The P story so far – an update on deep P research findings

Author: Mike Bell (University of Queensland), David Lester and Doug Sands (DAF), Rick Graham and Graeme Schwenke (NSW DPI) | Date: 02 Mar 2018

Call to action/take home messages

Depletion of subsoil phosphorus (P) reserves and uncertainty over the benefits of starter P use led to a detailed examination of crop responses to P fertilisers in terms of rates, products and placement over the last 5-6 years.

Trials have been conducted at 35 locations from central NSW to CQld. While large crop responses have been recorded at many locations, the impact of seasonal moisture availability (on yield potential and root activity), P placement (starter v deep P, band spacing, amount of soil disturbance), P rate, time of application and product choice will all impact on responsiveness.

Residual availability of deep P has generally been excellent, although crop response may be limited by shortages of other nutrients (e.g. N).nData continues to be incorporated into the Deep P calculator to provide an indication of the economic returns from P investment. To access the Deep-P Calculator, click here.

The trial program

There has been a staggered establishment of sites across the region since mid 2012, with a total of 30 experiments in UQ00063 by the 2016 winter crop season at locations indicated in Fig. 1. These trials consisted of rates of phosphorus (P) fertiliser (0 to 40 or 60 kg P/ha) applied in deep bands (at ~20cm depth), typically at band spacings of 50cm, along with an untilled Farmer Reference treatment. All main plots were then split to annual ‘with’ or ‘without’ starter P fertiliser applications at planting at rates ranging from 6 to 10 kg P/ha.

Crop choice at each site was dependant on the crop in the surrounding paddock (e.g. crop mix in the establishment years are shown in Table 1), and the residual benefit of the different rates of applied P was tracked through subsequent growing seasons. After the 2017 winter crop harvest, some sites have produced 4 or 5 crops, while some of the newer sites were only in the first or second crop after the initial P application. An extension to UQ00063 will allow the residual value of these later sites in particular to be extended to at least 4 crop seasons, weather permitting, while some of the earlier sites will be in their 5th to 7th crop after P application.

At each site, efforts were made to initially address other potential nutrient limitations, identified by soil testing, by applying a basal application of N, K, S and/or Zn as appropriate. Additional N applications were needed to balance different N additions when the P rates were applied as MAP, but also in response to a likely increase in crop yield potential after overcoming the P constraints. Whilst this initial ‘top up’ ensured crop responses to applied P should be expressed, problems emerged in later years at some responsive sites, as the higher yield potentials in the plots where P deficiency had been overcome did not have enough N to deliver the higher yield potentials (i.e. P responses were constrained by lack of N).

While mainly relevant to grain crops and not pulses such as chickpeas or mungbeans, this problem was often not evident until late in a growing season, or only from observations of sub-optimal grain protein content. That meant that remedial action could only be taken for the next crop in the cycle, and assessment of the residual P benefit may have been compromised in some crop seasons. An example of this is provided later in this paper.

 Figure 1 is a map of NSW and Qld showing the location of the 30 P trials established from 2012-2017 under UQ00063.

Figure 1. Location of the 30 P trials established from 2012-2017 under UQ00063.

The soil P tests for the 30 P trial sites in UQ00063 are shown in Table 1. Both Colwell and BSES P were higher in the 0-10cm layer than the 10-30cm layer, and BSES P was always at higher concentrations than Colwell P. Based on wheat critical soil test values for starter P responses, the median values would suggest a significant proportion of sites should have been P responsive. Similarly, we had hypothesized deep P responsive sites would have < 10 mg/kg Colwell P and <100 mg/kg BSES P in the 10-30cm layer, so most sites were predicted to respond to deep P.

Table 1. Median, maximum and minimum concentrations of Colwell and BSES P at sites sown to winter cereals (wheat and barley), chickpeas, grain sorghum or other species (sunflower and cotton) in the first year after fertiliser application.

 

Colwell P (mg/kg)

BSES P (mg/kg)

 

0-10cm

10-30cm

0-10cm

10-30cm

Winter cereal sites (11)

Median

19

6

66

18

Range

74-10

22-3

234-22

84-12

Chickpea sites (9)

Median

17

4

29

12

Range

22-6

6-1

57-13

16-5

Sorghum sites (9)

Median

14

3

71

15

Range

30-5

6-1

96-8

74-3

Other species (2 - Sunflower, cotton)

Median

17

7

42

19

Range

19-15

9-4

47-37

30-8

An additional series of trials have been established since 2015 in core sites at Jandowae, Lundavra, Terry Hie Hie and Bellata under UQ00078, with these trials looking at placement strategies (rate*band spacing interactions, liquid v granular fertilisers, form of applied P, degree of soil disturbance/mixing, effect of co-location of different nutrients). Crop choice was again dependent on the host farm crop rotation.

