The Improbability Pump
Natural selection has always been the most contested part of evolutionary theory. Many people who have no problem with evolution bridle at the thought that it's all driven by a mindless and unguided natural process. Indeed, while most scientists accepted the notions of evolution and common ancestry soon after Darwin proposed them in 1859, natural selection wasn't widely accepted by biologists until about 1930. The main problem was, and still is, a paucity of evidence. While the idea of natural selection seems eminently sound, people want to see it actually changing species in nature. And since the process is usually very slow, that evidence is hard to get for living organisms and nearly impossible for fossils.
It's this difficulty that leads Dawkins to observe that natural selection is on wobblier legs than the other tenets of evolutionary theory, such as evolutionary change and the branching pattern of life. "Nowadays it is no longer possible to dispute the fact of evolution itself--it has graduated to become a theorum," Dawkins writes, using a neologism for a scientific theory, "or obviously supported fact--but it could (just) be doubted that natural selection is its major driving force."
Well, maybe, but I think Dawkins is a bit too timid in his defense of natural selection. While biologists agree that natural selection is not the only cause of genetic change in populations, the evidence is strong that it's the only one that can produce the remarkable adaptations of animals and plants to their environment--the elephant's trunk, the cactus's spines, the tiger's fangs and so on--the designlike quality of organisms that, as Darwin put it, "most justly excites our admiration." For one thing, none of the alternatives seem to work. A once-popular rival of natural selection, for instance, was Lamarckism, the idea that the changes acquired by an individual during its lifetime could somehow be imprinted onto its genes and passed on to succeeding generations. This idea failed for two reasons. First, it is incapable of explaining "design" in general: most adaptations, like the tiger's fangs and cactus's spines, can't be credibly explained as acquired changes that later became genetic. Further, it turns out that almost no acquired changes are inherited. One example: despite millenniums of circumcision, Jewish boys are still obstinately born with foreskins.
In contrast, artificial selection has been stunningly successful. Virtually everything that we eat, grow or pet has involved transforming a wild species, through selective breeding, into something radically different. (Bear in mind that the ancestor of the Chihuahua is the wolf.) And of the thousands of selection experiments performed on species in the laboratory, I know of fewer than a dozen that have failed to elicit a response. Why is this relevant to natural selection? As Dawkins observes, "Artificial selection is not just an analogy for natural selection. Artificial selection constitutes a true experimental--as opposed to observational--test of the hypothesis that selection causes evolutionary change." That's because both processes inexorably result from genetic variation that is adaptive in the current environment, with the "environment" in one case dominated by humans who decide which individuals get to live and breed.
Further, the kinds of adaptations that we find--and don't find--in nature are precisely what we'd expect if they were built by natural selection. Natural selection, for instance, can't produce features in one species that are good only for members of a different species. And we never see such features: if one species does something that helps another species, it also helps itself (cleaner fish remove parasites from other species, but thereby get a free meal). Natural selection also predicts that "altruistic" behaviors should be preferentially directed at relatives, who carry the same genes. This, too, is what we see, starting with one's closest relatives--siblings and children. Natural selection builds features that benefit individuals, not populations or species. As expected, we find features beneficial to individuals but harmful to groups (when male lions usurp a pride of females, they kill the females' cubs, earning them the right to reproduce but reducing the population of lions).
Finally, we've observed natural selection in real time: bacteria evolve resistance to antibiotics, plants to herbicides, insects to insecticides. This is genuine natural selection, even though the species are responding to human interference with the environment. And if you don't like those, biologists have cataloged dozens of "real" cases of natural selection in which species ranging from plants to birds adapt to natural changes in the environment.
Of course, it's impossible to prove that every useful feature of organisms was built by selection. The evolution of some species happened when we weren't around, and other species are impossible to experiment on. Nevertheless, the evidence for selection is sufficiently strong that, although Dawkins notes that the importance of selection might "(just) be doubted," he also observes, "All reputable biologists go on to agree that natural selection is one of [evolution's] most important driving forces, although--as some biologists insist more than others--not the only one."
Although F&P aren't biologists, they couldn't disagree more. While accepting the notion of evolution (grudgingly admitting that it's "perfectly possible--in fact entirely likely"), they assert, loudly and repeatedly, that natural selection is not just wrong but "quite possibly fatally wrong," not just flawed but "irredeemably flawed." They conclude that natural selection "falls radically short of explaining the appearance of new forms of life."