The economics of deep Phosphorus use in marginal environments

Phosphorus (P) is an increasingly important fertiliser nutrient needed to support productivity and profitability of northern region cropping systems. It is a nutrient that is needed in small amounts (but at high plant tissue concentrations) to support the establishment of grain numbers at floral initiation in grain crops, very early in crop development. This is why the practice of using starter P fertilisers (placed with or very close to the seeding trench and hence the developing seedling root system) evolved. But P is also needed in increasingly larger quantities in later stages of growth to establish a high tiller density (in cereals), to help develop a vigorous root system and to grow biomass and ultimately fill grains (in all species). That later season P has traditionally come from native subsoil P reserves, but as we remove more and more in harvested grain the need to introduce fertiliser P sources to replenish that is becoming more urgent. As one can imagine, the placement of starter P fertiliser to meet the demands of a young seedling with a very small root system will be quite different to that needed to meet demands of a well established plant living on subsoil moisture during flowering and grain filling.

Some soils were low in available P reserves from the start and have required P inputs from early in their cropping history. However for many others the need to apply P is relatively new or still non-existent, due to either high native P fertility or relatively short crop histories. As a result, we in the north are still working out where and how best to use P in our fertiliser programs. This is a major contrast to southern and western systems where soil P was in its native state almost uniformly low, and long term P fertiliser use now has growers questioning when to stop (or at least reduce rates). In many ways the northern grains region (NGR) is at the opposite end of the P fertility continuum to the rest of Australia, in that we are starting to ramp up P fertiliser use.

Another reason we are different is how we farm. With the possible exception of parts of the central west of NSW on lighter soils, soil moisture storage during fallows and subsequent extraction and use during a crop season are as important, or in many cases more important, than in-season rainfall for achieving a profitable yield result in most of the NGR. Last winter (2013) was an extreme example, where many crops in Qld and NE and NW NSW were successfully grown on little or no effective in-crop rainfall. While not always to this extent, subsoils and root activity in them are the keys to our success in most northern cropping seasons. However it is not just water we extract from those subsoils, with availability of nutrients in those same layers essential to sustain growth of plants. Nutrients removed from those subsoil layers have to be replenished if we wish to keep successfully farming subsoil moisture, and while some nutrients like nitrogen (N) and sulfur (S) can be moved back into those layers as soil water stocks replenish, this transfer method doesn’t work for immobile nutrients like P and potassium (K). Replacing subsoil P and K requires either placing fertiliser into those layers directly, or moving fertilised topsoils deeper into the profile with some sort of inversion tillage. The latter is generally not a popular or feasible option in our largely zero tillage, heavy clay soils, given our reliance on stubble cover for water infiltration and the often unfavourable chemical characteristics of subsoils that would be brought to the soil surface.

This then generates a series of questions around (i) whether we would expect to get a crop response to applied P; (ii) how and when that P should be distributed across the soil profile; and (iii) perhaps most importantly, does it pay and when! The data and discussion presented here are a summation of a combination of field trial results (that are continuing), climate analysis and simulation modelling. They are a work in progress, but may help you think through the process of deciding if you may need to change your P fertiliser practice, what needs to be done and whether it might pay. In the short term the answers to the latter are driven as much or more by seasonal conditions as by fertiliser or crop management, but the longer term issues of efficient use of water to produce grain require us to develop longer term solutions if we are to be cropping into the future.

Where would we expect P responses?

Soil testing of both topsoil (0-10cm) and subsoil (10-30cm) layers is the key to determining whether a response to applied P (starter or deep P or both) could be expected. While we are currently working to refine these fairly broad soil categories, and to explore crop requirements, our best estimates for the Vertosols appear in Table 1 below.

Table 1. Critical P values and their relationship to P fertiliser decisions in northern Vertosols

Briefly, deep P will generally deliver a growth response if (i) Colwell P is <10 mg/kg in the 10-30cm layer, AND (ii) BSES P in that same layer is <30 mg/kg. If Colwell P in that layer is >10mg/kg responses are unlikely, but if Colwell is <10 mg/kg and BSES P is 30-100 mg/kg, you are in an area of uncertainty, as we can’t know the solubility of all the minerals measured in the BSES test. If that is the case, try a test strip.

If the subsoil is low, what size responses would we expect?

