Seasonal variation in the suitability of different methods for estimating biological nitrogen fixation by grain legumes under rainfed conditions

Abstract

Biological nitrogen fixation (BNF) by legume crops was estimated in a Mediterranean environment at ICARDA in northern Syria for 3 consecutive seasons beginning in 1991-92. Using the classical isotope dilution technique (NID), estimates ranged from 32 to 82 kg N/ha for chickpea and from 18 to 82 kg N/ha for lentil. In comparison the simple N-difference method gave lower, sometimes negative, estimates for BNF by both crops in the latter 2 seasons but a higher estimate for chickpea in the first year. Discrepancies in the estimates from N-difference were correlated with differences in the amount of soil N taken up by the legume and the non-fixing wheat reference crop. Since soil N uptake by lentil in the first year was similar to wheat, the estimates of BNF from the 2 methods for that season were similar. The indirect effects of an interaction of added N fertiliser on N derived from the soil and thus on N uptake and estimated BNF are discussed in relation to the use of the isotope dilution method with A-value modification (NAV). Despite some significant differences in A-value for soils receiving different amounts of fertiliser it is demonstrated that the A-value method used in this study, with fertiliser rates of 10 kg N/ha to the legume and 30 kg N/ha to the non-legume, resulted in BNF estimates for lentil similar to those obtained using classical isotope dilution. However, this was not the case for chickpea where a direct inhibitory effect of fertiliser N at 30 kg N/ha resulted in lower estimates of BNF from NID than NAV. Since the reference crops derived as much as 90% of their N from the soil, it is recommended that future BNF studies using isotope dilution techniques for lentil and chickpea crops at ICARDA use a fertiliser rate lower than that used in this study. An isotope dilution method utilising a slow-release source of 15N or the natural abundance technique for estimating BNF are suggested as potentially useful alternatives. The need for a basic understanding of the soil N dynamics pertinent to each site as a prerequisite for choosing an appropriate method for estimating BNF is highlighted.