Towards environmental DNA-based bioassessment of freshwater reservoirs with small volumes of water: robust molecular protocols

Bioassessment of freshwater quality via eDNA is rapidly developing into a powerful alternative to traditional methods involving collecting, sorting, and identifying macroinvertebrates based on morphology. Particularly attractive would be methods that can use remote-controlled boats for sampling because it would allow for cost-effective, and frequent monitoring at multiple sites. The latter will be particularly important for tropical reservoirs that require year-around surveillance. We here optimize molecular protocols for capturing reservoir-specific differences in metazoan communities based on small water volumes (15 mL). The optimization is based on samples from two freshwater reservoirs with very different water qualities ("reservoir signal"). Each reservoir was sampled at three sites ("biological replicates"). For each water sample, the DNA was extracted twice ("technical replicates"). We then tested how much DNA template (0.1 ng to 15 ng) and how many PCR cycles (25 or 35) minimized variance between technical replicates. We find that 15 mL is sufficient for capturing the reservoir signal regardless of sampling time, template amounts, or PCR cycle numbers. Indeed, extrapolation from our results suggests that <1 mL would be sufficient because only 17 of 59 metazoan mOTUs (mainly planktonic crustaceans and rotifers) detected with a 313bp COI minibarcode were shared. We find that the use of 35 PCR cycles significantly lowered the number of detected species and that template amounts <0.5 ng yielded somewhat higher variance between technical replicates. Despite extensive trials, the variance between technical replicates remained high (Bray-Curtis: 5-20%; Jaccard: 10-40%) and we predict that it will be difficult to reduce this variance further. However, the overall reservoir differences are so strong that all biological and technical replicates can be correctly assigned.