Infinitely Many Competing Species

• Infinitely Many Competing Species is a theoretical framework in mathematical ecology that models species coexistence and stability as the number of interactions grows without bound.
• The framework integrates classic and modern Lotka–Volterra dynamics with resource-mediated competition to derive stability criteria and predict macroscopic laws.
• It employs Lyapunov functionals, convex analysis, and mean-field approximations to resolve the diversity–stability paradox and facilitate algorithmic computation.

Infinitely many competing species arise in mathematical ecology through the analysis of ecological communities in the limit where the number of interacting species, $n$, tends toward infinity. This regime sits at the interface of stochastic processes, dynamical systems, statistical mechanics, and optimization. Key theoretical tools include the competitive Lotka–Volterra (LV) and Volterra models, resource-mediated models such as MacArthur’s consumer–resource framework, and their continuum extensions. The asymptotic behavior, stability, coexistence criteria, and emergent macroscopic laws in these systems are central concerns, with implications for both mathematical theory and quantitative ecology.

1. Classical and Modern Lotka–Volterra Systems in the Infinite-Species Limit

The $n$-species competitive Lotka–Volterra system is defined by the ODEs

$$\frac{dx_i}{dt} = x_i \left( r_i + \sum_{j=1}^n M_{ij} x_j \right)$$

where $x_i \geq 0$ is the abundance of species $i$, $r_i > 0$ its intrinsic growth rate, and $M$ a symmetric interaction matrix encoding competitive interactions. In the pure competition case, $M = -T - D$, with $T \geq 0$ encoding pairwise competition and $D$ diagonal (self-regulation).

A coexistence equilibrium $\mathbf{x}^* > 0$ satisfies $M \mathbf{x}^* = -\mathbf{r}$; global asymptotic stability is ensured by the negative-definiteness of $M$ (Goh’s theorem).

Mooij et al. (Mooij et al., 2024) show that, contrary to May’s classical result that “large systems are unstable,” stable coexistence of all species is attainable as $n \to \infty$—provided the underlying interaction network is sparse with bounded maximum degree. Feasibility ($x_k^* > 0$) and stability criteria are governed not by $n$, but by structural degree bounds. In contrast, for dense random graphs (Erdős–Rényi $G(n,p)$), where $d_{\max} \sim np$ diverges with $n$, the threshold for stability collapses. This clarifies that instability in large systems arises from degree outliers, not the sheer number of species.

2. Resource-Mediated Competition and the Continuum Limit

Resource competition models, particularly extensions of MacArthur’s consumer–resource equations with continuous traits, provide an analytic framework for systems with infinitely many species. The key continuum model is

$$\begin{aligned} \partial_t f(x,t) &= f(x,t) \bigg[ a(x) + \int_Y K(x,y) (R(y,t) - R^*(y))\,dy \bigg] \ \partial_t R(y,t) &= m(y)\left[ R^*(y) - R(y,t) \right] - R(y,t) \int_X K(x,y) f(x,t)\,dx \end{aligned}$$

where $f(x,t)$ is the consumer-trait density, $R(y,t)$ the resource, $K(x,y)$ the interaction kernel, $a(x)$ intrinsic growth, and $m(y),R^*(y)$ maintenance and carrying capacity (Cai et al., 2020).

Under non-degeneracy and boundedness hypotheses, this model admits a unique nonnegative steady-state $(\tilde{f},\tilde{R})$, termed the evolutionary stable distribution (ESD). The ESD is characterized as the minimizer of a convex functional; its existence, uniqueness, and convergence from initialization are proven using convex analysis and Lyapunov function techniques. Numerical algorithms for computing the ESD and for simulating the dynamics obey positivity and energy dissipation.

These results demonstrate that, in the continuum limit, resource-mediated competition selects a discrete set of “fittest” trait-values even from an initially continuous trait distribution.

3. Infinite-Species Integrable Volterra Models and Superintegrability

The integrable Volterra model, originally defined for a finite number of species, extends naturally to the case of infinitely many competing species, both countable ($n \in \mathbb{Z}$) and uncountable (continuous trait $x \in \mathbb{R}$) (Ragnisco et al., 21 Jan 2026). The countable model reads: $$\dot y_n = \epsilon_n + \sum_{m \in \mathbb{Z}} A_{nm} e^{y_m}$$ with logarithmic variables $y_n = \ln N_n$. The interaction matrix $A$ has rank two, fixed by parameter families $\epsilon_n$ and $\eta_n$.

