Weed Control, Environmental Impact and Profitability of Weed Management Strategies in Glyphosate-Resistant Corn

Eleven field trials were conducted over a three-year period (2006-2008) at three locations in southwestern Ontario, Canada to evaluate the effect of various weed management strategies in glyphosate-resistant corn on weed control, crop injury, corn yield, environmental impact and profit margin. No visible injury resulted from the herbicide treatments evaluated. Overall, the effect of all factors assessed were location specific. By 56 days after treatment, depending on location, glyphosate applied at the 7-8 leaf stage (LPOST), preemergence (PRE) herbicides followed by (fb) glyphosate LPOST and sequential glyphosate applications (EPOST (3-4 leaf stage) followed by LPOST) provided more consistent control of annual broadleaf weeds and annual grasses compared to glyphosate applied alone EPOST. Weed control at 56 days after treatment was lower when glyphosate was applied alone LPOST compared to sequential applications of glyphosate or PRE herbicides fb glyphosate. There were no differences in corn yield among the sequential programs evaluated; however, a yield benefit was found when a sequential program was used compared to glyphosate applied alone LPOST. Among the sequential programs the lowest environmental impact was isoxaflutole/atrazine fb glyphosate. The lowest profit margins were associated with atrazine, S-metolachlor/atrazine/benoxacor, dicamba/atrazine and glyphosate LPOST treatments compared to all other treatments. Overall, profit margins tended to be somewhat higher for treatments that included glyphosate applications. Based on these results, the most efficacious and profitable weed management program in corn was a sequential application of glyphosate; however, isoxaflutole/atrazine fb glyphosate was the treatment with the lowest environmental risk while also adding glyphosate stewardship benefits.


Introduction
The demand on growers to economically produce environmentally sustainable food while maintaining herbicide stewardship is increasing.To achieve this goal, data on weed control, crop yield, economics and environmental impacts of herbicides are needed to help identify the most advantageous herbicide program.Growers of glyphosate-resistant corn have several weed management options, including pre-emergence (PRE), post-emergence (POST), tankmixes and sequential applications.Traditional management strategies for corn have included atrazine [1,2]; however the future of atrazine use in corn is unclear [3,4].Since the introduction of glyphosate resistant corn, one-or two-pass glyphosate-only applications are now options that can simplify weed manage-ment and can be an effective method used to improve weed control [2,[5][6][7].However, sole reliance on glyphosate increases weed selection pressure, potentially selecting for glyphosate-resistant weeds [2,[8][9][10].Tank-mixes or sequential applications that utilize more than one herbicide modes of action can reduce selection pressure [2].
Timing of herbicide application is also critical for effective weed control [11][12][13][14].A single-pass herbicide program, PRE or POST can result in weed escapes if the program fails to control all weeds or has no soil residual [13,15].Weed escapes can be more difficult to control due to increased size, resulting in reduced herbicide efficacy of rescue sprays [16,17].Later germinating weeds are of particular concern with a single-pass application of glyphosate because of its lack of residual control [6,18].Sequential in-crop applications of glyphosate or combining a residual PRE herbicide with a POST application To justify the most appropriate weed management strategy, the decision-making process must include an assessment of herbicide efficacy, environmental impact and economic profitability, while recognizing that tradeoffs among those factors will occur.Therefore, the objective of this study is to determine which herbicide strategy for glyphosate-resistant corn will be most efficacious and economically profitable while providing low environmental impact.

