Deep P update 2019 – Multi-year grain yield impacts and economic returns for southern Queensland cropping

Take home messages

  • You know your paddock variability for yield – use that to prepare a soil sampling program
  • Look at the fertility and constraint status of the soil profiles
  • For southern Queensland, placing phosphorus at depth has produced statistically significant yield responses in 26 out of 35 crop seasons. The cropping program has been dominated by winter crops at these sites (27 of 35 crop-seasons), with wheat and barley responding positively in all 15 site-years and chickpea in 6 out of 12 crop seasons. Sorghum has responded in 4 out of the 6 site-years. Whether these differences in response frequency relate to seasonal moisture availability, soil P status or inherent differences in the ability of crops to utilise deep P bands is being explored in other projects and additional experiments
  • Effects of deep placed P on grain P concentration were small, so grain P removal (export) is primarily driven by crop yield
  • It is still challenging to estimate what re-application timeframes for deep P re-application look like, given other nutrient limits and varying seasonal conditions, plus the lack of a method to directly account for fertiliser P recovery and export.

The trial program and experimental design

Field research as part of UQ00063 commenced in the winter of 2012.  Since then a total of 12 sites across the eastern Downs (2), western Downs (8) and Maranoa (2) have been established.  Experiments have generally consisted of rates of Phosphorus (P) applied in bands at ~20cm depth on spacings of 50cm, along with an untilled farmer reference treatment. Application rates at depth, range from 0 to 60 or 80 kg P/ha. Table 1 displays the structure of deep treatments used in the two experiments established in the Maranoa.  Initial experiments in 2012 (4 sites) used triple superphosphate (TSP) as the P source for deep treatments while subsequent sites used monoammonium phosphate (MAP).  Due to poor efficacy at higher soil pH values, the TSP sites where not continued from 2016. Effects on yield are reported separating the TSP and MAP site responses.

A basal nutrient application of nitrogen, sulfur and zinc was added to the deep application to balance the rates of N added as MAP and lower the risk of other nutrient limitations constraining P responses. Potassium was also applied at one location. A location on the eastern Downs that subsequently proved to be potassium limited is not reported as part of this summary. Full agronomic details for experiments are contained in Gentry and Grundy (2018).

All main plots were split to annual ‘with’ and ‘without’ starter P fertiliser applications at planting, to test for any interactions between starter P (standard practice) and deep P applications (i.e. were effects independent or could one application method substitute for the other). The choice of starter product and rate represented grower practice at each site. The crop sequence at each site was dependent on the local rotation, and the residual benefit of the different rates of deep P was tracked through subsequent growing seasons.

Table 1. Experimental treatments for Mt Bindango deep placed P sites (FR=Untilled Farmer Reference treatment– no P fertiliser applied)

Deep P treatment nutrient application rates (kg/ha)

Treatment no

1

2

3

4

5

6

7

P rate (as Mono Ammonium Phosphate)

FR

0

10

20

30

40

60

N rate (from MAP and Urea)

-

40

40

40

40

40

40

Zn rate (Zinc Chelate)

-

2.0

2.0

2.0

2.0

2.0

2.0

Measurements of crop response typically comprised biomass cuts at physiological maturity, to determine crop growth response and nutrient acquisition, in addition to machine harvested grain yields. Grain was also analysed for nutrient composition to calculate nutrient export.

Yield effects of starter and deep P were determined using ANOVA (VSN International 2017).

A subsequent analysis investigated the potential impact of the deep tillage and basal nutrient (as a surrogate for ripping effect) using REML (VSN International 2017) comparing just the FR (untreated) and 0P treatments, with/without starter application.  This approach considers all the yield data for the site year, but just focuses on those two treatments, and their potential interaction.

