Can soil microbial diversity mitigate water stress and maintain crop yields in agricultural systems?

Intensive conventional agriculture allows us to increase crop yields in line with global demand, but it puts future yields and food security at a disadvantage due to the increased vulnerability to changes in the environment that intensive systems are less able to withstand. Arbuscular mycorrhizal fungi (AMF) are a key target for increasing agricultural sustainability as they moderate aspects of ecosystem functioning including plant productivity, nutrient cycling, soil structural maintenance, water relations and pathogen regulation. Restoring AMF functioning could be of importance to restoring degraded soil properties to confer climate change resistance and resilience. The main aims of this thesis were (a) to investigate how various agricultural management practices impact the community composition and diversity of AMF, and (b) to explore the causal pathways of AMF diversity and community composition in conferring benefits to soil health and functioning through crop yields and water stress mitigation. Overall, the thesis considered AMF communities against whole-community functional phenotypes and soil health under contrasting agricultural management practices. This included AMF inoculation of soils, the inclusion of grass-clover leys in arable rotations and contrasting tillage intensities. The thesis increases our knowledge of the link between management practices, soil health and crop yields that can be used to inform management choices in real-world agricultural situations. This is particularly important for grass-clover leys which are a relatively understudied management practice. From the findings of this thesis, it is recommended that future studies employ a reductionist approach to assessing AMF function under variable situations using a combination of targeted mechanistic experiments and larger holistic experiments. These experiments would assess individual AMF functions under different environmental contexts that will provide a trait-based framework of individual function and compliment them with larger scale community experiments that can expand upon the mechanistic knowledge gathered to begin to predict community-level function