Macroevolutionary patterns (e.g. rate of diversification, species-area relationship) ultimately result from microevolutionary mechanisms (competition, trait evolution, etc.). Nevertheless, the causal links between micro- and macro-timescales are usually very difficult to establish from empirical patterns. This explains why many macroevolutionary studies suggest historical scenarios compatible with the observed current patterns but do not prove that it is indeed what happened. I will discuss how models can be used to mechanistically link micro- and macroevolution. I will discuss how far such explicit link allows to establish causal relationships between microevolutionary processes and macroevolutionary patterns. I will illustrate the discussion with an individual-based model of diversification combining ecological, genetic and geographical causes of speciation and extinction, and producing macroevolutionary outputs such as phylogenetic trees.

Heterogeneous and complex molecular networks control cellular processes. While tremendous technical progresses led to the identification of molecular interactions, computational modelling is required to make sense of these data. I will first discuss the notion of molecular networks, the variety of related semantics, and provide a short overview of modelling approaches. I will then focus on the notion of (discrete) dynamical systems, on different types of properties and their biological relevance.

How mathematical models link with experimental data, with inputs from mathematics and physics to help us understand how biodiversity affects ecosystem functioning & stability. I’ll outline some of the research carried at the Centre for Biodiversity Theory and Modelling in Moulis.

The increase of atmospheric carbon dioxide is a cause of global climate change, yet the response of vegetation to ongoing changes is poorly resolved. Currently, about 25% of fossil fuel emissions are sequestered by vegetation, another 25% by the oceans and the rest, only 50%, is emitted into the atmosphere. It is therefore crucial to evaluate the futures of the carbon sequestration potential for lands and oceans. One approach to address this challenge is to develop models that combine the state of the art in our knowledge of plant physiology, water cycle, and remote sensing. Historically, these models have been seen as box models, where fluxes are modelled through a combination of empirical equations. This has resulted in largely phenomenological models and cumulatively complex ones. Even the cornerstone of these approaches, the Farquhar-von Caemmerer-Berry model of leaf carbon economy, is in part empirical. An even more serious drawback is that the carbon stock of an ecosystem has long been modelled as proportional to the ecosystem's uptake potential. This assumption is reasonable in annual systems, such as agricultural lands or savannas, but seriously departs from reality in forests. Here I contend that individual-based models of forest dynamics lift a number of assumptions in classic vegetation models, and provide a much more straightforward approach to the modelling of forest carbon dynamics. I will discuss the cost related to modelling trees one by one, versus the alternative of coarse-grained approaches. I will also discuss the implications of being able to track individuals through time explicitly, e.g., to explore the response of species to environment shifts. This study provides an example of how non-trivial macroscopic patterns emerge from microscopic individual-based models.

A fundamental question in ecology and evolutionary biology is the generation and maintenance of biodiversity. In epidemiology a parallel challenge concerns mechanisms underlying polymorphism in pathogens and microbes. Here I will show how with mathematical modelling we can address diversity and coexistence in co-colonizing microbial systems. I will describe the multi-type interactions, ranging from cooperation to competition, that arise in microbial co-colonization among N closely-related strains. Next, I will present an analytic framework for simplifying and understanding better such a system based on slow- fast dynamics. This framework can predict explicitly strain coexistence from the co-colonization interaction matrix, and enable very efficient computation of the dynamics for arbitrary number of strains. I will also discuss links with niche and neutral theory arising from this analysis and potential applications.

This presentation is focused on the relation between a novel of Stanislas Lem, Solaris, and the work of the geochemist James Lovelock. Behind the strong differences between a fiction based on the rule of suspension of disbelief, and a scientific assumption compatible with the rule of truth, we will show Gaia and Solaris are based on two common philosophical statements: life must be understood, not simply at the level of the organism, but also at the level of the biosphere; and life is some kind of primitive mind without any intentional purpose, and then without consciousness. Gaia knows how to proceed in order to regulate geochemical conditions that permit its own existence, even if it cannot know that it has this ability and this skill.

Cells display individual or collective behaviours that are described as analogous to decision making. At the individual level the capacity of spermatozoa of marine invertebrates to find their conspecific eggs during marine broadcast spawning events represents a form of mating choice. The ability of the lymphocytes of the adaptive immune system to tolerate body tissues while fighting infection and rejecting allografts is understood as a collective discrimination between self and not self. I will first overview our efforts to understand these two paradigmatic cell behaviours via mathematical modelling, simulations and artificial system engineering. Based on this overview, I will then discuss the epistemic value of analogies, simulations and models.

Models in biology are notoriously false. This property of models raises a subtle philosophical question: how can models be scientifically virtuous, in spite of being false? Or, rather, aren't models virtuous thanks to their being false?

Genetic and genomic data are increasingly used to reconstruct the demographic history of species. Some studies suggest that humans went through a bottleneck and expansions. Others claim that (some) humans admixed with Neanderthals. The aim of this talk is to try and discuss how the assumptions made (or the models used) by different authors may lead them to make (very different) statements on the evolution of humans, on the basis of genomic data. We will discuss in particular the importance of models that incorporate structure (the fact that mating is geographically constrained) and models that ignore structure (i.e. assume that population structure can be ignored for some inference). In a period where some studies are questioning the scenario whereby humans evolved in Africa, it is important to understand what genetic data allow to say and what they cannot allow us to say.

In his 1950, Turing proposed an "imitation game". The idea was to fool a person who, by asking questions via a teleprinter, seeks to establish whether the respondent is a woman or a machine. In no way Turing tried by this to find out how a human brain works, or to make a mathematical model of the brain. Turing's 1952 paper on morphogenesis, instead, describes a model of an action/ reaction/diffusion (non-)linear system that may allow to understand the genesis of (bio-chemical) forms. Can this help to understand today the differences between phenomenological modeling, analytic and computational simulations ? The proposal of (geometric) ''schemata'' is yet another use of mathematical tools for intelligibility which allowed us to de-spacialise biological time, in contrast to the prevailing physical modeling of time, increasingly identified with or subordinated to space, from Aristotle to Galileo and Einstein.
References
- https://www.di.ens.fr/users/longo/download.html:
G. Longo. Letter to Alan Turing. Invited, in Theory, Culture and Society, Posthumanities Special Issue, 2018
- https://www.di.ens.fr/users/longo/files/Letter-to-Turing.pdf
Francis Bailly, Giuseppe Longo, Maël Montévil. A 2-dimensional Geometry for Biological Time. In Progress in Biophysics and Molecular Biology: vol. 106, n. 3, pp. 474 – 484, 2011
- https://www.di.ens.fr/users/longo/files/CIM/2-dimTime.pdf