Competitive Three-Species Lotka–Volterra Model

• The competitive three-species Lotka–Volterra model is a mathematical framework that defines interspecific competition via a globally attracting carrying simplex and yields diverse dynamical behaviors.
• Zeeman’s classification organizes these systems into stable equivalence classes based on sign-patterns, providing explicit parameter criteria that govern system stability and the emergence of limit cycles.
• Recent analyses using bifurcation theory and focal value computations reveal multiple limit cycles and permanence regimes, highlighting the model's capacity to predict oscillatory coexistence and extinction dynamics.

The competitive three-species Lotka–Volterra (LV) model is a classical framework in mathematical biology for describing the interactions of three competing species. Its theoretical richness stems from the confluence of nonlinear dynamics, bifurcation theory, and invariant manifold analysis, giving rise to an array of behaviors including multistability, oscillatory coexistence, and complex boundary-driven phenomena. Research has systematically classified the global dynamics of such systems, especially through Zeeman’s taxonomy, which organizes competitive 3D LV systems by boundary dynamics and invariant sets. Recent work has established fundamental results on the number of possible limit cycles, the structure of the carrying simplex, and class-specific dynamical behavior.

1. Standard Formulation and Key Structures

The standard competitive three-species Lotka–Volterra ODE is given in vector and component form as

$$\dot x_i = x_i\Bigl(\sum_{j=1}^3 a_{ij}(x_j-1)\Bigr), \quad x_i > 0, \quad i = 1,2,3,$$

where the state variables $x_i$ are typically rescaled so that the unique interior equilibrium is positioned at $(1,1,1)$, and the interaction matrix $A = (a_{ij})$ satisfies $a_{ij}<0$ for $i \neq j$ (inter-specific competition) and $a_{ii}<0$ (self-regulation). In vector notation,

$\dot x = X[A(x-1)],$

where $X = \mathrm{diag}(x_1,x_2,x_3)$. All off-diagonal $a_{ij}$ being negative enforces strict competitiveness (Hu et al., 24 Mar 2026).

A central geometric object is the carrying simplex, a two-dimensional, globally attracting invariant manifold $x_i$0, homeomorphic to the standard simplex $x_i$1. Every nontrivial orbit is attracted to $x_i$2, and the global phase portrait reduces to the planar restriction of the flow on $x_i$3 (Niu et al., 2023).

2. Zeeman’s Classification and Sign-Pattern Taxonomy

The global dynamics have been classified into 33 stable equivalence classes (Zeeman classes), organized by the sign pattern of six key algebraic invariants derived from $x_i$4: $x_i$5 and $x_i$6, where $x_i$7 and $x_i$8 are explicit polynomials in the $x_i$9. Each class prescribes a distinct configuration of boundary fixed-points (axes and planes) and their local stability properties.

For instance, class 28 is defined by the sign pattern

$(1,1,1)$0

and systems in this class can be constructed with specific parameter choices to satisfy structural constraints, such as block-diagonalizability around the interior equilibrium (Hu et al., 24 Mar 2026).

The factorial classification via Zeeman’s approach is comprehensive even in the presence of seasonal succession or time-periodic coefficients, as the induced Poincaré maps are shown to admit the same combinatorial class structure (Niu et al., 2023, Niu et al., 2024).

3. Limit Cycles and Bifurcation Phenomena

Three-species competitive LV systems can exhibit multiple limit cycles, a behavior prohibited in two-species models. The number and type of limit cycles are class-dependent and require delicate normal form analysis.

The recent construction in class 28 demonstrates the existence of at least four limit cycles:

• Three small-amplitude limit cycles, nested and arising via successive higher-order Hopf (Bautin) bifurcations near the interior fixed point. These cycles are created through focus-type center manifold dynamics, with stability alternating (unstable–stable–unstable). The focal quantities $(1,1,1)$1 are computed to isolate parameter sets supporting multiple cycles.
• A fourth large-amplitude limit cycle, guaranteed by a Poincaré–Bendixson argument on the carrying simplex, judging from the presence of both an attracting boundary and a stable inner cycle. This outer cycle delineates the attraction basin of the interior cycles (Hu et al., 24 Mar 2026).

