High performance crop sequences for Southern Queensland

1 CSIRO Agriculture
2 Department of Agriculture and Fisheries, Queensland
3 University of Queensland
4 Innokas Intellectual Services

Take home message

  • Optimising crop sequence is critical to maximising system efficiency and sustainability.
  • No crop sequence is universally optimal against all system attributes, but sequences with higher water capture and use were also better for soil carbon, and sequences performed similarly for risk of weed herbicide resistance and propagating Pratylenchus thornei.
  • Crop sequence affected evaporative losses little, but higher crop intensity reduced runoff and drainage losses, increased fallow efficiency and increased proportion of rain transpired by crops.
  • Average sequence gross margin was highest in sequences with less time in fallow and a greater proportion of rainfall transpired by crops, but this increased risk in further west environments (e.g. Mungindi). While average gross margin was more linked to crop intensity, different crop sequences managed risk better than others.
  • Sequences using a double crop of chickpea or mungbean had higher variability in gross margin and a higher frequency of crop failures, particularly at lower rainfall environments. Short-fallow sorghum crops were also much riskier than long-fallow crops at Goondiwindi and Mungindi.
  • Sequences with low risk of developing P. thornei required a 2.5-3 year phase of resistant crops (e.g. sorghum) or fallow. Risks of developing herbicide resistant weeds were low when a phase of summer and winter crops were rotated, but high when summer fallows were frequent.  

Introduction

Many cropping systems in the northern grains region are facing significant sustainability and profitability challenges. Analysis of paddocks in southern southern Qld and northern NSW have shown that while many individual crops are highly water-use-efficient, in many cases crop sequences are performing less well. Current cropping systems are being challenged by declining soil resilience, build-up of soil-borne pathogens such as root lesion nematodes, and increasing reliance on external inputs. Furthermore most systems continue to deplete soil nutrients which would cost an additional $75-200/ha to be replaced.  Hence, there is a need to pay closer attention to crop sequences in order to enhance the capture and conversion of rainfall into production and profit and to maintain long-term productivity.

The analysis reported in this paper uses simulation models to compare the long-term performance of a range of common crop sequences against multiple elements of the farming system. That is, how do the various crop sequences compare in terms their relative profitability, riskiness, productivity, water-use-efficiency, fertiliser input use and efficiency and risks for multiplication of nematodes, herbicide resistance and soil carbon depletion. This comparison is done at 3 locations in southern Queensland (Cecil Plains, Goondiwindi, Mungindi) to capture environmental differences from east to west. Through this analysis we aim to identify opportunities to improve the efficiency of crop sequences, and quantify some future threats for the range of crop sequences currently used. It is important to note that there is no ‘best’ crop sequence, this will be affected by current prices, soil constraints, and other farm resources; here we hope to identify the relative benefits and costs of some of these.

Methods

Simulations

Using historical climate data for Condamine Plains, Goondiwindi and Mungindi long-term simulations (110 years) of 18 common cropping sequences were conducted in APSIM. APSIM is a farming systems model which predicts the production of crops and captures the dynamics of water and nutrients in the farming system. The sequences simulated vary in their cropping intensity (from 0.5 to 1.3 crops per year), mix of crops, and summer/winter dominance (Table 1). The simulated sequences were set, and were defined by rules which ensured each crop in the sequence was sown if a sowing opportunity occurred in their sowing window or at the end of the recommended sowing window, even when moisture levels were marginal. All cereal crops were fertilised to ensure 200 kg of N was available at sowing and legumes were not fertilised. All sequences simulated are based on a no-till system with full stubble retention using a common good cropping soil in each district (i.e. plant-available water-holding capacity (PAWC) for wheat at Condamine Plains = 290 mm, Goondiwindi = 190 mm, and Mungindi = 186 mm). Simulations generated long-term information on crop yields, dynamics of soil water and nitrogen accumulation and use during fallows and under crops, and changes in soil carbon over the 110 year record of climate.

Table 1. Summary of simulated crop sequences. The growing windows for each crop are indicated by the various colours (below), otherwise land is fallow.

