Call to action/take home messages

• Nitrogen (N) contributed by legumes is an important component of N supply to subsequent cereal crops, yet few Australian grain-growers routinely monitor soil mineral N before applying N fertiliser.
• Data collected from 16 dryland experiments conducted in eastern Australia from 1989–2016, showed soil mineral N (nitrate and ammonium) measured in autumn following legumes in the expected rooting zone (1.2m) of a subsequent wheat crop, was on average 35± 20 kg N/ha (n = 26) higher than after a previous wheat, barley or canola crop.
• The additional soil N availability was calculated to be equivalent to 0.15 ± 0.09 kg N/ha per mm summer fallow rainfall, 9 ± 5 kg N/ha per t residual legume shoot dry matter/ha and 18 ± 9 kg N/ha per t/ha grain harvested, representing 28 ± 11% total legume residue N. It was proposed that these 4 measures could be used as potential rules-of-thumb by grain-growers and their advisors to bench-mark the likely additional soil mineral N provided by pulse crops.
• The apparent recovery of legume residue N by wheat averaged 30 ± 11% for 16 legume treatments in a subset of eight experiments, which could be considered as a further rule-of-thumb to indicate the relative value of legume N to a following wheat crop.
• By comparison, the apparent recovery of fertiliser N in the absence of legumes in two of these experiments represented 64 ± 16% of the 51–75 kg fertiliser-N/ha supplied as top-dressing applications at stem elongation just prior to peak crop N demand.

Introduction

The concentrations of soil mineral (i.e. nitrate+ammonium) nitrogen (N) measured prior to sowing a cereal in dryland farming systems depends upon the relative balance between factors that either favour the build-up, or result in a reduction, of available soil N.

An accumulation of soil N can arise from the combined contribution of:

1. the carry-over of any mineral N not utilised by the previous crop (spared N; Herridge et al. 1995), and
2. the total N mineralised from above- or below-ground plant residues and the soil organic N pool by soil microbes (total N released).

Factors that reduce soil mineral N on the other hand include:

1. the extent to which crop residues or management influences the use of available soil N by soil microbes for growth (N immobilised),
2. assimilation of available N by weeds (weed N-uptake), and
3. leaching, erosion and gaseous losses (N lost; see papers cited by Peoples et al. 2017).

Consequently, the concentrations of soil mineral N observed at the beginning of a growing season represent the net effect of all these variables according to the following conceptual Equation [1]:

Soil mineral N = [(spared N)+(total N released)] – [(N immobilised)+(weed N-uptake)+(N lost)]

Equation [1]

Each of these processes can be influenced by:

1. the duration of the period of fallow between the end of one cropping season and the beginning of the next since this defines the time available for weed growth, and mineralisation or loss processes to occur,
2. rainfall amount and distribution during the fallow period, as soil moisture regulates soil microbial activity, determines the risk of N losses, and affects weed germination and growth, and
3. the quantity of plant residues remaining at the end of the previous growing season and the N content (or C:N ratio “quality” attributes) of those residues. Residue N content determines the amount of N potentially available for mineralisation, and C:N ratio influences whether a net release or immobilisation of mineral N occurs.

Many researchers have observed improved grain yields and/or N uptake by cereal crops grown after legumes compared to cereal-after-cereal sequences (Angus et al. 2015). This is usually attributed to elevated availability of soil mineral N and healthier crops recovering more soil N following legumes. However, few Australian dryland grain-growers routinely conduct pre-season soil testing, or monitor soil N fertility in all the cropping paddocks across their farms.

Computer models have been developed that can simulate the complex soil N and water dynamics in rainfed cropping systems, which have been used to investigate N and water use efficiency in different environments and soil types in response to agronomic practice. Grain-growers can now access the outputs from these sophisticated research tools through subscriptions to internet-based decision support services to predict soil N availability and grain yield prospects in response to different N fertilisation scenarios and seasonal conditions (Hochman et al. 2009). However, the costs associated with obtaining useful input data (including soil mineral N) to parametrise the model, and the soil characterisation required to improve the accuracy of predictions, remain major barriers to the widespread adoption of such technologies.

Apart from an Excel spreadsheet N budgeting approach recently proposed by Herridge (2017), there have been few convenient or simple ways by which farmers and their advisors can benchmark the expected net effect of including a legume in a cropping sequence on the accumulation of soil mineral N prior to sowing the next crop in the absence of soil tests, or to assess how much N from the preceding legume might subsequently be assimilated by a following wheat crop (Triticum aestivum).

