Why do most vegetation monitoring programs rely on permanent plots? If plots are randomly sampled over a homogeneous vegetation community, why not just establish new plots each time, rather than locating and re-sampling the same plots over and over again?

Why do most vegetation monitoring programs rely on permanent plots?

Take the context of a single ecological restoration site, for example. Let's say the variable of interest is cover by plant species, and monitoring is conducted throughout several seasons to track the relative/total cover of multiple species.

If plots are randomly sampled within homogeneous vegetation assemblages, why not just establish new plots each sampling period, rather than locating and re-sampling the same plots over and over again?

If the goal of sampling the is to represent characteristics of the population (i.e. the homogeneous area of vegetation), then why do researchers always sample permanent monitoring plots again and again rather than using a new set of randomly selected plots each time?