Abstract
Summary Longitudinal studies that integrate samples with variable biomass are essential to understand microbial community dynamics across space or time. Shotgun metagenomics is widely used to investigate these communities at the functional level, but little is known about the effects of combining low and high biomass samples on downstream analysis. We investigated the interacting effects of DNA input and library amplification by polymerase chain reaction on comparative metagenomic analysis using dilutions of a single complex template from an A rabidopsis thaliana ‐associated microbial community. We modified the Illumina Nextera kit to generate high‐quality large‐insert (680 bp) paired‐end libraries using a range of 50 pg to 50 ng of input DNA . Using assembly‐based metagenomic analysis, we demonstrate that DNA input level has a significant impact on community structure due to overrepresentation of low‐ GC genomic regions following library amplification. In our system, these differences were largely superseded by variations between biological replicates, but our results advocate verifying the influence of library amplification on a case‐by‐case basis. Overall, this study provides recommendations for quality filtering and de‐replication prior to analysis, as well as a practical framework to address the issue of low biomass or biomass heterogeneity in longitudinal metagenomic surveys.