Chloroplast Movement Imaging Under Different Light Regimes With a Hyperspectral Camera.

Abstract

Plants move chloroplasts in response to light, changing the optical properties of leaves. Low irradiance induces chloroplast accumulation, while high irradiance triggers chloroplast avoidance. Chloroplast movements may be monitored through changes in leaf transmittance and reflectance, typically in red light. We present a step-by-step procedure for the detection of chloroplast positioning using reflectance hyperspectral imaging in white light. We show how to employ machine learning methods to classify leaves according to the chloroplast positioning. The convolutional network is a method of choice for the analysis of the reflectance spectra, as it allows low levels of misclassification. As a complementary approach, we propose a vegetation index, called the Chloroplast Movement Index (CMI), which is sensitive to chloroplast positioning. Our method offers a high-throughput, contactless way of chloroplast movement detection. Key features • Protocol for detached leaves handled in laboratory conditions. • Based on differential (dark-adapted versus irradiated) hyperspectral images of plant leaves. • Data analysis includes machine learning methods and the calculation of a vegetation index. • Requires irradiation equipment apart from the hyperspectral camera set.

Plant phenotyping relevance

葉の反射ハイパースペクトル画像から葉緑体位置を検出・分類する手法と指標を開発し、高スループット測定として提示しており、植物表現型取得が中心である。

Code and data availability

The protocol explicitly deposits its authors' analysis code (HyperspectralImageProcessing.m, including the pretrained CNN classifier for chloroplast positioning) on GitHub and makes the original hyperspectral images of Arabidopsis and Nicotiana leaves used in the paper's figures available on figshare. Both are paper-­‐