In the south-eastern wheatbelt most of the nitrogen in cereal crops originates as biological fixation of clover-based pastures or grain legumes. It is unlikely, in the foreseeable future, that fertiliser-nitrogen will overtake pastures as the main contributor to crop nitrogen. However, there is increasing evidence that, in some circumstances, fertiliser can provide an important supplementary source of N which can profitably increase yield, grain protein or both.

Fertiliser nitrogen is not a new fashion for the region. About 20 years ago, an extensive series of N fertiliser trials were conducted over most of the wheatbelt. The yield responses found were highly variable but were, on average, low and unprofitable. In recent years CSIRO, the Departments of Agriculture, fertiliser companies and private groups have begun to re-examine the question of nitrogen fertiliser. Our reasons for persisting, in spite of previous failures, are because of changes in the wheat industry, including:

(i) more nitrogen-responsive wheat varieties;

(ii) lower soil nitrogen because of increased cropping intensity and poor pastures;

(iii) less disease because of break-crops;

(iv) availability of tissue tests.

The unchanging problem affecting nitrogen response of the region is the variability and unpredictability of rainfall; an adequate supply of water is essential for a yield response to nitrogen. Expressed in another way, the problem is to balance the supply of, and demand for, crop nitrogen. The ideal solution would be to apply nitrogen fertiliser only when seasonal weather conditions are likely to favour a large yield response. Since there are no forecasts at the time of sowing to predict weather for the whole season, a practical alternative is for wheatgrowers to delay the decision about N fertiliser until the season has progressed enough for a better decision to be made about its suitability. Research at CSIRO Canberra has been focussed on identifying the seasonal conditions and the N status of crops for which N fertiliser responses are most likely. This research approaches H fertilisation as a tactic, that is, a response to circumstances as they arise, rather than a strategy which ts followed in every season.

Results, 1981-84

Our earlier experiments were located on a well-drained red earth soil south of Harden (Table 1). These experiments were conducted in association with Dr Tony Fischer from 1981 to 1984 to compare the yield and protein responses to ammonium nitrate, broadcast either at sowing or at the terminal spikelet stage, about 1-2 weeks before the start of stem elongation. These showed excellent yield responses to N in 1981, 1983 and 1984. In the drought of 1982, there was little or no response, but in the following year, when crops were grown on the plots fertilised in 1982, we obtained virtually the same responses as those on nearby plots fertilised in 1983. In other words, in the drought of 1982 there was little or no uptake of the fertiliser and a strong residual effect in the following year.

The protein responses in these experiments were small, except in 1981, when both yield and protein content rose. In 1983 and 1984 the seasonal conditions in spring were so favourable that much of the fertiliser nitrogen was used in producing extra grain and none remained over for increased protein. In 1981 the protein response was associated with dry conditions during grain-filling. There was large between-season variability in protein content, largely explained by declining soil N as the cropping phase progressed, and by seasonal rainfall.

Table 1. Yield and protein responses to ammonium nitrate for Egret wheat grown on the same plots at “Kanimbla”, Harden, 1981-84. (The terminal spikelet stage occurred in mid-August).

Results, 1987

During 1987, with funding from the Wheat Research Council, Dr Fischer, Mr Anthony van Herwaarden and I conducted experiments on 30 sites throughout the Tablelands, Slopes and Riverina (Figure 1). We are grateful to AFL for analysing the samples for protein content. The 1987 season was relatively unfavourable for nitrogen response because of the dry spring. September rainfall in the eastern Riverina was lower than in 60-70% of years for which records are available. However, the period from mid-September to mid-October, which includes the important pre-flowering period, was drier than 80-90% of years.

Figure 1. Locations of 1987 CSIRO N fertiliser experiments indicated by •.

The yield responses to nitrogen averaged over all trials reflected the dry spring (Table 2). The most responsive crops gave profitable responses, at least to the lowest level of N fertiliser applied. However, many of the crops gave lower yields when N fertiliser was applied. This process, known as “haying-off” was associated with pinched grain and generally increased protein content. Some of the hayed-off crops were growing on acid soil or had symptoms of soil-borne diseases. However, the most common reason for haying-off appeared to be increased extraction of soil water by fertilised crops prior to flowering, so that there was less stored soil water for grain-filling. The increased nitrogen expressed itself as higher protein.

Table 2. Yield and protein responses of wheat to nitrogen as urea in the 1987 CSIRO experiments. DC 30 refers to stage 30 in the Decimal Code of wheat development, equivalent to the start of stem-elongation, which occurred from mid-August in the western trials to early September in the eastern trials.

Timing Of Fertiliser Application

The results of trials conducted since 1981 show no consistent advantage for applications from the time of sowing up to DC30. Individual experiments have shown advantages for one or other time, probably because rainfall soon after broadcasting usually gives high efficiency. However, for fertiliser applied before sowing, heavy rain can leach mineral N below the seedling roots. (See Appendix I for Decimal code description).

