Harvest weed seed control is less effective on annual ryegrass in the southern high rainfall zone

Author: | Date: 22 Aug 2019

Take home messages

  • In the southern high rainfall zone (HRZ), harvest weed seed control (HWSC) is likely to capture 30% of annual ryegrass (ARG) seeds compared to around 70% in other regions of Australia (Walsh et al 2014). Therefore, HWSC will assist ARG control but cannot be relied upon to dramatically reduce weed seedbanks.
  • It is difficult to increase the efficacy of HWSC in the southern HRZ. A harvest cut height of 15cm was no more effective than 30cm. Changing the crop cultivar and time of sowing did not manipulate ARG phenology and/or the harvest date enough to prevent high ARG seed shedding.
  • Be strategic with HWSC. The extra operating costs of weed seed impact mills (up to $34 per ha) will only be covered in the long-term in paddocks where there is moderate to high weed pressure. Consider cheaper HWSC options where annual ryegrass density is lower.


Harvest weed seed control (HWSC) is a term that refers to the capture of weed seeds during harvest for subsequent destruction by crushing, grazing, burning or rotting. It has been shown to be an extremely effective means of long-term annual ryegrass (ARG) control in Australia (Walsh et al 2014; Walsh et al 2017). However, while research has been conducted across all rainfall zones only a small proportion of it has been conducted in the southern HRZ. Not only is this a wet and productive environment – which changes the growth of the crop and ARG plants – it is also cooler, which prolongs the growing season and sets back harvest dates, possibly increasing the amount of pre-harvest ARG seed shedding. Therefore, the adoption of these technologies has been recommended in the southern HRZ on the assumption that HWSC has a similar fit in integrated weed management (IWM) packages as it does elsewhere. From 2015 to 2018 a GRDC investment led by Southern Farming Systems in collaboration with Riverine Plains, Mackillop Farm Management Group and FarmLink was implemented to test this assumption.


The research program had three components addressing three questions:

  1. Is HWSC effective? Small-plot experiments were conducted in Victoria (Lake Bolac, Rutherglen, Yarrawonga), South Australia (Conmurra) and Tasmania (Cressy). All trials except for Rutherglen were conducted for two to three consecutive years on the same experimental plots. Sowing date, crop cultivar choice and harvest cut height were hypothesised to affect the efficacy of HWSC. HWSC was simulated by catching all harvest trash from plot headers and taking it off-site (Figure 1).
  2. Is HWSC practical? On-farm trials were conducted in Victoria, South Australia and southern New South Wales to test the practicality of using HWSC technologies, ground-truth their efficacy and measure operating costs.
  3. Is HWSC profitable? The data from these trials were pooled to re-calibrate a farm systems model called LUSO (Lawes and Renton 2010) which was used to explore the long-term economic impact of adding HWSC to a wheat-barley-canola rotation.

Photo of harvest weed seed control simulation by using a plot header and capturing harvest trash to be discarded off-site

Figure 1. HWSC was simulated on a plot header by capturing harvest trash to be discarded off-site.

Results and discussion

High shedding and lodging reduce the efficacy of HWSC in the southern HRZ

ARG populations were not controlled or reduced by HWSC in any small-plot trial in any year with any treatment. This is because much fewer seeds were captured for HWSC in the southern HRZ compared to other environments. The proportion of weed seeds that was captured was calculated for each experimental plot and pooled across the research program. The median value of this data set was 29% of seeds captured, with half of the data falling between 46% and 10%. The primary causes of this low efficacy were ARG shedding and lodging.

The season length of the southern HRZ postpones harvest dates compared to other environments in Australia, prolonging the growing season. It appears that the longer growing season gives ARG extra time to reach maturity and shed seeds before harvest. Across trial years, ARG reached anthesis in November, and shedding was first recorded from the second week of November to mid-December (Table 1). All trials were harvested at least one week after the first day of shedding was measured, but most were harvested three to four weeks after shedding began.

Table 1. Summary of annual ryegrass phenology and harvest dates from small-plot trials across the Southern HRZ in 2015 to 2017.



