Systematic approach to validate and implement digital phenotyping tool for soybean: A case study with PlantEye

Abstract

Abstract Agriculturally beneficial soil microorganisms (biologicals) have long been recognized as a component to enhance crop growth and yield. However, efficacy and predictability of the microbes have been low because of a knowledge gap between lab‐based selection and field performance. Bayer's Crop Science research group focuses on a key prerequisite to close this gap by understanding the plant physiological parameters improved by the application of several hundred biologicals as seed coatings and then selecting the best performers using desirable phenotypes at greenhouse prior to field screening. In such large scale screening efforts, the challenge lies in obtaining manual phenotyping measurements, which is both tedious and inefficient. Digital phenotyping tools were adopted to address this challenge. PlantEye (PE) laser scanner was selected as the high‐throughput phenotyping technology for evaluation. Specifically, suitability of PE was evaluated in the greenhouse to screen biological seed treatments in soybean [Glycine max (L.) Merr.]. The adoption of PE digital phenotyping technology was based on successfully evaluating and establishing its comparability to a standard manual measurement using a leaf area meter (i.e., reference method). Initial study performed in soybean showed a strong correlation (0.91) between reference and PE measurements. However, an important bottleneck is defining criteria for validating the results in the context of application to a screening program. In this study, exploratory analyses were performed to evaluate effect of experimental parameters (e.g., soil type, plant location in greenhouse) on plant phenotypic variation. In addition, a standard methodology is proposed to establish the comparability between reference method and PE by formulating a quantitative approach. The approach can be applied to PE as well as similar digital phenotyping tools.