Erosion is the most deleterious form of soil degradation. Besides reducing the productivity capacity for crops, it can cause serious environmental problems, such as sedimentation and pollution of water sources. Yet the use of sound soil management systems and well-planned support conservation practices can solve erosion problems satisfactorily. In order to obtain quantitative information on the subject as guidelines for conservation plans, an erosion experiment under natural rainfall was conducted on a very clayey soil (Typic Haplortox), in Santo Angelo, region of Missoes, Rio Grande do Sul State, Brazil, December 1994 through May 1996, to quantify soil and water losses caused by rainfall erosion. Treatments of conventional tillage, reduced tillage, and no-till were evaluated under conditions of "soil with improved fertility" in slope-steepness classes of 0-0.04, 0.04-0.08, and 0.08-0.12 m m(-1) (with average slope gradients of, respectively, 0.035, 0.065, and 0.095 m m(-1)) and of "soil with actual fertility" in the 0.04-0.08 m m(-1) slope-steepness class. A crop sequence of two soybean (Glycine may, L) cycles (spring-summer 1994/95 and 1995/96) followed by black oat (Avena strigosa, S.) (fall-winter 1995) was used for an evaluation of erosion. Tillage and planting operations were performed along contour lines in all plots, except for the Wischmeier's unit plot, where plowing and disking operations were performed up and down the slope. The calculated rainfall erosivity index EI30 of the experimental period (1,5 year) was 10.236 MJ mm ha(-1) h(-1). Both in 1995 and 1996 a concentration in the period between January and March accumulated 70% of the total rainfall. Only under conventional tillage with no cropping and in vertical direction (control plot) there were great soil losses by rainfall erosion (up to 80 Mg ha(-1) in a single crop cycle), although the soil in the plot presented a very low erodibility value (K-factor of 0.0091 Mg ha h MJ(-1) mm(-1) ha(-1)). Under cropping, though in much lower values, soil losses were also highest in conventional tillage (about 13.0 Mg ha(-1) in 1,5 years), markedly in medium and high slope-steepness classes. Intermediate soil losses were associated with reduced tillage (about 4.0 Mg ha(-1) in 1,5 year), and the lowest with no-till (about 1.0 t ha(-1) in 1,5 year), while water losses were all very low and little differentiated. The condition "soil with improved fertility" greatly increased the crop aerial biomass, consequently the crop residue mass and, thus, the percentage of mulch covering the soil and soil loss by rainfall erosion, especially in the conventional tillage, although it did not appreciably affect water losses. In spite of tillage and planting operations along contours, improved soil fertility, and a relative high resistance of soil to erosion, the black oat-soybean crop in conventional tillage presented annual soil losses by rainfall erosion that lay very close to the tolerable level on slopes steeper than 0.04 m m(-1), even in the short slope-length of 21 m used in the experimental plots of this study.
