Take home message

• Queensland grain growers identified managing nitrogen and declining soil fertility levels as their ‘trickiest’ (riskiest) decisions
• Crop responses to the application of nitrogen and deep phosphorus are variable
• Trials showed an increase in crop growth early in the season under deep phosphorus with increases in nutrient uptake
• Yield responses varied. An increase in yield of ~500 kg/ha was recorded on a soil that had very low levels of subsurface P when both additional phosphorus and nitrogen was applied
• Paddock variability drives crop responses. Understanding paddock variability can help inform research outcomes and may assist in improved management
• Large strip trials utilising proximal sensing and precision technology are helping understand crop responses encompassing paddock variability.

Background

RiskWi$e (the National Risk Management Initiative) is a 5-year national initiative aimed at helping growers understand and improve the risk-reward outcomes when making decisions in the face of uncertainty. To deliver RiskWi$e, a participatory action research methodology (an approach to research that pro-actively involves members of communities affected by that research in the research itself) has been employed. In Queensland the project is centred around Capella, Gindie, Roma, Brigalow, Meandarra, St George and Goondiwindi. Issue identification sessions were conducted for all groups, during which we explored participants ‘trickiest/riskiest’ decisions. Managing nitrogen (N) and declining soil fertility levels (N, phosphorus (P) and potassium (K)) were identified as being the riskiest. As a result, 30+ growers, with support from the project team have implemented on-farm trials investigating nitrogen management, declining soil fertility and managing soil constraints. This paper will explore three of these trials, situated at Wallumbilla, Surat and St Geroge.

Trial design and implementation

Treatments, trial designs and logistics were negotiated one-on-one with each of the growers and their agronomist. This type of research is not a one size fits all, rather purpose built for each grower, their research question, paddock, machinery and skill set. All paddocks were mapped using the online tool ConstraintID to quickly, easily and cheaply identify paddock variability to inform trial location. Paddocks were soil sampled, and trials implemented in 2023. The N treatment strips were implemented using grower machinery, with the design allowing the grower to plant and harvest. Deep P (& K) was applied at right angles to the N strips in two of the trials (Wallumbilla and Surat) in a strip plot design. At Wallumbilla deep P was applied using experimental machinery due to limited capacity of grower machinery, whereas it was applied by the grower at Surat. In contrast, the trial at St George was a split plot design with all treatments running parallel down the paddock (12 m wide “deep” treatments in 36 m “N” strips), all implemented by the grower. “Deep” treatments were ripped in at ~20 cm. There were no “rip only” control treatments as deep placed fertiliser logistically cannot be applied unless ripped in so any potential response to ripping is inherently part of the overall “deep” treatment response (Table 1). However, this interaction is being investigated in the small plot research of the GRDC NGN Deep P in SW Queensland (DAQ2404-008BGX). All treatments were replicated twice. It is envisaged that these trials will have N treatments reapplied each season for the remainder of the project, approximately four years.

Research questions

These trials are investigating the following questions:

• What are the comparative efficiencies and legacies of different nitrogen strategies (including rate and timing of application)? What are the risks and rewards of these approaches? What are the impacts on N losses, movement of N in the soil profile, efficiency of crop N use, yield and gross margin?
• Is the application of deep P (and K) effective and viable for increasing yield and profit compared across a range of nitrogen strategies? What are the risks and rewards and impacts on yield and cumulative gross margin?

Treatments

Table 1 outlines the treatment list for each site.

Table 1. Outline of nitrogen (N), deep phosphorus (P) and potassium (K) treatments at each of the three Queensland sites.
Yield potential (YP) is calculated annually on modelled yield.

Treatment name	Treatment description	Wallumbilla	Surat	St George
Nitrogen treatments
Nil	No fertiliser
Early N   50% YP	N budget targeting 50% yield potential, applied early in the fallow
Early N   75% YP	N budget targeting 75% yield potential applied early in the fallow
Early N   100% YP	N budget targeting 100% yield potential applied in the fallow
Late N 50% YP	N budget 50% yield potential applied at planting
Late N 75% YP	N budget 75% yield potential applied at planting
Deep treatments
Deep P	Deep P (150 kg/ha Granulock® Z; 30 kg P/ha) ripped in at a depth of ~20 cm
Deep P + K	Deep P + K (150 kg/ha Granulock® Z; 30 kg P/ha) plus (100 kg/ha Muriate of Potash; 50 kg K/ha) ripped in at a depth of ~20 cm

Rates of nitrogen fertiliser are calculated annually via N budgeting based on modelled yield potentials in line with treatment rules and agreed to by the grower (Table 2). Due to sufficient soil N, additional N was not required on some treatments (note: only treatments with additional applied N will be presented). Nitrogen is not applied to pulse crops. All trials were planted to wheat in 2024. Approximate costs were calculated for each treatment (Tables 2 & 3).

Table 2. Target yield potential (YP) and quantity and cost of N fertiliser applied to wheat in 2024.

