free counter statistics
University of Manitoba Faculty of Agricultural and Food Sciences Department of Plant Science

Reducing Herbicide Rates:
How Far Can We Go?


In Manitoba in 2000, famers spent 69% more on herbicides than they had in 1993 (Anonymous, 2000). During this same period realized net income for farmers in Manitoba did not increase. Some farmers may consider reducing their inputs to try to cut costs and increase profits. A significant variable production cost for producers in western Canada is herbicides. In Manitoba herbicides are used at a very high frequency. In-crop applications are excluded from less than 5% of fields (table 1). When all applications on any one field are combined only 0.7% and 0% of spring wheat and canola fields, respectively, are cropped without the use of herbicides (Thomas et al. 1999).

Table 1. Frequency of herbicide use in fields in Manitoba in 1997. Adaped from Thomas et al. (1999).
Application Timing % Frequency of Herbicide Use
Pre-seed 17.7
In-crop 95.7
Pre-harvest 33.8
Post-harvest 11.1

The ability to exploit reduced herbicide rates comes with knowledge gained of the way in which certain herbicide products respond under certain conditions when used at reduced rates, and the species which are most readily controlled by certain herbicides. It is also important to have an understanding of how registerd herbicide doses are chosen. More importantly, however, is the fact that a reduction in herbicide dose, or herbicide use, cannot be broadly adopted by farmers if no change is made in their cropping systems. The most common cropping systems in western Canada are simple summer annual crop rotations (Thomas et al. 1999). These systems are biologically fragile and they are sustained by the use of fertilizer and pesticide inputs (Van Acker et al. 2001a). Robust cropping systems, systems that are inherently less susceptible to pest invasion, proliferation and interference are the systems that will allow producers to generally reduce herbicide rates and in some cases and years, exclude herbicide use completely (Van Acker et al. 2001b).

Biologically Effective Dose

The biologically effective dose (BED) is the herbicide dose which provides a 90% reduction in weed dry matter (Knezevic et al. 1998). The biologically effective dose is species specific and it does not necessarily imply a reduction in herbicide rate in comparison to registered rates. The concept of achieving good weed control with reduced herbicide rates is to apply them at the level of the biologically effective dose. The BED is not necessarily consistent, as one would expect, because the necessary dose required to achieve a 90% reduction in weed biomass is dependent upon the weed control scenario (Harker and Blackshaw 1996). Under oprimum conditions for herbicidal weed control the BED may be lower than registered rates. On average, however, it is likely that for most herbicides the BED is very close to the registered rate because of the way in which herbicide manufacturers select their registered rate.

How Registered Herbicide Rates are Determined

Herbicide manufacturers generally register a herbicide rate on the basis of the BED. In registering a herbicide they must submit data to the Pest Management Regulatory Agency (PMRA) on a species by species basis, demonstrating sufficient and consistent efficacy. Once the company registers a rate it is bound to provide a guarantee of efficacy for that rate for the species they include on their label. It is in the companies best interest to ensure that they rate they choose is reliable; that it will provide good efficacy under a broad range of conditions and that the crop on which the product is used will not be harmed by the herbicide at that rate and under various weather conditions.

The chosen rate for a given herbicide (herbicide 1 in figure 1) will be high enough to provide good weed control without causing a significant reduction in crop yield, even if the weather is less than ideal. There are some herbicides, however, for which the spread between the dose response of the weed and the crop is wide (herbicide 2 in figure 1). For these herbicides, the registered dose chosen by the manufacturer may be higher than the BED. The manufacturer may specifically choose this higher dose because it will tend to provide even more consistent high efficacy without risking crop safety. These herbicides are sometimes termed to be "over-labelled." With over-labelled herbicides there is a greater opportunity to reliably reduce rates without suffering reduced efficacy. Some herbicides are also labelled by species with specific rates for easy versus tough to control weeds. Additionally, tank mixes can alter the efficacy characteristics of herbicides. A common example is the efficacy reducing effect of group 4 products on group 1 products. In some cases this tank mix effect is exploited to provide crop safety as is the case for MCPA mixed with Fenoxaprop-p-ethyl to provide a safe herbicide product for use on barley.

