Towards quantitative DNA Metabarcoding: A method to overcome PCR amplification bias

Abstract

Metabarcoding analyses have recently undergone significant development due to the power of this technique in biodiversity monitoring. However, it is still difficult to draw accurate quantitative conclusions about the ecosystems studied, mainly because of biases inherent in the environmental DNA or introduced during the experimental process. These biases alter the relationship between the amount of DNA observed and the biomass or number of individuals of the species detected. Two of the biases inherent in metabarcoding have been measured: the ratio between total DNA and target DNA concentrations, and the PCR amplification bias. A method for their correction is proposed. All experimental tests were performed on mock alpine plant communities using the marker Sper01, which is expected to have low amplification bias due to its highly conserved priming sites. Our approach combines standard quantitative PCR techniques (qPCR and digital droplet PCR) with a realistic stochastic model of PCR dynamics that accounts for PCR saturation. The model was used to estimate PCR efficiencies for each species and to infer the true species proportions of the mock communities from the read relative frequencies. The corrections are easy to implement and can be applied to previously generated DNA metabarcoding data. This work demonstrates the relative importance of the two biases considered and is an open door to quantitative metabarcoding data, although many other biases remain to be considered.