Soil erosion rates on arable land frequently exceed the pace at which new soil is formed. This imbalance leads to soil thinning (i.e. truncation), whereby subsoil horizons and their underlying parent material become progressively closer to the land surface. As soil erosion is a selective process and subsurface horizons often have contrasting properties to the original topsoil, truncation-induced changes to soil properties might affect erosion rates and runoff formation through a soil erosion feedback system. However, the potential interactions between soil erosion and soil truncation are poorly understood due to a lack of empirical data and the neglection of long-term erodibility dynamics in erosion simulation models. Here, we present a novel model-based exploration of the soil erosion feedback system over a period of 500 years using measured soil properties from a diversified database of 265 agricultural soil profiles in the UK. For this, we adapted the Modified Morgan–Morgan–Finney model (MMMF) to perform a modelling experiment in which topography, climate, land cover, and crop management parameters were held constant throughout the simulation period. As selective soil erosion processes removed topsoil layers, the model gradually mixed subsurface soil horizons into a 0.2 m plough layer and updated soil properties using mass-balance mixing models. Further, we estimated the uncertainty in model simulations with a forward error assessment. We found that modelled erosion rates in 99 % of the soil profiles were sensitive to truncation-induced changes in soil properties. The soil losses in all except one of the truncation-sensitive profiles displayed a decelerating trend, which depicted an exponential decay in erosion rates over the simulation period. This was largely explained by decreasing silt contents in the soil surface due to selective removal of this more erodible particle size fraction and the presence of clayey or sandy substrata. Moreover, the soil profiles displayed an increased residual stone cover, which armoured the land surface and reduced soil detachment. Contrastingly, the soils with siltier subsurface horizons continuously replenished the plough layer with readily erodible material, which prevented the decline of soil loss rates over time. Although our results are limited by the edaphoclimatic conditions represented in our data, as by our modelling assumptions, we have demonstrated how modelled soil losses can be sensitive to erosion-induced changes in soil properties. These findings are likely to affect how we calculate soil lifespans and make long-term projections of land degradation.

中文翻译：

---

土壤变薄会改变土壤可蚀性吗？长期侵蚀反馈系统的探索

可耕地的土壤侵蚀速度经常超过新土壤形成的速度。这种不平衡导致土壤变薄（即截断），从而使底土层及其下面的母质逐渐靠近陆地表面。由于土壤侵蚀是一个选择性过程，地下层通常具有与原始表层土截然不同的特性，截断引起的土壤特性变化可能通过土壤侵蚀反馈系统影响侵蚀率和径流形成。然而，由于缺乏经验数据以及在侵蚀模拟模型中忽视了长期可蚀性动力学，人们对土壤侵蚀和土壤截断之间的潜在相互作用知之甚少。这里，我们使用英国 265 种农业土壤剖面的多元化数据库中测量的土壤特性，对 500 年来的土壤侵蚀反馈系统进行了基于模型的新探索。为此，我们采用修改后的 Morgan-Morgan-Finney 模型 (MMMF) 来执行建模实验，其中地形、气候、土地覆盖和作物管理参数在整个模拟期间保持不变。随着选择性土壤侵蚀过程去除表土层，该模型逐渐将地下土壤层混合到 0.2 m 的犁层中，并使用质量平衡混合模型更新土壤特性。此外，我们通过前向误差评估来估计模型模拟中的不确定性。我们发现 99% 的土壤剖面中模拟的侵蚀率对截断引起的土壤特性变化很敏感。除截断敏感剖面之一外，所有土壤流失均呈减速趋势，这表明模拟期间侵蚀率呈指数衰减。这在很大程度上可以解释为由于选择性去除了这种更易侵蚀的粒径部分以及粘土或沙质基质的存在，土壤表面的淤泥含量减少。此外，土壤剖面显示出增加的残留石覆盖，这保护了地表并减少了土壤剥离。相比之下，地下层较粉质的土壤不断用易侵蚀的物质补充犁层，从而防止土壤流失率随时间下降。尽管我们的结果受到数据中所代表的土壤气候条件的限制，但正如我们的建模假设一样，我们已经证明了模拟土壤流失如何对侵蚀引起的土壤特性变化敏感。这些发现可能会影响我们计算土壤寿命和对土地退化进行长期预测的方式。