Summary

Phosphorus rates required to optimise partial gross margins vary widely across paddocks according to soil type and production zones.

In trials across five paddocks in the Mid North and northern Yorke Peninsula regions, optimal rates were shown to vary from 0 kilograms per hectare to 40kg/ha.

Combining spatial data layers such as soil pH and historical satellite imagery in advance of seeding provided a good prediction of P response. A methodology for using data layers to create variable rate P prescriptions has been developed.

Background

Phosphorus fertiliser inputs are one of the largest variable input costs for grain farmers in SA. Spatial map layers such as grain yield, soil pH, soil EC and NDVI can infer useful information about potential P uptake, soil tie-up and P response.

It should be possible to use these data layers to develop variable prescription maps for P application to increase P fertiliser use efficiency and minimise over-application. However, the best methodology for integrating these data layers to produce variable rate P prescription maps has not been established.

Research Aims

The core objectives of the project were to:

• Increase the profitability derived from P fertiliser application.
• Better understand how spatial data layers can be integrated to produce variable rate prescription maps that optimise P rates across variable paddocks.

In The Field

Data layers for yield, soil pH, soil EC and NDVI were collated for five paddocks near Bute, Koolunga, Brinkworth, and Kybunga. Using computer software to combine data layers, areas of high and low yield, pH and NDVI were identified.

This provided 21 targeted response trial sites across these paddocks over the two seasons. Treatments of P at rates of 0, 5, 10, 20, 30 and 50 kilograms per hectare were applied as MAP for comparison with the P responses predicted by the data maps. Nitrogen rates were matched between the treatments using adjusted rates of urea.

Sixteen additional sites at the same locations were treated with chicken litter (5t/ha) or biosolids (5.0t/ha) to determine the P response from these products.

The relationship between the observed P response at the 21 trial sites and the predicted response was then used to develop an algorithm using the data layers to calculate optimal P rates and P prescription maps.

In season biomass was assessed in July and August/September using NDVI and biomass samples. Youngest emerged blade (YEB) samples were sent for tissue analysis to assess nutrient concentrations.

Grain yield data was collected using a plot harvester and grain samples were sent for grain nutrient concentration testing. Soil samples were taken and tested from a selection of responsive treatments.

Results

Combining soil pH maps with historical NDVI imagery was found to provide a good prediction of P response. This validated the premise that areas with high soil pH and low cereal crop growth would be responsive to higher P rates, whereas areas with lower soil pH and high growth rates would not be likely to require additional P.

It was found that targeting P to areas where a good response was likely delivered an average increase in partial gross margin of $41 per hectare. In addition, P fertiliser cost savings could be realised by identifying areas where lower to no P fertiliser could be applied without a yield penalty.