Common methods for fecal sample storage in field studies yield consistent signatures of individual identity in microbiome sequencing data

Abstract

Field studies of wild vertebrates are frequently associated with extensive collections of banked fecal samples-unique resources for understanding ecological, behavioral, and phylogenetic effects on the gut microbiome. However, we do not understand whether sample storage methods confound the ability to investigate interindividual variation in gut microbiome profiles. Here, we extend previous work on storage methods for gut microbiome samples by comparing immediate freezing, the gold standard of preservation, to three methods commonly used in vertebrate field studies: lyophilization, storage in ethanol, and storage in RNAlater. We found that the signature of individual identity consistently outweighed storage effects: alpha diversity and beta diversity measures were significantly correlated across methods, and while samples often clustered by donor, they never clustered by storage method. Provided that all analyzed samples are stored the same way, banked fecal samples therefore appear highly suitable for investigating variation in gut microbiota. Our results open the door to a much-expanded perspective on variation in the gut microbiome across species and ecological contexts.

Figure 1. Storage effects on alpha diversity.

(a) Shannon’s Diversity Index (SDI) values (y-axis) shown as a function of storage method (x-axis), with each individual plotted in a different color. Lyophilized samples have lower SDI values than other storage methods (Tukey’s HSD, lyophilized-frozen: p = 0.004; lyophilized-RNAlater: p = 7.6 × 10⁻⁵; lyophilized-ethanol: p = 0.063; Table S2). (b) SDI values are significantly correlated within individuals, between all storage methods (Pearson’s correlation, p < 0.05). Each dot represents an individual, and each panel shows the correlation between SDI values obtained from two different storage methods.

Figure 2. Storage effects on beta diversity.

(a) Bray-Curtis dissimilarity values (y-axis) comparing the same individual from samples collected under different storage conditions (red), different individuals with samples collected under the same storage conditions (blue), and different individuals with samples collected under different storage conditions (green). Median Bray-Curtis dissimilarity calculated from subsampling reads from the same sample (i.e., the minimum dissimilarity due to read resampling alone) is indicated by the gray dashed line. Because of the large number of data points, all pairwise comparisons are highly significant (Wilcoxon Rank Sum test, p < 1 × 10⁻⁹). However, the dissimilarity values for same individual/different storage are much lower on average (mean = 0.35 ± 0.11 s.d.) than dissimilarity values measured between individuals in either the same (mean = 0.51 ± 0.11 s.d.) or different (0.56 ± 0.11 s.d.) storage conditions. (b) Bray-Curtis dissimilarities cluster more strongly by individual (colors along the left-hand sidebar, with one color per individual) than by storage method (colors shown on the top, next to the dendrogram, and in the boxed legend).