Reintroduction training is instrumental in restoring the oral microbiota of giant pandas from "captivity" to "wildness"

Abstract

Reintroduction programs aim to restore wild populations, yet success is challenged by host microbiome adaptation to natural environments. Here, we characterized the oral microbiota of giant pandas undergoing pre-release training, comparing them to captive and wild conspecifics, to assess training-induced microbial shifts. We found that after one year of reintroduction training, multi-generational captive giant pandas exhibited increased oral microbiome diversity, with community structure, composition, and predicted functions converging toward wild-type profiles. Adaptive changes included reduced relative abundances of Actinobacillus and Bergeyella, and enrichment of Myroides and Psychrobacter. Functionally, these shifts correlated with decreased starch and sucrose, fructose and mannose, and various lipid metabolism pathways, alongside enhanced methane and galactose metabolism which align with the dietary constraints of a singular food source in the wild environment. Our study demonstrates that pre-release training drives oral microbiota convergence toward wild phenotypes, underscoring microbial adaptation as critical for successful captive-to-wild transitions in endangered species.

Keywords: Adaptive transition; Giant panda; Oral microbiota; Reintroduction.

Fig. 1

Overview of the study area and sample collection information. A The green area represents the Giant Panda National Park (GPNP), and the red spot indicates the sampling location. B The table of oral sample donors and detailed sampling information (C) The reintroduction pre-release training location: Daxiangling Nature Reserve, Daxiangling Mountains (DXL). D The pre-release reintroduction training stage for captive giant pandas: 3 individuals lived in 3 adjacent pre-release reintroduction training areas with fences in the Daxiangling Nature Reserve (green shape)

Fig. 2

The differences in OTU distribution, microbial diversity, and structure among the CA, PR, and WI groups. A The unique and shared OTUs among CA, PR and WI groups. The difference of (B) ACE, C Chao1, E Shannon and (F) Simpson index among CA, PR, and WI groups. (Description: asterisk indicates significant differences, p_FDR < 0.05, Wilcoxon test, error bars represent Mean ± SD); D PCoA based on the Bray-curtis distance matrix to find principal coordinates. (Description: X-axis, PCoA axis1 and Y axis, PCoA axis2. The scale of the X-axis and the Y-axis are the projection coordinates of the sample points in the two-dimensional plane, respectively. A dot represents each sample, and different colors represent different groups). G The relative abundance heatmap of shared OTUs among CA, PR and WI groups

Fig. 3

The oral microbial community and difference results. The microbial community bar plot at the (A) Phylum level and (C) Genus level. Bar charts showing the relative abundance of all phyla and genera detected in the oral microbiota collected from the giant pandas in the CA, PR and WI group. The identities of the microbiome were shown with color blocks on the right. The Linear Discriminant Analysis (LDA) demonstrated distinct microorganism on (B) Phylum level and (D) Genus level enriched in the CA, PR and WI group. The graph shows the LDA scores obtained from linear regression analysis of the significant microorganism groups in every two groups. When the default LDA value is > 4.0 and the p value is < 0.05, the result corresponds to a differential species

Fig. 4

The oral core microbiota and co-occurrence network analysis of captive, pre-released and wild giant pandas. A The core microbiota OTUs among CA, PR and WI group. The co-occurrence network analysis results in core oral microbiota of (B) CA group, (C) PR group and (D) WI group. The microbial correlations were inferred from OTUs abundance profiles using the Spearman method and only the robust and significant correlation (correlation values < − 0.8 or > 0.8 and p < 0.05) were kept for the construction of co-occurrence networks. Potential keystone taxa on the basis of co-occurrence network analyses of microbial communities, node sizes and colors are proportional to their scaled NESH score (i.e., score revealing important microbial taxa of microbial association networks). The size of the dots indicates the importance of each OTU, while the thickness of the connecting lines reflects the strength of the correlation. Red lines represent positive correlations, and blue lines indicate negative correlations

Fig. 5

The oral microbiota predictive function gene analysis of captive, reintroduced and wild giant pandas. A PCA plots representing the predicted metabolic pathways from oral samples of the CA, PR, and WI groups. (Description: X-axis, PCA axis1 and Y axis, PCA axis2. The scale of the X-axis and the Y-axis are the projection coordinates of the sample points in the two-dimensional plane, respectively. A dot represents each sample, and different colors represent different groups). The bar plot with the Wilcoxon rank-sum test and FDR for multiple test correction demonstrated the relative abundance of distinct KEGG functional metabolic pathways on (B) level 1and (C) level 2 in the CA, PR and WI group. The figure only displayed metabolic pathways with a relative abundance greater than 1%. The figure showed the significant KEGG functional metabolic pathways in every two groups. When the p value is less than 0.01, the result corresponds to the differential metabolic pathways, the asterisk indicates a significant difference between the two groups. D Biochemical synthesis iPath pathway map. Each line segment represents a distinct enzyme or metabolic pathway, with blue lines indicating pathways significantly enriched in the CA group, and red lines indicating those in the WI group. Nodes symbolize compounds, and lines connecting nodes are enzymes. Only KEGG orthologous genes with extremely significant differences (p < 0.01) were shown