Building climate resilience in farming systems

Take home messages

  • The profitability of a farming system should be quantified over the whole crop sequence accounting for fallow costs and the length of the crop sequence/rotation.
  • Cropping intensity is a key driver of system profitability and risk, more so than the mix of crops used. Crop systems with higher intensities (i.e. less time in fallow) have higher average profitability, but also higher risk — vice versa, crop systems with longer fallows have lower risk, but there are trade-offs of lower long-term gross margins (GMs).
  • It is critical to match cropping intensity to environment to optimise the risk-return trade-offs. Lower crop intensities (0.5-0.75) are optimal in harsher environments (e.g. western districts), moderate crop intensities (0.75-1.0) in the moderate environments, but crop systems with higher crop intensities (1.0-1.3 crops/yr) are optimal in higher rainfall environments.
  • Mixing summer and winter crops in the sequence also mitigates seasonal variability and helps stabilise gross margins and risk for key crops following transitional long-fallows.
  • Crop sequences with lower crop intensities and involving summer crop in rotations have been increasingly attractive (in terms of risk-returns) over the past 10-20 years.

Background

Farming in Australia’s subtropics is a risky pursuit. We experience one of the most variable climates in the world in a market environment exposed to significant fluctuations in prices for commodities and inputs. Hence, designing a cropping system that helps to mitigate these risks is central to long term viability of a farming enterprise. While there are several mechanisms that can be used to mitigate price risks, managing climate risks on production is critical. Central to this is the capacity of the farming system to capture and utilise rainfall to convert this into profit.

However, there is often a trade-off between optimising water use efficiency (i.e. $ return per mm) and riskiness of the farming system. For example, cropping systems can reduce risk by employing longer fallows and waiting until soil water in the profile has been built and thereby minimising risks of crop failures or lower crop margins. However, longer fallows clearly mean fewer crops are grown and these returns must be adjusted for the duration of the fallow prior. Systems with longer fallows typically have lower fallow efficiencies and less rainfall is used by the crops. Conversely, cropping systems with higher intensity (i.e. more crops per year) have shorter fallows, lower soil water at sowing, lower average margin for each crop, but more of them and higher risks. Hence, finding the right spot where returns per unit of risk are achieved is a significant challenge.

The analysis reported in this paper uses simulation models to compare the long-term performance of crop sequences in terms of their relative profitability, riskiness and their resilience to a changing climate. It is important to capture the dynamics over the whole crop sequence as there are benefits and costs transferred from one crop to another that influence the efficiency of the system as a whole (e.g. nitrogen (N) supply, residual soil water, ground cover influencing fallow water accumulation). These models can be used to predict the potential yield outcomes for different farming systems, but in reality, it is often difficult for growers to maximise this — maximising yields may not be economically optimal, higher inputs are often needed and crop management is also assumed to be optimal. For this reason, the relative performance of different crop sequences when yields are adjusted to 80% of water-limited potential is also explored – this level is regarded as an achievable level on-farm.

Method

Simulations

This study is a simulation analysis that uses the rotation features of the Agricultural Production Systems sIMulator (APSIM) framework (Holzworth et al. 2015) to simulate crop rotations from historic climate records (1900-2012). APSIM has a long history of simulating northern farming systems (Carberry et al. 2009; Whish et al. 2007) and uses environmental signals to trigger appropriate management decisions. However, these simulations only considered the dynamics of water and nutrients. Losses due to waterlogging, heat or frost shock events, disease, pests, weeds or crop nutrition other than N were not considered by these simulations.

The simulation of all crop sequences was phased, so that each crop in the sequence was sown in every simulated year — a rotation of three crops in four years would be phased four times with each crop and one fallow starting the rotation. This was done to avoid any bias associated with starting particular sequences in any particular year and to ensure data on every crop or fallow was available in every year.

Rotations

The rotations presented were identified through consultations with a wide diversity of growers and advisers across the northern region (Dubbo to Emerald). The rotations analysed cover a range of intensities (0.5 to 1.3 crops per year), but have been restricted to only include the region’s most commonly grown crops (Table 1). All crop rotations simulated here are set rotations, with a crop sown each year at the end of the sowing window irrespective of soil water conditions — hence, the capacity to mitigate risk through tactically avoiding crops sown on marginal soil water is not considered. Cotton was excluded from the simulations due to inadequacies of the Cotton model to simulate dryland cotton in a cropping system (particularly soil water extraction). However, crop sequences where sorghum is used could equally represent crop sequences involving cotton. However, there are likely to be higher investment risks and negative impacts on subsequent crops (e.g. lower fallow water accumulation and less soil water available at sowing) compared to sorghum.