This paper presents crop production and soil N data from 16 dryland cropping experiments undertaken at different locations across eastern Australia between 1989 and 2016 which aimed to quantify the N benefits derived from including pulses in cereal dominated cropping sequences. In these investigations soil mineral N concentrations immediately prior to sowing wheat for the next growing season were compared for soils after legumes and after various non-legume control treatments such as wheat, barley (Hordeum vulgare), or canola (Brassica napus). The productivity and N accumulated by the following wheat crop was also quantified. These studies differ from many previous investigations, as the majority of the experiments where conducted in farmers' fields in partnership with individual grain-growers and local grower associations.

The collated crop and soil data from these experiments were used to assess the potential of different approaches to develop simple ‘rules-of-thumb’ that could be used by farmers and their advisors as guides to the anticipated soil mineral N benefits derived from legumes harvested for grain (28 comparisons). The apparent recoveries of the legume N by a following wheat crop were calculated for eight experiments (16 estimates), and the recoveries of legume N were directly compared to the apparent uptake of fertiliser N applied to wheat when grown after a preceding wheat or canola crop in two experiments (three estimates).

Materials and methods

Full experimental details are described in Peoples et al. (2017), but briefly the studies described here were undertaken in partnership with FarmLink, Riverine Plains, Birchip Cropping Group (BCG), Mackillop Farm Management Group (MFMG), Mallee Sustainable Farming (MSF), or Eyre Peninsula Farming Systems grower groups, and/or in collaboration with the NSW Department of Primary Industries (NSW DPI), Victorian Department of Economic Development, Jobs, Transport and Resources (formerly Vic DPI), or the South Australian Research and Development Institute (SARDI). The experimental sites were located on different soil types across northern and southern New South Wales (NSW), Victoria (Vic), and South Australia (SA; Table 1).

Legume treatments included field pea (Pisum sativum), chickpea (Cicer arietinum), lupin (Lupinus angustifolius), faba bean (Vicia faba), lentil (Lens culinaris), and vetch (Vicia sativa). The non-legume control was wheat in nine experiments, barley in one (Breeza 1997), and canola in two studies (Tamworth 2009 and Loxton 2015). Three experiments included both wheat and canola as non-legume controls (Hopetoun 2009, Junee Reefs 2011, and Naracoorte 2011), but only the wheat data were used for subsequent calculations of soil mineral N benefits.

During the 25 years, the various cropping treatments experienced both above- and below-average rainfall during the growing season (April-October) and in the post-harvest summer-autumn fallow (Table 1). Most experimental plots consisted of at least six crop rows of plants sown 0.2–0.3 m apart in a randomized complete block design in plots 10 – 35 m long with three or four replicates sown in either late-April to early-May (legume crops and canola) or mid to late-May (wheat). In all cases weeds were controlled with registered herbicides using recommended commercial practices during the summer-autumn fallow between crops and during the growth of the following wheat so neither the accumulation of soil mineral N nor the subsequent uptake of N by wheat were confounded by the presence of weeds.

While wheat was subsequently sown into all legume and non-legume plots in May of the following growing season to quantify the impact of legumes for all 16 experiments, full N analyses of the wheat grain and stubble were only undertaken for 11 studies. In three of these experiments (one at Tamworth 2010 and two at Culcairn 2011), wheat growth was not constrained by N availability and there were no responses to either legume pre-treatments or applications of fertiliser N. The N uptake data presented in the current paper come from the remaining eight experiments and represented 16 pulse crops where significant wheat responses were observed (North Star 1990 and 1993, Breeza 1998, Gundibindjal 2001 and 2002, Junee Reefs 2012, Wagga Wagga 2012, and Loxton 2016).

The application of fertiliser N to wheat after wheat was included as an additional treatment to allow direct comparisons of the apparent recoveries of legume and fertiliser N in only two of these experiments (Junee Reefs and Wagga Wagga 2012). At Junee Reefs the wheat sown into the 2011 legume treatment plots received starter fertiliser of 25 kg/ha mono-ammonium phosphate (2.5 kg N/ha), and was top-dressed with 46 kg/ha urea-N at stem elongation. The 2011 wheat and canola plots were each split into 2 x 10 m sub-plots with one half receiving exactly the same N fertiliser treatment as the legume plots, and the other half being top-dressed with an additional 51 kg urea-N/ha (i.e. supplied with a total 97 kg urea-N/ha) just prior to stem elongation. At Wagga Wagga the wheat sown into the 2011 legume plots received no N fertiliser while the 2011 wheat plots received either no fertiliser or 75 kg fertiliser-N/ha (25 kg urea-N/ha at sowing with a further 50 kg N/ha at stem elongation).

