A High-Throughput Model-Assisted Method for Phenotyping Maize Green Leaf Area Index Dynamics Using Unmanned Aerial Vehicle Imagery.

Justin Blancon,Marie Weiss,Alexis Comar,Dan Dutartre,Sébastien Praud,Frédéric Baret,Marie-Hélène Tixier

doi:10.3389/fpls.2019.00685

Justin Blancon, Marie Weiss + Show 5 more

Open Access

https://doi.org/10.3389/fpls.2019.00685

Copy DOI

Abstract

The dynamics of the Green Leaf Area Index (GLAI) is of great interest for numerous applications such as yield prediction and plant breeding. We present a high-throughput model-assisted method for characterizing GLAI dynamics in maize (Zea mays subsp. mays) using multispectral imagery acquired from an Unmanned Aerial Vehicle (UAV). Two trials were conducted with a high diversity panel of 400 lines under well-watered and water-deficient treatments in 2016 and 2017. For each UAV flight, we first derived GLAI estimates from empirical relationships between the multispectral reflectance and ground level measurements of GLAI achieved over a small sample of microplots. We then fitted a simple but physiologically sound GLAI dynamics model over the GLAI values estimated previously. Results show that GLAI dynamics was estimated accurately throughout the cycle (R2 > 0.9). Two parameters of the model, biggest leaf area and leaf longevity, were also estimated successfully. We showed that GLAI dynamics and the parameters of the fitted model are highly heritable (0.65 ≤ H2 ≤ 0.98), responsive to environmental conditions, and linked to yield and drought tolerance. This method, combining growth modeling, UAV imagery and simple non-destructive field measurements, provides new high-throughput tools for understanding the adaptation of GLAI dynamics and its interaction with the environment. GLAI dynamics is also a promising trait for crop breeding, and paves the way for future genetic studies.

Highlights

Crop production is mainly driven by the plant’s capacity to intercept and use sunlight through photosynthesis
Comparison of Green Leaf Area Index (GLAI) dynamics between years shows that maximum GLAI was significantly higher and more variable in 2017 than in 2016
We proposed an innovative way of developing transfer functions, consisting in using spectral predictors r4∗50, r5∗32, r5∗68, r6∗75, r7∗30 concurrently with additional known variables itop, φGDD, d to predict GLAI dynamics of a large panel

Summary

Introduction

Crop production is mainly driven by the plant’s capacity to intercept and use sunlight through photosynthesis. Green leaf area is influenced by several stresses including nitrogen, water and temperature (Çakir, 2004; Ding et al, 2005; Chen et al, 2018), reducing dry matter production and yield. This underlines the importance of green leaf area estimation for several applications such as yield prediction (Baez-Gonzalez et al, 2005; Dente et al, 2008), precision farming (Walthall et al, 2007), and plant breeding (Yang et al, 2017b). Its dynamics throughout the crop cycle is considered as a crucial trait for improving grain yield and adapting a genotype to a particular environment and climatic scenario (Bänziger et al, 2000; Tardieu, 2012)

Objectives

Methods

Results

Conclusion