Advancing Marker‐Gene‐Based Methods for Prokaryote‐Mediated Multifunctional Redundancy: Exploring Random and Non‐Random Extinctions in a Catchment

ABSTRACT Multifunctional redundancy, which measures the extent of loss in multiple ecosystem functions with decreasing biodiversity, stands as a crucial index for evaluating ecosystem resilience to environmental changes. Previous studies have assessed functional redundancy by evaluating the overlap in multiple traits or functions among taxonomic units, but quantifying these experimentally for each taxonomic unit is challenging and resource‐intensive, especially for microorganisms. In this study, we aimed to refine the estimation of multifunctional redundancy in prokaryotic communities using a marker‐gene‐based methodology. We used the bioinformatic tool, PICRUSt2, to predict KEGG ortholog (KO) composition for each Amplicon Sequence Variant (ASV) of the 16S rRNA gene and assessed community‐wide KO richness as a metric of genomic multifunctionality. Additionally, we introduced a refined regression on KO richness vs. ASV richness curves, providing a reliable estimate of the power‐law exponent within computational time constraints, serving as the multifunctional redundancy index. To quantify variations between different species extinction scenarios, we incorporated various non‐random extinction scenarios alongside a random one. Finally, to elucidate potential mechanisms contributing to the variations in multifunctional redundancy, we applied our refined methodology to the prokaryote community in Lake Biwa and four of its inlet rivers. Our refined marker‐gene‐based methodology for quantifying multifunctional redundancy was effectively validated using experimentally quantified phenotypic multifunctionality. Our application to the Lake Biwa catchment revealed spatio‐temporal variations in multifunctional redundancy. Our analysis demonstrated lower redundancy in Lake Biwa compared to its inlet rivers, aiding in prioritising conservation targets and inferring distinct community assembly processes. Future directions include a deeper exploration of KO composition information for detailed multifunctionality quantification and the refinement of extinction scenarios. This study demonstrates the promising integration of bioinformatic functional prediction and modelling biodiversity loss, offering a valuable tool for effective ecosystem management.