Drone-measured canopy spectra can be a complementary predictor of belowground arbuscular mycorrhizal fungi variations in urban community gardens

Abstract

Arbuscular mycorrhizal fungi (AMF) form essential symbioses with plant roots, improving nutrient uptake, plant health, and soil function. However, assessing AMF abundance in urban ecosystems remains challenging. Drone-based multispectral imaging can estimate canopy functional traits, raising the question of whether variation in aboveground canopy metrics can serve as an ecological indicator of belowground AMF abundance. We sampled five community gardens over two growing seasons, combining quadrat-level plant surveys, soil sampling for AMF abundance, Unmanned Aerial Vehicle (UAV) multispectral imagery, and point clouds. From the UAV we derived vegetation indices, spectral diversity and texture metrics, and from the point clouds we extracted canopy height metrics. We related these traits to log AMF abundance using LASSO cross-validated regression and piecewise structural equation modelling (SEM). Our analyses showed that drone-derived canopy traits partially capture the response of plants to the variations in AMF abundance. Higher AMF abundance was associated with greener and more vertically heterogeneous canopies. UAV-derived canopy texture features explained up to ~36% of variance in AMF abundance. Piecewise SEM showed a non-linear relationship between climate and AMF abundance (β = 0.25; R² = 0.62). Vegetation indices (canopy greenness) had positive association with AMF (β = 0.21). Overall, UAV-derived canopy traits represent a scalable ecological indicator of relative AMF abundance in urban gardens, supporting non-destructive assessment of relative AMF variation and spatial targeting of field sampling. Establishing this link provides a baseline for developing stronger urban ecological indicators by integrating UAV data with soil measurements, garden management, and higher-resolution remote-sensing approaches.

Yasamin Afrasiabian

PhD student

My research interests include remote sensing, particularly hyperspectral UAV and satellite imaging, and machine-learning methods for ecosystem-biodiversity characterisation, precision agriculture, and hydrological analysis.

Prof. Dr. Kang Yu

Professor of Precision Agriculture

My research interests include precision crop farming, hyperspectral remote sensing, and AI in agriculture.