🧬 biology

Feasibility as a moving target: Fluctuating species interactions lead to universal power law in equilibrium abundances

This paper demonstrates that fluctuations in species interactions displace equilibrium points, causing equilibrium abundances to follow a universal power law and revealing that larger ecological communities are significantly more fragile to noise-induced feasibility loss, a prediction successfully validated across real-world networks.

Original authors: Cagatay Eskin, Vu Nguyen, Dervis Can Vural

Published 2026-02-24

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Cagatay Eskin, Vu Nguyen, Dervis Can Vural

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 ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed.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

The Big Idea: The Moving Target

Imagine you are trying to balance a broom on your hand. In traditional ecology (the study of how nature works), scientists have spent decades asking: "Is the broom stable? If I wiggle my hand a little, does it fall over?"

If the broom wobbles but stays upright, they call it "stable." If it falls, they call it "unstable."

This paper argues that this way of thinking is missing the real danger.

The authors say the real problem isn't that the broom wobbles too much. The real problem is that the spot where you are trying to balance the broom keeps moving.

In nature, the "balance point" (where all species live happily together) isn't a fixed spot on the floor. It's a moving target because the relationships between animals and plants (who eats whom, who helps whom) change constantly due to weather, seasons, and behavior.

If that moving target suddenly jumps into a "forbidden zone" (like a negative number, which in biology means a species goes extinct), the whole system crashes. The paper calls this "Feasibility Loss."

The Three Main Discoveries

1. The "Black Swan" of Nature (The Power Law)

The authors did some heavy math and found something surprising. Even if the changes in nature are tiny and gentle (like a light breeze), the risk of a massive crash is much higher than we thought.

The Analogy: Imagine a casino. Usually, if you play a game, you might win or lose a little bit. The odds of losing everything are tiny.
The Discovery: The authors found that in ecosystems, the odds of losing everything follow a "Power Law." This means that while small fluctuations happen often, huge, catastrophic shifts happen way more often than a standard bell curve would predict.
The Result: No matter how small the noise is, there is always a "heavy tail" of risk. A tiny, random change in how species interact can, surprisingly, push a whole ecosystem into a state where some species must go extinct.

2. The "Goldilocks" of Noise (Why Big Ecosystems are Fragile)

You might think a big, complex forest with 1,000 species is stronger than a small pond with 10 species. The paper says: Actually, the bigger the system, the more fragile it is.

The Analogy: Think of a house of cards.
- A small house of cards (5 cards) can handle a little breeze.
- A giant skyscraper of cards (1,000 cards) is so tall and complex that even a tiny, almost invisible draft of air can make it collapse.
The Discovery: The "critical noise" (the amount of change needed to break the system) gets smaller as the number of species gets bigger.
- Small community: Needs a huge storm to break.
- Large community: Needs only a whisper to break.
Why? Because in a complex web, a tiny shift in one relationship ripples through the whole network, eventually pushing the "balance point" into the extinction zone.

3. The Crystal Ball (Predicting Who Will Crash)

The authors didn't just find problems; they built a tool to predict them. They created a formula (a "risk score") that looks at the network of interactions and tells you:

Which specific species is most likely to cause the system to crash if things get noisy.
How long the ecosystem is likely to survive before a crash happens.

The Test: They took 98 real-world ecosystems (from mutualistic networks like bees and flowers, to food webs like lions and zebras) and ran computer simulations.
The Result: Their formula predicted exactly which species were the "weak links" and how long the systems would last. It matched the computer simulations almost perfectly.

What Does This Mean for Us?

1. Stability is an Illusion
Just because an ecosystem looks calm and stable today doesn't mean it's safe. It might be balancing on a moving target that is slowly drifting toward a cliff. We need to stop just checking if the system is "stable" and start checking if the target is "feasible" (i.e., is it still in the safe zone?).

2. Complexity is a Double-Edged Sword
We often think biodiversity makes nature stronger. This paper suggests that while biodiversity is great, it also makes the system more sensitive to noise. A complex system is like a high-performance race car: it's amazing, but it's also much harder to keep balanced than a slow, heavy truck.

3. Conservation Needs a New Focus
Instead of just trying to keep populations steady, conservationists should focus on preventing the "balance point" from drifting.

We need to identify the "weak links" (the species that, if they fluctuate too much, will drag the whole system down).
We need to realize that larger, more complex ecosystems are actually more vulnerable to small environmental changes than we thought.

The Bottom Line

Nature isn't just a static picture that can get blurry; it's a moving target. Even tiny, random changes in how species interact can, over time, push the whole system into a "no-go zone" where extinction becomes inevitable. The bigger and more complex the system, the easier it is to push it over the edge.

1. Problem Statement

Traditional theoretical ecology equates ecosystem persistence with the local stability of a fixed equilibrium point (i.e., whether small perturbations decay). However, this framework assumes the equilibrium point itself is static. In reality, ecological interactions fluctuate due to environmental noise (seasonality, climate, ontogeny), causing the equilibrium point to move.

The authors argue that the primary threat to persistence is not necessarily the loss of stability (instability of the fixed point), but the loss of feasibility: the scenario where the moving equilibrium point shifts into a "negative orthant" (where one or more species have negative abundances), leading to extinction. The paper addresses the following questions:

How do fluctuations in interaction strengths displace the equilibrium point?
What is the probability distribution of these displaced equilibrium abundances?
How does community complexity ( $N$ ) affect the risk of feasibility loss?
Can we predict which species are most vulnerable to driving the system into infeasibility?