All rotations were simulated with traditional sowing windows for each crop, and fertiliser needs were simulated to ensure each cereal crop had sufficient N supplied at sowing to maximise crop yield potential. The frequency of fallow sprays was also flexibly predicted, based on occurrences of events with 25mm or more of rain over three days and repeats were only included where these occurred more than 10 days apart.

Financial calculations

Average annual GM analysis was conducted for each phased crop sequence using the equation below. Long term average grain prices (2008-2017) and current variable input prices were used and these were held constant across all locations (Table 1). Insurance and levy costs together were 2% of the grain income value and were deducted from grain prices. Simulated yields were also adjusted to 80% of water-limited yields to approximate grower achievable yields. The price for N fertiliser applied was set at $1.30/kg N and each fallow spray was set at $17/ha. The simulations did not account for application losses of N fertilisers, therefore, an additional 30% of applied N was used to ensure fertiliser N reached the soil mineral N pool. The baseline ’variable cost’ for each crop included planting, non-N nutrients and in-crop pesticide applications. Harvesting costs, N fertiliser and fallow spray frequency were included separately as these varied between the crop sequences or if crops failed. Crops were not harvested if yields did not exceed the costs of harvesting. Machinery costs were based on an owner-operated production system, therefore, fuel, oil, repairs and maintenance (FORM) costs where included in the variable costs.

GMseq($/ha/yr = ∑{(Grain yield x price) – (kg N x 1.3) – (sprays x 17) – variable costs – harvest costs}
No. of years

Table 1. Assumptions of crop prices and variable costs used in GM calculations for crop sequences.

Crop

Average Price ($/t)#

Harvest cost ($/ha)

Variable costs ($/ha)

Wheat

264

40

175

Sorghum

225

55

218

Chickpea

569

45

284

Mungbean

710

55

276

#farm gate price with grading and additional harvesting costs already deducted

Results and discussion

Calculating crop sequence returns

Comparing the GM of different crops is commonly used by growers and their advisers to make tactical choices about which crops to choose for a particular season or set of conditions. However, only considering crops individually and on an annual basis fails to consider the implications on subsequent crops or the fallow leading up to that crop. It is important to consider and analyse the cropping system as a whole over the whole sequence. To highlight this, the following example compares two crop rotations involving the same crops, but where the intensity of the cropping system (i.e. number of crops per year) is altered.

The two similar crop rotations used the same crops (S – Sorghum, Ch – Chickpea, W – Wheat, x – Fallow) — one where a chickpea crop is double cropped (SChxWxx) following sorghum while the other chickpea is long-fallowed from sorghum (SxxChxWxx). The SxxChxWxx sequence achieves higher margins on each crop in the system (owing to higher soil water at sowing of each crop), and particularly the chickpea crop which was preceded by a long-fallow which yielded 0.7t/ha higher than a double crop. However, the aggregate return of the crop sequence as a whole must account for the additional fallow costs ($100) and must be averaged over the four-year sequence length compared to the three year sequence. Overall, there is only a little difference in sequence GM, with the higher intensity crop sequence (three crops in three years) achieving $15/ha/yr higher GM than the lower intensity crop sequence (three crops in four years).

Table 2. Computing System annual GM between different crop sequences (S – Sorghum, Ch – Chickpea, W – Wheat, x – Fallow). Example is based on simulated long term average yields for each crop at Narrabri.

Crop sequence

SChxWxx

SxxChxWxx

Sorghum*

Chickpea

Wheat

Sorghum*

Chickpea

Wheat

Yield (t/ha)

3.2

1.6

2.8

3.2

2.3

2.9

Costs ($/ha)

273

329

215

273

329

215

Crop GM($/ha)

447

581

524

514

980

551

Fallow costs ($/ha)

98

28

65

98

125

65

Total GM over sequence ($)

1362

1756

Rotation years

3

4

GM per year ($/yr)

454

439

*Equivalent dryland cotton yields to achieve similar GM is 2.1 bales per ha (assuming double input costs and $480/bale).

