Yield gaps in mungbean crops across the northern grains region
Take home messages
- Mungbean crops with low observed yield and high yield gaps were those with a low harvest index, which was not always associated with low crop biomass
- Management factors found to significantly increase yield were narrow row spacing (<50 cm) and crops sown on a fallow rather than double cropped
- Maximum water use efficiency (WUE) of approximately 7.5 kg/ha/mm of available water (in-crop rain + starting soil water) were found across the data set. Differences in starting water at this WUE explained observed yield differences between fallow and double crop mungbeans
- 35% of crops achieved > 80% of their water-limited yields, while 36% yielded < 60% of the water-limited yield potential. Nearly half of all monitored crops had yield gaps > 500 kg/ha
- No single biotic or abiotic factor was found to be associated with low mungbean observed crop yields or high yield gaps. However, 88% of crops with a high yield gap (i.e. < 65% water-limited yield) had either P. thornei present at > 3/g soil, or maximum temperatures > 39°C during flowering, or soil nitrate levels below 65 kg N/ha
- Consider soil tests for nutrient status and root lesion nematodes before sowing mungbeans to avoid risks of poor crop performance.
Mungbean crop yields in sub-tropical Australian farming systems are highly variable and the risk of low yields leads to grower perception they are a high-risk crop. The factors causing yield variability are poorly understood. A range of abiotic and biotic yield-reducing factors are likely to be important.
In this study a paddock survey approach was used to assess paddock conditions and yields across three main mungbean growing areas in the northern region. This was combined with simulation modelling to determine the water-limited yield potential and estimate yield gaps of mungbean crops across a diverse range of environments and growing conditions. The objective was to identify factors likely related to poor mungbean crop performance.
Field survey of mungbean crops
Consultants in Central Queensland (CQ), Darling Downs in southern Queensland (SQLD), northern New South Wales (NNSW; Moree and Liverpool Plains) were contracted to collect mungbean paddock survey data in season 2017-18. Supplementary data was collected from the GRDC farming systems and legume agronomy projects. Data was collected from 12 paddocks in CQ, 17 paddocks in SQLD and 13 paddocks in NNSW in addition to the supplementary data collected in the GRDC farming systems and legume agronomy projects. Paddock data including GPS latitude / longitude, soil type (if known), cropping history for past summer and winter seasons, fertiliser and herbicide applications were recorded, and in-season measurements completed. In-season measurements included starting / finishing soil water and starting soil nutrient analysis in 0-10, 10-30, 30-60 and 60-90cm soil layers, row spacing, cultivar, paddock rainfall, time of flowering, biomass and harvest cuts. In each paddock Predicta®B sampling was also completed. Crop measurements occurred across the 2017/18 season at five monitoring points (in the same location as the soil sampling) including plant counts, flowering date, biomass cuts, maturity date and harvest samples. Weeds, diseases and insects were monitored throughout the season.
Prediction of yield potential and gaps in mungbean crops
The APSIM-mungbean model was used to simulate the water-limited yield potential of mungbean crops corresponding to the 42 farm fields surveyed, along with 29 experimental sites obtained from either Mungbean Agronomy research (2013-2018) or Farming Systems research sites (2015-2018). Simulations for each crop, replicated crop management deployed in each field (e.g. same sowing date, configuration). Climate data was obtained from the nearest climate station available from the SILO database or from climate stations located at the experimental site. Soils used in simulations were sourced from the closest soil with the same classification in the APSoil database. Where possible, simulations were initiated with soil water and nutrient samples taken prior to or at sowing. In many cases, starting soil water was not measured or could not be reliably estimated at sowing because of uncertainties with crop lower limits on the specific soils at each site. In these cases, estimations were made by resetting plant available soil water to zero at the completion of the previous crop and allowing APSIM to model soil moisture accumulation during the fallow period leading up to mungbean sowing. The cultivar Emerald was chosen to most accurately represent Jade and Crystal in the APSIM model.
Estimation of yield gaps
In rain-fed crops, yield gaps were calculated as the difference between the water-limited yield potential (Yw) and the achieved grain yield (Ya). The water-limited yield potential is influenced by soil type, soil water status at sowing and climatic conditions over the crops growing season but is not limited by nutrients or biotic stresses. The yield gap was computed as the difference between APSIM simulated yield and the observed yield for each of the 71 mungbean crops. Hence, computed yield gaps are attributable to sub-optimal nutrient supply, biotic factors reducing grain yield or other stress events (e.g. high temperature) which are not captured in the APSIM simulation model. The model was also used to compute three separate stress indices from simulations of each crop to indicate the degree that crops are exposed to high temperature stress events, low soil moisture status and photosynthetic reductions due to moisture stress during flowering and grain filling periods.