Responses to starter P

Starter fertiliser applications were made at sowing – either by the trial operators with small plot equipment, or by the growers who turned starter fertiliser on and off in planned strip-plot designs. Unfortunately in the latter, starter was not always turned on and off as planned, so there was some loss of starter P contrasts at some trials. Overall, prior to the 2017 winter season (still being processed) there have been 42 site-years with starter P contrasts, split between wheat/barley (17), chickpeas (13) and sorghum (12), with crop responses assessed in relation to the initial pre-trial Colwell P concentration in the 0-10cm layer.

To cope with differing yields between sites and seasons, responses were assessed on the basis of relative yield (Yield no starter/Yield with starter), with values <1.0 indicating that the crop responded to starter P application. Responses for each crop species are shown in Fig. 2 (a, b, c) for winter cereals, chickpeas and sorghum, respectively.

Figure 2a is a scatter graph that shows response to starter P application in wheat and barley crops grown in the UQ00063 trials. Points indicate the relative yield of the plots without starter fertiliser, compared to those where starter P was applied.

Figure 2b is a scatter graph that shows the response to starter P application in sorghum crops grown in the UQ00063 trials. Points indicate the relative yield of the plots without starter fertiliser, compared to those where starter P was applied.

Figure 2c is a scatter graph which shows the response to starter P application in chickpea crops grown in the UQ00063 trials. Points indicate the relative yield of the plots without starter fertiliser, compared to those where starter P was applied.

Figure 2. Response to starter P application in winter cereals, chickpea and sorghum crops grown in the UQ00063 trials. Points indicate the relative yield of the plots without starter fertiliser, compared to those where starter P was applied.

There is a relationship that can be developed for the winter cereal crops, although it is poorly defined due to the limited number of sites with very low Colwell P in the 0-10cm. That said, there seemed to be no real evidence of responses to starter P for Colwell P > 15-17 mg/kg – a value somewhat lower than the 20-23 mg/kg suggested from the recent national database analysis (R. Bell et al. 2013). More data is needed to better define this relationship.

Despite some quite significant responses to starter P in chickpea (up to 20% yield increases), there is no clear relationship to Colwell P in the 0-10cm layer – possibly because the variable sowing depths between sites and seasons may mean the 0-10cm layer is not always relevant to the developing crop (i.e. it may have been planted below 10 cm!). The data for sorghum shows almost no response to starter P at all, even when Colwell P in the top 10cm layer is very low. The one site where a significant response was recorded was south of Emerald in 2015.

We had intended to undertake separate analyses to look at starter P responses in the presence or absence of moderate-high rates of deep P, to gain some understanding as to whether ameliorating low subsoil P could eliminate the need for starter P – an effective diversion of the current P inputs in the cropping system from smaller annual starter P applications to occasional larger deep P applications. However, given the limited responses to starter P across all crops this has so far been inconclusive.

Responses to deep P

Deep P applications were made using either mono-calcium phosphate (TSP), or more commonly mono ammonium phosphate (MAP) with equilibration of the N inputs for the different rates using urea. The responses to deep P (Figure 3) have been assessed relative to the nil P treatments that received tillage and background nutrient inputs, so any yield increases are directly attributable to P addition only. Comparisons between the deep P treatments and the ‘As is – Farmer Reference’ condition were also made, with the relative performance of the deep P treatments generally greater than when benchmarked against the nil P, but responses in these instances could be due to soil disturbance, other background nutrients or the deep P input. There were 19 winter cereal and 19 chickpea site years of data, and 17 for sorghum, prior to the 2017 winter season.

As indicated in Figure 3 below, all crops responded to deep P when subsoil P (in the 10-30cm layer) was low. There are a range of responses to deep P for any given soil P concentration, which were due to a combination of other yield constraints (e.g. a lack of water!) or an ability to access P from relatively enriched topsoil layers (i.e. wet years). As an example, ignoring the red point which represented a trial with low N, failing to apply deep P resulted in yield penalties of 10-25% for a Colwell P of 2 mg/kg.

It is difficult to make more definitive conclusions from the data at this stage (e.g. which species is the most responsive to deep P applications?), because the crops were grown on different sites, in different seasons with differing access to topsoil P, and with different deep P sources in some cases. For example, for sites where subsoil Colwell P was <5 mg/kg, the average yield loss due to low subsoil P from all site-years is 8% for wheat and chickpea crops but 13% for sorghum. It is impossible to tell whether this is because sorghum responds more to deep P than the other species, or because 90% of the sorghum crops were responding to deep P applied as MAP while for wheat and chickpea only 45-55% of the sites had MAP and the rest TSP.

