[Paper Review] Interaction patterns and diversity in assembled ecological communities

Guy Bunin|arXiv (Cornell University)|Jul 16, 2016

Evolutionary Game Theory and Cooperation3 references12 citations

TL;DR

This paper develops an analytical framework to show how ecological community assembly from a species pool systematically alters network properties: communities exhibit higher carrying capacities, weaker competition, and stronger mutualistic interactions than the pool, even when pool interactions are randomly drawn. Crucially, coexistence of all community species is enabled not by complex network structure, but by a simple, consistent relationship between interspecies interactions and species abundances.

ABSTRACT

The assembly of ecological communities from a pool of species is central to ecology, but the effect of this process on properties of community interaction networks is still largely unknown. Here, we use a systematic analytical framework to describe how assembly from a species pool gives rise to community network properties that differ from those of the pool: Compared to the pool, the community shows a bias towards higher carrying capacities, weaker competitive interactions and stronger beneficial interactions. Moreover, even if interactions between all pool species are completely random, community networks are more structured, with correlations between interspecies interactions, and between interactions and carrying capacities. Nonetheless, we show that these properties are not sufficient to explain the coexistence of all community species, and that it is a simple relation between interactions and species abundances that is responsible for the diversity within a community.

Motivation & Objective

• To understand how the process of community assembly from a species pool alters the statistical properties of ecological interaction networks.
• To determine whether random interactions in the species pool lead to structured, non-random patterns in the resulting community networks.
• To identify the specific network properties—such as interaction strength, carrying capacity distributions, and correlations—that emerge during assembly.
• To investigate the mechanistic basis for species coexistence in assembled communities, particularly whether network structure or interaction-abundance relationships are primary.
• To provide a quantitative, analytical framework that predicts community-level network properties from pool-level interaction statistics.

Proposed method

• Formalizing community assembly as a constrained selection process from a species pool, where only feasible (positive abundance) equilibria are retained.
• Using a generalized Lotka-Volterra model to describe species dynamics, with parameters for interspecific interactions ($\alpha_{ij}$) and carrying capacities ($K_i$).
• Applying statistical mechanics and random matrix theory to analytically derive the statistical properties of the resulting community interaction networks.
• Introducing a control parameter $u \simeq \sigma / \sqrt{S}$, where $\sigma$ is the standard deviation of interaction strengths and $S$ is pool size, to map analytical predictions to numerical simulations.
• Performing numerical simulations with Runga-Kutta 45 solvers to identify stable, uninvadable equilibria, using abundance cutoffs ($10^{-14}$) and iterative invasion testing.
• Comparing analytical predictions with simulations across different pool sizes ($S=15, 25, 200, 400$) and interaction distributions (Gaussian, uniform), validating results under finite-size conditions.

Experimental results

Research questions

• RQ1 How do the statistical properties of interaction networks in assembled communities differ from those in the original species pool?
• RQ2 To what extent do random interactions in the species pool give rise to non-random, structured patterns in the resulting community networks?
• RQ3 What network-level properties—such as interaction strength, carrying capacity distributions, and correlations—emerge due to the constraints of community feasibility?
• RQ4 What is the primary mechanism enabling coexistence of all species in the assembled community, given the network structure?
• RQ5 Can a simple, analytical relationship between interspecific interactions and species abundances fully account for community coexistence?

Key findings

• Even when all interactions in the species pool are randomly drawn and uncorrelated, the resulting community networks exhibit non-random structure, including correlations between interspecific interactions and between interactions and carrying capacities.
• Community networks show a systematic bias toward higher average carrying capacities and weaker competitive interactions compared to the pool, with stronger beneficial interactions also enriched.
• The analytical framework accurately predicts the statistical properties of community networks, with strong agreement between theory and simulations even for small pool sizes ($S=15$).
• The fraction of beneficial interactions ($\alpha_{ij} < 0$) in the community is significantly higher than in the pool, especially at larger interaction variance ($\sigma$), though still relatively low at $\sigma=0.5$ ($\sim2\%$) for $S=15$.
• Despite structural correlations in the network, the key determinant of species coexistence is not network complexity but a simple, consistent relationship between interspecific interactions and species abundances.
• The model shows finite-size convergence: results for $S=100$ and $S=200$ are nearly identical, with less than $5\%$ of species able to invade in the final community, indicating robustness.