Table 1. Summary of simulated crop sequences. The growing windows for each crop are indicated by the various colours (below), otherwise land is fallow.


Sequence annual gross margin ($/ha/yr) was calculated using simulated outputs of grain yield, N requirements and number of germination events during fallows, using the equation below. These assumed long-term average grain prices and current variable input prices for each crop (Table 2).

Downside risk for each sequence was calculated from the average gross margin in the worst 20% of simulated years.

Sequence annual gross margin formula

Table 2. Assumptions of crop prices and variable costs used in gross margin calculations for crop sequences

Crop

Average Price ($/t) (after transport)

Variable costs excl N fertiliser & fallow sprays ($/ha)

Wheat

240

175

Sorghum

205

218

Chickpea

400

284

Fababeans

380

341

Mungbean

550

276

Nematodes multiplication factor was calculated for Pratylenchus thornei based on the proportion of time in a sequence that a susceptible crop versus a fallow or resistant crop was present. The proportion of days over the simulation that a susceptible crop was growing increased was multiplied by 2.5 times, and the proportion of days that a resistant crop or fallow periods was multiplied by -0.4. The changes are based on the rates of increase and decline in populations observed during susceptible and resistant crops or fallow periods.

Herbicide resistance risk - The pyRAT model, which predicts the risk of weeds developing herbicide resistance to Group M, I and A herbicides was used to predict risk scores for each rotation. The maximum risk, either for summer or winter grasses or broadleaf weeds, was used for each crop sequence. Risks for development of Group A and Group I herbicides were low to negligible in the sequences simulated here and hence results presented are for Group M (glyphosate) resistance. 

Limitations of the analysis

When interpreting the results of this analysis it is important to be aware of some of the limitations. First the current crop models in APSIM used here don’t capture the effect of extreme heat stress, frost events, or sub-optimal supply of nutrients other than nitrogen. Hence, they may over-predict the likely yield due to losses from heat events at flowering in sorghum or mungbeans at Mungindi and Goondiwindi. Secondly, the yields in the crop sequences here are not reduced by losses due to weed competition or from soil-borne pathogens such as crown rot or nematodes; these capabilities are still under development and aim to be included in future analyses. Thirdly, the soils used here had no constraints for crop growth, for example, chickpea which might perform more poorly where subsoil constraints occur. Finally, we have used long-term average prices for each of the crops simulated; risk associated with price fluctuations amongst these are not considered here and the shifts in the relative prices of these crops is likely to shift the relative profitability of sequences.

Relative performance of common crop sequences

None of the 18 crop sequences simulated here performed best against all systems attributes. There were important trade-offs between crop sequence profitability, production risk, water capture and utilisation and long-term risks amongst the various crop sequences. Figure 1 illustrated the relative performance against 11 different attributes of the farming system for 6 common crop sequences at the 3 locations, Cecil Plains, Goondiwindi and Mungindi. From this there are some evident correlations between key attributes:

  • Sequences which transpired a higher % of rainfall also performed better in terms of fallow efficiency and had fewer losses to drainage and runoff. They also performed better in terms of maintaining soil carbon reserves. Conversely sequences with less rainfall used by crops had lower fallow efficiencies, more losses to drainage and run-off and depleted soil carbon more rapidly.
  • Systems performing poorly in terms of risk for propagating P. thornei, were also low ranked in terms of risks of developing herbicide resistant weed populations, and vice versa.

Figure 1. Relative performance of 7 common crop sequences at (a) Cecil Plains, (b) Goondiwindi, (c) Mungundi against several systems attributes. The point where the line for each sequence listed crosses each axis indicates the relative performances compared to all 20 simulated crop sequences. If the line is the closest to the centre, the crop sequence performed the worst for that attribute; if the line is on the outside it performed the best for that attribute.

Figure 1. Relative performance of 7 common crop sequences at (a) Cecil Plains, (b) Goondiwindi, (c) Mungindi against several systems attributes. The point where the line for each sequence listed crosses each axis indicates the relative performances compared to all 20 simulated crop sequences. If the line is the closest to the centre, the crop sequence performed the worst for that attribute; if the line is on the outside it performed the best for that attribute. 

