Degree Type

Dissertation

Date of Award

2018

Degree Name

Doctor of Philosophy

Department

Agronomy

Major

Crop Production and Physiology

First Advisor

Sotirios V. Archontoulis

Abstract

Optimizing nitrogen (N) management in maize (Zea mays L.) production systems is critical and essential to ensure profitability, productivity, and environmental sustainability. However, it represents a challenge because N is highly mobile within the soil-plant-atmospheric system. Therefore finding the optimum N rate for maize is a difficult task. The overall goal of this research was to evaluate crop model and statistical -based approaches to making N recommendations for maize and quantify prediction accuracy in two major maize production regions: Iowa, USA and Buenos Aires, Argentina. I addressed three questions: 1) how accurately process-based modeling and statistical based approaches can simulate yields and optimal N rates, 2) how does the accuracy change when models are used as a forecasting tools (with limited input data), and 3) which soil, crop, and atmospheric variables are most important to improve understanding of optimum N rate variability from year-to-year and from field-to-field? Data to test crop model predictions included yield response to N from a 16-year field experiment conducted in central Iowa, USA with two crop rotations totaling 31 N-trials. Data to test statistical models included a 5-year yield response to N from central-west Buenos Aires, Argentina with different rotations, soil properties, and landscape positions totaling 51 trials. The statistical-based approach predicted optimal N rates with higher accuracy than process-based models (root mean square error, RMSE of 42 vs 62 kg N ha-1, respectively). Yields that were predicted at the end of the season had a RMSE that ranged from 1 to 1.3 Mg ha-1. The accuracy of yield predictions at planting decreased more for optimal N rates when using process-based models. Optimal N rate at planting was predicted with similar accuracy to that predicted at the end-of-season (RMSE 60 and 47 kg N ha-1 for process- and statistical-based approach, respectively). Lastly, I found that the spring precipitation (April to June) and the precipitation events greater than 20 mm accumulated from planting to silking highly explained the variability in optimal N rates in both central Iowa and in central-west Buenos Aires.

Copyright Owner

Laila Alejandra Puntel

Language

en

File Format

application/pdf

File Size

157 pages

Included in

Agriculture Commons

Share

COinS