Interesting effect. I'd say the area around the plateau works better than the stringy ranges, because it IS an area. While symbolic ranges work okay as single linear beads-on-a-string, 3-d realistic ones don't. A real range would be all convoluted and multithreaded and of varying width - so the somewhat photorealistic style clashes. If you used that method and started with a whole spiderweb of ridgelines, you might get a better look.... but at the cost of maybe adding more detail than is warranted.

What with easy access to satellite imagery, we nowadays expect photo-like views to BE photo-like .... capturing every lick of data. It's a tough issue to find a compromise on.

The Andes is a seriously linear feature. Yet look at an orbital view of it around this scale, and you'll see many parallel ridges, gaps, peaks with ridges radiating in many directions, clumps and sparse spots.

I don't have a suggested way to handle the discrepancy, and if the way you're doing it works for you, ignore my blather :-).