Measurements

Soil mineral N - Treatment plots were sampled for soil mineral N immediately before sowing wheat in the following growing season. The soil mineral N data presented here, and the values subsequently used for calculations, were normalised for the anticipated rooting depth of wheat. At Mildura, Naracoorte, Minnipa and Loxton this represented 0-0.6 m since inhospitable subsoils (high salinity, sodicity or alkalinity), or rocks at these locations effectively restricted root exploration and/or prevented soil sampling deeper in the profile. At all other sites soil mineral N data were calculated to 1.2 m which approximates the average rooting depth of wheat in most eastern Australian soil types where there aren’t major subsoil constraints to root growth (average maximum rooting depth of wheat = 1.29 m ± a standard deviation of 0.3 m, n=36; Kirkegaard and Lilley 2007).

Examination of the distribution of soil mineral N in the soil profiles of treatments at eight of the experimental sites indicated that mineral N in the top 0.6 m was on average 68±9% of the total mineral N to 1.2 m (n=27), which in turn represented 88±5% of the soil mineral N detected to 1.5 m depth (data not shown). The distribution of soil mineral N in soil profiles following legumes (0-0.6 m equivalent to 66±8% of soil mineral N 0-1.2 m, n=17) was found to be similar to that observed after non-legumes (0-0.6 m equivalent to 71±9% of soil mineral N 0-1.2 m, n=9; data not shown).

Calculations

Total plant N - Since N associated with, or derived from, nodules and roots can represent a significant source of N for subsequent mineralisation and can play a major role in determining the N-balances of cropping systems (Peoples et al. 2009), below-ground N was estimated for each experimental treatment. For example, it was assumed that 25% of the whole plant N was below-ground for lupin and therefore a root-factor of 1.33 was used to convert shoot N to total plant N as described by Unkovich et al. (2010). Root factors used for the other crops were 1.47 for both field pea and vetch, 1.52 for faba bean, 1.56 for lentil, and 1.82 for chickpea. Below-ground N was also estimated for non-legume treatments using root-factors of 1.43 for canola and 1.52 for wheat, respectively.

The N accumulated in shoot and roots of grain crops at the end of the growing season was calculated as:

Total crop-N = [(vegetative DM) x %N/100]+[(grain DM) x %N/100] x root-factor

Equation [2]

Residue N - The total amounts of N remaining in crop vegetative residues and roots following grain harvest at the end of the growing season was calculated as:

Total residue N = (total crop N) – (grain N removed)

Equation [3]

Soil mineral N benefits of legumes - The net effect of growing legumes on available soil N (i.e. the integrated effect of all the factors described in Equation [1]) were determined from the differences in soil mineral N data after legumes and non-legume controls measured just prior to sowing wheat across all treatments in autumn the following year. The observed soil mineral N benefits derived from legumes were expressed in four different ways (Equations [4] to [7]) as the basis of developing simple predictive relationships that potentially could be used by grain-growers to assist decision-making about fertiliser N applications following legumes in their cropping sequence:

(a) Mineral N benefit per mm fallow rainfall (kg N/ha per mm)
= [(mineral N_{after legume}) – (mineral N_{after non-legume})]/(fallow rain)

Equation [4]

Where fallow rainfall (mm) represented the cumulative total between legume grain or BM harvest and sowing of the following wheat crop. In most studies where legume grain crops were grown, the fallow period began in late November or early December and finished in May.

(b) Mineral N benefit per t shoot residue DM (kg N/t shoot residue DM)
= [(mineral N_{after legume}) – (mineral N_{after non-legume})]/(legume shoot residue DM)

Equation [5]

Where shoot residue DM = (peak biomass DM) – (grain yield)

(c) Mineral N benefit per t legume grain yield (kg N/t legume grain yield)
= [(mineral N_{after legume}) – (mineral N_{after non-legume})] /(legume grain yield)

Equation [6]

(d) Soil mineral N benefit expressed as % total residue N
= 100 x [(mineral N_{after legume}) – (mineral N_{after non-legume})] /(total legume residue N

Equation [7]

Where total legume residue N was determined from Equation [3]

Apparent recovery of legume and fertiliser N - The apparent recoveries of legume or fertiliser N by the first wheat crop grown after the legume and non-legume (wheat, barley or canola) treatments were calculated as:

Apparent recovery of legume N (% total residue N)
= 100 x [(wheat N_{after legume}) – (wheat N_{after non-legume})] /(total legume residue N)

Equation [8]

Where wheat N represented an estimate of total N in the shoots + roots calculated as described in Equation [2].

Apparent recovery of fertiliser N (% additional N applied)
= 100 x [(wheat N uptake N_R2) – (wheat N uptake N_R1)] /(N_R2 – N_R1)

Equation [9]

Where wheat N represented an estimate of the total N present in shoots + roots, and N_R1 or N_R2 represent two different rates of fertiliser N applied to wheat grown after a non-legume. In the case of the Junee Reefs experiment N_R1 = 49 kg N/ha, and N_R2 = 100 kg N/ha, so the recovery of N referred to just the additional top-dressed 51 kg fertiliser-N/ha. At Wagga Wagga N_R1 = 0 kg N/ha and N_R2 = 75 kg N/ha.

