In signal processing, the Nyquist–Shannon sampling theorem says you need at least 2 samples of a frequency to be able to perfectly reconstruct it. So in your question, a sampling rate of $200\: \mathrm{MHz}$ means you can perfectly reconstruct frequencies in the range of $0 - 100\: \mathrm{MHz}$. So what happens when frequencies above $100\: \mathrm{MHz}$ are present? They fold over (are aliased) into the the $0 - 100\: \mathrm{MHz}$ range and the fold-over point at $100\: \mathrm{MHz}$ is the Nyquist Edge.
For example:
There is a pretty good article about this behavior at here.