Microbial sensors in Fennoscandian voles

Gut microbiome similarity as a measure of contact in wild rodents

Rodent Ecology
Microbiome
Disease Ecology
Sweden
Fellowship

Funding: Birgitta Sintring Fellowship Institution: Department of Ecology and Genetics, Uppsala University Collaborators: Elin Videvall, Heikki Henttonen, Grimsö Wildlife Research Station Code: GitHub

Overview

This project develops a “microbial sensor”: a tracer that uses gut microbiome similarity to measure social and environmental contact in wild rodent populations. It combines field ecology, long-read sequencing and Bayesian modelling to ask how functional connectivity, the contacts that actually occur, governs the spread of environmentally persistent pathogens such as Puumala orthohantavirus through host populations. It continues the work on small-mammal contact networks in Sierra Leone.

The problem

Rodent contact networks emerge from social organisation and movement, not from density alone, yet most models approximate transmission potential with population density. Methods that measure contact directly, such as RFID logging and capture–mark–recapture, are labour-intensive. They also reduce contact to a binary record, which misses the intensity, timing and cumulative duration of exposure through shared nests and burrows.

The approach

Gut microbiota similarity can integrate a history of past associations, but earlier work could not separate direct contact from a shared environment. PacBio HiFi sequencing resolves shared taxa at strain level, so similarity can be split into two channels:

  • Social channel (direct contact). Obligate anaerobic, non-spore-forming taxa such as Bacteroidales, which survive poorly outside the host and so mark recent physical proximity.
  • Environmental channel (indirect contact). Aerotolerant or spore-forming clades, such as some Lachnospiraceae. Comparing faecal profiles with paired soil and nest samples separates transient environmental taxa from those that persist in shared infrastructure.

Work packages

Both work packages use cyclic Fennoscandian vole populations.

  • WP1, comparative model development. Sequencing of 200 high-information pairs of individuals from long-term monitoring at Pallasjärvi, Finland, and Grimsö, Sweden. Pattern-oriented modelling tests which microbial assembly patterns host density and spatial overlap alone cannot reproduce.
  • WP2, validation. An intensive longitudinal live-trapping grid at Grimsö, with RFID logging and capture–mark–recapture giving time-ordered contact, paired with repeated faecal and environmental sampling.

Model

A Bayesian hierarchical model partitions the microbial similarity between hosts i and j into contributions from social contact, space sharing, shared environment and kinship:

\[S_{ij} = \beta_0 + \beta_1 \text{SocialContact}_{ij} + \beta_2 \text{SpaceSharing}_{ij} + \beta_3 \text{Environment}_{ij} + \beta_4 \text{Kinship}_{ij} + \epsilon\]

The coefficients give the relative weight of direct and indirect pathways, and the residual captures individual variation and unobserved environmental effects. The result is a pathogen-agnostic way to infer weighted contact networks in wild populations.

Last updated 5 October 2026