The intricate relationship between the gut microbiota (GM), which contains trillions of microorganisms (Budd et al., 2020), and the health of organisms is an emerging frontier in biology, and its significance extends even to the realm of conservation medicine for endangered species (Bahrndorff et al., 2016; West et al., 2019). Conservation efforts often focus on external factors such as habitat protection, breeding programs, and disease management. However, the internal microbial communities residing within animals, particularly their digestive tracts, are increasingly recognized as pivotal players in the overall health and well-being of these species (Bahrndorff et al., 2016; West et al., 2019). Endangered species often face challenges that disrupt their natural environments, alter their diets, and expose them to new pathogens. These disruptions can impact the balance of the GM, leading to health issues and reduced reproductive success (Lee and Hase, 2014; Shreiner et al., 2015; Keady et al., 2021). The GM plays a primary role in many physiological processes, including digestion and nutrient uptake, immunity, or reproduction, and dysbiosis (i.e., deviations from balanced microbiota) can lead to many pathological processes (Kau et al., 2011; Lee and Hase, 2014; Shreiner et al., 2015; Fabbiano et al., 2017; Keady et al., 2021). Accordingly, by studying the GM of endangered species, researchers can gain insights into how these microorganisms contribute to the overall health and adaptability of the animals. This understanding can inform conservation efforts by guiding dietary recommendations, disease management strategies, and even assisted reproductive techniques (Keady et al., 2021).

Elephants are non-ruminant, monogastric herbivorous, hindgut fermenters (Van Hoven and Lankhorst, 1981; Dumonceaux, 2006). In the wild, African savanna elephants (Loxodonta africana) have preferences in plant types and parts, and diet selection is optimized for energy, growth or reproduction (Biru and Bekele, 2012; Pretorius et al., 2012; Mwambola et al., 2014; Sach et al., 2019). As such, zoo-housed elephants should be provided with various types of hay and browse with differing energetic and digestibility coefficients (Bolechova et al., 2020). As diet has a direct influence on the gut microbiota of African herbivores (Kartzinel et al., 2019), GM differences between managed vs. free-ranging elephants are expected. Elephant fermentation primarily occurs in the colon, even if some fermentative bacteria can be found in their cecum (Van Hoven and Lankhorst, 1981; Clauss et al., 2003; Dumonceaux, 2006). As a consequence, the highest apparent digestibility is reported in the upper portion of their colon (Clemens and Maloiy, 1983), and energy supply is based on short-chain fatty acids originating from colonic fermentation of fibrous forage (Greene et al., 2019a).

Elephant dung boluses are easy to collect, non-invasive samples whose composition is a good predictor of their hindgut health, and the fecal/hindgut microbiome is often used as a proxy of the GM microbiome (Budd et al., 2020; Kandel et al., 2020; Keady et al., 2021; Moustafa et al., 2021). As other colonic fermenters with similar digestive physiology, equids are often used as a model for digestion in Proboscidae, and their GM has already been described (Kauter et al., 2019). Nevertheless, some differences exist (Clauss et al., 2003; Greene et al., 2019a). In particular, elephants exhibit faster ingesta passage rates, leading to lower nutrient digestibility coefficients (Hackenberger, 1987; Clauss et al., 2003). To optimize energy intake, elephants are able to increase food intake without dramatically raising ingesta passage (Clauss et al., 2007). These differences with horses reinforce the need for a precise description of the GM in elephants. Some data is available for wild African elephants (Budd et al., 2020), and one recent study evaluated the influence of clinical and hormonal findings in African savanna elephants housed in US institutions (Keady et al., 2021), but to the best of the authors’ knowledge the influence of husbandry and individual factors on the fecal microbiota of zoo-managed African savanna elephants is not available, nor is a comparison between captive and wild elephants.