Effects of starter P, deep P and starter P x seep P interactions

Results are based on analysis of each crop as an individual year, with a table summarising the response frequency for the different winter or summer species (Table 2). “Winter cereals” combines mainly wheat crops with only 2 barley crops grown. Individual site analyses are shown in the appendix Table 1. The number of experiment-years for each crop varies as two locations did not have starter applied.  Statistical significance for the starter treatment needs to be interpreted conservatively as the experimental design was only testing presence or absence of the fertiliser and so has limited the experimental ability to determine influence. It should be noted that while starter P responses are based on fresh applications in each crop season, deep P effects represent the average response across sites that have had up to five crop seasons after the application of deep P and the other basal nutrients.

Table 2. Summary of statistical significances (p <= 0.05) for southern Queensland sites to starter, deep or interactions for fertiliser P

 

Starter responses

Deep

Starter x deep

Winter Cereals

11 of 14 crops

15 of 15 crops

3 of 15 crops

Chickpeas

2 of 11 crops

6 of 12 crops

2 of 11 crops

Sorghum

0 of 6 crops

4 of 6 crops

0 of 6 crops

Mungbean

1 of 2 crops

1 of 2 crops

0 of 2 crops

Summary of results

Winter cereals – wheat and barley

Winter cereals consistently responded to having both starter fertiliser applied at sowing and to application of deep P, with very few crops showing an interaction between starter and deep P. From the 3 sites where statistically significant interactions were recorded, only 1 result appeared to be a genuine interaction, with the others a product of unexplained data variability. These results then reduce fertiliser P management into two independent decisions for winter cereals in southern Queensland: one about starter fertiliser use, and the other for deep placement.  Yield gain when starter P was applied averaged 210 kg/ha (7.6%) across all sites for wheat and barley, compared to no starter fertiliser.

Assuming P costs of $3.60/kg and typical starter-P rates of 6-12 kg/ha, applications represent a cost of approximately $20 - $40/ha. This cost is easily returned by the $84/ha from an average  210kg yield gain. At current prices, the response to starter provides a positive economic return to growers and so should be considered as a part of normal recommended practice. Grain prices would have to fall to below $200/t before this yield benefit would not add extra profit from 12kg of P, and below $100/t for 6kg/ha of starter P to not be profitable.

Deep P at 20 kg P/ha applied as either TSP or MAP has increased average grain yield at winter cereal sites by 9-13% (Table 3). Comparisons between TSP and MAP sites using cross-site statistical techniques, and in-field comparisons of P fertiliser choice are underway currently in UQ00078 to explore further the product choice options. With the MAP sites, increasing the deep P rate to 30 kg P/ha generated mean increases of 380 kg/ha (range 141-826) resulting in an additional 15% yield increase.

Table 3. Winter cereal yield change summary for deep placed TSP or MAP at 20 kg P/ha (FR=Untilled Farmer Reference – no P fertiliser applied)

Deep P source

Number of
crop-years

FR yield (kg/ha)

Average yield change with 20 kg P/ha deep

Range of responses with 20 kg P/ha deep

TSP

5

2426

+217 (9%)

+115-341

MAP

10

2522

+325 (13%)

+117-707

Chickpeas

Like the situation with winter cereals, chickpeas exhibited a low frequency of starter x deep P interactions, and again the two sites where these were significant were likely artefacts resulting from unexplained data variability. As outlined in Bell et al. (2016), chickpeas do not have an obligate requirement for starter application to set grain number (unlike cereal grain crops) and the very small number of responses to starter application (2 of 11 crops) is consistent with this (Table 2). However, there are situations where chickpeas were deep-sown into subsoils with very low available P, so the probability of starter P responses in these situations is greater.

The average chickpea yield without starter across all sites was 1747 kg/ha, compared to 1822 kg/ha with starter – a 75 kg/ha difference.  Where that comparison was restricted to the two significant ‘starter-responsive’ sites, the yield increase from starter application averaged 300 kg/ha (1710 kg/ha without starter, 2010 kg/ha with).