The relative performance of crop sequences shifts with the environment. A benchmark of a wheat-wheat-chickpea sequence is included at each location for comparison. It is clear that this sequence performs well against many attributes at Mungindi, but has high future risks for developing soil-borne pathogen and weed problems. Other crop sequences clearly outperformed the WxWxCh sequence at Cecil Plains and alternatives involving summer crops were available at Goondiwindi which had similar profitability.

Below we explore in more detail many of these performance attributes amongst the 18 simulated crop sequences at the 3 locations.

Sequence water-use efficiency

Across all locations, crop sequence had little if any effect on the amount of rain lost to evaporation – this was typically about 50-60 % of rain that falls; it is constant and difficult to change (Figure 2). The rain not lost as evaporation was either transpired (i.e. used by the crop) or was lost as runoff or drainage; the balance of these two was significantly influenced by the crop sequence.  Crop sequences with high crop intensity had the highest percentage of transpiration and hence had the lowest losses, while sequences with lower crop intensity had lower % of rainfall transpired with the rest of the water being lost as runoff or drainage.

The amount of rainfall transpired by crops, or conversely the proportion to time in fallow, were found to be critical drivers of water-use-efficiency and average sequence gross margin (i.e. the average annual GM over the full crop sequence). Figure 3 shows the different relationships at each location between average sequence gross margin (GM) and rainfall transpired (bottom) or proportion of time in fallow (top) for the full set of crop sequences.

  • At Cecil plains (Fig 3a), crop sequences with less time in fallow had higher average gross margins and as fallow increases the average sequence GM decreases. Despite the potential for higher crop intensity to increase risk, at Cecil Plains the GM in the worst 20% of sequences also declined as the time in fallow increased – that is, there was no substantial benefits for avoiding downside risks from longer fallow periods. The mix of crops in the sequence had little impact on the average sequence GM (this was most affected by proportion fallow), but did have some effect on managing risk in poor years. 
  • At Goondiwindi (Fig 3b), there was also a clearly higher average sequence GM in sequences with increasing crop transpiration and hence average GM declines as time in fallow increased. However, this relationship was less steep than at Cecil Plains. The GM in the worst 20% of years was not closely related to the time in fallow, with some sequences having lower downside risk compared to others with similar crop intensity. Like Cecil Plains for average GM the mix of crops in the sequence was less important than crop intensity, but large differences were evident between sequences for managing risk in poor years. 
  • At Mungindi the relationship between crop transpiration, or proportion of time in fallow, with average sequence GM was weak. That is, there was only a slight effect of time in fallow and crop transpiration on average GM of the sequences, the sequence of crops grown was more critical.  At the same time the gross margin in the worst 20% of years was better in sequences that had more fallow.  This means that sequences with higher crop intensity do not necessarily result in higher average GM and comes with greater risk.

Figure 2. Proportion of rainfall that is transpired by crops (green), lost as runoff or drainage (blue) and evaporated (orange) in 18 crop sequences at 3 locations in southern Queensland. Green

Figure 2. Proportion of rainfall that is transpired by crops (green), lost as runoff or drainage (blue) and evaporated (orange) in 18 crop sequences at 3 locations in southern Queensland. Green

Figure 3a. Cecil Plains: Relationships between time in fallow (top) and average sequence gross margin (GM) ($/ha/yr; top) or downside risk ($/ha/yr in worst 20% of years; middle) and crop transpiration (mm/yr) with average sequence GM (bottom) across 18 simulated crop sequences.

Figure 3a. Cecil Plains: Relationships between time in fallow (top) and average sequence gross margin (GM) ($/ha/yr; top) or downside risk ($/ha/yr in worst 20% of years; middle) and crop transpiration (mm/yr) with average sequence GM (bottom) across 18 simulated crop sequences.

Figure 3a. Cecil Plains: Relationships between time in fallow (top) and average sequence gross margin (GM) ($/ha/yr; top) or downside risk ($/ha/yr in worst 20% of years; middle) and crop transpiration (mm/yr) with average sequence GM (bottom) across 18 simulated crop sequences.

