Research Highlights

Research Highlights
Ongoing USB Research on the Effect of Cultural Practices on Seed Composition

A plot of non-nodulating soybean on the right versus a plot of nodulating soybean on the left. Non-nodulating soybean plots were included in the experiment so that the amount of nitrogen from nitrogen fixation could be estimated. Photo: Montserrat Salmeron


While average U.S. soybean yields have increased at a rate of 0.76 bu ac-1 yr-1 over the past 15 years, soybean seed protein concentration decreased from a national average of 35.5% to 34% over the same period, and this decrease has  been particularly severe for northern latitudes. Understanding how management practices can affect yield, total seed protein concentration, and the quality of seed (amino acid and fatty acid profiles) will ensure that soybean producers continue to provide a high-quality product even as yields increase. For this reason, the United Soybean Board (USB) funds research projects that quantify the effect of cultural practices on seed composition. The information in this publication summarizes preliminary data from a multi-state project conducted in 2019 to provide an overview of some of the USB supported research efforts aiming at improving soybean productivity and seed quality. Stay tuned for additional results from this project to learn about cultural practices that can help you achieve your productivity and seed quality goals.

Description of Research Study

Field experiments were carried out at 2019 in three locations (Fig. 1) to evaluate the interactive effects of genotype, environment, and management practices that affect crop nitrogen (N) dynamics and final seed composition as described in Table 1. These trials are being repeated in 2020.

Figure 1. Map of experimental sites. (1) St. Paul, MN; (2) Lexington, KY; (3) Fayetteville, AR

In brief, treatments evaluated in the experiment were two types of winter rotation (soybean after a winter fallow or soybean after a cereal cover crop), six soybean cultivars from 2 maturity groups in each location (MG1 and MG2 in MN, MG2 and MG4 in KY and AR), and three late-season input treatments (control, N fertilizer applications after R5, or Bradyrhizobium japonicum inoculant at R3).

Table 1 – Summary of factors and treatment levels in the field experiments.
1 Urea applied in two applications: 90 lb ac-1 at R5, and 90 lb ac-1 two weeks after R5. We acknowledge this is not an economically feasible rate. This high N fertilizer rate was selected to test the hypothesis that increased N availability during seed growth can modify seed composition.
2 Liquid B. japonicum inoculant (Cell-Tech liquid, Monsanto BioAg) applied at a rate of 1 oz per 1,000 ft row on the soil surface at R3 and irrigated into the soils.

Data collection and analysis

Cover crop growth was measured in the spring before cover crops were terminated and baled. Soil was sampled at this time to analyze inorganic soil N. Soybean yield was obtained by harvesting 16 to 45 square feet from bordered central rows from each plot and adjusted to 13% moisture. Seed protein and oil concentrations were determined using a Perten DA 7250™ near infrared spectroscopy (NIRs) analyzer.

Other key data collected in this study

Harvested seed was analyzed for amino acids and fatty acid composition using reverse phase HPLC and gas chromatography, respectively. The effect of the treatments evaluated on the relative amounts of different amino acids and fatty acids will be presented in another summary.

Soybean aboveground biomass and pod samples were collected during the growing season and analyzed for total inorganic N. Non-nodulating cultivars were included in our study to quantify the fraction of N from biological N fixation through analysis of δ15N and the natural abundance technique. This information will help us understand how the different treatments affected N accumulation in the plant and seeds, and final seed composition. This information will be also used to calibrate predictive tools that allow the estimation of the effect of environmental conditions and cultural practices in other years and locations in the US.


Location and MG effect

The locations and cultivar MG in the study provided a range of environmental conditions during seed filling that can impact soybean yield and seed composition (Table 2). Temperatures were highest in AR and coldest in MN, whereas solar radiation was highest in KY (Table 2). Each site included a MG cultivar considered well adapted for that location (MG 4 in AR and KY, and MG 1 in MN), and MG 2 cultivars were grown across locations to investigate the interactive effect of genotype by environmental conditions on yield and seed composition.

The later MG in all locations had yields that averaged 7.6 – 15.4 bu ac-1 more than the earlier MG, although this difference was not significant in MN. Yields were highest in KY, possibly due to the high solar radiation and relatively low temperatures compared to the other sites (Table 2).

Seed oil concentration showed a close relationship with average daily temperature during seedfill at each location, but was not different between MG (Table 2). In contrast, average protein concentration did not differ significantly across locations, except for MG 2 in AR that had the highest protein concentration compared to other sites and MG cultivars. These results support the conclusion that seed oil concentration is largely driven by environmental conditions, while seed protein concentration is impacted by the interactive effects of environmental conditions and carbon and N partitioning to the seed.

Table 2 –  Soybean yield, protein and oil concentration by maturity groups after fallow under no late-season input.

Cover crop biomass and residual soil N

Figure 2. Aboveground N in cover crop biomass (A) and soil inorganic N content (0-12 inch depth) (B) after fallow or after a cover crop in the spring before termination. * represent a significant difference in soil inorganic N content between cover crop and fallow treatments within a location at the 0.05 probability level.

