Results from this series of studies indicate that production systems based on intercropping upland rice with indeterminate legumes may substantially improve yields under low external input conditions. Upland rice is generally produced without inorganic fertilizer input on acid soils of low inherent fertility. The goals of this research are to develop intercropping systems which increase N availability to rice, increase the eco-sustainability by regenerating nutrients within cropping systems and to improve subsistence farmers' food security and nutrition by reducing the risk of complete crop failure and increasing the harvest of protein.

During initial intercrops, two-thirds of the cropland was planted in rice and one-third in grain legumes, resulting in increased harvests with Land Equivalent Ratios between 1.2 and 1.8, indicating considerably increased resource efficiency with intercropping. Per-plant N nutrition of rice in the intercrop was also generally much more favorable than in monoculture upland rice. However, maintaining total rice yields at monocrop levels, a vital consideration for subsistence farmers whose main food is rice, proved difficult as a third of the land was occupied by legumes.

Further studies thus focused on gaining intercropping incentives without reducing monocrop rice yield levels. Indeterminate varieties of cowpea and lablab (Lablab purpureus) were selected. Cowpeas were chosen because of their grain production and adaptation to acid soils. Lablab was selected for its nitrogen-fixing capacities, dry season growth and forage potential. Lablab pods and seeds are also edible. Rice and legumes were planted at the same time, with the legumes planted after every third row of rice. Both legumes displayed excellent vegetative growth and were first cut back after 40 days. The clipped foliage was used as mulch between rice rows once for cowpeas, adding 21 kg N/ha. Lablab was clipped three times adding 17, 28 and 48 kg N/ha to the soil and was allowed to continue growing during the five month dry season. Over this time, it produced more green manure and stabilized the soil before being incorporated at the onset of the next rice cropping period. The lablab had accumulated 19 t/ha of fresh biomass with over 100 kg N/ha.

Rice intercropped with lablab had significantly higher yields than monocrop rice, increasing the rice yield of the total area from 1.54 to 1.88 t/ha. The cowpea intercrop did not negatively affect the total rice yield and contributed 1.02 t/ha of dry cowpeas. Higher rice yields in the intercrop were associated with a 66% increase in rice N uptake per unit area of intercrop. The rice grain quality was observed to be substantially improved by intercropping, with a 30% rise in N content in the rice grain from the cowpea/rice intercrop and a 37% rise from the rice/lablab intercrop.