Stable equilibria in the Lotka-Volterra equations

We consider the Lotka-Volterra system and provide necessary conditions for an equilibrium to be stable. Our results naturally complement earlier fundamental results by N. Adachi, Y. Takeuchi, and H. Tokumaru, who, in a series of papers, give sufficient (and for some cases necessary) conditions for the existence of a stable equilibrium point.

💡 Research Summary

The paper investigates the stability of equilibrium points in the generalized Lotka‑Volterra (LV) system, aiming to complement the well‑known sufficient conditions established by Adachi, Takeuchi, and Tokumaru. The LV equations are written as
(x_i’(t)=r_i x_i(t)\bigl(1-\frac{1}{K_i}\sum_{j=1}^n\alpha_{ij}x_j(t)\bigr)),
with positive intrinsic growth rates (r_i), carrying capacities (K_i), and interaction coefficients (\alpha_{ij}) forming the matrix (A). A feasible equilibrium exists when (A) is invertible and the solution (x^*=A^{-1}K) lies in the non‑negative orthant (\mathbb R^n_+).

Previous work introduced the class S (there exists a diagonal matrix (W) such that (WA+A^TW) is positive definite) and showed that if (A\in S) then the LV system possesses a unique globally stable equilibrium. Moreover, when all off‑diagonal entries of (A) are non‑positive, the class S condition is equivalent to (A) being a P‑matrix (all principal minors positive). These results are sufficient for stability, but they do not address whether they are also necessary.

The central contribution of the present study is Theorem 3.2, which provides necessary conditions for stability, both for fully feasible equilibria (all species present) and for sub‑feasible equilibria (some species extinct). The authors first define a diagonal scaling matrix
(D^=\operatorname{diag}\bigl(\frac{r_1}{K_1}x^_1,\dots,\frac{r_n}{K_n}x^*_n\bigr))
and set (B=D^*A).

For a feasible equilibrium the theorem states that every principal minor of (B) of any order must be positive: for all (k=0,\dots,n) and for every (C\in D_k(B)) (the set of all (k\times k) principal sub‑matrices of (B)), (\det(C)>0). This condition is automatically satisfied if (A) is a P‑matrix because the diagonal entries of (D^*) are positive, but the converse does not hold in general. The gap originates from the possible presence of complex eigenvalues of the Jacobian at the equilibrium.

For a sub‑feasible equilibrium (where a subset (I) of species has zero density) the theorem adds two layers of requirements: (i) the growth functions (F_i(x^)=1-\frac{1}{K_i}\sum_j\alpha_{ij}x^_j) must be negative for all extinct species (i\in I); (ii) a more intricate sign condition involving mixed products of the negative (F_i)’s and the principal minors of the reduced matrix (B) (obtained by deleting rows and columns indexed by (I)). These conditions guarantee that all coefficients of the characteristic polynomial of the Jacobian are positive, which is necessary for all eigenvalues to have negative real parts.

The proof proceeds by writing the Jacobian at an equilibrium as (J=-D^*A). The characteristic polynomial (p(\lambda)=\det(J-\lambda I)) is expanded using the Leibniz formula; each coefficient of ((- \lambda)^{n-k}) is precisely the sum of determinants of all (k)-order principal minors of (B). By invoking Descartes’ rule of signs, the authors argue that stability (all eigenvalues with negative real part) forces every coefficient to be positive, leading directly to the conditions of Theorem 3.2. The analysis also shows that while the sign condition eliminates the possibility of positive real eigenvalues, it does not preclude complex conjugate pairs with negative real parts; therefore the necessary condition is slightly weaker than the sufficient class S condition.

The discussion highlights that the new necessary condition is “almost” equivalent to the P‑matrix condition for feasible equilibria, differing only when complex eigenvalues appear. For sub‑feasible equilibria, the additional requirement (F_i(x^*)<0) aligns with earlier lemmas in the literature and reflects the biological intuition that extinct species must experience a net negative growth rate at the equilibrium.

In summary, the paper enriches the theoretical understanding of LV dynamics by establishing matrix‑based necessary conditions for equilibrium stability. These results are particularly valuable for high‑dimensional ecological models where checking the full class S condition may be computationally demanding, while verifying positivity of principal minors of the scaled matrix (B) is more tractable. Moreover, the treatment of sub‑feasible equilibria provides a rigorous framework for analyzing partial extinction scenarios, a topic of growing interest in theoretical ecology and epidemiology.