NeMO: a flexible R package for nested multi-species occupancy modelling and eDNA study optimisation

Biodiversity monitoring using environmental DNA (eDNA) metabarcoding has expanded rapidly in recent years, offering a non-invasive and widely embraced tool among ecologists and stakeholders. However, eDNA surveys are prone to imperfect detection, and non-detections are often misinterpreted as true absences. Despite its critical implications for biodiversity assessments, detection uncertainty remains rarely quantified in eDNA studies. Occupancy modelling provides a powerful tool to address this limitation, yet it remains underused in eDNA applications, in part due to a lack of accessible and flexible tools. To bridge this gap, we developed NeMO (Nested eDNA Metabarcoding Occupancy), a user-friendly R package for fitting multi-species occupancy models in a Bayesian framework. NeMO explicitly accounts for the nested structure of eDNA metabarcoding studies, typically involving multiple replication steps: collecting several eDNA samples per site and running multiple PCR replicates for each sample. It further incorporates sequencing read counts and different PCR pooling strategies. This framework estimates species occupancy, the probability of eDNA collection when present, the probability of eDNA amplification when collected, and the expected sequence read count when amplified. It also enables users to assess how environmental or methodological covariates affect these probabilities. Importantly, NeMO helps estimate the minimum number of eDNA samples, PCR replicates, and sequencing depth needed to confidently confirm species absence, offering a retrospective tool to evaluate and optimise study design. We demonstrate its utility using a published fish biodiversity dataset from the Rhône River (France). NeMO integrates key modelling features into a single streamlined framework, making it broadly accessible for researchers and practitioners, to rigorously assess detectability and optimise resource allocation in eDNA metabarcoding surveys. Our results highlight the importance of quantifying detection uncertainty, with significant implications for monitoring elusive species and for guiding robust, cost-effective eDNA sampling strategies.