Document Type

Thesis - Open Access

Award Date

2023

Degree Name

Master of Science (MS)

Department / School

Agronomy, Horticulture, and Plant Science

First Advisor

Jason Clark

Keywords

Corn, Illite, Machine Learning, Potassium, Smectite, Soil Test Correlation

Abstract

Corn (Zea mays L.) is a vital commodity in South Dakota’s agricultural sector. Optimal corn production occurs when there are sufficient mineral nutrients in the soil, especially potassium (K). Applications of K fertilizer are used when soil test K (STK) levels are deficient. Therefore, producers need reliable, thoroughly tested fertilizer recommendations to make profitable decisions and maintain environmental stewardship. South Dakota K fertilizer recommendations have not been updated in nearly 20 years. Simultaneously, changes in corn genetics, management practices, and climate patterns suggest that the critical soil test value (CSTV) for STK may have shifted in that same time frame. Furthermore, the addition of other variables, notably clay mineralogy, could improve the accuracy of K fertilizer recommendations. Therefore, the objectives of this study were to 1) evaluate relationships among clay mineralogy, STK, and other common soil test parameters, and 2) use those relationships to improve K fertilizer recommendations for South Dakota. From 2019 to 2022, soil samples were collected from 43 locations, and field trials were conducted at 35 locations throughout central and eastern South Dakota. A correlation matrix and nonlinear regressions demonstrated significant relationships between STK and the smectite:illite ratio. Linear regressions between STK and several other soil parameters were influenced by smectite:illite ratio groupings: (illitic [1 but4.5]). Soil test K and several other soil test variables (water-soluble K, total K, soil organic matter [SOM], and clay content) were all positively related regardless of clay mineralogy, but STK was predicted to be lower by all soil test variables in highly smectitic soils as opposed to illitic and smectitic soils. Moreover, STK decreased as pH increased in highly smectitic soils. Random forest modeling identified STK as the most important variable for predicting the smectite:illite ratio. Therefore, the interactions between STK, the smectite:illite ratio, and other soil parameters should be further investigated for implementation in K fertilizer recommendations. Using soil test correlation techniques, seven nonlinear regression models displayed a wide range of CSTVs (111-196 mg kg-1 STK). Using model averaging, the optimal CSTV for improved corn yield response predictions was 144 mg kg-1, which was lower than the current South Dakota CSTV of 160 mg kg-1. While clay mineralogy variables were not identified as important predictors of yield responsiveness using random forest modeling, CEC, SOM, and permanganate oxidizable carbon (POXC), along with STK (CSTV = 144 mg kg-1) were important. Using these variables in a decision tree improved prediction accuracy from 62% to 72% compared to using STK alone (CSTV = 160 mg kg-1). Overall, these results demonstrated that there were significant relationships among STK, clay mineralogy, and other soil parameters, but clay mineralogy could not confidently be incorporated into K fertilizer recommendations. Rather, lowering the CSTV from 160 to 144 mg kg-1 STK and inclusion of CEC, SOM, and POXC resulted in improved accuracy of corn yield responsiveness to K fertilization. These results will help corn producers in South Dakota and abroad to improve farm profitability and reduce misapplications of fertilizer.

Library of Congress Subject Headings

Corn -- Fertilizers -- South Dakota.
Potassium fertilizers -- South Dakota.

Number of Pages

155

Publisher

South Dakota State University

Share

COinS
 

Rights Statement

In Copyright