A cereal rye cover crop was planted in the fall of 2018 in KY and AR on 10/18/2018 and 10/24/2018, respectively. Cover crop aboveground biomass the following spring was 3462 lb ac-1 in KY and 4559 lb ac-1 in AR by the time that cover crops were terminated and baled on 04/29/2019 and 05/14/2019, respectively. In MN, weather conditions did not allow planting a cover crop until the spring of 2019. Thus, an oat cover crop was planted on 04/14/2019 and terminated on 06/06/2019, that produced 526 lb ac-1. Nitrogen content in the cover crop biomass ranged 12 to 37 lb N ac-1 depending on the location (Figure 2A). The reduction in residual soil N (top 12 inches) by the cover crop compared to fallow treatments ranged from 5 to 20 lb ac-1 (Figure 2B). In all cases, the reduction of soil N in the spring after a cover crop was lower than the amount of N taken up by the cover crop. For instance, Figure 2 shows that in KY cover crop biomass contained 38 lbs N ac-1 at termination, but soil inorganic N only decreased by 5 lb N ac-1 after a cover crop compared to the fallow treatment. Thus, cover crops may increase N availability for the next crop when they are incorporated or left on the soil surface after termination.

Effect of cultural practices on soybean yield

Figure 3. Effect of late-season input treatments (control, inoculant application at R3, and N fertilizer application after R5) on soybean yield after fallow at each location.

Late-season input effect: The high rate of N fertilizer applied after R5 (180 lb ac-1 split in two applications) had a positive effect on soybean yields in two out of our three sites (Fig. 3). The experiment in KY was the only site with no effect of the N fertilizer treatment, coincident with high average yields (> 80 bu ac-1). The late N fertilizer application increased soybean yield after fallow by 5.5 bu ac-1 in AR, and by 10.7 bu ac-1 in MN (Fig. 3). The inoculant applications at R3 did not affect soybean yields compared to the control at any site (Fig. 3).

Figure 4. Cover crop effect on soybean yield by treatment. * Represent a significant yield difference between cover crop and fallow under the given late season input at the 0.05 probability level.

Cover crop effect: Cover crop biomass was baled and removed prior to soybean planting, to quantify any potential negative effect of growing soybean after a cover crop that may reduce soil inorganic N compared to fallow. Interestingly, yields still increased on average in soybean grown after a cover crop in KY and AR (4.9 to 9.6 bu ac-1 yield increase). This yield increase was significant in two out of the six cultivars grown in KY and AR (data not sown). The cover crop effect on soybean yield was opposite in MN.  When N fertilizer was applied to soybean, the spring oat cover crop in MN reduced soybean yield by 11.8 bu ac-1 compared to the fallow treatment (Fig. 4).

Effect of cultural practices on seed composition

Late-season input effect:  The high N fertilizer treatment increased protein concentration by 0.7 to 1.7% across all locations (Fig. 5A), indicating that protein concentration may be increased by increasing N availability after the R5 stage. The effect of the N fertilizer treatment on seed oil concentration was opposite to that observed for protein. Thus, oil concentration decreased by 0.2 to 0.6 % when high rates of N fertilizer were applied after R5 (Fig. 5B). However, total oil yield was still greater (AR and MN) or similar (KY) in the N fertilizer treatment compared to the control (Fig. 6). Similarly, total protein and oil plus protein yield increased in AR and MN with N fertilizer applications (Fig. 6). The treatment consisting of an inoculant application at R3 did not affect seed composition or oil and protein yield at any location (Fig 5).

Figure 5. Late-season input treatments effect on protein concentration (A) and oil concentration (B).
Figure 6. Late-season input treatments effect on protein content, oil content and total protein plus oil content. NS represent no significant differences and * represent a significant yield difference at 0.05 probability level within a group.

Cover crop effect: There was no significant effect of the cover crop under any of the late-season input treatments (Fig. 7). However, cover crops had a tendency to increase seed protein concentration in KY, and to decrease it in MN. This cover crop effect was significant in only a limited number of cultivars depending on the location (data not shown).

Figure 7. Cover crop effect on seed protein concentration under each late-season input treatment (control, late N fertilizer, and late inoculant application) and at each location. * and ** represent a significant difference at the 0.05 and the 0.1 probability level, respectively.


  • N fertilizer applications after R5 had a positive effect on yield and seed protein concentration, indicating that N availability during the seed filling phase is partially limiting protein concentration and yield, and this may be palliated with management practices.
  • Inoculant application at R3 had no effect on soybean yield and seed composition during our trials in 2019.
  • The effect of cover crop on soybean yield and seed protein concentration varied greatly from one location to another. Overall, the winter cereal rye cover crop had positive effect on soybean yield in KY and AR, whereas the spring oat cover crop had negative effect in soybean yields in MN. Further studies evaluating the effect of different cover crops species and management (residue baled vs. left on the soil surface) are necessary.


  • Trials are ongoing in 2020 to provide more robust conclusions on the potential of management practices that improve soybean yield and seed composition, including the relative amounts of amino acid and fatty acids. Follow up field trials and crop modeling research will be aimed at identifying economically optimal management practices that are tailored for producers in different locations and for any given year.

Research information and technical editing for this publication provided by: Montserrat Salmeron1, Ph.D., Anuj Chiluwal1, Ph.D., Larry Purcell2, Ph.D., Seth Naeve3, Ph.D., David Hildebrand1, Ph.D., Hanna Poffenbarger1, Ph.D., and Erin Haramoto1, Ph.D.

1University of Kentucky, Lexington, KY, USA, 2University of Arkansas, Fayetteville, AR, USA, 3University of Minnesota, St. Paul, MN, USA

To find research related to this Research Highlight, please visit the National Soybean Checkoff Research Database.