Site Descriptions and Procedures
Eleven field trials were conducted in southwestern Ontario at the Greenhouse and Processing Crops Research Centre, Agriculture and Agri-Food Canada, Harrow, Ontario in 2007 1.
Procedures at all sites were the same unless otherwise noted.Experiments were arranged in a randomized complete block design with four replicates.There were a total of thirteen treatments: a non-treated weedy control, a weed-free control, atrazine (1000 g•ai•ha −1 ), s-metolachlor/atrazine/benoxacor (1080 g•ai•ha −1 ), isoxaflutole (40 g•ai•ha −1 ) + atrazine (400 g•ai•ha −1 ), dicamba/atrazine (1000 g•ai•ha −1 ), glyphosate (900 g•ae•ha −1 ; 3 -4 leaf stage, EPOST), glyphosate (900 g•ae•ha −1 ; 7 -8 leaf stage, LPOST), dicamba/atrazine (1000 g•ai•ha −1 ) followed by (fb) glyphosate (900 g•ae•ha −1 ), atrazine (1000 g•ai•ha −1 ) fb LPOST glyphosate (900 g•ae•ha −1 ), s-meto-  Crop injury was estimated visually 7, 14 and 28 days after treatment (DAT), using a scale of 0 to 100% where a rating of 0 was defined as no visible plant injury and a rating of 100 was defined as plant death.Percent weed control was visually assessed 28 and 56 DAT using a scale of 0% to 100% where a rating of 0 was defined as no visible weed control and a rating of 100 was defined as complete control.Only data from 56 DAT are presented in this manuscript.Corn was mechanically harvested at physiological maturity using a plot combine at all sites.Corn yields were adjusted to a 15.5% moisture level.

Statistical Analyses
All data were subjected to analysis of variance and analyzed using the PROC MIXED procedure in SAS statistical software (Version 8. SAS Institute, Inc., Box 8000, SAS Circle, Cary, NC 27512).Variances were partitioned into the fixed effect of herbicide treatment and into the random effects of environment (year and location).When there was no significant interaction between environment and treatment the data were pooled.Contrast comparisons among herbicide application timings represent a priori orthoganol contrasts.The assumptions of the variance analysis were tested by ensuring that the residuals were random, homogeneous, with a normal distribution about a mean of zero using residual plots and a Shapiro-Wilk normality test.All percentage data required an arcsine square root transformation.Yield data did not require transformation.All percentage data presented in tables are on the back-transformed scale.Treatment means were separated at the 5% level of significance using a Fisher's Protected LSD test.

Environmental Impact
The environmental risk for each herbicide treatment was determined using published EIQ values for all active ingredients (a.i.) [24].However, the EIQ for atrazine, metolachlor, and isoxaflutole were recalculated based on PRE vs POST application, where the plant surface persistence value (P) of 1 was used instead of 3, respectively.The environmental impact of each treatment was calculated by multiplying herbicide EIQ by the amount applied in kg•ai/ae•ha −1 .For herbicide products and/or tank mixes that contain more than one a.i., the EI was calculated by summing EIQs at the appropriate proportion.

Profitability Analysis
The profitability analysis is based on the level of profit margins over weed control costs, measured as gross income less herbicide and application costs.Gross income for each replication was calculated as the yield multiplied by average price for corn between 2006  .All other costs of production are assumed to be constant across treatments, thus they are not considered in the analysis.Pairwise comparisons are made between treatments using SPSS (SPSS Software, Version 16.0 SPSS Inc., 233 S. Wacher Drive, Chicago, IL 60606) to test for significant differences in average profit margins between treatments.These pairwise comparisons are made across all locations and years as well as for each location (RCA, RCB, Exeter, and Harrow) in each year.This allows for testing for overall differences in profitability between treatments as well as for testing for variations in relative profitability for specific treatments between different locations, between different years, and

Weed Control
The dominant weed species in this study were redroot pigweed, common ragweed, common lambsquarters and velvetleaf.Because annual grass species varied by location, all species were grouped together for analysis.

