Nutrient Cycling in Agroecosystems 56: 165–176, 2000. © 2000 Kluwer Academic Publishers. Printed in the Netherlands. 165 Biological nitrogen fixation by two tropical forage legumes assessed from the relative ureide abundance of stem solutes: 15 N calibration of the technique in sand culture Bruno J.R. Alves, Alexander S. Resende, Segundo Urquiaga & Robert M. Boddey ∗ Embrapa-Agrobiologia, Km 47, Serop´ edica, 23851-970, Rio de Janeiro, Brazil ( ∗ Corresponding author; e-mail: bob@cnpab.embrapa.br) Received 3 November 1998; accepted in revised form 4 May 1999 Key words: Centrosema, Desmodium ovalifolium, 15 N dilution technique, N 2 fixation, Ureide technique Abstract The use of the relative ureide abundance (RUA) in the sap of mainly tropical ureide-producing legumes as a means to estimate the contribution of biological nitrogen fixation (BNF) is potentially an useful technique as it does not require the use of reference plants or additions of 15 N-labelled fertilizer, and the analyses necessitate only relatively simple equipment. However, one problem in the application of the technique arises from the difficulty of obtaining sap samples from such legumes, especially small-stemmed forage legumes under field conditions. This study was conducted to investigate the possibility of using RUA in hot-water extracts of the stems of two forage legumes, Desmodium ovalifolium and a Centrosema hybrid, to estimate the contribution of BNF. In this case only ureide and nitrate are analysed to calculate RUA (100 × ureide-N/(ureide-N + nitrate-N)). The technique was calibrated with the 15 N isotope dilution technique in sand culture where the plants were fed with 5 different levels of nitrate (0, 12.5, 25, 50 and 100 mg N pot −1 ). Despite the fact that in many stem extracts more than 90% of the N was neither nitrate or ureide, the colorimetric techniques utilised proved reliable and relatively immune to interference from other solutes in the extracts. One problem with the use of the 15 N dilution technique to calibrate the RUA technique is that the former gives an integrated estimate of the BNF contribution since planting (or between harvests) and the latter is a point estimate at the time of sampling. This was overcome by using a ‘plant to plant simulation technique’ where estimates of BNF are calculated from the daily accumulation of total N and the labelled N derived from the growth medium by the legumes using a curve-fitting strategy. These estimates of BNF for the days when stem extracts were analysed for nitrate and ureide showed linear correlations (r 2 = 0.82 and 0.90 for the D. ovalifolium and Centrosema hybrid, respectively). This indicated that RUA of stem extracts of these two legumes was a reliable indicator of the BNF contribution, at least under controlled conditions. Introduction In the central savanna (Cerrado) region of Brazil it is estimated that over the last 30 years almost 50 million ha of native vegetation have been replaced by pastures planted with grasses introduced from Africa, princip- ally of the genus Brachiaria (Macedo, 1995). The grasses are normally planted after the savanna has first been cleared and planted to rice or soybean, and for a few years forage production is good. Subsequently grass regrowth decreases and the pastures become in- vaded by non-palatable species, bare patches appear in the sward and often the soils become compacted and invaded by termite mounds. This process of degrada- tion is generally provoked by overgrazing, but persists because of phosphorus and nitrogen deficiency. Re- cent work by Oliveira et al. (1997), at three sites in the Cerrado region, showed that the vegetative pro- ductivity of such degraded areas of Brachiaria could be improved by application of N fertilisers only, but a concomitant addition of P were essential for a vig- orous production. Owing to its immobility in soil and low export from pastures in the animal carcass, P fertilisation can have long-lasting benefits on pro-