**Background:** Ecological corridors are a well-established conservation tool to mitigate habitat fragmentation by facilitating dispersal and species coexistence among macroorganisms. However, whether corridors influence microbial communities—which constitute the majority of biodiversity—remains largely unknown. Microorganisms, particularly root-associated endospheric fungi, may experience dispersal limitation at scales under 1 meter, and host plants may serve as preferential microhabitats. This study tested whether corridors composed of host plants (biotic corridors) shape the diversity, composition, and assembly processes of root endospheric fungal communities.
**Methods:** The authors established 20 mesocosms (1.30 m × 1.30 m) filled with homogenized substrate (20% sand, 80% silty clay soil) in a common garden at the University of Rennes 1, France. Each mesocosm contained two patches (diameter 0.40 m) of Trifolium repens (Fabaceae) embedded in a matrix of Brachypodium pinnatum (Poaceae). In 10 mesocosms, the T. repens patches were connected by a narrow corridor (0.90 m × 0.15 m) of T. repens; 10 mesocosms had no corridor (one was lost due to poor T. repens growth). The experiment began June 9, 2017, with sampling at five time points: October 2017 (t0+4 months), May 2018 (t0+11 months), June 2018 (t0+12 months), October 2018 (t0+16 months), and May 2019 (t0+23 months). At each time point, three T. repens ramets were collected from each patch (plus two from corridors where present) and two B. pinnatum ramets from the matrix, yielding 670 T. repens and 190 B. pinnatum samples. Root DNA was extracted, and the fungal SSU rRNA gene V4-V5 region was amplified and sequenced. After bioinformatics processing (Frogs pipeline, Swarm clustering, chimera removal), sequences were normalized to 4,292 reads per sample, yielding 266 fungal sequence clusters. Statistical analyses included linear mixed-effects models (LMMs/GLMMs) for diversity indices, PERMANOVA for community structure, Bray-Curtis dissimilarity for beta-diversity, Sloan neutral community models (NCMs) to partition stochastic vs. deterministic assembly, and functional guild assignment via FUNGuild, FUN^FUN, and FungalTraits databases.
**Key Results:** (1) T. repens and B. pinnatum harbored distinct root mycobiota (PERMANOVA, P < 0.001), confirming host preference. (2) Connected T. repens individuals showed higher fungal sequence cluster richness than isolated individuals at June 2018 and October 2018 (individual scale, P < 0.05); at the patch scale, richness was higher in connected patches in October 2018 (P < 0.05). (3) Bray-Curtis dissimilarity between patches was significantly lower in connected vs. isolated treatments at June 2018, October 2018, and May 2019 (patch scale, P < 0.05 to P < 0.001), indicating homogenization. Dissimilarity in connected patches decreased over time, while isolated patches remained stable. (4) Neutral community models showed that stochastic assembly predominated overall (46–51% of sequence clusters in connected, 51–57% in isolated). Connected patches had a lower proportion of stochastic sequence clusters and a higher proportion of overrepresented (deterministically selected) taxa, while isolated patches had more underrepresented (drift-prone) taxa. The Nm parameter (dispersal estimate) was higher in connected treatments. (5) Overrepresented taxa in connected conditions included 11 sequence clusters related to Mycoemilia scoparia, Rhizidium endosporangiatum, Rozella allomycis, and Triparticalcar arcticum. Underrepresented taxa in isolated conditions included sequence clusters related to Agaricus bisporus, Anomoporia bombycina, Geopyxis majalis, and Ambispora fennica. (6) Functional guild analysis (138 of 266 sequence clusters assigned) revealed no significant corridor effect on plant pathogen richness or abundance. Symbiotroph richness increased with corridor connection at individual scale (June 2018) and patch scale (October 2018). Saprotroph richness was higher with corridors at individual scale (June 2018, October 2018) and patch scale (October 2018).
**Clinical Implications:** While this is an ecological study without direct clinical application, the findings have translational relevance for agriculture and human health. Plant-associated microbiota influence host nutrition, stress tolerance, and pathogen resistance. Understanding how landscape connectivity shapes microbial communities could inform strategies to promote beneficial symbionts (e.g., mycorrhizae) or limit pathogen spread in agroecosystems. The concept of 'biotic corridors'—where host plants themselves serve as dispersal conduits for microbes—offers a framework for managing microbial ecosystem services. The absence of a corridor effect on plant pathogens is reassuring but warrants further study, as corridors could theoretically facilitate pathogen dispersal. These results also underscore that microbial biogeography operates at very small spatial scales (<1 m), challenging the assumption of ubiquitous microbial dispersal.