Corn Yield Response to Plant Density and Nitrogen: Spatial Models and Yield Distribution

Rai Schwalbert,Telmo J.C. Amado,Tiago A. N. Horbe,Lincon O. Stefanello,Y. Assefa,P.V.Vara Prasad,Charles W. Rice,Ignacio A. Ciampitti
DOI: https://doi.org/10.2134/agronj2017.07.0425
IF: 2.65
2018-05-01
Agronomy Journal
Abstract:Corn yield response to plant density and N rate were dependent on yield environment.Agronomic optimal plant density and N rate were positively correlated to yield level.Yield to density within a yield environment was independent on year, country, and hybrid.Similarity in yield frequency data distributions lead to similar yield–factor responses. Understanding the relationship of corn (Zea mays L.) yield responses to plant density and nitrogen (N) fertilization is critical to production decisions. The main objectives of this study were to (i) evaluate yield responses to plant density and fertilizer N rate at varying yields adjusting models considering a spatial component, (ii) perform a validation for the fitted models with an independent dataset, and (iii) identify key statistical parameters for the yield data distribution governing response models. Analyses were conducted with information from seven fields with 21 studies (one study per yield environment, with three environments per field) conducted from 2009 to 2017 in southern Brazil with geospatial data collected to evaluate yield response to plant density and fertilizer N rates (28911 data points) and one additional database with 12 field studies conducted from 2012 to 2015 in the US Midwest (1773 data points). Databases were divided into training and validation datasets. Field experiments evaluating both plant density and N rate were selected as training dataset. Key research findings were (i) yield–factor response models were dependent on yield environment and within a yield environment those models remained constant regardless the year, country, and hybrid for all evaluated fields, (ii) statistical models considering spatial correlation of the random errors outperformed those considering errors independent and identically distributed and, (iii) yield distribution with comparable 50% interquartile range and mode portrayed similar yield–factor relationship. In summary, fitting spatial yield–density models considering yield data distribution is critical to upscale site‐specific models to larger spatial domains.
agronomy
What problem does this paper attempt to address?