The spread of agriculture during the Neolithic period marked one of the most significant transitions in human history, particularly across regions rich with diverse climates and landscapes like the Western Mediterranean. A new study utilizes machine learning tools to unravel the story of how early farmers migrated and adapted their agricultural practices from approximately 5900 to 2300 cal. BC. Through this innovative approach, researchers have compiled substantial datasets, incorporating archaeological, paleoclimatic, and eco-environmental variables, shedding light on human adaptation to shifting climate conditions.
This research, part of the AgriChange project, aims to quantify the interplay between climate fluctuations and agricultural diversification, particularly focusing on how ecological constraints shaped early Neolithic farmers' choices. The Western Mediterranean region, which has seen varied climatic conditions, provides rich data for investigating how agricultural methods evolved as communities faced environmental changes.
By leveraging machine learning algorithms such as Random Forest (RF) and Maximum Entropy (MaxEnt), the study analyzed 3416 geo-referenced archaeological sites alongside 843 archaeobotanical records. The purpose was to not only assess where and when these farmers settled but also which crops thrived under the climatic constraints of the time. One of the notable findings suggests, "This study facilitates calculating the pace of spread of farming in the Western Mediterranean area, modeling and estimating the potential areas suitable for settlement location." This detailed analysis reveals how climatic conditions influenced the cultivation and movement of early agricultural communities.
The methodology involved the characterization of past climate variability using high-resolution datasets to provide insight on annual precipitation and temperature changes over time. The results indicated, for example, how summers have shifted to cooler and wetter patterns throughout the Neolithic, impacting the crop choices and settlement locations known to support farming activities.
Further, the study unveiled significant shifts on the maps generated, illustrating how the most suitable areas for Neolithic settlements transitioned from coastal to more inland regions as the climate evolved. During the earliest phases of farming, prime agricultural land was predominantly found along the Mediterranean shores, yet as conditions became more favorable for inhabitation, settlements migrated inland to the elevated Swiss Plateau and surrounding lakes. Such shifts reflect not only climatic adaptation but also indicate the social dynamics of Neolithic communities.
Researchers observed distinct variations among the crops cultivated as well. For example, naked wheat was largely preferred during the nascent stages of farming, clustering around higher temperatures and lower precipitation values. Conversely, changes around 5300 BC marked the decline of certain crops, including glume wheat, as communities adapted their practices to the changing climate, leading them to favor crops like emmer and einkorn, which thrived under wetter conditions. "Our results indicate how small-scale communities could have developed different adaptive strategies to face climatic-based limitations," states the research team, highlighting the resilience of these early farmers.
This research ends on the note of future investigations, aiming to incorporate more refined datasets and explore the complex relationships between early agricultural practices and climatic changes. The combination of machine learning techniques with rich archaeological datasets presents exciting opportunities for archaeologists seeking to elucidate the past, as well as offering insight relevant for contemporary agricultural sustainability challenges. The broader implication of such work extends beyond purely academic interests—it emphasizes the importance of adaptability and resilience, qualities necessary for thriving amid climatic uncertainty.