Parsing prehistoric patterns: Prospects and limitations of a big radiocarbon dataset for understanding the impact of climate on Late Palaeolithic and Mesolithic populations in northwest Europe (16–7.5 ka calBP)

Abstract

Hunter-gatherer populations in northwest Europe were variably affected by Late Glacial and Early Holocene climate fluctuations and their effects on sea level and the environment. We investigate the impact of these fluctuations with a dates-as-data approach to a large radiocarbon dataset. Radiocarbon dates are used as a proxy for past human activity, the intensity, nature and archaeological visibility of which will indirectly influence date density. The significance of changes is explored using Kernel Density Estimates and model tested Summed Probability Distributions. Whereas previous studies have focused on smaller highly curated datasets to minimise research and preservation biases, our more inclusive approach maximises sample size, which is essential for these methods to reliably reflect underlying patterns. To deal with biases, we test subsets of the dataset that are potentially affected by differences in formation processes.The summed radiocarbon dataset follows the general fluctuations of climate conditions, showing increased activity in temperate periods and decreased activity during cold phases. Our results indicate significant periods of interest where the data deviates positively or negatively from our models. Notably we observe the impact of the Younger Dryas, Preboreal Oscillation and the 8.2 ka event on the density of hunter-gatherer activity. Additionally we see peaks in activity in our dataset during the Early and Late Boreal. Permutation testing of different regions in the research area shows these patterns are geographically differentiated.Our exploration of biasing factors indicates that we should be careful to interpret the abovementioned patterns, as different sampling processes and national policies may lie at the basis of several patterns. Furthermore, calibration artefacts may also cause issues at key parts of the timeline. Dates-as-data approaches require an understanding of the archaeology, the timing of external events, the impact of the calibration curve and how biases inherent to the dataset and research area may have influenced the formation of patterns in the result.