Previous PageTable Of ContentsNext Page

Does control of summer fallow weeds improve whole-farm productivity in both Mediterranean and temperate environments? A simulation analysis

Andrew D. Moore and James Hunt

CSIRO Sustainable Agriculture Flagship, GPO Box 1600, Canberra, ACT 2061, Australia.
Email Andrew.Moore@csiro.au

Abstract

Weeds growing in summer fallows reduce the long-term water-use efficiency of paddocks by transpiring water and sequestering nitrogen that would otherwise be used for crop production. The effects of fallow weed growth on crop yields, livestock production, deep drainage and ground cover from representative rotation systems were examined at Hopetoun, Victoria and Temora, NSW by means of whole-farm simulation modelling. A hypothetical, palatable summer weed germinated in most summers. Modelled transpiration by weeds averaged about 15% of annual rainfall at both locations; somewhat less than half of this water was captured by crops when an aggressive weed control program was implemented. Crop yields responded strongly to fallow weed control: average simulated wheat yield increased from 1.5 to 1.9 t/ha at Hopetoun and from 2.8 to 3.6 t/ha at Temora. Effects on livestock production were much smaller despite the reduction of green biomass in summer. As a result, farm-scale partial budgets showed that complete weed control would be profitable. Simulations in which the starting date of weed control was varied showed that 63% of the crop yield benefit could be retained at Temora if the start of weed control was delayed from 1 December to 1 February. Control of the hypothetical summer weeds had substantial effects on NRM indicators: deep drainage increased at both locations, but the frequency of low ground cover increased at Temora and decreased at Hopetoun. From this modelling analysis, the N cycle effects of summer weed control appear to be roughly as important as the water balance effects.

Key Words

Summer weeds, stubble, whole-farm, APSIM, GRAZPLAN

Introduction

Weeds growing in summer fallows transpire water that could otherwise be used by subsequent crops. Summer weeds also reduce the supply of nitrogen (N) to following crops, by sequestering N in weed tissues (hence delaying its release to crops) and by drying the soil and thereby slowing mineralization of organic N. Plot-scale experimentation and modelling show that controlling these weeds can have a major effect on the yield of following crops. For example, Hunt et al. (2008) found that control of summer weeds led to an additional 70 mm of soil water being available at sowing, increasing subsequent wheat yield by 1.3 t/ha. Lilley (2007) used simulation modelling to estimate that control of summer grass weeds would increase subsequent wheat yield by 6 to 20%, depending on crop reliance on stored water. It is possible, however, for the benefits of a change to agricultural management at the paddock scale to diminish or even disappear once the constraints and tradeoffs that apply at the farm scale are taken into account. The area to which a practice change can be applied or the proportion of years in which it is useful may be limited, so diluting its effect. It may remove a biophysical resource, resulting in reduced production in other years or enterprises; in the case of summer weeds, control will reduce the amount of green herbage available to a livestock enterprise in the summer months. A practice change may also increase the risk of an undesirable natural resource management outcome, requiring other (unprofitable) adjustments to management to preserve the land resource. For example, storing extra soil water in crop fallows through weed control has the potential to increase deep drainage when large rainfall events fill the soil profile. These farm-scale constraints and tradeoffs are inherently complicated, especially in mixed farms. They also depend upon the frequency and magnitude of episodic events such as deep drainage or soil erosion that can only be evaluated over the long term. Biophysically-based simulation modelling can represent the interactions between different processes and outcomes and can be applied over many years, and so is a useful technique for understanding the potential for paddock-scale interventions to be “diluted out” or else limited by unintended consequences.

Methods

Mixed farming models

Mixed farming systems models were configured by linking the APSIM soil water, soil nutrient cycling and annual crop simulation models (Keating et al. 2003) to the GRAZPLAN pasture and ruminant simulation models (Donnelly et al. 2002). Surface residue dynamics were modelled with the GRAZPLAN pasture model so that grazing of stubbles could be represented. Management systems were described using a module that implements a rule-based language for specifying the occurrence of management events. Simplified mixed farming systems models were constructed for 2 locations: Hopetoun, Victoria (3544’S, 14222’E, average annual rainfall 328 mm) which has a Mediterranean climate and Temora, NSW (3425’S, 14731’E, 521 mm) which has a uniform rainfall climate. Patched Point weather data were acquired from the SILO database (www.longpaddock.qld.gov.au/silo). For simplicity, a single soil type was modelled at each site (Table 1) and the modelled farming systems consisted of two fixed pasture-crop rotations (Table 1) and a Merino ewe flock. Each phase of each rotation was present in the simulations each year. The soil humus and soil biomass pools were reset at the start of each pasture phase to keep soil organic matter near its starting value, so that each modelled year was directly comparable; otherwise the soil water and biomass dynamics in each phase were allowed to determine resource availability in the next phase of each crop-pasture sequence.

Table 1 Summary of site attributes and management practices in the two mixed farm models.

