Nitrogen fertiliser use efficiency ‘rules of thumb’ - how reliable are they?

Results and discussion

Available N at sowing

Table 1. Profile (0-120cm) soil nitrate-N (kg/ha) prior to sowing of the wheat phase in response to different rotation/tillage treatments in SCRIME (2001 to 2017). n.d. = not determined. n.s. = not significant (P < 0.05).

1 W=wheat; P=pulse; =barley; Gm=vetch green manure; ZT=zero tillage; RT=reduced tillage; CT=conventional tillage; C=canola; F=fallow; L=lucerne (Source: Armstrong et al. (2019)).

Figure 1. Relationship between rainfall (December of preceding year to March, inclusive in mm) and the amount of nitrate-N (kg/ha) in the profile (0-120cm) prior to sowing at SCRIME for 2001 to 2017. Analysis omits the value for 2011 when 375mm was received during this period.

Within season net N mineralisation

Studies of in-crop N mineralisation in southern region cropping systems indicate that this source of N potentially represents a significant proportion of a crop’s requirement (Dunsford 2019). A comprehensive review of DSS tools including simple RoTs for estimating in-crop N mineralisation has recently been published (Unkovich et al. 2020). Many of these tools were developed in an era when pasture legume-leys dominated cropping systems or on acid soils in southern NSW rather than current continuous cropping, reduced tillage/stubble retention practices and alkaline soils that dominate cropping in the southern region (Dunsford 2019). Dunsford (2019) found that the ‘Ridge Approach’, provided a fair estimate (R = 0.46) of net in-crop N mineralisation across a large section of cropping systems in western Victoria, based on soil organic carbon (SOC) of the topsoil and actual growing season rainfall (GSR, Table 2). The Ridge method is calculated as:

N mineralisation (k/ha) = 0.15 × SOC (%) × GSR (mm)………….Equation 2

If long-term average rainfall is used in the calculation, however, the reliability of the prediction is reduced considerably (data not presented). The recent extensive review of soil mineralisation by Unkovich et al. (2020) concluded that ‘none of the currently available tools appear to provide field and season-specific information (prediction) of N mineralisation on a useful time step’, or where more complex models existed (e.g., Yield Prophet), ‘they required detailed parameterisation unlikely to be conducted at an appropriate spatial scale‘.

Table 2. Correlation coefficients (r) of the relationships between net in-crop nitrogen mineralisation (net ICM) against growing season rainfall (GSR: April to October) and methods of predicting ICM, including soil tests and a simple calculator. Values are presented as r with significance indicated by *** P<0.001, ** P<0.01, and * P<0.05.

The efficiency of recovery of soil and applied N by the crop

A range of methods have been used to estimate fertiliser efficiency, and in particular, the quantity of fertiliser recovered by the crop. Traditionally, especially in the southern region, this was determined by measuring the difference between the quantity of N in the fertilised crop and a paired unfertilised treatment, expressed as a percentage, and referred to as the ‘difference measure’ (Strong 1995). Fertilised crops, however, frequently take up more soil N than unfertilised crops in a phenomenon referred to as the ‘added N interaction’ or a ‘priming effect’. As a result, difference measure generally overestimates the efficiency of use of fertiliser N. In contrast to difference methods, the efficiency of recovery of soil and fertiliser N can be assessed using ¹⁵N tracers. ¹⁵N is a stable (i.e., non-ionising) form of N that occurs ‘naturally’ at low levels in the environment (0.00367% of all N) and effectively behaves similarly at a chemical and physiological level to 14N. Although the applied ¹⁵N is subject to isotopic discrimination during mineralisation-immobilisation (MIT) (Strong 1995), it is considered a better method to assess fertiliser N utilisation (and losses), especially when measurements are made of residual ¹⁵N remaining in the soil after harvest. This is due to its ability to accurately detect crop uptake of a relatively small amount of fertiliser N (e.g., 25-100kg N/ha) against a much larger background quantity of soil N (e.g. 2,200kg N/ha assuming soil total N = 0.2% and bulk density of 1.1g/cm3).