Figure 1. A representation of the effect of herbicide dose on weed and crop <br>
biomass to demonstrate the positioning of herbicide dose to achieve weed control
efficacy and crop safety.

For perennial species the timing of herbicide application relative to the phenology of the weed can affect the herbicide dose required to achieve high and lasting efficacy. Dandelion control with glyphosate is best achieved with post-harvest applications, and with this timing a 900 g ai/ha rate (1 L Roundup/acre) provides equivalent control to applications of 1800 g ai/ha or 2700 g ai/ha (Froese and Van Acker 2001).

Selecting acceptable scenarios for reducing herbicide rates

For farmers wishing to reduce herbicide rates there are two general scenarios. In the first scenario the farmer does nothing to alter the croppings system in order to reduce weed pressure on the crop. If a farmer is employing a simple annual crop rotation and seeding an annual crop at the same time it is normally seeded then there is no reason to expect that there will be any fewer or any more weeds in the field than there were last year. In this case (the most common case) one cannot justify a reduced rate on the expectation of a reduced number of weeds, or that the weeds will be disdvantaged compared to the crop because nothing was purposefully done to offer advantage to the crop over the weeds (i.e. using a more competitive cultivar, delayed seeding, or very early seeding, or novel fertilizer placement to favour the crop). In this case, if farmers wish to attempt to use reduced herbicide rates they can assess the situation at herbicide application time. A number of factors must be considered when doing so and these have been summarized by Swedish researchers into simple wheel models; the "Swedish Models" (see appendix A and B). In these models the assessment factors include; crop stand vigour, crop competitiveness, ease of control of the weed, weed size, weather, and weed desnity. It is interesting to note from these wheel models that for a fall seeded crop, if the crop is rated as vigorous, under no circumstance (even with high density of large tough to control weeds in poor weather) is more than a ¾ herbicide dose recommended. The same is true if the crop is rated as normal and the weed desnity is low. For a spring seeded crop the same is true if the crop is rated as having a vigorous or normal start and the weed density is low or medium. Much of the theory that is incorporated within the wheel can be integrated into an assessment of how actively the weeds are growing. Actively growing weeds are more susceptible to herbicides and they generally reflect good crop growth conditions (adequate moisture and warm temperatures). Harker and Blackshaw (2001) have demonstrated a direct link between the rate of weed growth and the herbicide dose required to achieve acceptable efficacy.

In the second scenario the farmer actively changes the cropping systems and cropping practice to reduce weed pressure. In this case the farmer increases the likelihood that herbicide rates can be successfully reduced or that herbicides become unnecessary in some years.

Research Supporting Reduced Herbicide Rates

In the three following examples we summarize research relating to the two scenarios we have previously outlined. The first example provides support for reducing rates in common annual rotations. Examples 2 and 3 show the possibility of increasing the likeklihood of reducing dependence upon herbicides (or full herbicide rates) by altering cropping practice (AAFC-work at Brandon) or cropping systems (Glenlea rotation).

Example 1: Reduced herbicide rate studies

A number of studies on the effect of herbicide rate reduction on the efficacy of a number of common graminicides have been performed in western Canada recently. These provide some idea of the possibility of reducing rates in the case where no change in cropping system or cropping practice has been performed to reduce weed pressure. These studies have been done, therefore, in the context of common annual crop rotations. The results are then dependent upon the specific graminicide used, the tank mixes used, the weather experienced at each site-year, the crop vigour and the level of weed pressure. In none of these studies do the researchers pre-judge a scenario as being suitable or unsuitable for the application of reduced rates. It is left to chance.

