Take home messages

• A methodology called the P sufficiency index (pHnNDVI) has been developed for combining soil pH and NDVI data layers, and generating P fertiliser prescription maps for variable rate seeders and spreaders.
• Across 57 P fertiliser response trials conducted from 2019–2024, the optimal P rate to maximise partial gross margin ranged from 0 up to 50kg P/ha.
• Among different long-term P management strategies trialled, increases in DGT-P levels pre-seeding in 2024 generally only occurred where high rates of P fertiliser (50 or 90kg P/ha) had been applied repeatedly or the year prior to soil sampling.
• All three long-term sites were responsive to high P fertiliser rates. In four out of six trial years, an application rate of at least 50kg P/ha was required to maximise grain yield, and, for the remaining two trial years, grain yields were still increasing at the top rate of 90kg P/ha.
• Residual P available in the year following fertiliser application continued to increase grain yields in four out of six trial years, but generally only at rates greater than 50kg P/ha. This highlights, for highly P responsive soils, current district practice application rates of 10–20kg P/ha are unlikely to provide any useful residual P from the season prior.

Background

Fertiliser inputs are the single largest variable cost for grain growers producing a crop. The variability in rainfall experienced by growers, coupled with high fertiliser prices, has resulted in conservative fertiliser management.Consequently, P deficiency still causes yield losses in many environments and soil types across SA. This paper focuses on the key results from six seasons of field research under the following management areas:

• improving P fertiliser applications on variable soil types – the use of spatial data layers to provide information at the site-specific level and aid P fertiliser decision making.
• refining long-term P fertiliser management strategies on highly P responsive soils – evaluating four key management strategies on crop production (biomass and yield) and residual soil available P.

Soil type driving P fertiliser

The use of pH mapping has become common practice to identify areas within a paddock of low pH to improve lime application efficiency. While generating pH maps and comparing them with satellite NDVI imagery, it was observed that high pH areas on the map correlated with low crop vigour and P deficiency in many instances (Trengove et al. 2019, Mason et al. 2022). This finding resulted in the development of the P sufficiency index, which can provide information at the site-specific level and aid P fertiliser decision making. The P sufficiency index has been given the acronym pHnNDVI, as it is the soil pH value divided by NDVI normalised to the paddock average using the formula:

pHnNDVI = soil pH / (NDVI/paddock NDVI average)

Areas of a paddock with high soil pH (>7.5) and low relative normalised NDVI (<0.9) result in a high pHnNDVI value and are likely to be highly responsive to applied P (Figure 1). Areas with lower pH (<6.5) and high relative NDVI (>1.1) result in a low pHnNDVI value and are likely to be unresponsive to applied P.

Figure 1. Predicted P fertiliser response from the combined pH and NDVI data layers.

Field evaluation of P sufficiency index

In paddocks with substantial spatial variation, the P sufficiency index can accurately predict areas of low, medium and high P response in the Mid-North and Yorke Peninsula. More recently, this method has also been tested in areas of the Mallee and the Eyre Peninsula. To date, a total of 57 P fertiliser response trials have been monitored, providing a robust database to assess the capabilities of the pHnNDVI methodology (Figure 2).

Across six trial seasons, there was a strong in-season biomass response (measured by Greenseeker NDVI) to higher rates of P with increasing pHnNDVI (Figure 2). The P rate to achieve maximum biomass and pHnNDVI relationships were stronger than the yield response. This can be attributed to the fact that NDVI is assessed earlier in the season and is less likely to be influenced by as many factors as grain yield. As expected, at some of the sites, there was no yield gain from the application of any P fertiliser where pHnNDVI was low (Figure 2). In contrast, at other sites, the calculated optimum P level to achieve maximum partial gross margin was up to 50kg P/ha. This demonstrates the wide range of environments (e.g. soil types and yield potential) where the methodology was tested.

Figure 2. Relationship between pHnNDVI and P fertiliser rate (kg/ha) at 95% maximum NDVI for 57 sites (left) and maximum PGM for grain yield (right) for 47* sites across the Mid North, Yorke Peninsula and Eyre Peninsula. Partial gross margin has been calculated using the following pricing: MAP $1 200/t, wheat $350/t and barley $300/t.

*Trials that yielded poorly due to stress (e.g. soil disease, other confirmed nutrient deficiency, herbicide residues or frost) have been removed, as they do not adequately compare P fertiliser rates.