Figure 3a. Cecil Plains: Relationships between time in fallow (top) and average sequence gross margin (GM) ($/ha/yr; top) or downside risk ($/ha/yr in worst 20% of years; middle) and crop transpiration (mm/yr) with average sequence GM (bottom) across 18 simulated crop sequences.

Figure 3b. Goondiwindi: Relationships between time in fallow (top) and average sequence gross margin (GM) ($/ha/yr; top) or downside risk ($/ha/yr in worst 20% of years; middle) and crop transpiration (mm/yr) with average sequence GM (bottom) across 18 simulated crop sequences.

Figure 3b. Goondiwindi: Relationships between time in fallow (top) and average sequence gross margin (GM) ($/ha/yr; top) or downside risk ($/ha/yr in worst 20% of years; middle) and crop transpiration (mm/yr) with average sequence GM (bottom) across 18 simulated crop sequences.

Figure 3b. Goondiwindi: Relationships between time in fallow (top) and average sequence gross margin (GM) ($/ha/yr; top) or downside risk ($/ha/yr in worst 20% of years; middle) and crop transpiration (mm/yr) with average sequence GM (bottom) across 18 simulated crop sequences.

Figure 3b. Goondiwindi: Relationships between time in fallow (top) and average sequence gross margin (GM) ($/ha/yr; top) or downside risk ($/ha/yr in worst 20% of years; middle) and crop transpiration (mm/yr) with average sequence GM (bottom) across 18 simulated crop sequences.

Figure 3c. Mungindi: Relationships between time in fallow (top) and average sequence gross margin (GM) ($/ha/yr; top) or downside risk ($/ha/yr in worst 20% of years; middle) and crop transpiration (mm/yr) with average sequence GM (bottom) across 18 simulated crop sequences.

Figure 3c. Mungindi: Relationships between time in fallow (top) and average sequence gross margin (GM) ($/ha/yr; top) or downside risk ($/ha/yr in worst 20% of years; middle) and crop transpiration (mm/yr) with average sequence GM (bottom) across 18 simulated crop sequences.

Figure 3c. Mungindi: Relationships between time in fallow (top) and average sequence gross margin (GM) ($/ha/yr; top) or downside risk ($/ha/yr in worst 20% of years; middle) and crop transpiration (mm/yr) with average sequence GM (bottom) across 18 simulated crop sequences.

Figure 3c. Mungindi: Relationships between time in fallow (top) and average sequence gross margin (GM) ($/ha/yr; top) or downside risk ($/ha/yr in worst 20% of years; middle) and crop transpiration (mm/yr) with average sequence GM (bottom) across 18 simulated crop sequences.

Evaluating the water-use-efficiency of the whole crop sequence is important as it does not just consider the conversion of this available soil water and in-crop rainfall into grain product (crop water-use-efficiency) but also integrates the accumulation of soil water from rainfall during fallows (fallow-use-efficiency).  Because of the range of crops grown and their different value, gross crop returns (i.e. sum of yield x price) or gross margin per hectare are used to calculate WUEprofit. Below we provide some indicative benchmarks for sequence water-use-efficiency for the crop sequences at the 3 locations in southern Queensland (Table 3). This shows a significant difference in maximum and typical range for water-use-efficiency of the crop sequences between locations.

Table 3. Average water use efficiency benchmarks for crop sequences ($/ha/mm rain) at 3 locations in southern Queensland

WUEprofit

Cecil Plains

Goondiwindi

Mungindi

Gross returns ($/ha/mm)

Max

2.0

1.4

1.3

Typical range

1.4 - 1.7

1.0 - 1.2

1.0 - 1.2

Gross margin ($/ha/mm)

Max

1.20

0.60

0.50

Typical range

0.85 - 1.0

0.45 - 0.50

0.35 - 0.40

Variability in gross margin and risk of failed crops

Figure 4 shows the variation in sequence gross margin for the different crop sequences at each of the 3 locations. Note this variability is less than would be observed when considering individual crops, because the sequences includes several years which include both good and bad years that offset each other, and a diversity of crops that are grown across these years. Hence, longer crop sequences and those with a greater diversity of crops, particularly growing a mix of summer and winter crop had lower variability in gross margin.