Table 3. Cost of Deep P and K application and yield improvement to break even per hectare for wheat grown in 2024.

Data collection

The following data is collected for specific treatments for growers’ crop rotations over the life of the trial:

• Soil mineral N and water at planting and harvest
• Biomass cuts during vegetative growth, at anthesis and physiological maturity
• Maturity biomass cuts are processed to determine grain yield, protein, quality and nutrient concentration
• Grain yield and protein from grower harvester yield & protein monitor
• Rainfall data
• Satellite imagery (NDVI).

Satellite imagery from the PlanetScope constellation was used to monitor crop growth and responses throughout the growing season. Vegetation indices, typically the normalised vegetation index (NDVI), measuring crop greenness, was used as an indicator of crop biomass and its distribution and variation across the trial. A key benefit remote sensing is the ability to capture transient patterns in the crop which may not be present when physical measurements are collected.

Electromagnetic (EM) surveys were used determine the apparent soil electrical conductivity (EC_a). Several soil properties contribute to the EC_a of a soil, including soil water, texture (especially the clay fraction) and ions in the soil (mainly sodium). Mapping EC_a is a useful tool to identify changes in these soil properties and thus identify differing soil types within a paddock. This allows better interrogation of treatment responses, which may be affected by soil type.

Collection and analysis of harvester yield monitor data was used to assess treatment responses in this project and trial design was used to enable this methodology. Treatment strips run in line with traffic were applied at least one header width wide. Treatment strips running perpendicular to the traffic direction were wide enough to allow sufficient grain to allow yield monitoring to register discrete samples for each treatment, working within the limitations of the harvester size, processing and yield monitor systems. One header was used to harvest the trial area to avoid calibration issues between different headers. Raw data from the yield map was cleaned and analysed using statistical methods supported by AAGI, the GRDC’s Analytics for the Australian Grains Industry- however statistical results were not finalised at the time of writing this paper and as such all ‘results’ should be treated with caution. Headers allowed us to measure the entire plot in the trial (some are over 2 km long), as opposed to small individual locations when taking hand samples in the crop. These datasets provide insight into crop responses across the entire site and other factors which may be influencing plant growth.

Soil characterisation

The Wallumbilla, Surat and St. George sites were surveyed and characterised prior to implementing treatments (Table 4). The St George site had two distinct soil types in terms of clay content; hence they were soil sampled separately (Figure 1). Wallumbilla and Surat had very low levels of subsurface P while the St George site had acceptable levels. Current knowledge of critical limits suggests none of the sites are critically low in subsurface K.

Table 4. Soil characterisation for trial sites in 2023.

Figure 1. St George trial site paddock (left) with soil types and treatments overlayed onto trial area (right).

Biomass – weight, nutrient content and NDVI

A combination of methods were deployed to monitor responses across all trials during the growing season. This was particularly useful, as satellite imagery assisted tracking transient crop responses whilst hand cuts provided quantification of biomass production and nutrient uptake.

Vegetative biomass and nutrient uptake

Biomass was sampled at the Wallumbilla and Surat trials at the vegetative stage (~8 weeks after planting) to assess the impact of N status, with and without deep P and K. At the Wallumbilla site, no responses to either N timing or application rate were apparent (Table 5). At the Surat site, there appeared to be more biomass, N and P uptake in the 100% YP treatment compared to the Nil treatment (Table 6). At this stage of growth, responses to deep P were evident at both sites, producing more biomass and increasing P uptake relative to the Nil treatment (Tables 5 and 6). The application of deep K also increased K uptake at the Surat site (Table 6).

Table 5. Wheat biomass and nutrient uptake at vegetative growth state at Wallumbilla in 2024.

Anthesis biomass and nutrient uptake

Biomass was sampled again at the Wallumbilla and Surat trials at anthesis to assess the impact of N status, with and without deep P and K.

Responses to N at the Wallumbilla site were variable, with no clear trends apparent regarding effects of either N application rate, or timing on biomass or N uptake (Table 7). At the Surat site, biomass responses to N application were less apparent at anthesis compared to vegetative growth stage, but the 100%YP treatment appeared to have a higher N uptake than the Nil. (Table 8).

At the Wallumbilla site, a response to deep P appears to be present at anthesis with respect to both biomass and P uptake (Table 7). Across all N treatments there is a trend observed around increased shoot biomass and increased N uptake with the application of deep P. In contrast, at the Surat site, the P response which appeared to be evident early in the season had diminished by anthesis (Table 8).

Maturity biomass and nutrient uptake

Biomass was sampled at maturity at all sites to assess the impact of nitrogen status, with and without deep P and K.

Finalising data from the laboratory is still in progress. However, preliminary data suggests that the trends apparent at anthesis followed through to maturity at all sites.

Table 6. Wheat biomass and nutrient uptake at vegetative growth state at Surat in 2024.

Table 7. Wheat biomass and nutrient uptake at anthesis at Wallumbilla in 2024.

Table 8. Wheat biomass and nutrient uptake at anthesis at Surat in 2024.