We have a number of trials that have run for >1 year, and the cumulative yield responses are shown below –

• At Brookstead, our longest running site, we produced an additional 2100 kg/ha sorghum and 1200 kg/ha wheat over 6 crops from 2006/7 to 2011, which would have returned an extra $720/ha.
• At Capella we produced an extra 900 kg/ha chickpeas in 2012 and an extra 200 kg/ha wheat in 2013, which would be worth $410/ha for the first 2 crops after application using average crop values.
• At Gindie we produced an extra 600 kg/ha sorghum in 2011/12 and an extra 500 kg/ha chickpeas in 2013, which would return an extra $320/ha for the first 2 crops after application using average crop values.
• At Jandowae we produced an extra 0.5 t/ha wheat in a dry 2009 winter, followed by an extra 1.5 t/ha sorghum in 2010/11 which would return an extra $425/ha for the first 2 crops after application. Unfortunately a following chickpea crop was lost in the wet 2011 winter.
• At Wondalli we produced an extra 500 kg/ha sorghum in 2008/09 followed by an extra 650 kg/ha wheat in 2011, which would return an extra $263/ha for the first 2 crops after application using average crop values.
• At Biniguy we produced an extra 1.2 t/ha of wheat in 2011 followed by an extra 1 t/ha sorghum grain in 2011/12, returning an extra $500/ha.

Even last summer, under pretty tough conditions and in sorghum grown in single skip, double skip or 2m singles, responses were still significant. In these sites the first season after deep P was applied, grain yields typically averaged 3 t/ha under grower practice but with deep placed P yields increased by from 300-900 kg/ha, or 10-30%.

If we had to summarise all our trial experience to date, we would conclude that

• The frequency of grain yield responses to starter P would be in the order of wheat > long fallow sorghum, mungbean and chickpea > short fallow sorghum. The magnitude of the responses varied, but typically ranged from 0-10%, except in a mungbean crop where soil P was very low and responses were much larger in relative terms.
• All crops (wheat, chickpea, sorghum and mungbean) have shown responses to deep P application.
• Responses to deep P, assuming other nutrients like K and S were sufficient, have averaged ~20% in crop yields in the 1^st and 2^nd season (few sites have been monitored any longer). Some responses have been larger (50-70%) under particular agronomic (e.g. heavy nematode pressure) or climatic (limited or no effective in crop rainfall) circumstances.
• The strongest responses to deep P in grains (wheat and sorghum) occurred when post-planting rainfall allowed the establishment of secondary roots and tillers, which are the main pathways to plant P uptake and increased yields. In wet years when the topsoil was readily accessible, or in extremely dry years, responses were more limited. By contrast, the strongest responses in chickpeas were in years with little effective rainfall, especially up to pod loading, although consistent responses were still recorded in wetter seasons.

How often would we expect to see these responses?

Clearly there is an effect of seasonal moisture availability and in-crop rainfall on yields and fertiliser responses, and we can loosely classify seasons into three basic types with characteristics that influence crop response to deep P. We have loosely termed these as

Dry starts - those years with little or no effective rainfall from planting until after tillering. Secondary root growth and tillering are seriously affected;

Wet start/dry finish - those with enough rain to ensure good early growth, secondary root development and tillering but serious later stress ensuring a strong reliance on subsoil moisture; and

Wetter years– no severe crop stress, with an expectation that more regular rainfall will ensure the top soils have plenty of active roots (although we may be battling foliage diseases in the hope of securing high yields).

The frequency of these types of seasons obviously varies (winter v summer, and region to region). For example, the frequency of ‘dry start’ seasons is much higher in summer (average of 40% for sorghum and 21% for mungbean across the region) than in winter (typically only 5-10%) of years, except in CQld. However while the frequency of season types varies so does the impact, with the soil water holding capacity and how full the profile is at sowing key factors. While a ‘dry start’ that limits secondary root growth and tillering in cereals has a similar effect on soils with high and low water holding capacity (eg. 120mm or 240mm PAWC), the effects of the ‘wet start, dry finish’ in a low PAWC soil is much more pronounced than in a high PAWC soil. This is shown in the figure below for wheat. In addition, having a higher PAWC soil reduces the frequency of seasons that are ‘wet years’ versus ‘wet start, dry finish’ with the higher PAWC reducing the frequency of late stress seasons by an average of 30% (wheat) to 40% (chickpeas), and by 16% (sorghum) to 30% (mungbeans) in summer.

Figure 1. Simulated yields of wheat in ‘wetter year’ and ‘wet start/dry finish’ seasons at Gunnedah, Walgett and Condobolin for soils with 120mm or 240mm PAWC that were at least 2/3 full at sowing.

