Proving a negative; estimating species ′Confidence in Absence for Decision-Making′ (CIADM) using environmental DNA monitoring

Policy-driven decision-making is an important aspect of environmental management globally, often focused on protecting priority species. However, declining trends in freshwater biodiversity have resulted in a lack of up-to-date knowledge regarding the distribution of rare and elusive species. Furthermore, when dealing with priority species, it is sometimes more important to provide a confident assessment of absence, than merely confirm their presence. Without such confident assessments, resource intensive management plans may be misplaced, and not adequately targeted to conserve important remaining populations. Here, we present a framework to estimate confidence in absence, referred to as ′Confidence in Absence for Decision-Making ′ (CIADM), based on single-visit environmental DNA metabarcoding data obtained from water samples. It uses a case study of European eel presence / absence upstream of 44 water pumping stations, given their critically endangered status and the legislative drivers for remediation (EC Eel Regulation 1100/2007, Eels (England and Wales) Regulation 2009). Through a high degree of biological (sample) and technical (PCR) replication, we retrospectively assigned ′confidence in absence′ values and proposed various strategies to achieve the required confidence levels in future surveys. Our findings indicate that 17 out of 44 pumping stations tested positive for eel, and we were able to assign a >99% confidence level that the remaining 27 sites were negative for eel DNA at the time of sampling. Increasing both biological and technical replication increased ′confidence in absence′ values. For example, using three PCR replicates per sample, required four replicate biological samples to achieve >95% and six to achieve >99% confidence in eel absence given non-detection. However, we estimate that by using seven PCR replicates per sample a >99% confidence in eel absence following non-detection could be achieved with only three replicate biological samples. Furthermore, we found that eel positive sites had significantly higher species richness, and fish communities differed between eel positive and eel negative sites. This study highlights the importance of optimising workflow specific replication, and provides an adaptable framework to produce confidence estimations of priority species absence given non-detection.