how do you know which part of a book is significant and important?
It's all important. Since you mention physics, I'll use the example of Griffiths' E&M book, which goes into quite a bit of detail about waveguides. Many professors (quite reasonably) skip this material, as it is "less significant" than other chapters which discuss general topics like Maxwell's Equations, the Wave Equation, radiation, and relativistic electrodynamics. Yet, waveguides are an entire industry (of which Griffiths only scratches the surface), so I would hesitate to say that the discussion is insignificant or unimportant.
Or is the described above just a feature of lectures?
I think the real feature of lectures is brevity. By turning a long book (or a portion of the book) into 42 bite-sized pieces, it is easier to make sense of things. And once you've had 42 bites, it is easier to read the book and fill in the rest of the gaps. In the absence of lectures, then, you could instead treat each subsection as a "bite", and try to understand it in isolation.
Context subjects: Physics and Mathematics
The key to most physics and math courses is the problem solving. Many introductory students make the mistake of reading the textbook, making beautiful notes, memorizing the examples, etc., but their first time working on a previously-unseen problem is during the exam. In my experience, the ideal balance is almost exactly the opposite -- students should spend ~90% of their time working problems, even if this means they don't read much of the book. For self-study, this means that the availability of a solutions manual is an absolutely essential criterion when choosing a textbook.