Study System
Sable Island is an oceanic sandbar, located 175 km off the eastern coast of Nova Scotia, Canada (Figure 1). Since the 18thcentury, the island has been home to a population of free-living horses, which have remained largely unmanaged since their introduction. Importantly, no individual of the population has ever been administered antibiotics . Using 16S rRNA gene amplicon sequencing, we previously demonstrated evidence for dispersal limitation in horse-associated gut microbiota across the island’s narrow (49 km) length We also characterized spatially structured variation in the microbiome, which we hypothesized is caused by dietary variation.
Marram grass (Ammophila breviligulata ) comprises the most abundant component of the Sable Island horse diet across the island. However, the semi-succulent plant sea sandwort (Honckenya peploides ), only found at the eastern and western extremes of Sable Island is preferred by the horses when present in their home-range. To generate coarse predictions of diet, we previously estimated the relative coverage of major Sable Island plant community types within 150-metre radius buffers centred on the point of fecal sample collection (Stothart et al., 2021). The radius of this spatial buffer is proportional to the median daily movement observed in the horses. Therefore, plant community composition within these spatial buffers are predicted to be coarsely indicative of the forage to which horses had access, during the 24 hours preceding defecation. Using these estimates, we observed greater diversity of amplicon sequence variants (ASVs) and greater phylogenetic clustering in the microbiome of horses without sea sandwort within their spatial buffer . Conversely, lower bacterial diversity and greater evidence for stochasticity was observed in the microbiome of horses with access to sea sandwort. We hypothesized that sandwort did not support fibrolytic niches in the gut microbiome to the same extent as marram grass. However, we were unable to test this hypothesis, as 16S amplicon data do not allow for direct estimates of microbiome functional potential.