There are a lot of different theories for how は works. One theory is that は can follow case particles, but when it follows が or を, that particle is deleted:
が ＋ は ＝ は
を ＋ は ＝ は (although there is a literary をば < を + は)
に ＋ は ＝ には (although sometimes に is dropped)
で ＋ は ＝ では
へ ＋ は ＝ へは
と ＋ は ＝ とは
から ＋ は ＝ からは
より ＋ は ＝ よりは
According to Martin, は replaces が thirteen times more often than it replaces を, and I think it especially often replaces が at the beginning of a sentence.
So what is は replacing in your sentence, if anything? In your sentence, both を and に seem unlikely:
- With this verb, を marks the thing being discussed (in your example paraphrased with について).
- With this verb, で would be used instead of に, and で wouldn't be deleted. Also, if に were used anyway, I don't think it could be dropped before は—see discussion at the end of the answer.
Instead, I'll suggest that は is replacing が. The subject is この本:
この本が 自然主義について論じている（こと） (before topicalization)
がは 自然主義について論じている。 (after topicalization, が is deleted before は)
The verb 論じる can be used in several different ways. Although the book can be the subject, as in your example, so can the author, in which case the book could be mentioned in an adjunct marked with で:
著者が この本で 自然主義を論じている（こと） (before topicalization)
この本では 著者が自然主義を論じている。 (after topicalization)
The flexibility in this case reminds me of English, where we can say:
This book talks about naturalism.
The book is the subject.
In this book, the author talks about naturalism.
The book is part of an adjunct. The author is the subject.
And so we find examples like the following (collected from the web and BCCWJ):
At the end of your post, you give an example where に can be dropped. So the question is, why can に sometimes be dropped? Well, I linked to a paper above that discusses the topic in more detail, but I think we can explain the example at the end of your question like this:
1a. 彼が 英語ができない（こと） (before topicalization)
がは 英語ができない。 (after topicalization, が is deleted before は)
2a. 彼に 英語ができない（こと） (before topicalization)
2b. 彼には 英語ができない。 (after topicalization)
Your examples are 1b and 2b. Of course, it's natural to use は there, but if we add something like こと and turn it into a subordinate clause, は disappears, leaving behind the basic case marking of the sentence (1a and 2a).
This is what Shibatani calls a non-canonical construction, and in this case either が or に is possible:
- When you add は to が, you get は.
- When you add は to に, you get には.
Since you can do either one, it gives the appearance that に is optional, but the real difference is が versus に. Of course, this doesn't apply to your 論じている example because に and が aren't in alternation there, so you're comparing apples and oranges.
(As an aside, adding こと does complicate things slightly, because it allows が-の conversion. But in this answer, I'm ignoring that fact.)