We study a spatially explicit harvesting model in periodic or bounded environments. The model is governed by a parabolic equation with a spatially dependent nonlinearity of Kolmogorov–Petrovsky–Piskunov type, and a negative external forcing term $-\delta$. Using sub- and supersolution methods and the characterization of the first eigenvalue of some linear elliptic operators, we obtain existence and nonexistence results as well as results on the number of stationary solutions. We also characterize the asymptotic behavior of the evolution equation as a function of the forcing term amplitude. In particular, we define two critical values $\delta^*$ and $\delta_2$ such that, if $\delta$ is smaller than $\delta^*$, the population density converges to a “significant" state, which is everywhere above a certain small threshold, whereas if $\delta$ is larger than $\delta_2$, the population density converges to a “remnant" state, everywhere below this small threshold. Our results are shown to be useful for studying the relationships between environmental fragmentation and maximum sustainable yield from populations. We present numerical results in the case of stochastic environments.