Modelling Host-Microbiome Evolution

Bob Week - Kiel University

Who am I

  • Background: Mathematical evolution and ecology
  • PhD: BCB, UI, Scott Nuismer ~ Coevolutionary Theory
  • Postdoc 1: MSU, Gideon Bradburd ~ Host-Parasite Coevolution
  • Postdoc 2: UO, Brendan Bohannan ~ Host-Microbiome Evolution
  • Current: KiTE postdoctoral fellow with Schulenburg Group at Kiel University
    • Host-Microbiome Evolution
    • Two Secondments:
      • Regional non-academic (industry)
      • International academic
        • Hosted by Brendan Bohannan at UO
        • Purpose: travel to share science and meet scientists

We live in a microbial world

Andrew Rae for Quanta Magazine

Microbes interact with hosts

metabolism / digestion / nutrition

immune response / pathogen defense

development / life-history / phenology

behavior / neurophysiology

Microbiomes mediate host trait variation

Modified from Henry et al. (2021), Nat. Comm.

Microbiomes are heritable

  • Fraction of attributes heritable: ~5–40%
    • relative abundance, alpha diversity, etc …
  • Broad sense heritability: H² ≈ 0.01 - 0.5
    • perfect heritability = 1.0

Modified from Morris & Bohannan (2024), Nat. Micr.

Microbiome inheritance modes

Modified from
Morris & Bohannan (2024), Nat. Micr.



  1. Direct parent-offspring transmission

  2. Host genes associated with microbes

  3. Social transmission

  4. Shared environment

  5. Habitat modification

Can Microbiomes facilitate adaptation?

Modified from Uller & Helanterä (2017), Extended Evolutionary Synthesis blog

But Microbial Inheritance not Mendelian …


Modified from Miller, Svanbäck & Bohannan (2018), Trends Ecol. Evol.



  • Community assembly (microbial dispersal, within-host selection, etc …)

  • Problem: Complicated!

  • Need: Simple general frameworks

  • Q: Can we apply existing frameworks?

Extended inheritance frameworks

Modified from
Morris & Bohannan (2024), Nat. Micr.



  1. Maternal effect


  2. Indirect genetic effect


  3. Niche construction

Maternal Effect

Modified from
Week et al. (2024), Nat. Evo. Eco.

Definition: Parent trait explains offspring trait

(controlling for genetic variation)

Indirect Genetic Effect

Definition: Genotype of one individual influences trait of another individual

Modified from
Week et al. (2024), Nat. Evo. Eco.

Niche Construction

Modified from
Week et al. (2024), Nat. Evo. Eco.

Definition 1:

Organism activity alters environment

Definition 2:

Organism activity alters selection pressures

Two perspectives:

  1. Microbiome as environment

  2. Microbiome modifying environment

Conclusion


PROS

  • Maternal Effects ~ vertical parent-offspring inheritance
  • Indirect Genetic Effects ~ horizontal social inheritance
  • Niche Construction ~ interaction of microbiomes and selection


CONS

  • Each only cover one aspect
  • Microbiome inheritance follows several modes
  • Need: unified/sythetic framework

Unified Framework for Extended Inheritance

Modified from Day and Bonduriansky (2011), Am. Nat.


  • Models: ME’s, IGE’s, etc.
  • Limitation: requires assumptions on
    • transmission
    • fidelity
    • mutability

Framework for Microbiome Inheritance

Modified from Roughgarden (2023), Am. Nat
  • Collective inheritance facilitates adaptation
  • Advantage: Potential to explain adaptation without transmission details

A Gap

Done: Collective inheritance applied to gene-centric perspective 🦠 🚚 🧬

Gap: Connect concepts of microbiome inheritance and host trait variance to predict microbiome-mediated adaptation


microbiome inheritance

host trait adaptation

Is a Quantitative Genetic Approach Useful?





  • Explains genetic adaptation of complex traits ✅

  • Can account for trait variation explained by microbiomes ✅

  • Adaptation predictions assume Mendelian inheritance ❌

    • Interface with microbiome inheritance to close gap ✅

Classical Partitioning of Trait Variation

mini review of quantitative genetics




  • \(P = G + E\)
  • \(G = G_A + R\)
  • Adaptation Prediction:
    • Resp. to Sel. = \(G_A\) × Sel. Grad.




Additive Genetic Variance \(G_A\)

Partitioning of Variation including Microbiome




  • \(P_A = G_A + M_A + C_A\)
    • \(M_A=\) microbial variation
    • \(C_A=\) gene-microbe covariance
  • Adaptation Prediction:
    • Resp. to Sel. = \(P_A\) × Sel. Grad. ?