For both discrete and continuum species, the Hamiltonian structure persists, with the system being maximally superintegrable: infinitely many Casimirs (constants of motion), plus integrals associated with the kernel of the rank-2 interaction. Dynamics reduce to a two-dimensional system $(P,Q)$, corresponding to collective coordinates (e.g., total abundance, trait mean), while all orthogonal directions are invariant. As a result, even with infinitely many species, macroscopic dynamics admit exact integration and periodic orbits, confirmed by numerical solutions.

4. Macroscopic Laws and Mean-Field Limits in Random Ecosystem Models

In models with random species-resource couplings, the thermodynamic ($N, M \to \infty$) behavior is rigorously investigated via dynamical path-integral methods (Batista-Tomas et al., 2021). After averaging over disorder and performing a saddle-point analysis, the dynamics reduce to two coupled stochastic processes representing an “effective” species and an “effective” resource: $$\frac{\dot{n}(t)}{n(t)} = \cdots,\qquad \frac{\dot r(t)}{r(t)} = \cdots$$ with colored noise and retarded (non-Markovian) self-interactions.

The steady state is characterized by truncated-Gaussian solutions for abundance and resource levels, with associated algebraic self-consistency equations. The stability of these solutions is determined by the smallest eigenvalue of the community matrix, which depends on heterogeneity parameters $\sigma_c$ (metabolic strategy) and $\sigma_K$ (carrying capacity). The framework recovers well-known competitive exclusion principles: typically, $N_s \le M_s$ surviving species and resources. Increasing heterogeneity in strategies enhances species survival but can destabilize the system if excessive.

The dynamical mean-field theory (DMFT) formalism reveals that, in the infinite limit, system behavior is completely encoded by a finite set of macroscopic order parameters.

5. Structural Criteria and the Resolution of the Diversity–Stability Paradox

The classical “diversity–stability paradox,” stemming from May’s result that interactions must vanish ($\sigma^2 \sim 1/n$) for stability as $n\to\infty$, is refined by recent work (Mooij et al., 2024). The key insight is that the critical factor for instability is the growth of the maximum node degree ($d_{\max}$) rather than $n$ itself.

If $d_{\max}$ remains bounded (sparse network), then there exists $\tau < \Omega$, where $\Omega$ is the smallest real root of a cubic in terms of $d_{\max}, d_{\min}$, for which global coexistence and stability are guaranteed, even as $n \to \infty$. In contrast, if $d_{\max} \to \infty$ (e.g., for dense random graphs), stability collapses, recovering May’s scaling. Thus, network sparseness constitutes a natural resolution to the paradox, aligning mathematical prediction with empirical observations from sparse ecological networks.

6. Conserved Quantities, Lyapunov Functions, and Algorithmic Computation

In both resource-mediated and integrable Volterra models, existence proofs and algorithmic computation rely on the construction of Lyapunov (energy-dissipation) functionals, ensuring global convergence to equilibrium under broad conditions (Cai et al., 2020). For finite or discretized models, projected-gradient and convex optimization algorithms efficiently compute steady-state solutions, with extensions to fully discrete schemes in time preserving positivity and dissipativity.

In maximally superintegrable systems (Ragnisco et al., 21 Jan 2026), infinitely many functionals commute with the Hamiltonian, and the remaining nontrivial dynamics are confined to a finite-dimensional manifold. This duality—between the infinite-dimensional state space and finitely many active degrees of freedom—suggests a general principle for model reduction in large ecosystem dynamics.

---

References:

• "Stable coexistence in indefinitely large systems of competing species" (Mooij et al., 2024)
• "The integrable Volterra system in the case of infinitely manyspecies, either countable or uncountable" (Ragnisco et al., 21 Jan 2026)
• "Path-integral solution of MacArthur's resource-competition model for large ecosystems with random species-resources couplings" (Batista-Tomas et al., 2021)
• "Dynamics of many species through competition for resources" (Cai et al., 2020)