Sowing date(s)

ARG anthesis

First recorded shedding

Harvest date



15 May, 3 June, 26 June

28 October


22 December

Lake Bolac


20 April, 15 May, 17 June


26 November

4 Dec (TOS1) or

23 Dec (TOS2,3)



17 May

23 November

19 December

4 January

Lake Bolac


25 May

21 November

30 November

23 December



28 April


11 November

11 December



12 April, 10 May


15 December

5 January



22 May

10 November

24 November

4 January



12 May


10 November

10 December

Seed shedding was measured and calculated for each individual plot and the data from all trial years was pooled. The median value of this data is 51% of seed shed, but there is substantial variation (Figure 2b). The sowing and harvest dates in the trials are appropriate for the region, so data can be taken to be representative of the likely situation for HWSC on farms in the southern HRZ.

It was hypothesised that changing the sowing date and crop cultivar would manipulate the phenology of ARG, and as a consequence, change the efficacy of HWSC. However, there were no significant differences (P>0.05) between sowing date treatments in any trial in any year. In other words, this research suggests that the growing season in the southern HRZ is too long to change ARG phenology to an extent that prevents its shedding in cereals. Canola was not included in the small-plot trials but given that it is harvested before cereals it is likely that it will be less affected by ARG shedding, but still more affected in the southern HRZ when compared to other environments.

Around 20-25% of ARG seeds were located below either a 15cm or 30cm harvest cut height. This was due to lodging or to mature seed heads detaching from the stem and falling to the ground. Again, there was substantial variation in the dataset and while the median value of seeds captured was slightly higher for a 15cm cut height compared to a 30cm cut height, analysis of the small-plot experiment showed that there were no long-term effects of a low-cut height. ARG plants were counted in early 2018 at Yarrawonga after applying HWSC at 15 or 30cm in 2016 and 2017. After two years of HWSC, ARG establishment remained above 100 plants/m2 with no difference between cut heights (P>0.05). This suggests that while it may be beneficial in some years, cutting at 15cm may not be necessary in the southern HRZ. This supports the finding by Walsh et al (2018) who showed that greater crop biomass competition resulted in ARG setting seed higher in the crop canopy.

Column bar graph showing the incidence of pre-harvest seed shedding by annual ryegrass in cereals, recorded from 2015 to 2017

Figure 2. The incidence of pre-harvest seed shedding by annual ryegrass in cereals across the southern HRZ from 2015 to 2017, presented as the proportion of shed seed to total seed set.

The high productivity potential of the southern HRZ makes HWSC a valuable weed control tool

To estimate the costs and benefits of controlling ARG with HWSC it is essential to be able to estimate the yield loss ARG causes. Several models and parameters are available in the literature, but the data set of the research program made it possible to choose parameters specifically for the southern HRZ. Remarkably, it was not necessary to adjust the parameters for weed and crop competitiveness; the values in LUSO that are used in the medium and low rainfall zones also work in the southern HRZ. As can be seen in Figure 3, yield can be predicted accurately for a given crop and winter ARG population.

Scatter line graphs showing actual plot yield for wheat and barley at four different high rainfall zone locations (left hand side) compared with yield predicted by LUSO modelling based on weed and crop competition parameters (right hand side)

Figure 3. Actual plot yield and the yield predicted by LUSO for wheat and barley based on default weed and crop competition parameters using the number of annual ryegrass plants established in winter.

Despite very high weed burdens (>100 plants/m2 in many plots in this data set), the relative competitiveness of weeds and crops remain similar to that recorded for other rainfall environments with lower weed pressures. The theoretical point where an ARG population would cause crop failure was not reached. Intensive cropping rotations in the southern HRZ can tolerate a higher weed burden than other environments.

On the other hand, because of the southern HRZ’s high yield potential the value of ARG control is high. According to these parameters, 50 weeds/m2 established in a cereal crop in winter reduces the yield potential by about 10%. If the yield potential of a given environment is 4t/ha, this equates to a 0.4t/ha yield loss to weed competition. If the yield potential is 8t/ha (as in the southern HRZ), this equates to a 0.8t/ha yield loss to weed competition. So, while more weeds can be tolerated in a cropping system in the southern HRZ because the penalty is cushioned by a high yield potential, the gains from controlling weeds are greater too. This makes the contribution of HWSC to ARG control more valuable for the southern HRZ than might be expected from its low efficacy rate of 30%.

