27An ongoing challenge for ecological studies has been the collection of data with high precision 28and accuracy at a sufficient scale to detect effects relevant to management of critical global 29 change processes. A major hurdle for many workflows has been the time-consuming and 30 challenging process of sorting and identification of organisms, but the rapid development of 31 DNA metabarcoding as a biodiversity observation tool provides a potential solution. As high-32 throughput sequencing becomes more rapid and cost-effective, a 'big data' revolution is 33 anticipated, based on higher and more accurate taxonomic resolution, more efficient detection, 34 and greater sample processing capacity. These advances have the potential to amplify the 35 power of ecological studies to detect change and diagnose its cause, through a methodology 36 termed 'Biomonitoring 2.0'. 37Despite its promise, the unfamiliar terminology and pace of development in high-38throughput sequencing technologies has contributed to a growing concern that an unproven 39technology is supplanting tried and tested approaches, lowering trust among potential users, 40 and reducing uptake by ecologists and environmental management practitioners. While it is 41 reasonable to exercise caution, we argue that any criticism of new methods must also 42 acknowledge the shortcomings and lower capacity of current observation methods. Broader 43 understanding of the statistical properties of metabarcoding data will help ecologists to design, 44 test and review evidence for new hypotheses. 45We highlight the uncertainties and challenges underlying DNA metabarcoding and 46 traditional methods for compositional analysis, focusing on issues of taxonomic resolution, 47 sample similarity, taxon misidentification, sample contamination, and taxon abundance. Using 48 the example of freshwater benthic ecosystems, one of the most widely-applied non-microbial 49 applications of DNA metabarcoding to date, we explore the ability of this new technology to 50 3 improve the quality and utility of ecological data, recognising that the issues raised have 51 widespread applicability across all ecosystem types. 52

No abstract

Metabarcoding is capable of delivering consistent and accurate fine-resolution biodiversity data, and offers great promise for improving aspects of environmental assessment and research. Even so, many ecologists are keen to make further inferences about species’ abundances and the number of sequence reads has proven to be a poor proxy for abundance. The conservative interpretation has been to treat metabarcoding data as presence/absence, and although such data are less rich, occurrence and abundance are only different expressions of the same phenomenon. Interestingly if we assume the probability of detecting individuals is constant, it should be possible to use changes in the frequency of detection to infer changes in the underlying abundance. We tested the possibility that changes in the abundance structure of benthic macroinvertebrate communities could be recovered using replicated metabarcoding. We conducted 5 monthly surveys from Jun-Nov 2019 at the Catamaran Brook, a small tributary of the Little Southwest Miramichi River in New Brunswick, Canada. Each survey collected 30 benthic samples divided between control and treatment cages that excluded predatory fish. A further 6 samples were taken for traditional microscopic identification and counting. Analysis of the metabarcoding data demonstrated that we could recover plausible changes in abundance from occurrence data, including significant responses to both seasonal dynamics and the experimental exclusion of predators. The microscopy samples merely confirmed that count data are highly stochastic, and therefore while specific estimates of expected abundance from our model are highly uncertain, they capture those differences we could validate. In summary, while we confirmed that occurrence data are more robust for routine bioassessment, it is possible to recover fine-resolution changes in abundance that can inform ecological studies using metabarcoding.