Multivariate analysis reveals spatial variability of soil geochemical signals in the area of a medieval manorial farm

•Relicts of medieval landscape were identified using LiDAR survey.•Compositional analysis of geochemical data enabled to determine anthropogenic activities.•Medieval impact in a marginal landscape was connected to agricultural and industrial activities.•Past human activities can best be understood by linking archaeology and the soil sciences. This study presents the results of a multi-element and multivariate geochemical analysis of a deserted medieval settlement and its agricultural landscape, including a monastic (Cistercian) manorial farm, situated in the Czech Republic. We used LiDAR survey and historic maps to detect relict landscape features and identify past changes in land use-patterns. Comparing archaeological evidence and soil chemistry, we found that geochemical signals - derived from samples collected in the courtyard of the farm and in the surrounding area and processed by PCA on log-transformed and isometrically log-transformed data - responded to different agricultural activities, which could be linked to the farm and - more broadly - to current and past land-uses. The results generally illuminate the environmental impact of monastic settlement in a marginal landscape, with regard to different economic activities, and the study also demonstrates that this type of analysis allows tracing anthropogenic and natural phenomena over a wider area, beyond the limits of the more narrowly defined archaeological site.