Introduction
Cultural weed-management techniques can have a substantial impact on crop competitiveness and crop productivity. One important method of cultural weed management is optimal seeding rate. Soybean is generally a poor competitor with weeds but seeding rate can improve crop competitiveness due to both increased plant stand and more rapid canopy development (Blackshaw et al. Reference Blackshaw, O’Donovan, Harker and Li2002; Guillermo et al. Reference Guillermo, Pedersen and Hartzler2009).
The recommended seeding rate for soybean in Saskatchewan is approximately 493,000 to 630,000 seeds ha−1 (44 to 57 plants m−2) (Saskatchewan Pulse Growers 2017). Increasing soybean seeding rate has a significant effect on reducing weed populations by improving crop competitiveness (Guillermo et al. Reference Guillermo, Pedersen and Hartzler2009; McWhorter and Barrentine Reference McWhorter and Barrentine1975; Nice et al. Reference Nice, Buehring and Shaw2001; Norsworthy and Oliver Reference Norsworthy and Oliver2001). Several studies have also reported a yield increase in soybean with higher seeding rates, largely due to improved weed control, light interception, and rapid canopy development (Cox and Cherney Reference Cox and Cherney2011; Elmore Reference Elmore1998; Place et al. Reference Place, Reberg-Horton, Dunphy and Smith2009). However, high seeding rates also can pose agronomic challenges, because increased plant density increases intraspecific competition for resources, and increases potential for disease (Krupinsky et al. Reference Krupinsky, Bailey, McMullen, Gossen and Turkington2002; Pennypacker and Risius Reference Pennypacker and Risius1999)
Planting date can also have an impact on crop yield and competitiveness with weeds, although studies results have been inconsistent with regard to competitiveness. Soybean has a higher competitive ability with weeds when planted early (Klingman and Oliver Reference Klingman and Oliver1994), but others have reported soybean to be more competitive with weeds with delayed planting (Coulter and Nafziger Reference Coulter and Nafziger2007; Liebman et al. Reference Liebman, Mohler and Staver2001; Rushing and Oliver Reference Rushing and Oliver1998). The effects of planting date on soybean yield are also inconsistent. Delaying planting significantly reduces soybean yield (De Bruin and Pedersen Reference De Bruin and Pedersen2008; Hardman and Gunsolus Reference Hardman and Gunsolus1994; Robinson et al. Reference Robinson, Conley, Volenec and Santini2009), and other studies have reported yield increases with late soybean planting (Buhler and Gunsolus Reference Buhler and Gunsolus1996; Rushing and Oliver Reference Rushing and Oliver1998). These differences in crop competitiveness and soybean yield response to planting date are due to many factors, such as the species and emergence timing of weeds, time of weed removal, and environmental conditions. Nevertheless, integrating various seeding rates and planting dates can maximize productivity and significantly improve soybean competitiveness (De Bruin and Pedersen Reference De Bruin and Pedersen2008; Lee et al. Reference Lee, Egli and TeKrony2008).
A major weed of soybean in western Canada is volunteer canola, which is an early emerging species (Lawson et al. Reference Lawson, Van Acker and Friesen2006). Dicot weeds such as volunteer canola tend to cause greater yield loss in soybean compared with yield loss associated with monocot weed competition (Nave and Wax Reference Nave and Wax1971). Volunteer glyphosate-resistant (GR) canola poses a challenge to producers growing GR soybean, because of limited herbicide options for control; thus, integrated methods are needed. However, integrating methods may not always be successful. For example, mechanical weed control combined with banded herbicide application provided no difference in weed density or soybean crop yield compared with conventional herbicide control (Swanton et al. Reference Swanton, Shrestha, Clements, Booth and Chandler2002).
There is potential for integrated weed management in soybean crops to better manage volunteer GR canola, for which herbicide options are limited (Geddes and Gulden Reference Geddes and Gulden2018). Geddes and Gulden (Reference Geddes and Gulden2018) found a positive response of soybean to integrated weed management practices, although they did not look at the impact of planting rate and date combined. It is not known whether these two cultural control methods can affect competition between volunteer GR canola and GR soybean. The present study was conducted to evaluate the impact of GR soybean planting date and rate on GR volunteer canola.