 

Hopetoun

Temora

Soil type

Vertic Calcarosol

Brown Chromosol

Available water holding capacity (wheat, mm)

174

132

Rotation 1*

50% AWCW

60% 4AWCWB

Rotation 2

50% AWBW

40% 5UWCWBW

Annual pastures

Barley grass+medic

Ryegrass+clover

Stocking rate (ewes/winter pasture ha)

3.0

6.0

Lamb sale period

1 Nov – 31 Jan

1 Dec – 31 Jan

* W=wheat, C=canola, B=barley, A=annual pasture, U=lucerne. Numerals denote consecutive years in a land use.

Locally-adapted crop cultivars were used. Crops were sown if 25mm of soil water was available in the surface 400mm on the first day of a fixed sowing window (1 May-15 June for cereals and 20 Apr-15 May for canola); otherwise they were sown once 15mm of rain fell over 5 days or dry-sown at the end of the sowing window. Preliminary simulations were used to identify fixed N fertilizer rates for each cropping phase that produced a marginal benefit:cost ratio of 2.0. At Temora, the first wheat crop after pasture was sown to cv Wedgetail and then grazed during the winter if a sowing opportunity occurred by 15 May. A flock of Medium Merino ewes (50 kg standard reference weight) grazed the pastures, dual-purpose crops and stubbles. Each February, sufficient ewes to produce a cohort of replacements were mated to Merino rams, and the remainder were mated to Border Leicester rams for prime lamb production. Male lambs were castrated after birth and all lambs not required as replacements were sold during a defined period (Table 1) when they either reached 50 kg live weight or started to lose weight once pastures dried off. A flexible grazing management scheme was used in which stock were moved to the paddock that would offer the best diet. Paddocks with cover below 0.75 or with growing crops were excluded from grazing (except during dual-purpose wheat grazing). Stubble paddocks were grazed immediately after harvest to use spilt grain and then whenever they provided the best source of forage for ewes. Ewes were supplemented with wheat grain to maintain their body condition score if it fell below 2.0.

Hypothetical summer weed

The GRAZPLAN pasture model was used to model a hypothetical summer-growing forb. For the purposes of this analysis we chose to model a species that has high potential rates of water and N uptake, is palatable to livestock and is of high nutritive value when green. The genotypic parameter set for capeweed (Arctotheca calendula) was therefore adapted to represent a forb that germinated readily during the summer, was more resilient than capeweed to drought stress during seedling growth but could be killed more easily by extended drought once established. 5 kg/ha of summer weed seed was assumed to germinate between 1 December and 31 March each year whenever 25 mm or more of rain fell in an event (a sequence of days with no more than one dry day at a time). This assumption resulted in modelled germinations in most summers, averaging 1.2 events/year at Hopetoun and 2.5 at Temora. The annual pasture species (Table 1) also germinated and grew in the stubble paddocks whenever environmental conditions were suitable for them.

Simulation experiment

A simulation experiment was carried out with 3 factors: location (Hopetoun and Temora), stubble grazing (permitted or not permitted) and start of summer weed control (half-month intervals from the typical harvest date to 15 March, plus simulations with no weed control). All simulations were run from 1957 to 2011; the first 3 years were omitted from the calculated results. During the weed control period, all plant species were killed (with 100% efficiency) whenever total green mass reached 100 kg/ha. Each weed control event was assumed to cost $20/ha. Water balance, ground cover and production results (crop yields, livestock sales and wool shorn) were recorded from each simulation run and the latter were used to compute farm-scale partial budgets, using the simulation with grazed stubbles and no weed control as a reference.

Results

Water balance and crop yield effects of stubble grazing and weed control

Ceasing to graze crop stubbles had little or no effect on modelled farm-scale water balance at either location (Table 2). Changes in modelled crop yields at Hopetoun were negligible, but at Temora the yield of barley crops was reduced slightly by not grazing the preceding summer fallow. The supplementary feeding requirement increased at both locations when the grazing of summer fallows was omitted from the models, making this change to management unprofitable at both locations.

Table 2 Consequences for the long-term average water balance, crop and livestock production and area-weighted frequency of low ground cover of changing summer fallow management in two simulated mixed farms by either controlling summer weeds prior to all annual crops, or by ceasing to graze crop stubbles.

 

Hopetoun

Temora

Weed control

+

+

+

+

Grazing of stubbles

+

+

+

+

Transpiration by crops (mm)

78

77

95

95

133

132

152

152

Transpiration by fallow weeds (mm)

54

53

12

12

78

79

20

19

Runoff (mm)

2

2

2

2

6

6

7

7

Deep drainage (mm)

6

6

20

20

50

53

65

68

Yield of wheat (t/ha)

1.45

1.43

1.90

1.90

2.79

2.77

3.63

3.58

Yield of canola (t/ha)

1.14

1.14

1.49

1.48

1.82

1.80

2.33

2.31

Yield of barley (t/ha)

1.24

1.23

2.29

2.28

2.45

2.33

3.50

3.45

Pasture NPP - Apr-Oct (t/ha)

2.13

2.16

3.12

3.13

3.86

3.70

3.70

3.58

Pasture NPP - Nov-May (t/ha)

0.43

0.44

0.15

0.16

1.52

1.50

1.20

1.20

Fallow weed NPP (t/ha)