2. Methodology

Theoretical Framework

Model: The study uses the Generalized Lotka-Volterra (gLV) equations:
$\frac{dn_i}{dt} = n_i \left[ r_i + \sum_{j=1}^N A_{ij} n_j \right]$
where $n_i$ is biomass, $r_i$ is the intrinsic growth rate, and $A_{ij}$ is the interaction matrix.
Fluctuation Model: Unlike previous studies assuming random matrix structures, the authors assume the interaction matrix $A(t)$ fluctuates around a mean structure $\bar{A}$ following an Ornstein-Uhlenbeck (OU) process:
$d(\delta A_{ij}) = -k_{ij}\delta A_{ij}dt + \sqrt{D_{ij}}dW_{ij}$
This models interactions as having a "restorative force" (returning to a canonical value) and random noise.
Equilibrium Analysis: The equilibrium abundance vector is $\vec{y} = -[A + \delta A]^{-1}\vec{r}$ . The authors derive the probability distribution $h(\vec{y})$ of this moving target.
Feasibility Metrics:
- Escape Rate ( $\lambda_i$ ): Derived using Kramers' formula for first-passage time, estimating the rate at which a specific species $i$ hits zero abundance.
- Mean Time to Loss ( $\langle T_{loss} \rangle$ ): The inverse of the sum of individual escape rates, representing global system fragility.
- Critical Noise ( $\sigma_c$ ): The noise level at which feasibility loss becomes "near certain."

Empirical and Simulation Validation

Empirical Data: Analysis of two large datasets:
1. Paul Lake: 33 years of whole-lake measurements (control group).
2. Baltic Sea Mesocosm: 2,300 days of plankton community data under constant conditions.
- Statistical Approach: Used Kolmogorov-Smirnov (KS) tests, semi-parametric bootstrapping, and Likelihood Ratio tests to fit power laws against exponential and log-normal distributions.
Simulations: Stochastic simulations (Euler–Maruyama integration) performed on 98 empirical networks (82 mutualistic, 16 food webs) from the "Web of Life" database to test analytical predictions against dynamic outcomes.

3. Key Contributions and Results

A. Universal Power Law in Equilibrium Abundances

The most significant theoretical finding is that even light-tailed (Gaussian) fluctuations in interaction strengths lead to heavy-tailed power law distributions for equilibrium abundances.

The Law: The marginal probability distribution for the abundance $y$ of a single species scales as:
$P(y) \sim \frac{1}{y^\alpha}$
Universality: The exponent is theoretically derived as $\alpha = 2$ . This value is universal, independent of community size ( $N$ ), interaction structure ( $A_{ij}$ ), or noise strength.
Mechanism: Large fluctuations can make the interaction matrix nearly singular, causing massive shifts in the equilibrium vector. The "heavy tails" represent a non-negligible risk of catastrophic shifts to negative abundances (extinction).

B. Empirical Confirmation

Analysis of the 34 species in the empirical datasets revealed:

A power law signal was detected in most species.
The observed exponents clustered between $\alpha \approx 2$ and $3$, with a median of $\alpha \approx 2.56$ .
The deviation from the theoretical $\alpha=2$ $α = 2$ is attributed to:
1. Holling-type functional responses: Non-linear responses modify the exponent to $\alpha = m+1$ (where $m$ is the Holling exponent, typically 1–3).
2. Relaxation times: Slow-relaxing species cannot track the moving equilibrium, effectively "shaving off" the extreme tails and increasing the apparent $\alpha$ .

C. Fragility of Large Communities

The study derives a scaling law for the critical noise ( $\sigma_c$ ) required to cause near-certain feasibility loss:
$\sigma_c(N) \sim N^{-1}$

Implication: As community size ( $N$ ) increases, the system becomes exponentially more fragile. Larger communities require infinitesimally small interaction noise to push the equilibrium into an infeasible state, even if the equilibrium remains mathematically stable.

D. Predictive Metrics for Risk

The authors developed analytical metrics that successfully predict fragility in real-world networks:

Species-Specific Escape Rate: Correctly ranks which species are most likely to trigger system-wide feasibility loss.
Mean Time to Feasibility Loss: Highly correlates with persistence times in stochastic simulations.
Validation: The analytical predictions matched simulation results for 98 networks with high concordance (CCC = 0.96 for food webs; CCC = 0.72–0.90 for mutualistic networks).

4. Significance and Implications

Redefining Persistence: The paper shifts the paradigm from "stability of a fixed point" to "feasibility of a moving target." A system can be locally stable (eigenvalues have negative real parts) yet biologically non-viable because the equilibrium drifts into negative space.
Black Swan Events: The heavy-tailed distribution explains "black swan" events where robust-looking ecosystems suddenly collapse due to rare, large fluctuations in interaction strengths, rather than gradual degradation.
Complexity-Stability Paradox: It provides a quantitative mechanism for May's complexity-stability tension: larger, more complex communities are inherently more fragile to noise-induced feasibility loss, regardless of their interaction topology.
Conservation Application: The derived metrics ( $\lambda_i$ and $\langle T_{loss} \rangle$ ) offer a practical, data-driven framework for identifying vulnerable species and prioritizing conservation efforts based on their potential to drive the entire ecosystem into infeasibility.

In summary, the paper establishes that environmental noise acting on species interactions creates a universal risk of extinction characterized by power-law statistics, fundamentally altering how ecological risk and community persistence should be modeled and measured.