Materials and Methods
Site Description
Field experiments were conducted in 2014 and 2015 at the Kernen Crop Research Farm (52.25°N, 106.88°W) in Saskatoon, Saskatchewan, Canada; at the Western Applied Research Corporation Research Field (52.35°N, 108.82°W) in Scott, Saskatchewan; at the Indian Head Agricultural Research Foundation Research Farm (50.52°N, 103.65°W) at Indian Head, Saskatchewan; and at the University of Manitoba Research Farm (49.50°N, 98.00°W) in Carman, Manitoba, Canada. Saskatoon and Scott are located on a dark brown soil; Indian Head and Carman are located on black soils. Soil descriptions are presented in Table 1.
Soil classification and descriptions for each site-year.
a Saskatchewan, Canada.
b Manitoba, Canada.
Experimental Procedure
The experimental design was a split plot with 15 treatments and four replications. Main plots were planting date (early, intermediate, and late) and subplots were seeding rates (targeted 10, 20, 40, 80, and 160 plants m−2 corresponding to 101,880; 203,775; 407,550; 815,100; and 1,630,200 seeds ha−1). Plots were seeded in late May, early June, and mid-June in 2014 and in mid-May, late-May, and early June in 2015. Planting dates in 2014 were later than targeted, because of environmental conditions, which delayed seeding. Actual planting dates are listed in Table 2. Main plots at Saskatoon and Scott measured 10 m wide by 6 m long, main plots at Indian Head were 13.5 m wide by 10.7 m long, and plots at Carman were 12.5 m wide by 8 m long. Each subplot at Saskatoon and Scott measured 2 m wide by 6 m long, subplots at Indian Head were 2.7 m wide by 10.7 m long, and subplots at Carman were 2.5 m wide by 8 m long. Border plots were seeded at all sites to minimize border effects. All plots received a 450 g ae ha−1 application of glyphosate immediately after seeding to control emerged weeds.
Planting dates and growing degree days for planting date treatments at Saskatoon, Scott, and Indian Head, Saskatchewan, Canada; and Carman, Manitoba, Canada, in 2014 and 2015.
a GDD, growing degree day (base temperature:10 C).
The soybean cultivar used was P001T34R (DuPont Pioneer, Mississauga, Ontario, Canada). It was pretreated with a copack of thiamethoxam plus fludioxonil plus metalaxyl (Cruiser Maxx® Beans; Syngenta Canada, Guelph, Ontario, Canada) and sedaxane (Vibrance® 500 FS) applied at rates of 195 mL plus 10 mL 100 kg−1 of seed, respectively (i.e., Cruiser Maxx Vibrance Beans; Syngenta Canada, Guelph, Ontario, Canada). Soybean seed was preinoculated with Bradyrhizobium japonicum (Optimize®; Syngenta Canada, Guelph, Ontario, Canada) inoculant, and granular Penicillium bilaii (TagTeam®; Syngenta Canada, Guelph, Ontario, Canada) was applied at a rate of 3 kg ha−1 at the time of seeding. Soybean was seeded 3-cm deep with a cone seeder equipped with disc openers spaced at 40 cm at the Saskatoon location; hoe openers were used at the other locations. A soybean survival rate of 75% (OMA 2009) was used to determine seeding rates; therefore, actual seeding rates were 16, 27, 53, 106, and 215 seeds m−2. Actual planting dates and cumulative growing degree-dates (base temperature 10 C) (Zhang et al. Reference Zhang, Wang and Hesketh2001) for each planting date are presented in Table 2.
Volunteer canola was seeded at a rate of 80 seeds m−2 using a 50% survival rate (Canola Council of Canada 2015) to establish a target plant density of 40 plants m−2. Canola (‘Dekalb 72-65 RR’; Bayer Crop Science Canada, Calgary, Canada) was cross-seeded with a plot drill across the entire trial immediately following each soybean planting date.
Volunteer canola biomass sampling was conducted at the canola podding stage. Aboveground shoot-biomass samples were collected in two 0.5-m2 quadrates per plot from the front and back of each plot. Samples were cut just above the ground surface, with the canola separated and placed in brown paper bags. All material was oven dried at 80 C for 72 h and weighed. Soybean crop height was measured just prior to biomass sampling by measuring the distance from the ground to the top of the plant on five to 10 plants per plot. Plots were harvested with a small plot combine and samples were dried, cleaned, and weighed to determine final seed yield. Soybean is considered dry at 14% moisture content; therefore, yields were adjusted to 14% moisture content. Volunteer canola seeds that were cleaned out of soybean samples were also weighed to determine volunteer canola seed contamination.
