Here are my two cents:
1. A Fragility Curve (FC), like the one shown above, correlates 'probability' of occurrence of a well-defined 'damage state' or condition, in case a 'structure' is subjected to a given earthquake parameter.
2. In case of given FC, the 'damage state' is the Permanent Vertical Ground Displacement (PVGD), sustained by the 'structure' (which are actually two retaining walls (RWs), one 6.0 m & the other 7.5 m high), as a result of being subjected to various Peak Ground Accelerations (PGA) shown along horizontal axis of the curve. The 'probability' is shown along vertical axis. It is for a specific soil profile type C.
3. Various 'damage states' applicable to this FC are, as under:
a. 'Minor' means a PVGD value between 0.02 and 0.08.
b. 'Moderate' means a PVGD value between 0.08 and 0.22, whereas
c. 'Extensive/complete' means a PVGD value between 0.22 and 0.58.
4. This FC is comparing the probability of sustaining (or exceeding) the PVGD of the two RWs, when subjected to a given PGA. Continuous lines in the FC are for 7.5 m high RW, & dashed-lines for the 6.0 m high RW.
5. Now, if we want to know 'What is the probability of occurring 'Extensive/complete' damage (i.e., PVGD value of the two RWs at 0.6 PGA?', it is lesser for 6.0 m wall, and more for 7.5 m wall (Actual values may be read from the two lowest Red curves).
6. No. It is is not that way. This actually means that there is 25% probability that building will experience extensive damage (h=7.5m), 55% probability that it will experience moderate damage and 90% probability that will experience slight damage only. These probabilities are however independent of each other, and AFAIK, these are not added together.
7. This FC is part of a research paper that may be consulted at the following link, in case a deeper insight on the subject is required:
https://www.researchgate.net/publication/257102194_Development_of_fragility_functions_for_geotechnical_constructions_Application_to_cantilever_retaining_walls
Regards.