Gut microbial ecology of the Critically Endangered Fijian crested iguana (Brachylophus vitiensis): Effects of captivity status and host reintroduction on endogenous microbiomes

Abstract Animals often exhibit distinct microbial communities when maintained in captivity as compared to when in the wild. Such differentiation may be significant in headstart and reintroduction programs where individuals spend some time in captivity before release into native habitats. Using 16S rRNA gene sequencing, we (i) assessed differences in gut microbial communities between captive and wild Fijian crested iguanas (Brachylophus vitiensis) and (ii) resampled gut microbiota in captive iguanas released onto a native island to monitor microbiome restructuring in the wild. We used both cloacal swabs and fecal samples to further increase our understanding of gut microbial ecology in this IUCN Critically Endangered species. We found significant differentiation in gut microbial community composition and structure between captive and wild iguanas in both sampling schemes. Approximately two months postrelease, microbial communities in cloacal samples from formerly captive iguanas closely resembled wild counterparts. Interestingly, microbial communities in fecal samples from these individuals remained significantly distinct from wild conspecifics. Our results indicate that captive upbringings can lead to differences in microbial assemblages in headstart iguanas as compared to wild individuals even after host reintroduction into native conditions. This investigation highlights the necessity of continuous monitoring of reintroduced animals in the wild to ensure successful acclimatization and release.