Nitrogen contained in compound fertilisers is much cheaper than pure forms of nitrogen such as urea or ammonium nitrate. For a clearly deficient paddock, compound fertiliser applied with the seed is the most efficient means of applying N. The maximum safe application of N with the seed is about 20 kg N/ha, costing about $3.00/ha. Although this amount is not much to lose if a crop does not respond, it must be remembered that extra nitrogen applied to a crop growing on a soil which is already high in N could lead to haying-off and yield loss in a dry spring.

Other factors to consider are the cost of aerial application if the land is too wet for traffic at DC30, and the delay in sowing when fertiliser is applied as a separate operation in autumn.

How To Identify Responsive Crops

In view of the 1987 results, it is as important to identify those crops for which yield will decrease with applied N, as it is to identify the crops for which yield and/or protein will increase. The various sources of information about likely responses differ in the availability, cost and reliability of the information and the advance warning time that they give.

Paddock history

Recent pasture history and the history of crop yield and protein content provides a first guide about likely yield responses to fertiliser nitrogen. However, a simple count of the number of years of pasture, cereal, grain legumes, etc, may be misleading if the vigour of the pasture or the yield of the crop is not considered. For example, the years of generally clover-dominant pastures from 1983 to 1986, associated with early autumn breaks, has probably led to above average build-up of soil nitrogen in Riverina farms, which is currently contributing to cereal yields. Similarly, poor clover-growing seasons in the mid-late 1970s may have been associated with large N responses found in the region in the early 1980s.

Soil tests

Tests of topsoil mineral-N taken it autumn also provide a guide to the likely yield responses to N fertiliser. However, the tests underestimate mineral-N if samples are taken during a dry autumn because the breakdown of organic matter depends on the soil being wet. They may also under-estimate mineral-N after heavy rainfall because nitrate may be leached below the sampling depth.

Crop-tissue tests

Tests of tissue nitrate concentration are probably the most reliable indicator of likely response, provided they are taken at a known developmental stage and during the morning. During 1987, CSIRO compared three such tests attheDC30 developmental stage and found them similar in accuracy. We found that fresh stem-bases contained nitrate-N concentrations 2.4 times those of the equivalent youngest leaves, as sampled by the Farmlab test. Dry stem-bases contained 6.5 times the nitrate-N concentration of equivalent fresh stem-bases.

In the 1987 experiments we found that crops containing less than 200-300 ppm nitrate-N in the fresh leaf tissue mostly gave positive yield responses to N. this was equivalent to nitrate-N concentrations of 500-700 ppm in fresh stem-bases and 3300-4500 ppm nitrate-N in dry stem-bases. As mentioned previously, the 1987 spring was unusually dry, and it is likely that the critical nitrate concentrations will be higher in most seasons.

Tiller counts

The number of tillers per unit of land area, also measured at DC30, seems to be a good indicator of the nitrogen status (provided plant number is within the normal range of 20-30 plants per metre of 7-inch row). In the 1987 CSIRO experiments, a tiller number of less than 110-120 per metre of 7-mci row was associated with yield responses to N, while numbers greater than 130-140 per metre of row were associated with yield declines.

Test strips

If N fertiliser is applied in strips across a paddock at or before sowing, and there are visible responses in crop colour or vigour before the DC30 stage, then it is likely that fertiliser applied at DC30 will produce a yield and/or protein response on the rest of the paddock.

Effects Of N-Fertiliser On Protein

Protein responses and yield responses tend to be alternatives. Protein response is likely to be small when yield response is large, for example after a wet spring. It is likely to be largest when yield responses are small or negative, for example after a dry spring or when fertiliser is applied to a high N crop. The experiments reported in Tables 1 and 2 do not support fertilisation for increased protein content. For example, 1 kg of N costing $0.80 applied, could, depending on seasonal conditions, produce potential responses of yield or protein as follows:

At current prices, the all-grain response gives a gross margin of about ($3.00-$0.80) $2.20. For the grain protein response to give a similar return, the price premium would have to be $10/tonne for each additional percent protein. At a more realistic premium of $5/tonne per additional percent protein, the gross margin would be (2 t/ha x 0.10% x $5 - $0.80) $0.20.

Conclusions

Nitrogen fertiliser has given profitable responses of yield and protein in the following conditions:

Research is continuing to define these conditions more closely, particularly about the latest possible timing for N application, the soil water and the crop nitrogen status.

Our research so far has not emphasised the amount of N to apply, because the optimum varies so much from year to year. If any N is to be applied, it should be at least 20 kg/ha, because it is difficult to be certain, with amounts less than this, whether responses are due to the effect of the fertiliser or to random variations. In addition, small applications can lead to reductions in protein when growing conditions are favourable. Large amounts - more than 60-80 kg N/ha - are risky because of haying off. Between these limits, higher applications are justified when crop measurements easily meet the above conditions.

Appendix 1 Zadoks’ decimal code for the growth stages of cereals