Redroot Pigweed
All treatments provided greater than 90% control of redroot pigweed (Table 2).When glyphosate was applied EPOST redroot pigweed control was reduced by 7% compared to LPOST glyphosate, a PRE herbicide fb glyphosate or a sequential glyphosate application.Stewart et al. [7] also previously demonstrated that an EPOST application of glyphosate provided 7% -11% lower redroot pigweed control compared to a sequential application of glyphosate in corn.In contrast, a sequential application of glyphosate did not improve redroot pigweed control in comparison to a single EPOST application in soybean [13].Nurse et al. [6] reported a decrease of 35.9 plants m −2 of redroot pigweed when flufenacet + metribuzin fb glyphosate was applied compared to when glyphosate was applied alone in corn.The results of several other studies support that late-emerging weeds are controlled most effectively using sequential applications of glyphosate or by following a PRE herbicide with a POST application of glyphosate [6,18,20].

Common Ragweed
Preemergence application of S-metolachlor/atrazine/benoxacor provided less than 77% control of common ragweed except at RCA in 2007 (   4).A PRE herbicide fb glyphosate only had higher common lambsquarters control compared to a sequential glyphosate application at Exeter in 2006, otherwise the two programs did not differ.Generally, sequential glyphosate application increased control by 3% -28% and 2% -21% compared to glyphosate applied alone EPOST or LPOST, respectively, depending on location.The benefits of sequential glyphosate applications on common lambsquarters have been previously reported in both corn and soybean where control was between 5% -9% and 4% -9% higher compared to a single application of glyphosate [7,20].This makes sense because sequential in-crop applications of glyphosate offer a grower an opportunity to control weeds escaping an EPOST glyphosate application while late emerging weeds may have been too large at the time of application to be completely controlled by a LPOST glyphosate application [5,16,18].However, under certain environmental conditions, a single application of glyphosate has been shown to provide adequate season long control of common lambsquarters eliminating the need for a sequential application [13].

Velvetleaf
Preemergence herbicides applied alone provided less than 75% control of velvetleaf within Environment 1 (Table 5).As expected, this is largely due to the inadequate control of velvetleaf with atrazine alone or S-me- Gonzini et al. [20] found that in glyphosate-resistant soybean control of giant foxtail improved by 2% -15% with sequential applications of glyphosate POST or PRE herbicides followed by a POST application of glyphosate compared to a single-pass application of glyphosate POST.

Crop Injury and Yield
There was no visual crop injury at 7 and 28 DAT for all treatments in this study (data not shown).At most locations, PRE herbicides fb glyphosate resulted in higher yields than when glyphosate was applied alone LPOST (Table 7).Nurse et al. [6] demonstrated that corn yield increased when a PRE herbicide (flufenacet + metribuzin) was fb glyphosate compared to glyphosate alone.They attributed this response to early season weed control.In our study, sequential applications of glyphosate also had higher corn yield by up to 1.9 MT•ha −1 compared to glyphosate applied LPOST at most locations.We attribute these higher yields to increased weed control.

Environmental Impact
The type of application impacts the EIQ of a pesticide.Kovach et al. [23] considered that all POST herbicides have a greater risk due to plant surface persistence, which factors into the farm worker, consumer and environmental components of the EIQ calculation.The EIQ of PRE applied atrazine, metolachlor, and isoxaflutole (Table 8) were on average 8.8 lower than the same herbicide applied POST.With approximately an order of magnitude difference in the EIQ, it is critical to consider the type of application when evaluating the relative risk of weed control strategies.The lowest EI of 6.0 was atrazine + isoxaflutole, due to the low application rate.In contrast, due to higher EIQ values, the highest EI of herbicide products applied alone was atrazine + dicamba at 24.6.Clearly, adding another active ingredient to the weed management strategy will increase the EI, but for all treatments other than atrazine + isoxaflutole, adding glyphosate less than doubles the EI.For atrazine + isoxaflutole the addition of glyphosate more than triples the EI.
Sequential glyphosate applications had a higher EI than all products applied alone and atrazine + isoxaflutole fb glyphosate.The EI of applying glyphosate twice was equivalent to applying atrazine + dicamba alone, atrazine fb glyphosate, atrazine + metolachlor fb glyphosate and atrazine + isoxaflutole fb glyphosate.Thus, the addition of the aforementioned PRE herbicides to a Data were pooled by environment (location and year) when the interaction between environment and treatment was non-significant.Means are presented on the back-transformed scale.Means followed by the same letter within a column are not significantly different according to Fisher's Protected LSD (P < 0.05).b a priori orthogonal contrasts.* = significant (P < 0.05); Abbreviations: PRE, preemergence; POST, postemergence; EPOST, early postemergence; LPOST, late postemergence; WF, weed-free; fb, followed by; RCA, Ridgetown Site 1; RCB, Ridgetown Site 2; NS, not significant.glyphosate-resistant corn production is strongly recommended based on equivalent environmental risk as well as resistance management.

