Diamond has provided (elsewhere) a convenient definition for "collapse" as "drastic decrease in human population numbers and/or in political, economic, or social complexity"; a definition which, of course, includes all civilizations that ever existed. His definition also seems to include any place on Earth where new civilizations replaced old civilizations (for example, Diamond often mentions Java as a place where "societies have existed continuously for thousands of years", but the people of today's Java are Muslims, which is not what Java was "thousands of years ago").

The fundamental weakness of his approach is that he does not use a word of caution at the very beginning. Much of what he writes is based on our understanding of how the environment used to be. That is an approximation at best. There were certainly a lot more forests, but that is true for all places on Earth. Beyond such trivial statements, it is really hard to reconstruct how the environment looked like when a civilization was at its peak.

Diamond's thesis is that some societies ended up destroying the very source of their livelihood. The thought never occurs to Diamond that a society might simply change source of livelihood. If we run out of wood, we can build houses using a different material. If we run out of boats, we can simply start farming or hunting instead of fishing. The history of humankind is not a mechanical repetition of the same lifestyle day after day. Quite the contrary. Many individuals of the Silicon Valley have been software engineers all their lives. Now that jobs for software engineers are scarce, many of those engineers are simply changing career.

His list of pre-industrial societies that collapsed include such obscure civilizations as Cahokia (in Missouri) but does not include such world-famous civilizations as the Mongols and the Arabs, which built empires slightly bigger than Cahokia. He also spends precious little time on Athens, Rome and Persia. He carefully avoids civilizations such as China and Britain, that completely destroyed their original territory and became huge powers (China "is" the only civilization that has lasted thousands of years). He also neglects India, a good example of a region where societies caused little or no environmental damage but had a tendency to disappear very quickly: perhaps precisely because, unlike the British and the Chinese, they did not massively exploit their natural resources?

Diamond offers an explanation for the Rwandan holocaust (a such vague explanation that I would not know how to summarize it), but ignores the fact that many overcrowded places in the world never experienced anything even remotely similar to what happened in Rwanda. He discusses Haiti (which he considers "the poorest country in the New World" because he never traveled to Cuba) but fails to compare it with comparable countries such as Philippines or Indonesia that had similar histories. What conclusions to draw from these examples is not clear.

In general, he offers one remote, obscure, and often unverifiable example that supports his theory but blatantly ignores obvious counterexamples that prove just the opposite.

Diamond seems to think that Japanese society qualifies as a "thousand-year old" civilization, and praises the fact that Japan has been capable of re-foresting what it deforested. Even assuming that the Japanese civilization was not destroyed by the USA in 1945, Diamond neglects the fact that the Japanese people, far from being proud of their environmental wisdom, commit suicide at a rate higher than any other people on the planet.

The truth is that the well-known civilizations don't quite fit his rules, and thus he prefers to dwell on little-known civilizations such as the Mayas and the Vikings of Greenland (one hundred pages!), that have one thing in common: they wrote very little, so we know very little, so we can claim pretty much anything we like about the causes of their "collapse". He thinks that "the clearest examples of collapses of isolated societies involve remote Polynesian islands. One wonders how this collapse could be so "clear" if those societies did not leave us any document: we don't even know how they functioned. We might have "theories" on what happened there, but "clearest" sounds a bit exaggerated.

(And, by the way, far from collapsing, the Norse did pretty well: Norway is now the wealthiest country in the world).

Personally, I think that the secret in survival of a civilization rests in the ability to exploit the environment. Some civilizations did it better than others, that's all. And Darwinian selection took care of the ones that did it less well. Deforestation per se did not cause anybody's collapse: in some cases (from ancient Mesopotamia to western Europe) caused a political, economic, cultural and technological boom. In other cases (the ones that Diamond focuses on), it may (may) have contributed to their decline. It's called "mismanagement".

Diamond would have a hard time explaining why Napoleon lost and why Alexander won. Others have explained both facts, quite convincingly, without any need to discuss environmental issues.

A more interesting title for Diamond's book would have been: "How Societies Choose to Read Books like This One". There is a passion for hearing that we are laying the foundations for our own apocalypse, a passion that has been around for thousands of years.
Other than that, there is little substance in what Diamond writes. His premises are way too weak to justify his (very strong) conclusions.
I recommend instead "Forest and Civilisations", edited by Yoshinori Yasuda, written by leading scholars of environmental archeology.