Crop sequences with higher variability were those that had a high probability of unprofitable or failed crops (i.e. a negative GM), shown in Figure 5. At all locations, sequences with regular long-fallows had a much lower frequency of failed crops, though at Cecil Plains this reduction was only small. The frequency of failed crops was much lower at Cecil Plains, but at Goondiwindi and Mungindi more aggressive crop sequences with higher crop intensity had higher frequencies of failed crops. Sequences with a particularly high frequency of failed crops were those involving a chickpea double crop (e.g. SChxW, SChxWMgx); at Goondiwindi and Mungindi around 40% and 60% of chickpeas grown as double crops in the sequence did not produce a positive gross margin. Crop sequences involving double crops of mungbean, were also risky with about 30% and 40% of these crops failing at Goondiwindi and Mungindi, respectively. At all locations sequences where the sorghum crops were grown after a short fallow had a higher frequency of failure than those proceeded by a long-fallow (from a previous winter crop), though this was much lower at Cecil plains than at Goondiwindi or Mungindi.

Figure 4. Variability in sequence gross margins ($/ha/yr) for 18 crop sequences at 3 locations in southern Queensland. Light shaded boxes indicate 25th - 50th percentile, dark shaded boxes indicate 50th – 75th percentile and bars the 95th and 5th percentile years. Table 1 provides price and cost assumptions used.

Figure 4. Variability in sequence gross margins ($/ha/yr) for 18 crop sequences at 3 locations in southern Queensland. Light shaded boxes indicate 25th - 50th percentile, dark shaded boxes indicate 50th – 75th percentile and bars the 95th and 5th percentile years. Table 1 provides price and cost assumptions used.

Figure 5. Frequency of crop failure, that is a crop with a negative gross margin, for 18 crop sequences at 3 locations in southern Queensland. Critical crop yields for each crop used are provided in the legend.

Figure 5. Frequency of crop failure, that is a crop with a negative gross margin, for 18 crop sequences at 3 locations in southern Queensland. Critical crop yields for each crop used are provided in the legend. 

Crop sequence consequences for root-lesion nematodes and herbicide resistance risk

Table 4 categorizes the various crop sequences according to their predicted risk for propagating root lesion nematodes. The key results are:

  • Sequences with low or negligible risk of propagating P. thornei are those that have 2.5 – 3 years phase without a host plant, including crops of sorghum and/or fallow
  • A single sorghum crop in a system dominated by wheat and chickpeas will slow the rate of build-up but does not prevent build-up of nematodes
  • Resistant winter crops (e.g. canola) are required to reduce risks in winter-dominated sequences. Data not shown here. 
  • Several crop sequences in the lower risk categories that have only a small opportunity cost compared to the crop sequence with the highest average GM. If losses of around $100/ha or 0.5 t/ha from nematodes then alternative ‘low-risk’ rotations could provide alternatives.

Table 4. Estimated shadow costs (i.e. difference in average GM from highest returning sequence) for crop sequences that have differing levels of risk of propagating P. thornei.

Nematode propagation risk

Sequence

Cecil Plains

Goondiwindi

Mungindi

Negligible (<1.1)

SxSxSChxx

-101

-76

-53

SxSxSxxWMgx

-189

-59

-54

SxSxSxxChxWxx

-260

-83

-62

SxSxMgWxx

-188

-3

-27

SxSxSChxWxx

-144

-68

-68

Low (1.1-1.25)

SxxWxWxx

-332

-130

-106

SxSChxWxx

-139

-82

-80

SxxChxWxx

-277

-97

-62

WxxxChxxx

-436

-149

-75

SxSxSFbxWxChxWxx

-66

-40

0

Moderate (1.25-1.5)

SxxWxChxWxx

-218

-80

-50

SChxWxx

-130

-84

-65

SxxChxWxChxWxx

-221

-78

-46

SxxChxWxFbxWxx

-240

-94

-61

SxxWxChxWMgx

-118

-46

-33

High (>1.5)