| INTRODUC TI ON
Gastrointestinal microbial communities are critical to host health, contributing to an array of functions that impact host fitness and reproductive success such as nutrient acquisition based on digestive efficiency, hormone balance, and immune response (Cho & Blaser, 2012;Colston & Jackson, 2016;Fraune & Bosch, 2010;Ley et al., 2008). Given that gut microbiota serve essential roles in maintaining host well-being, the study of these communities is a novel tool for wildlife conservation initiatives, particularly in programs involving ex situ animal care (Bahrndorff et al., 2016;Jiménez & Sommer, 2017;Redford et al., 2012;West et al., 2019). With few exceptions, a variety of species housed in captivity show disparate gut microbiomes compared with wild counterparts which may be caused by dietary differences, antibiotic treatments, exposure to other species in captivity, or various other potential drivers that alter microbial compositions (Alfano et al., 2015;Cheng et al., 2015;Clayton et al., 2016;Eigeland et al., 2012;McKenzie et al., 2017;West et al., 2019;Zhu et al., 2011). Such differences may be signs of dysbiosis, or perturbations of microbial communities that hinder system function and are often associated with negative health outcomes in hosts (Gilbert et al., 2016;West et al., 2019). For example, captivity has been linked to increases in potential pathogens within gastrointestinal microbial communities in mammals (Amato et al., 2016;Cheng et al., 2015;Wan et al., 2016;Wasimuddin et al., 2017), birds (Xie et al., 2016), and reptiles (Jiang et al., 2017;Kohl et al., 2017).
Distinct gut microbiota between captive and wild hosts is especially significant in headstart and reintroduction conservation programs, as altered microbial communities or introduced pathogens in captive animals slated for release could hinder reintroduction success and survivorship in the wild due to reduced dietary efficiency or compromised immune response affecting survivorship (Bahrndorff et al., 2016;Jiménez & Sommer, 2017;Redford et al., 2012;West et al., 2019).
Headstart programs have become increasingly common management strategies to supplement declining wildlife populations at risk of extinction (McGowan et al., 2017;Redford et al., 2011;Tear et al., 1993). In these programs, young animals are reared in captivity past their most vulnerable life stages before being released to reinforce wild populations (Alberts, 2007;Ferguson et al., 1982).
Historically, however, effective reintroduction of captive animals into the wild has been rare, with as few as 13% of such projects being deemed successful (Fischer & Lindenmayer, 2000;Mathews et al., 2005). Multiple factors have been linked to animal headstart and reintroduction difficulties including individual animal behavior (Alberts, 2007;Mathews et al., 2005) and ill-suited release sites (Pérez-Buitrago et al., 2008). More recently, microbial incompatibilities also have been suggested as possible impediments to reintroduction success (Bahrndorff et al., 2016;Jiménez & Sommer, 2017;Redford et al., 2012;West et al., 2019). However, no studies to date have examined gut microbiota in reintroduced species both preand postrelease to analyze microbial composition and acclimation of these communities to native habitats. Improved understanding of host natural microbiomes and microbial shifts associated with captivity and headstart animal release could help management practitioners to better prepare animals for reintroduction and increase headstart success of imperiled species.
The Fijian crested iguana (Brachylophus vitiensis) is an herbivorous lizard species endemic to dry and littoral forests in western Fiji (Fisher et al., 2019;Harlow, Fisher, & Grant, 2012). Since the species' discovery in 1981, it has experienced sharp population declines throughout most of its limited range due to habitat loss and introduced predators (Fisher et al., 2019;Gibbons, 1981;Harlow et al., 2007). The Fijian crested iguana is listed on CITES Appendix S1 and as Critically Endangered by the IUCN Red List (Fisher et al., 2019;Harlow et al., 2012). To ensure the long-term viability of this species in Fiji, a captive breeding and headstart program was established in 2010 with a specific focus on animals from the uninhabited island of Monuriki (Chand et al., 2016;Fisher et al., 2019). Monuriki Island crested iguanas are genetically distinct from all other crested iguana populations (Keogh et al., 2008), and the 2008 Iguana Species Recovery Plan prioritized Monuriki as the single most important site for immediate conservation action for this taxon (Fisher et al., 2019;Harlow et al., 2008). From 2010 to 2012, 20 adult iguanas were caught in the wild from Monuriki Island and transported to Kula Eco Park on the large island of Viti Levu to develop a captive breeding colony (Chand et al., 2016). Over the next six years, these 20 wildcaught individuals were successfully bred in managed care at Kula Eco Park with the intention of headstarting and returning the offspring to their source island of Monuriki (Chand et al., 2016;Fisher et al., 2019). In mid-May 2015, 32 captive-bred crested iguanas were released onto Monuriki Island, with an additional 32 captive-bred iguanas and 16 of the original adult wild founder iguanas released onto Monuriki in February 2017.
In 2017, we completed extensive sampling of gut microbial communities from Fijian crested iguanas in captivity at Kula Eco Park, wild iguanas on Monuriki Island, and previously captive iguanas released onto Monuriki to better understand how endogenous microbiomes are influenced by both human care and host reintroduction.
In this study, we not only compare gut microbiomes in captive and wild lizards of a Critically Endangered species, but also assess the restructuring of microbiota in headstart animals reintroduced into native habitats. Additionally, by inventorying gut microbiota in Fijian crested iguanas using two sampling techniques, cloacal swabs and fecal samples, we address how sampling regime influences microbial data recovered and subsequent downstream analyses. While gut microbial diversity reported from cloacal and fecal sampling is often similar, significant discrepancies in relative abundances of microbial taxa between sampling types are well noted (Colston et al., 2015;Kohl et al., 2017;Stanley et al., 2015). We used both techniques to maximize our understanding of gut microbial ecology in B. vitiensis and to mitigate potential shortcomings associated with employment of a single sampling technique (Colston et al., 2015;Ren et al., 2016). The Fijian crested iguana headstart initiative represents a unique opportunity to address two important research questions: (i) How does captivity status effect the diversity and structure of gut microbiomes? and (ii) How do such communities respond to host reintroduction into native habitats? The results of this study have direct implications for the management and conservation of this Critically Endangered reptile species and for headstart and reintroduction programs globally. Hatchlings were fed in the same manner as adults and juveniles, but salads were cut into smaller pieces. We implanted unique passive Brachylophus (Harlow et al., 2007) on Monuriki Island to monitor iguanas and sample gut microbial communities in the wild. Once iguanas were captured, the presence of a PIT tag allowed us to determine whether the individual was a WCF, CB2017, or CB2015 iguana, while all iguanas lacking PIT tags were classified as Wild individuals.

