Organic soybean producers must rely on a variety of tactics for weed management. The use of soybean varieties with enhanced ability to compete with weeds may be one such tactic to increase weed control. Experimental design utilized for this project was a split plot design with soybean trait of interest as the main plot, genotype as the subplot and weedy and weed free strips across the block for side by side comparison of treatments in weedy and weed free conditions. Differences in weed biomass were detected between genotypes. Larger seeded genotypes resulted in increased percent ground cover estimated at three weeks after emergence but seed size had little influence on percent ground cover estimates at five weeks after emergence. Optimum models from multiple regression showed seed size to be the most significant trait measured in overall genotype competitive ability.
Currently over 90% of soybeans grown in the United States utilize glyphosate resistance technology (Cerdeira and Duke 2006). An even greater percentage of soybeans are cultivated with the use of herbicides. But the acreage of organic soybeans is increasing as the organic milk, beef and egg markets grow each year (Dimitri 2008). The profit margin for organic soybeans can be substantial (Archer et al., 2007) but farmers making the transition to organic soybean production cite weed management as their top challenge (Archer and Kludze 2006; Cavigelli et al. 2008; Hamilton et al. 2007; Walz 1999). Without the use of herbicides as a management option, organic soybean producers must rely on a variety of tactics to reduce weed pressures (Liebman et al., 1997). The utilization of more competitive soybean cultivars may be another supplemental weed management tactic. A highly competitive cultivar would not only be useful for organic producers but conventional producers could also benefit if less herbicide applications are needed in season (Norsworthy & Shipe, 2006).
Genotypic differences in competitiveness for weeds have been identified for several agricultural species including: wheat (Triticum aestivum L.) (Ramsel & Wicks 1988), rice (Oryza sativa L.) (Haefele et al., 2004), cowpea (Vigna unguiculata L.) (Remison 1978), corn (Zea mays L) (Wooley & Smith 1986) and many others (Callaway 1992). Similar studies have suggested differences in competitiveness of soybean genotypes (Jannink et al., 2000; Rose et al., 1984) but high variation often overshadows possible differences (Norsworthy & Shipe, 2006; Bussan et al., 1997).
Soybean breeding programs have typically focused on improving characteristics such as yield and disease resistance with little or no attention to weed competitive improvement since the majority of soybean breeding trials are conducted in weed free conditions (Baenziger et al. 2006; Egli 2008; Gepts and Hancock 2006; Heisey et al. 2001). However, genetic variation in soybean competitive ability has been described by other research groups. Reports of traits that may be related to competitive ability have included height (Jannink et al. 2000), leaf area (Jordan 1993), and early vigor (Guneyli et al., 1969; Rose et al., 1984). However, identifying characteristics imparting competitive advantage has been difficult (Norsworthy & Shipe, 2006). Root characteristics may also influence soybean competitiveness (Dunbabin 2007; Place et al. 2008), but screening for and selection on canopy characteristics that improve competitive ability will be most feasible for soybean breeders.
Because traits of interest for increased competitiveness such as canopy cover or height may be variable depending on the growth stage, such traits should be investigated during the most critical period for weed competition. This period is defined as the interval in the life cycle of the crop when it must be kept weed free to prevent yield loss (Zimdahl 1980; Van Acker et al., 1993). This period is variable depending on environmental conditions (Van Acker et al., 1993) but has been estimated between the soybean stages V2 and V8 (Eyherabide et al., 2002) which occur at approximately 2 and 7 weeks after emergence (WAE), respectively.
We investigated the weed competitiveness of 27 soybean genotypes that were selected based on differing characteristics of seed size, petiole length, petiolule length, leaflet width and length, and main stem height. Our main objectives were to (1) determine if differences in weed competitive ability exist between cultivars of varying canopy traits and seed sizes and (2) determine the relation of these traits on the competitive ability of a genotype.
