I didn't find definitions either, at first, but then I understood that they 'define by examples'. They are on page 398.
About Variability Risks, they say "uncertainty exists about some key characteristics of an event, activity," etc. Maybe you cannot properly analyze (qualitative, quantitative) and therefore cannot plan an appropriate Response. In that case they suggest to use Monte Carlo analysis to "reduce the spread of possible outcomes".
Likewise, they 'define' Ambiguity Risks with situations where you simply don't know enough to build a solid plan and get to work without worries. The suggested approach here is to get external expert help, benchmark, prototype, etc.
I hope it helps!