| Animal maintenance and sample collection
Gut microbial samples were collected using the same methodologies as for iguanas in captivity at Kula Eco Park. the index primers and PCR protocols of Kozich et al., (2013). PCR products were cleaned, normalized, and pooled using a SequelPrep Normalization Plate Kit (Invitrogen). Pooled libraries were purified using Agencourt® AMPure® magnetic bead capture and sent to the University of Oklahoma's Consolidated Core Lab (CCL) for sequencing using 515F and 806R primers targeting 2x300bp reads on an Illumina MiSeq sequencing platform (Caporaso et al., 2012).
Taxonomies were assigned to OTUs using GreenGenes v13.8 (DeSantis et al., 2006). Archaea, chloroplast, mitochondria, PhiX, and other nonbacterial sequences were removed from processed OTU tables to ensure only bacterial sequences were included in downstream analyses. All 16S rRNA sequences have been deposited in the Sequence Read Archive (SRA) under accession no. PRJNA702127.
Among all samples (n = 94), a number were either duplicates (i.e., multiple subsamples of a single fecal deposit or cloacal swabs collected from the same host consecutively) or failed to generate sufficient sequencing coverage to produce meaningful microbial assessments. In instances where duplicate samples existed (n = 9), we retained only the sample with the greater sequencing depth. Of the remaining samples, those with fewer than 500 sequences (n = 2) were also removed to maximize sample inclusion against OTU coverage. The finalized dataset used for all subsequent analyses consisted of 83 samples (46 cloacal and 37 fecal) from 38 Fijian crested iguanas (Appendix S1). Within these datasets, five Fijian crested iguana hosts had complete time-series sets (pre-and postrelease sampling) via cloacal swabbing and five had them through fecal sampling.
Three individuals occurred in both groups and had complete sampling sets from the two methodologies (Appendix S1).
Rarefaction depths varied by comparison based on Good's coverage estimates (Good, 1953) and rarefaction curves to maximize sample inclusion against OTU coverage ( Figure S1). We compared a variety of community membership metrics across samples from Fijian crested iguana hosts. For all comparisons, we first calculated alpha-diversity measurements including number of observed OTUs, the Shannon index (Shannon, 1948), and Faith's phylogenetic diversity (Faith's PD;Faith, 1992). Alpha-diversity measurements were compared using analysis of variance (ANOVA) tests in R v3.6.2 (R Core Team, 2013) with the Tukey test used for post hoc analyses. The Kruskal-Wallis tests with Bonferroni's corrections were used in QIIME to compare relative abundances of bacterial taxa between treatment groups. In examining specific OTUs, BLAST (Altschul et al., 1990) was used to compare novel sequences against those available in the National Center for Biotechnology Information's (NCBI) nucleotide database.
Community diversity and structure were compared using principal coordinates analysis (PCoA) on beta-diversity metrics including weighted and unweighted UniFrac distances (Lozupone & Knight, 2005) and the binary Jaccard index (Jaccard, 1901). Betadiversity matrices and PCoA plots were generated from the same rarefied datasets used to measure alpha-diversity metrics. The adonis function in the vegan v2.3_4 package (Oksanen et al., 2016) of R v3.3.1 (R Core Team, 2013) was used on beta-diversity distance matrices with 999 permutations to compare community composition between groups statistically.

| Sample comparisons
We first analyzed bacterial composition across all 83 samples (Appendix S1) and then split the dataset into cloacal and fecal subsets to examine general patterns between sample types. Following broad overviews of the data, we tested the effects of captivity status on gut bacterial communities in crested iguana hosts and examined for microbial restructuring in reintroduced lizards postrelease.
To determine the influences of captivity status on gut microbial communities, we used snapshot analyses of cloacal and fecal samples taken from WCF, CB2017, CB2015, and Wild lizards. For cloacal comparisons, we included 35 samples collected between 22 February and 2 March 2017 (Appendix S1). This subset included 10 WCF, 13 CB2017, three CB2015, and nine Wild individuals. In our subsequent fecal analyses, we included 26 fecal samples collected between 22 February and 1 March 2017 (Appendix S1). This dataset encompassed fecal samples from nine WCF, nine CB2017, two CB2015, and six Wild iguanas. In addition to comparing microbial communities across four treatments, we also ran all analyses between just two conditions, captive (WCF and CB2017 grouped) and noncaptive (CB2015 and Wild grouped) (Ren et al., 2016).
We sought to assess the effects of release on lizard microbiota using both cloacal and fecal samples collected roughly 2 months after host reintroduction to Monuriki. We collected cloacal samples from five recently released lizards, one WCF and four CB2017, between 24 April and 11 May 2017 (Appendix S1). We compared microbial communities from these samples against those in the initial 23 captive animal cloacal samples (10 WCF, 13 CB2017) and the initial 12 noncaptive samples (nine Wild, three CB2015). We also compared six novel fecal samples (one WCF, five CB2017) collected between 2 and 17 May (Appendix S1) against the 18 initial captive fecal samples (nine WCF, nine CB2017) and eight noncaptive fecal samples (two CB2015, six Wild). In both instances, we sought to determine whether gut microbiomes were more similar to captive communities or noncaptive communities two months after host reintroduction.

