Rainfed grain production, based on winter cereals, is marginal in south-west Queensland, Australia, because of low and variable rainfall and high evapotranspiration. Also, grain yield and grain quality have decreased as soil fertility, particularly soil nitrogen supply, has declined on older cropping lands. An option for improving soil N supply is to include legume-based pastures in rotation with winter cereals. The objective of this study was to determine the effects of short-term (18 months) legume pastures (annual medics and lucerne + annual medics), and longer term (3 years) mixed grass (Bambatsi panic) and legume (lucerne + annual medics) pasture phases on sheep production and on soil water and N supply and production of subsequent wheat crops on a grey Vertosol soil.

Two separate phases of annual medics and lucerne + annual medics pastures produced mean total aboveground dry matter yield of 7.10 t/ha of annual medics and 5.80 t/ha of lucerne + annual medics over the 18-month periods. For two phases of the grass + legume pastures, mean total aboveground dry matter yield was 3.95 t/ha for grass and 8.19 t/ha for legume over 3 years. Over an 18-month period, sheep bodyweight gains and fleece weights were similar for the annual medics, lucerne + annual medics and grass + legume pastures and were approximately five times greater than those from native pasture as a result of the greater stocking rate possible on the sown pastures.

Greater drying of the soil profile occurred following lucerne + annual medics and grass + legume pasture phases than continuous wheat, resulting in lower soil water content at sowing of wheat crops following these pasture phases on several occasions. Mean soil nitrate-N benefits before wheat sowing in the first year following termination of the 18-month annual medics, lucerne + annual medics, and the 3-year grass + legume pasture phases were 45, 44 and 42 kg N/ha, respectively. Grain N yields and gross margins of the first wheat crops following the 18-month annual medics, lucerne + annual medics, and the 3-year grass + legume pasture phases were similar in value to continuous wheat with ~60, 80, and 40–60 kg N/ha fertiliser applied at sowing, respectively. Improvements in grain N yield and gross margin were still evident in the fifth wheat crop following annual medics and lucerne + annual medics pastures and in the third wheat crop following grass + legume pasture, compared with continuous wheat without N fertiliser addition.

Total gross margins from 1996 to 2005 were 1.6–2.5 times greater for the pasture–crop rotations than continuous wheat where no N fertiliser was applied to wheat. However, gross margins were greater in continuous wheat than in pasture–crop rotations where N fertiliser was applied to target prime hard grade grain protein in wheat.

The 3-year grass + legume pasture phase showed potential to improve surface soil structure and water infiltration and to reduce decline in soil organic carbon concentration at 0–0.1 m depth, compared with continuous wheat cropping and shorter-term legume pasture phases.