Holm et al. (2000) summarized the effectiveness of Horizon, Achieve, Assert and Puma on wild oat control in a study spanning two sites (Scott and Saskatoon, SK) and including a total of 10 site years of data. In general the control of wild oat was unaffected for all 4 herbicides when rates were dropped from full rate to 2/3 rate. There was a significant drop in efficacy for all 4 herbicides when rates were dropped from 2/3 rate to 1/3 rate, this dropped also resulted in a significant drop in wheat yield and an average drop in net return across site-years and herbicides of $5/ha/year. When considering differences in efficacy between the 2/3 and full rate the authors noted that application timing according to weed leaf stage was a more influential factor than rate of herbicide, especially for Achieve and Assert, where earlier applications (2-3 leaf vs. 3-4 leaf) provided more cosnsitently high effficay.

Kirkland et al. (2001) considered the effect of rate and tank mixtures on the efficacy of Everest herbicide at 1 site (Scott, SK) over four years. There was no effect on wild oat efficacy when rates were reduced from full rate to 2/3 rate, regardless of carrier volume and wheat yield was also unaffected by the rate change. Tank mixes with either MCPA + 2,4 D, or Fluroxypyr + 2,4-D, or dicamba + mecaprop + MCPA resulted in significantly reduced wild oat efficacy.

Blackshaw and Harker (1996) studied the efficacy of Horizon at two sites (Lethbridge and Lacombe, AB) over two years. At Lethbridge under dry conditions biologically effective dose (BED) for wild oat required full rate or 1.5 X full rate in 1992 and 1993, respectively, while at Lacombe (a wetter site) the BED for wild oat was 1/3 and 2/3 registered rate in 1992 and 1993, respectively. They also tested the effect of tank mixes on efficacy and found that mixes with either dicamba or Ally reduced efficacy while mixes with 2,4-D, or Bromoxynil or Lontrel did not reduce efficacy on wild oat.

Example 2: wild oat management study - AAFC-Brandon

In a rotation of wheat with peas, fuel, fertilizer and herbicide inputs can be reduced without reducing yield or crop quality. The project consisted of wheat and peas grown in a tight rotation where fertilizer and herbicide rates varied in wheat following peas while the pea crop was all treated the same. In this way, the residual effects of peas on wheat could be determined and the impact of various management strategies in wheat on subsequent peas could also be ascertained. Research was conducted at two sites near Brandon MB. Soil types were sandy loam and clay. Fertilizer treatments in wheat were 25%, 50%, 75%, and 100% of recommended rates or 25, 50, 75, and 100 kg/ha of actual nitrogen (applied as 46-0-0 sidebanded at time of seeding along with P at constant rate of 40 kg/ha in form of 11-52-0). Herbicide rates were 100% and 66% of recommended rates for Horizon plus Target at each of the fertilizer rates for wheat. Each of these treatments was implemented in a high and low disturbance seeding system in wheat (HDS and LDS). For the LDS a pre-seeding burnoff of 0.5 L/acre of Roundup was used each year. Peas were also seeded in both seeding systems but fertilizer and herbicide rates were not varied (Odyssey was used as the herbicide in peas). Treatments were kept the same through time. For instance, seeding systems were kept constant in pea and wheat, and the low fertilizer and herbicide treatments in wheat were always in the same plots through time.

Wheat yield was the same regardless of fertilizer rate, herbicide rate, or seeding systems at the clay site. Despite 4 years of reduced inputs, using 25 kg/ha of N and 66% herbicide rate provide the same yield as 100 kg/ha of N and the full herbicide rate. Pea yields were the same regardless of seeding system or previous fertilizer rates used in wheat. Yields were lower where herbicide rates were previously reduced in wheat in the HDS system, but this did not occur in the LDS system. Results were different at the sandy loam site. Wheat yields were greater in the low disturbance seeding system and where low rates of fertilizer where used. At this site, yields dropped as rates of N went above 25 kg/ha, particularly in the HDS system. In some instances, full rates of herbicides provided higher yields. particularly in the HDS system. Pea yields at this site were the same between seeding systems and where rates of N and fertilizer were varied in the previous wheat crop. Wheat protein did not differ among treatments at both sites, therefore, concerns about lower protein levels where fertilizer rates are reduced in wheat after pea may not be well founded.

