Functional-Structural Modelling of the Palisade Mesophyll Air Channel

The photosynthetic pathway has been extensively researched at the subcellular level and for the whole leaf, with established biochemical and mechanistic models. While it is known that the balance of intercellular airspace and cellular material depends on patterns of cell division and growth, it is less well understood how different intercellular airspace arrangements impact photosynthesis. What is the relationship between the amount of internal mesophyll surface area exposed to air, photosynthetic capacity and cell size? Is there an optimal airspace-tissue balance, and if so, how will it change under future levels of CO2? With recent advances in micro-computer tomography (μ-CT), it is possible to quantify the intercellular mesophyll airspace at a local scale and obtain paired datasets of global gas exchange measurements. In this thesis, we propose a mathematical framework to model air channels in the palisade mesophyll layer as microscopic air channels by applying Fick’s first law of diffusion. By validating this mechanistic model with experimental observations of paired μ-CT/gas exchange data in a series of Arabidopsis thaliana mutants, we can explore the trade-off between the amount of cellular material and air channels in the palisade layer. More importantly, this structure-function modelling framework allows us to identify which of these structural properties are favoured by mutants with better photosynthetic performance, and quantify whether there is any consistent behaviour leading to an ‘optimal’ trade-off between cellular material and air channels. The analysis shows that the intercellular CO2 uptake rate is a function of the palisade mesophyll air channel depth. This observation implies a trend in which leaves with shorter intercellular diffusion pathways within the palisade mesophyll have a higher CO2 assimilation rate, which could suggest an optimisation of the inner mesophyll cell structure. Hence, from a physical diffusion-based model supported by data-driven estimation, we are able to predict function from structure.