Captivity reduces diversity and shifts composition of the Brown Kiwi microbiome

Background Captive rearing is often critical for animals that are vulnerable to extinction in the wild. However, few studies have investigated the extent to which captivity impacts hosts and their gut microbiota, despite mounting evidence indicating that host health is affected by gut microbes. We assessed the influence of captivity on the gut microbiome of the Brown Kiwi (Apteryx mantelli), a flightless bird endemic to New Zealand. We collected wild (n = 68) and captive (n = 38) kiwi feces at seven sites on the north island of New Zealand. Results Using bacterial 16 S rRNA and fungal ITS gene profiling, we found that captivity was a significant predictor of the kiwi gut bacterial and fungal communities. Captive samples had lower microbial diversity and different composition when compared to wild samples. History of coccidiosis, a gut parasite primarily affecting captive kiwi, showed a marginally significant effect. Conclusions Our findings demonstrate captivity’s potential to shape the Brown Kiwi gut microbiome, that warrant further investigation to elucidate the effects of these differences on health. Supplementary Information The online version contains supplementary material available at 10.1186/s42523-021-00109-0.

by protozoan Eimeria spp., is a common disease in captive kiwi [13]. However, the relationship between captivity, coccidia, and gut microbial communities has garnered little attention. We sought to compare gut bacteria and fungi between captive and wild kiwi. We tested the hypothesis that captivity status and history of coccidiosis would decrease diversity and modify composition of the gut microbiome.

Results
Fresh fecal samples were collected from seven sites on the north island of New Zealand (Fig. 1a, Supplementary Table 1) during January -April 2019. Bacterial 16S rRNA (V4 region) [14] and fungal ITS genes [15] were ampli ed using isolated DNA from captive (n = 38) and wild kiwi fecal samples (n = 68).
To determine whether captivity in uences kiwi gut microbiota, we used PERMANOVA and NMDS analyses. Bacterial communities clustered by captivity across spatially independent sites with little overlap between the ellipses (Fig. 1b, PERMANOVA, r 2 = 0.07, p = 0.001). Captivity signi cantly decreased alpha diversity of bacteria ( Fig. 2a, ANOVA, p < 0.005) and fungi (Supplementary Fig. 1, ANOVA, p = 0.012) by 33% and 74% respectively. To assess the spread of variation among kiwi microbiomes in captive and wild treatments, we calculated distance to centroid, a metric for beta diversity. No discernible pattern was observed for bacteria ( Supplementary Fig. 2a, ANOVA, p = 0.948), but a marginally signi cant pattern was detected in fungal communities ( Supplementary Fig. 2b, ANOVA, p = 0.051). We also tested if site (a factor nested within captivity status) and history of coccidiosis (positive or negative) had an in uence on variation in microbial communities using PERMANOVA (Supplementary Table 2). Site showed a signi cant effect on bacteria (r 2 = 0.129, p = 0.001) and fungi (r 2 = 0.183, p = 0.001). Coccidiosis history showed a marginally signi cant trend with bacteria ( Fig. 2b, r 2 = 0.048, p = 0.095) and fungi (r 2 = 0.074, p = 0.087). Although bacterial phyla composition was variable within and across treatments, Firmicutes was more prevalent in wild kiwi, while Proteobacteria dominated in captive kiwi (Fig. 2c).
We conducted a simper analysis [16] to determine the most in uential OTUs that differentiate captive and wild kiwi samples for both bacteria and fungi. Eighty-seven bacterial OTUs and fteen fungal OTUs accounted for about 70% of the differences between wild and captive samples (Supplementary Tables 3   and 4). Using a multinomial species classi cation method (clamtest) [16] we categorized OTUs into four classes: rare, generalist, wild specialist, and captive specialist. For bacterial OTUs, 10% were classed as generalist, 53% as rare, 20% as wild specialist, and 17% as captive specialist (Fig. 3a, Supplementary Table 5). For fungal OTUs, 0% were classed as generalist, 47% as rare, 27% as wild specialist, and 27% as captive specialist (Fig. 3b, Supplementary Table 5).