| General patterns in Fijian crested iguana microbiota
Our curated dataset of 83 samples generated 898,625 reads with a minimum read depth of 540, a maximum of 30,503, and a median of 9,883 reads per sample. Among the 46 cloacal samples only, 410,545 reads were recovered with a minimum read depth of 540 sequences per sample, maximum of 25,304, and median read depth of 8,521.5.
The 37 fecal samples produced 488,080 reads with a minimum, maximum, and median read depth of 3,378, 30,503, and 12,558 reads per sample, respectively.

| Comparison of microbiota in captive and noncaptive iguanas via cloacal samples
Comparisons of cloacal samples from Fijian crested iguanas of treatment groups WCF, CB2017, CB2015, and Wild yielded no significant differences in measured alpha-diversity metrics ( Figure S2). This lack of differentiation remained even when samples were grouped as captive (WCF and CB2017 grouped) and noncaptive (CB2015 and

| Temporal variation of cloacal microbiota in captive crested iguanas postrelease
Comparisons of microbial communities from five cloacal samples taken shortly after host reintroduction against both captive and noncaptive microbial communities revealed no significant variation in alpha-diversity metrics ( Figure S6)

| Temporal variation of fecal microbiota in captive crested iguanas postrelease
We compared microbial communities in six fecal samples from rein- Relative abundances of these phyla varied between conditions ( Figure S10). In comparing OTU relative abundances, Kruskal-Wallis tests retrieved nine OTUs that differed between all three treatments (Appendix S2). These included one Acetobacterium sp.

