Evolution of a trait distributed over a large fragmented population: Propagation of chaos meets adaptive dynamics

Amaury Lambert, Hélène Leman, Hélène Morlon, Josué Tchouanti · 2025

Read Evolution of a trait distributed over a large fragmented population: Propagation of chaos meets adaptive dynamics by Amaury Lambert, Hélène Leman, Hélène Morlon, Josué Tchouanti (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-13154.txt.

Introduction

Amaury Lambert and co-authors model evolution across a split population in a 2025 math paper. Imagine many islands, each home to a fixed crowd carrying a heritable trait. Migrants and mutants arrive slowly, so each island is nearly uniform at any moment. The authors ask what happens to the trait map as the number of islands grows without bound.

In that limit, the whole map follows a fixed equation rather than chance. Any few islands then evolve on their own, since each migrant is drawn from the global mix. Under even dispersal, the trait at any spot obeys a random process they pin down. The bridge between island-level chance and continent-level law serves ecology and genetics alike.

Catalog record

CREATORAmaury Lambert, Hélène Leman, Hélène Morlon, Josué Tchouanti
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE65 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-13154

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Evolution of a trait distributed over a large fragmented population: Propagation of chaos meets adaptive dynamics Amaury Lambert, Hélène Leman, Hélène Morlon, Josué Tchouanti Abstract We consider a metapopulation made up of K K demes, each containing N N individuals bearing a heritable quantitative trait. Demes are connected by migration and undergo independent Moran processes with mutation and selection based on trait values. Mutation and migration rates are tuned so that each deme receives a migrant or a mutant in the same slow timescale and is thus essentially monomorphic at all times for the trait (adaptive dynamics). In the timescale of mutation/migration, the metapopulation can then be seen as a giant spatial Moran model with size K K that we characterize. As K → ∞ K\to\infty and physical space becomes continuous, the empirical distribution of the trait (over the physical and trait spaces) evolves deterministically according to an integro-differential evolution equation. In this limit, the trait of every migrant is drawn from this global distribution, so that conditional on its initial state, traits from finitely many demes evolve independently (propagation of chaos). Under mean-field dispersal, the value X t X_{t} of the trait at time t t and at any given location has a law denoted μ t \mu_{t} and a jump kernel with two terms: a mutation-fixation term and a migration-fixation term involving μ t − \mu_{t-} (McKean-Vlasov equation). In the limit where mutations have small effects and migration is further slowed down accordingly, we obtain the convergence of X X , in the new migration timescale, to the solution of a stochastic differential equation which can be referred to as a new canonical equation of adaptive dynamics. This equation includes an advection term representing selection, a diffusive term due to genetic drift, and a jump term, representing the effect of migration, to a state distributed according to its own law. Keywords: Moran model \cdot Metapopulation \cdot Propagation of chaos \cdot Adaptive dynamics \cdot Trait substitution sequence \cdot Canonical equation \cdot Migration \cdot Diffusion with jumps \cdot Ecology \cdot Macroevolution MSC 2020: 60F17 \cdot 60J25 \cdot 60J70 \cdot 60J76 \cdot 60K35 \cdot 35Q70 \cdot 92D15 \cdot 92D40 1 Introduction To understand the origin of the diversity, through time and among taxa, of a phenotypic trait (body mass, beak shape, leaf size, wing color…) is one of the central questions in evolutionary biology. Several approaches to address this question exist, depending on the scale studied. On the slowest timescale, i.e., the scale of paleontology and macroevolution, the evolution of a trait, or rather of its species/population average, is frequently modeled by a stochastic process such as Brownian motion [ Fel73 , HM96 ] , the Ornstein-Uhlenbeck process [ Han97 , BK04 ] or a Lévy process [ LSL13 , MMML20 ] . This top-down approach notably serves in comparative phylogenetic methods for the inference of macro-evolutionary dynamics from the knowledge of traits of present or fossil species, see [ PH13 ] for a review. On the fastest timescale, i.e., the scale of a few generations, the distribution of a quantitative trait in a population is often assumed to be normal and its evolution is described through the dynamics of its mean and variance. In quantitative genetics, the value of a trait is the sum of an environmental component and of a genetic, heritable component, often modeled as the sum of allelic effects at different loci. Selection on the trait (assumed here to be one-dimensional) is typically modeled by a Gaussian-like fitness function of the trait (stabilizing selection) with variance \omega^{2} and a possibly moving optimum (directional selection). Russell Lande and his coauthors showed in a series of foundational papers [ Lan76 , Lan79 , LA83 , BL94 ] that under these normality assumptions, the average trait follows an autonomous equation pushing it toward the optimum, with a Gaussian deviation with second moment A/N due to genetic drift, where A is an increasing function of the additive genetic variance \sigma_{g}^{2} , with A(0)=\omega^{2}/2 (in the absence of environmental noise) and A(x)\sim x when x is large. The additive genetic variance \sigma_{g}^{2} measures the degree of heritable polymorphism in the population. Its value at equilibrium decreases with population size/strength of selection (i.e., increases with \omega ) and increases with mutation rate. It is essentially zero when selection is strong enough to deplete diversity and mutation rate is too small to replenish it. In this latter case, one observes a separation of timescales: ecological dynamics occur quickly while mutations are rare, so that only a few trait values coexist at any given time. The theory of adaptive dynamics pushes this assumption to the extreme case where fixation occurs before the next mutation event and the average trait becomes the only trait value in the population, except during the short period when the resident trait and the mutant trait compete. Under this assumption, Metz et al [ MGM+95 ] and Dieckmann & Law [ DL96 ] derived what is known as the canonical equation of adaptive dynamics, which describes the evolution of the so-called dominant trait in the timescale of mutations. Later, Champagnat et al [ CFM06 ] and Baar et al [ BBC17 ] formally derived this equation in the context of large populations. At the cost of suppressing all the information on polymorphism, this alternative approach offers several benefits: 1) by considering two-player games (resident vs. mutant), it allows for richer ecological dynamics than stabilizing selection resulting from an absolute fitness function, 2) the fitness of the mutant in the resident background emerges naturally from the ecology (see below) rather than being given a priori, and 3) the evolution of the dominant trait in the population can be rigorously derived mathematically, without the normality assumptions required for the average trait in a polymorphic population to have autonomous dynamics. See the companion paper [ LLMV24 ] for a tentative synthesis. Although adaptive dynamics theory traditionally deals with large populations, the assumption that fixation occurs rapidly compared to mutations should be more relevant for small populations. Champagnat & Lambert [ CL07 ] extended the approach of adaptive dynamic to finite populations, where the possible fixation of traits with suboptimal fitness introduces stochasticity into the canonical equation of adaptive dynamics. Here, the fitness of the mutant is its probability of fixation in a resident population at stochastic equilibrium. In the limit of small mutations, trait dynamics are then described by a diffusion process, called the canonical diffusion of adaptive dynamics, with a similar advection term as derived by Dieckmann & Law [ DL96 ] , as well as a diffusion term due to genetic drift. As in [ Lan76 ] , this diffusion term scales like 1/N , but it depends on the covariance matrix of the mutation kernel rather than on the Hessian of the fitness function at the optimum. The two approaches (quantitative genetics and the canonical diffusion of adaptive dynamics) have the merit to unveil the micro-evolutionary underpinnings of trait evolution but suffer from inherent contradictions regarding population size: The first one (quantitative genetics) needs population size to be both finite (for genetic drift to play a role) and infinite (for normality of trait distribution to hold, which is required for the

Read the complete text