The “Big Four,” part I: Natural selection

This post is the first in a special series about four fundamental forces in evolution: natural selection, mutation, genetic drift, and migration.

This post was chosen as an Editor's Selection for ResearchBlogging.orgAmong non-biologists, the best-known of the Big Four forces of evolution is almost certainly natural selection. We’ve all heard the catchphrase “survival of the fittest,” and that’s a pretty good, if reductive, summing up of the principle. In more precise terms, here’s how natural selection works:

  • Natural populations of living things vary. Deer vary in how fast they can run, plants vary in how much drought they can tolerate, birds vary in their ability to catch prey or collect seeds—no two critters of the same species are exactly alike.
  • Some of those variable traits determine how many offspring living things have. How well you avoid predators, fight off disease, and collect food all determine how many babies you can make.
  • Many of those variable traits are heritable, passed on from parents to offspring. Faster deer usually have faster fauns; drought-tolerant plants make drought-tolerant seeds.

With these three conditions in place, natural selection occurs: heritable traits that help make more babies become more common. That is, if you have a trait that lets you support more offspring than your neighbor, you’ll have more children than your neighbor, and they’ll have more children than your neighbor’s children, and so on.

Fitness-versus-phenotype regressions for directional, stabilizing, and disruptive selection. Graphic by jby.

Measuring selection

Put this way, natural selection is simply a relationship between fitness, the number of offspring an organism can produce (often reported in comparison to the rest of the local population), and phenotype, the value of one or more traits of that organism (wing length, running speed, number of flowers produced, &c). Biologists can measure selection in natural populations by estimating this relationship between fitness (or a proxy for fitness, like growth rate), and phenotypes. Such an analysis should produce something like the regression graphs to the right, in which the relationship might be directional, with greater- (or smaller-) than-average phenotypes having greater fitness; stabilizing, with the average phenotype value having greater fitness; or disruptive, with extreme phenotype values having greater fitness. The slope of the line, or the shape of the curve, is a measure of the strength of natural selection [PDF] on an organism’s phenotype. This approach to measuring selection has been widely applied, and in 2001 a group of biologists led by Joel Kingsolver collected more than 2,500 estimates of the strength of natural selection [PDF].

How strong is selection?

Kingsolver et al. found that selection was usually surprisingly weak. Studies with the largest sample sizes, and the most statistical power to detect selection, mostly found directional selection strength (that is, the slope of the fitness-phenotype regression) less than 0.1, and the strength of stabilizing or disruptive selection was similarly low. Does this mean selection doesn’t matter in the short-term evolution of natural populations?

Probably not. The average selection strength estimates from the Kingsolver et al. dataset are actually stronger than selection strength assumed in most mathematical models of evolution. Furthermore, the collected estimates of selection had “long tailed” distributions—a small number of studies found quite strong selection, up to ten times as strong as the average. So maybe rare but strong bouts of selection have disproportionate impact over the long term.

Peter and Rosemary Grant have documented decades of shifting natural selection on Darwin’s finches (Geospiza spp.). Photo by Igooch.

Taking the finch by the beak

Part of the problem with assessing selection in nature is that most datasets measure selection over just one or a few years. One exception is the case of Darwin’s finches in the Galapagos Islands. The Galapagos offers a wide variety of habitat types, and experiences substantial year-to-year environmental variation—a landscape that should exert all sorts of natural selection on its occupants. Peter and Rosemary Grant have studied Galapagos finches for decades now, and found that selection is continuously at work on these unassuming birds. (The Grants’ book How and Why Species Multiply sums up their research program for a lay audience.)

Much of the Grants’ work has focused on the finches’ beaks, which largely determine what food the birds can eat. The distribution of seed sizes available on different Galapagos islands strongly predicts [PDF] the size of finches’ beaks on those islands. In 1989, the Grants published estimates of selection on beak size in the finch species Geospiza conirostris following a drastic wet-to-dry climactic shift that radically changed what foods were available to the finches. They found strong selection [$a], with fitness-phenotype regression slopes as high as 0.37. What’s more, the direction of selection changed dramatically from a very wet year to the dry year immediately afterward, as the finches were forced to move from feeding on small seeds and arthropods—which gave the advantage to shorter beaks—to hard-to-crack seeds, which required deep beaks.