Statistical Analysis
Residuals were initially tested to ensure that the assumptions of ANOVA were met. The Shapiro-Wilk test in PROC UNIVARIATE (SAS Institute, Cary, NC) was used to assess normality and the Levene test was used to assess homogeneity of variance. Where there was heterogeneity between sites, the REPEATED statement was used to account for this heterogeneity. If model fit was improved by modeling heterogeneity, then this model was used. Where model fit was not improved, the original PROC MIXED model was used.
Data were analyzed with the PROC MIXED procedure in SAS, version 9.3. Rate, date, and rate*date treatments were considered fixed effects in the model, whereas random effects consisted of site-year, block, and site-year interactions with fixed effects. To assess the significance of random effects and their interactions with fixed effects, covariance parameters were examined using the COVTEST option of PROC MIXED in SAS, version 9.3, to determine if the site-years could be combined and if conclusions could be drawn from a broader (population-based) inference space (SAS Institute 2014).
Orthogonal polynomial contrasts were calculated to determine whether variables had a linear or quadratic response to seeding rate. Analysis of covariance was used to calculate linear or quadratic regression coefficients for seeding-rate responses (Yang and Juskiw Reference Yang and Juskiw2011). Nonlinear curves were fit using SigmaPlot 12® (Systat Software, Inc., San Jose, CA) Contrasts were used to determine if regression coefficients were significantly different between sites. Sites with similar regression coefficients were combined for analysis.
Economic Analysis
An economic analysis was conducted wherein the soybean market price was CAD$0.44 kg−1 (CAD$11.85 bushel−1), which is an average price based on the market price projection for 2016 of CAD$0.39 kg−1 (Saskatchewan Crop Insurance Corporation 2016), current market price of CAD $0.42 (Rayglen Commodities 2016) and average market price of CAD $0.49 from 2013, 2014, and 2015 (Agriculture and Agri-Food Canada 2015). Based on the recommended seeding rate of 40 plants m−2, the average soybean seed cost is CAD $233.17 ha−1 for seed and seed treatment (Government of Manitoba 2016). Gross income was calculated by multiplying soybean seed yield by market price. A contribution margin was calculated by subtracting the seed cost from the gross income.
Differences in soybean yield and volunteer canola-shoot biomass were determined by comparing each seeding rate with the standard rate of 40 plants m−2.
Soybean seed yield and dockage predictions were computed for all seeding rates using the quadratic equation, y = ax 2 + bx + c, where a is the quadratic coefficient, b is the linear coefficient, c is the y intercept, and x is the seeding rate. Prediction values used are from combined analysis of each variable. Coefficient values for parameters a, b, and c to predict seed yield were as follows: a = −0.034; b = 13.277; and c = 189.81. Coefficient values for parameters a, b, and c to predict dockage were as follows: a = 0.00213; b = −0.6335; and c = 73.67.
Volunteer canola-shoot biomass predictions for all seeding rates were calculated using the linear response formula y = ax 2 + b, where a is the linear coefficient and b is the y intercept. Coefficient values for parameters a and b to predict canola shoot biomass are as follows: a = −15.95 and b = 3,134.18. Predicted yields for seeding rates used in the experiment, as well as seeding rates that were not used in the experiment, are shown in Table 3.
Predicted soybean yields at various seeding rates from 10 to 160 plants m−2.
Results and Discussion
Soybean
The main effects of planting date and site were not significant when site-years were combined, but there was a site-year*date interaction for soybean plant height (Tables 4 and 5). Overall, soybean plant height exhibited a positive linear relationship with increasing seeding rate (P < 0.0001) (Table 4; Figure 1). For example, soybean plant height increased by 9.25% as seeding rate was increased from 10 plants m−2 to 160 plants m−2. When plant height data were combined across site-years, plant height tended to be greater at intermediate and late planting dates (56 and 56 cm, respectively) when compared with the early planting dates (48 cm), although the difference was not significant (P = 0.09). The planting-date effect was statistically significant at Indian Head in 2015 and at Saskatoon in 2014 and 2015, but not for the remaining four site-years, which likely accounted for the overall site-year*date interaction. At the three sites where planting date was significant, the early planting date was shorter than the intermediate and late planting date (Table 4), which is consistent with the overall trend from the combined analysis.