Experimental design consisted of a split plot design. The main plots consisted of 3 soybean trait groupings: 14 genotypes with varying seedsize, 6 genotypes with varying soybean leaf morphologies, and 7 genotypes with varying petiole and petiolule lengths. Subplots consisted of soybean genotypes, each planted six rows, 4 m in length and spaced at 96.5 cm. Half of each subplot (3 rows) was maintained weed free. Canopy traits were measured in the weed free area and weed biomass was taken from the weedy half. Which half was weed free was randomly assigned for each block. One location was planted with 9 replications.
Prior to soybean planting, three random subsamples of 100 seed for each genotype were collected, weighed and tested for germination using germination chambers set at 30˚ C. Soybean seeding rates were adjusted for each genotype to achieve 39 live seed row m-1 based on germination tests. A custom designed plot cone planter was used to plant soybeans at the Kinston Research Station in Kinston, NC on May 21, 2009 on a Pocalla Loamy Sand.
Immediately following soybean planting, the weedy subplots were overseeded with redroot pigweed (Amaranthus retroflexus L.) seed to increase weed pressure uniformity. Weed free plots were treated with alachlor at 5.84 L ai ha-1 immediately following soybean planting. Weed free plots were maintained weed free with weekly hand weeding. Between row areas for both weedy and weed free plots were maintained weed free with between row cultivation at 4 and 7 weeks after emergence (WAE), leaving only the weeds within 10 cm of the crop row in the weedy plots. At 1 and 4 WAE, sethoxydim with crop oil adjuvant was sprayed over the entire trial at 1.75 L ai ha-1 to limit weed presence to broadleaf weeds for more uniform weed pressures.
Soybean measurements were taken in the weed free plots over the 7 week period following soybean emergence to quantify canopy characteristics. Measurements were taken in all 9 replications. Stand counts were taken at 1 WAE (early) and 2 WAE (late). Soybean height was measured at 3 WAE (early) and 7 WAE (late). Overhead photographs taken at 3 WAE (early) and 5 WAE (late) were used to estimate soybean canopy percent ground cover in weed free plots. Images were processed utilizing SigmaScan Pro with a macro language software for batch analysis as described by Karcher and Richardson (2005) with the hue settings from 47 to 107 and the saturation setting from 10-100. Photos were taken with a Canon PowerShot A360 Digital Camera (8.0 mega pixels) using a custom built camera stand which was centered over the middle row of each weed free plot. Plots were shaded during photography to avoid shadow effects in image processing. The digital image size was 1.37 m wide and 1.87 m of the row, capturing approximately 70 plants. Leaf petiole, leaf petiolule, leaflet length and leaflet width were measured on the 3rd most fully expanded leaf at 4 WAE. At 7 WAE node number was measured.
At 7 WAE, an area 0.36 m wide and 3.05 m long of the center row of each weedy and weed free genotype subplot was harvested using a Haldrup forage plot harvester. Weeds and soybean plants were separated by hand in weedy plots. Fresh weight biomass was measured for the entire harvest area. For each genotype subplot, 3 biomass measures were recorded: the soybean biomass maintained weed free, the soybean biomass in weedy conditions, and the weed biomass. Every individual biomass measure was sub-sampled, dried, and weighed to estimate biomass moisture percentage. Total plot dry biomass was calculated for each individual biomass measure for analysis.
Statistical analysis was conducted using SAS 9.1. Year effects were treated as random. Year*treatment effects were significant for many treatments, thus results were reported by separate years. Model predicted values versus residual error graphs were utilized to confirm assumptions of error variance. Weed weight was square root transformed for analysis. The ratio of weedy to weed free soybean biomass was natural log transformed. All other dependent variables met model assumptions. Least squares mean values were utilized for reporting of treatment means. Multiple regression was utilized to determine which soybean characteristics were most influential in end of season weed biomass and percent ground cover estimation.
