Water's gateway to heaven: 3D imaging and modeling of transient stomatal responses in plant leaves under dynamic environments

Stomata are tiny pores on the surface of plant leaves. The pores cover less than 5% of the leaf area but facilitate the majority of the exchange of gases between the atmosphere and terrestrial vegetation. Leaves typically reduce the size of the stomatal pore to reduce water loss and the speed of stomatal closure in response to limited water availability has been shown to affect crop and ecosystem productivity. However, stomata sometimes briefly and paradoxically open in response to a water deficit, potentially increasing transpiration and accelerating leaf wilting. This so-called wrong-way response is thought to be caused by a change in the shape or volume of cells on the leaf surface (the epidermal cells). To better understand the relation between leaf structure and stomatal responses during a wrong-way response, we employed fast, high-resolution X-ray microcomputed tomography (microCT). Using this technique, we imaged a portion of a living leaf, containing nearly 100 stomata, during wilting. A custom-built gas-exchange chamber was used to measure water loss and photosynthesis during the imaging. This resulted in a unique dataset combining physiological measurements with multiple, high-resolution 3D images of the leaf structure. Surprisingly, we found that the wrong-way response did not depend on a measurable change in epidermal thickness. To deal with the vast amount of data generated, novel image analysis approaches were used to organize the information into hierarchical levels. In addition, we utilized deep-learning methodology to rapidly identify different cells and tissues within the leaf structure. We showed that quantifying uncertainty in machine learning could better leverage incorporating human expertise in image analysis, resulting in improved quantification of leaf anatomical characteristics. To further investigate the relationship between epidermal structure and stomatal function, we used fluorescence and scanning electron microscopy to examine plant lines with varying epidermal cell size and shape. Unlike chemical fixation, which can distort cell structure, rapid freezing provided a viable method for obtaining 2D images of the leaf during dehydration. By using this technique, we found a correlation between the wrong-way response and several anatomical characteristics, including the size of epidermal cells. This interdisciplinary research project provided valuable insights into the interplay between leaf anatomy and stomatal function. We made significant methodological advancements by integrating ultrafast 3D imaging, physiological measurements, and innovative image analysis techniques to quantify tissue properties.