ANOVA results for the fixed effects (rate, date, and rate by date interaction) and random effects (site-year and site-year by treatment interactions) on soybean and canola variables.
a P values for random effects (site-year and site-year by treatment interactions) were assessed using the Wald Z test (COVTEST in SAS).
b Abbreviations: SC, seed contamination; TSW, thousand seed weight.
Mean soybean plant height at three planting dates at two sites in Saskatchewan, Canada, where height had a site-year*date interaction.
a Abbreviation: LSD0.05, least significant difference at the α level of 0.05.
b Means followed by different letters are significantly different within site-years at given LSD values.
Soybean plant height response to seeding rates of 10, 20, 40, 80 and 160 plants m−2. Data points represent the means of all site-years at each seeding rate. Bars indicate ±1 standard error of the mean. The linear equation is y = 0.0313x + 51.55. R 2 = 0.9622.
Soybean seed yield increased consistently across site-years with regard to seeding-rate effects (P < 0.0001), but due to significant differences between regression coefficients and site-year*date*rate interactions, data were not combined across all site-years (Table 4). When site-years were analyzed separately, five of the seven site-years had a date*rate interaction, and the remaining two (Carman and Scott in 2015) had a seeding-rate effect only. Of the five site-years that had a date*rate interaction, the early planting date had the highest soybean yield at Indian Head in 2014 and Saskatoon in 2014 (Figure 2), whereas the late planting date yield was highest at Carman in 2014, Indian Head in 2015, and Saskatoon in 2015 (Figure 2). The date*rate interaction at these five site-years may have been due to a difference in the magnitude of the response to seeding rate at different planting dates. For example, at Indian Head in 2014, a seeding rate increase from 40 plants m−2 to 80 plants m−2 produced a soybean seed yield increase of 43% at the intermediate planting date and 82% at the early planting date (Figure 2). Similarly, soybean seed yield at Carman in 2014 increased by 52%, 69%, and 39% at the early, intermediate, and late planting dates, respectively, when the seeding rate was increased from 40 plants m−2 to 80 plants m−2. Conversely, at Carman in 2014, seed yield increased by 55%, 25%, and 21% at the early, intermediate, and late planting dates when seeding rate was increased from 80 plants m−2 to 160 plants m−2 (Figure 2).
Effect of seeding rate and planting date on soybean yield at (a–e) site-years where there were site-year*date*rate interactions; and (f) Carman in 2015 and Scott in 2015, where there were no site*date*rate interactions. Bars indicate ±1 standard error of the mean. Line equations for planting dates: (a) Early: y = −0.0434x 2 + 16.794x + 76.618. R 2 = 0.9928. Intermediate: y = −0.001x 2 + 3.6416x + 35.209. R 2 = 0.9901. (b) Early: y = −0.0461x 2 + 23.054x + 541.87. R 2 = 0.9616. Intermediate: y = −0.0188x 2 + 14.358x + 237.1. R 2 = 0.9605. Late: y = −0.0322x 2 + 12.107x − 52.423. R 2 = 0.9944. (c) Early: y = −0.0182x 2 + 7.1369x + 0.8078. R 2 = 0.994. Intermediate: y = −0.0422x 2 + 12.421x + 6.5059. R 2 = 0.9999. Late: y = −0.0839x 2 + 23.785x + 253.42. R 2 = 0.9947. (d) Early: y = −0.032x 2 + 14.315x + 131.45. R 2 = 0.9864. Intermediate: y = −0.098x 2 + 28.647x + 203.9. R 2 = 0.9816. Late: y = −0.1155x 2 + 30.608x + 320.99. R 2 = 0.9872. (e) Early: y = −0.006x 2 + 5.6412x − 14.268. R 2 = 0.9984. Intermediate: y = −0.0064x 2 + 8.7189x + 34.508. R 2 = 0.9945. Late: y = −0.0303x 2 + 13.441x + 1.7394. R 2 = 0.9998. (f) Scott 2015: y = −0.0438x 2 + 15.612x + 45.428. R 2 = 0.9997. Carman 2015: y = 2.7054x + 596.86. R 2 = 0.8698.
