NITROGEN-FIXATION BY 3 TROPICAL FORAGE LEGUMES IN AN ACID-SOIL SAVANNA OF COLOMBIA

The proportion of nitrogen derived from fixation was measured by N-15 isotope dilution in three tropical grass-legume pastures on two Oxisol s differing in texture over a 3 year period using three non-fixing con trols. The use of the companion pasture grass, Brachiaria dictyoneura, was satisfactory as a non-fixing control when compared with a non-fix ing Panicum maximum ecotype KK16 and native savanna grasses. Amounts o f nitrogen fixed ranged from 0.3 to 40 kg N ha(-1) 12 weeks(-1) during the wet season and were greatest with Stylosanthes capitata followed by Centrosema acutifolium and Arachis pintoi mainly as a result of gre ater legume biomass in the former compared with the latter two species . There were little or no differences amongst the three legumes in kg N fixed t legume DM-1. Amounts of nitrogen fixed per unit area decreas ed over the 3 year period mainly as a result of a loss of legume bioma ss. The % N derived from fixation (%Ndfa) was generally greater than 8 0% in all legumes on both soil types and there were little or no diffe rences in %Ndfa with two levels of fertilization given at establishmen t. The %Ndfa did not change over time. In a separate experiment with A . pintoi, %Ndfa did decrease with increasing legume proportion in the sward. The results indicate that in tropical pastures sown on low fert ility acid soils the amounts of nitrogen fixed by forage legumes are d ependent on legume growth and persistence. Soil type, level of fertili zation or age had little effect on the N-2-fixation process as measure d by %Ndfa. Based on these results approximate amounts of nitrogen fix ed may be estimated by simple measurements of legume biomass-N x 0.8. (C) 1997 Elsevier Science Ltd.