P fertiliser management strategies

At the beginning of 2021, three highly P responsive sites were identified using the P sufficiency index methodology (Table 1). These were soil sampled (0–10cm) pre-seeding in 2021. Soil pH levels ranged from 7.7–7.9 pH CaCl2, which is categorised as moderately alkaline. The DGT-P values were low ranging from 18–23µg/L (critical limit for wheat 60µg/L). Full comprehensive soil test analysis was conducted for each site and no other nutritional constraints were identified. Pre-seeding in 2024, all plots were soil sampled (0–10cm) and analysed for DGT-P and Colwell P. The crop sown at each location/year were chosen based on the hosting growers’ rotation (Table 1).

Table 1: Average growing season rainfall (GSR = April–October), soil properties and crop sown for long-term P response sites.

The long-term P fertiliser trials sown at all three sites aimed to assess four main management strategies (Table 2). Phosphorus fertiliser was applied as MAP and N was balanced at seeding with urea, to match the amount of N in the 90kg P/ha treatment. In the main treatments, the fertiliser (MAP and urea) was applied below the seed using a knife point press wheel system on 250mm row spacing.

Table 2: Treatment list showing units of P (kg P/ha) applied, the equivalent rate applied as MAP fertiliser (kg/ha) and cumulative P rate for the long-term P response trials in the Mid-North, SA.

*75kg P/ha spread prior to sowing as MAP + 15kg P/ha banded as MAP in 2021 and 2024.

**78 and 75kg P/ha spread as CL prior to sowing + 15kg P/ha banded as MAP in 2021 and 2024, respectively.

Residual soil P

The P use efficiency (PUE) of fertilisers was generally low in the year of application, ranging from 2–26% in this trial series, however, it continued to provide P to crops for several years. All three trials were soil sampled pre-seeding in 2024, (following three trial seasons) to understand if the various P management strategies had built up or mined soil available P compared to year one.

At Hart, all DGT-P values remained below the critical limit (60µg/L). There was a greater range and higher number of treatments above the critical DGT-P at both Crystal Brook and Spalding (Figures 3 and 4). Among the three trial sites, Hart had the highest PBI and pH (stronger ability to bind added fertiliser P), which most likely contributed to the lower P availability and lower variation in DGT-P values at this site.

Among all the strategies trialled, the only P rates that had an impact on starting DGT-P were generally where high rates of P fertiliser had been applied repeatedly each year (Figure 3) or in year three only, prior to testing in year 4 (Figure 4). This shows a portion of the fertiliser P applied in these high rates (last season or cumulatively) had carried over in plant available forms and should be available to the subsequent crop. It also highlights P fertiliser rates of <50kg P/ha applied repeatedly or in a single season, were not sufficient to increase DGT-P the following season on P fixing soils.

Figure 3. Pre-seeding 2024 DGT-P following three seasons (2021–2023) of repeated applications of P fertiliser rates ranging from 0–90kg P/ha for Hart (R2=0.78), Spalding (R2=0.94) and Crystal Brook (R2=0.96). Treatments 3 and 10–15 in Table 1.

Figure 4. Pre-seeding 2024 DGT-P following once off applications of P fertiliser rates ranging from 0–90kg P/ha the year prior (2023) to sampling at Hart (R2=0.89), Spalding (R2=0.93) and Crystal Brook (R2=0.79). Treatments 3 and 16–21 in Table 1.

Crop responses

Year of application

All three sites were responsive to high rates of P in the year of application. In four out of six trial years, an application rate of 50kg P/ha was required to maximise grain yield. In the remaining two trial years, grain yields were still increasing at the top rate of 90kg P/ha. This demonstrates that district practice rates of 10–20kg P/ha are not meeting the crop demand on these soil types.

As an example, at Hart, the average grain yield increase was 138% of the nil applied in year one (range 129–147%) from the application of 90kg P/ha (Figure 5). These responses to applied P fertiliser were in line with the ‘high’ predicted P response based on the P sufficiency index (pHnNDVI) for each site.

