Abstract. Wet canopy and dry canopy evaporation from young and mature plantations of Pinus caribaea on former grassland soils under maritime tropical conditions in southwestern Viti Levu, Fiji, were determined using micrometeorological and hydrological techniques. Modeled annual evaporation totals (ET) of 1926 and 1717 mm were derived for the 6-and the 15-year-old stands, respectively. Transpiration made up 72% and 70% of annual ET, and modeled rainfall interception by the trees and litter layer was 20-22% and 8-9% in the young and the mature stands respectively. Monthly ET was related to forest leaf area index and was much higher than that for the kind of tall fire-climax Pennisetum polystachyon grassland replaced by the forests. Grassland reforestation resulted in a maximum decrease in annual water yield of 1180 mm on a plot basis, although it is argued that a reduction of (at least) 500-700 mm would be more realistic at the catchment scale. The impact of reforesting grassland on the water resources in southwest Viti Levu is enhanced by its location in a maritime, seasonal climate in the outer tropics, which favors a larger difference between annual forest and grassland evaporation totals than do equatorial regions.
IntroductionOut of concern about the potentially adverse effects on soils, hydrology, and climate of the continuing conversion of the world's tropical rain forests to other, mostly agricultural, land uses, considerable effort has gone into the evaluation and prediction of such effects during the last two decades [Lal, 1987;Bruijnzeel, 1990Bruijnzeel, , 1996Bruijnzeel, , 1998Fritsch, 1992 [Bruijnzeel, 1990], the associated range reported for the initial increase in streamflow after total forest clearance is quite large (125-820 mm yr-•). This variation can be explained only partially by differences in rainfall between locations or years. Other factors affecting postclearing rates of growth and water uptake (e.g., soil fertility status, elevation), runoff disposal (catchment steepness, soil depth), or both (notably the degree of surface disturbance by machinery or fire) play a role as well [Bruijnzeel, 1996[Bruijnzeel, , 1998]. In addition, Shuttleworth [1989] suggested that compared to midcontinental sites, tropical deforestation is likely to have the most effect on river flow (though not necessarily on climate) at continental edge and island locations. In the absence of exper-