At $800/t, even the overall average of 75 kg/ha increase in chickpea yield easily covers the cost of $20-40/ha of starter P, and the observed upper end responses would generate over $200 in additional profit for growers.  To improve the reliability of starter P responses, growers should consider further on-farm experimentation – especially comparing responsiveness under deep sowing or normal sowing conditions.

It has been more difficult to make conclusive interpretations of deep P effects in chickpea crops in southern Queensland, with only half of the crops (6 from 12) showing statistically significant responses to deep P (Table 2).

Table 4. Chickpea yield change summary for deep placed TSP or MAP at 20 kg P/ha (FR=Untilled Farmer Reference – no P fertiliser applied)

Deep P source

Number of
crop-years

FR yield (kg/ha)

Average yield change with 20 kg P/ha deep

Range of responses with 20 kg P/ha deep

TSP

7

2007

+56 (3%)

-172 to +249

MAP

5

1203

+133 (11%)

-144 to 535

The reasons for this more variable response are unclear, particularly in the light of the more consistent responses recorded in central Queensland. Dry matter responses to deep P were larger and more consistent than grain responses with an average increase of 500 kg/ha (10%).  As harvest index for pulse crops is not relatively constant (compared to grass crops), this suggests that growth responses to P are not necessarily translating into yield responses. Further investigation into the relationship between P supply, biomass growth and establishment of grain yield in chickpea would appear to be needed to explain these interactions.

Sorghum

None of the sorghum crops grown in the current study period (2013-14 to 2017-18) recorded any statistical effect of starter application.  Average grain yields without/with starter application also indicate a negligible effect (3404 kg/ha without vs 3376 kg/ha with).  Warm soil conditions allowing rapid root expansion, combined with high potential evaporative loss in surface layers, may allow rapid early exploitation of P in the top soil layers but then limit the duration of access to the starter P band.

Deep P as MAP at 20 kg P/ha increased average grain yield from 3431 kg/ha in untreated plots by 311 kg/ha (Table 5).  Application of 30 kg P/ha increased average yields slightly more with an average 372 kg/gain (11%, 319 – 514 kg/ha range). With only two crops on the TSP sites it is difficult to make much assessment on performance.

Table 5. Sorghum yield change summary for deep placed TSP or MAP at 20 kg P/ha (FR=Untilled Farmer Reference – no P fertiliser applied)

Deep P source

Number of
crop-years

FR yield (kg/ha)

Average yield change with 20 kg P/ha deep

Range of responses with 20 kg P/ha deep

TSP

2

2924

69 (2%)

54 - 84

MAP

4

3431

311 (9%)

-44 – 517

Mungbean

The very limited set of mungbean data makes robust recommendations challenging.  Like sorghum and chickpea, starter application showed negligible effects on mean yield (876 kg/ha without starter vs 908 kg/ha with starter).  Similarly, deep P has provided only small average yield increases of 67 kg/ha for mean untreated yields of 837 kg/ha.

Economic assessment of deep P

As Deep-P involves large upfront costs (~$100/ha for 20kg P) it is important to identify how many crops it takes for this investment to be repaid, and how long this investment will continue to generate additional income. Of the 11 sites for deep P experiments in southern Queensland, 8 had repaid the investment in 20 kg/ha P and returned increased profit within 2 years and 5 of those had managed to do so in the first year.  The 20 kg/ha P treatment at Jimbour West, which has had 5 crops between winter 2014 and winter 2018, has returned almost $800/ha in increased profit over this time period.

Potential tillage impacts on response with deep P

As outlined earlier, a contrast analysis just comparing the factorial effects of +/- starter and +/- deep rip plus basal nutrients was conducted.  This approach focuses on these treatments inside the broader yield data for the site and crop in that year.

Results of this analysis indicate there was no substantial impact of deep tillage and basal application on their own, relative to current grower practice, with only 2 of the 35 crop seasons showing any statistically significant response.

