Scalable computer vision-based assessment of bait lamina sticks to quantify soil fauna activity (Part 2)
Autoren: Adrija Roy, Lukas Thielemann, Masahiro Ryo, Juan Camilo Rivera-Palacio, Konlavach Mengsuwan, Kathrin Grahmann
Abstract
Soil erosion remains a major global concern affecting agricultural productivity and land sustainability. This study investigates the magnitude and variability of soil erosion in a long-term experiment (LTE) established in 1963 in Colonia, Uruguay, aiming to compare the performance of the 137Cs tracer technique and the RUSLE model across different crop rotations. We hypothesized that pasture inclusion into rotations reduces soil erosion compared to continuous cropping, regardless of the model. Four contrasting LTE treatments were selected: continuous cropping without fertilization, continuous cropping with fertilization, and two crop-pasture rotations with 33% or 50% time under pasture. Each plot had a size of 25 × 200 m positioned along a hillslope and was sampled at six sampling points along a transect from the hilltop towards the footslope. Soil erosion was assessed using two methods: (1) the 137Cs technique, estimating net soil redistribution based on deviations from a reference site, and (2) the RUSLE model, calibrated with national agronomic and climatic data. The 137Cs method showed erosion rates from −73 to +51 Mg ha−1 year−1, with accumulation at footslopes and highest losses in eroded gullies under unfertilized continuous cropping. RUSLE predicted losses from −1 to −20 Mg ha−1 year−1, based on sheet erosion assumptions. The comparison was limited to five sampling points where erosion processes were dominant and compatible with RUSLE assumptions. Nevertheless, results confirmed that crop-pasture rotations substantially reduced erosion and preserved soil over six decades. We conclude that integrating pastures into rotations is an effective erosion control strategy in sloping agricultural systems and helps maintain long-term productivity.