The paper presents runoff and soil erosion measurements from plots on outward-sloping rainfed agricultural terraces in the Likhu Khola drainage basin, Middle Hills, Nepal, for the pre-monsoon and monsoon periods of 1992 and 1993. Runoff coefficients ranged from 5% to over 50%, depending on the nature of the rainfall event and the characteristics of the terrace. Total rainfall amount provided the highest level of explanation for the variation in runoff. Soil losses ranged from 2.7 to 8.2 t ha¯¹ for 1993 and up to 12.9 t ha¯¹ for 1992. The higher losses were associated with red, finer-grained soils. The majority of these rates are lower than the rates of soil loss that have been commonly perceived for the Middle Hills of the Himalaya. However, they are broadly similar to rates obtained from the few other studies that have examined runoff and erosion under traditional rainfed cultivation. The results suggest that a re-evaluation of the degree of land degradation in such areas may be necessary. Relationships between soil loss and rainfall characteristics were highly variable but were improved consider- ably when vegetation cover was included. This indicates that the maintenance of some form of ground cover is advisable if runoff and erosion are to be minimized.

Introduction

Soil and water are widely recognized as very important resources in the mountain kingdom of Nepal (HMG Nepal, 1988). Over the past 20 years, significant concerns have been raised over the degradation of the soil resource in the Middle Hills of Nepal as a result of the expansion of agricultural land and the increase in cropping intensity. These concerns, most notably expressed by Eckholm (1976) and the World Bank (1979), led to the formulation of the Theory of Himalayan Degradation. This‘Theory’ has been reviewed critically by Ives and Messerli (1989). Its central tenet was that deforestation and land use change were causing accelerated runoff and soil erosion on the steep Himalayan hillslopes. Implicit if not explicit in the ‘Theory’was the view that, if accelerated runoff and soil loss were occurring, the local culti- vators were either oblivious to what was happening or, if aware, were somehow unconcerned. This would clearly be an unsustainable situation.

By the late 1980s these ideas were coming under increasing scrutiny; a number of detailed studies were suggesting that deforestation in the Middle Hills was not a recent phenomenon, and that it was currently occurring at rates of less than 0.1% per annum (Land Resource Mapping Project, 1986; Mahat, Griffin, & Shepherd, 1986). The contribution of human activity to landsliding was also being questioned (Ramsay, 1985, 1986). Also some studies, such as that of Fleming (1983), were indicating that soil erosion on cultivated slopes was not as high as was thought. Above all, it was recognized that reliable and accurate data with which to test the various components of the ‘Theory’, both at the hillslope and catchment scales, were largely unavailable (Ives & Messerli, 1989). The data that did exist were frequently based on perceptions and observations over relatively short time periods (Gilmour, 1991), usually with a western, developed perspective, and were often contradictory and fraught with uncertainty (Thompson & Warburton, 1985). In terms of soil ero- sion, the paucity of data relating to rates, timing and processes was highlighted. Thus, most studies concluded that there was a clear need to evaluate carefully the ways by which human activity may be causing accelerated soil erosion through changing land use, land cover and land management practices.