Agronomic drivers of mungbean yield - nutrition
There were limited relationships between yield and both key macro and micro-nutrients across the surveyed paddocks. Across the sites, nitrate N in 0-90cm ranged from 10-300 kg/ha with an average of 115 kg N/ha and average grain yield of 1.0 t /ha. Higher yields were associated with higher starting soil N, however, this relationship was highly variable. Vegetative biomass ranged from 0.75 – 7.0 t dry matter (DM)/ha and there was no significant correlation between grain yield / biomass and phosphorus (P), zinc (Zn), sulfur (S), magnesium (Mg) and potassium (K) in the soil. This is not to say these nutrients aren’t critical for crop growth and yield, but more that they weren’t singularly the key factor driving yield.
Agronomic drivers of mungbean yield – harvest Index
Unsurprisingly, grain yield was highly correlated with biomass (0.84) and harvest index (HI) at 0.68 (Figure 1). High biomass didn’t guarantee high yields, with biomass being poorly correlated with harvest index (0.22). Crops with poor harvest index were strongly associated with lower yields and breeding / management factors that improve harvest index should be a future focus.
Figure 1. Relationship between grain yield (kg/ha) (A) and biomass kg DM/ha; (B) harvest index across mungbean paddocks in season 2017-18. Spearman correlation coefficient (CC) and significance indicated.
Agronomic drivers of mungbean yield – row spacing and WUE
Management factors that improved yields included narrow row spacing and planting onto fallow. Row spacing < 40 cm yielded 33% higher than crops on row spacing > 50cm, at 1.2 and 0.8 t/ha, respectively (Figure 2).
Crops achieving yields higher than 1.25 t/ha, generally required a minimum of 200 mm of available water (starting soil water + in-season rainfall) (Figure 2). Benchmark water-use efficiency of 7.1 kg/ha/mm (n = 10) was observed in the best survey paddocks, with mungbean WUE across the survey paddocks ranging from 0.2 to 9.4 kg/ha/mm with an average of 4.4 kg/ha/mm. While overall available water was not different between double-crop and fallow, mungbeans planted onto fallow had an extra 27 mm water in the soil at planting (starting water after fallow averaged 133 mm and when double cropped, 106 mm).
This difference in starting soil water accounts for the observed yield increase of 241 kg/ha for crop preceded by fallow compared to double-cropped. Crops preceded by fallow also had significantly higher biomass than double cropped mungbeans (4.1 and 3.0 t/ha, respectively). Crops that achieved yields over 1 t/ha generally had a minimum of 85 mm water in the starting profile. However, available water wasn’t the only factor driving yield, as many paddocks with over 200 mm available water also yielded poorly (Figure 2).
Figure 2. Relationship between grain yield and (A) available water (starting soil water and in-season rainfall mm) and (B) row spacing’s across mungbean paddocks in season 2017-18. (C) Relationship between yield gap and paddock history (fallow versus double-crop) / region (CQ, SQLD and NNSW). Univariate ANOVA used to determine significance.
Regional differences in mungbean yields
Comparison of paddocks between regions (CQ, SQLD, and NNSW) found no significant differences between regions for yield, biomass, harvest index and available water (starting soil water + in-season rainfall mm). On average, paddocks across the three areas all had > 100 kg nitrate N kg/ha, > 90 mg/kg Colwell P and > 14 mg/kg Mg. NNSW had higher P, K and Zn levels compared to SQLD and CQ. While nutrient thresholds for mungbeans are currently uncertain, it appears unlikely that most sites did not have nutrient status low enough to limit yields. However, low P levels in both CQ and DD may be contributing to yield limitations based on data in other legumes. In-crop rainfall was not significantly different across the regions, however much of the crop in NNSW suffered significantly from prolonged high temperatures in January 2018. Nematode pressure was significantly higher in SQLD with > 3 nematodes/g soil. The threshold for significant yield loss in intolerant crops including mungbeans is 2 nematodes/g soil. P. thornei levels in other regions were well below this level and there were no differences in P. thornei levels between fallow and double-crop paddocks
Mungbean yield gaps
The model reasonably predicted the grain yields and harvest index of crops that achieved their potential. However, there were large differences between simulated and observed grain yields for a large number of crops analysed. These differences were used to estimate mungbean yield gaps. Across the whole dataset, simulated mungbean yields ranged from 0.5 – 2.6 t/ha across the spectrum of double-cropped following a winter crop the same year, following a short-fallow after a summer crop the previous year or where partial irrigation was provided. There were no clear differences due to location or data origin in terms of the simulated yield potentials or observed grain yields across this data set.
One third (24 of 71) of the crops analysed had no significant yield gap (< 200 kg/ha) or observed yields exceeded simulated yields (Table 1). Several observed crop yields from the both paddock surveys and mungbean agronomy datasets exceeded the simulated yield potential, but in most cases, this was within the boundaries of variation for those observed crop yields. Some uncertainties with information used to simulate these crops may also explain these differences. Forty-five % of crops had a yield gap > 500 kg/ha. Seven (10%) of the 71 crops were identified that had large yield gaps of greater than 1 t/ha (Table 1). The remaining 16 crops (22%) had yield gaps between 200 and 500 kg (Table 1). Interestingly, crops with yield gaps of different magnitudes and proportions were found across both data from experimental and paddock survey origins. As the crops grown under experimental conditions were managed using optimal weed, insect and disease control, this suggests that yield gaps are unlikely to be explained by these yield reducing factors alone.