The late and intermediate planting dates had highest soybean yields at Indian Head in 2015, with the early planting date yielding significantly less (Figure 2). The magnitude of the seed-yield increase at higher seeding rates also varied with planting date. For example, increasing seeding rate from 20 plants m−2 to 40 plants m−2 increased soybean seed yield by 30%, 60%, and 57% at the early, intermediate, and late planting dates, respectively. However, when seeding rates were increased from 40 plants m−2 to 80 plants m−2 and 80 plants m−2 to 160 plants m−2, the magnitude of seed-yield increase was highest at the early date in both cases (Figure 2). Results at Saskatoon in 2015 were similar, with the late planting date having the highest soybean yield and the early planting date having the lowest yield. The magnitude of the yield increase again varied with planting date at these sites.
Although planting-date treatments tended to show inconsistent effects that depended on conditions at each site-year, soybean seed yields tended to be greatest when planting occurred after May 20 and before June 11, indicating an optimal planting date range for soybean in western Canada. However, planting around June 11 is likely to be too late for the short growing season of western Canada, and soybean planted this late is unlikely to mature before the first fall frost. The long-term average date for the first frost in Saskatchewan is between September 9 and 15, and the first frost generally occurs from September 11 to 16 at Carman (Saskatchewan Crop Insurance Corporation 2017; Manitoba Agriculture, 2017). Volunteer canola would also be well established by June 11, which may present a large disadvantage to soybean if the volunteer canola is not well managed before planting. Therefore, we recommend an optimal planting date range from May 20 to June 1 in western Canada.
Soybean seed yield at the Carman and Scott locations in 2015 was combined across planting dates because date had no effect at these site-years and there was no rate*date interaction (Table 4). At both site-years, soybean seed yield increased consistently with increasing seeding rates (Figure 2). At Carman in 2015, seed yield had a linear relationship with seeding rate, and the overall range of yield was much lower compared with the Scott site. Yield ranged from 670 to 1,040 kg ha−1 as density increased from 10 plants m−2 to 160 plants m−2, whereas seed yield ranged from 190 to 1,420 kg ha−1 at Scott (Figure 2).
Soybean seed yield consistently increased with increasing seeding rates in this study, but the incremental response tended to decrease with increasing seeding rates. This is likely due to the law of constant final yield, where total standing plant biomass initially increases in proportion to density, levels off, and then remains constant as density increases more (Weiner and Freckleton Reference Weiner and Freckleton2010). However, in most cases, maximum yield was not achieved at the densities tested in this experiment, and we did not achieve constant final yield. Plant densities required to achieve maximum yield, calculated by differentiating the quadratic formula, ranged from 142 to 250 plants m−2. Several studies have also reported that increasing soybean seeding rate increased yield (Cox and Cherney Reference Cox and Cherney2011; Elmore Reference Elmore1998; Place et al. Reference Place, Reberg-Horton, Dunphy and Smith2009). Studies have also found that increasing seeding rate in other pulse crops, such as field pea (Pisum sativum L.) and lentil (Lens culinaris Medik.), results in higher yield and lower weed biomass in organic production systems (Baird et al. Reference Baird, Walley and Shirtliffe2009a, Reference Baird, Walley and Shirtliffe2009b).
Volunteer Canola
Seeding rate had a significant effect on volunteer canola-shoot biomass (P < 0.0001) and there was also a significant site-year*date interaction for canola biomass (Table 4). However, there was no planting-date effect when site-years were combined. It is likely that there was a site-year*date interaction because two of the seven site-years (Indian Head in 2014 and Saskatoon in 2014) exhibited a planting-date effect, whereas the remaining five had no statistically significant date effect (Table 4). However, trends were consistent across all site-years and so site-years and planting dates were combined.
A significant site-year*rate interaction also existed for volunteer canola-shoot biomass and seed contamination (Table 4). At site-years when canola biomass was less than 1,500 kg ha−1 (Carman and Indian Head in 2014, and Carman in 2015), increasing seeding rate had a smaller effect on canola-shoot biomass (Figure 3). This led to a site-year*rate interaction, because the response to seeding rate differed within site-years as a function of the amount of volunteer canola present at each site. In contrast, the remaining four site-years with volunteer canola biomass greater than 1,500 kg ha−1 had much steeper declines in volunteer canola-shoot biomass as seeding rates increased (Figure 3).
