Flows in Arbuscular Mycorrhizal Fungi Networks

Abstract

Arbuscular mycorrhizal fungi (AMF) form extensive extraradical hyphal networks in soil that mediate nutrient exchange between the fungus and its host plant. AMF, through this symbiotic relationship with its host plant, has been shown to increase survivability for its host plant when faced with external stressors, such as nutrient-depleted soil or droughts. Understanding how the geometry and topology of these networks govern internal fluid transport is critical for predicting nutrient delivery efficiency, yet the relationship between network architecture and flow remains poorly characterized due to the difficulty of measuring cytoplasmic streaming at the whole-network scale. This thesis presents a computational framework for generating realistic AMF-like tubular networks and solving a model of steady-state pressure and flow fields within them, enabling systematic investigation of how branching structure, anastomosis, and geometric taper shape transport performance. The transport model decomposes the flow in each hypha into a motor-driven and a pressure-driven component, with the inner and outer radii, edge lengths, and conductivities derived self-consistently from the network geometry and lipid demand. These design choices are informed primarily by data from researchers at the Vrije Universiteit Amsterdam, who are collaborating on the study of AMF. Results demonstrate that the equilibrium pressure field increases monotonically from the root toward hyphal tips, with the steepest gradients concentrated in proximal trunk edges, consistent with motors actively pushing fluid outward against an accumulating backpressure. Anastomosis, which is when hyphae fuse into one another to form closed loops in the network, is shown to be the primary conduit for net bulk water redistribution between branches, while tree-like portions of the network carry negligible net flow. Batch simulations across ensembles of randomly seeded networks reveal how statistical transport properties, such as mean pressure, maximum water flux, and loop count, vary with network size and geometry. Comparisons to actual biological networks are made. Limitations of the current framework are discussed, and directions for future work addressing these gaps are outlined.