I'm not sure I completely understand your question (the document does have a combination of path redundancy and daisy-chain), but I can explain the problem with daisy-chains.
Normally, you want path redundancy. That is why the Internet was developed in the first place. It allowed traffic to automatically be routed in a different direction in case of damage (being funded by the DoD, nuclear war). Path redundancy eliminates single-points-of-failure (SPoFs).
When you daisy-chain paths, you introduce SPoFs. If one link or device in the daisy-chain fails, then the two resulting parts of the daisy-chain are isolated. This can be a problem, and the daisy-chain is depicted in Figure 5 of the document to which you link:
If any one link fails, then you have isolated one or more sites.
Figure 6 shows the combination of redundant paths and a daisy-chain that resolves the SPoF:
If any one link fails, then there is a redundant path that will prevent any site from being isolated. Routing must take into account that there is a loop, and it must prevent traffic from trying to loop. That is built into most routing protocols, and IP itself has a failsafe with the TTL.