Diagnosis of yield limiting factors
In examining possible causes of yield gaps driven by paddock nutrient and pathogen status there were weak correlations between estimated yield gap and Colwell P (R2 = 0.11), sulfur (R2 = 0.13), or potassium (R2 = 0.18). Yield and biomass gap differed across the regions and paddock history with higher yield gaps (500 – 700 kg/ha) in CQ and SQLD compared to NNSW. This suggests there are factors affecting crops reaching their predicted yield potential in those regions which are not well predicted by the mungbean APSIM model. Yield gaps in double-crop mungbeans were significantly higher than in fallow mungbeans (Figure 2). The yield gap observed in double-crop mungbeans strongly suggests there are other factors or combination of factors that are playing a role in crops not reaching their yield potential.
Table 1. Percentage of crops by yield gap (absolute t/ha or proportional %) (n in brackets) for mungbean crops analysed from different sources across the subtropical grain’s region of eastern Australia.
Yield Gap (t/ha)
Multiple yield reducing factors
Only 43 or the 71 crops had a full complement of corresponding soil data; these were examined in more detail in an attempt to identify critical yield reducing factors. Of these a group of 26 had relative yields of < 65% of modelled water-limited yield potential (high yield gap group) and 17 were found to have relative yields > 65% of water-limited yield potential (low yield-gap group) (Table 2).
While discrepancies in the frequency of a range of factors between these two groups were examined, only a few factors were found to occur at different levels between the groups. These were also factors found to be more important in statistical analyses, giving us confidence that they are important potential factors to consider further. The population of P. thornei was the most prominent single factor discriminating amongst crops in the high and low yield gap groups. Of the high yield gap group, 53% of crops had P. thornei levels greater than 3/g soil, while in the low yield gap group this was 23%. This factor alone was not statistically significant (Table 2), but when additional factors were added, significant differences in the populations were evident. Thus, the model that distinguished best between high and low yield gaps was when crops had either high P. thornei populations (> 3/g soil).
Additional factor identified as reducing mungbeans crops yield were when more than 1 day of maximum daily temperatures of over 39oC during flowering, or the crops had soil nitrate levels of less than 65 kg N/ha in the top 60 cm at sowing.
A combination of at least one of these stresses occurred in 88% of crops with a high yield gap. This set of diagnostic criteria were found to provide a significantly higher probability of occurring in the crops that had a high yield gap compared to the group with the low yield gap. Hence, a combination of one of these 3 yield reducing factors are suggested to be possible foci for further work to understand their impact on mungbean yield accumulation and in particular low harvest index.
Table 2. Frequency of crops experiencing a combination of one or more stress criteria amongst groups with a high yield gap (i.e. relative yield < 65%) or low yield gap (> 65%). P was calculated using a Fischer exact test, n is the number of crops in each group for which a full complement of data was available.
High YG (% group)
Low YG (% group)
P. thornei > 3.0/g soil
+ Max. temp > 39oC during flowering
+ NO3 (< 65 kg/ha)
Overall, the project confirmed industry experience that there is a large proportion of mungbean crops that fail to achieve their water-limited yield potential. Crops with low observed yield and high yield gaps were those with a low harvest index, which was not always associated with low crop biomass. One third of crops achieved < 60% of the water-limited yield potential and nearly half had yield gaps > 500 kg/ha. Low mungbean yields or high yield gaps were not due to a single biotic or abiotic factor. Rather there was a multitude of factors that appear to be associated with high yield gaps (root lesion nematodes, high temperatures at flowering and soil nitrogen status). Being a single season study these findings should be interpreted with this in consideration.
The research undertaken as part of this project is made possible by the significant contributions of growers through both trial cooperation and the support of the GRDC, the author would like to thank them for their continued support. We would like to thank the consultants, Leigh Norton, Mike Balzer, Hugh Reardon-Smith, Rob Evans, Josh Bell and Jim Hunt, who collaborated by collecting soil and crop samples on the mungbean fields monitored here. We also thank the many farmers who allowed for this sampling to occur. Finally, we would recognise the contributions of previous research projects (Farming Systems CSA00050 and DAQ00192, and Mungbean Agronomy UQ00067) for provision of their experimental data.
CSIRO Agriculture and Food
203 Tor St, Toowoomba Qld, 4350
Ph: 0409 881 988
La Trobe University
Ph: 0409 160 355
Varieties displaying this symbol beside them are protected under the Plant Breeders Rights Act 1994
® Registered trademark
GRDC Project code: CSP1801-002RTX
Was this page helpful?