| D ISCUSS I ON
Our findings show that captive and noncaptive Fijian crested iguanas harbor distinct microbial communities regardless of sampling regime (cloacal versus fecal). These results expand on a growing body of evidence that suggests animals housed in captivity have distinct microbiomes when compared to wild conspecifics (Alfano et al., 2015;Cheng et al., 2015;Clayton et al., 2016;Eigeland et al., 2012;Jiang et al., 2017;Kohl et al., 2017;McKenzie et al., 2017;Ren et al., 2016;West et al., 2019;Zhu et al., 2011). In both cloacal and fecal sampling, captive (WCF and CB2017 grouped) and noncaptive (CB2017 and Wild grouped) iguanas harbored significantly different microbial communities in at least two beta-diversity metrics (Figure 4).
Further, Jaccard distances were consistently lower within captive treatments, suggesting a greater degree of shared OTU breadth and potentially homogenization among captive individuals. These findings are consistent with those seen in Anolis sagrei where alphadiversity measures were generally higher in captive animals compared with wild conspecific hosts, yet gut communities were more homogenous, and beta-diversity metrics separated wild and captive hosts (Ren et al., 2016). In addition to harboring distinct microbial communities, a number of specific OTUs, particularly potential pathogens, were seen in greater abundances in captive over noncaptive Critically Endangered Fijian crested iguanas.
The introduction of potentially pathogenic bacteria has been documented previously in wild reptiles brought into temporary captivity (Jiang et al., 2017;Kohl et al., 2017) but not in a conservation initiative specifically designed to release captive animals into the wild. In cloacal samples from captive Fijian crested iguanas, one Brachybacterium sp., one Brevibacterium sp., and two Nesterenkonia spp. were present in rarefied reads while absent from noncaptive counterparts (Appendix S2). All three of these genera have species implicated as potential pathogens at least in humans (Gruner et al., 1993;Nakayama et al., 2009;Tamai et al., 2018). Fecal samples produced similar results where strains from multiple genera, including Bacteroides, Cloacibacillus, and Ruminococcus were found commonly in captive samples but absent in rarefied, noncaptive reads (Appendix S2). These three genera are also potentially pathogenic strains in humans (Domingo et al., 2015;Titécat et al., 2014;Wexler, 2007). Although determining the exact pathogenic capacities of particular microbes is outside the realm of this investigation,  high abundances of potential pathogens in animals under human care support the possibility that headstart animals can harbor diseasecausing bacteria at significantly higher rates than animals living in the wild (Redford et al., 2012). Although microbial communities in hosts can shift rapidly on the scale of days to even hours in some cases Ren et al., 2016), the impacts of releasing animals with elevated levels of what could be pathogenic microbiota have received little attention to date (Redford et al., 2012).
Reintroduction of captive Fijian crested iguanas into native habitats promoted restructuring of gut microbiomes toward noncaptive communities. After 2 months on Monuriki Island, cloacal samples from reintroduced iguanas appeared to harbor gut microbial communities more similar to noncaptive than to captive compositions ( Figure 4 and S6). Additionally, noted potential pathogens in captive individuals were either absent or diminished in reintroduced hosts.
Microbial assemblages generated from fecal samples, however, did not produce similar results. Instead, microbiota from fecal samples of reintroduced lizards seemingly resembled captive hosts more closely rather than noncaptive hosts (Figures 4 and 5). Potential pathogens also displayed differing trends with Ruminococcus spp.
becoming less abundant in host iguanas two months after release and Bacteroides sp. and Cloacibacillus sp. becoming more abundant in samples taken from individuals after reintroduction. Such findings support previously proposed hypotheses that pathogens associated with human care may continue to impact headstart or reintroduced animals even after release (Bahrndorff et al., 2016;Redford et al., 2012;West et al., 2019). Despite fecal samples from reintroduced iguanas being significantly distinct from noncaptive samples, this differentiation does appear to be temporary. Released animals relocated onto Monuriki Island in 2015 (CB2015) contained gut microbial assemblages more closely associated with true wild iguanas rather than captive ones in both cloacal and fecal samples, suggesting that reacclimation of wild-type microbiomes can occur after prolonged survival in native habitats (i.e., two years; Figure 2).
Although both cloacal and fecal sampling techniques recovered significant differentiation in gut microbial communities between captive and noncaptive Fijian crested iguanas (Figures 2 and 4), specific OTUs that varied between treatments were inconsistent.
Further, differences were apparent in comparing assemblages from reintroduced lizards to those in captive and noncaptive hosts based on sampling regime (Figure 4). Cloacal samples from reptiles generally encapsulate the breadth of gut microbial diversity but vary significantly in abundances compared directly to hindgut samples while fecal samples tend to better represent gut diversity and abundances (Colston et al., 2015;Kohl et al., 2017). When assessing microbial communities in captive lizards for potential disease-causing microbes, or in evaluating the restructuring of host microbiomes postrelease, multiple nonlethal gut microbial sampling techniques may be necessary to fully elucidate trends of interest.
Gut microbial communities in captive Fijian crested iguanas are distinct from those in noncaptive iguanas and this differentiation prevails for some time postrelease. However, the duration in which a host's microbial composition shifts to closely resemble true wild counterparts remains unclear. A continued need exists to monitor microbial communities in headstart animals postrelease to track animal well-being (Bahrndorff et al., 2016;Jiménez & Sommer, 2017;Redford et al., 2012;West et al., 2019). Such studies could determine the influences of potential disease-causing bacteria associated with captive upbringings on host survival, growth, and reproduction in the wild. Further, wild conspecifics in populations with introduced animals should be monitored closely for introduction of novel pathogens brought on from interaction with animals sourced from headstart programs (West et al., 2019). Such scenarios may justify the use of soft releases or probiotics prior to animal release to acclimatize gut microbiota in headstart individuals to natural conditions and eliminate possible disease-causing agents before complete reintroduction to the wild (Redford et al., 2012;West et al., 2019). Along with increased monitoring of animal health, additional scrutiny of specific OTUs seen in differential abundances between headstart and wild animals that may be pathogenic is necessary to determine the virulence of such bacterial strains. Should these OTUs be min-