Case study: A weed seed impact mill is profitable only if there is already a weed problem

Extra costs of a weed seed mill

In 2017, an on-farm demonstration was conducted at Wolesley, SA, comparing two New Holland CR8090 headers, one with an integrated Harrington Seed Destructor (iHSD) and one without an iHSD. Wheat was harvested at 15cm. The data collected from this demonstration was used to calculate the extra operating costs of using an iHSD by making use of the data from the PIRSA Gross Margin Guide 2019 (https://grdc.com.au/resources-and-publications/all-publications/publications/2019/farm-gross-margin-and-enterprise-planning-guide). The value was $34/ha, which is comparable to a value calculated by the Kondinin Group ($4.81/t, or $33.67 for a 7t/ha yield).

Table 2. Values used to estimate the extra operating costs of using an integrated Harrington Seed Destructor in a cereal crop in Wolesley, South Australia.


GM Guide 2019 ($/ha)

Conversion Factor

WSM Cost ($/ha)

Repairs & Maintenance


+15% to engine load




+37% to fuel usage


Contractor Harvest


-36% to harvest efficiency



Total extra cost:


Scenario analysis

The long-term impact of a weed seed mill (WSM) on ARG populations and profitability was modelled in a wheat-barley-canola rotation over 12 years using the values presented in Table 3. Fixed costs were set at $180, and the cost of nitrogen (N) was set at $1.50/unit of N. The model was run with and without HWSC at two starting weed seedbank levels. Then the model was run again with the same conditions but the proportion of weeds surviving until maturity was raised from 5% to 10%. In the small-plot trials, low intensity herbicides were used to focus on the impact of HWSC and so the average survival rate in these trials was 10%. This value was used in the scenario analysis to illustrate the role of herbicide efficacy which may be affected by numerous factors including herbicide resistance.

Table 3. Values used in the economic analysis of HWSC in a wheat-barley-canola rotation in LUSO.




Grain Price

Variable cost

N required

Wheat + WSM










Feed barley + WSM





Feed barley





Canola + WSM










After 12 years, a similar pattern in ARG control was observed when the starting seedbank was 100 or 500 seeds/m2 (Figure 4). Populations increased exponentially if 10% of weeds survived and a WSM was not used. Adding a WSM to this situation significantly reduced weed population growthbut the final number of weed seeds in the seedbank was much higher after 12 years than the initial value. Reducing weed survival to 5%, the default LUSO value, achieved this to a greater extent. Only when a WSM was used and weed survival was 5% did ARG numbers decline in absolute terms, but they only declined by about 30% over 12 years.

Column bar graph showing the final seedbank levels after twelve years of a wheat, barley and canola rotation with 5 or 10 per cent of weeds surviving until maturity, with and without the use of a weed seed mill and starting at 100 or 500 seeds per metre squared

Figure 4. Final seedbank after 12 years of a wheat-barley-canola rotation with 5% or 10% weeds surviving until maturity and with or without a weed seed mill (WSM), starting at 100 seeds/m2 or 500 seeds/m2.

Although the pattern in final seedbank was similar between the 100 and 500 seeds/m2 scenarios, the pattern in profitability differed depending on the starting seedbank (Figure 5). With a 5% survival rate, adding a WSM did not increase profitability when starting at 100 seeds/m2. It was, however, profitable when starting at 500 seeds/m2. Using a WSM was always profitable when 10% of weeds survived to harvest. These results suggest that there must be enough of a yield penalty caused by high weed numbers of cover the extra costs of a WSM. It should be noted that these calculations only account for extra operating costs, and do not account for extra capital costs associated with purchasing a WSM.

Column bar graph showing total profit after 12 years of a wheat, barley and canola rotation in response to starting at 100 seeds per metre squared (left hand side) or 500 seeds per metre squared (right hand side) and comparing 5 or 10 per cent weeds surviving until maturity and with or without weed seed mill

Figure 5. Total profit after 12 years of a wheat-barley-canola rotation with 5% or 10% weeds surviving until maturity and with or without a weed seed mill (WSM), starting at (a) 100 seeds/m2 or (b) 500 seeds/m2.


