Weed Management in White Bean With Pre-plant Incorporated Herbicides

Five field experiments were conducted in Ontario Canada during 2018-2020 to determine the level of crop injury, weed control and white bean yield with up to four-way mixtures of herbicides applied preplant incorporated (PPI). The trials were arranged in a factorial design: Factor 1 was “Grass herbicide” including no grass herbicide, trifluralin, S-metolachlor and trifluralin + S-metolachlor and Factor 2 was “Broadleaf herbicide” including no broadleaf herbicide, halosulfuron, imazethapyr and halosulfuron + imazethapyr. At 2 and 4 weeks after emergence (WAE), there was minimal (≤ 4%) white bean injury. At 8 weeks after herbicide application (WAA), trifluralin, S-metolachlor or trifluralin + S-metolachlor averaged across Factor 2 controlled velvetleaf 69, 71 and 62%, respectively; halosulfuron, imazethapyr and halosulfuron + imazethapyr averaged across Factor 1 controlled velvetleaf 75, 95 and 97%, respectively. At 8 WAA, trifluralin, S-metolachlor and trifluralin + S-metolachlor controlled pigweed 93, 90 and 97%, respectively, and halosulfuron, imazethapyr and halosulfuron + imazethapyr controlled pigweed 97, 79 and 98%, respectively. At 8 WAA, trifluralin, S-metolachlor and trifluralin + S-metolachlor provided poor (≤ 32%) control of common ragweed while halosulfuron, imazethapyr and halosulfuron + imazethapyr controlled common ragweed 86, 53 and 87%, respectively. The 4-way tankmix of trifluralin, S-metolachlor, halosulfuron + imazethapyr controlled common ragweed 95%. At 8 WAA, trifluralin, S-metolachlor and trifluralin + S-metolachlor controlled common lambsquarters 81, 38 and 91%, respectively, and halosulfuron, imazethapyr and halosulfuron + imazethapyr controlled common lambsquarters 94, 97 and 99%, respectively. At 8 WAA, trifluralin, S-metolachlor and trifluralin + S-metolachlor provided poor (≤ 46%) control of wild mustard while halosulfuron, imazethapyr and halosulfuron + imazethapyr provided excellent (≥ 97%) wild mustard control. At 8 WAA, trifluralin, S-metolachlor and trifluralin + S-metolachlor controlled barnyardgrass 70, 85 and 94%, respectively, and halosulfuron, imazethapyr and halosulfuron + imazethapyr controlled barnyardgrass 9, 50 and 59%, respectively. At 8 WAA, trifluralin, S-metolachlor and trifluralin + S-metolachlor controlled green foxtail 89 to 98% and halosulfuron, imazethapyr and halosulfuron + imazethapyr controlled green foxtail 19, 69 and 67%, respectively. Weed interference reduced white bean yield 76%. Generally, white bean yield reflected the level of weed control. Based on these results, the 2and 3-way tankmixes of herbicides evaluated generally provide similar weed control as the 4-way tankmixes.


Introduction
Dry bean (Phaseolus vulgaris L.) is a legume crop grown and consumed in many parts of the world (OBG, 2021). Canada is an important dry bean producer in the world. White (navy) bean has been grown in Ontario since the early 1900's and is the most popular dry bean market class grown in the province (OBG, 2021). Most of the white bean produced in Ontario is exported to the United Kingdom for baked beans and the rest is consumed domestically (OBG, 2021). Farmers in Ontario produce approximately 56,000 tonnes of white bean grown on 27,000 hectares with a farm gate value of nearly $50 million (OMAFRA, 2021). Weeds can reduce white bean yield substantially if not controlled. Manuscripts published by the Yield Loss Committee of the Weed Science Society of America (WSSA) estimated yield loss of 71% in dry bean due to weed interference which was substantially greater than other field crops such as corn (50%), soybean (52%) and winter wheat (23%) (Flessner, 2021;Soltani et al., 2016Soltani et al., , 2017Soltani et al., , 2018. Despite a drastic potential yield loss due to weed interference, the number of herbicides available to white bean growers is much fewer than corn, soybean and winter wheat (OMAFRA, 2020). Soil-applied herbicides commonly used by white bean producers include trifluralin, S-metolachlor, halosulfuron and imazethapyr (OMAFRA, 2020).
