Dataset description
Twenty six female southern hairy-nosed wombats were captured, including 14 adult and 12 subadult animals, ranging in weight from 9.1 kg - 26.8 kg (Figure 1; SI table 1). To examine females and pouch young at different developmental stages, samples from 8 individuals and 1 joey were collected on the first trip in April 2019, 9 individuals and 1 joey were collected in August, and 9 individuals and 3 joeys were collected in October (Figure 1). According to head length measurements (SI table 1) and a growth rate equation estimated by [23]; Age = (Head length - 5.2) / (0.393 ± 0.26), the age of the joeys sampled were estimated to be ~168 days for the joey caught in April, ~28 days for the joey caught in August, and ~37, ~73 and ~78 days for the joeys caught in October.
A total of 104 microbial samples were collected from the 26 female SHNWs captured, including 26 pouch samples, 26 oral samples, 17 faecal samples, 17 skin samples, 9 joey samples (5 skin samples, 2 oral samples and 2 cloaca samples), 5 cloaca samples, and 4 milk samples.
A total of 130 samples were sequenced (104 biological, 26 negative control) on an Illumina MiSeq (2 x 150 bp), which yielded 12,313,465 paired end sequences, with an average of 94,718 reads per sample. After merging paired end reads and denoising with deblur (250 bp trim), 5,265 ASVs were identified across the dataset.
Determination of sample biomass and limit of detection using qPCR
Because 16S rRNA gene amplicon methods are susceptible to DNA contamination [15], it is important to determine the limit of detection by comparing biological samples to negative controls [16]. This is especially important when investigating new sample types such as the marsupial pouches in this study, where the microbial biomass has not been previously estimated. We tested whether the sample types collected in this study contained higher absolute amounts of microbial DNA compared to negative controls by using qPCR of the V4 region of the 16S rRNA gene. The faecal samples showed the lowest cycle threshold (ct) values, with the extraction blank control and sampling blank controls having the highest ct values (Figure 3). Milk, joey, and pouch samples had ct values lower than the negative controls, suggesting that biological samples contained microbial DNA at levels higher than the limit of detection of this study (ct of ~27). As expected, milk samples contained low quantities of DNA (ct values 17.6-24.4), as we were only able to collect ~3 drops per wombat. Overall, we conclude that SHNW pouch samples contain relatively high quantities of microbial DNA that is not a result of contamination from the sampling or laboratory environment.
Decontamination of dataset using decontam
To explore and remove potential contaminant taxa from our dataset, we exported the feature (amplicon sequence variant) table from QIIME2 into the phyloseq R package [41] and identified putative contaminants using the prevalence-based method in decontam [36]. This approach exploits the widely observed signature that contaminant taxa are likely to have higher prevalence in negative control compared to biological samples. The score statistic ranges from 0 to 1 and based on a histogram of decontam scores we chose a decontam score threshold of 0.5 (SI figure 2), which resulted in 60 features being classified as contaminants (SI figure 3, SI table 2). Using a threshold value of 0.5 will classify features as contaminants if they are present in a higher fraction of negative controls than biological samples. Taxa classified as putative contaminants include Acinetobacter, Bradyrhizobium, Pseudomonas, Ralstonia, and Sphingomonas (SI table 2), taxa that have previously been reported in the negative controls of multiple studies [16]. We filtered these putatively contaminant features from our feature table prior to subsequent analyses.
Microbial diversity and composition of Southern Hairy-nosed Wombats
To provide context for the pouch samples, we sought to characterize the microbial diversity and composition of microbes at the different body sites collected. The skin and subadult pouch samples contained the highest microbial diversity (~900 features) when compared to other sample types (Figure 4A, richness pairwise Kruskal-Wallis p-values < 0.05; SI table 3). As expected, skin and subadult pouch samples contained similar levels of microbial diversity (Pairwise Kruskal-Wallis p-value > 0.05; SI table 3) as subadult pouches are underdeveloped, open to the external environment, and generally resemble skin in appearance (Figure 2A). Faecal samples contained the next highest level of microbial diversity (~450 features), followed by cloacal (~100 features) oral samples (~50 features). The reproductive samples (adult pouch, milk, and joey) contained the lowest microbial diversity (<50 features). We found that subadult pouches contain significantly higher microbial diversity than adult pouches (Figure 4A; pairwise Kruskal-Wallis richness p-values < 0.001; SI table 3).
Analysis of microbial composition revealed that subadult pouch and skin samples clustered together (ANOSIM R = 0.003592, p-value = 0.425; Figure 4B). Faecal samples formed a tight cluster of their own, separated from the subadult pouch and skin/subadult pouch samples across PC3 (ANOSIM R = 0.818670, p-value = 0.001; Figure 4B). Faecal and skin/subadult pouch samples were separated from the oral and reproductive samples across PC1, which explained 33% of the variation (Figure 4B; see SI tables 4 & 5 for pairwise ANOSIM and permdisp tests). The oral and reproductive samples were separated along PC2 (ANOSIM R=> 0.75, p-values < 0.002; SI table 4; Figure 4B). Overall, we found statistically significant differences in microbial composition between body sites, with reproductive samples (adult pouch, milk, joey) clustering separate to other body sites sampled.