Statistical analyses - Throughout the paper mean values are presented along with ± standard deviations to provide measures of variability. Analysis of variance was undertaken of the soil mineral N, crop DM and N data for each experimental site/year to provide least significant difference (LSD) determinations (P<0.05). Regression analysis was used to evaluate the potential value of different prospective relationships to predict soil mineral N benefits, and to explore the relationship between legume grain yield and total legume residue N.

Results

Legume growth and N accumulation

Data for shoot residue and grain DM, N accumulation, and calculations of net inputs of total residue N collated from all 16 experiments conducted in different locations, years, soil types and environments across the eastern Australian dryland cropping zone are presented in Table 2. Field pea was the most frequently used legume treatment (included in eight experiments), while vetch grown for grain was used the least (only once).

Large differences in growing season rainfall (GSR, 92–447 mm; Table 1) for the different experiments resulted in the accumulation of a wide range of legume shoot residue DM (1.4–9.8 t/ha), and grain yields (0.5–3.9 t/ha; Table 2). The harvest index (i.e. grain yield as a proportion of total above-ground DM calculated from Table 2) for the different legume crops grown for grain ranged from 0.17–0.20 (field pea and chickpea at Loxton in 2015 which experienced a heat wave during grain-filling) to 0.56–0.58 (lentil at Junee Reefs 2011 and faba bean at Breeza in 1997), and were 0.26–0.40 for the remaining 20 of the 24 crops (mean 0.34±0.09 across all 28 crops), values within the range commonly observed for Australian legume crops.

The N contents of the legume shoot residues remaining after grain harvest was higher, and C:N ratios were typically lower (0.9–1.4% N, C:N ratios 33–56), than either canola (0.7–0.8% N, C:N ratio 50–60), or wheat and barley stubble (0.3–0.6% N, C:N ratio 75–160). The estimates of net inputs of legume N associated with the vegetative residues and nodulated roots at the end of the growing season ranged 52–330 kg N/ha (Table 2). Comparisons of the contributions by individual grain crops in different experiments suggested that the largest net inputs of total residue N were often achieved by faba bean and lupin (Table 2).

Trends in available soil N

Soil mineral N concentrations measured after the non-legume controls in autumn varied widely from 36–141 kg N/ha, which presumably reflected key differences in inherent background fertility at the various study sites and rainfall during the preceding fallow period, but were on average 68±25 kg N/ha after wheat (n=13), 59 kg N/ha after barley (n=1) and 90±30 kg N/ha canola (n=5; Table 3).

Soil mineral N was significantly greater (P<0.05) after legumes than the non-legume controls for 26 of the 28 legume crops (non-significant data occurred following field pea and lentil at Loxton in 2015; Table 3). The difference in autumn soil mineral N after legume crops and non-legume treatments ranged from 11–89 kg N/ha (mean 35±20 kg N/ha, n=26; Table 4).

Soil mineral N benefits

The soil mineral N benefits calculated for the 31 legume treatments where autumn measures of soil mineral N were significantly different from non-legume controls, were equivalent to 0.03–0.36 kg N/ha per mm fallow rainfall (mean 0.15±0.08), 3–20 kg N/t above-ground residue DM (mean 9±4), representing 13–48% of the N remaining in vegetative and below-ground legume residue at the end of the previous growing season (mean 27±10%; Table 4). Despite large differences in crop performance, inputs of residue N, GSR and fallow rainfall across the 16 experiments (Tables 1 and 2), the average relationships derived for soil mineral N benefits calculated on the basis of either fallow rainfall, residue DM or N were remarkedly similar between legume species (Table 4).

In recognition of a previous observation that the size of the effects of lupin on subsequent wheat in Western Australia was related to the grain yield by the lupin crop (Seymour et al. 2012), our data were further examined to ascertain whether there might also be a useful relationship between legume grain yield and subsequent soil mineral N benefit. Values ranged from 7–46 kg additional soil mineral N/t legume grain harvested, with estimates for 17 of the 26 grain crops of between 11 and 26 kg N/t grain. The mean relationship between soil mineral N benefit and grain yield across all 28 grain crops was 18±9 kg mineral N/t legume grain (Table 4).

