Background

Plants show distinct chloroplast positioning, dependent on light conditions. In low light, chloroplasts gather at cell walls lying perpendicular to the direction of incident light, a phenomenon known as the accumulation response. In high light, chloroplasts show avoidance response by moving to the cell walls parallel to the direction of incident light. Distinct dark positioning is also observed, with chloroplasts located at the bottom of the palisade cells [1].

Chloroplast movements have been examined using light microscopy for over a century [2,3]. Such observations are still important for assessing chloroplast responses, providing a detailed description of the movements of chloroplasts within the cell [4]. However, microscopic observations are laborious and usually restricted to the outer layers of the mesophyll in unfixed leaves. An alternative approach is to take advantage of the substantial changes in the optical properties of the whole leaf that result from chloroplast movements. Their optical effects are often visible to the naked eye. The slit assay is based on a simple observation that chloroplast avoidance makes leaves look pale, while chloroplast accumulation renders them darker [4]. Changes in leaf transmittance yield information on averaged chloroplast positions within the mesophyll cells and enable quantitative assessment of chloroplast movements. Photometric devices are used for measuring blue-light-induced changes in red light leaf transmittance [5–7]. They provide a continuous recording of the changes in leaf transmittance, which allows for studies of the kinetics of the process, though only one leaf can be examined at a time. Chloroplast movements may also be measured by changes in leaf reflectance [8–10]. Red light reflectance was used to monitor chloroplast movements with photosynthetic efficiency during the growth of Arabidopsis thaliana [11]. Two previous works [12,13] employed an integrating sphere to measure chloroplast movement-induced changes in diffuse reflectance. This approach involves the collection of whole (i.e., hemispherical) leaf reflectance, which is independent of the observation angle. This reduces the sensitivity of measurements to the leaf orientation and its surface undulations. However, such measurements are limited due to the requirement of direct contact with the leaf, which results in low throughput. Hyperspectral imaging is a fast, non-contact method of acquiring reflectance spectra in a spatially resolved manner; thus, it can be used to assess chloroplast positioning in several leaves at the same time. Such high throughput may be useful for phenotyping multiple mutants or studies on leaves collected in the field, in which a large sample is necessary due to the heterogeneity of the light conditions in natural environments. However, while a detector connected to an integrating sphere can measure light reflected to the whole hemisphere, the camera collects light reflected to a narrow cone. As a result, specular reflection from the leaf surface may reduce the accuracy of leaf classification according to chloroplast positioning, especially when leaves contain pronounced veins or their surface undulates. Specular reflectance may be reduced for such samples using light polarizers [10]. In an alternative approach, used in this protocol, classification of leaves is performed not based on raw reflectance, but using the complete visible light reflectance spectrum, through application of pretrained machine learning classifiers. The workflow of the protocol is shown in the Graphical overview.