AbstractThe West Usambara Highlands in north‐eastern Tanzania have many smallholder agro‐ecosystems with unknown composition, management, and vulnerability to erosion. Their specific locations and spatial extent are difficult to trace by satellite images or remote sensing imagery alone. To address these limitations, we combined ground soil surveys, geographic information system, and erosion modelling to (a) locate and map smallholder agro‐ecosystems, (b) determine their biophysical characteristics, and (c) model their soil losses. Land resource information was collected from 301 random 0.1‐ha plots sampled from a total area of 200 km2. Annual soil losses were estimated using the universal soil loss equation. The study located six dominant agro‐ecosystems with the following spatial extent: maize‐bean (24.9%), maize‐bean‐agroforestry (31.2%), maize‐bean‐agroforestry‐high value trees (18.9%), tree farms (7.0%), forests (15.6%), and grazing lands (2.3%). Agroforestry and other tree‐based agro‐ecosystems dominate the area due to historical land use change and later institutional interventions. This study finds combined use of soil surveys, geographic information system, and modelling to be reliable in locating, mapping, and assessing soil losses in smallholder agro‐ecosystems. The agro‐ecosystems differ significantly (p < 0.05) in slope, vegetation cover, soil conditions, and soil losses. Soil loss in the maize‐bean agro‐ecosystem (28.3 t ha−1 yr−1) was 18 times higher compared with natural forests (1.57 t ha−1 yr−1) due to lower soil cover and inefficient conservation and cultivation practices. Our results show that adoption of soil conservation measures and improved vegetation cover technologies across the agro‐ecosystems reduces soil losses by 37% and increases organic carbon levels by 16%.
Read full abstract