THE POOP EXPERIMENT: DEFINING SAMPLING PROCEDURES FOR METABARCODING DIETARY STUDIES IN BATS
Vanessa Mata (CIBIO-InBIO, Portugal) | August 2, 2017 | European Bat Research Symposium, Donostia – The Basque Country, Spain
Dietary studies based on DNA metabarcoding have boomed during the last decades. This technique offers a powerful tool to tackle old ecological questions by providing ecologists with unprecedented data resolution. At the moment, diet analysis of species, or populations, based on DNA metabarcoding techniques are commonly limited to the use of frequencies of occurrence (FO), i.e., the frequency a prey is present in a number of samples. This means that in order to have good FO estimates, an adequate number of samples is required. However few studies have assessed the impact of sampling design on molecular diet analysis and the advantages and disadvantages of the different approaches. Here we aim to assess how different experimental designs affect the accurate estimation of FO in metabarcoding studies of bat diets. We investigated the effects on FO estimates of: sequencing depth, number of PCR replicates per fecal sample, number of fecal samples per individual, and pooling of samples before extraction. One of the main goals is to understand the sources of variability in FO estimates, thereby devising sampling and analytic strategies that maximize FO estimate accuracy. For this we assessed the diet of the European free-tailed bat Tadarida teniotis in an orthogonal set-up, and sequenced a small fragment of the COI gene of 20 bat individuals, using 15 pellets per individual as well as a pool of 15 pellets (all from the same foraging night), 3 PCR replicates, and 2 levels of sequencing depth. Overall, results suggest that maximizing the number of individuals analyzed is the most important factor for accurately estimating FO, with the number of pellets per individual also affecting significantly FO estimates for each given sample size of individuals. PCR replicates were fairly consistent given similar sequencing depths, with rarer species disappearing at lower sequencing depths. These results will give us fundamental baseline information to propose guidelines for the sampling design of future studies.
Image credits: Vanessa Mata