Lotka–Volterra Competition Equations

• Lotka–Volterra competition equations are deterministic models describing the dynamics of species competing for shared resources through ordinary and partial differential equations and stochastic variants.
• They extend the classical predator–prey framework by incorporating explicit competition terms, enabling analysis of global stability, bifurcation, and spatial pattern formation.
• These models are pivotal in mathematical biology and ecology, offering insights into invasion dynamics, coexistence regimes, and spatial segregation phenomena.

The Lotka–Volterra competition equations are a family of deterministic models that quantify the population dynamics of two or more biological species competing for shared resources. They serve as canonical models for interspecific competition, generalizing the original prey–predator Lotka–Volterra system by including explicit competition terms, and form a cornerstone of mathematical biology, nonlinear dynamics, and quantitative ecology. Their study encompasses ordinary differential equations (ODE), partial differential equations (PDE), systems on graphs, stochastic variants with diffusion or jumps, time-scale models, and spatially structured or heterogeneous frameworks. The equations' analytical tractability and biological relevance have fostered rich theoretical developments, including global existence and convergence, bifurcation theory, spatial pattern formation, and invasion/spreading dynamics.

1. Mathematical Formulation and Derivation

The classical $N$-species competitive Lotka–Volterra ODE system is

$$\dot n_i = n_i \left( r_i - \sum_{j=1}^N b_{ij} n_j \right), \qquad i = 1,\dots,N$$

where $n_i(t) \geq 0$ is the population size of species $i$, $r_i > 0$ its intrinsic growth rate, and $b_{ij} \geq 0$ the competitive effect of species $j$ on $i$ (Champagnat et al., 2010). The system can be derived as a mean-field limit of pairwise stochastic competition with replacement rules $X_m + X_\ell \to 2X_m$ at rate $p_m^\ell$; the deterministic rate equations for species fractions $$\dot n_i = n_i \left( r_i - \sum_{j=1}^N b_{ij} n_j \right), \qquad i = 1,\dots,N$$0 then take the antisymmetric form

$$\dot n_i = n_i \left( r_i - \sum_{j=1}^N b_{ij} n_j \right), \qquad i = 1,\dots,N$$1

with the traditional Lotka–Volterra form emerging upon inclusion of birth/death terms as a limiting case (Zia, 2010).

Extensions encompass:

• Resource-based models: growth is limited via resource-dependent nonlinearities, giving

$$\dot n_i = n_i \left( r_i - \sum_{j=1}^N b_{ij} n_j \right), \qquad i = 1,\dots,N$$2

with symmetry and strict monotonicity in $$\dot n_i = n_i \left( r_i - \sum_{j=1}^N b_{ij} n_j \right), \qquad i = 1,\dots,N$$3 leading to global convergence results (Champagnat et al., 2010).

• PDE competition–diffusion systems: in spatial domain $$\dot n_i = n_i \left( r_i - \sum_{j=1}^N b_{ij} n_j \right), \qquad i = 1,\dots,N$$4

$$\dot n_i = n_i \left( r_i - \sum_{j=1}^N b_{ij} n_j \right), \qquad i = 1,\dots,N$$5

with Dirichlet, Neumann, or mixed conditions (Hu et al., 2022).

• Discrete/graph or time-scale models (Streipert et al., 27 May 2026).

2. Dynamical Classification: Local and Global Behavior

ODE Regimes and Global Stability

For $$\dot n_i = n_i \left( r_i - \sum_{j=1}^N b_{ij} n_j \right), \qquad i = 1,\dots,N$$6-species systems under weak competition and symmetry/irreducibility conditions ($$\dot n_i = n_i \left( r_i - \sum_{j=1}^N b_{ij} n_j \right), \qquad i = 1,\dots,N$$7), all positive initial conditions converge to a unique globally attracting equilibrium. Analysis is based on construction of a strict Lyapunov functional

$$\dot n_i = n_i \left( r_i - \sum_{j=1}^N b_{ij} n_j \right), \qquad i = 1,\dots,N$$8

with $$\dot n_i = n_i \left( r_i - \sum_{j=1}^N b_{ij} n_j \right), \qquad i = 1,\dots,N$$9 and strict monotonicity of $n_i(t) \geq 0$0, guaranteeing uniqueness and global attractivity of the positive steady state (an ESS in classical game theory: $n_i(t) \geq 0$1 for all $n_i(t) \geq 0$2 with $n_i(t) \geq 0$3) (Champagnat et al., 2010).