Profitability Analysis
The results of the profitability analysis indicate that across all locations and years no significant differences in profit margins exist between the following treatments: isoxaflutole + atrazine, glyphosate EPOST, glyphosate LPOST, dicamba/atrazine followed by (fb) glyphosate, atrazine fb glyphosate, S-metolachlor/atrazine/benoxacor fb glyphosate, isoxaflutole + atrazine fb glyphosate and sequential applications of glyphosate (Table 9).Aside from the weedy check treatment, the lowest profit margins were found in the atrazine and S-metolachlor/ atrazine/benoxacor treatments.Other than the weedy check, relatively few significant differences existed between treatments within each location and within each year.At both Ridgetown locations, profit margins across all three years only were reduced for the atrazine and S-metolachlor/atrazine/benoxacor treatments compared to almost all other treatments.At Exeter, profit margins for the S-metolachlor/atrazine/benoxacor treatment were lower than the profit margins for dicamba/atrazine, glyphosate EPOST and sequential applications of glyphosate.At Harrow, only the atrazine and dicamba/atrazine treatments were significantly lower in profit margins than most of the other treatments.Similarly, other than the weedy check, there are few differences in profit margins between treatments found within each year.In 2006 and 2007, atrazine and S-metolachlor/atrazine/benoxacor have reduced profit margins compared to all other treatments.In addition, in 2007 isoxaflutole + atrazine had reduced profit margins compared to several other treatments (Table 9).In 2008, there are no differences in profit margins between any of the herbicide treatments.
Relatively little change in these results can be found when examining profit margins within each location-year (Table 10), as the significant differences that exist are consistent for the most part with the results discussed above.Among the 11 location-years, treatments that are found to have recurring lower profit margins compared to other treatments include the atrazine, S-metolachlor/ atrazine/benoxacor, dicamba/atrazine and glyphosate LPOST treatments.Overall, profit margins tend to be somewhat higher for treatments that include glyphosate applications.

Conclusions
In summary, PRE herbicides fb glyphosate LPOST and sequential glyphosate applications provided greater control of annual broadleaf weeds and annual grasses compared to single glyphosate applications.This is due to control of weed escapes through residual control of PRE herbicides and/or the control provided by a second glyphosate application LPOST.Generally, weed control with glyphosate applied LPOST was greater than weed control with glyphosate applied EPOST; however, this a Means within columns that are followed by the same letter are not significantly different from each other (P < 0.05).Abbreviations: EPOST, early postemerence; LPOST, late postemergence; fb, followed by; RCA, Ridgetown Site 1; RCB, Ridgetown Site 2. g generally did not translate into a yield benefit and often resulted in yield losses due to prolonged early season weed competition.A yield benefit was found with both sequential herbicide programs compared to a single application of glyphosate LPOST.However, corn yield did not differ between the sequential herbicide programs and glyphosate applied EPOST.
Based on the EIQ, the sequential herbicide programs had a greater environmental impact than one-pass herbicides programs, except dicamba/atrazine alone, which was equivalent to several two-pass treatments.Isoxaflutole/atrazine fb glyphosate had the lowest environmental impact of the sequential herbicide programs evaluated.Overall, profit margins were moderately higher for treatments that included a glyphosate application (glyphosate alone or applied LPOST following a PRE herbicide).While glyphosate applied EPOST may appear to achieve the same management goals, the impact of reduced weed control over time could negatively affect future profit margins.Overall, these data showed that glyphosate fb glyphosate was the most efficacious and profitable treatment; however, for the purposes of glyphosate stewardship, and reduced environmental impact, isoxaflutole/atrazine fb glyphosate is also recommended as a profitable weed management system.
and 2008, at the Huron Research Station, Exeter, Ontario in 2006, 2007 and 2008 and two different sites at the University of Guelph, Ridgetown Campus, Ridgetown, Ontario in 2006, 2007 and 2008 (RCA, Site 1 and RCB, Site 2).Soil descriptions from each location can be found in Table