SChxWMgx

-39

-62

-104

SxMgWxCh

0

0

-28

xWxChxWMgx

-182

-66

-49

xWxWxCh

-165

-67

-46

Table 5 categorizes the various crop sequences according to their predicted risk for developing glyphosate resistant weed populations. The key results found were:

  • Highest risks were for summer weeds and hence the level of risk were lower at Mungindi due to the lower frequency of summer rainfall events.
  • Sequences with lowest risk were those that had a mixture of winter and summer crop phases and crop frequencies of around 1 crop per year.
  • Highest risk crop sequences were those that were dominated by winter crops, with high requirements for summer fallow herbicides placing high selection pressure on summer weeds.
  • Again, several crop sequences in the lower risk categories have only a small opportunity cost compared to the crop sequence with the highest average GM. In particular, crop sequences with higher crop intensity performed better in terms of glyphsate resistance risk and had higher long-term average gross margins.

Table 5. Estimated shadow costs (i.e. difference in average GM from highest returning sequence) for crop sequences that have differing levels of developing herbicide resistance in weeds.

Herbicide resistance risk

Sequence

Cecil Plains

Goondiwindi

Mungindi

Low

SxSxMgWxx

-188

-3

-27

SxSxSChxWxx

-144

-68

-68

SxSChxWxx

-139

-82

-80

SxSxSxxWMgx

-189

-59

-54

SxSxSxxChxWxx

-260

-83

-62

Moderate

SxMgWxCh

0

0

-28

SChxWxx

-130

-84

-65

SxSxSChxx

-101

-76

-53

SxxWxChxWxx

-218

-80

-50

SChxWMgx

-39

-62

-104

SxSxSFbxWxChxWxx

-66

-40

0

SxxWxChxWMgx

-118

-46

-33

SxxChxWxx

-277

-97

-62

SxxWxWxx

-332

-130

-106

High

SxxChxWxChxWxx

-221

-78

-46

SxxChxWxFbxWxx

-240

-94

-61

xWxChxWMgx

-182

-66

-49

WxxxChxxx

-436

-149

-75

xWxWxCh

-165

-67

-46

Changes in soil carbon

Most of the crop sequences simulated saw soil carbon levels decline over the period of the simulation. Figure 4 compares the long-term simulated changes in soil carbon between 5 differing crop sequences at the 3 locations. The critical result here was that crop sequences with a higher frequency of crops maintained soil carbon at higher levels, and conversely longer periods in fallow reduced soil carbon more rapidly. Winter cereals also appeared to be better at maintaining soil carbon levels that sorghum, but this was a less important effect. The differences between sites were related to the soil characteristics used in the simulations more than their association with climatic conditions – for example the soil used at Goondiwindi had high initial levels of labile soil carbon and hence saw faster and larger losses than the soils used for Dalby and Mungindi which had lower initial labile soil carbon.

Figure 6. Change in labile soil carbon (t/ha) over 110 simulated years for a selection of crop sequences varying in cropping intensity (SChxWMgx = 1.3 crops/yr; xWxWxCh and SxSChxWxx = 1.0 crops/yr; SxxWxChxWxx = 0.8 crops/yr; and SxxChxWxx = 0.75 crops/yr) at 3 locations in southern Queensland.

Figure 6. Change in labile soil carbon (t/ha) over 110 simulated years for a selection of crop sequences varying in cropping intensity (SChxWMgx = 1.3 crops/yr; xWxWxCh and SxSChxWxx = 1.0 crops/yr; SxxWxChxWxx = 0.8 crops/yr; and SxxChxWxx = 0.75 crops/yr) at 3 locations in southern Queensland.

Acknowledgements

The research undertaken as part of CSA00050 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 also like to thank the growers and advisors involved in meetings to understand the range of crop sequences being applied across the northern grain zone.

Contact details

Lindsay Bell
CSIRO, 203 Tor St, Toowoomba Qld, 4350
Ph: 0409 881 988
Email: Lindsay.Bell@csiro.au

GRDC Project code: CSA00050