  • By itself, HWSC only decelerates the population growth of ARG. The final weed seedbank output from LUSO matches what was seen in the small-plot trials – ARG numbers continue to grow between seasons despite the use of HWSC when many plants survive until harvest. HWSC must be combined with consistently effective herbicide and other weed management options that reduce the number of weeds setting seed, and even then, the reduction in ARG seedbanks will be small and slow. However, with an efficacy of 30% a WSM had a large decelerating effect on ARG population growth in every scenario. Therefore, HWSC cannot be relied upon in the southern HRZ to drastically reduce ARG numbers but can support IWM packages by decelerating ARG population growth.
  • There must already be a weed problem for a WSM to pay for itself. A WSM was only profitable if the initial weed seedbank was high burden (i.e. 500 seeds/m2 is equivalent to about 70 plants/m2 in mid to late winter) or the number of weeds surviving to harvest was high (i.e. low herbicide efficacy). Therefore, expensive to use HWSC technologies like WSMs could be targeted to problem paddocks to cover the extra costs and should not necessarily be used in clean paddocks with low resistance levels.
  • Furthermore, cheaper HWSC options with lower operating costs should be considered over expensive technologies. Effort should be made to reduce the operating costs associated with weed seed mills if they are going to have a place in farm businesses in the southern HRZ. The new vertical iHSD selling for $85,000 instead of $200,000 is a step in the right direction, but ways to reduce operating costs should also be pursued. Chaff lining or decking may be a low-cost HWSC option that fits the needs of the southern HRZ, but its efficacy has not been tested in this region. Recent findings by Broster et al (2018) in Wagga Wagga showed that high concentrations of chaff (>24t/ha) will suppress ARG germination by 40 to 80%. Large amounts of chaff would be produced in the southern HRZ, but it is unknown whether the different conditions would affect its ability to suppress weeds.


HWSC is likely to only capture 30% of ARG seeds in the southern HRZ, but this 30% is valuable to farm businesses because of the high productivity of the region. Therefore, HWSC has a fit in the southern HRZ farming systems, but it fits differently – it is not the ‘holy grail’, but it can play a supporting role for IWM packages. Adding HWSC to weed management strategies in the southern HRZ is recommended, but growers should focus on the cheaper HWSC options and/or use the expensive HWSC options strategically.

Useful resources

The 'Big 6' of the WeedSmart plan

How much does harvest weed seed control cost?

Chaff lining…too good to be true?


Broster, J, Rayner, A, Ruttledge, A & Walsh MJ 2018, ‘Impact of stripper fronts and chaff lining on harvest weed seed control’, GRDC Update Paper, 25 July 2018.

PIRSA, 2019, ‘Farm Gross Margin and Enterprise Planning Guide’, accessed 9 Aug 2019.

Lawes, R & Renton M 2010, ‘The Land Use Sequence Optimiser (LUSO): A theoretical framework for analyzing crop sequences in response to nitrogen, disease and weed populations’, Crop and Pasture Science, vol. 61, pp. 835-843.

Walsh, MJ & Powles, SB 2014, ‘High seed retention at maturity of annual weeds infesting crop fields highlights the potential for harvest weed seed control’, Weed Technology, vol. 28, pp 486-493.

Walsh, MJ, Broster, JC & Powles SB 2017, ‘iHSD mill efficacy on the seeds of Australian cropping systems weeds’, Weed Technology, vol. 32, 103-108.

Walsh, MJ, Broster, JC, Aves, C & Powles SB 2018, ‘Influence of crop competition and harvest weed seed control on rigid ryegrass (Lolium rigidum) seed retention height in wheat crop canopies’, Weed Science, vol. 66, pp 627-633.

White, B, Giumelli, J & Saunders M 2018, ‘Residue management at harvest: Weed seed options’, Kondinin Group Research Report, no. 97.


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. Paul Breust, formerly of SFS, managed this project for the majority of its implementation.

Contact details

James Manson
23 High Street, Inverleigh VIC 3321
0488 600 509

GRDC Project Code: SFS00032,