In the absence of an independent data set to test the relative value of the mean relationships between soil mineral N benefits and either fallow rainfall, residue biomass, residue N, or grain yield calculated above as prospective predictive tools, the different approaches were evaluated by using each mean relationship to estimate soil mineral N from the original experimental biophysical data. Those predictions were then compared to the actual mineral N benefits measured across the 16 experiments. The proportion of the variance explained by the various simple relationships were determined from r² assessments obtained from regression analyses of predicted soil mineral N benefits vs actual data. On the basis of such analyses it was suggested that:

• 0.15 x mm fallow rainfall explained 24% of the variance
• 9 x t shoot residue explained 35% of the variance
• 18 x t grain explained 27% of the variance
• 28% x total legume residue N explained 57% of the variance.

Other relationships based on fallow rainfall combined with either grain yield, shoot DM or residue N were also examined, but in all instances comparisons of predicted vs actual indicated that the ability to calculate subsequent soil mineral N was not improved. Indeed regression analysis of all approaches that combined more than one parameter suggested that less than 10% of the variance was explained by the process.

It was concluded that as a decision-making tool the individual relationships could only provide a rough approximation of the expected additional soil mineral N after a legume. Of the four different expressions, predictions based on legume residue N were likely to be the most reliable because it explained the largest fraction (i.e. 57%) of the observed variation. Unfortunately, residue N is a particularly difficult parameter for farmers to measure directly. Since grain yield is usually related to above-ground residue biomass, the data were examined to ascertain whether grain yield might also provide a guide to the amount of total residue N remaining at the end of the legume growing season.

Analyses of the experimental data were confounded by the clustering of yield below 1.5 t/ha and above 2.5 t/ha (solid circles; Fig. 1), and an apparent outlier (faba bean at Culcairn 2010, solid triangle; Fig. 1), so data for a further 20 legume crops obtained from previously published and unpublished sources for another 16 experimental studies conducted in Victoria and NSW between 1995 and 2014 (open circles; Fig. 1) were included to improve the prospects of devising a relationship between grain yield and residue N. These additional data exhibited a degree of variation in residue N across similar yield values, presumably reflecting differences in seasonal conditions and harvest index. This was despite restricting the selection of data to legume crops with a similar range in harvest index to that measured for the majority of treatments in the original 16 experiments (i.e. between 0.26 and 0.40).

Regression analysis of the combined data in Figure 1 indicated that total residue N = [54 + (30 x legume grain yield)] (r² = 0.56). Therefore, another more slightly complex rule-of-thumb for estimating the additional soil mineral N derived from crop legumes would be: 0.28 x [54 + (30 x t legume grain harvested)].

Apparent recovery of legume N by the following wheat crop

Pre-sowing soil mineral N following all legume and non-legume treatments, and subsequent measures of wheat total N uptake from eight of the 16 experiments are presented in Figure 2 and Table 5. Wheat N at harvest failed to exceed soil mineral N at sowing (i.e. fell below the 1 : 1 line depicted in Fig. 2) at only two locations, North Star 1993 and Breeza 1998. The lower than expected wheat uptake at Breeza could have been associated with potential denitrification losses of N from the soil as a result of waterlogging due to the record high rainfall experienced during the growing season (866 mm compared to 334 mm long-term average, Tables 1 and 5).

Examination of the data across all trials and treatments except Junee Reefs (where applications of fertiliser N would have confounded the calculations), indicated that on average wheat accumulated 1.34±0.67 kg N/ha for every 1 kg/ha of soil mineral N present at sowing. Given that the ratio of wheat N uptake : pre-sowing soil mineral N exceeded 2.5 : 1 at one location (Gundibindjal 2001), and fell between 1.1–1.7 : 1 across all other experiments and treatments, it was concluded that soil N mineralisation during crop growth play an important role in contributing N for wheat uptake at most locations in most years (Fig. 2).

All legume treatments significantly increased (P<0.05) above-ground wheat biomass (by 0.6–4.5 t/ha mean 2.4±1.2 t/ha) and total N uptake (by 8–86 kg N/ha, mean 38±23 kg N/ha), but grain yield was significantly greater (P<0.05) than measured in the neighbouring wheat crops grown after wheat, barley, or canola following 12 of the 16 legume pre-cropping treatments (Table 5). The increased N uptake represented apparent recoveries of legume N by wheat equivalent to 12–48% of the residue N estimated to be remaining at the end of the previous growing season (mean 30±10%; Table 5).