Multiple regression optimal model selection was determined using model information criteria methods. It was shown that in simulated model testing, optimal model selection using information criteria methods selected the true models significantly more often than heuristic methods or model diagnostic methods (Beal 2005). The model chosen for predicting soybean competitive ability was the model with the lowest value for the Schwarz Bayesian Criteria (SBC), Bayesian Information Criteria (BIC), and Akaike’s Information Criteria (AIC) and dependent on all parameters being significant at p
Competitive cultivars have sometimes been classified in two main ways, weed tolerant and weed suppressive (Callaway 1992; Jannink et al. 2000). Weed tolerant cultivars maintain yield or show minimal yield reductions when growing with weeds. In contrast, weed suppressive cultivars reduce weed biomass. Weed suppressive ability is the preferred component in crop competitiveness (Jordan 1993) for reducing weed seed rain and future weed infestations, although the tolerance to weeds may be important for acceptable yields (Norsworthy & Shipe, 2006).
In this experiment, less weed biomass associated with a soybean genotype was interpreted as a greater weed suppressive ability. Weed suppressive differences were detected in the grouping of genotypes with variable seed size. Differences in weed suppressive ability were also detected in the varying petiole and petiolule grouping of genotypes. The natto (small seeded and narrow leaflet) soybean genotypes N7103, N94-7440, N96-6429, and TCAXBXX-717 were the least effective genotypes in suppressing weeds in the variable seed size group.
Weed free soybean biomass differences between genotypes were detected in the same three genotype groupings that showed weed suppressive differences. Weedy soybean biomass differences between genotypes were detected in the seed size grouping of genotypes as well as the petiole/petiolule group. No weedy soybean biomass differences were detected.
Because each genotype subplot was split by a weedy and weed free treatment, the soybean biomass reduced by weedy conditions for each genotype was calculated. A smaller soybean biomass reduction due to the presence of weeds was interpreted as a greater ability to tolerate the presence of weeds. No significant differences were detected, suggesting no obvious differences between genotypes in ability to tolerate the presence of weeds.
The ability of a soybean genotype to effectively intercept light and increase canopy cover during the critical period for weed competition is crucial for suppressing weed growth (Peters et al. 1965; Yelverton and Coble 1991). Measurement of canopy coverage is often made to estimate leaf area index (LAI), a critical canopy characteristic for light competition (Gibson et al. 2003). The digital imaging technique used in this experiment was previously found by Stewart et al. to have a strong relationship (r² = 0.74) with measured leaf area index (LAI). Larger percent ground cover estimates from digital images were interpreted as greater ability to compete for light. Early images taken at 3 WAE showed light competitive differences between all genotype groupings. Late images taken at 5 WAE showed genotypic differences in 2 of the 3 groupings.
All of the canopy traits measured were included in multiple regression models to determine which of these characteristics most influenced competitive ability. Separate multiple regression models were developed with the dependent variables: early estimated percent ground cover, late estimated percent ground cover, and weed biomass. By utilizing models with dependent variables specific to light competition (percent ground cover estimates) and dependent variables for overall competition (weed biomass) we can discern the importance of these measured canopy traits for competition for light and overall competition with weeds.
Seed size, early soybean stand density, and leaflet width were the most influential traits in competitive ability for light. Seed size was the most influential trait for early light competition, suggesting that larger seeded genotypes resulted in improved ability to compete for light at 3 WAE. Two weeks later, the influence of seed size on percent ground cover had subsided dramatically. The reduction of seed size influence on soybean canopy traits over time has been previously reported (Oexemann 1942). Later competition for light was improved by longer leaves and wider leaflets. Wells et al. (1993) reported that narrower leaflets were less effective at PAR interception. However, Suh et al. (2000) points out that soybean genotypes with narrow leaves have better light distribution through their canopy and a higher photosynthetic rate than those genotypes with oval leaf shape. The narrow leaflet trait may demonstrate a trade-off advantage of improved light distribution through the canopy in weed free conditions but a disadvantage in light competition in weedy conditions.
