Nonvulnerability of ecosystems in unpredictable environments

One way to bracket the effects of a real environment on an ecosystem during a finite time interval is to use the concept of vulnerability. If a deterministic model ecosystem has a good Lyapunov function, it may be possible to derive simple and useful tests for the system to be nonvulnerable. For a subset of Lotka-Volterra models, the system is nonvulnerable if the smallest eigenvalue of a certain matrix is not only positive, but is greater than a positive number, which depends on a priori estimates for the bounds on the unpredictable forcing functions. The bounded but unknown functions which act on the Lotka-Volterra equations also can be interpreted as errors in the system's equations which can be tolerated without a qualitative change in the behaviour of its solutions.