The pursuit of the core microbiome has gained prominence in microbial ecology, shedding light on the interactions between microbial communities and their hosts or environments (Neu et al., 2021). Essentially, a core microbiome encompasses shared microbial taxa and their associated genomic or functional attributes, characterizing specific hosts or ecosystems (Turnbaugh et al., 2007; Hamady and Knight, 2009; Risely, 2020). This endeavor has prompted investigations into genes (Turnbaugh et al., 2009), functional pathways (Jiang et al., 2016), and metabolic profiles (Stefanini et al., 2017) common among microbial communities in various settings. Typically, uncovering a core microbiome involves identifying shared taxa across multiple microbial communities within a given host species or environment, hypothesizing these shared taxa represent vital microbial associates of the host or environment under the conditions studied (Shade and Handelsman, 2012). These core members are assumed to have ecological and functional importance, with potential implications ranging from human health (Zaura et al., 2009; Bäckhed et al., 2012) to responses to climate change (Ainsworth and Gates, 2016; Hutchins et al., 2019) leading to a surge in core microbiome studies (Neu et al., 2021). However, core microbiome analyses differ in their criteria to define the core. The process typically involves determining the proportion of samples sharing a set of microbial taxa, their relative abundances, or a combination of both. This quantification spans different taxonomic levels, from amplicon sequence variants (ASVs) to phyla, and across various spatial and temporal scales (Neu et al., 2021). More recently, core association networks (CAN) were proposed (Röttjers et al., 2021) as a novel approach to examine the concept of the core microbiome. These networks extend the understanding of microbial interactions beyond individual taxa by investigating the associations and co-occurrences among them. A core association network represents a subset of microbial interactions that are conserved across different samples or environments, providing insights into the shared microbial partnerships that might underlie core microbial communities (Röttjers et al., 2021).

The objectives of the present study were (1) to characterize the fecal microbiota of European captive-bred African savanna elephants based on amplicon sequencing (16S rRNA) of dung samples, (2) to measure the influence of several factors (age, sex, daily activity, residence zoo, type of diet including provision of branches, birthplace, and social status), and (3) to compare its composition and diversity to previously reported data in free-ranging African savanna elephants. We hypothesized that (a) captive and wild animals of the same species develop different microbiomes due to their differing environments, (b) individual and environmental factors affect microbiome composition, and proper diet and exercise correspond with a healthier microbiome. If these hypotheses hold true, we should observe that (a) the microbiota of captive-bred African savanna elephants differs from that of their wild counterparts, (b) differences are observed as a function of influence factors, and (c) branch intake and diversity and increased exercise lead to positive health adaptations.

The study provides insights into the GM of managed African savanna elephants, offering a foundation for clinical comparisons, probiotic use, and transfaunation (Coverdale, 2016; Greene et al., 2019b), particularly for elephants facing changes due to captivity and anthropogenic activities (Moustafa et al., 2021). However, the scope of this study’s findings is limited to managed African savanna elephants, as GM composition varies across species, environments, and diets (Budd et al., 2020; Keady et al., 2021).

Limitations of this study encompass the small sample size (n = 21) and divergent sampling methods between French elephants (sterile collection of fresh dung boluses on FTA cards) and prior free-ranging data (20 g of fecal sample boiled, then stored at −20°C; Budd et al., 2020). Varied sample collectors necessitated a simple, rapid, and consistent method. FTA cards mitigated challenges of preserving fibrous samples and eliminated biases from freezing logistics (Supplementary Figure 1). While immediate freezing at −80°C is considered ideal, FTA cards showed high intraclass correlation coefficients for α- and β-diversities compared to freezing (Wang et al., 2018). Chosen for ease of use, repeatability, and moisture preservation (Stępień et al., 2019), FTA cards’ primary limitation was lower read counts compared to frozen dung boluses (Budd et al., 2020). An additional limitation lies in the different sequencing technologies used for the zoo and wild elephants, specifically Ion Torrent and Illumina, respectively. In our study, the Ion Torrent method targeted seven regions of the 16S gene. While Ion Torrent generally has lower sequencing depth and unique base-calling errors compared to Illumina, the multi-region targeting partially counterbalances these technological differences. To further ensure the rigor and validity of our comparative analysis, we implemented a consistent bioinformatics pipeline for both datasets, for instance (i) minimized noise through the DADA2 approach, (ii) standardized taxonomic assignments using the SILVA v138 database, and (iii) conducted network and statistical analyses based on centered log-ratio transformation (Aitchison, 1986), which made the data symmetrical.

Institutions (zoo of residence) significantly impacted β-diversity differences (Table 2), echoing US African savanna elephants’ findings (Keady et al., 2021). Different sampling methods were employed in that study (Keady et al., 2021), but husbandry practices, encompassing diet, training, and enrichment, uniquely shaped African savanna elephant GM (Table 2; Figures 1C, 2). Notably, regardless of pellet brand (horse, cattle, or elephant-designed), β-diversity remained similar, implying limited pellet impact (Figure 1C). Similar gut microbiomes among geographically proximate zoo elephants (Kartzinel et al., 2019; Keady et al., 2021) imply influential environmental and nutritional factors in GM, including hay types (Keady et al., 2021). Only one zoo in our study analyzed the analytical components of hay, so we did not include this parameter in the statistical analyses. However, variations in fruits, green vegetables, and seeds led to β-diversity differences, highlighting the significance of varied diets (Figure 1C).

