• This approach facilitates more accurate cross-disciplinary studies in prehistoric demography.