Comparisons of recoveries of legume N with fertiliser N

The effect of legumes on wheat N uptake could only be directly compared to fertiliser N in two experiments. In these, applications of fertiliser N increased total N uptake by wheat grown after canola or wheat by 25 and 31 kg N/ha, respectively at Junee Reefs, or by 61 kg N/ha, at the Wagga Wagga site (Table 6). The apparent recoveries of the fertiliser N were calculated to be equivalent to 49–81% of the fertiliser N supplied (mean 64±16%; Table 6). While these determinations of apparent recoveries of fertiliser N by wheat were somewhat higher than the recovery of legume N in the same experiments (mean recovery of 30±5% of residue N from nine legume treatments), the additional quantity of N accumulated by wheat in response to fertiliser N was lower than observed after all legume treatments (38–84 kg N/ha) at Junee Reefs.

Discussion

Effect of legumes on soil mineral N

In keeping with the findings of other previous studies undertaken in Australia and elsewhere in the world, concentrations of soil mineral N were significantly greater following legumes compared to after non-legumes. In absolute terms the magnitude of the effect of legumes varied across locations and years (Tables 3–5) reflecting the influence of rainfall within the growing season on biomass production, and rainfall during the subsequent fallow. Improvements in soil N availability tended to be lowest after lentil and vetch, and highest after faba bean (Table 4), which was consistent with faba bean’s reputation as a species with a capacity for the accumulation of high biomass and the symbiotic fixation of large amounts of atmospheric N₂ (Peoples et al. 2009).

Reasons for the unusually high soil mineral N following the chickpea crop at Junee Reefs, compared to the lupin or lentil (see Tables 3 and 5), remains unresolved. Some of the additional soil mineral N could have arisen from chickpea’s tendency to be less efficient at recovering soil mineral N during growth than wheat (Herridge et al. 1995). Unfortunately, soil mineral N was not determined following chickpea grain harvest so the presence of unutilised ‘spared’ mineral N cannot be confirmed in the Junee Reefs experiment. Chickpea is also known to partition a larger proportion of the plant N below-ground in nodules than most other legume species. Nodules tend to have high N contents (4–7% N) and low C:N ratios which is conducive to rapid decomposition rates, so it is possible that the observed effect of chickpea on soil N dynamics could have reflected a higher nodule load combined with high rainfall during the summer-autumn fallow period to stimulate microbial activity and mineralisation processes (Tables 1 and 2).

Soil mineral N benefits derived from legumes

Pre-season deep soil testing for soil mineral N (0-60cm) is considered to be the most accurate data that grain-growers and their advisors can utilise for decision-making about fertiliser N applications before or at sowing. Similarly deep soil sampling in-crop could be used to better inform further top-dressing decisions in light of weather forecasts for the remainder of growing season and grain prices. Unfortunately, few Australian farmers routinely monitor soil N availability in all their fields prior to cropping, and the logistics and cost of in-crop soil sampling and analysis for soil mineral generally precludes adoption of this practice by most growers.

In the absence of pre-season soil testing, the most valuable information that could be provided to farmers would be some means of predicting the soil mineral N prior to sowing wheat which could be used as a basis for decisions about rates of N fertiliser to apply to meet target yields and grain quality. The large location and year variability in the soil mineral N data observed following the non-legume control treatments in all 16 experiments (Table 3) exemplifies the underlying influence that different soil types, soil organic N contents, and preceding rainfall can have on the end result, and emphasises the challenge in devising such a tool. However, it was hoped that through the interrogation of data collated from 25 years of cropping systems studies that it might be possible to identify some simple relationships that could be utilised to benchmark the likely incremental improvement in soil mineral N as a result of growing a N₂-fixing legume rather than a non-legume.

Two key parameters used in the calculations of soil mineral N benefits of legumes (Table 4) that all farmers routinely monitor or measure are rainfall and grain yield. Therefore, of the four potential measures of mineral N benefits examined here, relationships described as 0.15 kg N per mm fallow rainfall, and 18 x t legume grain yield, are the ones that could most easily be applied by farmers. Given the relationship between grain yield and shoot DM reported here (i.e. average harvest index = 0.34±0.09), growers might also be able to estimate shoot residue biomass by assuming grain yield generally represents around one-third of above-ground biomass (i.e. shoot residue DM = 2 x t legume grain yield).

By combining this knowledge with the estimate of 9 kg additional mineral N/t shoot residue DM the added contribution of crop legumes to soil mineral N could also be calculated to approximate: 18 x t legume grain harvested. However, the most reliable estimate of soil mineral N benefit is likely to be calculated on the basis of net inputs of total residue N after legume cropping (28% residue N; Table 4). By utilising the relationship between grain yield and legume residue N presented in Figure 1, soil mineral N benefit calculated on a % N residue basis could also be re-expressed in a form that farmers can extrapolate from grain yield as: 0.28 x [54 + (30 x t legume grain harvested)].

