High-throughput soil health assessment to predict corn agronomic performance

Abstract

Developing high-throughput methods to quantify soil health can support decision making to minimize soil constraints for crop growth and increase sustainability in agroecosystems. This study evaluated the ability of soil health assessment conducted at the beginning of the crop growing season to predict corn agronomic performance in nutrient fertilization trials. We compared different methods to assess soil health, from the most traditional approach of measuring individual soil biochemical indicators to more novel approaches using diffuse reflectance infrared Fourier transform spectroscopy (mid-DRIFTS). Soil health was quantified using: I) traditional and emerging biochemical soil health indicators (up to 20 soil indicators), II) mineral and organic functional groups abundances estimated as peak areas in mid-DRIFTS, and III) mid-DRIFTS with partitioned spectra and partial least squares regression. Corn response to fertilizer was evaluated in a) phosphorus, potassium, and sulfur omission trials (2016–2018; 41 sites across 17 Ohio counties) as well as in b) N rates trials (2016–2017; 25 fields across 14 Ohio counties). Modeling using either biochemical soil health indicators or mid-DRIFTS accurately predicted corn yield across nutrient treatments in omission trials (R2V=0.61–0.84 in validation sets) and unfertilized controls in N rates trials (R2V=0.68–0.81) as well as corn N-driven yield gap (i.e., delta yield defined as grain yield at agronomic optimum N rate minus grain yield of the control, R2V=0.64–0.76) in N rates trials. Mid-DRIFTS approaches equaled or outperformed the predictive ability of measured biochemical soil health indicators in estimating crop performance (up to 38 % improvement in R2V), unveiling a promising method to more efficiently assess soil health. Early season soil health assessment via mid-DRIFTS has potential to directly estimate corn agronomic performance.