Summary

This project demonstrated that grain protein maps from the previous season can help guide more accurate nitrogen (N) fertiliser decisions within paddocks. Across three of four paddocks, historic protein was a more reliable indicator of soil available N than grain yield or N removal. Areas with higher protein generally required less applied N to maximise returns, while low-protein zones were more responsive to fertiliser. Optimal N rates varied widely within paddocks, ranging from zero to 129kg/ha of N. The findings suggest protein mapping can support variable rate N strategies, improve fertiliser efficiency and reduce unnecessary input costs. However, further research is needed across a broader range of paddocks and seasonal conditions to validate and refine this approach.

Background

Nitrogen fertiliser is the highest variable input cost for many grain growers, yet N deficiency remains a major cause of unrealised yield potential. Growers often apply conservative N rates due to uncertainty around rainfall, haying-off risk and difficulty predicting in-season demand. Traditional soil testing provides only a snapshot of soil mineral N and often fails to represent the high spatial variability found within paddocks. The increasing availability of harvester-mounted grain protein analysers has created an opportunity to use historic protein data as an additional decision-support layer for N management.

Research Aims

The core objectives of the project were to:

• Examine the relationship between pre-seeding soil mineral N to grain yield and protein maps from the previous season
• Assess how historical grain yield and protein maps explain spatial variability in N response across paddocks
• Explore whether protein maps can be used to develop variable rate N fertiliser prescriptions.

In The Field

The project was conducted over the 2023 and 2024 seasons in four paddocks at Bute and Redhill. Two collaborating growers provided historic grain yield and protein maps, where data were generated using standard yield monitors and CropScan grain analysers. Within each paddock, four zones were selected to represent different combinations of yield and protein.

Small plot N response trials were established within each zone, with urea applied post-emergent at rates from 0 to 200kg/ha of N. Soil mineral N was measured before seeding, and grain yield and protein were measured at harvest. These trials allowed N responses to be compared directly with historic protein and yield data.

Results

Grain protein in the previous season was a more reliable indicator of soil available N across sites and seasons than grain yield or N removal (yield x protein). On average, each one per cent increase in grain protein from the previous season corresponded to an increase of approximately 8.5kg/ha of soil N measured in the following year pre-seeding.

Protein data also helped explain large differences in fertiliser responses within paddocks. Across three paddocks and 12 trial sites, the N rate that maximised partial gross margin ranged from 0 to 129kg/ha of N, with an average of 76kg/ha. Zones with higher historic protein generally required lower N rates to achieve maximum returns.

At Redhill, each one per cent increase in historic protein reduced the optimal N rate by approximately 16kg/ha. At Bute, the response varied by season, with reductions of around 15kg/ha per one per cent protein increase in 2024. In 2023 there was a steeper response driven by one highly responsive site, with a one per cent protein increase reducing optimal N rate by 42kg/ha. In one moisture-stressed paddock in 2024, yield responses to N were negative, highlighting the influence of seasonal conditions.

Sites varied in the absolute N requirement for a given historical protein. For a historical protein level of 10 per cent, the N rate required to maximise partial gross margin (PGM) for Redhill in 2023 was 129kg N/ha, and for Bute was 68kg N/ha in 2023 and 51kg N/ha in 2024. Protein data was shown to support management of within paddock variability and allocation of N within the paddock. However, complementary data and decision-making tools are still required to determine optimal rates between paddocks, from which the protein map can then redistribute according to expected relative response.

The project also demonstrated how protein maps could be converted into variable rate N prescription maps. In a case study paddock at Bute, recommended N rates ranged from 0 to 110kg/ha, reflecting strong spatial variability in protein and N demand.