A generalised value of 50% recovery of the fertiliser and available soil N within the season of application by the crop is widely assumed (Unkovich et al. 2020). This value is based on experimental data from field trials undertaken in the 1980s cited in Chen et al. (2008) in southern Australia and Strong (1995) in southern Queensland, which indicate that wheat crops recover 25 to 60% of the applied N. More recent publications cite fertiliser N recoveries by wheat of 49 and 57% in an acid chromosol in southern NSW (Smith et al. 2019).

Data on crop utilisation of ¹⁵N labelled urea collected from western Victoria using current grower management practices and rates across a range of environments and seasons indicate a wide range of fertiliser N recoveries (2 to 75%) depending on the zone/cropping system (Table 3). The average crop recovery of fertiliser N (34 to 35%) was surprisingly similar between environments and cropping systems. The key determinant of crop recovery was rainfall (or irrigation) following application. Very low recovery of fertiliser N was measured where GSR post application was <66mm, a common occurrence in low and medium rainfall situations during the seasons tested. Crop recovery was also sensitive to the rate of application (as expected), with higher crop recovery at lower application rates. However, there was no statistical evidence (P > 0.05) that applying fertiliser produced an added N interaction (i.e., inducing an increased utilisation of soil N; data not presented), a finding that contrasts with the NANORP study where the interaction was relatively large (Wallace et al. 2020).

Table 3. Mean (and range) of crop recovery of fertiliser N using both ¹⁵N labelling and difference methods, errors between these methods and losses of fertiliser N from the crop:soil system within the season of application using pooled data from nine sites (2014 to 2016). N was applied as top-dressed urea during vegetative/early tillering stage.

Data from a smaller spatial scale study (three experiments, all in the Wimmera) indicated a narrower range of crop fertiliser N recoveries, ranging from 12 to 56% (Table 4). This study highlighted the strong effect of seasonal conditions on the effectiveness of different N management strategies. For example, in 2012, N management had no significant effect on crop recovery of fertiliser N (range 49 to 55%), although it did affect grain yield response. In contrast, in 2014, maximum fertiliser recovery by the crop occurred when N was applied early whereas later applications (topdressing) resulted in very low crop recoveries and a trend to poorer grain yield responses due to very low rainfall received from August onwards. A positive finding, however, was that the low crop recovery under these dry conditions generally corresponded to high rates of recovery in the soil at harvest, rather than being lost. In contrast to the AonG survey data, significant interactions were recorded, indicating that utilisation of soil N was stimulated by fertiliser treatment (Wallace et al. 2020).

Table 4. Crop total N uptake at maturity, N uptake from fertiliser, total N uptake, recovery of fertiliser N in crop (grain + straw) and estimated loss of fertiliser N from crop:soil system (0-40 cm) using different N management strategies in the Wimmera from 2012 to 2014. (Source: Wallace et al. 2020).

0N = no N applied. 50N = urea incorporated at sowing; 50 N DMPP = urea + nitrification inhibitor incorporated at sowing; 0:50 N: urea top dressed at early-mid tillering; 0:50 N NBPT = urea + urease inhibitor top-dressed at early-mid tillering. Fertiliser N applied at equivalent of 50 kg N/ha.*Superscripts indicate significant differences (P <0.05) compared with other treatments within a given year.

Net off-site losses of N or immobilisation

Significant losses of N can occur from the soil by a range of pathways, including gaseous losses of ammonia (via volatilisation), denitrification (predominantly as dinitrogen- N₂) and leaching (of nitrate) below the root zone.

The use of a ¹⁵N labelled fertiliser mass balance approach allowed a quantitative assessment of the potential irretrievable loss of N derived from fertiliser. In our field studies, losses of fertiliser N ranged from 2 to 47% in low and medium rainfall zones (mean = 25%), 4 to 53% (mean = 32%) in the HRZ and 26 to 57% (mean = 41%) in irrigated cropping sites in northern Victoria (Table 3). Similar to crop recoveries, N management strategy significantly affected losses of fertiliser N depending on seasonal conditions, with nitrification inhibitors (DMPP) reducing losses in above average rainfall conditions and urease inhibitors (NBPT) producing significant benefits in reducing losses of top-dressed urea under dry seasonal conditions (Table 4). While the data directly estimated the loss of N derived from the labelled fertiliser applied to the crop:soil system, it could not unfortunately identify the primary N loss processes responsible. Measurements following harvest indicated little movement of ¹⁵N below the topsoil (0-10cm and occasionally 10-20cm), suggesting that gaseous loss (denitrification or volatilisation) was most likely responsible rather than leaching. Furthermore, total losses of N from the system are likely to be greater than our data indicate, as our procedure only accounted for fertiliser N and not for losses of ‘background’ soil mineral N which was not labelled with ¹⁵N.