Volunteer canola-shoot biomass response to seeding rate across all seven site-years. Bars indicate ±1 standard error of the mean. Line equations: Carman 2014: y = −0.0024x 2 − 3.8977x + 1656.8. R 2 = 0.6356. Carman 2015: y = 0.005x 2 − 8.6171x + 1689.3. R 2 = 0.9985. Indian Head 2014: y = 0.0293x 2 − 8.4452x + 1392.7. R 2 = 0.8503. Indian Head 2015: y = 0.1125x 2 − 35.422x + 3319.2. R 2 = 0.9783. Saskatoon 2014: y = 0.0789x 2 − 26.284x + 4727.2. R 2 = 0.7678. Saskatoon 2015: y = −0.0276x 2 − 9.698x + 4720.6. R 2 = 0.952. Scott 2015: y = 0.0495x 2 − 16.785x + 3852.8. R 2 = 0.8231.
A consistent trend was observed in all site-years wherein volunteer canola-shoot biomass (P < 0.0001) and seed contamination (P = 0.0012) tended to decrease linearly with increasing seeding rate (Table 4; Figures 3 and 4). Volunteer canola biomass and seed contamination were highest at lower seeding rates due to poor crop competition. For example, when seeding rate was increased from 40 plants m−2 to 80 plants m−2, the decrease in volunteer canola biomass ranged from 17% to 45% across site-years, with the exception of the Scott location in 2015, which increased only 6% (Figure 2). Similarly, volunteer canola seed contamination declined between 9% and 34% across all site-years when seeding rate was increased from 40 plants m−2 to 80 plants m−2 (Figure 4). When seeding rate was increased from 80 plants m−2 to 160 plants m−2, the decrease in volunteer canola biomass ranged from 6% to 62% (Figure 3), whereas the decrease in canola seed contamination ranged from 21% to 56% across all site-years (Figure 4).
Volunteer canola contamination response to seeding rate at five all site-years. Bars indicate ±1 standard error of the mean. Line equations: Carman 2014: y = 0.0124x 2 − 3.6442x + 431.67. R 2 = 0.9903. Carman 2015: y = 0.0007x 2 − 0.4482x + 73.264. R 2 = 0.9253. Saskatoon 2014: y = 0.0253x 2 − 9.2581x + 1525.3. R 2 = 0.9429. Saskatoon 2015: y = 0.0145x 2 − 7.3056x + 1730.3. R 2 = 0.9857. Scott 2015: y = 0.0111x 2 − 3.7214x + 710.71. R 2 = 0.9756.
Seeding rate effects observed in this study generally were consistent across both the soybean crop and volunteer canola. Increasing the seeding rate resulted in greater soybean height and seed yield, and also positively influenced soybean’s competitive ability against volunteer canola. Lower weed densities and biomass have been reported in several studies when soybean was seeded at high densities (Guillermo et al. Reference Guillermo, Pedersen and Hartzler2009; McWhorter and Barrentine Reference McWhorter and Barrentine1975; Nice et al. Reference Nice, Buehring and Shaw2001; Norsworthy and Oliver Reference Norsworthy and Oliver2001). Crops seeded at a higher population density tend to have a competitive advantage over weeds, due to rapid canopy development and, therefore, improved competitiveness, as well as increased plant stand (Guillermo et al. Reference Guillermo, Pedersen and Hartzler2009; Place et al. Reference Place, Reberg-Horton, Dunphy and Smith2009).
Economic Analysis
Maximum soybean yield was not reached at any seeding rate in most site-years; therefore, an economic analysis was conducted to determine the optimal seeding rate for growers and the economic benefit of different seeding rates. The highest contribution margin (i.e., net income) was observed at a seeding rate of 10 plants m−2 at $90.37 ha−1, but it became negative at seeding rates greater than 80 plants m−2, because of high seed costs (Table 6). Net income consistently decreased with increasing seeding rates, because yield increases were not great enough to offset the increased seed cost. As seeding rate increased, the decline in contribution margin became larger. For example, the difference in contribution margin between 10 and 20 plants m−2 was only −$1.22, whereas the difference between 40 and 50 plants m−2 was −$10.75.
Soybean and volunteer canola production and economic factors for all seeding rates.
a Soybean seed yield and dockage predictions were calculated using the quadratic equation: y = ax 2 + bx + c.
b Canola-shoot biomass predictions were calculated using the linear response formula: y = ax 2 + b
c Difference in soybean yield and canola-shoot biomass was determined by comparing each seeding rate with the standard rate of 40 plants m−2.
d Gross income was calculated by multiplying soybean seed yield by market price.
e Contribution margin was calculated by subtracting the seed cost from the gross income.
