Microbiota fingerprints within the oral cavity of cetaceans as indicators for population biomonitoring

Abstract

The composition of mammalian microbiota has been related with the host health status. In this study, we assessed the oral microbiome of 3 cetacean species most commonly found stranded in Iberian Atlantic waters (Delphinus delphis, Stenella coeruleoalba and Phocoena phocoena), using 16S rDNA-amplicon metabarcoding. All oral microbiomes were dominated by Proteobacteria, Firmicutes, Bacteroidetes and Fusobacteria bacteria, which were also predominant in the oral cavity of Tursiops truncatus. A Constrained Canonical Analysis (CCA) showed that the major factors shaping the composition of 38 oral microbiomes (p-value < 0.05) were: (i) animal species and (ii) age class, segregating adults and juveniles. The correlation analysis also grouped the microbiomes by animal stranding location and health status. Similar discriminatory patterns were detected using the data from a previous study on Tursiops truncatus, indicating that this correlation approach may facilitate data comparisons between different studies on several cetacean species. This study identified a total of 15 bacterial genera and 27 OTUs discriminating between the observed CCA groups, which can be further explored as microbiota fingerprints to develop (i) specific diagnostic assays for cetacean population conservation and (ii) bio-monitoring approaches to assess the health of marine ecosystems from the Iberian Atlantic basin, using cetaceans as bioindicators.

Figure 1

Comparison of the oral microbiomes of cetaceans according to the species of the sampled animals. Panel (a) shows the Canonical Correspondence Analysis (CCA) (p-value = 0.001) which was performed after subsampling the OTU table to even sequencing depth and at the bacterial species level, with Hellinger transformation of the abundances. Similar results were obtained when grouping the OTUs by bacterial genera. The colour frames in both CCA plots group the samples according to the constrained variable. The major contributions of the “host species” variable are shown as % in the first and second component of the CCA plot (CCA1 and CCA2, respectively). Panel (b) shows the richness and diversity measures calculated between species (as the average observed OTUs and Shannon diversity index of each sample), using the OTU table normalized by total sum scaling. Significance of these alpha-diversity metrics was tested with Kruskal-Wallis chi-squared test followed by pairwise Wilcoxon test between groups (n.s, statistically not significant). Panel c shows the number of total OTUs in each group and the number of OTUs shared between groups. Delphinus delphis (n = 18), Phocoena phocoena (n = 10) and Stenella coeruleoalba (n = 10).

Figure 2

Comparison of the oral microbiomes of cetaceans according to the development stage/age class. The Canonical Correspondence Analysis (CCA) in panel (a) (p-value = 0.001) was performed after subsampling the OTU table to even sequencing depth and at the bacterial species level, with Hellinger transformation of the abundances. Similar results were obtained when grouping the OTUs by bacterial genera. The colour frames in both CCA plots group the samples according to the constrained variable and the symbols represent the species of the sampled animals: D. delphis (●), P. phocoena (▲) or S. coeruleoalba (■). The major contributions of the “development stage/age class” variable are shown as % in the first and second component of the CCA plot (CCA1 and CCA2, respectively). Panel (b) shows the richness and diversity measures calculated between species (as the average observed OTUs and Shannon diversity index of each sample), using the OTU table normalized by total sum scaling. Significance of these alpha-diversity metrics was tested with Kruskal-Wallis chi-squared test followed by pairwise Wilcoxon test between groups (a, p-value = 0.008; b, p-value = 0.002; c, p-value = 0.001; d, p-value = 0.002; e, p-value = 0.001). Panel c shows the number of total OTUs in each group and the number of OTUs shared between groups. Adult (n = 14), Subadult (n = 12) and Juvenile animals (n = 12).

