🧬 biology

A stochastic population model with hierarchic size-structure

This paper establishes that a deterministic delay equation model for a hierarchically structured population accurately approximates the corresponding individual-based stochastic model in the large population limit, specifically by demonstrating that the deterministic stationary birth rate converges to the stochastic quasi-stationary birth rate.

Original authors: Carles Barril, Àngel Calsina, József Z. Farkas

Published 2026-07-03

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Carles Barril, Àngel Calsina, József Z. Farkas

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Imagine a crowded room where everyone is trying to grow taller. In this room, your ability to grow depends entirely on how many people are taller than you. If you are the shortest person, you have plenty of room to grow. But if you are the second shortest, the tallest person is "stealing" some of your resources, slowing your growth down. The more people taller than you, the slower you grow.

This paper is about building a mathematical model to understand how such a population behaves over time. The authors compare two different ways of looking at this problem:

1. The Two Ways of Looking at the Crowd

The "Smooth" View (Deterministic Model):
Imagine looking at the crowd from a high-altitude helicopter. You can't see individual people; you just see a smooth, flowing river of people. You assume the number of people is so huge that random fluctuations don't matter. You use a set of rules (equations) to predict exactly how the "average" person grows and how many babies are born. This is the Deterministic Model. It's like predicting traffic flow on a highway: you assume a steady stream of cars.

The "Pixelated" View (Stochastic Model):
Now, imagine you are standing in the middle of the crowd. You see individual people. Sometimes, just by bad luck, a tall person might trip and die early. Sometimes, a short person might get lucky and have a baby sooner than expected. The total number of people jumps up and down randomly. This is the Stochastic Model. It treats every person as a unique agent with a random life span and random birth events. It's like watching a chaotic dance floor where everyone moves slightly differently.

2. The Big Question: Do They Agree?

The authors wanted to know: If the crowd gets huge, does the "Pixelated" view eventually look exactly like the "Smooth" view?

In many simple models, the answer is "yes." If you have a million people, the random bumps and glitches average out, and the chaotic dance floor looks like a smooth river.

However, this specific model has a tricky feature: Hierarchy.

In a normal crowd, if one person leaves, it barely changes the average.
In this hierarchical crowd, the "tallest" people (the leaders) have a massive impact. If the tallest person dies, everyone below them suddenly gets more resources and starts growing faster. Because the top of the hierarchy is always made of very few people, their random deaths and births create a lot of "noise" that doesn't easily average out, even in a large crowd.

3. The Key Findings

The authors did two main things:

A. They built a "Quasi-Stationary" Bridge:
Since the stochastic model (the chaotic one) eventually runs out of people and goes extinct (everyone dies), it doesn't have a permanent "steady state." However, for a long time before extinction, it settles into a "quasi-stationary" rhythm. The authors created a clever formula to estimate this rhythm. They found that while the chaotic model is hard to solve exactly, this formula gives a very good approximation.

B. They Proved the Connection:
They showed mathematically that as the "area" of the room gets infinitely large (meaning the population gets huge), the "Quasi-Stationary" rhythm of the chaotic model converges to the steady rhythm of the smooth, deterministic model.

In simple terms: If you have a small room, the chaotic, random nature of the individuals makes the population behave differently than the smooth equations predict. But if you have a massive stadium full of people, the smooth equations become a very accurate prediction of what the chaotic crowd is doing.

4. The Surprising Twist

The paper also found a nuance. Even in a large crowd, the deterministic model (the smooth view) tends to underestimate the birth rate of the stochastic model (the chaotic view).

Why?
Think of the "top of the hierarchy" (the tallest people). In the chaotic model, these top individuals grow very fast because there are so few people above them. Because they grow fast, they get big quickly, and big individuals have more babies.
In the smooth model, this "super-fast growth" of the elite few gets diluted by the average. The smooth model misses the fact that the "lucky" few at the top are producing a disproportionate number of offspring. So, the real, chaotic population actually reproduces slightly faster than the smooth equations predict.

Summary Analogy

Imagine a game of musical chairs where the chairs are "resources."

The Smooth Model assumes there are so many players that the game is perfectly predictable.
The Stochastic Model realizes that if the person in the "best seat" (the tallest person) suddenly leaves, the person who was second-best suddenly gets a huge boost.

The paper proves that if you have a stadium full of players, the Smooth Model is a great guide. But if you have a small room, the Smooth Model misses the excitement of the "lucky" few at the top who grow fast and have many babies, causing the real population to be slightly more active than the smooth math suggests.

