WATER EROSION IN NATURAL GRASSLAND UNDER DIFFERENT MANAGEMENTS: WATER AND SOIL LOSSES AND PHOSPHORUS, POTASSIUM AND AMMONIUM IN THE RUNOFF

WATER EROSION IN NATURAL GRASSLAND UNDER DIFFERENT MANAGEMENTS: WATER AND SOIL LOSSES AND PHOSPHORUS, POTASSIUM AND AMMONIUM IN THE RUNOFF Native pastures in southern Brazil are used for cattle and sheep grazing. The forage offer in the spring-summer season is greater than the animal requirement, so the forage left over is usually burned. Pasture burning mineralizes nutrients and leaves the soil bare, triggering water erosion. This study aimed to quantify total soil and water losses and P, K and NH(4)(+) contents in the runoff water from an Oxisol with native pasture. In the treatments, the native pastures, were i) unburned and unfertilized, ii) unburned and P-fertilized, iii) burned and unfertilized, and iv) burned and P-fertilized. Rain was simulated at an intensity of 75 mm h(-1) and duration of 3 h on the plots (11 in long along the slope and 3.5 in wide). Burning of the native vegetation decreased the times until runoff began and peaked and reduced soil water infiltration, but increased the maximum runoff rate, soil and water losses by water erosion and the P, K and NH(4)(+) concentration and total losses in the runoff water, compared to no burning. The runoff sediment concentration was higher in the treatment with vegetation burning than without. The highest and lowest values of sediment concentration were detected after a shorter runoff period in the treatment with vegetation burning than in that without, which is related to the 8.9 times higher total soil loss from burnt than from unburnt grassland. The P concentration and total loss, in the runoff water were higher in the treatment with than without phosphate fertilizer, in the treatment with burning as well as that without. For K and the NH(4)(+), the concentration and total losses were higher in the treatment with burning than in that without. Phosphorus, K and NH(4)(+) concentration in the runoff water fit the exponential model y = ae(-bx) in relation to runoff duration (p < 0.01).