Fig. 1 

A We are testing the impact of urbanization on the composition of small mammal communities. Additionally, we are investigating whether the diversity and composition of the bacterial gut microbiota are influenced by the sites sampled along the urbanization gradient or by the small mammal species in presence. B Decision tree for interpreting changes in bacterial composition to highlight the processes that may influence the bacterial gut microbiota and its response to urbanization. Taxonomic and functional compositions are represented by icons, with blue indicating a change and grey an absence of change. Yellow diamonds represent ecological processes, and grey rectangles indicate the statistical tests performed to infer underlying ecological processes. Step 1. The GLM & PERMANOVA tests assesses whether there is a change in GM diversity and composition between sites that could be associated with urbanization. If there is no taxonomic change, it suggests that the bacterial gut microbiota remains stable despite urbanization, whereas if there are changes, the GM is considered to be flexible. When taxonomic changes occur, but functional changes are not necessarily present or are weak, redundancy processes may underlie the assembly of microbial communities. Step 2. Redundancy analysis is applied to identify situations where bacterial taxa express the same function (considered redundant) or where each taxon expresses its own function (not redundant). This test is followed by a null model test (NTI) to determine the mechanisms that may underlie the redundancy. These mechanisms can be selective processes through overdispersion (indicating competition between bacterial taxa) or underdispersion (indicating cooperation between bacterial taxa), resulting from abiotic or biotic effects among bacterial taxa. Additionally, it is worth noting that phylogenetic dispersion between bacterial taxa can also be influenced by neutral effects. Otherwise, when taxonomic and functional changes occur, stochastic or deterministic processes may underly GM assembly. Step 3. The betadisper test examines differences in intragroup variances. High variance may result from strong selective pressures, such as dietary variance, or stochastic effects indicating processes favoring dysbiosis (see Anna Karenina’s principle, [117]). Conversely, healthy species tend to express similar essential functions, resulting in lower variance. The null model helps to determine whether GM assemblage is primarily driven by stochasticity or determinism. Step 4. We use DESeq2 test to identify the functions that may be subject to selection. A heatmap enables to illustrate these functional changes between different host species and sites, reflecting the different levels of urbanization. The colored rodent icon indicates the likely impact of GM changes on its health. Grey represents a neutral effect, red a detrimental effect and green a beneficial effect