Skip to content
MC

ResearchThree areas

Studying what cannot always be observed directly.

Mathematics, statistics and machine learning make it possible to explore mechanisms that are hard to isolate in experiments: population dynamics, biological invasions, plant responses to stress.
  1. Simulate

    Turn a biological mechanism into equations or agents, then run it.

  2. Test the mechanisms

    Vary a parameter, a behavioural rule or a release scenario, and watch what changes.

  3. Interpret

    Relate the results back to biology, and state their conditions of validity and their limits.

A

Population dynamics and biocontrol

The doctoral work. Under what conditions do releases of sterile males reduce a pest population, when sterilisation is imperfect (residual fertility), when females mate several times (polyandry), or when releases are periodic and limited by a budget?

Methods: Differential equations · Bifurcation analysis · Impulsive models · Individual-based models

  • Python
  • SciPy
  • SymPy
  • python-control
  • Mesa
  • NumPy
  • pandas
  • statsmodels
  • seaborn

B

Invasiveness of a species and its functional response

Do invasive species consume more prey than other species, and does that measure predict their impact? A statistical analysis of functional responses already published for 45 freshwater fish species, carried out with Chloé Souques, Yann Voituron, Loïc Teulier, François-Xavier Dechaume-Moncharmont (LEHNA, Université Claude Bernard Lyon 1) and Vincent Médoc (ENES, Saint-Étienne).

Methods: Comparative analysis of published data · Bayesian mixed models (MCMCglmm) · Phylogenetic correction

  • R

CIn progress

Integrating multi-omics data to study plant responses to multiple stresses

Work in progress. How does tomato respond to a stress that combines heat and salinity? Transcriptomic and metabolomic data are integrated with deep learning models, including a variational autoencoder (VAE), built with PyTorch. No result has been published yet.

Models and tools

  • Deep learning
  • Variational autoencoder (VAE)
  • Transcriptomics
  • Metabolomics
  • PyTorch
  • Python

No publication or result yet: this work is in progress.

The work in figures.

Each figure illustrates part of areas A and B, with the source it comes from: a thesis chapter or an article.
  1. Area AThesis, chapter 2 · Bulletin of Mathematical Biology 2025 · PLOS Computational Biology 2026

    The sterile insect technique

    Releasing sterilised males to disrupt the reproduction of a pest.

    The sterile insect technique relies on releasing sterilised individuals to reduce the reproduction of a target population. How well it works depends on many biological and technical factors, whose effects can be explored with models.

    Read the detail: The sterile insect technique
    Bifurcation diagram: equilibrium density of females as a function of the density of sterilised males, with a stable branch, an unstable branch and a saddle-node bifurcation point at S_c.
    Thesis figure — fig_chapter2_thesis.ipynb
  2. Area AThesis, chapter 3 · Bulletin of Mathematical Biology 2025

    Residual fertility

    When sterilisation is imperfect.

    Sterilisation can be imperfect. Residual fertility in the released insects can change the outcome of control strategies. This work examines when these effects become decisive.

    Read the detail: Residual fertility
    Female density curves over 300 days: without releases (constant), and with low or high releases, with or without 1 % residual fertility.
    Thesis figure — Script_Manuscrit_SIT_RF.ipynb
  3. Area AThesis, chapter 4 · PLOS Computational Biology 2026

    Multiple matings

    Polyandry, re-mating and sperm use.

    In some species, females can mate several times. Re-mating and the way sperm is used can then influence how well sterile insect releases work.

    Read the detail: Multiple matings
    Two panels of larval density over time for three release intensities, comparing the “First” (blue) and “Last” (orange) sperm-use scenarios under high (A) and low (B) initial infestation.
    Thesis figure — Compartmental_model_output_vAvril.ipynb
  4. Area AThesis, chapter 4 · PLOS Computational Biology 2026

    Individual-based modelling

    Representing each insect, its behaviour and its interactions.

    Individual-based models represent each organism, its age, its matings and its egg laying. They can include individual variability and behavioural rules that are hard to express in an aggregated model.

    Read the detail: Individual-based modelling
    Left, stochastic trajectories of the number of larvae over 100 days for several sperm-use rules; right, reduction of the area under the curve for each scenario.
    Thesis figure — ABM_model_output.ipynb
  5. Area AThesis, chapter 5

    Release strategies and deployment constraints

    Frequency, size and a limited budget.

    In practice, sterile males are released at discrete times, with a limited budget. The thesis chapter on the agricultural context studies how to split that budget between the frequency and the size of releases.

    Read the detail: Release strategies and deployment constraints
    Population of sterilised males under periodic releases every 7 days: the simulation converges to the analytical periodic solution.
    Thesis figure — Notebook_part_impulsive_releases.ipynb
  6. Area BJournal of Animal Ecology 2026

    Invasion ecology

    Functional responses, invasiveness and trophic impact.

    Studying functional responses helps understand how resource consumption, species traits and the ability to become invasive are linked. The geographical and ecological context can change how these links are read.

    Read the detail: Invasion ecology