Diet and daily activity are interrelated, with foraging and exploration increasing when branches are available (Lasky et al., 2021). In our study, diet, especially branch supply, significantly influenced zoo African savanna elephant GM. Elephants displaying higher daily activity exhibited elevated Lentisphaerae proportions, a pattern linked to better sleep in humans (Anderson et al., 2017). Augmented activity could offer improved sleep for managed elephants, potentially through increased access to branches in outdoor enclosures. This is all the more interesting since sleep behavior is recognized as a major indicator of elephant wellfare (Schiffmann et al., 2018). Factors like enclosure design complexity and NDF differences between diets with and without browse might also influence daily activity and GM (Clauss and Dierenfeld, 2008; Katole et al., 2014; Scott and LaDue, 2019; Wood et al., 2020). Encouragingly, mimicking wild diets with diverse branch types could enhance GM composition (Table 2), adjusting throughout the year according to tree species’ availability (Wood et al., 2020). Additionally, branches and browse can supply minerals and enhance vitamin uptake in elephants, but excluding pellets is not advised due to their role in meeting nutrient requirements, including vitamins D and E (Dierenfeld, 2006; Wood et al., 2020). Despite the benefits of branches, pellets play a vital role in balanced diets for captive elephants (Wood et al., 2020). Providing branches aligns with guidelines and enhances welfare by improving foraging and exercise opportunities (Bolechova et al., 2020; Lasky et al., 2021).

An influence of sex was seen on selected bacterial phyla, as Spirochaetes and Synergistetes were overrepresented in males. To the authors’ knowledge, it has not been assessed in adult African elephants so far, and hormonal differences between males and females may explain these differences. As Spirochaetes are overrepresented in horses fed exclusively on hay (Fernandes et al., 2021), male elephants may eat a higher proportion of hay than females. In other mammals, such as humans, the composition of intestinal microbiota differs over time between the sexes (Valeri and Endres, 2021), and in the closely related Asian elephants (Elephas maximus), the metabolizable energy intake is maintained from their diet across all seasons to reach sufficient needs to complete sex-specific physiological needs (Koirala et al., 2019). In addition, there is a link between hormones (including prolactin) and the structure of the gut microbiome in captive elephants (Keady et al., 2021). Such differences justify hindgut microbiome sex-specific research in African savanna elephants.

The GM of managed elephants was different than wild individuals, as significant differences were seen in its taxonomic structure between zoo and wild African savanna elephants (Figure 3A), with only 99 bacterial taxa in common (Figure 3C) and 46 bacterial genera with significantly different abundance (Figure 4). Digestive disorders may be more frequent in zoo elephants (Dumonceaux, 2006; Miller et al., 2015; Greene et al., 2019a; Scharling et al., 2021), as lower relative abundances of Ruminococcae and Lachnospiraceae have been consistently reported in horses with gastrointestinal disease (Costa et al., 2012; Schoster et al., 2017), and these bacterial families were underrepresented in zoo elephants (Figure 4). These members of the Firmicutes phylum are cellulolytic bacteria that degrade cellulose to produce energy-providing short-chain fatty acids (Kauter et al., 2019). Such bacteria are likely responsible for the production of specific enzymes responsible for colonic fermentation in elephants (Sthitmatee et al., 2011). Their lower abundance in zoo elephants may be linked to a lower diversity in diet items compared to the wild (Biru and Bekele, 2012; Pretorius et al., 2012; Mwambola et al., 2014; Sach et al., 2019), and with seasonal changes in food availability which is dramatically reduced in yearlong homogeneous pellet-based diet of captive-bred elephants (Codron et al., 2006; Dierenfeld, 2006; Clauss and Dierenfeld, 2008; Wood et al., 2020). This may explain markedly reduced bacterial occurrence networks in captive individuals as evidenced in Figure 5 and a markedly reduced network robusteness (Figure 6). Through zootechnical and nutritional modifications, increasing the robustness of microbiota networks in zoo elephants would bring them closer to those of wild elephants, resulting in a more resilient microbial community, better able to withstand disturbance, and improving their ability to adapt to changing conditions, respond to stressors and maintain optimal health.