The yield losses due to low P incurred in the different seasonal types are clearly pivotal in any analysis of the costs and benefits of ensuring adequate P nutrition. The estimates we have come up with are preliminary and based on (in some cases very) limited data. However we think the discounts or yield reductions shown in Table 3 below are realistic for soils where we are very confident P responses will be obtained.  We have presented these as total yield response to applied P (the sum of starter and deep P), although we would note that responses to starter P range from 0% to 10% of yield potential, with most consistent responses in higher yielding seasons. Under the right seasonal conditions, overall P response can be much greater. These estimates of P responses have been used to work out the value of lost yield if low soil P is not addressed adequately (ie. right rate, right place and right time). These values are being used to estimate the costs and benefits of P fertiliser application. We have used average prices for each crop: $200/t for sorghum, $250/t for wheat, $400/t for chickpeas and $700/t for mungbeans.

Table 2. Estimates of relative yield losses that would be experienced if fertiliser P was not applied for different soils and season types.  These effects are due to a combination of starter P and deep P, and would reduce the simulated potential yields such as shown in Table 2.

As these loss factors are percentages, it is fairly obvious that regions/soil types with higher potential yields will deliver a larger yield benefit if low P is addressed. Conversely, in sites or seasons growing low yielding crops, the payoff for deep P applications (eg. 10-20% yield increase) is likely to be limited, as water or some other factor is a much greater limitation.

How much do we need to apply and what is the residual value of applied P?

This is currently part of the focus of the existing trial program in UQ00063, with responses to different P rates assessed in both the initial crop year and subsequent seasons through a crop rotation. We know from work at IPL long term fertiliser trial sites like Colonsay and Tulloona that roughly half of the net P removal in our cropping soils has come from the subsoil (Wang et al., 2007), so we assume that future fertiliser programs will ultimately be a mix of starter and periodic deep P applications if they are to be sustainable.

In our analyses here we have assumed deep P application rates of either 20 or 40 kg P/ha (eg. 100 or 200 kg MAP/ha), with responses available undiminished over a 3 (20P) or 5 (40P) year cropping period.  While most of this will be a ‘new’ fertiliser input in some areas, in others at least part of this P may be able to be diverted from what is currently being applied in starter fertilisers. Our experience suggests that, provided metering systems are adequate to ensure an even fertiliser distribution along the seeding trench (easier with liquid P forms), effective rates of starter P can be as low as 3-4 kg P/ha in row spacings of 100cm or greater (eg. summer sorghum) and 5-6 kg P/ha row spacings of 25-50cm (wheat and chickpeas). The response from higher rates of starter P is likely to be marginal, and although available to roots in subsequent seasons provided topsoils are wet, could probably be used more efficiently as part of deep applications.

Our experience in soils with low Phosphorus Buffer Indices (PBI) across the region (ie. most of the major cropping soils) is that residual value of these large P applications into subsoils is excellent and covers multiple crop years. We have typically used applications of 40 kg P/ha and seen consistent yield responses over up 6 consecutive crop seasons. Additional P uptake in crop biomass can be as high as 10 kg P/ha in some seasons, but the rate of removal of fertiliser P in grain is often only 20-25% of that found in biomass, so continued yield responses are consistent with the slow removal rate. Whether the P remains deep, or is returned to the topsoil in residue is still being determined.

What does it cost to deep place P?

We are indebted to growers in central and southern Qld for providing estimates upon which to base these figures. These operators have used purpose-bought strip tillage equipment or modified planters capable of deep sowing chickpeas to place fertiliser at ca. 20cm depth. Both sets of equipment were 12m wide and placed fertiliser at 50cm or 75cm band spacings (we would recommend bands 50cm apart or closer if possible to get the best response), and operated at 7-8 km/h, using about 60-65L/h in diesel. We have estimated total application costs at between $30 and $40/ha, which are added to costs of MAP or a mix of MAP and KCl (if both P and K are limiting) that vary from $750-$850/t bulk. Thus the cost of deep placement of 100 kg/ha product is estimated at $105-115/ha after every 3^rd crop, or $180-$210/ha for 200 kg/ha after every 5^th crop. Allowing for the interest on borrowing to afford those outlays over the residual period would raise the total cost of deep placement by 25% (3 year) or 40% (5 year), assuming borrowing at 8% interest. We have assumed these costs are additional to what is being incurred now (including any use of starter P). We think this is reasonable given possibly reduced costs associated with lower starter P rates but an increased N fertiliser cost associated with higher yield potentials in the cereal crops.  