Apparent recoveries of legume and fertiliser N by wheat

While the observed range of estimates of the apparent recovery of legume N by wheat was large (12–48%), 16 of the 20 determinations fell between 19-39%, and the mean represented 30±10% across all legume treatments (Table 5). This provides new insights into the value of including legumes in a cropping sequence in the rainfed grains belt of eastern Australia. Applying a similar approach to the one described above to estimate total residue N from grain yield, farmers could also calculate the likely recovery of legume N by a following wheat crop as representing: 0.30 x [54 + (30 x t legume grain harvested)].

The data strongly suggested that for most crops wheat’s enhanced N uptake reflected improvements in N availability both prior to sowing wheat and during crop growth. A reduced incidence of cereal root disease following a legume break crop was also likely to have assisted wheat’s ability to more fully exploit the soil mineral N pool.

The mean apparent total recovery of fertiliser N by wheat calculated from two experiments conducted in southern NSW (64%; Table 6) was comparable to the mean value previously reported for wheat in Australia (38% on a shoot basis, which is equivalent to 58% when re-calculated as total shoot + root N, n=42; Krupnik et al. 2004).

That the apparent recoveries of fertiliser N were higher than calculated for legume residue N in the same studies (30%, Table 5) was not surprising given that either two-thirds (Wagga Wagga), or >90% of the fertiliser N applied (Junee Reefs) was supplied at the stem elongation phase of crop development immediately prior to a period of high plant demand for N, and that only a fraction of the organic legume N would have become available for crop uptake (Peoples et al. 2009). However, it should be noted that the soil mineral N generated after legume cropping should be just as effective a source of N to support wheat growth as N released from fertiliser, and that legumes contribute a large pool of organic N that becomes available for the benefit of more than one subsequent crop and sustains to the long-term fertility of the soil.

Conclusions

In the absence of any direct measures of soil mineral N, the four predictive relationships reported here could be used by grain-growers and their advisors in the dryland cropping areas of eastern Australia to estimate the additional pre-sowing soil mineral N following legume grain crops as they can be calculated directly or indirectly from readily-available information such as rainfall and legume grain yield. Growers could also potentially apply the relationship developed in the current paper to estimate N remaining in legume residues from grain yield to benchmark the subsequent recovery of legume N by a wheat crop.

Recognising that none of the relationships will provide perfect predictions, and acknowledging that there are potential consequences in over- or under-estimating available N and wheat N uptake, it is recommended that all five expressions be used as a means of providing some measure of uncertainty. The risks of either under-fertilising in a wet growing season and not realising yield potential, or over-supplying fertiliser N to wheat when there is a prolonged period of drought during spring which can lead to yield reductions due to haying-off, would also be lowered, and the efficiency of fertiliser N uptake improved, if decisions on applications of N fertiliser can be delayed until later in the growing season when there is more confidence about anticipated rainfall.

More experimentation following the accumulation of soil mineral N and crop recovery of N after legumes still needs to be undertaken across different soil types, farming practices and years to evaluate and validate the preliminary simple predictive relationships proposed here and to further refine them. Studies should also be initiated to explore whether similar approaches to those described in the current paper might usefully deployed in dryland grain production systems beyond eastern Australia.

Figure 1. Relationship between grain yield of various legume crops and total (above- + below-ground) legume residue N remaining at the end of the growing season using data from the 16 cropping systems experiments presented in Table 2 (●), together with additional published and unpublished data for 20 legume crops from 16 other studies conducted in Victoria and NSW between 1995 and 2014 (○). The line of best fit calculated without the apparent outlier (faba bean at Culcairn; ▲) was described by: Total residue N = [54 + (30 x legume grain yield)] (r² = 0.56)

Figure 2. Relationship between pre-sowing soil mineral N and wheat total N uptake across eight different experiments conducted at two locations in northern NSW (North Star 1990 and 1993,; Breeza 1998,■), three in southern NSW (Gundibindjal 2001 and 2002,▲; Junee Reefs 2011, ; Wagga Wagga 2011,) and one site in SA (Loxton 2016,▼). The symbols represent data for non-legumes (open) and legumes (closed) grown prior to sowing wheat.
The dashed line indicates the 1 : 1 line.

^a Soil type as described by Isbell (2016). Soil pH below 6.0 were determined 1:5 in CaCl₂ and above 6.0 were determined 1:5 in water.

^a Above-ground data adjusted to include an estimate of below-ground N using Equation [2].
^b Calculated using Equation [3].
^c Nd indicates no data available.

^a Year in which the crop was sown, which followed the fallow period after the legume or non-legume crops were harvested.
^b Autumn measures of soil mineral N detected to a depth of either 0-0.6 m (Naracoorte, Minnipa, Mildura and Loxton), or 0-1.2 m (all other sites).
^c LSD = Least significant difference.
^d There are two sets of data for wheat at Culcairn as both the lupin and faba bean experimental plots, which were located in separate parts of the same field, included wheat controls.