Growers often need to tankmix grass herbicides such as trifluralin and S-metolachlor with broadleaf herbicides such as halosulfuron and imazethapyr for broad-spectrum weed control in white bean production. Earlier studies with trifluralin, S-metolachlor, halosulfuron, and imazethapyr have mostly focused on two-way tankmixes of these herbicides for broad-spectrum weed control in white bean (Li et al., 2016(Li et al., , 2017Soltani et al., 2010;Soltani et al., 2012aSoltani et al., , 2012bSoltani et al., 2014a). White bean growers have seen inconsistent control of some problematic weed species such as common ragweed with two-way tankmixes in Ontario. Three-or four-way tank-mixtures of these herbicides may improve the efficacy and consistency of weed control in white bean production. To our knowledge, no previous study has cumulatively compared the crop safety and weed control efficacy of a four-way tankmix of trifluralin, S-metolachlor, halosulfuron, and imazethapyr, applied PPI in white bean under Ontario environmental conditions. More research is needed to evaluate crop safety and consistency of weed control with various tankmix combinations of these herbicides to improve weed control efficacy, increase seed yield, and elevate net returns to white bean growers in Ontario.
The objectives of this study were to determine the level of crop injury, weed control and white bean yield with trifluralin (600 g ai ha -1 ), S-metolachlor (1050 g ai ha -1 ), halosulfuron (26.25 g ai ha -1 ), and imazethapyr (37.5 g ai ha -1 ), applied preplant incorporated (PPI) alone and in two-, three-and four-way tank-mixtures.

Materials and Methods
A total of five field experiments were completed over at three-year period with three at the Huron Research Station (one in 2018, 2019 and 2020) near Exeter, Ontario (43°19′1.21″N, 81°30′3.87″E) and two at the University of Guelph Ridgetown Campus (one in 2019 and 2020) near Ridgetown, Ontario (42°26′26″N, 81°53′3″W). The soil at Exeter was a Brookston clay loam (Orthic Humic Gleysol, mixed, mesic, and poorly drained) and the soil at the Ridgetown location was a Watford/Brady sandy loam. Seedbed preparation at all sites consisted of fall moldboard plowing followed by seedbed preparation in the spring with a field cultivator with rolling basket harrows.
The experimental design was a two-way factorial, established in the field as a randomized complete block design (RCBD) with 4 replicates. Factor 1 was "Grass herbicide" and Factor 2 was "Broadleaf herbicide". Treatments are listed in Table 1. Each plot was 3.0 m wide and 10 m long at Exeter and 8 m long at Ridgetown and consisted of four rows of 'T9905' white bean spaced 0.75 m apart. White bean was planted at a rate of approximately 240,000 seeds ha -1 in late May to early June of each year. Means followed by a different letter within a column (a-b) or row (X-Z) within each section are significantly different according to a Tukey-Kramer multiple range test at P < 0.05. 1 Means for white bean injury are based on data from Exeter in 2018 to 2020; Ridgetown trials showed no visible injury and were excluded from analysis due to zero variance. 2 No yield data for Ridgetown in 2020.
Herbicide treatments were applied using a CO 2 -pressurized backpack sprayer calibrated to deliver 200 L ha -1 at 240 kPa. The boom was 1.5 m long with four ultra-low drift nozzles (ULD120-02, Hypro, New Brighton, MN) spaced 50 cm apart. The surface area sprayed was the center 2.0 m of each plot. There was a 1.0 m unsprayed area between adjacent plots. Preplant incorporated herbicides were applied 1-2 days before planting and were immediately incorporated into the soil with two passes (in opposite directions) of an S-tine cultivator with rolling basket harrows.
