Why Evolution Is True (30 page)

Exactly how these barriers arise puzzled biologists for a long time. Finally, around 1935, biologists began to make headway in both the field and laboratory. One of the most important observations was made by naturalists, who noticed that so-called “sister species”—species that are each other’s closest relatives—were often separated in nature by geographical barriers. Sister species of sea urchins, for example, were found on opposite sides of the Isthmus of Panama. Sister species of freshwater fish often inhabited separated river drainages. Could this geographic separation have something to do with how these species arose from a common ancestor?

Yes, said the geneticists and naturalists, and they eventually proposed how the combined effects of evolution and geography could make this happen. How do you get one species to divide into two, separated by reproductive barriers? Mayr argued that these barriers were merely the by-products of natural or sexual selection that caused geographically isolated populations to evolve in different directions.

Suppose, for example, that an ancestral species of flowering plant was split into two portions by a geographic barrier, like a mountain range. The species may, for example, have been dispersed over the mountains in the stomachs of birds. Now imagine that one population lives in a place having a lot of hummingbirds but only a few bees. In that area, the flowers will evolve to attract hummingbirds as pollinators: typically the flowers would become red (a color that the birds find attractive), produce copious nectar (which rewards birds), and have deep tubes (to accommodate hummingbirds’ long bills and tongues). The population on the other side of the mountain may find its pollinator situation reversed: few hummingbirds but many bees. There the flowers will evolve to attract bees; they may become pink (a color bees favor), and evolve shallow nectar tubes with less nectar (bees have short tongues and don’t require a large nectar reward) as well as flatter flowers whose petals form a landing platform (unlike hovering hummingbirds, bees usually land to collect nectar). Eventually, the two populations would diverge in the form of their flowers and amount of their nectar, and each would be specialized for pollination by only a single type of animal. Now imagine that the geographic barrier disappeared, and the newly diverged populations found themselves back in the
same
area—an area containing both bees and hummingbirds. They would now be reproductively isolated: each type of flower would be served by a different pollinator, so their genes would not mix via cross-pollination. They would have become two different species. This is in fact the likely way that the monkeyflowers we considered earlier
did
diverge from their common ancestor.

This is just one way that a reproductive barrier can evolve by “divergent” selection—that is, selection that drives different populations in different evolutionary directions. You can imagine other scenarios in which geographically isolated populations diverge so that later they could not interbreed. Different mutations affecting male behaviors or traits could appear in different places—say, longer tail feathers in one population and orange color in another—and sexual selection might then drive the populations in different directions. Eventually, females in one population would prefer long-tailed males, and females in the other, orange males. If the two populations later encountered each other, their mating preferences would prevent them from mixing genes, and they would be considered different species.

What about the sterility and inviability of hybrids? This was a big problem for early evolutionists, who had trouble seeing how natural selection could yield such palpably maladaptive and wasteful features. But suppose that these features were not selected directly, but were simply accidental by-products of genetic divergence, divergence caused by natural selection or genetic drift. If two geographically isolated populations evolve along different pathways long enough, their genomes can become so different that, when they’re put together in a hybrid, they just don’t work well together. This can disrupt development, causing hybrids to either die prematurely or, if they live, turn out to be sterile.

It’s important to realize that species don’t arise, as Darwin thought, for the purpose of filling up empty niches in nature. We don’t have different species because nature somehow needs them. Far from it. The study of speciation tells us that
species are evolutionary accidents
. The “clusters” so important for biodiversity don’t evolve because they increase that diversity, nor do they evolve to provide balanced ecosystems. They are simply the inevitable result of genetic barriers that arise when spatially isolated populations evolve in different directions.

In many ways biological speciation resembles the “speciation” of two closely related languages from a common ancestor (an example is German and English, two “sister tongues”). Like species, languages can diverge in isolated populations that once shared an ancestral tongue. And languages change more rapidly when there is less mixing of individuals from different populations. While populations change genetically via natural selection (and sometimes genetic drift), human languages change by linguistic selection (appealing or useful new words get invented) and linguistic drift (pronunciations change due to imitation and cultural transmission). During biological speciation, populations change genetically to the extent that their members no longer recognize each other as mates, or their genes can’t cooperate to produce a fertile individual. Likewise, languages can diverge to the extent that they become mutually unintelligible: English speakers don’t automatically understand German, and vice versa. Languages are like biological species in that they occur in discrete groups rather than as a continuum: the speech of any given person can usually be placed unambiguously in one of the several thousand human languages.

The parallel goes even further. The evolution of languages can be traced back to the distant past, and a family tree drawn up, by cataloging the similarities of words and grammar. This is very like reconstructing an evolutionary tree of organisms from reading the DNA code of their genes. We can also reconstruct protolanguages, or ancestral tongues, by looking at the features that descendant languages have in common. This is precisely the way biologists predict what missing links or ancestral genes should look like. And the origin of languages is accidental: people don’t start to speak in different tongues just to be different. New languages, like new species, form as a by-product of other processes, as in the transformation of Latin to Italian in Italy. The analogies between speciation and languages were first drawn by—who else?—Darwin, in
The Origin.

But we shouldn’t push this analogy too far. Unlike species, languages can “cross-fertilize,” adopting phrases from each other, like the English use of the German
angst
and
kindergarten.
Steven Pinker describes other striking similarities and differences between the diversification of languages and species in his engrossing book
The Language Instinct.