Multispecies pairwise competitive systems ($n_i(t) \geq 0$4 species, $n_i(t) \geq 0$5 the frequency) reduce to $n_i(t) \geq 0$6 with $n_i(t) \geq 0$7 antisymmetric. Odd $n_i(t) \geq 0$8 guarantees at least one interior fixed point, invariants, and neutrally stable cycles; even $n_i(t) \geq 0$9 with nonzero determinant implies exponential collective variables (Zia, 2010).

Bifurcation and Pattern Formation

In systems with spatial structure and diffusion, or with competition–cooperation terms, bifurcation phenomena can occur. For example, treating the competition strength as a parameter, the positive constant equilibrium may lose stability at a critical threshold, giving rise to branches of spatially inhomogeneous steady states; in the symmetric case and strong competition limit ($i$0), coexistence states segregate spatially, converging to solutions of a scalar free-boundary problem (Li et al., 2024).

Planar systems with density-dependent intraspecific terms $i$1 exhibit Hopf bifurcations; a branch of limit cycles is born at a critical parameter and can "blow up," with the oscillation amplitude becoming unbounded as $i$2 varies, leading to global cycles of arbitrarily large amplitude (Bouse et al., 2010).

Stochastic and Jump-Diffusion Systems

Stochastic Lotka–Volterra competition models with Brownian motion and random jumps admit unique global positive solutions, uniform moment bounds, and explicit criteria for persistence or extinction based on Lyapunov exponents. For each species $i$3, persistence requires positivity of

$i$4

almost surely; extinction occurs if the averaged $i$5 is negative (Bao et al., 2011).

3. Spatial Extensions: Graphs, Domains, and Anomalous Diffusion

Finite Graphs

On finite graphs with general boundary conditions, the two-species Lotka–Volterra competition–diffusion system preserves the four equilibrium structure from the continuum theory:

• Mutual extinction,
• Single-species persistence,
• Coexistence: $i$6, $i$7 and admits a dichotomy based on parameter thresholds. The qualitative trichotomy—winner-takes-all, coexistence, or bistability depending on inequalities among intrinsic growth and competition rates—mirrors the classical continuous-space scenario (Hu et al., 2022).

The extension of the upper–lower solution method on graphs yields a maximum principle and monotone iteration schemes enabling rigorous classification of long-term population outcomes.

PDEs and Fractional Diffusion

In $i$8, competition–diffusion PDEs reveal intricate invasion and replacement dynamics, sensitive to the competition strength (weak/strong) and the geometry of initial supports:

• With strong competition ($i$9), traveling wave fronts connecting $r_i > 0$0 and $r_i > 0$1 exist, with the species whose intrinsic KPP speed ($r_i > 0$2 for $r_i > 0$3, $r_i > 0$4 for $r_i > 0$5) is higher generally winning and excluding the other at maximal speed; the loser is confined to a shrinking region, except at the bistable invasion interface propagating at $r_i > 0$6 (Bao et al., 2024, Guo, 25 Feb 2026).
• For general measurable initial supports, spreading occurs anisotropically with direction-dependent speeds determined by the geometry of the supports; precise domains of influence (spreading sets) for each species are characterized explicitly (Guo, 25 Feb 2026).
• Under weak competition ($r_i > 0$7), monotone traveling fronts connect the extinction equilibrium to a positive coexistence equilibrium, ensuring global coexistence behind the invasion front (Wang et al., 2017).
• Fractional diffusion $r_i > 0$8 ($r_i > 0$9) yields nonlocal competition; in the strong competition limit, species segregate, and regularity theory guarantees Hölder/Lipschitz continuity of the limiting profiles (and sharp free boundary conditions) (Verzini et al., 2013).

