Data from more than 580 field experiments conducted in South Australia over the past 30 years have been re-examined to estimate extractable soil phosphorus (P) levels related to 90% maximum yield (C90) for 7 crop species (wheat, barley, oilseed rape, sunflower, field peas, faba beans, potato) and 3 types of legume-based pasture (subterranean clover, strawberry clover, annual medics). Data from both single-year and longer term experiments were evaluated. The C90 value for each species was derived from the relationship between proportional yield responsiveness to applied P fertiliser rates (determined as grain yield in crops and herbage yield in ungrazed pastures) and extractable P concentrations in surface soils sampled before sowing. Most data assessments involved the Colwell soil P test and soils sampled in autumn to 10 cm depth. When all data for a species were considered together, the relationship between proportional yield response to applied P and soil P status was typically variable, particularly where Colwell soil P concentration was around C90. When data could be grouped according to common soil types, soil surface texture, or P sorption indices (selected sites), better relationships were discerned. From such segregated data sets, different C90 estimates were derived for either different soil types or soil properties. We recommend that site descriptors associated with the supply of soil P to plant roots be determined as a matter of course in future P fertiliser experiments in South Australia. Given the above, we also contend that the Colwell soil P test is reasonably robust for estimating P fertiliser requirements for the diverse range of soils in the agricultural regions of the State. In medium- and longer term experiments, changes in Colwell soil P concentration were measured in the absence or presence of newly applied P fertiliser. The rate of change (mg soil P/kg per kg applied P/ha) appeared to vary with soil type (or soil properties) and, perhaps, cropping frequency. Relatively minor changes in soil P status were observed due to different tillage practices. In developing P fertiliser budgets, we conclude that a major knowledge gap exists for estimating the residual effectiveness of P fertiliser applied to diverse soil types under a wide range of South Australian farming systems.

The accumulation of phosphorus applied as superphosphate over a 25-year period to pastures growing on sand over clay soils was measured in three areas of different rainfall in the south-east of South Australia. Total soil phosphorus levels increased in the 0-10 cm, 10-30 cm and 30 cm-clay layers for all areas, but no accumulation occurred in the clay layer. For every 100 kg ha-1 of single superphosphate (9.6% phosphorus) applied, 2.0, 3.2 and 3.5 kg ha-1 of phosphorus accumulated in the 0-10 cm layer for the Hundreds of Coles, Willalooka and Senior respectively. The recoveries of applied phosphorus in the sand layer for the three areas were 42, 57 and 75% for Coles, Willalooka and Senior respectively. The annual accumulation rates of organic phosphorus in the top 30 cm of soil ranged from 2.4 to 3.5 kg ha-1. By using a 'balance sheet' approach, the annual maintenance phosphate requirements for the three areas were calculated. These ranged from 60 to 165 kg ha-1 of superphosphate, and were associated with differences in leaching losses, stocking rates and organic phosphorus accumulation.

The retention of phosphorus applied as superphosphate was measured in 31 multi-rate experiments on siliceous sands in the south-east of South Australia. The amount of phosphorus retained in the top 30 cm of soil and recovered in harvested pasture 5 months after application ranged from 0 to 100% and averaged 57% for an average dressing of 32 kg phosphorus ha-1. Of the eight soil parameters measured, 'active' iron, 'active' aluminium, organic carbon and phosphorus sorption were most closely correlated with phosphorus retention. Combinations of these variables accounted for far more of the variation than did single factors. 'Active' iron and organic carbon accounted for 53.6% of the variation. The inclusion of quadratic 'active' iron and aluminium terms improved the fit with 63.9% of the variation accounted for. Rainfall during the experiments varied from 168 to 484 mm but had no effect on retention. Some agronomic implications of the leaching losses are discussed. It is considered that a measure of 'active' iron and organic carbon could be used as indicators of leaching and could assist in determining phosphorus recommendations in the region.

No abstract

No abstract