New technology for improved herbicide use efficiency

Take home messages

• Site specific herbicide application can optimise weed control while minimising herbicide cost.
• The economic return from site specific weed management (SSWM) with herbicides cannot exceed the cost of the highest cost herbicide applied unless the herbicide has a phytotoxic effect on the crop that reduces yield.
• High density weed patches should be targeted with high efficacy treatments over several years to deplete the seed bank.
• Next generation weed identification sensors are being investigated for use in Australia.

Background

Site specific weed management (SSWM) has the potential to deliver significant improvements in weed control efficiency, through the targeted application of weed control measures only to where the weeds are located. Improvements in weed control efficiency will typically be achieved through reduced herbicide usage where herbicide is not required. SSWM has four principal components:

• Weed identification: Locate and identify weeds.
• Treatment decision: Make decision on appropriate treatment to control the weeds.
• Application: Apply appropriate treatment to the weeds.
• Documentation: Record weed location and the applied treatment.

This presentation will discuss the current state of play for weed identification sensors and review recent results of site specific herbicide trials.

Weed Identification

Presently, the only commercial weed sensors are spot spray systems that are only for use in fallow situations, where all green plants are considered weeds and sprayed, such as the WeedSeeker® and WEEDit systems. However, numerous groups around the world have been working on sensing systems that can identify different weed species within a growing crop, including several groups in Australia.  There are however, no commercially available products yet.

Agricon is a precision ag company in Germany that is developing and commercialising a weed ID sensor for the European market. This sensor uses near infrared and red imagery and leaf shape parameters to differentiate different weed types from crops. SAGIT is funding a project led by SPAA Precision Agriculture Australia to assess this weed ID sensor in Australian crops and to produce new adapted classifiers for identifying important Australian weeds in Australian crops. Examples of its application are discussed.

Variable Rate Herbicide Application

A number of trials have been conducted to assess the best weed control options for targeting high and low weed densities, with particular focus on annual ryegrass and brome grass. With most in crop selective herbicides no longer being effective due to herbicide resistance, particularly for annual ryegrass, many of these treatments are reliant on pre-emergent herbicides. The weeds are not growing at the time when pre emergent herbicides are applied, so to effectively target pre emergent herbicides site specifically an historical weed map is required. The relevance for historical weed maps from one season to the next is dependent on weed patch stability. Weeds that disperse seeds long distances via wind or other means are likely to have low patch stability and are poorly suited to SSWM using historical weed maps.  Consequently, these weeds should be targeted in the season they are mapped. Weeds that disperse seed only short distances will have more stable patches and are more suitable for using historical weed maps to target pre emergent herbicide applications. For example, natural dispersal of annual ryegrass is less than 1m, however combine harvesters can pick up and move ryegrass seed up to 18-20m in the direction of harvester travel. To overcome this problem in this example and to avoid missing ryegrass with a pre emergent herbicide application a 20m buffer could be added to the patches to ensure patch expansion is covered. Figure 1 shows that even over several seasons in these two paddocks there was a strong relationship between ryegrass densities.

Figure 1. The relationship in annual ryegrass density in two paddocks between 2006 and 2009.

The economic return from SSWM with herbicides can’t exceed the cost of the herbicide unless the herbicide has a phytotoxic effect on the crop that reduces yield.  Therefore, when costs for weed mapping and variable rate application are considered, it is apparent that SSWM with low cost herbicides will not be economical.  In this study it was found that variable rate application of clethodim is not likely to pay, as it is a relatively low cost herbicide.  The maximum potential saving from clethodim was $5.42/ha and declines as infestation area increases (Figure 2).  However, in high density annual ryegrass patches in two paddocks of lentils the desired level of ryegrass mortality was not achieved even at the highest clethodim rates.  If alternative control measures such as crop and weed desiccation with non-selective herbicides are employed, which sacrifice all crop yield to achieve the necessary weed control, then obviously the economics of patch management change significantly (Figure 2).