Grain P concentrations with deep P treatment

Across the southern Queensland trial program, there were contrasting effects on grain P concentration between the grass and pulse species.  For wheat and barley, 20 kg P/ha at depth increased grain P concentrations by an average of 150 mg P/kg (or 0.15 kg P/t).  Grain P concentration in FR plots averaged 2260 mg P/kg (2.26 kg P/t) and with the 20 kg deep P/ha it increased to 2410 mg P/kg (2.41 kg P/t). At average grain yield for 20 kg P/ha applied deep, only an additional 1.1 kg P/ha leaves the paddock compared to the treatment without deep P.

Chickpea grain P concentrations showed greater responses to deep P applications, with 20 kg deep P/ha increasing grain P concentration by 330 mg P/kg (0.33 kg P/t). However, grain yield increases were smaller, so the change in P removed from the field with a 20 kg P/ha treatment was an increase of 1.2 kg P/ha, comparable to that of winter cereals.

The small differences in P removal rates with deep P application suggest that it will be difficult to use “cheque book” accounting to monitor depletion of deep placed P treatments.  These data also suggest that while P applications can generate significant yield responses and improved profitability, they are also not having a large impact on crop P status. Grain P concentrations <2500-2900 mg P/kg are purported to indicate suboptimal crop P status in wheat, but even with a combination of deep P and starter P applications, average grain P concentrations still average only 2400 mg P/kg. These data therefore highlight the fact that once profile P becomes severely depleted, restoring soil P status with fertilizer applications will be a slow process that requires careful ongoing management.

Average additional P removal of ≈ 1.0-1.2 kg P/ha would appear to suggest an extended lifespan from a deep P application of 20 kg P/ha. However, it is unknown how residual fertiliser P availability will be impacted over time by chemical reactions that occur in the fertiliser band, and how those reactions might vary between soil types. Additional work to explore the chemistry of residual P availability is currently being conducted by a GRDC-funded postdoctoral fellow at University of Queensland.

Suggested on-farm research treatments

This field research was conducted under carefully managed experimental conditions.  Before commencing a large-scale nutrient application program, growers are urged to appropriately soil test their fields to establish available nutrient concentrations in the surface and subsurface layers, and to identify the potential constraints to yield. They are then encouraged to evaluate the crop responses to fertiliser applications designed to address those yield constraints using an appropriate program of strip-trials and on-farm exploration to validate the diagnosis of nutrient constraints.

There are four suggested treatments to explore the effects of a deep P application before starting a larger program (Table 6):

  1. Treatment 1 is current practice or “do nothing”, which benchmarks current system performance;
  2. Treatment 2 involves the physical tillage of soil to a depth or roughly 20-25 cm, which simulates the deep placement operation without any fertiliser application. While not a long-term solution, simply loosening soils can sometimes allow better root exploration of those profile layers and allow more efficient uptake of scarce soil P resources.
  3. Treatment 3 is tillage with additional nitrogen.  In many sites, nitrogen status is in equilibrium with the existing ‘normal’ yields from that field, and if deep P improves field yield potential, extra N has to be applied to achieve the higher yield target.  Applying additional N alone in this treatment allows growers to separate responses from tillage, extra N, and extra N and P.
  4. The last treatment is deep P application.  Given that MAP is the most effective form of P, and soil Zn is often also low, an application of 100-150 kg/ha of an ammonium phosphate product with Zinc is typically used.  Suggested rates for use in strip trials are 20-30 kg P/ha of an ammonium phosphate-based product. Placement of the P needs to be such that crops are going to be likely to access it. Plant roots must have a high probability of encountering the applied P early in the growth stage, so band spacings of 50 cm or less are suggested to maximise the chances of roots from each crop row encountering some of the applied nutrients.

Table 6. Suggested on-farm deep P treatments

Treatment

Rip (≈ 20-25 cm)

Deep N
(≈ 30/50 kg N/ha)

Deep N+P+Zn
(30/50 kg N/ha + 20/30 kg P/ha + Zn/ha)

1

   

2

Y

  

3

Y

Y

 

4

Y

Y

Y

Treatments should be done in a way to make recording of yield response simple. The easiest strategies involve full-length field strips (ideally two or three header widths together) and also replicated several times within and across fields. Talking with precision ag practitioners, the minimum treated area to produce reliable yield estimates with harvester yield monitors is 1 ha in 5-6 header widths.