There were fewer weeds in the LDS system at both sites in wheat and especially in peas, therefore, weed competition may be lower in LDS than HDS. In the HDS system at the clay site, weed densities declined as rates of fertilizer increased, but not in the LDS system. Higher rates of fertilizer may be required to provide a competitive crop in HDS than LDS. In wheat at the sandy loam site, the lowest weed densities occurred in the 50% fertilizer treatment, therefore, if rates of fertilizer are too low, ie: 25%, competition is reduced and if rates are too high then weeds may access fertilizer to a greater extent than wheat. The previous fertilizer rate in wheat did not effect weed densities in LDS peas; however, in HDS, weed densities increased with increasing rates of N at both sites. This may be due to HDS and fertilizer reducing N fixation and the competitive ability of peas. Reducing herbicide rates to 66% of recommended in the previous wheat crop did not consistently increase weediness in peas.

2001 will be the last year of the trial and provide data to determine if the trends observed in previous years prevail over time. To date, the results suggest that the recommended rate of nitrogen applied to wheat after peas need to be lowered, and that the consequences of reducing herbicide rates in wheat are not that great in a subsequently grown less competitive crop like peas.

Example 3: The Glenlea rotation - University of Manitoba

A long-term crop rotation study was established at the University of Manitoba's Glenlea research station in the spring of 1992. The objectives are:

  1. to compare the biological and economic performance of conventional, low input, pesticide-free and organic crop production systems;
  2. to monitor the impact of crop rotation and crop input level on pest population dynamics.

The rotation treatmnents include:

  1. wheat-pea-wheat-flax
  2. wheat-sweetclover green manure- wheat-flax
  3. wheat- alfalfa (two years harvested for hay)-flax.

The individual mainplots are 2 acres in size. The main rotations are split into 4 subplots which involve different levels of crop inputs (full inputs- +herbicides/+ fertilizers; low input - +herbicide/-fertililzer or -herbicide/+fertilizer; organic - no fertilizer or herbicide inputs).

After the first eight years of this experiment, treatment differences have become obvious. In 1995 and 1999 all plots were seeded to flax and a complete comparison was made of the influence of crop rotation on the dependence on fertilizer and pesticide inputs. The rotation by input treatment interaction was significant in both 1995 and 1999. Achieving high flax yield and low weed biomass in the annual rotation was very dependent upon the use of fertilizer and pesticide inputs, while the same was not true for the rotation containing a two year break of alfalfa (Table 3). The density of some weeds was more affected than others. The populations of weed species that are seed or propagule limited (have relatively short-lived seed banks) such as wild oat and Canada thistle, are more greatly decreased by the inclusion of alfalfa in rotation than are populations of non-seed limited species such as wild mustard, redroot pigweed and green foxtail (Ominski et al. 1999). The highest average yearly return and the lowest cash risk (lowest average yearly input costs) were achieved in the alfalfa containing rotation under the no-input (F-H-) system (table 4). Even under the annual rotation a higher return was gained under the no-herbicide no-fertilizer system compared to the full inputs system, despite the difference in yield between the two systems (table 4). This indicates that the returns from inputs such as herbicides must always be scrutinized, even when they are used in systems that are dependent upon them to achieve high yield.

Table 3. Dry matter (g m-2) and grain yield (Kg/ha) for the Glenlea long-term cropping systems study test crop (flax) in 1995 and 1999. Statistical analysis performed on log transformed data (including all rotations) and means were considered significant when p<0.05. Results for only two rotations are presented here. Rotation 1 = wheat-pea-wheat-flax; Rotation 2 = wheat - alfalfa -alfalfa - flax. (Adapted from Humble 2001)
    1995   1999  
Rotation Inputs Weed DM Crop Yield Weed DM Crop Yield
Rotation 1 F+H+ 16.5 1877 6.3 1378
  F-H- 117.9 961 123.7 606
Rotation 2 F+H+ 3.4 1712 19.9 1454
  F-H- 43.2 1373 112.2 1379

