PROJECT RESULTS

The Impact of Herbicide Mode of Action on Low-Drift Spray Effectiveness

Applicant: 

Dr. Rene Van Acker

Department of Plant Science

University of Manitoba

Winnipeg, Manitoba  R3T 2N2  Canada

Table of Contents:

• Extension Report 1 - Response of Canola (Brassica napus L.) and Field Bean (Phaseolus vulgaris L.) to Simulated Drift of Glyphosate, MCPA or Thifesnulfuron:tribenuron
• Extension Report 2 - The Impact of Herbicide Mode of Action on Low-Drift Spray Effectiveness

ARDI Project:

#99-305

Total Approved:

$84,000

Date Approved:

February 8, 2000

Project Status:

Completed December, 2003

extension report 1:  response of canola (brassica napus l.) and field bean (phaseolus vulgaris l.) to simulated drift of plyphosate, mcpa or thifesnulfuron:tribenuron

Researchers:  Rene Van Acker, Jaret Sawchuk and Lyle Friesen, Department of Plant Science, University of Manitoba

Background and Objective:

There are very few good quality days for post-emergent herbicide application in Western Canada and the off-target movement of herbicide solutions can be relatively common.  Spray drift during herbicide application remains an ongoing economic and environmental concern for farmers.  Large-scale farming has lead to an increased use of custom applicators.  The machines used by custom applicators have the ability to cover a large amount of ground in a relatively short time.  This is desirable for Manitoba farmers because there are only 6 to 8 suitable post-emergence spraying days in a typical spraying season but it also means that herbicide application may often occur in less than ideal environmental conditions.  The effect of herbicide drift on non-target crops is important for farmers who wish to protect non-target crops and for those affected by drift who may wish to claim for damages.  This is especially true for high value crops.  In Manitoba, canola (Brassica napus) and navy beans (Phaseolus vulgaris) are two commonly grown high value crops which are susceptible to commonly used herbicides.  There is limited information available on the potential impact of herbicide drift on these crops.

The objective of this project was to characterize the effect of simulated herbicide drift of three commonly used herbicides; MCPA, glyphosate and thifensulfuron:tribenuron (2:1) on canola and navy beans.

Procedure and Project Activities:

A field study was conducted during the summers of 2001 and 2002 at two different sites.  Experiments were conducted at the Carman Research Station, in Carman, Manitoba, and at the Point in Winnipeg, Manitoba.  The experiments were each a randomized complete block design composed of 20 treatments and 4 replications.  Individual plots were 1.8 meters wide and 5 meters long.  For each of the two crops, agronomic practices did not differ between years.  However, agronomic practices between the two crops differed.  Standard fertilizer and seeding rates were used in all site-years.  Herbicidal weed control in-crop was used when necessary.  The canola variety was 46A65 (Proven Seed, Winnipeg, Manitoba, Canada) and the bean variety was Envoy (Aggasiz Seed Farms, Homewood, Manitoba, Canada).  When the canola reached the 2 to 4 true leaf stage, if it was necessary to over-spray for general weed control.  A tank-mix of ethametsulfuron-methyl (Muster), clopyralid (Lontrel 360) and sethoxydim (Poast Ultra) was used.  All herbicides were used at recommended label rates; Muster - 12 g/acre (22 g a.i. ha^-1), Lontrel - 0.34L/acre (302 g a.i. ha^-1), Poast Ultra - 0.27L/acre (300 g a.i. ha^-1).  Merge (adjuvant) was used at a rate of 1% v.  In the navy beans, trifluralin (Treflan EC) was applied at a rate of 0.93L/acre (1103 g a.i. ha^-1).  The trifluralin was incorporated the same day with a field cultivator to a depth of 7.5 cm.  When the beans reached the 2 to 3 trifoliate leaf stage, if it was necessary to over-spray for weed control.  The herbicide bentazon (Basagran) with the adjuvant Assist was used.  Approximately four days later, sethoxydim (Poast Ultra) was applied.  All herbicides were applied at recommended label rates; Basagran - 0.91 L/acre (1079 g a.i. ha^-1), and Assist - 0.74L ha^-1 and Poast Ultra - 300 g a.i. ha-¹ and Merge - 0.74L ha^-1.