Table 2 . Mean percent control of AMARE in response to weed management strategies 56 days after treatment at Exeter and Ridgetown, ON from 2006 to 2008 and Harrow, ON from 2007 to 2008 a .
b WF vs glyphosate EPOST * WF vs glyphosate LPOST NS Glyphosate EPOST vs PRE fb glyphosate LPOST * Glyphosate LPOST vs PRE fb glyphosate LPOST NS Glyphosate EPOST fb glyphosate LPOST vs PRE fb glyphosate LPOST NS Glyphosate EPOST vs glyphosate EPOST fb glyphosate LPOST * Glyphosate LPOST vs glyphosate EPOST fb glyphosate LPOST NS Glyphosate EPOST vs glyphosate LPOST * a Data were pooled by environment (location and year) when the interaction between environment and treatment was non-significant.Means are presented on the back-transformed scale.Means followed by the same letter within a column are not significantly different according to Fisher's Protected LSD (P < 0.05); b a priori orthogonal contrasts; * = significant (P < 0.05).Abbreviations: AMARE, redroot pigweed; PRE, preemergence; POST, postemergence; EPOST, early ostemergence; LPOST, late postemergence; RCA, Ridgetown Site 1; RCB, Ridgetown Site 2; WF, weed-free; fb, followed by; NS, not significant p Weed Control, Environmental Impact and Profitability of Weed Management Strategies in Glyphosate-Resistant Corn 1598

Table 3 )
. Furthermore, when applied alone, atrazine PRE had no control of common ragweed at RCA in 2006 and less than 30% control at RCB in 2006 and 2007.By 56 DAT common ragweed may have escaped the soil residual provided by atrazine, resulting in the poor control observed with Smetolachlor/benoxacor/atrazine or atrazine alone at these locations.Glyphosate applied LPOST had as much as 17% higher common ragweed control at Exeter in 2006 compared to when glyphosate was applied EPOST.This is most likely due to common ragweed emerging after the EPOST application.Sequential applications of glyphosate increased common ragweed control compared to glyphosate applied EPOST by 19%, 2%, 22%, and 2% at RCA, RCB, and Exeter in 2006 and RCB in 2007, respectively (Table 3).Sequential glyphosate application also increased common ragweed control by 21% and 5% at RCA 2006 and Exeter 2007, respectively, compared to a LPOST application of glyphosate.Generally, the appli-

Table 3 . Mean percent control of AMBEL in response to several weed management strategies 56 days after treatment at Exe- ter and Ridgetown, ON from 2006 to 2008 and Harrow, ON, 2008 a .
a Data were pooled by environment (location and year) when the interaction between environment and treatment was non-significant.Means are presented on the back-transformed scale.Means followed by the same letter within a column are not significantly different according to Fisher's Protected LSD (P < 0.05); b a priori orthogonal contrasts.* = significant (P < 0.05).Abbreviations: AMBEL, common ragweed, PRE, preemergence; POST, postemergence; EPOST, early ostemergence; LPOST, late postemergence; RCA, Ridgetown Site 1; RCB, Ridgetown Site 2; WF, weed-free; fb, followed by; NS, not significant.

