Archived

This content is available here for research, reference, and/or recordkeeping.

Author ORCID Identifier

https://orcid.org/0009-0003-0549-0168

Date Available

7-29-2026

Year of Publication

2026

Document Type

Doctoral Dissertation

Degree Name

Doctor of Philosophy (PhD)

College

Agriculture, Food and Environment

Department/School/Program

Plant and Soil Sciences

Faculty

Edwin Ritchey

Faculty

Arthur Hunt

Abstract

Soil scientists did not develop soil test-based fertilizer recommendations to support precision nutrient management. Knowledge of crop responses to applied essential nutrients, such as phosphorus (P), and the processes governing these responses at high spatial resolution are necessary to improve precision agronomic recommendations. This dissertation investigates the limitations of current soil sampling methodologies and recommendation frameworks for crop response to fertilizer, with particular emphasis on P fertility in Kentucky (KY) corn production. The first study examined the relationship between the resolution of systematic grid sampling and the spatial structure of soil test variables (Mehlich-3 P (M3P), Mehlich-3 potassium (M3K), and soil pH) across four fields in KY. Results demonstrated that the conventional 100-m grid sampling resolution failed to capture the structural component and spatial variability of soil test variables in these fields. These findings challenge the universal application of standard grid sampling protocols and highlight the mischaracterization of nutrient spatial heterogeneity within fields.

The second study evaluated corn yield response to applied P fertilizer at low M3P levels, using a spatial resolution higher than the conventional grid sampling protocol across two fields in Eastern and Western KY. Corn yields increased significantly with fertilizer P application when M3P values fell below 30 mg kg-1; however, fewer than half of the main plots were yield-responsive. Mehlich-3 P also failed to significantly predict nutrient response in three of six site-years. Additionally, geographically weighted principal component analysis (GWPCA) and random forest (RF) modeling were used to assess soil chemical, physical, and landscape properties influencing yield response. Random forest modeling and GWPCA identified spatially heterogeneous variables beyond M3P that contribute to understanding yield responsiveness. Spatio-temporal models of the probability of corn yield response to P fertilization were developed using spatial generalized additive models (GAM) and state-space modeling frameworks. Mehich-3 P alone was insufficient to explain the observed spatial and temporal patterns in the probability of yield response. Persistent spatial patterns in response likelihood were identified, indicating that integrating the spatial structure of soil physical, chemical, and landscape variables with temporal variation is necessary to characterize fertilizer responsiveness when M3P falls below 30 mg kg⁻¹. Collectively, this research demonstrates that current soil test-based fertilizer recommendations inadequately support high-resolution P management. High-resolution sampling, spatially explicit analytical methods, and probability-based decision frameworks that incorporate spatial structure and temporal variation are essential for advancing site-specific nutrient management while acknowledging inherent vulnerabilities in current fertilizer recommendation methods.

Digital Object Identifier (DOI)

https://doi.org/10.13023/etd.2026.373

Archival?

Archival

Funding Information

This study was supported by the United States Department of Agriculture (USDA-ARS) project, Precision Nutrient Management Through Improved Understanding of Soil Dynamics (no.: 3098-13610-009-092-A).

Download

Share

COinS