For the application of simulated drift treatments carrier volume remained constant while the herbicide dose varied.  All treatments were applied with a bicycle plot sprayer using standard 80015 flat fan nozzles and application pressure was 275 kpa to produce a spray solution application rate of 110 L ha^-1.

Table 1.  Herbicide rates used to simulate drift of MCPA ester, glyphosate or thifensulfuron:tribenuron (2:1) on canola in 2001.  Registered field rate of each herbicide is also included.

Table 2.  Herbicide rates used to simulate drift of MCPA ester, glyphosate or thifensulfuron:tribenuron (2:1) on canola in 2002.  Registered field rate of each herbicide is also included.

Above ground crop biomass samples were taken approximately 14 to 21 days after simulated herbicide drift treatments were applied.  Biomass samples were taken by clipping crop plants at ground level within quadrants set within each plot.  For canola, biomass was collected from within two 0.25 m² quadrants per plot.  For beans a full m² of plant material was taken to measure above ground bean shoot biomass.  For grain yield harvest, canola was hand-swathed using sickles.  Individual canola plots had to be monitored separately for appropriate time for swathing because the various herbicide treatments caused differences in crop maturity.  The plots to which the lowest doses of each herbicide were applied matured first.  Canola swaths were combine harvested approximately 14 to 21 days after the last of the plots had been swathed.  Grain was screened for dockage and crop yield after removal of dockage was expressed as a percentage of the dockage-free untreated control for each site-year.  For navy beans, two full m² were clipped at ground level in each plot.  The plant biomass was put into burlap sacks and the sacks were hung in a drying room at 25°C for approximately two weeks.  After the two-week drying period the contents of each sack were threshed.  Yield samples were screened for dockage (cracked seeds were left in samples).  Bean yield after removal of dockage was expressed as a percentage of the dockage-free untreated control for each site-year.

Table 3.  Herbicide rates used to simulate drift of MCPA ester, glyphosate or thifensulfuron:tribenuron (2:1) on field beans in 2001.  Registered field rate of each herbicide is also included.

Table 4.  Herbicide rates used to simulate drift of MCPA ester, glyphosate or thifensulfuron:tribenuron (2:1) on field beans in 2002.  Registered field rate of each herbicide is also included.

^a Herbicide not applied to beans in Carman in 2002.

^b Herbicide not applied to beans in Carman in 2002

Statistical analysis of the dose-response curves from the field experiments closely followed the procedure outlined by Seefeldt et al. (1995).  Data initially were fitted to the log­-logistic model

y = C + (D - C)/(1 + exp(b(ln(x) - 1n(I₅₀))))

where y = crop yield or shoot dry matter (percentage of untreated control), x = herbicide dosage (g ha^-1; a small positive value of 1.0 was assigned to 0 g ha^-1 dosage to calculate natural logarithms), C = lower limit (asymptote) of the response curve, D = upper limit, b = slope, and I₅₀ = dose (g ai ha^-1 of the herbicide that reduced shoot dry matter by 50% relative to the untreated control).  The exp refers to e (the base of the natural logarithms) raised to the specified power, and 1n is the natural logarithms.  Individual curves were statistically tested systematically for common C and D, common b, and common I₅₀ using the lack-of-fit F test at the 0.05 level of significance.  Models for individual site-years were then combined when possible.  Models not fitting the log-logistic model were analyzed using a linear model

y = mx + B

where y = crop yield or shoot dry matter (percentage of untreated control), x = the herbicide dose, m = slope, and B = y axis intercept.  Data were fitted to the linear model using PROC GLM in SAS (Version 8.2).  Data that did not fit either of these models was graphed as treatment means with error bars representing standard deviation.

