Abstract
Stools are commonly used as proxies for studying human gut microbial communities as sample collection is straightforward, cheap and non-invasive. In large-scale human population surveys, however, sample integrity becomes an issue as it is not logistically feasible for researchers to personally collect stools from every participant. Instead, participants are usually given guidelines on sample packaging and storage, and asked to deliver their stools to a centralised facility. Here, we tested a number of delivery conditions (temperature, duration and addition of preservative medium) and assessed their effects on stool microbial community composition using 16S rRNA gene amplicon sequencing. The largest source of variability in stool community composition was attributable to inter-individual differences regardless of delivery condition. Although the relative effect of delivery condition on community composition was small compared to inter-individual variability (1.6% vs. 60.5%, permutational multivariate analysis of variance [PERMANOVA]) and temporal variation within subjects over 10 weeks (5.2%), shifts in microbial taxa associated with delivery conditions were non-systematic and subject-specific. These findings indicated that it is not possible to model or accurately predict shifts in stool community composition associated with sampling logistics. Based on our findings, we recommend delivery of fresh, preservative-free stool samples to laboratories within 2 hr either at ambient or chilled temperatures to minimise perturbations to microbial community composition. In addition, subsamples from different fractions of the same stool displayed a small (3.3% vs. 72.6% inter-individual variation, PERMANOVA) but significant effect on community composition. Collection of larger sample volumes for homogenisation is recommended.