Risk-return relationships in farming systems

Crop intensity (% of time in crop) was found to be a major driver of GMs of the crop sequence, irrespective of the mix of crops used (Figure 1). Figure 1 plots the relationship between period of time in fallow (i.e. no crop actively growing) and the average GM across the 22 crop sequences, varying in their mix of crops and intensity of crops in the sequence. This shows that environment has a large influence on these relationships. Mungindi has a very flat relationship indicating that crops across a wide range of intensities (from 55%-80% time in fallow) generate very similar average GMs. Meanwhile, at Trangie and Narrabri there is a reduction in average sequence yield as time in fallow increases (i.e. crop intensity declines). This relationship is even steeper at the higher rainfall and more favourable location of Cecil Plains.

While it is important to consider the average annual returns of the crop sequence as a whole, it is also important to quantify the risk associated with the crop sequence. While there are a range of ways ‘risk’ can be quantified, here the average GM is calculated in the worst 20% of years. Hence, systems with a higher return in these poor years are those that are less prone to negative or very poor GMs and are less ‘risky’.

Figure 1 shows that all site rotations with more time in fallow were those with lower risks (due to higher soil water available at sowing) while higher crop intensities (i.e. less fallow time) increased the risk of low returns in poor seasons (due to more marginal soil water sowing conditions). Again this relationship was more associated with crop intensity than it was with the mix of crops grown. This relationship between higher risk with increasing crop intensity was most severe in lower rainfall or more marginal environments (e.g. Mungindi), but Narrabri and Trangie also demonstrated a strong negative relationship. At Cecil Plains, cropping intensity (i.e. time in fallow) was less strongly related to risk, meaning that the downside of higher crop intensities was lower.

To illustrate this further, Table 3 compares five selected crop sequences that differ in crop intensity from 0.5 to 1.3 crops per year. For each location, the % of times that crops are sown with plant available water below 100mm and the % of crops that do not break even (i.e. returns do not exceed variable and fallow costs) are shown. This clearly shows that as crop intensity increases, the % of crops that are sown when soil moisture levels are marginal (defined here as < 100mm plant-available water) increases significantly. This is because there is less time for fallows to accumulate moisture and crops become more reliant on in-crop rain. Hence, as crops are grown on marginal soil water, the risk that any particular crop may not break even is increased. This may be managed in a cropping system through tactical decisions to avoid an ‘at risk’ crop in the sequence, but this is not considered in the current analysis.

Figure 1. Relationship between proportion of time in fallow and average crop sequence GM (left) and downside risk (i.e. average GM in the worst 20% of years) (right) for 22 crop sequences across four differing locations in the northern grains zone.

Table 3. Frequency of crops being sown on marginal soil water (i.e. < 100mm plant-available water (PAW)) and % of all crops with a negative GM across crop sequences varying in intensity and environments.

 

Crops/yr

Mungindi

Narrabri

Trangie

Cecil Plains

% of crops sown with < 100mm PAW

WxxxChxxx

0.5

4

0

4

0

SxxChxWxx

0.66

25

14

25

4

SxxWxChxWxx

0.8

26

16

26

3

SChxWxx

1.0

44

36

50

13

SChxWMgx

1.3

66

54

66

37

% of crops with GM < $0/ha

WxxxChxxx

0.5

8

7

1

0

SxxChxWxx

0.66

30

9

16

6

SxxWxChxWxx

0.8

28

18

12

4

SChxWxx

1.0

49

21

32

10

SChxWMgx

1.3

55

44

45

14

These analyses clearly show that in most grain production environments, there is a significant trade-off between higher potential average GMs per year with increasing risk, and this is closely associated with cropping intensity. So, for a particular environment, it is important to know which crop systems maximise the returns per unit of risk. In Figure 2, the average long-term returns are plotted against the downside risk for the full range of 22 crop sequences to see which crop sequences and their associated cropping intensities are optimal against these two competing factors. Sequences located further to the top right are found to be more optimal in terms of maximising the return-risk trade-off for each location (i.e. higher return with lower downside risk or higher returns in the worst years). If a crop sequence achieves a lower GM for a given unit of risk, then it is sub-optimal in terms of risk-return. The crop sequences at the frontier of this trade-off have been highlighted for each location. This figure shows that at Mungindi low intensity crop sequences (0.5-0.75 crops/yr -SxxWxChxWxx, WxxxChxxx) were optimal, while at Cecil Plains higher intensity crop sequences (1.0-1.3 crops/yr) were optimal. Narrabri and Trangie were intermediate with varying crop sequences ranging from 0.5-1.0 crops/yr presenting different risk-return propositions that may be tailored to the particular grower’s risk appetite or financial position.

