Intensive agricultural management significantly affects soil chemical properties. Such impacts, depending on the intensity of agronomic practices, might persist for several decades. We tested how current soil properties, especially heavy metal concentrations, reflect the land-use history over a 24,000 ha area dominated by intensive apple orchards and viticulture (South Tyrol, ITA). We combined georeferenced soil analyses with land-use maps from 1850 to 2010 in a space-for-time approach to detect the accumulation rates of copper and zinc and understand how present-day soil heavy metal concentrations reflect land-use history. Soils under vineyards since the 1850s showed the highest available copper concentration (median of 314.0 mg kg^-1, accumulation rate between 19.4 and 41.3 mg kg^-1·10 y^-1). Zinc reached the highest concentration in the same land-use type (median of 32.5 mg kg^-1, accumulation rate between 1.8 and 4.4 mg kg^-1·10 y^-1). Using a random forest approach on 44,132 soil samples, we extrapolated land-use history on the permanent crop area of the region, reaching an accuracy of 0.72. This suggests that combining current soil analysis, historical management information, and machine learning models provides a valuable tool to predict land-use history and understand management legacies.

集约化农业管理显着影响土壤化学性质。这种影响，取决于农艺实践的强度，可能会持续几十年。我们测试了当前的土壤特性，尤其是重金属浓度，如何反映以集约型苹果园和葡萄栽培为主的 24,000 公顷地区（南蒂罗尔，ITA）的土地利用历史。我们以时空方法将地理参考土壤分析与 1850 年至 2010 年的土地利用图相结合，以检测铜和锌的积累率，并了解当今土壤重金属浓度如何反映土地利用历史。自 1850 年代以来，葡萄园下的土壤显示出最高的可用铜浓度（中位数为 314.0 mg kg ^-1，积累率在 19.4 和 41.3 mg kg ^-1之间·10 年^-1 )。锌在同一土地利用类型中达到最高浓度（中位数为32.5 mg kg ^-1，积累率在1.8-4.4 mg kg ^-1 ·10 y ^-1之间）。我们对 44,132 个土壤样本使用随机森林方法，推断该地区永久作物面积的土地利用历史，准确度达到 0.72。这表明，结合当前的土壤分析、历史管理信息和机器学习模型，为预测土地利用历史和了解管理遗产提供了一种有价值的工具。