Dryland agriculture is both a potential source and potential sink for CO 2 and other greenhouse gases. Many carbon accounting systems apply simple emissions factors to production units to estimate greenhouse gas (GHG) fluxes. However, in Australia, substantial variation in climate, soils, and management across >20 Mha of field crop sowings and >30 Mha of sown pastures in the intensive land use zone, provides substantial challenges for a national carbon accounting system, and simple emission factors are unlikely to apply across the region. In Australia a model framework has been developed that requires estimates of crop dry matter production and harvested yield as the first step to obtain carbon (residue) inputs. We use Australian Bureau of Statistics data to identify which crops would need to be included in such a carbon accounting system. Wheat, barley, lupin, and canola accounted for >80% of field crop sowings in Australia in 2006, and a total of 22 crops account for >99% of the sowing area in all States. In some States, only four or six crops can account for 99% of the cropping area. We provide a ranking of these crops for Australia and for each Australian State as a focus for the establishment of a comprehensive carbon accounting framework. Horticultural crops, although diverse, are less important in terms of total area and thus C balances for generic viticulture, vegetables, and orchard fruit crops should suffice. The dataset of crop areas presented here is the most comprehensive account of crop sowings presented in the literature and provides a useful resource for those interested in Australian agriculture. The field crop rankings presented represent only the area of crop sowings and should not be taken as rankings of importance in terms of the magnitude of all GHG fluxes. This awaits a more detailed analysis of climate, soils, and management practices across each of the regions where the crops are grown and their relationships to CO 2, nitrous oxide and methane fluxes. For pastures, there is a need for more detailed, up to date, spatially explicit information on the predominant sown pasture types across the Australian cropping belt before C balances for these can be more reliably modelled at the desired spatial scale.
