The removal of nitrogen (N) in grain cereal and canola crops in Australia exceeds 0.3 million t N/year and is increasing with improvements in average crop yields. Although N fertiliser applications to cereals are also rising, N-2-fixing legumes still play a pivotal role through inputs of biologically fixed N in crop and pasture systems. This review collates Australian data on the effects of grain legume N-2 fixation, the net N balance of legume cropping, summarises trends in the soil N balance in grain legume-cereal rotations, and evaluates the direct contribution of grain legume stubble and root N to wheat production in southern Australia. The net effect of grain legume N-2 fixation on the soil N balance, i.e. the difference between fixed N and N harvested in legume grain (Nadd) ranges widely, viz. lupin -29-247 kg N/ha (mean 80), pea -46-181 kg N/ha (mean 40), chickpea -67-102 kg N/ha (mean 6), and faba bean 8-271 kg N/ha (mean 113). Nadd is found to be related to the amount (Nfix) and proportion (Pfix) of crop N derived from N-2 fixation, but not to legume grain yield (GY). When Nfix exceeded 30 (lupin), 39 (pea) and 49 (chickpea) kg N/ha the N balance was frequently positive, averaging 0.60 kg N/kg of N fixed. Since Nfix increased with shoot dry matter (SDM) (21 kg N fixed/t SDM; pea and lupin) and Pfix (pea, lupin and chickpea), increases in SDM and Pfix usually increased the legume's effect on soil N balance. Additive effects of SDM, Pfix and GY explained most (R-2 = 0.87) of the variation in Nadd. Using crop-specific models based on these parameters the average effects of grain legumes on soil N balance across Australia were estimated to be 88 (lupin), 44 (pea) and 18 (chickpea) kg N/ha. Values of Nadd for the combined legumes were 47 kg N/ha in south-eastern Australia and 90 kg N/ha in south-western Australia. The average net N input from lupin crops was estimated to increase from 61 to 79 kg N/ha as annual rainfall rose from 445 to 627 mm across 3 shires in the south-east. The comparative average input from pea was 37 to 47 kg N/ha with least input in the higher rainfall shires. When the effects of legumes on soil N balance in south-eastern Australia were compared with average amounts of N removed in wheat grain, pea-wheat (1:1) sequences were considered less sustainable for N than lupin-wheat (1:1) sequences, while in south-western Australia the latter were considered sustainable. Nitrogen mineralised from lupin residues was estimated to contribute 40% of the N in the average grain yield of a following wheat crop, and that from pea residues, 15-30%; respectively, about 25 and 15 kg N/ha. Therefore, it was concluded that the majority of wheat N must be obtained from pre-existing soil sources. As the amounts above represented only 25-35% of the total N added to soil by grain legumes, the residual amount of N in legume residues is likely to be important in sustaining those pre-existing soil sources of N.
