A multi-method approach with machine learning to evaluating the distribution and intensity of prehistoric land use in Eastern Iberia

Abstract

The present study seeks to better understand the coupling of social and biophysical systems during the late Pleistocene and Holocene, a period characterized by changing interglacial conditions as well as human population expansion and intensified ecosystem management. The approach consists of a combination of patch-based archaeological survey methods, sediment column sampling for paleoenvironmental data, geospatial analysis, and machine learning for chronological unmixing, allowing the systematic evaluation of the distribution and intensity of prehistoric land use in the study area of eastern Mediterranean Iberia. Occupational and Land Use Intensity maps developed from continuous distributions of surface artifacts as well as a summed probability density curve developed from 14C dates indicate low but steady human presence in the study area during the Middle Paleolithic and Upper Paleolithic with a marked decrease of human presence across the Pleistocene/Holocene boundary. The Early (and Middle) Neolithic saw the most ubiquitous and intensive occupation of the study area followed by a significant decline during the Late Neolithic/Chalcolithic. The charcoal analysis also supports this pattern. Early Neolithic farming strategies may not have been damaging initially during the climatic regime of the Early Holocene but exacerbated the impacts of higher temperatures and summer droughts, with a loss of the most productive farmland, seen at the onset of the Late Neolithic. Similar boom-bust population trends have been documented throughout Europe during these same time spans and may indicate a recursive interaction or “coupling” between global and regional climate events and human land use strategies.