The Grants’ longer-term study of selection on Galapagos finches confirms this image of selection swinging back and forth unpredictably [PDF]. From 1972 to 2001, they tracked populations of the finch species G. scandens and G. fortis, and saw both more gradual long-term changes in the finches’ body size and beak measurements as well as sudden sharp shifts. These changes continually altered the ability of the two species to hybridize, so that some years they were more reproductively isolated than others—and conditions in any one year were poor indicators of what would be going on five, ten, or twenty years later.

So when does selection matter?

The Grants’ study makes natural selection look as shifting and impermanent as the wind. How can it shape patterns of evolution over millions of years, then? One possibility is that trends may emerge over longer periods of time, as wobbly selection moves species in new directions in a drunkard’s walk, with two steps forward, then one step back, then four steps forward. Another is that lasting trends only occur when speciation intervenes to lock in fleeting changes due to variable natural selection [$a]. Much also depends on how selection interacts with mutation, genetic drift, and migration, as I’ll discuss in the rest of this series.

And here’s a shameless plug for a t-shirt. Photo by jby.


Futuyma, D. (1987). On the role of species in anagenesis. The American Naturalist, 130 (3), 465-73 DOI: 10.1086/284724

Grant, B.R., & Grant, P.R. (1989). Natural selection in a population of Darwin’s finches. The American Naturalist, 133 (3), 377-93 DOI: 10.1086/284924

Grant, P.R., & Grant, B.R. (2002). Unpredictable evolution in a 30-Year study of Darwin’s finches. Science, 296 (5568), 707-11 DOI: 10.1126/science.1070315

Grant, P.R. and B.R. Grant. (2008) How and Why Species Mutliply: The Radiation of Darwin’s Finches. Princeton University Press. Google Books.

Kingsolver, J., Hoekstra, H., Hoekstra, J., Berrigan, D., Vignieri, S., Hill, C., Hoang, A., Gibert, P., & Beerli, P. (2001). The Strength of phenotypic selection in natural populations. The American Naturalist, 157 (3), 245-261 DOI: 10.1086/319193

Lande, R. (1976). Natural selection and random genetic drift in phenotypic evolution. Evolution, 30 (2), 314-34 DOI: 10.2307/2407703

Johnson, T., & Barton, N. (2005). Theoretical models of selection and mutation on quantitative traits. Phil. Trans. R. Soc. B, 360 (1459), 1411-25 DOI: 10.1098/rstb.2005.1667

Schluter, D., & Grant, P. (1984). Determinants of morphological patterns in communities of Darwin’s finches. The American Naturalist, 123 (2), 175-96 DOI: 10.1086/284196


Back to basics: The “Big Four”

ResearchBlogging.orgThe nice thing about a field season away from all regular internet access is that it gives you a real sabbatical of a sort—a chance to reassess plans and set new goals. One of the new goals I set myself this last field season was to introduce a new kind of topic here at Denim and Tweed.

Most of my writing about science at D&T focuses on recently published discoveries in evolution and ecology. It’s fun writing, and it coincides neatly with my regular journal reading, and I intend to keep doing it. But I’ve discovered that when I want to put new work in context, I often need to discuss fundamental concepts of evolutionary biology that aren’t necessarily common knowledge, such as genetic drift or sexual selection. However, I rarely have room to explain these concepts in depth within a blog post devoted to something else.

So maybe the solution is to devote some posts to explaining these “basics.” I’m going to start with a series of posts on the “Big Four” processes of population genetics. These are the four processes that account, in one way or another, for every change in the frequency of genes within natural populations. In other words, the Big Four account for much of evolution itself. They are:

  • Natural selection, changes in gene frequencies due to fitness advantages, or disadvantages, associated with different genes.
  • Mutation, the source of new forms of genes;
  • Genetic drift, or changes in gene frequencies that arise from the way probability works in finite populations; and
  • Migration, or changes in gene frequencies due to the movement of organisms from site to site.