Technical Summary: A Stochastic Population Model with Hierarchic Size-Structure

Problem Statement
The paper addresses the relationship between deterministic and stochastic formulations of hierarchically size-structured population models. In such systems, an individual's access to resources (and thus its growth rate) is determined by the presence of larger individuals. While deterministic models for these populations, often formulated as partial differential equations (PDEs) or delay equations, are well-studied, their justification as approximations of underlying individual-based stochastic processes remains a critical theoretical question. Specifically, the authors investigate whether the stationary birth rate of a deterministic renewal equation can be recovered as the large-population limit of the quasi-stationary birth rate of a stochastic individual-based model. A key challenge is that stochastic systems with an absorbing state (extinction) do not possess a true stationary distribution; instead, they are characterized by a quasi-stationary distribution (QSD) conditioned on non-extinction.

Methodology
The authors develop two parallel models based on the same biological assumptions:

Deterministic Model: A renewal equation for the population birth rate $b(t)$ per unit area. The model assumes individuals are born with a minimal size $x_m$ , grow at a rate $g$ dependent on the density of larger individuals, have size-dependent fertility $\beta(x)$ , and a constant mortality rate $\mu$ . The growth rate depends on the integral of the population density of individuals larger than the focal individual.
Stochastic Model: An individual-based process defined on a finite area $A$ . The state is represented by the history of birth events. Individuals are born with size $x_m$ and grow according to a rate determined by the number of currently larger individuals (which, due to the common birth size and positive growth, is equivalent to the number of older individuals). Birth and death events occur as Poisson processes with rates dependent on the current population state.

To compare these models, the authors:

Derive a heuristic approximation $\bar{B}(A)$ for the quasi-stationary birth rate of the stochastic model. This is achieved by assuming that, in the quasi-stationary regime, past birth events follow a Poisson process with rate $\bar{B}(A)$ . This allows for the calculation of the expected size of an individual at a given age, conditioned on the number of older individuals present at birth.
Formulate a consistency equation for $\bar{B}(A)$ by requiring that the expected number of offspring per individual equals one (the condition for a non-trivial equilibrium).
Analyze the limit of the ratio $\bar{B}(A)/A$ as the area parameter $A \to \infty$ .
Utilize numerical simulations to compare the theoretical predictions of both models against stochastic realizations, specifically examining the expected size-age relationship and the distribution of individual sizes.

Key Contributions and Results

Derivation of the Stochastic Quasi-Stationary Rate: The authors derive an explicit consistency equation (Equation 19) for the approximate quasi-stationary birth rate $\bar{B}(A)$ . They prove that if the basic reproduction number $R_0 > 1$ , this equation has a unique positive solution, provided the growth function $g$ is decreasing and the fertility function $\beta$ is linear ( $\beta(x) = \beta_0 x$ ).
Convergence Theorem: The central theoretical result (Theorem 2) establishes that the stationary birth rate $\bar{b}$ of the deterministic renewal equation is the limit of the scaled stochastic quasi-stationary birth rate:
$\lim_{A \to \infty} \frac{\bar{B}(A)}{A} = \bar{b}$
This confirms that the deterministic model is a valid mechanistic approximation of the stochastic system in the large population limit.
Numerical Validation: Simulations demonstrate that for moderate values of the area parameter $A$ (e.g., $A=1$ ), the deterministic model provides a surprisingly accurate approximation of the stochastic system's behavior, particularly when the competition parameter $z_0$ is large (indicating weaker non-linear effects).
Analysis of Discrepancies: The paper identifies that the deterministic model tends to underestimate the quasi-stationary birth rate of the stochastic model, particularly when competition is strong (small $z_0$ ). The authors attribute this to the high volatility of the population birth rate in the stochastic model, driven by the stochastic behavior of the few individuals at the top of the hierarchy (the largest/oldest), whose dynamics cannot be averaged out even as the total population grows.

Significance and Claims
The paper claims that its primary significance lies in rigorously linking the deterministic and stochastic descriptions of hierarchically structured populations. By reformulating the dynamics in terms of birth event histories, the authors simplify the state space, allowing the quasi-stationary birth rate to be treated as a scalar value rather than an element of an infinite-dimensional space. This simplification facilitates the proof of the convergence result.

The authors modestly note that their derivation of the quasi-stationary rate relies on a heuristic assumption (that past birth events follow a Poisson process) and that a rigorous proof of the convergence of the real quasi-stationary rate to this approximation is not provided. However, they argue that the numerical evidence supports the validity of this approximation, especially as $A$ increases. Furthermore, the work highlights that the equivalence between the deterministic and stochastic models is most robust when the system's non-linearity is softened (large $z_0$ ) or when the population is near the extinction threshold, where the dynamics are dominated by single individuals. The paper concludes that while the birth-history formulation is powerful for models where size is strictly a function of age, generalizations involving size-dependent competition or variable birth sizes would require more complex state variables beyond the birth history.