Results and Discussion:

At the highest MCPA dose, canola yield was reduced to approximately 0% of the yield of the untreated check in three of four site-years.  The Winnipeg 2002 site-year was much different than the other three site-years.  The MCPA seemed to have much less of an effect in this site-year.  Canola yield dropped to only 70% of the yield of the untreated check at the highest MCPA dose (280 g ae ha^-1) (Figure 1).

It was relatively easy to fit a log-logistic model to the canola yield data from plots treated with glyphosate, and it may, therefore be relatively easy to predict canola injury as a result of glyphosate drift.  At the highest simulated herbicide drift dose of glyphosate (200 g ai ha^-1), canola yield was reduced to between 0 and 30% of the yield of the untreated check.

Canola treated with simulated herbicide drift of thifensulfuron:tribenuron (2:1) produced data that was difficult to model suggesting that it may be difficult to predict canola yield loss for any given year as a result of thifensulfuron:tribenuron (2:1) drift.  In 2001, canola yield from plots treated with the highest dose of simulated herbicide drift of thifensulfuron:tribenuron (2:1) (2 g ai ha^-1) equaled approximately 15 to 20% of the yield of the untreated check (Figure 1).  Canola yield in 2002 was reduced to only 90% of the yield of the untreated check in Carman, and canola yield was actually 103% of the yield of the untreated check in Winnipeg at the highest dose of thifensulfuron:tribenuron (2:1) (2 g ai ha^-1) (Figure 1).

A log-logistic model fit the data for field bean biomass for all three herbicides used for simulated herbicide drift and for all site-years (data not shown) but it was not necessarily possible to model the effect of herbicide drift on bean yield because there was considerable recovery of the beans from the effect of herbicide drift, especially for the MCPA treatments.  At the highest dose of MCPA, field bean biomass was reduced to between 30 and 50% of the biomass of the untreated check (data not shown) yet at the highest dose of simulated herbicide drift of MCPA, field bean yield ranged from 50 to 120% of the yield of the untreated check.

At the highest glyphosate dose, field bean yield was reduced to between 10 and 20% of the bean yield of the untreated control for all site-years (Figure 2).

It was difficult to fit a log-logistic model to field bean yield data for beans treated with thifensulfuron:tribenuron (2:1).  Field bean yield would be very difficult to predict after drift from thifensulfuron:tribenuron (2:1) because at the highest simulated herbicide drift dose, field bean yield ranged between 50 and 90% of the yield of the untreated check.

It can be difficult to estimate yield based on low herbicide dosages applied early in the growing season because there is a great deal of time for plants to compensate, thus minimizing the effect of the herbicide.  As herbicide rate increases, visual injury rating becomes more consistent (data not shown).  For both canola and field beans, visual injury estimates were much more accurate at the highest herbicide doses.  Standard deviations of the visual injury ratings decreased as herbicide dose increased.  At the lowest herbicide doses, standard deviations were quite high.

Glyphosate was the most effective of the three herbicides in our experiments.  Field beans may be much more tolerant to herbicide drift than we initially suspected.  Canola was much more sensitive than the field beans to all three herbicides used to simulate herbicide drift.  There is not much information available regarding the sensitivity of field beans to sublethal herbicide doses.  Both canola yield and field bean yield were highly correlated with the visual injury rating for glyphosate (r = -0.88 and r = -0.66, respectively).  There was a very poor correlation between field bean yield and the visual injury rating for MCPA or thifensulfuron:tribenuron (2:1) (r = -0.36).  Yield and visual injury ratings were not always well correlated because it appears as though both canola and field bean were able to compensate to various degrees for the damage caused by the simulated herbicide drift.  In the canola plots treated with the simulated herbicide drift of glyphosate, correlation between yield and number of seeds per pod was r = 0.5.  In the canola plots treated with MCPA or thifensulfuron:tribenuron (2:1), the correlation between yield and number of seeds per pod was r = 0.80 (Table 4.2).  This indicates that when canola is injured by sublethal doses of herbicide, it attempts to compensate by producing more seeds per pod.  In the field bean plots treated with glyphosate, correlation between yield and number of seeds per pod was quite low ( r = 0.31), but the correlation between yield and thousand kernel weight was much higher (r = 0.66).  In the field bean plots treated with simulated herbicide drift of MCPA or thifensulfuron:tribenuron (2:1), yield was not well correlated with either number of seeds per pod (r = 0.45) or thousand kernel weight (r - 0.44).  This indicates that when field beans are injured by sublethal doses of glyphosate, they compensate by producing heavier seeds.

