Relative efficacy of two biophysical approaches to assess soil loss tolerance for Doon Valley soils of India

Abstract

One of the major factors in favor of soil conservation measures is the prevention of top fertile soil removal, which adversely affects the crop productivity, depending upon the type of crop, soil, and erosion intensity. This reasoning has generally been assigned qualitatively and has rarely been supported through a quantitative relationship between soil loss and crop productivity. The soil loss tolerance limit (T value) is one of the several indicators to properly explain this phenomenon. The T value can be used as a guide to decide the maximum soil loss that can be removed before the long-term soil productivity is adversely affected. In this paper, two methods have been compared in determining T value on a regional scale for Doon Valley conditions in India. The first approach is based on assessment of the productivity index (PI) that considers permissible soil productivity loss rate (δ) and planning horizon (H) for sustainable land use. Productivity index is assessed and then related with tolerable rate of soil loss. The second approach is based on a quantitative weighted additive model, which has been used to define the current state of the soil resource. Both methods have been found to be good indices of soil loss tolerance value. However, the PI-based approach requires a complicated depth-wise dataset, including available water capacity, bulk density, and pH, which at present is not available for most of the ecological regions of India. Generating such a database may require long time and large investment. On the other hand, the weighted additive model requires a minimum data set of six soil attributes, which are readily available. Using the sensitivity index, the different T value at each of the study sites was separately compared. The overall mean of the sensitivity index was statistically insignificant at p < 0.05 for each location. Both methods were able to provide a reliable estimate of T value at different locations. However, the weighted additive model proved to be more reasonable as it requires a readily available dataset.