General discussion

Despite its importance to on-farm profitability, N management remains problematic for most growers and advisers. The guiding ‘4R’ principles of the right source, rate, time and place of fertiliser requires knowledge of both crop demand and soil N supply. Seasonal conditions remain the primary driver of crop demand for N in dryland cropping systems and to a significant extent also fertiliser use efficiency. Similarly, seasonal conditions also strongly influence the underlying assumptions of soil N supply, via its effect on both the rate of N mineralisation during the summer/autumn fallow and the rate of N mineralisation in-crop, as well as crop utilisation and losses/immobilisation of fertiliser N.

Although no one can control seasonal conditions, seasonal forecasting is steadily improving and it is thought that growers/advisers can significantly reduce uncertainty of potential N supply through measuring both soil profile N prior to or just after sowing and in-crop N mineralisation using currently available procedures. The previous widely held assumption of 50% recovery of fertiliser and soil N was reputedly based on higher recovery of soil mineral N (70%) and 30% of fertiliser N (Mike Bell, quoting Wayne Strong and Chris Dowling). This value was formulated, however, in a period when soil N supplied most of the crop’s N. In current cropping systems, however, where there is less legume/pasture leys and tillage, and organic matter levels have declined, fertiliser N is the dominant source of crop N supply, a more appropriate figure for crop utilisation of fertiliser and soil of 35% is suggested for the southern region. This revised figure appears to be consistent across diverse cropping systems, although this may reflect differing influences. For example, better soil moisture in higher rainfall environments allowing for greater crop access to fertiliser being balanced by lower losses of N in drier environments.

Similarly, losses of fertiliser and presumably soil N appear to be an inherent feature of current cropping systems and based on the data appear to have been underestimated. Although many advisers in the 2015 survey were aware of the mechanisms by which N could be lost from the cropping system, most had difficulty in identifying rates of losses occurring in their environments. Furthermore, because the emphasis of many previous field-based experiments has been on greenhouse gas emissions (N₂O), where total quantities of N in terms of kg/ha are relatively low (Wallace et al. 2018), there was a lack of awareness that N₂ losses from denitrification could be of much greater ‘agronomic significance’, as suggested by the ¹⁵N mass balance data. ‘Unaccounted for’ N which was assumed to be lost averaged 25, 32 and 41% in low-medium, high rainfall and irrigated cropping systems, respectively, but instances of losses of > 50% were recorded across all these systems. These losses occurred at agronomically relevant rates and could be expected to become relatively larger at higher application rates. Previous research (also using ¹⁵N mass balances) has found large losses (up to 90% depending on application strategy) of applied fertiliser N from HRZ cropping systems in western Victoria (Harris et al. 2016). In the AonG study, fertiliser N was applied via topdressing of urea (mostly during tillering to first node), so an assumption was that volatilisation was the main loss mechanism. However, there were circumstances, especially in the HRZ and irrigated cropping trials and on sodic soils in the MRZ, that background soil conditions may have been conducive to denitrification losses driven by anaerobic waterlogged conditions. Knowing the mechanism of this N loss is important, as the NANORP study clearly showed that use of an appropriate fertiliser management strategy can significantly reduce losses of N and enhance supply to the target crop if applied in the correct situation.

Conclusions

Growers can potentially improve their fertiliser N management by (i) undertaking deep N soil testing, (ii) using current RoT predictions of in-crop N mineralisation and (iii) reducing the assumed crop utilisation of soil + fertiliser N to approximately 35 rather than current 50%.

Large losses of N appear to occur regularly across low and medium rainfall and irrigated cropping systems, not just in the HRZ as previous thought. However, the ability to mitigate these losses is hampered by uncertainty as to the loss pathways. Knowledge of this information would facilitate the identification of appropriate fertiliser management strategies, which have been shown to significantly reduce losses and improve yields.