For canola, the most damaging herbicide in terms of effect on yield vs. percent of field rate was thifensulfuron:tribenuron (2:1).  The simulated herbicide drift of thifensulfuron:tribenuron (2:1) at a rate of 13% of the field rate caused a 46% decrease in yield.  A 50% yield loss in the canola plots treated with simulated herbicide drift of MCPA occurred at 18% of the field rate.  A 50% yield loss in the canola plots treated with glyphosate occurred at 23% of the field rate.

For field beans, the most damaging herbicide in terms of effect on yield vs. percent of field rate was glyphosate.  The simulated herbicide drift of thifensulfuron:tribenuron (2:1) at a rate of 80% of the field rate caused only a 25% decrease in field bean yield.  A 25% yield loss in the bean plots treated with simulated herbicide drift of MCPA occurred at 400% of the field rate.  A 48% yield loss in the bean plots treated with glyphosate occurred at 28% of the field rate on field bean yield loss.

Conclusion:

In general, canola was much more sensitive than field beans to the simulated drift of the three herbicides we tested.  Crop yield resulting from herbicide drift early in the field season may be very difficult to predict.  Herbicide injury symptoms that usually are worrisome to growers, such as mottled chlorosis and leaf stunting, crinkling, and discoloration caused by sulfonylureas; cupping of terminal leaf, crinkling and leaf stunting caused by MCPA; chlorosis of the upper leaves caused by glyphosate; and chlorosis and discolouration of growing points caused by thifensulfuron:tribenuron, can occur at rates much lower than required to reduce yield.  We had hoped to produce a scale or index that could be used by farmers, custom applicators and insurance claim adjusters that would help them settle any claims resulting from herbicide drift of MCPA, glyphosate or thifensulfuron:tribenuron (2:1) on canola and field beans.  This may not be immediately possible because of the large variations in the effect of simulated drift on yield between years for both canola and field beans.  However, both canola and field bean appear to be more tolerant to MCPA, glyphosate and thifensulfuron:tribenuron (2:1) than was previously thought, but this does not mean that these crops are safe from the potential effects of drift from these herbicides.  Caution and care must still be exercised when applying herbicides around these crops.

extension report 2:  THE IMPACT OF HERBICIDE MODE OF ACTION ON LOW-DRIFT SPRAY EFFECTIVENESS

Researchers:  Rene Van Acker and Rufus Oree, Department of Plant Science, University of Manitoba; Tom Wolf, Agriculture and Agri-Food Canada Saskatoon; Rick Holm and Ken Sapsford, Crop Development Centre, University of Saskatchewan;  and Linda Hall, Alberta Agriculture, Food and Rural Development

Background and Objectives:

Management of spray drift continues to dominate in discussions of how to reduce the environmental impact of pesticides.  Factors leading to decreased tolerance to drift include greater use of non-selective herbicides during times of acute sensitivity of the ecosystem (i.e. in herbicide tolerant crops), products with greater activity, and more diversification to sensitive special crops.  Recent initiatives by the Pest Management Regulatory Agency (PMRA) to protect environmentally sensitive areas (such as shelter belts and water bodies) from drift damage through "buffer zones" can significantly restrict pesticide use on farm land.

