Siligîtes: a QGIS extension to investigate mobility networks in archaeology

For about fifteen years, french Collective Research Projects have significantly advanced research on the origin and processing of silicites (commonly referred to as ‘flint’) during Prehistory. The method of inventory and characterization of these lithic resources, described from a dynamic perspective using the concept of the evolutionary chain, is now applied during systematic prospecting as well as in the inventory of reference collections known as ‘lithothèques’. All this data is freely available through a dedicated webmapping platform: www.cartosilex.fr. The wealth of available data acquired in this context allows us today to fundamentally question the geography of Paleolithic ‘cultures’ in Western Europe from a landscape archaeology perspective. Among this research, studies related to lithic transfer networks play a central role in better understanding the spatial occupation by prehistoric populations. Until recently, the preparatory data required to conduct these studies, as well as the analysis of the network’s components, represented very time-consuming and repetitive stages. Moreover, the possibilities offered by Least Cost Path (LCP) analysis were underused, despite it becoming an essential tool for such studies by enabling the integration of topographical constraints into models of prehistoric population movements. This underuse is due to the complexity and low visibility of some of these tools, even though they are available in several GIS software packages (such as QGIS). To simplify the analysis process, we have developed a QGIS extension and propose appropriate modules and tools for prehistoric archaeology to: 1) assist in the acquisition and preparation of data for the creation of transfer networks; 2) facilitate the use of various LCP analyses by guiding users through the specific choices available; and 3) offer tools to simplify network analysis. For this purpose, we utilized the Python language, along with the tools provided by QGIS and PyQt, complemented by R scripts to access functions in the Movecost library. This library enables LCP-based analyses and offers a wide range of models, along with detailed explanations to avoid errors during use. Three modules were developed to perform the analysis steps within the QGIS extension, named Siligîtes. The first module, Silextracteur, allows users to create a vector layer representing specific parts of geological formations from a specific WebFeatureService (WFS). This tool reduces the data preparation time for archaeologists by a factor of six. The second module, SilLCP, focuses on preparing the previous data to create networks, particularly those based on LCP. It enables the use of the resources provided by the R library Movecost, including tools for creating LCP networks, isolines, Least Cost Corridor analyses, and more. A wide variety of LCP models based on time and energy costs are also included. The third module, Silanalyses, is designed to assist users in analyzing the networks they have created, whether by simplifying the network to include only links with a specific distance, time, or energy cost; studying the centrality of nodes (degree and betweenness); or determining the shortest path between two nodes (with or without intermediate steps). To test these tools, we used the example of the ‘Magdalénien inférieur à Lamelles à Dos Dextre Marginal (LDDM),’ a prehistoric ‘culture’ dated between 21.5 and 20.5 ka calBP. This ‘culture’ includes 13 prehistoric sites, including Lascaux, and features lithic objects that can be linked to nearly fifty geological sources. The area covered by these locations include a large part of southwestern part of France, from the Loire to the Pyrenees, and from the Atlantic Ocean to the eastern Massif Central. Using the tools developed with different datasets, we compared the results obtained by studying networks, whether using straight-line connections between locations or LCP-based networks. The two types of networks are quite similar in terms of the surfaces covered, with a margin of error of five to ten kilometers. Their nodes also display a similar level of connectivity, particularly in terms of betweenness. However, LCP-based networks offer new hypotheses, especially in topographically contrasting landscapes. This paper focuses on the efficiency of the GIS tools compared to manual execution of the various stages of data treatment, as well as exploring the limitations, whether related to the tools themselves or more globally in the application of LCP to prehistoric archaeology. Our various test allowed us to highlight these improvements and limits in terms of analysis provided by our QGIS extension, while demonstrating the advantages the different tools offer to researchers working on Paleolithic silicates in France, or on any other subject and period of archaeological study. In addition to providing easy access to various tools, including those for creation and analysis using LCP, the QGIS extension facilitates reproducibility of analyses and can help those interested in their future research.