The putative maintaining mechanism of gut bacterial ecosystem in giant pandas and its potential application in conservation

Abstract

Animals living in captivity and the wild show differences in the internal structure of their gut microbiomes. Here, we performed a meta-analysis of the microbial data of about 494 fecal samples obtained from giant pandas (captive and wild giant pandas). Our results show that the modular structures and topological features of the captive giant panda gut microbiome differ from those of the wild populations. The co-occurrence network of wild giant pandas also contained more nodes and edges, indicating a higher complexity and stability compared to that of captive giant pandas. Keystone species analysis revealed the differences between geographically different wild populations, indicating the potential effect of geography on the internal modular structure. When combining all the giant panda samples for module analysis, we found that the abundant taxa (e.g., belonged to Flavobacterium, Herbaspirillum, and Escherichia-Shigella) usually acted as module hubs to stabilize the modular structure, while the rare taxa usually acted as connectors of different modules. We conclude that abundant and rare taxa play different roles in the gut bacterial ecosystem. The conservation of some key bacterial species is essential for promoting the development of the gut microbiome in pandas. The living environment of the giant pandas can influence the internal structure, topological features, and strength of interrelationships in the gut microbiome. This study provides new insights into the conservation and management of giant panda populations.

Keywords: gut microbiome; keystone species; maintaining mechanism; module analysis; rare taxa; wild and captive giant panda.

FIGURE 1

Network co‐occurrence analysis of giant panda gut microbiome among three different populations, GPCAP (a), GPMS (b), and GPXXL (c). The nodes are colored according to different genera and modules, respectively. A connection represented a strong (Spearman's ρ ≥ 0.6 or ρ ≤ −0.6) and significant (p‐value ≤ 0.05) correlation. The size of each node was proportional to the number of connections (degree value). Each major module in GPCAP, GPMS, and GPXXL had more than 11 nodes, 55 nodes, and 49 nodes, respectively. Other modules included all small modules with ≤11, ≤51, and ≤32 nodes per module in GPCAP, GPMS, and GPXXL, respectively. Each circle represented one individual operational taxonomic unit (OTU). For each OTU, abundance was averaged over all samples from each population. Positive and negative correlations were shown as green and red edges, respectively.

FIGURE 2

Taxonomic composition of modules in terms of relative abundance of OTUs of phyla in gut microbiome network among three populations (GPCAP, GPMS, and GPXXL). The length of the bars on the outer rings and inner rings represented the percentage and relative abundance of each module type and gut microbiome group (phylum level) in their respective sections, respectively. Each phylum was represented by a specific ribbon color, and the width of each ribbon showed the abundance of each phylum in each module type. I–VI represented the six main modules of the microbial network of three giant panda populations (GPCAP, GPMS, and GPXXL), respectively.

FIGURE 3

Co‐occurrence patterns of OTUs in all giant panda samples. The nodes are colored according to different modules (a), genera (b), and relative abundance (c), respectively. Ternary plots displayed the relative abundance of OTUs from modules I to X in the three giant panda populations (d). A connection stands for a strong (Spearman's ρ ≥ 0.6 or ρ ≤ −0.6) and significant (p‐value ≤0.05) correlation. The size of each node was proportional to the number of connections (degree value). Each major module had more than 10 nodes. Other modules included all small modules with ≤10 nodes per module. Each circle represents one individual OTU. For each OTU, abundance was averaged over all samples from each population. Positive and negative correlations were shown as green and red edges, respectively.

FIGURE 4

Properties of the co‐occurrence network of OTUs based on correlation. (a) Roles of nodes from the giant panda populations (GP, combing GPCAP, GPMS, and GPXXL) in Zi‐Pi parameter space. Each node in a network could be characterized by its within‐module connectivity (Zi) and its among‐module connectivity (Pi). Nodes with Zi ≥2.5 and Zi <2.5 were classified as module hubs and nonhubs, respectively. Nodes were classified as module hubs (Zi >2.5, Pi <0.62), network hubs (Zi >2.5, Pi >0.62), peripheral nodes (Zi <2.5, Pi <0.62), and connectors (Zi <2.5, Pi >0.62). (b) Comparison of node‐level topological features (degree value, betweenness centrality value, closeness centrality value, and eigenvector centrality value) among three giant panda populations (GPCAP, GPMS, and GPXXL). The top and bottom boundaries of each box indicated the 75th and 25th quartile values, and the lines within each box represented the median values. *** indicated highly significant differences (p < 0.001).