Taxonomically, we found differences in the microbial communities between the body sites sampled. At the phylum level, faecal samples were dominated by Firmicutes (58.4%), Bacteroidota (19.4%), and Spirochaetota (14%) (Figure 5). Within faecal samples, the most dominant families were Christensenellaceae (17.6%), Spirochaetaceae (13.9%), Oscillospiraceae (10%), Rikenellaceae (7.4%), and Lachnospiraceae (6.1%) (SI file 1). The Firmicutes to Bacteroidetes ratio in faecal samples was calculated to be 3.1:1 (SD=1.1). In oral samples, the most abundant phyla were Proteobacteria (55.5%), Firmicutes (25.5%), and Actinobacteria (11.6%), with the most dominant families being Pasteurellaceae (26.5%), Streptococcaceae (16.3%), Moraxellaceae (14%), Neisseriaceae (9.4%), and Micrococcaceae (6.6%) (SI file 1). The most abundant phylum for reproductive samples (milk, adult pouch, joey) was Actinobacteriota(81.7%-90.6%; Figure 5), with the most abundant families being Brevibacteriaceae (20.9%-36.5%), Corynebacteriaceae (12.4%-22.2%), Microbacteriaceae (15%-21.1%), and Dietziaceae (8.6%-19%) (SI file 1). For the full taxonomy, both collapsed by sample type and at the individual sample level, see SI files 1 and 2.
Effect of female reproductive state on pouch microbial communities
Given that the SHNW pouch undergoes both morphological and physiological changes throughout the reproductive cycle (Figure 2), we next sought to test the hypothesis that changes in the reproductive cycle of female SHNWs influence the microbial diversity and composition of the pouch. To examine this, we filtered the feature table to only include pouch samples and their corresponding skin samples as controls. The data indicated that microbial diversity declined from reproductively inactive wombats (anoestrus, post-lactation, subadult) to reproductively active animals (cycling, lactating) (Figure 6A). These differences were statistically significant for cycling and lactating vs. subadult wombats (richness pairwise Kruskal-Wallis p-values <0.002 SI table 6), although due to insufficient sample size we were unable to statistically test anoestrus (n=2) and post-lactation females (n=1). For microbial composition, we observed a similar trend across PC1 (42% of variation) that corresponded to female reproductive status (Figure 6B), with cycling and lactating animals having statistically significant differences in microbial composition compared to subadult and skin samples (Unweighted UniFrac ANOSIM R values =>0.91, p-values 0.001; see SI tables 7 & 8 for all pairwise ANOSIM and perdisp comparisons). Taxonomically, the cycling and lactating pouch samples were dominated by the phylum Actinobacteriota, which contained five taxa that accounted for >90% of the total relative abundance: Corynebacterium, Brevibacterium, Dietzia, Microbacteriaceae, and Helcobacillus (Figure 6C).
Characterisation of milk and joey samples
Next, we explored whether the age of pouch young influenced the microbial composition detected in the pouch or on the joey, and whether milk samples contained a distinct microbial signature. Samples were collected from five animals (trip 1: animal 11, trip 2: animal 99, trip 3: animals 207, 208, and 209) that had a pouch young with estimated ages of ~168, ~28, ~37, ~73 and ~78 days, respectively. The weights of the joeys ranged from ~16 to 500 grams. Taxonomically, milk and joey samples resembled their corresponding pouches (Figure 7). However, the taxonomic composition of microbes from family 11 (i.e. animal 11 and the corresponding joey samples) obtained from the most mature joey sampled, differed from the other, less mature joey families (Figure 7). Samples from family 11 had a higher relative abundance of Dietzia and Horanjiania, and a lower relative abundance of Microbacteriaceae. The joey oral and milk samples from this animal both contained a Neisseriaceae feature that was also found in the oral sample taken from the mother (Figure 7; SI figure 5). The joey cloaca sample also contained a higher abundance of Escherichia-Shigella, Enterobacteriaceae, and Enterococcus.
Finally, we sought to better classify the five most abundant microbes (accounting for >90% of the relative abundance) in female wombats from cycling/lactating pouch, milk, and joey samples. We first created a heatmap of features from all cycling/lactating pouches, milk, and joey samples with minimum frequencies of 100 to determine the exact features that dominated these samples (SI figure 4). We then used BLAST (against the NCBI nt database) to find the closest reference alignment of these features (Table 1). Three of the five top features had best (or second-best for the g__Helcobacillus; s__uncultured_bacterium feature) hits to uncultured 16S rRNA gene clones isolated from the pouches of Tammar wallabies. One of the features, which was only classified to the family Microbacteriaceae, was highly divergent from any reference (239/250 nucleotide matches).
FeatureID
|
SILVA_138 QIIME2 classified taxonomy
|
Sequence identity
|
Best hit
|
53fe3d733108e7d7b71644dd4f51b8b8
|
g__Brevibacterium; s__uncultured_bacterium
|
248/250
|
No best hit, multiple hits to Brevibacterium spp. and uncultured prokaryotes.
|
9532794e312626243216651ea0765320
|
g__Dietzia
|
247/250
|
Uncultured pouch clone 1530-P-3B from Chhour et al. 2010.
|
6fba0a71cfb9da18a3b61c9d3090de55
|
f__Microbacteriaceae
|
239/250
|
No best hit, multiple hits to Gulosibacter spp.
|
c0a3a4e78ac14be9f65454f6e10e163e
|
g__Corynebacterium
|
248/250
|
5 best hits to uncultured pouch clones from Chhour et al. 2010.
|
91548fb32f63077f9653391f4f1205a0
|
g__Helcobacillus; s__uncultured_bacterium
|
247/250, 246/250
|
Best hit (247/250) to a Hyena anal pouch clone, 3 second-best hits (246/250) to uncultured pouch clones from Chhour et al. 2010.
|
Table 1 | BLAST results from the top five most abundant features in the cycling/lactating pouch, joey, and milk samples. Two features have best hits, and one feature has multiple second-best hits to bacterial 16S clones isolated from Tammar wallaby pouches.