Figure 2. Herbicide saving from variable rate application of clethodim based on average savings in three paddocks to achieve a head density less than 20 heads/m² and the sensitivity to area of infestation compared with the opportunity cost of desiccating high density annual ryegrass patches in lentils where sufficient ryegrass mortality rates cannot be achieved with clethodim.

Variable rate applications of pre-emergence herbicides in cereals were more economically viable, as these herbicides are typically more expensive.  The herbicide savings are dependent on infestation level, but in one paddock where 35% of the paddock was infested:

• Variable rate application targeting Boxer Gold® to the high density patch and trifluralin to the low density patch would have generated a saving of $15.30/ha. 
• Variable rate applications reduced the risk of low returns from using high cost herbicides across the whole paddock.
•  Variable rate application made it economic to treat smaller patches. To make an economic return in the year of application with a uniform high cost treatment (Boxer Gold® in this paddock trial) required at least 11% of the paddock to have a high density ryegrass patch. With variable rate application it was economic to treat patches less than 6% of paddock area. This assumes $7.50/ha for uniform rate application costs and $15/ha for variable rate application costs.

Across a number of paddocks in the 'high' density weed patches the highest efficacy treatments were also generally higher cost, being greater than $25/ha in all cases. In high weed densities these higher costs were returned through increased yields. The exception being where the herbicides caused phytotoxic effects on the crop.  The benefit of high efficacy treatments at high density weed sites was often observed in subsequent years with reduced seedling recruitment in following years.  However, due to the high background seed bank associated with the high density patches the populations were still elevated and required ongoing targeted management with high efficacy treatments to deplete the seed bank further.

Decision making at the 'low' density sites is potentially more difficult. At 'very low' density weed sites (untreated had less than 2 ryegrass heads/m²) the level of weed control from the high cost treatments was not substantially different from the untreated, and was not yield responsive. At these sites the most economic treatment was the untreated. The untreated did not lead to any substantial increase in weed numbers in the subsequent year after treatment in these 'very low' density sites. At 'low' density sites (untreated had 2-20 ryegrass heads/m²) the response was less predictable. At these sites the higher efficacy treatments provided some weed control over the untreated, but did not yield substantially more and did not provide the best economic outcome in the 1st year of application. However, the population in the untreated tended to increase in the following year and if treatment was not implemented in the 2nd year then yield penalties were incurred. Therefore in the long term these low density sites were likely to benefit from treatment, but the question remains, how often should these populations be treated to maintain the low ryegrass density and maximise economic return? With improved weed mapping systems and an annual mapping/surveillance program these areas can be monitored and treatments adjusted for the subsequent seasons where required.

Variable Rate Seed

In addition to varying herbicides, crop seed rates can also be varied. Increased seed rates in the weed patch are used to increase crop competition and reduce weed vigour. This is generally simpler to apply than variable rate herbicide, too.

Variable Rate Herbicide to Soil Type

Several soil applied residual herbicides make label statements indicating different label rates for different soil types, with different soil types often defined by soil texture and organic matter levels. This information could form the basis for variable rate applications of herbicide based on soil type, with data layers such as EM38 potentially being suitable for defining soil types. While some growers may manually change rates on-the-go according to their assessment of soil type change, there are few examples of this process being automated and used widely.

Conclusion

High density weed patches should be targeted with high efficacy treatments over several seasons to drive weed numbers down. Herbicide savings can be made by reducing inputs into low density populations and these savings are greatest when using high cost herbicides. It is important to monitor weed populations where herbicide application has been reduced for density increase and be prepared to treat where large increases occur. Improved weed mapping systems and an annual weed surveillance program will help to ensure population increases are monitored and managed.

Contact details

Sam Trengove
PO Box 6, Bute, SA, 5560
0428 262 057
samtrenny34@hotmail.com
@TrengoveSam