Study 2: How does sampling methodology (swab vs feces) affect recovered community composition?
Shannon diversity index values were significantly higher in fecal pellets than in pre-defecation cloacal swabs (t = 16.29, df = 7, p < 0.001; Fig 4a), as was richness (t = 8.39, df =,7, p < 0.001; Fig 4b) and phylogenetic diversity (F = 10.94, df = 7, p < 0.001; Fig 4c). Beta diversity analysis using Bray-Curtis distance showed that the microbial composition of fecal pellets was significantly more similar to other fecal pellets than it was to that of the pre-defecation cloacal swabs (F = 9.65, df = 1,14, p = 0.002; Fig 5a). The groups were also dispersed differently (F = 5.54, df = 1,14, p = 0.034). Fecal pellets also differed from pre-defecation cloacal swabs in both UniFrac metrics (weighted: F = 18.91, df = 1,14, p = 0.001, Fig 5b; unweighted: F = 15.13, df = 1,14, p = 0.001, Fig 5c), and dispersion of the communities also varied for each measurement (weighted: F = 13.83, df = 1,14, p = 0.002; unweighted: F = 4.78, df = 1,14, p = 0.046).
There was no significant difference between pre- and post-defecation swabs in any alpha diversity metric (Shannon: t = 1.40, df = 7, p = 0.204; Richness: t = 1.45, df = 7, p = 0.190; Phylogenetic diversity: t = 1.72, df =,7, p = 0.129; Fig 4), although the post-defecation swab tended to have higher diversity. Swab types did not cluster separately based on Bray-Curtis distances (F = 1.24, df = 1,14, p = 0.231, Fig 5a), and the groups were dispersed similarly (F = 2.57, df =1,14, p = 0.131). Pre-defecation swab communities differed from post-defecation swab communities in dispersion based on weighted (F = 7.08, df = 1,14, p = 0.019, Fig 5b) and unweighted (F = 36.14, df = 1,14, p < 0.001, Fig 5c) UniFrac distances, as well as composition of unweighted UniFrac distance (F = 2.20, df = 1,14, p = 0.035), although this could be due to differences in dispersion rather than true clustering of groups. The different swab types did not cluster separately based on weighted UniFrac distance (F = 1.14, df = 1,14, p = 0.281).
The most abundant taxa recovered from fecal pellets was Lachnospiraceae, which made up 38.7 ± 5.2% on average (Fig 6). Enterobacteriaceae, Bacteroidaceae, andRuminococcacea each accounted for ~10% of the recovered communities, and several others made up between 1-5% of the recovered community including Tannerellaceae, Erysipelotrichaceae, Eggerthellaceae, Marinifilaceae, and Akkermansiaceae.Pre-defecation cloacal swab communities were dominated byEnterobacteriaceae (84.9 ± 5.3%), and the next most abundant taxa was Helicobacteraceae, which only accounted for 5.1 ± 2.2% of the community. A corncob analysis confirmed thatLachnospiraceae was significantly more abundant (t = -7.20, p < 0.001) and Enterobacteriaceae was significantly less abundant (t = 3.83, p < 0.001) in fecal samples compared to pre-defecation cloacal swabs.
While the same two taxa dominated the pre- and post-defecation swabs, post-defecationEnterobacteriaceae was reduced to only 44.8 ± 13.2%, andHelicobacteraceae increased to 19.0 ± 9.5%. There was also an increase in the abundance of feces-associated taxa, particularlyLachnospiraceae which accounted for 15 ± 7.4% on average. However, there was a distinct bifurcation in the post-defecation swab samples. In five samples, Enterobacteriaceae andHelicobacteraceae account for >75% of the community, similar to the pre-defecation swabs, while the remaining three samples had a more “feces-like” community, with a higher abundance of Lachnospiraceae and Bacteroidaceae .