MANAGEMENT OF SESBANIA-ROSTRATA GREEN MANURE CROPS GROWN PRIOR TO RAIN-FED LOWLAND RICE ON SANDY SOILS

Extremely infertile sandy soils and erratic rainfall are major constra ints to rainfed lowland rice yields in northeast Thailand. Response to inorganic fertilizers is low and variable. Farmers rely on farmyard m anure and leaf Litter from trees and shrubs to stimulate yield respons es in rice. Field experiments on an Aeric Paleaquult soil were conduct ed in two successive wet seasons to develop production systems for Ses bania rostrata as a green manure (GM) crop prior to lowland rice. In t he first experiment sesbania was grown at seeding rates of 15, 30 and 60 kg ha(-1) with or without P application at 22 kg/ha. Sesbania was s oil-incorporated at 46 or 61 days after seeding (DAS). Biomass product ion and accumulation and their effects on rice were determined in comp arison with the effect of zero or 50 kg ha(-1) of inorganic N. The sec ond experiment evaluated the effect of 2, 14, and 21 day intervals bet ween incorporation of 50 day old sesbania and rice transplanting. P ap plication, and increased growth duration (61 vs 46 DAS), significantly improved sesbania total dry matter yield (TDMY) and nutrient accumula tion. The effect of seeding rate on yield was positive only when P was applied, unless incorporation was delayed to 61 days. P application t o sesbania and delayed sesbania incorporation significantly increased rice grain and TDM yields. GM seeding rates did not significantly affe ct rice yield or TDMY. GM effects compared favorably to inorganic N. R ice yield response to inorganic N application was significant, but not to unfertilized, incorporated weeds. In the second trial GM (2.26 t h a(-1) dry weight) contributed 76 kg N ha(-1) to the rice crop and sign ificantly increased grain yield and TDMY. Differences in the interval between sesbania incorporation and rice transplantation did not affect crop performance. Products of GM decomposition tended to depress seed ling vigour and survival in the 2d interval treatment but grain yield was still 159% higher than the no-GM control after replanting the affe cted rice hills. The optimum incorporation-to-transplanting interval a t the site is about 5-7 days to avoid adverse effects of decomposition .