I have spoken of different continents to make it clear, but the principle of geographical separation as a barrier to interbreeding can apply to animals on different sides of a desert, a mountain range, a river, or even a motorway. It can also apply to animals separated by no barrier other than sheer distance. Shrews in Spain cannot interbreed with shrews in Mongolia, and they can diverge, evolutionarily speaking, from shrews in Mongolia, even if there is an unbroken chain of interbreeding shrews connecting Spain to Mongolia. Nevertheless the idea of geographical separation as the key to speciation is clearer if we think in terms of an actual physical barrier, such as the sea or a mountain range. Chains of islands, indeed, are probably fertile nurseries for new species.
Here, then, is our orthodox neo-Darwinian picture of how a typical species is 'born', by divergence from an ancestral species. We start with the ancestral species, a large population of rather uniform, mutually interbreeding animals, spread over a large land mass. They could be any sort of animal, but let's carry on thinking of shrews. The landmass is cut in two by a mountain range. This is hostile country and the shrews are unlikely to cross it, but it is not quite impossible and very occasionally one or two do end up in the lowlands on the other side. Here they can flourish, and they give rise to an outlying populations of the species, effectively cut off from the main population. Now the two populations breed and breed separately, mixing their genes on each side of the mountains but not across the mountains. As time goes by, any changes in the genetic composition of one population are spread by breeding throughout that population but not across to the other population. Some of these changes may be brought about by natural selection, which may be different on the two sides of the mountain range: we should hardly expect weather conditions, and predators and parasites, to be exactly the same on the two sides. Some of the changes may be due to chance alone. Whatever the genetic changes are due to, breeding tends to spread them within each of the two populations, but not between the two populations. So the two populations diverge genetically: they become progressively more unlike each other.
They become so unlike each other that, after a while, naturalists would see them as belonging to different 'races'. After a longer time, they will have diverged so far that we should classify them as different species. Now imagine that the climate warms up so that travel through the mountain passes becomes easier and some of the new species start trickling back to their ancestral homelands. When they meet the descendants of their long-lost cousins, it turns out that they have diverged so far in their genetic makeup that they can no longer successfully interbreed with them. If they do hybridise with them the resulting offspring are sickly, or sterile like mules. So natural selection penalises any predilection, on the part of individuals on either side, towards hybridising with the other species or even race. Natural selection thereby finishes off the process of 'reproductive isolation' that began with the chance intervention of a mountain range. 'Speciation' is complete. We now have two species where previously there was one, and the two species can coexist in the same area without interbreeding with one another.