Figure 3

Comparison of the oral microbiomes of cetaceans according to the stranding location of the specimen. The Canonical Correspondence Analysis (CCA) in panel (a) (p-value = 0.046) was performed after subsampling the OTU table to even sequencing depth and at the bacterial species level, with Hellinger transformation of the abundances. Similar results were obtained when grouping the OTUs by bacterial genera. The colour frames in both CCA plots group the samples according to the constrained variable and the symbols represent the species of the sampled animals: D. delphis (●), P. phocoena (▲) or S. coeruleoalba (■). The major contributions of the “stranding location” variable are shown as % in the first and second component of the CCA plot (CCA1 and CCA2, respectively). Panel (b) shows the richness and diversity measures calculated between species (as the average observed OTUs and Shannon diversity index of each sample), using the OTU table normalized by total sum scaling. Significance of these alpha-diversity metrics was tested with Kruskal-Wallis chi-squared test followed by pairwise Wilcoxon test between groups (n.s., statistically not significant). Panel c shows the number of total OTUs in each group and the number of OTUs shared between groups. Animals sampled in the northern Atlantic Iberian coast (n = 11) and animals sampled in the western Atlantic Iberian coast (n = 27).

Figure 4

Comparison of the oral microbiomes of cetaceans according to their cause of death. The Canonical Correspondence Analysis (CCA) in panel (a) (p-value = 0.013) was performed after subsampling the OTU table to even sequencing depth and at the bacterial species level, with Hellinger transformation of the abundances. Similar results were obtained when grouping the OTUs by bacterial genera. The colour frames in both CCA plots group the samples according to the constrained variable and the symbols represent the species of the sampled animals: D. delphis (●), P. phocoena (▲) or S. coeruleoalba (■). The major contributions of the “cause of death” variable are shown as % in the first and second component of the CCA plot (CCA1 and CCA2, respectively). Panel (b) shows the richness and diversity measures calculated between species (as the average observed OTUs and Shannon diversity index of each sample), using the OTU table normalized by total sum scaling. Significance of these alpha-diversity metrics was tested with Kruskal-Wallis chi-squared test followed by pairwise Wilcoxon test between groups (n.s., statistically not significant). Panel c shows the number of total OTUs in each group and the number of OTUs shared between groups. Sampled animals resulting from bycatch (n = 27) and stranded diseased animals (n = 11).

Figure 5

CCA of the oral microbiomes of the 4 cetacean species, according to their health status. Canonical Correspondance Analysis of the oral microbial communities of cetaceans sampled in this study and by Bik et al., constrained according to the health status of the sampled animals (p-value = 0.001). The color frames in the plot group the samples according to the constrained variable, comprising 3 groups: dead animals derived from bycatch (n = 27), stranded diseased animals (n = 11) and healthy animals sampled by capture and release procedure (n = 25) in Bik et al.. Both ordinations were performed after subsampling the OTU tables to even sequencing depth of 1019 sequences and at the genus level, with Hellinger transformation of the abundances. In both panels, the species of the sampled animals are represented by symbols: D. delphis (●), P. phocoena (▲) or S. coeruleoalba (■) or T. truncatus (♦). The major contributions of the “health status” variable are shown as % in the first and second component of the CCA plot (CCA1 and CCA2, respectively).

Figure 6

Core OTUs among the different cetacean species (a) and present in all 38 samples (b). In panel (a), the number of OTUs comprising the core microbiome of each group is represented by the OTUs present in all respective samples. In panel (b), the taxa abundance of the OTUs present in all samples is shown as the average relative frequency for each genera (or family when a genus was not assigned).

Figure 7

Bacterial genera and OTUs with different relative abundance between the groups of variables under study. The significance of potential microbial fingerprints (highlighted with “*”, p-value < 0.05) were identify by the Indicator Species Analysis and the linear discriminant analysis (LDA) effect size algorithm (LEfSe), detailed in Supplementary Table S6. Only the taxa showing significant differential abundance at the genus and OTU levels by both methods were considered. For representation and due to a broad range of values, the relative abundances were Z-scaled to highlight their comparisons between groups. Taxonomic validation of the representative sequences of the selected OTUs was performed with BLASTN analysis against the NCBI non-redundant nucleotide database.