The idea that geographic isolation is the first step in the origin of species is called the
theory of geographic speciation.
The theory can be stated simply: the evolution of genetic isolation between populations requires that they first be geographically isolated. Why is geographic isolation so important? Why can’t two new species just arise in the same location as their ancestor? The theory of population genetics—and a lot of lab experiments—tell us that splitting a single population into two genetically isolated parts is very difficult if they retain the opportunity to interbreed. Without isolation, selection that could drive populations apart has to work against the interbreeding that constantly brings individuals together and mixes up their genes. Imagine an insect living in a patch of woods that harbors two types of plants on which it can feed. Each plant requires a different set of adaptations to use it, for they have different toxins, different nutrients, and different odors. But as each group of insects within the area begins adapting to one plant, it also mates with insects adapting to the other plant. This constant intermixing will keep the gene pool from splitting into two species. What you will probably wind up with is just a single “generalist” species that uses both plants. Speciation is like separating oil and vinegar: though striving to pull apart, they won’t do so if they’re constantly being mixed.

What is the evidence for geographic speciation? What we’re asking about here is not
whether
speciation happens, but
how.
We already know from the fossil record, embryology, and other data that species diverged from common ancestors. What we really want to see is geographically separated populations turning into new species. This is no easy task. First of all, speciation in organisms other than bacteria is usually slow—much slower than the splitting of languages. My colleague Allen Orr and I calculated that, starting with one ancestor, it takes roughly between 100,000 and five million years to evolve two reproductively isolated descendants. The glacial pace of speciation means that, with a few exceptions, we can’t expect to witness the whole process, or even a small part of it, over a human lifetime. To study how species form we must resort to indirect methods, testing predictions derived from the theory of geographic speciation.

The first prediction is that if speciation depends largely on geographical isolation, there must have been lots of opportunities during the history of life for populations to experience that isolation. After all, there are millions of species on earth today. But geographic isolation is common. Mountain ranges rise, glaciers spread, deserts form, continents drift, and drought divides a continuous forest into patches separated by grassland. Each time this happens, there is a chance for a species to be sundered into two or more populations. When the Isthmus of Panama was formed about three million years ago, the emerging land separated populations of marine organisms on either side, organisms that originally belonged to the same species. Even a river can serve as a geographical barrier for many birds that don’t like to fly over water.

But populations don’t have to become isolated by the formation of geographic barriers. They might simply become separated by accidental long-distance dispersal. Suppose that a few wayward individuals, or even a single pregnant female, go astray and end up colonizing a distant shore. The colony will thereafter evolve in isolation from its mainland ancestors. This is just what happens on oceanic islands. The chances for this kind of isolation through dispersal are even greater on archipelagoes, where individuals can occasionally move back and forth between neighboring islands, each time becoming geographically isolated. Each round of isolation provides another chance for speciation. This is why archipelagoes harbor the famous “radiations” of closely related species, such as the fruit flies of Hawaii, the
Anolis
lizards of the Caribbean, and the finches of the Galapagos.

There’s been ample opportunity for geographic speciation, then, but has there been enough time? That too is not a problem. Speciation is a splitting event, in which each ancestral branch splits into two twigs, which themselves split later, and so on as the tree of life ramifies. This means that the number of species builds up exponentially, although some branches are pruned through extinction. How fast would speciation need to be to explain the present diversity of life? It’s been estimated that there are 10 million species on earth today. Let’s raise that to 100 million to take into account undiscovered species. It turns out that if you started with a single species 3.5 billion years ago, you could get 100 million species living today even if each ancestral species split into two descendants only once every 200
million years.
As we’ve seen, real speciation happens a lot faster than that, so even if we account for the many species that evolved but went extinct, time is simply not a problem.
³⁸

What about the critical idea that reproductive barriers are the by-product of evolutionary change? That, at least, can be tested in the laboratory. Biologists do this by performing selection experiments, forcing animals or plants to adapt through evolution to different environments. This is a model of what happens when isolated natural populations encounter different habitats. After a period of adaptation, the different “populations” are tested in the lab to see if they’ve evolved reproductive barriers. Since these experiments take place over tens to dozens of generations, while speciation in the wild takes thousands of generations, we can’t expect to see the origin of full species. But we should occasionally see the beginnings of reproductive isolation.

Surprisingly, even these short-duration experiments quite often produce genetic barriers. More than half of these studies (there are about twenty of them, all done on flies because of their short generation time) give a positive result, often showing reproductive isolation between populations within a year after selection begins. Most often, adaptation to different “environments” (different types of food, for example, or the ability to move up versus down in a vertical maze) results in mating discrimination between the populations. We’re not sure exactly what traits the populations use to discriminate against each other, but the evolution of genetic barriers in such a short time confirms a key prediction of geographic speciation.

The second prediction of the theory involves geography itself. If populations must usually be physically isolated from one another to become species, then we should find the most recently formed species in different but nearby areas. You can get a rough idea of how long ago species arose by looking at the amount of difference between their DNA sequences, which is roughly proportional to the time elapsed since they split from a common ancestor. We can then look for “sister” species in a group who will have the greatest similarity in their DNA (and are thus most closely related), and see if they’re geographically isolated.