The completion of explicit four-cycle examples for Zeeman classes 26–29 underscores the nontrivial dynamical possibilities in 3D competition models.

4. Carrying Simplex and Global Attractors

Hirsch’s theory asserts the existence and uniqueness (up to homeomorphism) of a carrying simplex $(1,1,1)$2, to which all positive orbits converge. Dynamics on $(1,1,1)$3 encodes global behavior:

• Every nontrivial omega-limit set is contained in $(1,1,1)$4.
• The interior dynamics may include unique or multiple positive fixed points, attracting cycles, or even invariant closed curves (quasi-periodic behavior).
• The phase-portrait (in class 28, for example) reveals concentric small cycles surrounding the coexistence fixed point, and a peripheral cycle shadowing the boundary heteroclinic contour (Hu et al., 24 Mar 2026, Niu et al., 2024).

The carrying simplex reduction is also valid for discrete or periodically forced competitive models, providing a unifying framework for permanence, coexistence, and extinction regimes (Niu et al., 2023, Niu et al., 2024).

5. Parameter Criteria, Focus Values, and Permanence

Key scalar invariants such as the invasion exponents $(1,1,1)$5 and mixed-competition terms $(1,1,1)$6 govern the possibility of permanence and coexistence:

• Mutual invadability conditions ($(1,1,1)$7 for all $(1,1,1)$8) are necessary and sufficient for robust three-species coexistence (class 33) (Niu et al., 2024).
• Classes 27, 29, 31 permit more subtle forms of permanence (e.g., with repelling heteroclinic cycles), subject to algebraic positivity of further determinants such as $(1,1,1)$9 (Niu et al., 2024).
• Invariant closed curves (discrete Neimark–Sacker bifurcations) may arise within permanent classes, precluding global asymptotic stability of the interior fixed point but maintaining bounded coexistence (Niu et al., 2023, Niu et al., 2024).

Permanence, as distinguished from global attractivity, is thus determined by explicit inequalities in $A = (a_{ij})$0 and associated scalar exponents, and its violation signals the onset of boundary attractors or partial extinction dynamics.

6. Lyapunov Functions, Integrals, and Special Examples

Explicit global Lyapunov functions can be constructed for certain subclasses of the competitive 3D LV model. In classical "generalized gradient" forms, such functions (e.g., quadratic polynomial in $A = (a_{ij})$1 or composite rational functions in total and product variables) yield global decay and exclude nontrivial closed orbits.

In the canonical May–Leonard model,

$A = (a_{ij})$2

the Lyapunov function transitions between Hamiltonian-like (neutrally stable orbits), global convergence, and boundary exclusion as $A = (a_{ij})$3 changes sign (Tang et al., 2012). The existence/nonexistence of limit cycles and invariant tori is deeply linked to the presence of (generalized) gradient structure or antisymmetric feedbacks, as formalized by Ao et al.'s decomposition framework.

7. Broader Implications and Extensions

The competitive three-species LV model serves as a touchstone for multi-dimensional competitive systems, spatially distributed analogues (reaction–diffusion with interfaces), and non-autonomous extensions (periodic seasonal succession). In higher-dimensional or spatially structured settings, further complexity arises:

• Interface-mediated systems allow coexistence regimes forbidden by homogeneous ODEs, due to exchange-driven shifts in effective invasion eigenvalues (Álvarez-Caudevilla et al., 2024).
• Extensions include stochastically forced models, structured demography, or time-heterogeneous environments, invoking the same topological and bifurcation-theoretic toolkit.

Recent advances underline the role of focal value computations, parameter sign-pattern classification, and carrying simplex theory in delivering rigorous, computable descriptions of global outcomes for competitive ecological systems.