Although maximum soybean yield was not reached at most of the site-years in this study, it is very likely that seeding rates higher than those included in this study will not be economically feasible for growers. Based on the economic analysis, the grower’s contribution margin will be negative when seeding rates exceed 80 plants m−2, due to high seed costs (Table 6). Surprisingly, the highest net income was reached at 10 plants m−2, which is one-fourth of the current recommended seeding rate. However, this seeding rate produced very high volunteer canola-shoot biomass and contamination. If volunteer canola is not well controlled at such a low seeding density, it will set seed and replenish the seed bank for the following years—volunteer canola can produce up to 3,600 seeds m−2 (Gulden et al. Reference Gulden, Shirtliffe and Thomas2003) and can persist in soil for several years. This adds an additional cost, because the volunteer canola will require control in the following years and continue to compete with sequential crops for several years.
Given the aforementioned information, it appears that during years with low or average market prices for soybean, 40 plants m−2 is likely the best option for growers, because seeding rates below 40 plants m−2 have very high levels of volunteer canola contamination, but the contribution margin continues to decrease at seeding rates higher than 40 plants m−2. However, when market prices are high, growers will see a benefit to increasing the seeding rate to 50 to 60 plants m−2, because this will potentially increase soybean yield by 15% to 30% and offset seed costs while minimizing volunteer canola competition. Seeding rates higher than 70 plants m−2 are generally not economic for growers because the yield benefits are not great enough to offset the high seed costs.
Taken together, our results demonstrate that cultural weed control methods, such as altering soybean seeding rate and planting date, can substantially affect soybean yield and reduce competition from volunteer canola. However, the effects of planting date on soybean development were inconsistent and differed among site-years. The early planting date had the greatest soybean emergence and seed yield at Indian Head and Saskatoon in 2014. These two site-years were also the only two in which planting date affected volunteer canola biomass, with the early planting date having the lowest canola-shoot biomass. This indicates that early planted seeds were able to effectively compete with volunteer canola and the finding is similar to that of Klingman and Oliver (Reference Klingman and Oliver1994), who reported that soybean yield loss due to weed interference increased as planting date was delayed from early May to early June. In our experiment, plants from the early planting date matured earlier and may have reduced yield losses due to frost damage. Several studies have reported higher soybean yields with early planting (De Bruin and Pedersen Reference De Bruin and Pedersen2008; Hardman and Gunsolus Reference Hardman and Gunsolus1994; Kane et al. Reference Kane, Steele and Grabau1997; Parvez et al. Reference Parvez, Gardner and Boote1989; Robinson et al. Reference Robinson, Conley, Volenec and Santini2009). However, some of these studies used planting dates that are either too early or too late for western Canada. The latest planting date produced the greatest yield at Carman in 2014, Indian Head in 2015, and Saskatoon in 2015, and soybean emergence was greatest at many of these sites as well. This contrasts with results at the two Saskatchewan sites in 2014, but other studies have also reported that yields of later-seeded soybean can be higher than early-seeded soybean (Buhler and Gunsolus Reference Buhler and Gunsolus1996; Rushing and Oliver Reference Rushing and Oliver1998). The authors suggested this was often due to inadequate rainfall early in the season and decreased weed competition with later planting.
In summary, planting-date effects were variable across site-years, whereas seeding rate effects were fairly consistent. Earlier seeding tended to improve the crop’s competitiveness with volunteer canola, whereas late seeding may reduce the ability of the soybean crop to compete with early emerging volunteer canola. Based on planting date results, the optimal planting date range for soybean in western Canada is May 20 to June 1. Higher seeding rates resulted in greater soybean yield and lower volunteer canola biomass and seed contamination. Based on the economic analysis, the optimal seeding rate is 40 plants m−2 in years with low or average market prices. When market prices are high, increasing the seeding rate to 50 to 60 plants m−2 will increase soybean yield significantly, decrease volunteer canola competition and dockage, and increase net income for the grower. Improving crop competition with higher seeding rates will also decrease the contribution of canola seed to the seedbank and, therefore, decrease volunteer canola populations in sequential crops.
Acknowledgements
The authors thank the Western Grains Research Foundation, the Saskatchewan Ministry of Agriculture’s Agriculture Development Fund, Pioneer Hi-Bred, and Monsanto for graciously funding the research. We also wish to thank the University of Saskatchewan for in-kind support including facilities and technical support staff. This work would not have been possible with the collective efforts of all technical support staff at the various locations. No conflicts of interest have been declared.
