Human encroachment into wildlife gut microbiomes

The study investigates whether changes in wildlife gut microbiomes are driven by habitat fragmentation alone or by fragmentation combined with additional anthropogenic disturbances (e.g., contact with humans, domesticated animals, invasive species, and their pathogens). Given the critical role of gut microbiota in host nutrition, immunity, and disease resistance, and the increased frequency of zoonoses in a globalized world, disentangling these effects is important for conservation biology and disease ecology. The focal system is Tome’s spiny rat (Proechimys semispinosus), a generalist rodent and potential reservoir host, sampled across Panamanian landscapes that differ in fragmentation and human disturbance. The central hypothesis is that habitat fragmentation per se does not alter gut microbiome diversity and structure, whereas fragmentation combined with additional anthropogenic disturbance will reduce alpha diversity, shift beta diversity, increase dispersion among individuals, and alter predicted functional potential.

Prior work highlights that anthropogenic changes (habitat reduction, isolation, and altered host-environment interactions) can influence disease dynamics and wildlife health. Generalist species may act as reservoirs for pathogens and often persist in disturbed habitats. The gut microbiome is integral to host health, shaping nutrition and mucosal immunity; disturbances beyond normal variation (dysbiosis) are associated with disease susceptibility. However, adaptive potential of the microbiome (metagenomic plasticity) under human-driven changes remains underexplored. Literature points to both stochastic and deterministic processes in shaping microbial beta diversity and notes the Anna Karenina Principle, where stressed hosts exhibit more variable microbiomes. Studies of other taxa (e.g., primates, bats) suggest land-use change impacts gut microbiota, but whether fragmentation alone versus additional human disturbance drives microbiome shifts had been unclear.

Study design: 384 Tome’s spiny rats (Proechimys semispinosus) sampled in the Panama Canal area across three landscape types: (C) protected continuous tropical forests; (I) protected forest islands formed by canal construction (fragmentation without additional human activity); (A) forest fragments at uncultivated edges embedded in an agricultural matrix representing fragmentation plus additional anthropogenic disturbance (contact with humans, domesticated animals, invasive species, pathogens). This design allows separating effects of fragmentation per se from added disturbance. Sampling and lab procedures: Fecal samples collected from wild individuals. DNA extracted with NucleoSpin Soil Kit (bead-beating via SpeedMill PLUS; extraction blanks included). Bacterial 16S rRNA gene V4 region amplified using primers 515F/806R with a two-step PCR and barcoding (Fluidigm Access Array), followed by Illumina sequencing. Bioinformatics: Reads processed in QIIME 2 (versions 2018.6–2020.8) using DADA2 for primer removal, denoising, chimera removal, and ASV inference/merging. Taxonomy assigned with a pre-trained classifier (SILVA-based). Outputs (feature table, taxonomy, rooted phylogeny) imported into R (phyloseq) for downstream analyses; contaminants and controls evaluated. Diversity analyses: Alpha diversity metrics included observed ASVs, Shannon diversity, and Faith’s Phylogenetic Diversity (PD). Generalized linear mixed models (GLMMs; lme4/glmmTMB) modeled alpha diversity with fixed effects (landscape, sequencing depth, season/year) and random effects (site nested within landscape; extraction batch), using an information-theoretic (AIC/AICc) approach for model selection; marginal and conditional R² reported; spatial autocorrelation considered. Beta diversity and dispersion: Phylogeny-aware distances computed using weighted and unweighted UniFrac. PERMANOVA (vegan::adonis) tested compositional shifts with permutations constrained by site within landscape (strata). Cohen’s d effect sizes reported using first two PCoA axes. PERMDISP2 evaluated differences in dispersion (heterogeneity of variances) among landscapes with site effects considered. Differential abundance: ANCOM identified differentially abundant ASVs between disturbed (A) and the combined protected landscapes (C and I), controlling for compositionality and multiple testing; threshold w ≥ 0.95. Functional potential: PICRUSt2 predicted metagenomes and MetaCyc pathway abundances. Differential pathway abundance tested similarly to ASVs, identifying pathways over-represented in disturbed vs protected landscapes; stratified outputs linked ASVs to pathways. Pathways not known in bacteria were removed. Reproducibility: Analyses conducted in R; code availability noted; sequencing data deposited on NCBI BioProject.