New "venturi" nozzle technologies that dramatically reduce spray drift were introduced into Canada in 1997.  These nozzles incorporate air into the spray pattern, and reduce drift by up to 95%.  Since their introduction, ten different venturi-type nozzle models have become available and these are very popular with applicators.  In 2000, new low-drift nozzles that do not use air amendment were introduced, and these will likely also have good market potential.  It is not known how the lack of air-induction will affect biological performance of low-drift herbicide sprays.

The limiting factor to low-drift nozzle adoption is lack of knowledge in efficacy performance.  Applicators are reluctant to use application methods that may reduce pest control, for which they assume all risk.  Conversely, agrichemical companies are reluctant to endorse low-drift measures if product performance may be compromised.  The only way out of this dilemma is to provide good information to all parties.  Lowering the risk, coupled with demonstration and communication of effective weed control with low-drift sprays, will enhance adoption of new methods, ultimately protecting the environment.  Buffer zones may be reduced for low-drift application methods, providing additional encouragement for applicators to spray in a safe manner.

The objectives of this project were:

1. To establish the effect of conventional and air-amended drift reducing sprays on herbicide effectiveness.

2. Identify the limits of spray coarseness for various weed - herbicide combinations.

Procedure and Project Activities:

Field studies were carried out at the University of Saskatchewan Kernen Research Farm, the University of Manitoba, Carman Research Farm and field sites operated by Alberta Agriculture, Food and Rural Development in 2000, 2001 and 2002.  Natural weed populations were used at each site.  In some cases weeds were seeded into the plots to ensure the presence of certain species at certain sites-years.  Herbicides were always applied when weeds and crops were at the recommended stage for treatment and when weather conditions were at least reasonable (if not optimal) for efficacy.

Three spray qualities (medium, coarse and very coarse) were applied using conventional flat fan and air-included tips.  Spray qualities are intended only to differentiate nozzles in a generic manner, and only approximately reflect ASAE quality classification.  Nozzles included in this study were:

• Non-air induced standard and pre-orifice flat fans at 40 psi; Wilger ComboJet ER8002 (medium spray), Delavan RF8002 (coarse spray), Wilger ComboJet DR8002/04 (very coarse spray)

• Air-induced tips, each at about 70 psi, delivering the equivalent to an 8002 at 40 psi; Air Bubble Jet 110015 (medium spray), Greenleaf TurboDrop XL 110015 (coarse spray), SprayMaster Raindrop Ultra 110015 (very coarse spray)

Nominal flow rates were lower, and spray pressures were higher, for the air-induced tip to be consistent with current recommendations for optimum efficacy, and also to more closely match spray deposits of paired nozzles.  In-flight spray droplet sizes for each nozzle were determined on a Phase / Doppler Particle Analyzer, using a traverse scan along the x-axis (mean droplet size) and using y-axis traverses at 5-cm increments to document changes in spray characteristics along the x-axis.  Nozzles were calibrated by collecting output for 30 seconds from each nozzle.  Pressures were adjusted as needed to ensure that all nozzles had equivalent output so that a single travel speed could be used.

Ten herbicide treatments, comprising 14 active ingredients and eight modes of action were applied.  Each herbicide was applied at two non-zero rates. An untreated control plot was included for each herbicide.  All trials were conducted with a 4-nozzle sprayer.  Nozzle height was maintained at 50 cm above the targets for all treatments.  Travel speeds were about 8.9 km/h.  Weed control was visually assessed at two dates during the growing season.  Weed biomass was determined by sampling two 1/4 m² quadrants in each plot.  Weed counts were conducted as required.  Grain yield and dockage as determined at harvest.

