Exploring Plant Responses to Changing Environments: Integrating Phenotyping and Modeling Across Scales

Abstract

Climate change and the depletion of essential resources like phosphorus are challenging agriculture by reducing water and fertilizer availability and ultimately threatening the security of the human food supply. Knowledge of how plants respond to changing environmental conditions is required to cope with these challenges. Plant growth information and corresponding environmental data are key to unraveling stress responses and revealing the underlying mechanisms. Understanding architectural and functional plant adaptations to stresses, such as water and nutrient limitation, is crucial to exploring new pathways to sustainable agriculture. It is vital to consider all organs, including the often-overlooked root system and surrounding soil, that are essential for water and nutrient uptake. Plant phenotyping and functional-structural plant modeling are key technologies for understanding plant responses to changing environments, making their continued development and application imperative. This doctoral project is dedicated to advancing the field of plant research by 1. developing a novel in situ phenotyping method for roots, 2. applying this method to assemble a comprehensive collection of in-field root and soil data, 3. investigating the architectural responses of Zea mays to phosphorus deficiency, 4. gaining a deeper understanding of the responses to stress by investigating the effects of phosphorus deficiency on the root system’s conductance, and 5. placing the findings into an overall context. <br /> First, a new method combining deep neural networks and automated feature extraction was developed and validated to analyze root images, reducing processing time by 98% while achieving high precision compared to manual annotation (r=0.9). Second, besides other technologies, this method was applied to assemble a comprehensive collection of in-field root and soil data over time in two minirhizotron facilities in distinct soil domains. The resulting open-access, time-series dataset includes dynamic crosshole ground-penetrating radar, minirhizotron camera measurements, and static soil sensor observations at a high temporal and spatial resolution over five years of Zea mays and Triticum aestivum experiments, including drought stress treatments and crop mixtures trials. Third, a combined approach of the developed phenotyping workflow and functional-structural plant modeling was used to investigate the responses of Zea mays to varying phosphorus availability. Combining measured architectural plant parameters with root hydraulic properties enabled time-dependent simulations of plant growth and root system conductance under different phosphorus regimes, revealing that only plants with optimal phosphorus availability sustained a high root system conductance. In contrast, all other phosphorus levels led to significantly lower root system conductance under light and severe phosphorus deficiency. It was also shown that root system organization is critical for its function rather than mere total size. Finally, this thesis contributes to collaborative studies aiming to enhance phenotyping methods and further investigate Zea mays responses to environmental changes. We found that ground-penetrating radar could be employed as a root-sensing tool in the future. By linking aboveground crop data to the belowground dataset, we revealed that maize responses to water stress vary significantly with soil conditions. We combined the automated analysis method with functional-structural modeling to show that Zea mays domestication was driven by water availability, with seminal root number emerging as a critical adaptation trait, possibly providing key information for breeding drought-tolerant varieties. Lastly, we applied an in silico approach using a game engine that visualizes plant models in high-performance computing environments to generate virtual data for neural networks, enhancing their precision and informative power. <br /> This work explores different methods, data, and models to understand plant responses to a changing environment across scales and provides new insights into the combined stress responses and development of Zea mays.