Discussion
Our results indicate that captivity changes both bacterial and fungal communities in the Brown Kiwi gut.
Bacterial composition clustered by captivity (Fig. 1b), suggesting that kiwi from the wild are more similar to each other than their captive counterparts, even across geographically distinct sites. The consequences of reduced microbial diversity between wild and captive kiwi remain unclear, but several studies have linked dysbiosis to higher disease prevalence in a variety of animals from humans to corals [17,18]. Coccidiosis history showed a marginally signi cant effect. However, these results may be affected by small sample size. Our results suggest a potential link between changes in the microbiome to disease states that requires further exploration.
The shift in dominant bacterial phyla, Firmicutes to Proteobacteria, from wild to captive samples may be caused by sterile captive facilities and probiotic supplementation, a common practice particularly after antibiotics. Frequent surface disinfection [10] and probiotic treatment [19] have been shown to increase Proteobacteria in human subjects. Lactobacillus (OTU 49), a common genus in probiotics, was grouped as a captive specialist. Other captive-associated taxa include Corynebacterium (OTU 62), which has been found in the cloaca of penguins and the preen gland of turkeys [20], and Bacteroides (OTU 544), normally found in animal hosts but can include potential pathogens [21]. One wild specialist was Bacillaceae (OTU 101), a family associated with soil and plant pathogen defense [22], suggesting kiwi are ingesting environmental microbes when foraging [23]. Blautia (OTU 290) is a genus found in the human gut and associated with visceral fat accumulation [24]. Faecalitalea cylindroides (OTU 687), a butryrate producing microbe, has been detected in chicken [25]. These taxa may be indicative of nutrient acquisition in the wild, where food may be less available.
No fungal OTUs were categorized as generalists suggesting fungi in kiwi re ect their local environment. Some captive specialists include Cladosporium (OTU 151) and Aureobasidium (OTU 2), both associated with indoor environments and plant material [26,27], implicating the contribution of added soil and ferns to enclosures. Trichosporon (OTU 171), another captive specialist, is a common human skin taxa [28], suggesting close human interaction may shape kiwi fungi. One wild specialist, Rhizopogon (OTU 159), has been identi ed as a dietary component of small mammals, suggesting kiwi may be consuming and dispersing these fungi [29]. Preussia (OTU 181) and Saitozyma podzolica (OTU 37), both associated with soil and litter, were grouped as wild specialists [30,31].

Conclusions
In captivity, factors that shape gut microbial communities may include arti cial diet, sterile built environments, and human interaction [6,10]. A follow-up study investigating the establishment of gut microbes throughout kiwi development with sampling of the captive environment can elucidate how factors inherent to captivity contribute to the kiwi gut microbiome. Overall, our data suggest that captivity in uences the gut microbiome of the Brown Kiwi with potential for health and disease assessment for captive-reared individuals.

Study system
Captive The National Kiwi Hatchery is located at the Rainbow Springs Nature Park in Rotorua, New Zealand. It is the leading facility in kiwi husbandry, egg incubation, and kiwi rearing.

Sample collection
Fresh fecal samples (n = 108) were collected using sterile spatulas. The interior of the fecal pellet was collected to ensure minimal environmental exposure. Fecal samples were stored in 5 mL Eppendorf tubes suspended in molecular grade ethanol and when accessible refrigerated in -4ºC. DNA was extracted using Macherey-Nagel NucleoSpin Soil Kit (Macherey-Nagel, Duren, Germany) on Agilent extraction robot, suspended in TE buffer, and stored in -4ºC.

Metabarcoding
Using a metabarcoding approach, we ampli ed a highly variable region (V4) of the bacterial 16S rRNA gene using 515F/806R primers [14] and the fungal ITS gene [15]. We used a SeraMag solution to clean PCR products to isolate bacterial and fungal DNA [32]. We pooled DNA according to the number of samples. Qubit (dsDNA HS Assay Kit, Invitrogen, Carlsbad, United States) was used to quantify DNA concentration and libraries were diluted to 4 nM prior to nal pooling. We used LabChip GX Touch Nucleic Acid Analyzer to determine DNA concentration and assess quality. Samples were sequenced using Illumina MiSeq platform at Auckland Genomics Facility (University of Auckland), phiX spike 10%, 250 × 2 cycles. Bioinformatics pipeline Claident was used to demultiplex raw sequences. Pear merged paired end reads. VSEARCH ltered noisy reads, removed chimeras, and clustered sequences into operational taxonomic units (OTUs). Claident generated an OTU and assigned taxonomy with RDP classi er database [33].

Statistical analysis
We calculated Shannon index to test for a relationship between microbial alpha diversity and captivity. To better understand the intraspeci c variation in microbial composition and assess if certain bird species have higher variance than others, we calculated beta diversity using a multivariate version of Levene's test for homogeneity of variances. We reported the distance to centroid value.
We used R packages phyloseq and vegan [16,34]. We used non-metric dimensional scaling (NDMS) with Bray-Curtis dissimilarity matrices to reduce multivariate data and spatially visualize microbial communities. NMDS was used to nd patterns across captivity. We used permutational analysis of the variance (PERMANOVA) also with Bray-Curtis distance matrices to determine whether different factors, such as captivity status (wild/captive), site (geographic area), microsite (i.e. in brooder box, soil, etc.), age (days old of captive individuals), weight (mass in grams for captive individuals), collection date, and history of coccidiosis (positive/negative) can explain microbial community variance. We conducted a simper analysis [16] to determine which OTUs explain over 70% of the differences between treatments. We used a clamtest [16] to categorize bacterial and fungal OTUs into groups: generalist, too rare, and treatment specialist (wild-, captive-, positive-, negative-). Positive and negative correspond to individual kiwi who have had a history of coccidiosis.  The Brown Kiwi bacterial community differs both in diversity and composition due to captivity status. (A) Alpha diversity of captive kiwi is signi cantly reduced compared to wild individuals (B) NMDS shows distinct clusters between wild and captive samples and differences between individuals with and without a history of coccidiosis (C) Relative abundances of bacterial phyla between captive and wild kiwi.