Figure 3 is a scatter graph that shows the response to deep P application in winter cereals, chickpea and sorghum crops grown in the UQ00063 trials. Points indicate the relative yield of the plots with starter fertiliser but no deep P, compared to those where starter P and deep P was applied, and represent both fresh and residual P responses. Points in the upper LHS of the curve (red and green squares) are where deep P was applied too close to the planting date, or where the site was N limited.

Figure 3. Response to deep P application in winter cereals, chickpea and sorghum crops grown in the UQ00063 trials. Points indicate the relative yield of the plots with starter fertiliser but no deep P, compared to those where starter P and deep P was applied, and represent both fresh and residual P responses. Points in the upper LHS of the curve (red and green squares) are where deep P was applied too close to the planting date, or where the site was N limited.

If we used the lower (more responsive) boundary of the trial data combined for all crop types (dashed line in Fig. 3) as an indicator of the potential crop responses to deep P in favourable seasons, this analysis would suggest that 90% of maximum yield potential (the value normally used to define a critical soil test concentration) will be achieved when Colwell P in the 10-30cm layer is ~10 mg P/kg. This agrees quite well with the original predictions at the beginning of this project in 2012.

Optimum application rates and residual value of applied P

To explore these effects we use case studies from 2 sites – one at Dysart (deep applied in Aug 2013, crop 4 harvested in 2017 winter) and the other at Wondalli (deep P applied in May 2013 and crop 4 harvested in 2017). Both sites were responsive to deep P, and responses are still evident in the 4th crop after application.

It is difficult to tell whether the P response is diminishing over time, or whether the rate required to reach P-unlimited yields is changing, due to other variables like seasonal conditions and N availability. In the first 3 crop years of sorghum, 20 kg P/ha was enough to reach maximum yields, but the P response was increasing all the way to the top rate (40 kg P/ha) in the chickpea crop in season 4. While this would suggest that the residual benefits of the lower P rates were diminishing by the 4th crop, this ignores the fact that yields of sorghum (crop 3, and to a lesser extent 2) were constrained by N availability, so the full potential P response may not have been able to be expressed.

When a legume crop (chickpea) was grown in season 4, low N was less likely to be limiting crop performance, and the P response was very strong (and very profitable!). Further site-years with higher N inputs will help explore this response. However, to date the combination of tillage and deep P has produced an additional 700 kg/ha of sorghum in crop 1, 550 kg/ha of sorghum in crop 2, 450 kg/ha sorghum in year 3 and >850 kg/ha of chickpeas in year 4.

Figure 4a is a column graph that shows grain yields at the deep P site at Dysart for the first 4 crop seasons after deep P application in 2013. Deep P was applied as MAP at rates ranging from 0 to 40 (Dysart) or 60 (Wondalli) kg P/ha, and each site had an unripped Farmer Reference (local practice) treatment as a benchmark.

Figure 4b is a column graph that shows grain yields at the deep P site at Wondalli for the first 4 crop seasons after deep P application in 2013. Deep P was applied as MAP at rates ranging from 0 to 40 (Dysart) or 60 (Wondalli) kg P/ha, and each site had an unripped Farmer Reference (local practice) treatment as a benchmark.

Figure 4. Grain yields at deep P sites at Dysart and Wondalli for the first 4 crop seasons after deep P application in 2013. Deep P was applied as MAP at rates ranging from 0 to 40 (Dysart) or 60 (Wondalli) kg P/ha, and each site had an unripped Farmer Reference (local practice) treatment as a benchmark.

There was no evidence of N limitations at Wondalli, although this site did lose its chickpea crop in season 3 due to the wet conditions in 2016. At this site, yields increased with increasing P rate up to 60 kg P/ha in the initial sorghum and subsequent wheat crops. However, while there were significant P responses in the 2017 wheat crop, there was no yield increase to rates >20 kg P/ha. The very dry seasonal conditions and water-limited yields in this season may have limited any response to higher P rates, so again, further site-years are needed to determine how long the residual effects last. To date, the combination of tillage and deep P has produced an additional 880 kg/ha of sorghum in crop 1, 650 kg/ha of wheat in crop 2 and 400 kg/ha of wheat in year 4.

When do I re-apply?

As more growers and advisors start deploying deep P strategies in their fields, this question is asked more frequently. Unfortunately there are no easy answers, other than the use of test strips with accurate yield monitoring. The amount of time and effort to effectively sample the subsoil layer to account for residual fertiliser bands, and assess the P bioavailability, would suggest soil sampling will not be the answer. Even a budgeting approach will not provide reasonable estimates. This is because, experimentally, we are unable to identify whether additional P uptake or removal by the crop accurately reflects fertiliser P recovery, as most crops are capable of proliferating roots in a P band and so can preferentially take P from the band while sparing P from the surrounding soil. We do not have any tracers that we can put into a P band to indicate fertiliser P recovery over a series of crop seasons, although Phil Moody and the group at DSITI are evaluating some options in the lab at present.