^aAdditional soil mineral N detected to a depth of either 0-0.6 m (Naracoorte, Minnipa, Mildura and Loxton), or 0-1.2 m (all other sites) following legumes compared to wheat, barley or canola treatments calculated from data shown in Table 3.
^b For convenience data from one lentil crop and a single vetch crop grown for grain were pooled for analysis.
^c The data from field pea and lentil crops grown at Loxton in 2015 were not included as soil mineral N measured in 2016 following these two treatments were not significantly different from the non-legume control.

^a Soil mineral N data from Table 3. Estimates of total wheat N uptake were calculated using Equation [2], and apparent recovery of legume N using Equation [8].
^b Data in brackets represent cumulative growing season rainfall (GSR) experienced by the wheat crop.

^a Total wheat N uptake derived from above-ground N data assuming 34% of the total crop N was below-ground using Equation [2].
^b Calculated using Equation [9]. Note there was no nil N fertiliser control at Junee Reefs so it was not possible to estimate N recovery for the 49 kgN/ha treatment.

Acknowledgements

The collaborating grower groups and farmer co-operators are thanked for participating in the research and facilitating access to on-farm sites to undertake experimentation. We are grateful to the Grain Research & Development Corporation (GRDC) and the South Australian Grains Industry Trust (SAGIT) for financial support.

References

Angus JF, Kirkegaard JA, Hunt JR, Ryan MH, Ohlander L, Peoples MB (2015). Break crop and rotations for wheat. Crop & Pasture Science 66, 523-552.

Herridge DF (2017) Validation of NBudget for estimating soil N supply in Australia’s norther grains region in the absence of soil test data. Soil Research 55, 590-599.

Herridge DF, Marcellos H, Felton WL, Turner GL, Peoples MB (1995) Chickpea increase soil-N fertility in cereal systems through nitrate sparing and N₂ fixation. Soil Biology & Biochemistry 27, 545-551.

Hochman Z, van Rees H, Carberry PS, Hunt JR, McCown RL, Gartmann A, Holzworth D, van Rees S, Dalgliesh NP, Long W, Peake AS, Poulton PL, McClelland T (2009) Re-inventing model-based decision support with Australian dryland farmers. 4. Yield Prophet® helps farmers monitor and manage crops in a variable climate. Crop & Pasture Science 60, 10567-1070.

Isbell RF (2016) The Australian Soil Classification – Second Edition. CSIRO Publishing, Clayton South, Australia.

Kirkegaard JA, Lilley JM (2007) Root penetration rate – a benchmark to identify soil and plant limitations to rooting depth in wheat. Australian Journal of Experimental Agriculture 47, 590-602.

Krupnik TJ, Six J, Ladha JK, Paine MJ, van Kessel C (2004) Assessment of fertilizer nitrogen recovery efficiency by grain crops. In: AR Mosier, JK Syers, JR Freney (Eds.), Agriculture and the Nitrogen Cycle: Assessing the Impacts of Fertilizer Use on Food Production and the Environment, Scientific Committee on Problems of the Environment (SCOPE) report 65. Island Press, Washington DC, USA, pp. 193-207.

Peoples MB, Brockwell J, Herridge DF, Rochester IJ,  Alves BJR, Urquiaga S, Boddey RM, Dakora FD, Bhattarai S, Maskey SL, Sampet C, Rerkasem B, Khan DF, Hauggaard-Nielsen H, Jensen ES (2009) The contributions of nitrogen-fixing crop legumes to the productivity of agricultural systems. Symbiosis 48, 1-17.

Peoples MB, Swan AD, Goward L, Kirkegaard JA, Hunt JR, Li GD, Schwenke GD, Herridge DF, Moodie M, Wilhelm N, Potter T, Denton MD, Browne C, Phillips LA, Khan DF (2017) Soil mineral nitrogen benefits derived from legumes and comparisons of the apparent recovery of legume or fertiliser nitrogen by wheat. Soil Research 55, 600-615.

Seymour M, Kirkegaard JA, Peoples MB, White PF, French RJ (2012) Break-crop benefits to wheat in Western Australia—insights from over three decades of research. Crop & Pasture Science 63, 1–16.

Unkovich MJ, Baldock J, Peoples MB(2010) Prospects and problems of simple linear models for estimating symbiotic N₂ fixation by crop and pasture legumes. Plant & Soil329, 75-89.

Contact details

Mark Peoples
CSIRO Agriculture & Food
Black Mountain Science & Innovation Park, GPO Box 1700 Canberra, ACT 2601, Australia
Ph: 02 6246 5447
Email: mark.peoples@csiro.au