- Habitat fragmentation per se did not alter within-host gut bacterial diversity, but fragmentation combined with additional anthropogenic disturbance did. Alpha diversity (observed ASVs, Shannon, and Faith’s PD) was significantly lower in the anthropogenically disturbed fragments (A) than in protected continuous forests (C) and protected forest islands (I). Model selection supported a landscape effect: observed ASVs (ΔAICc = 19.63; R²m = 0.473; R²c = 0.592), Shannon diversity (ΔAICc = 5.84; R²m = 0.115; R²c = 0.171), Faith’s PD (ΔAICc = 6.50; R²m = 0.200). The effect was less pronounced when considering abundance (Shannon), suggesting rare taxa disproportionately contributed to the decline. - Beta diversity shifted in disturbed fragments (A), indicating compositional changes relative to protected landscapes. PERMANOVA on UniFrac distances showed significant effects of landscape: weighted UniFrac p < 0.001 (999 permutations), R² = 0.142, F = 37.00; unweighted UniFrac p < 0.001, R² = 0.652, F = 13.17. Effect sizes (Cohen’s d) indicated strong separation along PCoA axes for both weighted and unweighted metrics. - Dispersion (heterogeneity among individuals) was greater in disturbed fragments (A), consistent with the Anna Karenina Principle. PERMDISP2 supported increased dispersion for both weighted and unweighted UniFrac distances when accounting for site effects. - Differential abundance (ANCOM) identified 142 ASVs (of 4213) differing between protected (C+I) and disturbed (A) landscapes. Taxa over-represented in protected landscapes included SCFA-producing groups such as Clostridium sensu stricto 1, Roseburia, and the Eubacterium coprostanoligenes group. Taxa over-represented in disturbed landscapes included Gastranaerophilales, Acidobacteria, and members of Mollicutes RF9; several of these are associated with domesticated animals and include potential pathogens. - Predicted functional potential shifted: Protected landscapes had higher abundance of pathway 2701 (pyrimidine deoxynucleotide biosynthesis from nucleosides). Disturbed landscapes showed enrichment of pathway 1861 (formaldehyde assimilation II—RuMP cycle), pathway 2941 (L-lysine biosynthesis II), and teichoic acid (poly-glycerol) biosynthesis—functions implicated in stress responses, cell wall biogenesis, and potential pathogenesis. - Together, results indicate that added human disturbance, not fragmentation alone, reduces alpha diversity, shifts composition, increases inter-individual variability, introduces/boosts taxa linked to domestic animals and potential pathogens, and alters predicted metabolic functions.

Findings demonstrate that, in a generalist rodent, gut microbiomes are resilient to fragmentation per se but are disrupted when fragmentation is combined with additional anthropogenic disturbances. Lower alpha diversity and higher beta-dispersion in disturbed landscapes suggest reduced microbiome stability and resilience, aligning with the Anna Karenina Principle where stressed hosts exhibit more idiosyncratic microbiomes. Compositional shifts appear driven more by changes in relative abundances of taxa already present (deterministic filtering and/or stochastic colonization/extinction) than by wholesale gain/loss of taxa, consistent with metagenomic plasticity. Enrichment of taxa associated with domesticated animals and potential pathogens in disturbed sites underscores risks of pathogen spillover and altered host-pathogen dynamics as humans and domestic species encroach on wildlife habitats. Predicted functional changes (e.g., teichoic acid biosynthesis, RuMP formaldehyde assimilation, lysine biosynthesis) may reflect stress adaptation, alterations in substrate availability near agricultural matrices, and potential increases in pro-inflammatory or pathogenic traits. These microbiome alterations have implications for wildlife health and zoonotic risk, emphasizing the importance of mitigating human-wildlife overlap and monitoring gut microbiome health as an early warning indicator.

The study disentangles the effects of habitat fragmentation from additional anthropogenic disturbances on a wildlife gut microbiome. Fragmentation alone did not affect gut microbial diversity or structure, whereas added human disturbance reduced alpha diversity, shifted and dispersed beta diversity, increased representation of taxa associated with domestic animals and potential pathogens, and altered predicted metagenomic functions. While patterns suggest some metagenomic plasticity, increased dispersion and loss of presumed beneficial taxa indicate potential maladaptation and reduced resilience. Future research should (1) validate functional predictions with shotgun metagenomics and experimental studies, (2) assess temporal dynamics and causality, (3) expand to other wildlife species and ecosystems to test generality, and (4) integrate host health metrics and pathogen surveillance to evaluate consequences for zoonotic risk.

- Functional inferences are based on 16S rRNA gene profiles (PICRUSt2) and thus are predictive rather than direct measurements; species/strain-level resolution is limited, potentially obscuring pathogenic vs commensal lineages. - Cross-sectional sampling across seasons/years and sites, despite mixed modeling and permutation constraints, cannot fully exclude unmeasured confounders. - Some methodological details (e.g., OCR errors) indicate potential reporting ambiguities; however, core analyses (GLMMs, PERMANOVA/PERMDISP2, ANCOM) are standard and robust. - Differential abundance and pathway analyses were constrained by compositional data considerations and dataset partitioning for computational reasons; rare taxa may be underpowered. - Study focuses on a single generalist rodent in one geographic region; generalizability to other taxa and biomes requires further study.