Holocene climate variability in the Mediterranean Basin is often cited as a potential driver of societal change, but the mechanisms of this putative influence are generally little explored. In this paper we integrate two tools - agro-ecosystem modeling of potential agricultural yields and spatial analysis of archaeological settlement pattern data - in order to examine the human consequences of past climatic changes. Focusing on a case study in Provence (France), we adapt an agro-ecosystem model to the modeling of potential agricultural productivity during the Holocene. Calibrating this model for past crops and agricultural practices and using a downscaling approach to produce high spatiotemporal resolution paleoclimate data from a Mediterranean Holocene climate reconstruction, we estimate realistic potential agricultural yields under past climatic conditions. These serve as the basis for spatial analysis of archaeological settlement patterns, in which we examine the changing relationship over time between agricultural productivity and settlement location. Using potential agricultural productivity (PAgP) as a measure of the human consequences of climate changes, we focus on the relative magnitudes of 1) climate-driven shifts in PAgP and 2) the potential increases in productivity realizable through agricultural intensification. Together these offer a means of assessing the scale and mechanisms of the vulnerability and resilience of Holocene inhabitants of Provence to climate change. Our results suggest that settlement patterns were closely tied to PAgP throughout most of the Holocene, with the notable exception of the period from the Middle Bronze Age through the Early Iron Age. This pattern does not appear to be linked to any climatically-driven changes in PAgP, and conversely the most salient changes in PAgP during the Holocene cannot be clearly linked to any changes in settlement pattern. We argue that this constitutes evidence that vulnerability and resilience to climate change are strongly dependent on societal variables.
|Date made available||2018|