Dawkins' suggestion is that religious belief is not something that natural selection has selected for, but a by-product of something it has selected for. He draws an analogy with the moth and the flame - natural selection did not select moths that fly into flames, it selected for e.g. a form of navigation (keeping a fixed point of light - such as the moon - at a certain position in the visual field) that has this unfortunate by-product: moths navigating by a candle will spiral in and fry.
Dawkins provides an example of the sort of thing he has in mind. Children need to acquire a lot of information very quickly if they are to have a good chance of survival. They cannot carefully reason though things before accepting - they must take advice from the elders on trust. So natural selection selects for this. But this feature of young humans has a down side - it makes them vulnerable to bullshit beliefs. Bullshit beliefs can slip in, and, once they slip in, they'll be passed down the generations along with the genuinely useful ones. Religion takes advantage of precisely this vulnerability.
Dawkins suggests we may also have an innate propensity towards dualism, falling in love, and understanding things intentionally (as intentional behaviour done by an agent), which again religion can take advantage of.
In the second part of the chapter, Dawkins applies memetics to religion, saying "memetic natural selection of some kind seems to me to offer a plausible account of the detailed evolution of particular religions" (p. 233) Dawkins was saying this kind of thing in "Viruses of the Mind" (1991 - available here) in which he compares religious belief to a computer virus.
I find very plausible the suggestion that religious belief can, to a large extent, be explained in these ways. But of course the suggestion upsets many religious folk. And of course memetics has its own critics, in any case.
The question I will put to you is - assuming that these explanations do have much to them, to what extent is this really a threat to the reasonableness of religious belief? To what extent does this kind of causal explanation of religious belief threaten its claims to truth?