The effect of including phases of long-growing-season annuals and herbaceous perennial pastures on water use was examined at 2 sites (deep sand and duplex soil) in Western Australia. Herbaceous perennials used were lucerne (Medicago sativa), and a mix of C3 grasses comprising phalaris (Phalaris aquatica), tall wheat grass (Thinopryum ponticum), and tall fescue (Festuca arundinacea) (perennial grass treatment). The long-season annual treatment was a mix of yellow and pink serradella (Ornithopus sp.) and Casbah biserrula (Biserrula pelecinus). These treatments were compared with annual-based pasture that was a mixture of subterranean clover with capeweed and Brassica species, and annual crops. Pasture treatments were first sown in 1998. High senescence of C3 grasses over the 1998–99 summer and poor germination of serradella/Casbah biserrula in the autumn of 1999 necessitated the re-seeding of the long-season annual and the perennial grass treatment in 1999. Wheat was sown in 1998, lupin in 1999, and barley in 2000 in an annual crop treatment. Soil water content to 1.5 m was measured hourly using frequency domain reflectometer probes, and a neutron probe was used monthly to measure changes in soil water to 5 m. Herbage production and species composition were determined. In each year of the study, annual pasture species senesced by November. About 20 lucerne plants/m2 persisted through the first summer–autumn in deep loamy sand and 40 lucerne plants/m2 in a duplex soil. Perennial C3 grass species did not survive the summer–autumn in sufficient density and distribution to evaluate their effect on soil water. Annual dry matter (DM) production in lucerne-based and subterranean clover-based pasture was not significantly different. Dry matter production in lucerne between 1 December and the following May–June, when germination of annual-based pastures occurred, was 1.2–1.9 t/ha at one site and 0.2–1.6 t/ha at another site. Long-season annual pastures produced significantly more DM than either lucerne or subterranean clover-based pastures in one season at one site but produced significantly less DM than either lucerne or subterranean clover-based pasture at another site in another season. Long-season annual-based pastures extracted amounts of soil water to a depth of 5 m similar to subterranean clover-based pasture when these were grown on deep sand and a duplex soil. In contrast, lucerne removed an additional 128 mm of water to 5 m, with 70 mm of this water being drawn from 2.5–5 m, compared with subterranean clover-based pasture. Lucerne was comparatively less effective in extracting water from a duplex soil where rooting depth was restricted to 2 m by a saline watertable. Early germination of annual pastures appeared to reduce drainage compared with a crop treatment where weeds were killed in autumn and early winter ahead of seeding. The need for studies at landscape scales that include concurrent measurements of groundwater levels and changes in soil water content to a depth of at least 5–6 m under perennial-based production systems is highlighted.

Decision support systems (DSS) that provide advice on best N fertiliser management practice ideally need to assess the effect that fertiliser type, rate and management have on N loss. Currently only nitrate leaching and denitrification losses can be assessed in Australian DSS that use output from the Agriculture Production Simulator (APSIM). This paper describes a simple spreadsheet-based model for estimating NH3 loss from urea, urea ammonium nitrate and ammonium sulfate that can be used in conjunction or independent of current DSS. The use of scaling factors that reduce NH3 emission on the basis of factors known to be key determinants of NH3 loss, including type of fertiliser used, rate of fertiliser applied, management of fertiliser, rainfall and crop development was found to account for 85% of the variance between predicted and observed values when tested against 40 case studies. The model also had good predictive power with a root mean square error (RMSE) equivalent to 4.8% of N applied.

The effect of annual applications of gypsum at rates of 0, 5, 15, and 30 kg S/ha on grain yields of lupin, wheat, and canola were studied between 1992 and 1995, and the residual value of these sulfate (SO42–) inputs was examined in 1996 and 1997. The top 0.1 m of soil was collected in 1994, 1996, and 1997 to determine KCl-40 extractable S. Extensive soil sampling was undertaken in 1996 and 1997 to determine quantities of SO42– in soil layers to 0.5 m, and to measure rates of net SO42– mineralisation. Application of gypsum resulted in significantly higher grain yields in 2 out of 3 wheat crops, and in canola, but SO42– fertiliser use did not increase lupin yield. Wheat yields were also significantly higher in treatments with a prior history of gypsum use but lupin grain yields were not enhanced by previous gypsum applications. The quantities of SO42– in the top 0.2 m of soil decreased sharply in the first year after SO42– fertilisers were withheld with further declines in SO42– to low levels across all treatments in the second year gypsum was not applied. A buildup of SO42– was evident in soil between 0.3 and 0.5 m. However, S budgets for the top 0.5 m suggested that little SO42– leached below 0.5 m in the years studied. Previous sulfur fertilisation did not significantly affect the total growing season net SO42– mineralisation, which was 2.2–8.0 kg S/ha. Mineralisation of organic S over a summer fallow period (0–5.8 kg S/ha) and the total S net mineralised over 17 months (7–17 kg S/ha) was significantly affected by previous S fertiliser history. Net SO42– mineralisation was not correlated with the labile organic S fraction extracted using the KCl-40 soil test.

Net nitrogen (N) mineralisation in soil and N uptake by wheat from mature shoots and roots of subterranean clover, capeweed, and annual ryegrass, and from clover burrs were assessed with 15N-labelled plant material in 2 field studies, using confined micro-plots. In the first study, shoot residues of the 3 species (150 g DM/m2) were placed on the soil surface, and roots of the 3 species (75 g DM/m2) were mixed into 0–10 cm soil. The treatments were applied in March 1991. The shoot residues were incorporated into soil in early June 1991. Net 15N mineralisation from the clover, capeweed, and ryegrass shoots during the 8-month experimental period was estimated to be, respectively, 15%, 12%, and 12%, and for the corresponding roots was 10%, 7%, and 6%. Negligible net mineralisation of 15N occurred during the 2.5 months that the shoot residues were on the soil surface. Crop 15N recoveries in wheat, at maturity, in November 1991 were 9%, 7%, and 7%, respectively, of that applied in the clover, capeweed, and ryegrass shoot residues. The respective crop recoveries from the root residues were 6%, 5%, and 3%. Less than 5% of N taken up by wheat was obtained from shoot or root residues. In a second similar study, 15N-labelled subterranean clover shoots (200 g DM/m2) and burrs (75 g DM/m2) were applied in December 1992; 3% of 15N in the clover shoots was net mineralised during the 5 months they were on the soil surface. Crop recoveries of 15N in October 1993, at the time of wheat anthesis, from the clover shoots and burrs were, respectively, 14% and 17% of applied 15N. The results of these field studies suggest that mature shoot residues and the associated intact roots (recoverable by wet-sieving), and clover burrs, make only a small direct contribution to the N response of cereals immediately following ley pasture in southern Australia. They also indicate that, under Mediterranean climatic conditions, generally very little net N mineralisation occurs from mature shoot residues until the shoots are incorporated into soil.