We are currently in year 1 of a 3 year extension to UQ00063 where we will continue to monitor a number of the more recently established sites as well as some of the longer term ones, and that may provide better estimates of the likely length of residual P benefits.

Products, placement strategies, timing

This remains an area of active research in both Qld and NSW under the co-funded UQ00078 GRDC project. Research has so far showed that

  • There seems to be no advantage to be gained by using liquid v granular forms of MAP, as has been reported on the calcareous soils of the Eyre Peninsula in South Australia, and recent results from the NSWDPI site near Bellata have suggested the liquid MAP was actually inferior to the granular product.
  • Studies have not been able to demonstrate an interaction between P rate and P band spacing (i.e. in band concentration), so the important thing is to get enough P into the subsoil layers in a way that is practical but maximizes the chances of crop P recovery. Current recommendations for band spacings of 40-50cm therefore remain valid.
  • To date, there has been no evidence of benefits of co-locating P and Zn in deep P bands, as there were suggestions that improved P uptake may limit Zn acquisition – possibly through a less extensive mycorrhizal network. However these effects have only been observed in sorghum crops in UQ00063, and as yet the specific P-Zn trial sites have not hosted a sorghum crop.
  • The question of type of P fertiliser (MAP or TSP in particular) is currently the subject of trials sown in 2017 winter in NSW and in 2017/18 summer in Qld, and will also form a component of research conducted by a new postdoctoral appointment at UQ commencing in early 2018.
  • Trials have also shown that there are no obvious negative impacts of combining MAP and KCl in deep bands in sites where both P and K are limiting, and that in this case, the root proliferation associated with the P band will encourage root activity and K uptake. Again, further work on the interaction between these products at different in-band concentrations and in different soil types will be undertaken by the UQ postdoctoral appointment from early 2018.
  • Finally, trials are in the process of being established in Qld to explore the impact of amount of soil disturbance/mixing and the volume of soil enriched with P and K fertilisers on crop nutrient uptake. This will involve comparing discs, tynes and strip tilling units for their efficacy in making P available to the crop, and help to provide better guidance on the type of fertiliser rig required. A particularly interesting observation from this authors recent study tour to Europe and the UK has been the impact of voids/large pores on root branching and proliferation. This is illustrated in X-ray CT images produced in the University of Nottingham for maize and wheat (Fig. 5). Their results suggest that if the deep placement operation leaves large voids around the fertiliser bands and/or there is insufficient time and rain to allow profile reconsolidation, the chances of achieving the vigorous root proliferation needed to get good fertiliser uptake may not occur.

Conclusions

Research to better understand the effects of low subsoil P, agronomic strategies to overcome these limitations and the implications for fertiliser P management across northern cropping systems has come a long way. Rudimentary but functional diagnostic indicators of when to use starter and deep P fertilisers are becoming available, guidelines for effective application methods to address P limitations are developing, and economic assessments of the profitability of deep banding in a cropping system are showing strong returns in low P fields.

Factors that limit deep P response include the availability of other nutrients (particularly N) and a lack of plant available water, both of which can restrict the achievement of higher yield potentials. The quantum of deep P response is also affected by seasonal conditions that impact on root activity in different parts of the soil profile, so obtaining benefits under some seasonal conditions will remain somewhat problematic. This risk is countered by the excellent residual value of deep P that is being documented in field trials, allowing the benefits of applications to occur across a rotational sequence rather than solely in the year of application.

Figure 5 is two pictures of root systems side by side showing the impact of soil voids/air gaps on localized root branching in maize (left) and wheat (right). Image supplied by Prof Malcolm Bennett, University of Nottingham, and is reproduced from Morris et al (2017) Shaping Root Architecture. Current Biology, 27, R919–R930.

Figure 5. Impact of soil voids/air gaps on localized root branching in maize (left) and wheat (right). Image supplied by Prof Malcolm Bennett, University of Nottingham, and is reproduced from Morris et al (2017) Shaping Root Architecture. Current Biology, 27, R919–R930.

Acknowledgements

The extensive field research program undertaken as part of these projects is made possible by the significant contributions of growers through both trial cooperation and the support of the GRDC, the author would like to thank them for their continued support. The authors also gratefully acknowledge the efforts of the many QDAF and NSW DPI technical staff involved in conducting these research trials.

References

Bell, R., Reuter, D., Scott, B., Sparrow, L., Strong, W., Chen, W., (2013), Soil phosphorus-crop response calibration relationships and criteria for winter cereal crops grown in Australia, Crop and Pasture Science: an international journal for crop and pasture science, 64, 5, pages 480 - 498.

Contact details

Prof Mike Bell
University of Queensland
Building 8117A, Gatton Campus, 4343
Ph: 07 5460 1140
Email: m.bell4@uq.edu.au