0.64

0.63

0.19

0.19

1.15

1.15

0.32

0.32

Clean wool sold (kg/farm ha)

2.2

2.2

2.3

2.3

12.9

12.7

12.9

12.8

Lamb live weight sold (kg/farm ha)

7.0

6.9

8.7

8.4

46.1

43.5

45.2

43.5

Supplement requirement

44

57

23

38

23

39

25

39

Mean change in net income ($/ha/year)

 

-4

74

70

 

-21

78

60

S.E. of mean change in net income

 

1

14

13

 

3

12

11

Frequency of ground cover < 0.70

0.058

0.061

0.013

0.010

0.141

0.160

0.179

0.190

A change from no control of summer weeds to full control had much larger effects on the modelled farm water balance, with the proportion of rainfall that was transpired by crops increasing by 23% at Hopetoun and 14% at Temora (Table 2). The increase in crop transpiration was noticeably smaller than the decrease in transpiration by summer weeds at both sites. Crop transpiration efficiency for grain production also increased (by 21% overall at Hopetoun and 16% at Temora), resulting in large increases in modelled long-term average crop yields. The relative crop yield increases were largest for the simulated barley crops at both locations. Summer weed control had very little effect on the productivity of the sheep enterprise at Temora, while at Hopetoun it actually increased lamb turnoff and decreased the supplementary feeding requirement. The latter result was linked to an increase in growing-season pasture productivity in the Hopetoun simulations when summer fallow weeds were controlled (Table 2).

Effect of the timing of weed control on crop yield and profit

As expected, a delayed start to control of stubble weeds reduced the modelled yield benefit to following crops; the pattern shown in Figure 1(a) for wheat was also followed for canola and barley. At Hopetoun the simulated crop yield benefits were proportional to the time over which weeds were controlled. At Temora, however, the yield foregone by not controlling weeds in each successive month increased steadily; 63% of the yield benefit could be obtained by starting weed control on 1 February (i.e. 50% of the time). The profit increases from increased crop yields were larger than the changes in profitability in the livestock enterprise caused by summer weed control, regardless of the time at which control was commenced (Figure 1(b)).

Figure 1. Effect of control of stubble weeds starting on different dates on (a) the increase in long-term average yield of wheat crops (all rotations and phases) and (b) the change in long-term average operating profit from the cropping and livestock enterprises, relative to yield and profit with no control of stubble weeds.

Natural resource management implications

Control of the hypothetical summer weeds had substantial effects on NRM indicators. At Temora, simulated farm-scale deep drainage increased from 50 to 65 mm/year and the frequency of low ground cover (below 0.70) increased from 0.14 to 0.18 when weeds were controlled (Table 1). At Hopetoun, simulated deep drainage changed from 6 to 20 mm/year but low ground cover was less frequent due to more rapid canopy closure by both crops and pastures.

Discussion and Conclusions

Our modelling results suggest strongly that control of weeds is far more important than grazing management when managing summer fallows, and that even partial control of summer weed growth is likely to be well worth while. At Temora, there is little evidence of residual water or N effects on pasture production. At Hopetoun, the modelled increases in pasture production must be due to the availability of residual water and/or N, since weed control was not practiced in the summer prior to the pasture phase. These residual effects are estimated to make a small contribution to profit but also to usefully enhance the overall level of ground cover. The average crop yield increases reported here are somewhat smaller than those reported by Hunt et al. (2008) for a single year, but larger than the long-term increases estimated by Lilley (2007) from paddock-scale simulations with wheat. The latter difference is most likely due to the different crop N fertilizer policies that were assumed. In this study the direct effects of increased water supply due to summer weed control accounted for only about half the crop yield increases, with other factors – presumably dominated by N supply – contributing the rest. Lilley (2007) used high N fertilizer rates that would have masked the effects of summer weeds on N dynamics in the fallows. The primary tradeoff that should be considered when managing summer fallows is not that between crop and livestock production, but between crop production and the risk of undesirable NRM outcomes. In particular, the water saved by weed control is only partly re-directed to crop transpiration, and much of the remainder appears as increased deep drainage.

Acknowledgments

This work was financially supported by the Grains Research and Development Corporation as part of its Water Use Efficiency Initiative.

References

Donnelly JR, Freer M, Salmon EM, Moore AD, Simpson RJ, Dove H and Bolger TP (2002). Evolution of the GRAZPLAN decision support tools and adoption by the grazing industry in temperate Australia. Agricultural Systems 74, 115-139.

Hunt JR, Cousens RD, Knights SE (2008) The biology of Australian weeds 51. Heliotropium europaeum L. Plant Protection Quarterly 23, 146-152.

Keating BA, Carberry PS, et al. (2003). An overview of APSIM, a model designed for farming systems simulation. European Journal of Agronomy 18, 267–288.

Lilley JM, Kirkegaard JA (2007) Seasonal variation in the value of subsoil water to wheat: simulation studies in southern New South Wales. Australian Journal of Agricultural Research 58, 1115-1128.

Previous PageTop Of PageNext Page