A five-species counterexample to the Hofbauer--Kon--Saito conjecture on qualitative permanence
Abstract
Hofbauer, Kon and Saito showed that a Lotka-Volterra interaction matrix A with negative diagonal can be qualitatively permanent only if (C1) every cycle of length ≥ 2 of A is nonpositive and (C2) every negative cycle contains exactly one negative entry. They conjectured that these conditions are also sufficient, and proved this for n≤ 3. We give a 5× 5 integer matrix with seven nonzero off-diagonal entries that satisfies (C1) and (C2) but is not qualitatively permanent. For an explicit integer growth vector, in which one species is an obligate consumer as the definition allows, the system has an interior equilibrium, and an asymptotically stable heteroclinic cycle on the boundary of the orthant attracts an open set of interior orbits. Adding non-interacting species gives counterexamples for every n≥ 5. The interactions form a one-way feedback loop together with a predator of two prey. The verification is elementary except for one heteroclinic connection, which lies in a face that is not Volterra-Lyapunov stable. For it we prove a connection lemma based on invasion graphs, which reduces the question to a smaller, stable face. The failure is robust. It persists under small perturbations, including weak additional interactions, so every sign pattern that satisfies (C1) and (C2) and contains ours is a counterexample, and it occurs for an open cone of growth vectors. For n≥ 6 there are also counterexamples in which every species grows when alone. Finally, we announce that the conjecture holds for n=4 and that, for n=5, four minimal sign patterns characterise its failure.
Related
- Preliminary Evaluation of Two iiPCR Platforms for Qualitative Field Detection of Cannabis sativa
- Species Distribution, Antifungal Susceptibility, and Machine Learning-Based Prediction of Optimal Therapies Against Candida Species
- Latitudinal Variation in the Phenology of the Emergent Species Phragmites australis and the Submerged Species Potamogeton crispus in Shallow Lakes of the North China Plain
- A new and unusual water mite species of the subgenus Truncaturus Thor, 1901 from Tasmania, Australia, and notes on other Arrenurus species from Australia (Acari: Hydrachnidia: Arrenuridae)
- The Importance of Different Species of Aspergillus for Stimulation of Plant Growth
BibTeX
@misc{mirunamihaela2026fivespecies,
title = {{A five-species counterexample to the Hofbauer--Kon--Saito conjecture on qualitative permanence}},
author = {Beldiman Miruna-Mihaela and Halanay Andrei-Dan},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.38374},
eprint = {2610.38374},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.38374},
}
RIS
TY - UNPB
TI - A five-species counterexample to the Hofbauer--Kon--Saito conjecture on qualitative permanence
AU - Beldiman Miruna-Mihaela
AU - Halanay Andrei-Dan
PY - 2026
DA - 2026/10/28
PB - arXiv
DO - 10.48550/arxiv.2610.38374
UR - https://arxiv.org/abs/2610.38374
AB - Hofbauer, Kon and Saito showed that a Lotka-Volterra interaction matrix A with negative diagonal can be qualitatively permanent only if (C1) every cycle of length ≥ 2 of A is nonpositive and (C2) every negative cycle contains exactly one negative entry. They conjectured that these conditions are also sufficient, and proved this for n≤ 3. We give a 5× 5 integer matrix with seven nonzero off-diagonal entries that satisfies (C1) and (C2) but is not qualitatively permanent. For an explicit integer growth vector, in which one species is an obligate consumer as the definition allows, the system has an interior equilibrium, and an asymptotically stable heteroclinic cycle on the boundary of the orthant attracts an open set of interior orbits. Adding non-interacting species gives counterexamples for every n≥ 5. The interactions form a one-way feedback loop together with a predator of two prey. The verification is elementary except for one heteroclinic connection, which lies in a face that is not Volterra-Lyapunov stable. For it we prove a connection lemma based on invasion graphs, which reduces the question to a smaller, stable face. The failure is robust. It persists under small perturbations, including weak additional interactions, so every sign pattern that satisfies (C1) and (C2) and contains ours is a counterexample, and it occurs for an open cone of growth vectors. For n≥ 6 there are also counterexamples in which every species grows when alone. Finally, we announce that the conjecture holds for n=4 and that, for n=5, four minimal sign patterns characterise its failure.
ER -