The field experiment comprised a total of 142 treatments (6 application methods * 2 rates + control * 10 herbicides, plus 2 carrier volumes for Reglone).  Plots were laid out in a split plot design, with rates as main plots.  Treatments were replicated 4 times.  A total of 16 herbicide active ingredients were applied alone or in tank mixes at 2 rates (lx and 0.5x or 0.75x) (Figure 1). Applications were done in 100 L/ha carrier at the recommended weed stage.  The main grassy weeds analyzed were wild oat (Avena fatua), green foxtail (Setaria viridis), and barnyardgrass (Echinochloa crusgalli). The main broadleaf weeds were wild mustard (Sinapsis arvensis), wild buckwheat (Polygonum convolvulus), redroot pigweed (Amaranthus retroflexus), green smartweed (Polygonum spp.), cleavers (Galium aparine), chickweed (Stellaria media), hempnettle (Galeopsis tetrahit), lambsquarters (Chenopodium album), dandelion (Taraxacum officinale), and Canada thistle (Cirsium arvense).  ANOVA was conducted on weed control for each weed/herbicide combination at each site, each year.  A total of 266 ANOVAs were summarized (73 on grassy weeds, 193 for broadleaf weeds).  The frequency of statistically significant effects (p=0.05) for each weed type and herbicide mode of action was calculated for all site years.  A total of 798 statistical effects were analyzed.

Figure 1.  Herbicide and Crop Treatments for the Field Experiments

Results and Discussion:

Herbicide rate was the most important determinant of weed control, having significant effects in 49% of cases (63% for grasses, 44% for broadleaves).  Groups l, 8 and 9 were most sensitive to changes in rate for grasses, and Group 4, 9 and 22 were most sensitive for broadleaves (Figure 2).  This means that herbicides in Groups 2 and 10 were relatively unaffected by rate of herbicide (all other factors combined).  Herbicide rate was the factor which most frequently affected herbicide efficacy.  Of course we used only two rates in this experiment (1/2 and full rate), but this result shows that herbicide rate is a much more important factor than nozzle type when it comes to herbicide efficacy.  This was especially true for grassy weed species, including wild oat and green foxtail.

Spray quality had a significant effect on weed control 21% of the time (34% and 17% for grasses and broadleaf weeds, respectively).  The frequency of significant effect for this factor was less than half of the frequency of effect for herbicide rate.  Groups 1, 9 and 10 were most sensitive to spray quality for grasses, whereas Groups 6 and 9 were most sensitive for broadleaves (Figure 3).  It is somewhat surprising that spray quality effect was not more frequent for Group 22 (Reglone) or for Group 10 (Liberty) on broadleaf weeds.  Both of these products are contact herbicides for which spray quality might be expected to significantly influence efficacy.  In all cases we used 1101/ha (10 gallons/acre) of spray solution volume, and it is possible that for contact herbicides spray volume may be a more critical factor than spray quality when it comes to efficacy.

Air induction had relatively minor effects, being significant in only 15% of cases for both grasses and broadleaves (less than 1/3 the frequency of effect of herbicide rate).  Groups 2, 8 and 9 were least sensitive to air-induction on grasses, Groups 2 and 10 were least affected for broadleaves (Figure 4).  On grasses, Group 2 products were less sensitive to herbicide rate, spray quality, and air-induction than Group 1 products (Figure 5).  Group 2 products appear better suited than Group 1 products for the use of low-drift sprays.

Conclusion:

Grassy weeds were more sensitive to application method than broadleaf weeds.  Group 1 and Group 9 products were most sensitive to application method.  In terms of efficacy, the contact herbicides (Groups 6, 10 and 22) were not necessarily those most sensitive to spray quality.  Group 2 products were least sensitive for both grassy and broadleaf weed control.  Herbicide rate affected weed control much more frequently than spray quality.  With low-drift spray nozzles, weed control efficacy levels equivalent to that of regular spray nozzles (non-air induced, non-low drift) can be achieved in the majority of cases.  Some caution may be warranted when using low drift nozzles to control grassy weed infestations with Group 1 or Group 10 herbicides, especially if the low-drift nozzle delivers a coarse spray pattern.  However, rate of herbicide is a much more significant consideration than spray nozzle type.  The most ideal situation, one which allows for spray drift reduction and the most reliability in efficacy, may be to employ low-drift nozzle types which also offer traditional spray quality (medium spray pattern) such as the Air BubbleJet nozzle used in this study.

ARDI Home - Contact ARDI