Identifying Changepoints in Population Size through Radiocarbon Determinations

An 8-month project to develop, and test, novel statistical tools that will enable environmental, and archaeological, scientists to accurately identify changes in the past size of populations from samples dated via radiocarbon. Studying changes in patterns of occupation is key to understanding our responses to past environmental and societal changes, and to forecasting our resilience to potential future climate change. It is essential to provide a statistically-rigorous alternative to replace some of the incoherent approaches currently used.

A frequently-used archaeological approach to estimating changes in the size of a population is to consider how the number of samples relating to that population (e.g., human/animal bones, or other evidence of occupation) varies over time. Time periods for which there are large numbers of samples are expected to correspond to a larger population being present; periods when fewer samples are found suggest a smaller population size. Such an approach can be applied to individual sites, or by collating samples across multiple sites.

The reliability of such a "dates-as-data" approach is highly dependent upon our ability to estimate the calendar ages of the discoveries. When studying the last 55,000 years, the most common way to obtain the necessary dates is via radiocarbon. Unfortunately, the need to calibrate all radiocarbon determinations introduces considerable, and complex, uncertainties into the calendar ages of each of the samples. This uncertainty should be incorporated into later inference such as when using the density of the dates as a proxy for population size. However, it is well recognised that the techniques currently used by the radiocarbon community fail to do so fully. This limits the reliability of the inference they provide. New statistically-rigorous tools are required. This project will provide them.

We will develop an integrated Bayesian framework that is able to jointly calibrate the radiocarbon determinations belonging to a set of samples, and to identify if there are statistically-significant changes in the rate at which the samples arise. We will demonstrate our approach on a range of geoscientifically-interesting questions including the expansion of humans into the Yukon and Alaska in the late Pleistocene and early Holocene. We will aim to investigate both the timings of such migrations in comparison with the climatic changes known to have occurred during this period, and to study the potential interactions between humans and other megafauna in the region.

In addition to developing and publishing the statistical framework, we will also work to ensure our work is of maximum benefit to the target radiocarbon user community. We will disseminate our research through provision of usable and easily-accessible software; and a user-guide providing illustrative cases studies and worked examples, that is appropriately written for the target archaeological and environmental science community.