Estimating Aquatic Plant Diversity Using Spectral and Phylogenetic Metrics

As aquatic ecosystems are threatened globally, the conservation of aquatic plant diversity is becoming a priority. In the last decade, remote sensing and phylogenetics have opened up new ways of measuring biodiversity, and linking spectral and phylogenetic features to characterise plant communities can further advance this issue. In this study, we explored the use of phylogenetic features derived from a fully resolved supertree of spectral features extracted from centimetre resolution hyperspectral images collected by a drone to estimate functional diversity (richness, divergence, and evenness) using non-linear parametric and machine learning models within communities of floating hydrophytes and helophytes sampled across a trophic gradient. Our results show that all three functional diversity metrics can be estimated from spectral features using machine learning models (Random Forest; R2=0.89-0.91), while parametric models perform worse (GAMs; R2=0.35-0.79), especially for community evenness. Merging phylogenetic and spectral features improves modelling performance for all diversity metrics (R2=0.91-0.97) using machine learning, but only benefits community evenness estimation when GAMs are used. Combining imaging spectroscopy and phylogenetic analysis can provide a quantitative means of capturing variability in macrophyte communities across scales and gradients, to the benefit of aquatic ecologists focusing on the study and monitoring of biodiversity and related processes.