Residual response

The residual effect of P fertiliser was assessed in year two and year three, where district practice applications (15kg P/ha) followed the range of rates (0–90kg P/ha) in the 2021 or 2023 season. The residual responses from P applications were variable across sites and seasons. In four out of six trial years, there was evidence of residual P responses in year two. However, similar to the soil test results, these responses only occurred from high application rates (generally greater or equal to 50kg P/ha) (Figure 5).

The two seasons where a grain yield response to residual P was not observed can be attributed to the dry conditions in 2024 (below decile 1 rainfall). These two trials were sown to lentils and grain yields averaged 0.65t/ha. The low yield potential likely masked any possible residual P response.

To date, this research has only evaluated the third-year residual response to P fertiliser rates in three trials (Figure 5). Two of the three trials showed residual P fertiliser responses in year three. The application rate of 90kg P/ha was required to have any carryover effect on grain yield in year three at Hart (111% of nil applied). At Crystal Brook, there was a third-year residual response to 50kg P/ha however, the 90kg P/ha was not different to the untreated. At Spalding there was no response in year three to P rates applied in year one, indicating no legacy effect of the high P rates carrying into the third season. While the results show there have been some residual P responses to P rates applied in year one, the outcomes were not consistent and suggest the application rates required are well above current district practice rates.

Figure 5. Relative grain yield in the year of application and residual response in year two and three following various P rates (0–90kg P/ha) for Hart (left), Crystal Brook (middle) and Spalding (right). Lines ending with symbols refer to grain yields following P rates applied in 2021 (circle) and 2023 (triangle) and their residual season responses.

Implications for P management

It is common for growers in the southern region to use a P replacement strategy based on the amount of P removed in the grain (that is, 3kg P/t cereal grain) to determine fertiliser P application rates. Using this strategy, ‘district practice’ P fertiliser rates are generally in the range of 10–20kg P/ha per annum. Given below average grain yields last season, it would be fair to assume <5kg P/ha has been exported in the grain in many areas. The P replacement strategy would therefore assume a reduction in P fertiliser rates going into this season. Using the field trials above, we explore the question – can we cut back to 5kg P/ha as replacement this season?

This research has shown at district practice P fertiliser application rates (<20kg P/ha), a grower cannot rely on residual P from the season prior if the zone or paddock is P responsive with a moderate PBI (range 77–110 at these sites). In our trials repeated applications of >20kg P/ha were required to shift pre-seeding DGT-P soil levels enough to have any implications on crop growth and grain yield.

The results from the various field sites (Figure 5) showed a yield response to residual P, when high rates were used in year one and then returning to ‘district practice’ in year two and year three. However, these graphs do not show how much economic benefit is lost by not applying the optimum P rate or continuing with repeated high fertiliser rates. The graphs above, in fact, are a demonstration of what not to do on P responsive soils. Reducing P fertiliser rates coming into 2025 will limit the yield potential of this season’s crop (year one response), and the yield potential of the subsequent crop may also be limited (year two response), even when ‘district practice’ rates are reapplied in future years.

Conversely, P fertiliser management for non-responsive zones or paddocks requires a different approach. For these areas, there is great value in residual fertiliser P from previous applications. In some cases, soils are not responsive to P at all, and it is rare that they respond to greater than replacement levels. The pHnNDVI methodology can help to identify where these areas are, and it can be used to make considerable savings on P fertiliser application on these soil types.

Acknowledgements

The research undertaken as part of this project is made possible by the contributions of growers through both trial cooperation, and the authors would like to thank them for their continued support. We would also like to thank the Wundke, Stephenson and Sargent families through trial cooperation for long term trials. Other growers involved in the projects include Bill Trengove, Leigh Fuller, James Venning, Joe and Jess Koch, Scott Weckert, Kenton Angel, Tristan Baldock, Andrew Thomas and David Cooper. This program is co-funded by SAGIT (project code TCO 06024, TC221 and TC219) and GRDC. We would also like to acknowledge the previous SAGIT investments (TC221 and TC219) to establish and monitor the single year and long-term field trials.

References

Mason S, Trengove S, Sherriff S, Bruce J (2022) An informed approach to phosphorus management in 2022. Proceedings GRDC Grains Research Update, Adelaide, February 2022,

Trengove S, Sherriff S, Bruce J (2019) Improved phosphorus prescription maps – beyond phosphorus replacement. Hart Trial Results 2019, pp. 72–78.

Contact details

Sam Trengove
0428 262 057
samtrenny34@hotmail.com