Do we have to actually place the P there with cold steel and diesel?

We are often asked whether deep placement is the only way to address this issue, and whether fertiliser could be surface applied to soil that was dry and well cracked. A rainfall event or a harrowing would then result in at least some of the fertiliser falling into deeper profile layers. While we have not addressed this in trials, we expect results will be disappointing – for a couple of reasons. Firstly, cracks represent only a small proportion of the soil surface area. As a result the P that does drop down the cracks will be a small fraction of that applied, and will not be represent an evenly distributed P source that all plants along a crop row can access. Secondly, profiles don’t conveniently crack whenever we want to apply P, so there are limited opportunities when this can be tried. Finally, when soils are strongly cracked, the cracks can be very deep. Putting P very deep is not necessarily effective, as the best chance of good P recovery is out of the 10-30cm layer, where there are lots of crop roots most of the time.

So does it pay to deep place P?

Simulations – We have used the simulated potential yields and the discounts that would be incurred from low P (Table 2) to estimate the average responses to applied P for each crop in each season type. Then for each location and PAWC soil we have calculated an average yield response across the climate record which is weighted for the different frequencies of each season type. For example, on a 240mm PAWC soil in Gunnedah growing wheat, there were 6% ‘dry start’ seasons with an average yield of 2.5 t/ha, 30% ‘wet start, dry finish’ seasons with an average yield of 3.3 t/ha and 65% ‘wetter years’ with an average yield of 5.4 t/ha. Using these figures and the relative yield losses that would be incurred without adequate P in each season type shown in Table 2, we estimate the average annual loss in wheat yield resulting from low subsoil would be 840 kg/ha.

We have made similar calculations for each crop, soil type and location, and assembled some 5 crop rotation sequences (Fig. 1) which show the estimated cumulative lost production ($/ha) that we predict would be overcome by a 200 kg/ha deep MAP application at a cost of $180-$210/ha plus any allowances for borrowing costs. Data clearly show greater returns from applying P on a higher PAWC soils (reflecting the greater productivity of such soils in most seasons), and to some extent higher returns on summer only or mixed summer-winter crop rotations. The latter would appear to be due to the relatively high returns generated by mungbeans valued at $700/t, and given the quality/price uncertainties with this crop, those findings should be treated with caution.

Perhaps the key finding is that positive returns were generated from deep placement of P in responsive soils in most situations – the exceptions being on low PAWC soils with summer-dominant rotations in Walgett and Condobolin (not commonly grown anyway), and exclusively winter crop rotations at Emerald.  On low PAWC soils these returns barely exceeded costs of application and would represent a relatively poor return on investment, but in high PAWC soils returns typically ranged from $2-$5 for each dollar invested in deep P application. These results are consistent with the responses we showed earlier from a variety of our field trials.

Figure 2. Estimated returns from investment in deep P on responsive soils with 120mm or 240mm PAWC in centres across the northern grains region. Returns are based on median yield responses to applied P across 5 year crop sequences involving wheat (Wh), sorghum (Sg), chickpea (Cp) or mungbeans (Mb).

Collectively, the experimental results and this analysis support the conclusion that deep placed P would likely provide returns to growers well in excess of the application costs. The fact that most of our experimental sites achieved those returns within 2 crop seasons after deep P application may at least partly reflect the string of wetter seasonal conditions in recent years, although the results from Jandowae, Capella and Gindie all included dry seasons in at least one of the crops followed. A key factor that we need more data on is the consistency of the yield response to deep P applied every 3-5 years on differing soil types. This is currently a focus of the UQ00063 trial program.

For those considering deep P applications in their cropping program, a few key factors need to be considered –

• Make sure you test the paddocks first, to ensure a P response is likely.
• Make sure P isn’t the only limiting nutrient – secondary limitations to yield from K or S can limit an expected P response.
• If you are raising grain crop yield potentials above what is normally experienced, make sure there is adequate N available to achieve that higher yield goal.
• If you are diverting some of the P fertiliser used in starter applications to periodic deep P applications, make sure the rate you apply is still adequate to meet seedling P demand, with the key factor likely to be uniformity of fertiliser distribution along the seeding row.

Contact details

Prof Mike Bell
Queensland Alliance for Agriculture and Food Innovation (QAAFI), Univ of Qld
Ph: 07 4160 0730
Fx: 07 4162 3238
Email: m.bell4@uq.edu.au