Soil Testing Helps Improve Corn Fertilizer Decisions

By Scout Nelson

Soil testing helps farmers understand the nutrients available to crops and make better fertilizer decisions. For corn growers, phosphorus is an important nutrient that supports root growth, energy transfer, and healthy plant development.

However, corn does not always respond to phosphorus fertilizer in the same way, even when soil test results appear similar. New research in South Dakota examines whether adding more soil measurements can help farmers better predict when corn needs phosphorus fertilizer.

The research was conducted at 35 locations across central and eastern South Dakota between 2019 and 2022. The findings are presented by Jason Clark, Assistant Professor and SDSU Extension Soil Fertility Specialist, along with Jeffrey Svdin, Curtis Ransom, and David Clay. The team studied how soil fertility, texture, and biological properties affect corn yield response to phosphorus fertilizer.

Why Soil Testing Matters?

Previous studies in South Dakota showed strong relationships between corn yield response and Olsen Soil Test Phosphorus (STP) levels. Researchers found that using a critical STP value of 16 ppm correctly identified responsive and unresponsive sites about 70% of the time.

However, variability remained a challenge. The confidence interval for the critical STP value ranged from 7 ppm to 21 ppm. In addition, about 40% of sites with Olsen STP values greater than 16 ppm still responded to phosphorus fertilizer, while 22% of sites with levels below 16 ppm did not respond.

These findings suggested that factors beyond soil test phosphorus might influence crop response. Researchers wanted to determine if adding information about soil health, soil texture, and other soil characteristics could improve prediction accuracy.

Research Locations and Field Trials

The study was conducted at 35 locations across central and eastern South Dakota between 2019 and 2022.

Researchers established 117 fertilizer response monitoring sites. Each location included between 3 and 5 fertilizer response experiments in areas with different landscape positions and soil test phosphorus levels.

Two phosphorus treatments were evaluated:

• No phosphorus fertilizer application
• 100 pounds P₂O₅ per acre applied as triple superphosphate (0-46-0)

The phosphorus rate was selected because it was large enough to maximize the chance of observing a yield response and matched South Dakota recommendations for an expected corn yield of 200 bushels per acre.

Nitrogen fertilizer was applied using each farmer's normal practices. Farmers also selected hybrids and managed planting and crop production using their standard methods.

Soil Measurements Evaluated

Before planting, soil samples were collected from a depth of 0 to 6 inches.

Researchers analyzed several soil properties, including:

• Olsen phosphorus
• Soil pH
• Soil organic matter (SOM)
• Apparent cation exchange capacity (CECa)
• Sand, silt, and clay percentages
• Total carbon
• Total nitrogen
• Permanganate oxidizable carbon
• Soil respiration
• Arylsulfatase enzyme activity
• Acid phosphatase enzyme activity
• Beta glucosidase enzyme activity

Corn grain yield was measured after harvest and adjusted to 15.5% moisture for fair comparisons.

Machine Learning Analysis

Researchers used a random forest machine-learning model to evaluate which soil properties were most useful in predicting corn response to phosphorus fertilization.

The analysis ranked variables using two measurements:

• Increase in node purity
• Increase in mean square error (MSE)

Results showed that Olsen Soil Test Phosphorus was the most important predictor of corn yield response.

The second most influential factor was apparent cation exchange capacity (CECa). This measurement reflects a soil's ability to hold nutrients and is often related to clay content.

Researchers found that soils with higher CECa values can hold phosphorus more tightly, which may reduce its availability to plants. This suggests that CECa may help improve future fertilizer recommendations.

Results of Combining Soil Measurements

The study found that adding common soil fertility measurements produced only modest improvements in prediction accuracy.

Using only soil test phosphorus produced:

• Calibration R²: 0.18
• Calibration RMSE: 7%
• Validation R²: 0.02
• Validation RMSE: 11%

Adding soil fertility measurements such as pH, SOM, and CECa increased:

• Calibration R² to 0.24
• Validation R² to 0.18
• Validation RMSE improved to 10%

Including soil texture measurements increased validation R² to 0.31 and lowered RMSE to 9%.

Combining soil fertility and soil health measurements resulted in a validation R² of 0.28 with an RMSE of 9%.

The model that included soil fertility, texture, and soil health together achieved the highest validation R² of 0.41 while maintaining a 9% RMSE.

Practical Value for Farmers

Although some statistical improvements were observed, researchers determined that soil biological and physical measurements provided only limited practical benefits when compared to their testing costs.

Adding pH, SOM, and CECa improved prediction accuracy by approximately 6%, while soil texture and soil health measurements improved R² values by 3% or less in many cases.

Because these additional analyses can increase testing expenses, researchers concluded that they may not provide enough economic return for routine phosphorus management decisions.

Photo Credit: pexels-jan-kroon