Modified from James et al. (2021), An Introduction to Statistical Learning

Not all microbes contribute to adaptation

Ancestral Concordance Classification

Lineal Microbes:

Ancestry overlaps with host ancestry

Non-lineal Microbes:

Ancestry in host population

Novel Microbes:

No previous ancestry with host population


classifies individual microbial cells

Partitioning Var. by Ancestral Concordance

\(M_L=\) lineal variance

\(M_N=\) non-lineal variance

\(M_V=\) noVel variance

  • (Additive Phenotypic Variation) \(P_A = G_A + {\color{magenta}{M_A}} + C_A\)
  • (Additive Microbial Variation) \({\color{magenta}{M_A}} = M_L + M_N + M_V + {\color{gray}{C_{LNV}} }\)

Hypothesis

  • Adaptation:
    • Lineal: \(M_L\) most likely
    • Non-Lineal: \(M_N\) less likely
    • NoVel: \(M_V\) no contribution
  • How to test:
    • Host Pedigree
    • Metagenomic Data for Microbiomes of Hosts in Pedigree
      • Strain Resolved
    • Classify Microbes by Ancestral Concordance
    • Partition host trait variance
    • Selection Experiment
      • Predict response including different combinations of \(M_L,M_N,M_V\)
      • Measure response to selection and assess predictions

Simulations

Non-lineals sampled before selection

[ 🖤 lineal | 🩶 non-lineal | 🤍 novel ]

Simple inheritance model:

  • Lineal abundances
    • sampled from parents
  • Non-lineal abundances
    • sampled from random host
    • before or after selection
  • Novel abundances
    • drawn independently

Additive host trait:

  • \(z=\sum_i\alpha_i\,n_i\)

Simulations

Non-lineals sampled after selection

[ 🖤 lineal | 🩶 non-lineal | 🤍 novel ]

Simple inheritance model:

  • Lineal abundances
    • sampled from parents
  • Non-lineal abundances
    • sampled from random host
    • before or after selection
  • Novel abundances
    • drawn independently

Additive host trait:

  • \(z=\sum_i\alpha_i\,n_i\)

Results

\(\pmb\Delta\mathbf{\bar z}\) = Resp. to Selection, G = Gen. var, L = Lineal, N = Non-lineal, V = NoVel

Conclusion: include only microbe variation shaped by selection

The Selective Microbes and Heritability

Definition: Subset of microbes that contribute to host adaptation

  • \(M_A^\psi\) = trait variance explained by variance of selectives
  • \(C_A^\psi\) = trait variance explained by covariance of selectives and genes

By definition: (Resp. to Sel.) = (\(G_A+M_A^\psi+C_A^\psi\)) × (Sel. Grad.)

Without microbe fx: (Resp. to Sel.) = (\(G_A\)) × (Sel. Grad.)

Narrow-sense heritability: \(h^2\) \(=\)\(\frac{G_A}{P}\), Resp. to Sel.: \(R=\) \(h^2\) \(S\)

Extending Heritability




Narrow-sense heritability: \[h^2=\frac{G_A}{P}\]

\(R=h^2\,S\)

(w/o micr. fx)

Narrow-sense total transmissibility:

\[t^2=\frac{M_A^\psi+C_A^\psi}{P}\]

\(R=t^2\,S\)

(w micr. fx)

Read all about it!




Week et al. (2025), Evolution

Thanks Co-Authors




Hannah Tavalie

Bill Cresko

Peter Ralph

Brendan Bohanna

What about Theory?

  • Quant. Gen. approach:
    • Statistical framework
    • Aimed at empirical studies
    • Still very complicated
  • Initial theoretical insights:
    • Simple models
    • Motivates some current projects…

Microbe-Mediated Host Rescue

Hildegard Uecker
MPI Plön

  • Scenario: Doomed host population

  • Microbiome: Presence/absence single microbe

    • Probability of lineal inheritance
    • Rate of social transmission

  • Fitness: Host gene determines microbe benefit

  • Results: Delayed host extinction time analysis

  • Birth-Death Diffusion \(dn=[(r+m)\,n+\psi]\,dt+\sqrt{n}\,dB\)

Microbiome-Mediated Niche Construction

Hinrich Schulenburg
Kiel University
  • Mesocosm experiments:
    • C. elegans shape microbiome in compost
    • Compost conditions degrade
    • Microbiome manipulation buffers against \(\Delta E\)
  • Quant. Gen. Theory
    • Modify \(\Delta\bar z=G\beta\)
      • Modelling framework (instead of statistical)
    • Microbiome community dynamics ~ mutation
    • Feedback with external microbiome
    • External microbiome changes effect of env.

Q: What eco-evo processes and what host-microbiome attributes shape adaptation via niche construction?

Multilevel Selection Framework

Arne Traulsen
MPI Plön
  • Mesocosm experiments:
    • C. elegans shape microbiome in compost
    • Compost conditions degrade
    • Microbiome manipulation buffers against \(\Delta E\)
  • MLS Model
    • Modify \(\Delta\bar z=G\beta\)
      • Modelling framework (instead of statistical)
    • Microbiome transmission + dynamics ~ mutation
    • Feedback with external microbiome
    • External microbiome changes effect of env.

Q: What eco-evo processes and what host-microbiome attributes shape adaptation via niche construction?

Social Microbiome Metacommunity Framework

Aura Raulo
University of Oxford
  • Unique Biological Scales:
    • Asymmetry: astronomical local abundances
    • Constrained Dispersal: social ntwk, discr. pulses
    • Transmission effect: composition, not abundance
    • Implications: unique biology, distinct diversity
    • Math: jump-diffusion processes
  • Viscocity Hypothesis:

Pathogens accumulate in
clusters

Mutualists accumulate in
bridges

Entangled Banks

Entangled Host-Microbiome Systems

Andrew Rae for Quanta Magazine