The optimal models for how genotype canopy characteristics affected weed suppressive ability were somewhat unexpected. Seed size was the most influential trait with larger seeded genotypes showing an improved ability to suppress weeds. Surprisingly, models showed that a shorter leaflet length resulted in more weed suppressive ability. When it was seen that the very competitive genotype N00-7153 with a large seed and short leaflet may have been disproportionately influencing the model, that genotype was removed and the multiple regression was re-calculated. Even with the absence of that influential genotype, the relationship of a larger seed and a shorter leaflet improving overall competitive ability was maintained. Why a shorter leaflet or petiole would improve competitive ability is not clear, particularly since longer leaves resulted in improved percent ground cover at 5 WAE. One possible explanation is that a shorter leaflet or petiole length is a trait that is linked to another trait not measured in this experiment which may be improving the overall competitive ability of the genotype.
The influence of seed size was strong 3 WAE but was barely detectable by 5 WAE. Yet, seed size was the most influential soybean trait for overall competitive ability (reduced weed biomass) measured at 7 WAE. Two general interpretations can be made. The first possible explanation is that overall competition was influenced by factors other than competition for light. Longer et al. (1986) showed large soybean seed was superior to smaller seed in root mass. Such a root mass advantage could have reduced weed biomass. The second interpretation is that the light competition at 3 WAE was more important than light competition at 5 WAE in the overall weed competition. Early vigor in seedling development is important in competitive ability (Guneyli et al., 1969). Regardless of these interpretations, seed size seems to be the single most important trait measured in this study for overall competitive ability with weeds. However, how seed size is imparting competitive ability is more difficult to discern.
Some insight into the influence of seed size may be seen in simple linear regression of seed size on all the soybean canopy traits measured. Increased soybean seed size was associated with wider leaflets, increased height at 3 WAE, increased height at 7 WAE, and a reduced soybean stand density. The association of larger seeds and wider leaflets was expected since a high number of seeds per pod and the narrow leaflet trait is considered to be a pleiotropic effect of the same allele (Johnson and Bernard 1962). More seeds per pod associated with narrow leaflets results in a lower individual seed weight. Burris et al. (1973) found that larger soybean seeds resulted in cotyledonary area increases and greater heights than small seed seedlings. A correlation between larger soybean seed and petiole length was also demonstrated by Oexemann (1942). The reason for the influence of seed size on stand count is not obvious but it could have been related to soil moisture. Smaller seeds may have imbibed and germinated faster than larger seeds. If larger seeds experienced a germination delay this may have ultimately resulted in reduced soybean stand density. Investigations of soybean seeding rate effects on weed competition show that a reduced stand density would be a disadvantage for ability to compete with weeds (Burnside and Moomaw 1977; Légère and Schreiber 1989; Nice et al. 2001, Place et al. 2009). Such a reduced stand count disadvantage may suggest a slight competitive ability trade off with the larger seed size.
Educational & Outreach Activities
At this time there have been no publications from this work. We have mentioned the project to producers and some excitement has been generated over the prospective of initiating breeding programs for organic farming systems.
Overall, competitive ability was most affected by seed size with larger seeded genotypes better able to suppress weed biomass. Narrow leaflet genotypes were poor competitors with weeds. Estimations of early canopy ground cover also showed larger seeded genotypes to have an early advantage in light competition. Although larger seed size has been implicated in improved seedling vigor and height, more research is needed to clarify what advantages in weed competition are imparted from a larger seed size.
This project has led us to a larger investigation of competitive soybean genotypes and potential variety development. This project allowed us to test and establish better protocols in effectively identifying competitive plant introductions. We have also realized that there is great potential in the development of more competitive soybean cultivars for organic producers.
An economic analysis of this project is not applicable at this stage. However, if a more competitive cultivar results in 20% improvement in weed management this could result in 5 to 10% higher yields. Such a yield increase on the 10,000 acres of organic soybean production in North Carolina could result in 10,000 to 20,000 additional bushels of organic soybeans to organic producers. If the selling price is estimated at $15 per bushel for organic soybean this would be an economic advantage potential of $150,000 to $300,000 for organic soybean producers in the state.
There is currently not a variety that farmer’s can begin using.
Areas needing additional study
Further research is needed to identify competitive soybean genotypes, the traits involved, and how easily such traits could be introgressed into future soybean cultivars.