A New Insight on Corn Yield Responses to Plant Density and Nitrogen Rates: Data Distribution Analysis.

Plant density and fertilizer nitrogen (N) rate have a large impact on maize (Zea mays L.) yield components and consequently on the attainable yield. Optimum plant density and fertilizer N rate are known to present high variability between- and within-field. Therefore, identifying the optimum plant density and N rate is fundamental to achieve high maize profitability. The main objectives of this study were to: (i) evaluate yield responses to plant density and fertilizer N rates at varying yield environments, (ii) develop and validate yield-factor models using data from Brazil (BRZ) and US and (iii) perform a sensibility analysis to verify what are the key factors leading to similarities in the yield-plant density, and yield-N rate relationships. Analyses were conducted in a database from field studies carried out from 2009 to 2017 in southern Brazil and from 2012 to 2015 in the US Midwest region. The database from BRZ was divided into two categories, aiming model construction, “training dataset”, and for validation purposes, “validation dataset”. The US database was only utilized for validation purposes. Data analysis comprised three steps: i) yield-factor response analysis by yield environment, ii) model validation utilizing independent databases, and iii) identification of key parameters in the yield data distribution with the highest impact on yield-density response models. Main outcomes from this study were: 1) a significant interaction of plant density and N rate with yield environment, resulting in different yield-factor response models at each yield environment, 2) optimum plant density and N rate increased from low- to high-yield environments, 3) spatial validation of yield-factor models – yield-density relationship was more dependent on similarities in yield frequency distribution rather than in geographical distances, and 4) yield data distribution factors such as 50% interquartile range and mode were identified as relevant for obtaining more “universal” yield-density response models.