Table 4 . Mean percent control of CHEAL in response to several weed management strategies 56 days after treatment at Exe- ter and Ridgetown, ON from 2006 to 2008 and Harrow, ON from 2007 to 2008 a .
and RCA and in 2008 within Environment 1. Delaying glyphosate application improved common lambsquarters control by 12%, 12%, 2% and 3% at Exeter in 2006 and Harrow, RCA and RCB in 2007, respectively, compared to glyphosate applied EPOST (Table a Data were pooled by environment (location and year) when the interaction between environment and treatment was non-significant.Means are presented on the back-transformed scale.Means followed by the same letter within a column are not significantly different according to Fisher's Protected LSD (P < 0.05); b a priori orthogonal contrasts; * = significant (P < 0.05); Abbreviations: CHEAL, common lambsquarters, PRE, preemergence; POST, postemergence; EPOST, early postemergence; LPOST, late postemergence; RCA, Ridgetown Site 1; RCB, Ridgetown Site 2; WF, weed-free; fb, followed by; Env1 = Exeter and Harow 2008; Env2 = RCA and RCB 2008; NS, not significant.r

Table 5 . Mean percent control of ABUTH in response to several weed management strategies 56 days after treatment at Ridgetown, ON from 2006 to 2008 a .
a Data were pooled by environment (location and year) when the interaction between environment and treatment was non-significant.Means are presented on the back-transformed scale.Means followed by the same letter within a column are not significantly different according to Fisher's Protected LSD (P < 0.05); b a priori orthogonal contrasts; * = significant (P < 0.05); Abbreviations: ABUTH, velvetleaf; PRE, preemergence; POST, postemergence; EPOST, early postemergence; LPOST, late postemergence; RCA, Ridgetown Site 1; RCB, Ridgetown Site 2; WF, weed-free; fb, followed by; Env1 = RCA 2006, RCB 2006, RCA; nv2 = RCA 2007 and RCA 2008; NS, not significant.
There was no difference in corn yield between glyphosate applied EPOST and sequential glyphosate applications; however, when glyphosate was applied EPOST yield increased by 1.7 MT•ha −1 compared to glyphosate applied LPOST at Exeter in 2007.Greater control of common ragweed contributed to higher corn yield at this location with glyphosate applied EPOST.In contrast, glyphosate applied EPOST reduced yield by 1.1 MT•ha −1 compared to glyphosate applied LPOST at RCB in 2008.Again, we attributed this to a decrease in weed control.At all other locations, there was no difference in corn yield between EPOST and LPOST glyphosate.

Table 7 . Mean corn yield in response to several weed management strategies at Exeter and Ridgetown, ON from 2006 to 2008 and Harrow, ON from 2007 to 2008 a .
Data were pooled by environment (location and year) when the interaction between environment and treatment was non-significant.Means are presented on the back-transformed scale.Means followed by the same letter within a column are not significantly different according to Fisher's Protected LSD (P < 0.05); b a priori orthogonal contrasts.* = significant (P < 0.05); Abbreviations: PRE, preemergence; POST, postemergence; EPOST, early postemergence; LPOST, late postemergence; WF, weed-free; fb, followed by; RCA, Ridgetown Site 1; RCB, Ridgetown Site 2; NS, not significant. a

Table 8 . Environmental impact quotient (EIQ) and environmental impact (EI) of weed management strategies used at Exeter and Ridgetown, ON from 2006 to 2008 and Harrow, ON from 2007 to 2008.
Kovach et al. (1992)h a.i.obtained fromKovach et al. (1999), except for atrazine, metolachlor, and isoxaflutole which were calculated according to formula developed byKovach et al. (1992)using PRE vs POST application; b EI values for products with more than one a.i.were obtained by summing the relative proportion of each a.i.; Abbreviations: EPOST, early postemergence; LPOST, late postemergence; fb, followed by; EIQ, environmental impact quotient; EI, environmental impact.