White bean injury [2 and 4 weeks after crop emergence (WAE)] and weed control [4 and 8 weeks after herbicide application (WAA)] were visually estimated on a scale of 0% (no injury/control) to 100% (complete plant death). Weed density and dry weight were evaluated 8 WAA by counting and cutting plants at the soil surface in two 0.5 m 2 quadrats per plot and separating by species. Each weed species was dried at 60 °C to a constant moisture and then weighed. White bean was combined at harvest maturity using a small plot combine; seed moisture content and weight were recorded. Seed moisture content was adjusted to 18% prior to analysis.
Data analysis was carried out using Proc Glimmix in SAS (Ver. 9.4, SAS Institute Inc., Cary, NC), with grass herbicide, broadleaf herbicide and their interaction as the fixed effects, and year-location combinations (environment), replicate within environment and environment by grass herbicide by broadleaf herbicide interaction as the random effects. Evaluation of potential distributions for each parameter was accomplished by using studentized residual plots to control for departures from the assumption of homogeneous variance, the Shapiro-Wilk statistic and normal probability plot to confirm the assumption of normality, the Chi-square/df ratio to check for overdispersion, and information criteria such as AICC to compare fit between models where possible. White bean injury and yield were analyzed using the normal distribution, percent visible weed control was arcsine square-root transformed prior to analysis with the normal distribution, and weed density, weed biomass and white bean moisture at harvest were analyzed using the lognormal distribution. All least square mean pairwise comparisons were adjusted using the Tukey-Kramer method and the significance level was set at p < 0.05. Main effect least-square means were separated only if the grass herbicide by broadleaf herbicide interaction was negligible; when this interaction was non-negligible, least-square means comparisons for simple effects are presented. All comparisons were conducted on the model scale. However, means for data analyzed using a non-normal distribution or requiring transformation for analysis were back-transformed for presentation of results.

White Bean Injury and Yield
There was minimal visible white bean injury (≤ 4.1%). at 2 and 4 WAE ( Table 1). At 2 WAE, trifluralin, S-metolachlor or trifluralin + S-metolachlor averaged across Factor 2 caused 2% to 3% white bean injury which decreased slightly at 4 WAE. Halosulfuron, imazethapyr and halosulfuron + imazethapyr averaged across Factor 1 caused 2 to 4% and 1 to 4% white bean injury at 2 and 4 WAE, respectively. There was no difference in white bean seed moisture content within the "Grass herbicide" or "Broadleaf herbicide". There was trend to slightly lower seed moisture content with the use of a herbicide although differences were not always statistically significant, which indicates that the presence of weeds results a slight delay in white bean maturity. Weed interference reduced white bean yield as much as 76% in this study (Table 1). Reduced weed interference with trifluralin, S-metolachlor, halosulfuron, or imazethapyr applied alone or in a two-, three-or four-way tank-mixtures resulted in increased white bean yield. Generally, white bean yield reflected the level of weed control. Results are similar to other studies in which weed interference reduced white bean seed yield by 70% (Soltani et al., 2020). In the same study two-way tankmixes of trifluralin + halosulfuron and S-metolachlor + halosulfuron, applied PPI resulted in white bean yields that were up to 95% of the weed-free control (Soltani et al., 2020).
Means followed by a different letter within a column (a-b) or row (X-Z) within each section are significantly different according to a Tukey-Kramer multiple range test at P < 0.05. Rows without an uppercase letter have no difference between the check and broadleaf treatment.
Means followed by a different letter within a column (a-b) or row (X-Z) within each section are significantly different according to a Tukey-Kramer multiple range test at P < 0.05. Rows without an uppercase letter have no difference between the check and broadleaf treatment.
Means followed by a different letter within a column (a-c) or row (X-Z) within each section are significantly different according to a Tukey-Kramer multiple range test at P < 0.05. Rows without an uppercase letter have no difference between the check and broadleaf treatment.