Great knot faecal samples were largely comprised of Bacilli, particularly Mycoplasmataceae and Erysipelotrichaceae, while communities in pipis were dominated by spirochaetes, and sand samples had more evenly diverse communities with many bacterial classes (Fig. 1a). Among other birds, Firmicutes and Proteobacteria [3, 7] are the dominant bacterial phyla in gut communities. Both phyla had high relative abundance in our pre-migration samples, though Firmicutes (Bacilli) replaced other taxa after migration.

Bacterial communities differed among sample types but not between sampling periods. a Barplots of the major bacterial classes in field samples, including classes accounting for > 3% of the relative abundance in any sample. b ASV richness was greatest in sand samples and lowest in faecal samples (Tukey’s HSD, P < 0.0001 each). c Bray–Curtis principal coordinates analysis revealed no clustering in faecal samples and strong clustering in the other sample types. d Bray–Curtis network (maximum distance = 0.4) supports minimal connectivity of great knot faecal samples with their food or the environment. Eight samples were collected per sample type per sampling period, though three faecal samples and one sand sample failed during library prep and sequencing

We found differences among the sample types in both alpha and beta diversity metrics. In alpha diversity, sand samples had significantly greater richness than the other two sample types, while faeces had the lowest richness (Fig. 1b; Contents: F2,38 = 1080.56, P < 0.0001). Beta diversity also differed among the sample content types (Fig. 1c; Contents: F = 17.83, R2 = 0.46, P = 0.001), though pipi samples had differing beta dispersion from the other sample types. Pipi samples clustered tightly in principal coordinates space, while faecal samples were spread quite broadly. In faecal samples, low richness could be due to low resident biomass, with faecal microbes mainly associated with foodborne transient taxa, while dispersion of the faecal samples in principal coordinates space supports previous results of low phylogenetic signal in bird gut communities ([6], but see [9]). Lewis et al. [10] found that changing environment and food source can quickly alter gut microbes of passerines, and Dion-Phénix et al. [11] also found overlap in gut microbes in blue tits and their available food source. However, had the great knots’ faecal bacteria been strongly affected by their food source in this study, we would expect connectivity between sample types in network analysis. In our samples, only the pipis showed connectivity in network analysis (Fig. 1d).

In comparisons between the two sampling time points, we found similar richness (Sampling: F1,38 = 1.697, P = 0.2; Contents × Sampling: F2,38 = 0.514, P = 0.6) and beta diversity (Fig. 1c; Sampling: F = 1.63, R2 = 0.02, P = 0.09; Contents × Sampling: F = 1.42, R2 = 0.04, P = 0.1) across time. The lack of a seasonal component suggests that the communities are relatively stable across time, especially in pipis, which have strong connectivity in network analysis. In the birds, statistical similarity over time may be a consequence of their quick digestion and low microbial retention, resulting in varied communities among individuals in general. Studies of other migratory birds have found that, though bird gut microbiomes differ between breeding and overwintering ranges, the communities can shift within the first day of returning to a site [10, 31]. Consequently, apart from migratory birds temporarily retaining key bacterial taxa perhaps linked to energy storage, their gut communities can become similar to resident conspecifics within a few days of returning from migration [32]. We might expect that local food source should drive such rapid community changes, but the absence of an overlap between the bird and pipi samples suggests that not just dietary shifts, but other concurrent factors (e.g., physiological and environmental) may also contribute to these shifts.

Regarding the initial hypothesis of greater connectivity before than after migration, we found that relaxing the maximum distance for the network analysis actually showed stronger grouping among the post-migration samples. Along with the relative increase in Bacilli (or decrease in other microbes) after migration, these results could be because physiological stress due to migration had altered the gut bacteria. Exercise and stress can affect gut microbiomes [33] and the birds sampled in September 2022 had recently flown several thousand kilometers, as indicated by their visibly poorer body condition. Our results suggest that migration status and low microbial specificity greatly affected the great knot gut microbiomes represented in the faecal samples collected after the birds returned.

All three sample types contained potentially pathogenic Vibrio spp., including V. campbellii and V. owensii (emerging aquatic animal pathogens), and V. parahaemolyticus (a human pathogen) (Fig. 2). Pipi and sand samples diverged in principal coordinates space, but we found minimal connectivity within the sample types, and no connectivity in Vibrio communities between sample types. Interestingly, the three bird samples with detectable Vibrio each had a single different dominant species, which could be a sign of one species outcompeting the others in the gut, or differential colonisation success. One bird sample was dominated by V. xiamenensis, which was not detected in either sand or pipis.

Faeces, pipis, and sand have diverse Vibrio species with minimal community connectivity. a Barplots of Vibrio spp. b Bray–Curtis principal coordinates analysis. c Bray–Curtis network analysis (maximum distance = 0.4)

Though the data presented here stem from DNA samples, in a pilot study (February 2022) we cultured samples on two Vibrio growth media, confirming that great knot faeces (as well as the two other sample types) contain living Vibrio spp. (unpublished data). Elsewhere, pathogenic V. parahaemolyticus has been cultured from faecal samples of other waterbirds [34]. As migratory birds are implicated in the spread of multiple human pathogens by other means (e.g., as biological carriers through infection, as transport of pathogenic spores or pathogen-carrying ectoparasites; [2]), their capacity to spread bacteria, including potential pathogens, via their gastrointestinal tract warrants further research. In particular, we hypothesise that animals may act as vectors for Vibrio movement via trophic interactions and migration in addition to the known dispersal route of warm currents.

The identification of Vibrio spp. in this study was based on a sequence database from 2019 [20]. The high relative abundance of unassigned Vibrio ASVs highlights the limitations of our knowledge of the genus. Most of the unassigned reads were found in the pipi samples (Fig. 2a), revealing tropical invertebrates and their bacterial symbionts as an understudied system deserving more focus.