It is also worth noting that in these locations, mixtures of summer and winter crops offer lower risks than systems dominated by summer or winter crops only. This is associated with mixing the range of crops up to utilise good or poor summer or winter seasons to help buffer the system against climate variability. Such systems where summer and winter crops are used in the crop sequence also allow for mitigating risk though employing some long-fallows leading into key crops (e.g. long-fallow from winter to a summer crop (e.g. sorghum here, but alternatively cotton) or stabilising the yield of the first winter crop after a long-fallow). Further risk mitigation would also be provided through buffering against price variability by using a variety of crops where prices are not linked. There are also likely to be a range of agronomic benefits from using summer crops regularly in the crop rotation.

Figure 2. Relationship between long term average annual GM and downside risk (i.e. GM in the worst 20% of years) amongst 22 crop sequences at four locations across the northern grains zone. Crop sequences with the highest return per unit of risk are highlighted (○) and labelled.

Long term changes in relative risk-returns of crop systems

The above analyses have taken a long-term view considering a very long climate record at each of the focus locations. However, indications of changing climatic conditions and increasing variability may mean that the relative performances of different cropping systems may be changing. How relative GMs and risk have changed over the past 50 years was examined, looking at 10-year long periods from 2012 back to 1971. Overall, this shows that the average GM of all cropping sequences at all of the study locations has declined over this time, but particularly since 1996. The relative profitability of different crop sequences has also shifted significantly over this time. For example, at Narrabri, the most profitable crop sequence over a 10-year period has changed between several crop sequences over the past 50 years — showing that flexible and adaptable farming systems are critical. However, there appear to be some more fundamental shifts in the relative profitability of crop sequences across the sites. At Trangie, the Wheat-Wheat-Chickpea rotation was superior to other cropping sequences throughout the period up until the decade ending in 2008. Since that time, other lower intensity crop sequences involving summer crops of sorghum (e.g. SxxChxWxx and SxxWxChxWxx) have achieved similar average sequence GM with lower levels of risk. A similar trend is also evident at Narrabri. At Cecil Plains, while the higher intensity system (SChxWMgx) has maintained the highest GM over the whole period, the difference to other lower intensity systems has decreased at the same time as its riskiness has increased significantly. Finally, the analysis at Mungindiclearly shows that low intensity crop sequences (e.g. WxxxChxxx) have become equally profitable to other crop sequences with significantly lower levels of risk in the past 15-20 years. Overall, these trends show that over the past 10-15 years, lowering the intensity of the cropping system has had advantages of lower risk, but the opportunity cost of the lower intensity system compared to the higher intensity system has declined.

Figure 3. For the preceding decade finishing in 2012 to 1971, changes in 10-year moving average (left) and coefficient of variation (right) in GM of six crop sequences varying in crop intensity across four locations in the northern grains region.

Conclusion

Simulation analysis has allowed a long-term view of the relative profitability and risk-return relationships for cropping systems commonly deployed across the northern grains zone. This has shown that cropping intensity is a key driver of system profitability and risk, but this relationship varies significantly with cropping environment. Tailoring the cropping intensity suitable for your environment is a critical factor to balance the trade-off between risk and return across the crop system. There are many crop sequences that are suboptimal in a particular environment, and the gaps can be significant, hence, there is significant opportunity to alter the farming system to fit the risk appetite of the grower and their enterprise. Finally, this analysis shows that across the study environments, the relative attractiveness of summer crops and lower intensity farming systems has grown over the past 10-20 years, indicating that growers need to continually re-evaluate their farming system as a whole.

Acknowledgements

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.

Contact details

Lindsay Bell
PO Box 102, Toowoomba Qld, 4350
07 4571 3201 or 0409 881 98
Lindsay.Bell@csiro.au
@lindsaywbell

GRDC Project Code: CSA0050,