​​High-throughput phenotyping methods of guard and subsidiary cells in maize

Stomata, small pores on land plant leaves, facilitate gas exchange. Despite kidney-shaped guard cells being common, stomata exhibit diverse morphology. Subsidiary cells' number, shape, and size vary. Opening stomata lets CO2 in but increases water loss. Plants with fast stomata opening may use water more efficiently. Grass stomata respond fast due to uniquely dumbbell-shaped guard and triangle-shaped subsidiary cell shapes 1 . Our research suggests subsidiary cells aid stomatal closure 2 . What features of guard and subsidiary cells optimize closure efficiency? To tackle this question, we measured stomatal closure speed and visualized stomatal morphology across 29 maize inbred lines. To address this challenge of lacking high-throughput phenotyping methods, we developed a deep-learning model to precisely identify and segment guard cells and subsidiary cells. Comparison between hand-traced and model-predicted sizes revealed no significant differences in both guard cells and subsidiary cells, indicating the high accuracy of our model predictions. Subsequently, we utilized Cell Profiler 3 , an open-source software for modular image analysis pipelines, to extract high-dimensional anatomical features from segmented cells. Our study unveiled variations in stomatal dynamics and morphology among maize genotypes. Subsequent analysis indicates these variations are genotype dependent. We aim to share our pipeline and novel research with the plant science community.