DEVELOPMENTS IN SOME ASPECTS OF REACTIVE PHOSPHATE ROCK RESEARCH AND USE IN NEW-ZEALAND

There has been over 50 years of use and research into the agronomic ef fectiveness of reactive phosphate rocks (RPR) directly applied to New Zealand pastures. In recent years RPR-carrying fertilisers made up abo ut 16% of phosphatic fertiliser sales in the North Island of New Zeala nd. Most is applied, as maintenance fertiliser, to hill country sheep and beef farms. Use has been recommended on soils with pH <6 and in an nual rainfall regimes >800 mm. This is based on the poor performance o f Sechura phosphate rock in summer dry areas receiving <750 mm of rain fall annually. Phosphate rocks that have more than 30% of their total phosphate soluble in 2% citric acid have been classed as 'reactive' an d suitable for direct application. More recent research indicates that extraction with 2% formic acid, or a dissolution test performed in a simulated soil solution at a fixed pH, will provide improved measures of RPR quality. Field trials, undertaken by the New Zealand Ministry o f Agriculture and Fisheries [MAF; now AgResearch Crown Research Instit ute (CRI)] and others, to evaluate the relative agronomic effectivenes s of RPR versus soluble P fertilisers in adequate to marginally P-defi cient soils have proven to be a painstaking task. Long periods (3-6 ye ars) of fertiliser withdrawal were required for pasture growth on some soils to become significantly responsive to applied P. Only then did differences between P sources become significant. This problem has enc ouraged efforts to relate measurements of the extent of RPR dissolutio n in soils to their agronomic effectiveness. Three main modelling appr oaches have been used to achieve this objective: Kirk and Nye (1986a, 1986b, 1986c); Sinclair et al. (1993a); and Watkinson (1994b). These m odels are reviewed and their explanation of RPR dissolution in mowing trials tested. Components of each model have then been combined to pro duce models to predict the agronomic effectiveness of RPR. The develop ment of P tests for soils receiving RPR-containing fertilisers is revi ewed. Separate Olsen P test-yield response calibration curves are requ ired for soils fertilised with soluble P fertilisers and soils fertili sed with sparingly soluble P sources or soluble P in the presence of h eavy lime applications. Whereas alkaline P tests such as Olsen or Colw ell underestimate the amount of plant-available P in these soils, acid P tests such as Bray I are likely to overestimate the available P. Te sts involving cation and anion exchange resin membranes appear to be m ore appropriate for soils with unknown histories of soluble P and RPR use and may permit the use of single calibration curves. Trends observ ed in Olsen P soil test values, from farms on the North Island of New Zealand that have a history (3-15 years) of RPR use are presented. A p redictive dissolution model is used to explain these trends but it is evident that spatial and temporal variation in soil test results on fa rmers' paddocks will be a major constraint to the precision to which t his or similar models may be used. The model, however, may provide the basis for sound advice on the strategic use of RPR for direct applica tion to New Zealand pasture soils. It may prove useful in explaining t he variation in RPR effectiveness in a wider range of climates and soi ls.