Acknowledgements

The research undertaken as part of project UQ00063 was made possible by the significant contributions of growers who host the field trials and the financial support of the GRDC. The authors would like to acknowledge these contributions and thank both groups for their continued support.

References

Bell, MJ, Lester, DW, Graham, R, Sands, D, Brooke, G (2016) Phosphorus and potassium nutrition. In 'GRDC Adviser Update - 2016. Goondiwindi', Mar 2016. (GRDC. Available at https://grdc.com.au/resources-and-publications/grdc-update-papers /tab-content/past-update-proceedings/2016/grdc-grains-research-update-goondiwindi-2016

Gentry, J, Grundy, T (Eds) (2018) 'Queensland Grains Research - 2017-18 Regional Agronomy.' (Department of Agriculture and Fisheries (DAF): Brisbane, Qld).

VSN International (2017) 'Genstat for Windows 19th Edition.' (VSN International, Hemel Hempstead, UK)

Contact details

David Lester
Department of Agriculture and Fisheries
PO Box 2282 Toowoomba
Mb: 0428 100 538
Email: david.lester@daf.qld.gov.au

GRDC code: UQ00063

Supplementary Table 1. Individual site by statistical significance for starter, deep or interaction

Crop

Starter

Deep P

Starter * Deep P

Westmar - ANOVA

2014 Wheat

*

***

n.s.

2015 Chickpea

n.s.

n.s.

n.s.

2016 Chickpea

n.s.

**

n.s.

Inglestone - ANOVA

2013 Wheat

*

***

***

2014 Chickpea

**

**

n.s.

2015 Chickpea

n.s.

***

***

Lundavra #1 - ANOVA

2013 Wheat

No Starter

*

No Starter

2014 Chickpea

n.s.

*

*

2015 Wheat

n.s.

***

n.s.

2016-17 Sorghum

n.s.

n.s.

n.s.

Lundavra #2 - ANOVA

2013 Chickpea

No Starter

n.s.

No Starter

2014 Wheat

n.s.

*

*

2015-16 Sorghum

n.s.

***

n.s.

2016 Chickpea

n.s.

n.s.

n.s.

Wondalli - ANOVA

2013-14 Sorghum

n.s.

***

n.s.

2015 Wheat

***

***

n.s.

2017 Wheat

***

*

n.s.

Condamine #1 - ANOVA

2015-16 Sorghum

n.s.

n.s.

n.s.

2017-18 Mungbean

n.s.

n.s.

n.s.

Condamine #2 - ANOVA

Crop

Starter

Deep P

Starter * Deep P

2014 Chickpea

n.s.

n.s.

n.s.

2015 Wheat

*

*

***

2017 Wheat

**

**

n.s.

2018 Wheat

**

***

n.s.

Mount Carmel - ANOVA

2013-14 Sorghum

n.s.

**

n.s.

Jimbour West #1 - ANOVA

2014 Barley

*

***

n.s.

2014-15 Mungbean

*

**

n.s.

2015-16 Sorghum

n.s.

*

n.s.

2017 Chickpea

*

***

n.s.

2018 Barley

**

***

n.s.

Mt Bindango #1 – ANOVA

2016 Wheat

*

***

n.s.

2017 Wheat

*

**

n.s.

2018 Chickpea

n.s.

*

n.s.

Mt Bindango #2 – ANOVA

2016 Chickpea

n.s.

n.s.

n.s.

2017 Wheat

n.s.

*

n.s.

2018 Chickpea

n.s.

n.s.

n.s.

n.s. = not significant, * = p < 0.05, ** = p < 0.01, *** = p < 0.001

GRDC Project Code: UOQ1207-001RTX,