Table 4. Average yearly input costs and net returns for Glenlea rotations from 1992-1999.
Crop Rotation F+H+
Full Inputs
Low Input
Low Input
Organic System
Input cost $104.14
Net return $27.87
Input cost $77.17
Net return $30.87
Input cost $71.36
Net return $26.67
Input cost $43.44
Net return $40.23
Input cost $71.68
Net return $77.83
Input cost $51.92
Net return $93.42
Input cost $55.92
Net return $73.73
Input cost $36.08
Net return $93.77


  • Reducing herbicide rates is possible. Even if the farmer does nothing to decrease the weed pressure in the cropping system there is still the possibility of using reduced herbicide rates on the basis of a real-time judgement of the scenario.
  • In order to consistently employ reduced herbicide rates, however, farmers must look to altering their cropping systems and cropping practices to favour the crop over the weeds and to reduce weed pressure.
  • It is even possible to develop cropping systems which are robust enough to function almost independent of herbicides.

Further Reading:

Anonymous. 2000. Manitoba agricultural review 2000. Manitoba Agriculture and Food, Program and Policy Analysis. Winnipeg, MB, Canada. Pp23.

Harker, K.N., and R.E. Blackshaw. 2000. Predicting when low herbicide rates may be effective. The Barley Country - Summer 2000 - Vol 9 No 2.

Harker, K.N., and R.E. Blackshaw. 1996. Growth stage and broadleaf herbicide effects on CGA 184927 efficacy. Weed Tech. 10:732-737.

Holm, F.A., K.J. Kirkland, and F.C. Stevenson. 2000. Defining optimum herbicide rates and timing for wild oat (Avena fatua) control in spring wheat (Triticum aestivum). Weed Tech. 14:167-175.

Humble, S. M. 2001. Weeds and ground beetles (Coleoptera:Caribidae) as affected by crop rotation type and crop input management. MSc. Thesis, University of Manitoba. Winnipeg, Canada, Pp 220.

Kirkland, K.J., E.N. Johnson, and F.C. Stevenson. 2001. Control of wild oat (Avena fatua) in wheat with MKH 6562. Weed Tech. 15:48-55.

Froese, N.T. and R.C. Van Acker. 2001. Dandelion distribution, interference and control in Roundup Ready canola. WSSA Abstract 41:100.

Knezevic, S.Z., P.H. Sikemma, F. Tardif, A.S. Hamill. K. Chandler and C.J. Swanton. 1998. Biologically effective dose and selectivity of RPA 201772 for preemergence weed control in corn (Zea mays). Weed Tech. 12: 670-676.

Ominski, P. D., M.H. Entz, and N. Kenkel. 1999. Weed supression my Medicago sativa in subsequent cereal crops: a comparative study. Weed Sci. 47:282-290.

Thomas A.G., J.Y. Leeson and R.C. Van Acker. 1999. Farm Management practices in Manitoba, 1997 Manitoba weed survey questionnaire results. AAFC Weed survey series 99-3. Saskatoon, SK, Canada.

Van Acker, R.C. A.G. Thomas, J.Y. Leeson, S.Z. Knezevic and B.L. Frick. 2000. 1997 weed survey of cereal and oilseed crops in Manitoba. Canadian Journal of Plant Science. 80:963-972.

Van Acker, R.C., D.A. Derksen, M.H. Entz, G. Martens, T. Andrews and O. Nazarko. 2001a. Pesticide-Free Production (PFP): an idea drawing farmers to implement Intergrated Pest Management. Proceedings of the Brighton Crop Protection Conference - Weeds. Pp 269-276.

Van Acker, R.C., D.A. Derksen, M.H. Entz, G. Martens, T. Andrews and O. Nazarko. 2001b. Pesticide-free production: a reason to implement integrated weed management. In Integrated Weed Management: Explore the Potential. Ed R.E. Blackshaw and L.M. Hall. Expert Committee on Weeds, Sainte-Anne-de- Bellevue, Qc, Canada. pp 61-73.

Copyright and Liability

This page created August 2004.