Ichthyophis asplenius, Taylor, 1965, Taylor, 1965

Host taxonomic View in CoL assessment

The Temburong specimens closely agreed in morphology with the holotype and paratopotype of I. asplenius by having a continuous lateral body stripe (present in I. asplenius , Figure 1 View Figure 1 (A)); no splenial teeth (absent in I. asplenius , Figure 1 View Figure 1 (C)); body folds and grooves not crossing the back on the anterior portion of the body, instead becoming complete approximately one-third down the length of the body (body folds complete approximately one-third down the length of the body, grooves complete about two-thirds down the length of the body in holotype of I. asplenius ; given as folds becoming ‘complete near the middle of body’ and grooves becoming complete ‘on latter fourth of body’ by Taylor 1965); having 240–272 body folds (247–270 in I. asplenius ); and 106–107 vertebrae (105–106 in I. asplenius ; Table 1 View Table 1 ). However , the Temburong specimens notably differed from the holotype and paratopotype of I. asplenius by having fewer teeth, with only 30–45 premaxillary-maxillopalatine teeth (52–55 in I. asplenius ), 36–43 vomeropalatine teeth (53–58 in I. asplenius ) and 35–40 dentary teeth (50 in I. asplenius ), as well as having a shorter tail bearing only a single body fold posterior to the vent (three folds in I. asplenius ; Taylor 1965; Table 1 View Table 1 ). The number of teeth in caecilians is known to change with age, with new teeth being added distally to tooth rows over time ( Taylor 1977). The Temburong specimens were generally similar in total body lengths (180–201 mm) to the holotype and paratopotype of I. asplenius (207 and 202 mm, respectively; Taylor 1965), but their ages cannot be discerned. Taylor’s method of manually examining teeth after sometimes removing concealing gingivae on one side ( Taylor 1977) would be more likely to undercount (miss) teeth than the CT scanning method employed here, and so differences in methods would not explain lower tooth counts in the Temburong specimens. Rather , the fewer tooth numbers observed in the Temburong specimens are attributed to ontogenetic variation, and the Temburong specimens are taxonomically referred here to I. asplenius .

In mt DNA, the Temburong specimens closely matched a series of eight specimens belonging to Clade N of Nishikawa, Matsui , Yong , et al. (2012) from Kuching, Sarawak, Malaysia and Tawau, Tenom, and Ulu Senagang , Sabah, Malaysia that those authors considered to be true I. asplenius (Nishikawa, Matsui, Yong, et al. 2012, p. 720). The Temburong and Clade N specimens of Nishikawa, Matsui, Yong , et al. (2012) exhibited maximum uncorrected pairwise (p) distances of 2.5% within the coding region of cytochrome b and 2.7% within the 16S ribosomal RNA gene.

Microbiome sampling

We collected 171,510 high-quality sequences representing 2,214 distinct ASVs from all three caecilian specimens across various body locations (n = 15). Five archaeal phyla and 43 distinct bacterial phyla were found to be present across all samples ( Figure 2 View Figure 2 ). Proteobacteria constituted the dominant percentage of microbial communities with an average 58% relative abundance. The next most common bacterial phyla were Firmicutes at 17%, Bacteroidota at 14%, Actinobacteriota at 6%, and Acidobacteriota at 0.6% relative abundance. The most abundant species included Acinetobacter baumannii, Empedobacter stercoris, Acinetobacter brisouii, Faecalibacterium prausnitzii and Streptococcus infantarius . As a comparison, in the caecilian soil environment (n = 3) we had 34,302 sequences representing the same number of ASVs but with the relative abundance of Acidobacteriota sequences (12%) higher than that of Bacteriodota (11%), Actinobacteriota (8%) and Firmicutes (4%). However, Proteobacteria was still the dominant phylum at 49% relative abundance. The most abundant bacterial species in the soil were VBCG01 sp005881895, Microbacter jiangxiensis , Pseudolabrys sp001426945, Acinetobacter brisouii , and Solibacter usitatus _A. The caecilian cutaneous microbiome community did not significantly differ by body location sampled for both presence and abundance of microbial taxa (PERMANOVA – Jaccard: R2 = 0.149, DF = 4, F-model = 0.4408, p -value = 0.284; Bray-Curtis: R2 = 0.113, DF = 4, F-model = 0.319, p -value = 0.244). There was, however, a significant difference of the microbial community between specimens (PERMANOVA – Jaccard: R2 = 0.561, DF = 2, F-model = 7.669, p -value = 0.001; Pairwise PERMANOVA p = 0.011 for all comparisons; Bray-Curtis: R2 = 0.684, DF = 2, F-model = 12.996, p -value = 0.001; Pairwise PERMANOVA p = 0.012 for all comparisons; Figure 2 View Figure 2 and Figure 3 View Figure 3 (A)). Soil samples differed from all skin samples independently of animal and body part (PC1 36.01% of explained variance; Figure 3 View Figure 3 (A)). We found 163 ASVs with different abundance between caecilian and soil samples ( Figure 3 View Figure 3 (B) and Supplementary File S1). A total of 160 of these ASVs increased their abundance in the caecilian skin. For bacterial sequences that matched those with known antifungal, particularly anti- chytrid, properties, 113 sequences from our overall dataset matched at 100% identity to those from the Antifungal Isolates Database (Supplementary File S2) and 37 of the 160 ASVs with an increased abundance on caecilian skin were also 100% identical (Supplementary File S3).