Re-examination of a published sample-set using shallow shotgun
metagenomics
Amplicon and shotgun metagenomic sequencing painted similar portraits of
the Sable Island horse microbiome, but we sought to determine whether
shallow shotgun sequencing would recapitulate results first described in
an amplicon dataset . A primary finding in this amplicon dataset—a
clear beta-diversity by spatial distance relationship—was present (but
weaker) in the shallow shotgun dataset. Although partial-length 16S
amplicon reads cannot reliably be assigned to species, and frequently
not to genus, grouping reads to ASVs can still provide a high-resolution
picture of the microbiome. In fact, this approach can sometimes be too
sensitive, separating 16S rRNA gene reads derived from the same
bacterial genome, into separate ASV bins By contrast, the read-based
classification of shotgun reads is constrained by taxonomic demarcations
recorded in reference databases. Therefore, amplicon sequencing—or
deep shotgun sequencing which allows for de novo MAG assembly and
strain-resolved tracking—might be better suited than shallow shotgun
sequencing for characterizing relationships in the microbiome which are
mechanistically driven by microbiota dispersal.
The assembly of Sable Island horse specific MAGs could help increase the
resolution of shallow shotgun read classification, however, increased
taxonomic resolution is not always useful, since it can obscure
biological patterns or cause spurious correlations if explanatory
variables are spatially structured; like sea sandwort on Sable Island .
We observed that horse access to sea sandwort was correlated with
microbiome beta-diversity differences in both the amplicon and shallow
shotgun metagenomic datasets. Although the taxa identified by
differential abundance were not identical, both broadly support the same
biological conclusion. Namely, that marram grass better supports
fibrolytic niches in the microbiome than sea sandwort. In the 16S
amplicon dataset, plant fibre specialist bacteria were more abundant
when sandwort was absent (Fibrobacteraceae , ;Spirochaetaceae , ), while simple sugar fermenters were more
abundant when horses had access to sea sandwort
(Christensenellaceae , ). Similarly, in the shallow shotgun
metagenomic dataset, plant fibre specialists were more abundant when
sandwort was absent (Fibrobacteraceae and a marginally
non-significant trend in Spirochaetaceae ) while a functionally
diverse clade of plant, host, and alpha glycan foragers,Muribaculaceae , were more abundant when sea sandwort was
present.
Interestingly, sea sandwort was associated with bacterial families which
contain members capable of degrading the plant toxin oxalate, in both
the 16S amplicon (Burkholderiaceae , Hervé, Junier, Bindschedler,
Verrecchia, & Junier, 1982), and shallow shotgun datasets
(Muribaculaceae , ; Oxalobacteraceae , ). Despite
discrepancies in the taxa implicated, data from both sequencing methods
suggest sea sandwort might be a source of oxalate in the Sable Island
horse diet. Conventional analysis of 16S rRNA gene amplicon data would
stop at this point; using taxon differential abundance results to make
ecological inferences.