Interestingly, our study revealed that despite differences in their environments, zoo elephants shared a core group of taxa with their wild counterparts. The shared “core microbiota” among zoo and wild African savanna elephants indicates a remarkable degree of consistency in certain microbial taxa across distinct environments. This finding suggests that while elephants in captivity settings and the wild experience contrasting diets and living conditions, there are underlying factors that drive the preservation of specific microbial communities. These core taxa, including Rikenellaceae, Prevotellaceae, Lachnospiraceae, and Ruminococcaceae, are not solely determined by host phylogeny, but also by the shared hindgut fermentation strategy common to herbivores like horses, rhinoceroses and smaller domestic species (Bian et al., 2013; Dougal et al., 2013; O'Donnell et al., 2017). This implies that the evolutionary heritage of hindgut fermenters contributes to the establishment of these core microbial communities, emphasizing the functional importance of these microbes in herbivorous digestion.

The discovery of core association networks (CANs) builds upon the concept of the “core microbiota” by revealing consistent patterns of microbial interactions. These CANs represent stable relationships that persist across different conditions (Röttjers et al., 2021), regardless of the varying dietary and environmental contexts experienced by zoo and wild elephants. The presence of these non-random associations in both captive and wild populations signifies meaningful microbial interactions that contribute to host health and functioning. The CANs provide insights into the coexistence and cooperation of microbial species within the gut ecosystem, contributing to the complex dynamics of hindgut fermentation and nutrient processing. The correlation between the shared core taxa and the CANs becomes particularly intriguing. The presence of certain microbial groups in the core microbiota may influence the formation of specific microbial associations within the CANs. For instance, the cellulolytic capabilities of members of the Ruminococcaceae (Froidurot and Julliand, 2022) family among the core taxa can shape interactions within CANs by affecting the breakdown of complex plant material. These interactions, in turn, might impact nutrient availability for the host (Froidurot and Julliand, 2022), and potentially contribute to the observed adaptations in digestive strategies.

This study holds significant implications for comprehending the impact of living environment on the gut microbiota (GM) of the endangered species Loxodonta africana, the African savanna elephant. The findings not only provide crucial insights for enhancing the management of these animals in zoos but also contribute to a relatively nascent area of knowledge: the intricate ecosystems constituted by host-associated microbiota.

It is evident from this research that, irrespective of their habitat, African savanna elephants share a fundamental core composition of gut microbiota, characterized by the prevalence of bacterial types commonly associated with hindgut fermenters, notably Rikenellaceae, Prevotellaceae, Lachnospiraceae, and Ruminococcaceae. However, a striking revelation emerges as zoo-dwelling elephants exhibit a discernible reduction in both bacterial diversity and intricate interactions within the structure of their hindgut microbiota, in comparison to their wild counterparts.

Zoo husbandry factors emerged as pivotal determinants shaping the composition of the microbiota in zoo elephants. Among these factors, diet and daily activity stand out prominently. The influence of these factors, particularly diet and its seasonal variations, can elucidate the observed differences between captive and wild individuals. To optimize the gut health of zoo elephants and to mirror their natural dietary patterns, it becomes imperative to provide a diverse array of branches that can be foraged throughout their enclosures. This dietary enrichment has the potential to positively impact the composition and diversity of the microbiota.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found at: https://www.ncbi.nlm.nih.gov/, PRJNA939400 (SRX19520099, SRX19520098, SRX19520097, SRX19520096, SRX19520095, SRX19520094, SRX19520093, SRX19520092, SRX19520091, SRX19520090, SRX19520089, SRX19520088, SRX19520087, SRX19520086, SRX19520085, SRX19520084, SRX19520083, SRX19520082, SRX19520081, SRX19520080, SRX19520079) https://www.ncbi.nlm.nih.gov/, PRJNA587772.

Ethical approval was not required for the study involving animals in accordance with the local legislation and institutional requirements because the study required only the collection of fresh boluses of elephant dung. These samples were collected immediately after the animal’s defecation, in a totally non-invasive manner (on the ground). No animals were handled to collect these samples, and no animal’s daily routine was altered for the study.

MT, BM, P-YM, AC-C, and AL conceived the study. P-YM acquired the data. DO, AM, and AW-C performed the bioinformatic and statiscal microbiome analyses, and visualized the results. LM-H curated the data. MT, DO, and AC-C drafted the first version of the manuscript. AW-C edited and corrected the manuscript. All authors revised and accepted the last version of the manuscript.