Acknowledgements

Much of the original data used in this project originated from the Australian Department of Agriculture and Water Resources (AOTGR2- 0073) project ‘Reducing on-farm nitrous oxide (N₂O) emissions through improved nitrogen use efficiency in grains’ and Filling the Research Gap’ NANORP program. We wish to acknowledge the invaluable discussions and insights provided by our colleagues Murray Unkovich, Mike Bell, Louise Barton and Sean Mason. Katherine Dunsford’s PhD was co-funded by Agriculture Victoria, LaTrobe University and GRDC through Project DAN00168. We wish to acknowledge the significant contributions of the many grain growers of western Victoria who assisted us with conducting these experiments on their properties.

References

Angus JF (2001) Nitrogen supply and demand in Australia agriculture. Australian Journal of Experimental Agriculture 41, 277-288.

Armstrong, RD, R Perris, M Munn, P Hekmeijer, K Dunsford, K Bell, F Waldner and Z Hochman (2018) Genetics, management or environment: what are the key drivers of water use efficiency in wheat in western Victoria? GRDC Adviser Update, Bendigo February 2018.

Armstrong RD, Perris R, Munn M, Dunsford K, Robertson F, Hollaway GJ, O’Leary GJ (2019) Effect of long-term rotation and tillage practice on grain yield and protein of wheat and soil fertility on a Vertosol soil in a medium rainfall temperate environment. Crop & Pasture Science 70, 1-15.

Chen D, Suter H, Islam A, Edis R, Freney JR, Walker CN (2008) Prospects of improving efficiency of fertiliser nitrogen in Australian agriculture: a review of enhanced efficiency fertilisers. Soil Research 46(4): 289–301.

Dunsford K (2019) ‘Understanding causes of poor grain yield response of wheat to phosphorus and nitrogen fertilisers’ PhD thesis, LaTrobe University, Bundoora, Melbourne

Gan Y, Campbell CA, Janzen H, Lemke RL, Basnyat P, McDonald CL (2010) Nitrogen accumulation in plant tissues and roots and N mineralisation under oilseeds, pulses, and spring wheat. Plant & Soil 332, 451-461.

Harris RH, Armstrong RD, Wallace AJ & Belyaeva ON. (2016) Delaying nitrogen fertiliser application improves wheat ¹⁵N recovery from high rainfall cropping soils in south eastern Australia. Nutrient Cycling in Agroecosystems. 106, 113-118

Smith CJ, Hunt JR, Wang E, Macdonald BCT, Xing H, Denmead OT, Zeglin S, Zhao Z, (2019) Using fertiliser to maintain soil inorganic nitrogen can increase dryland wheat yield with little environmental cost. Agriculture, Ecosystems and Environment 286, 106644.

Strong W (1995) Nitrogen fertilization of upland crops. In: Bacon PE, Ed. Nitrogen fertilization in the environment. New York: Marcel Dekker Inc; p. 129–70.

Unkovich, M, Herridge, D, Denton, M, McDonald, G, McNeill A, W, L (2016) Nitrogen mineralisation and N decision support tools in cropping systems of southern Australia. A milestone report for the Grains Research and Development Corporation by The University of Adelaide., Adelaide.

Unkovich MJ, Herridge DF, Denton MD, McDonald GK, McNeill AM, Long W, Farquharson R & Malcolm B. (2020) A Nitrogen Reference manual for the Southern Cropping Region. GRDC.

ISBN: 978-1-921779-99-2

Wallace AJ, Armstrong RD, Harris RH, Belyaeva ON, Grace PR, Partington DL, Scheer C (2018) Reducing N₂O emissions of rainfed wheat in a semi-arid environment through fertiliser timing and use of inhibitors. Nutrient Cycling in Agroecosystems 112(2), 231-252

Wallace AJ, Armstrong RD, Grace PR, Scheer C,Partington DL (2020). Nitrogen use efficiency of ¹⁵N urea applied to wheat based on fertiliser timing and use of inhibitors" Nutrient Cycling in Agroecosystems 116, 41-56.

Contact details

Prof. Roger Armstrong
Grains Innovation Park, 110 Natimuk Rd, Horsham VIC 3400.
0417 500 449
roger.armstrong@agriculture.vic.gov.au