4. Novel Regimes: Spatial Heterogeneity, Interfaces, and Time Scales

Competition models with spatial barriers introduce additional phenomena. For example, in domains partitioned by interfaces where only some species mix, coexistence regions can emerge even under adverse local growth conditions, violating classical exclusion principles; the window for coexistence is explicitly delimited by principal eigenvalue thresholds associated with the interface and diffusion parameters (Álvarez-Caudevilla et al., 2024).

Competition models on general time scales (unifying continuous and discrete settings) are analyzed using dynamic augmented phase-plane methods. The global classification of dynamics—competitive exclusion, coexistence, bistability, and degenerate continua—holds under mild regularity, with attractors determined by nullcline structure, invariance regions, and the sign patterns of "root-operators" tracing time step dynamics (Streipert et al., 27 May 2026).

5. Invasion, Spreading, and Phase-Plane Dynamics

The spatial models allow precise characterization of invasion and replacement:

• In the strong-competition regime with disjoint initial supports, each species may invade only along directions where the support is sufficiently thick; front speeds and spreading sets depend on geometric projection paths and anisotropic KPP-type variational formulas (Bao et al., 2024, Guo, 25 Feb 2026).
• In the weak competition regime, invasion by either species leads to ultimate coexistence, as the traveling fronts settle on the positive equilibrium.
• Dynamic phase-plane analysis for two-species models elucidates all possible qualitative outcomes, governed by the location and stability of fixed points, nullclines, and invariant regions; the time scale setting admits this analysis for both continuous and discrete models (Streipert et al., 27 May 2026).

6. Bifurcation, Limit Cycles, and Complex Behaviors

Beyond equilibrium and invasion, Lotka–Volterra equations exhibit richer behaviors under specific constructions:

• Hopf bifurcations yield limit cycles, which can persist and grow unboundedly in amplitude as a parameter is varied (e.g., in systems with cooperation–competition transitions) (Bouse et al., 2010).
• Bifurcation analysis in reaction-diffusion models reveals critical thresholds where constant equilibria lose stability and new (spatially structured or segregated) branches of solutions emerge, with limiting profiles governed by free boundary problems (Li et al., 2024).

7. Summary Table: Core Regimes (Two-Species LV ODE)

Parameter Regime	Dynamics	Attractor
$b_{ij} \geq 0$0, $b_{ij} \geq 0$1	Global coexistence	Interior fixed point
$b_{ij} \geq 0$2, $b_{ij} \geq 0$3	$b_{ij} \geq 0$4 excludes $b_{ij} \geq 0$5	$b_{ij} \geq 0$6
$b_{ij} \geq 0$7, $b_{ij} \geq 0$8	$b_{ij} \geq 0$9 excludes $j$0	$j$1
$j$2, $j$3	Bistability (initial data determine outcome)	Semi-trivial
Edge cases	Continuum/line of neutrally stable equilibria	Neutral segment

($j$4: competition coefficients; $j$5: carrying capacities) (Streipert et al., 27 May 2026).

• "Convergence to equilibrium in competitive Lotka-Volterra equations" (Champagnat et al., 2010)
• "General Properties of a System of $j$6 Species Competing Pairwise" (Zia, 2010)
• "Blow Up of a Cycle in Lotka-Volterra Type Equations with Competition-Cooperation Terms and Quasi-Linear Systems" (Bouse et al., 2010)
• "Lotka-Volterra competition models on finite graphs" (Hu et al., 2022)
• "Bifurcation for the Lotka-Volterra competition model" (Li et al., 2024)
• "Phase Plane Analysis on Time Scales for a Lotka-Volterra Competition Model" (Streipert et al., 27 May 2026)
• "Strong competition versus fractional diffusion: the case of Lotka-Volterra interaction" (Verzini et al., 2013)
• "Competitive Lotka-Volterra Population Dynamics with Jumps" (Bao et al., 2011)
• "A three population Lotka-Volterra competition model with two populations interacting through an interface" (Álvarez-Caudevilla et al., 2024)
• "Asymptotic speeds of spreading for the Lotka-Volterra system with strong competition in $j$7" (Bao et al., 2024)
• "Spreading dynamics for the Lotka-Volterra system with general initial supports: the strong competition" (Guo, 25 Feb 2026)
• "Entire Solutions for the Classical Competitive Lotka-Volterra System with Diffusion in the Weak Competition Case" (Wang et al., 2017)