Cultivated uplands in tropical latitudes are severely affected by soil water erosion. Conservation agriculture (CA) is specifically intended to control erosion. The aim of the present study is to analyse the effectiveness of CA measures to reduce the erosion in maize cultivation (Zea mays L.) on andosols in the mountains of southern Guatemala. Resolving this question is crucial to the development of sustainable soil policies appropriate to the conditions prevailing in this region. The study was conducted over a three-year period, from 2017 to 2019, on three experimental plots managed under conventional tillage (CT), reduced tillage (RT) and no-tillage (NT). The general-to-specific (GETS) method, based on reduction theory, was used to generate explanatory models of the erosion-affected surface. The study was carried out at two scales, first, with the complete database, and second, the database was segmented according to the agricultural management systems. The results showed different rates of eroded soil surface between the three management systems: 73.2% under CT, 41.3% under RT and 20.4% under NT. Analysis of the complete database (n = 36) showed that the litter cover (ryl.p = –0.86, p < 0.001) and the soil disturbance (ryp.l = 0.57, p < 0.001) were, in that order, the factors with the greatest explanatory power of the eroded surface. The segmented analysis (n = 12) showed that the management system adopted had a decisive influence on the ground cover (litter and weed cover) and, therefore, on the soil erosion. Under CT, the eroded surface was only correlated with the weed cover (ryw.l = –0.68, p < 0.05), under NT only with the litter cover (ryl.w = –0.89, p < 0.001) and under RT the erosion did not correlate with either of the vegetal layers. Three conclusions are derived from this study. First, litter layer was the key explanatory factor of erosion. Second, this factor is highly influenced by the agricultural management system. The proportion and distribution of the litter layer in each management situation were key to explaine the different soil erosion rates between the three management systems. And finally, it is proposed for the area of this study the soil management under NT with a dense and well distributed litter cover.

热带纬度的耕地高地受到水土流失的严重影响。保护性农业 (CA) 专门用于控制侵蚀。本研究的目的是分析 CA 措施对减少玉米种植侵蚀的有效性（Zea maysL.) 在危地马拉南部山区的山脊上。解决这个问题对于制定适合该地区普遍条件的可持续土壤政策至关重要。该研究在 2017 年至 2019 年的三年期间进行，在常规耕作 (CT)、减耕 (RT) 和免耕 (NT) 下管理的三个试验地块上进行。基于还原理论的一般到特定 (GETS) 方法用于生成受侵蚀影响表面的解释模型。该研究在两个尺度上进行，一是使用完整的数据库，二是根据农业管理系统对数据库进行细分。结果表明，三种管理系统之间侵蚀土壤表面的比率不同：CT 下为 73.2%，RT 下为 41.3%，NT 下为 20.4%。p  < 0.001) 和土壤扰动 (ryp.l = 0.57, p  < 0.001) 依次是对侵蚀表面具有最大解释力的因素。分段分析 (n = 12) 表明所采用的管理系统对地被植物（凋落物和杂草覆盖）具有决定性影响，因此对土壤侵蚀也有决定性影响。在 CT 下，侵蚀表面仅与杂草覆盖相关 (ryw.l = –0.68, p  < 0.05)，在 NT 下仅与凋落物覆盖相关 (ryl.w = –0.89, p < 0.001) 并且在 RT 下，侵蚀与任一植物层都无关。本研究得出三个结论。首先，凋落物层是侵蚀的关键解释因素。其次，这一因素受农业管理体制的影响很大。每种管理情况下凋落物层的比例和分布是解释三种管理系统之间不同土壤侵蚀率的关键。最后，建议本研究区域在北领地下进行土壤管理，具有密集且分布良好的凋落物覆盖。