Lay readers may be surprised both by what we know, and what we don’t, about how these four processes operate in nature. Natural selection is relatively easy to measure, and apparently ubiquitous [PDF] in natural populations—but we don’t know how often the resulting short-term changes impact evolution over millions of years. Mutation, the source of variation on which natural selection acts, seems to vary widely among living things. Genetic drift means that a trait can come to dominate a population even if it has no fitness effect—or sometimes a deleterious one. Finally, migration across variable landscapes can interact with selection, drift, and mutation [$a] to completely alter their effects.

I’ll devote one post each to selection, mutation, drift, and migration, discussing classic findings as well as more recent scientific discoveries about each. They’ll arrive as my usual mid-week science posts for the next four weeks, and I’ll update this post with links to the others as they go online—so if this looks worth following, you can either bookmark this post, or subscribe to D&T’s RSS Feed.

Natural selection, mutation, genetic drift, and migration act together to shape the evolution of natural populations. Photo by jby.


Drake JW, Charlesworth B, Charlesworth D, & Crow JF (1998). Rates of spontaneous mutation. Genetics, 148 (4), 1667-86 PMID: 9560386

Kingsolver, J., Hoekstra, H., Hoekstra, J., Berrigan, D., Vignieri, S., Hill, C., Hoang, A., Gibert, P., & Beerli, P. (2001). The strength of phenotypic selection in natural populations. The American Naturalist, 157 (3), 245-61 DOI: 10.1086/319193

Slatkin, M. (1987). Gene flow and the geographic structure of natural populations. Science, 236 (4803), 787-92 DOI: 10.1126/science.3576198

Wright S (1931). Evolution in Mendelian populations. Genetics, 16 (2), 97-159 PMID: 17246615


Caught between birds and squirrels, limber pines go both ways

ResearchBlogging.orgResponding to natural selection often means compromising between different selective forces. A brief paper published online early at Evolution documents one such case – limber pine trees’ compromise between protecting their seeds from squirrels, and making them accessible to the birds that disperse them. Pulled between these conflicting selective sources, some limber pine populations grow cones in a wider variety of shapes [$a].

Cones of the limber pine must balance protection against squirrels with accessibility for seed-dispersing Clark’s nutcrackers. Photos by Fool-On-The-Hill and almiyi.

Seeds of the limber pine are cache-dispersed by Clark’s nutcrackers. That is, the birds collect pine seeds to cache as a winter food source, but they collect many more than they need, and forget lots of them, and forgotten seeds are often able to sprout. However, squirrels also like pine seeds, and they harvest them before the birds start caching seeds. These two seed-harvesters generate conflicting selection on limber pine cones [$a]. Nutcrackers go for cones with lots of seeds protected by thinner scales; but so do squirrels.

However, pine-nut-eating squirrels are not present everywhere limber pines grow. The new study’s authors, Siepielski and Benkman, take advantage of this quirk of distributions to perform a natural experiment, comparing pines that only need to satisfy their seed dispersers with pines that also need to defend against seed predators. Surveying cone shapes in populations of each class, they found that limber pine populations facing conflicting selection were bimodal, with trees mainly growing either squirrel-defended short, thick-scaled cones, or nutcracker-friendly longer, thin-scaled cones. Populations growing in regions without squirrels produced only nutcracker-friendly cones.

This apparently simple pattern conceals more complicated dynamics – in fact, as the authors disclose in the Discussion section, many other limber pine populations are solely composed of trees producing squirrel-defended cones. This is because, when pines establish in areas with large squirrel populations, nutcrackers may never colonize the area, or may visit less frequently and disperse fewer seeds. Without nutcracker dispersal, seeds are mainly dispersed after the cones fall, by rodent species that (unlike squirrels) forage on the ground. This makes squirrel defense the only selective priority. Populations displaying both cone types probably only arise in unique conditions, the authors say, where squirrels are present but not at high density.


Siepielski, A., & Benkman, C. (2007). Convergent patterns in the selection mosaic for two North American bird-dispersed pines. Ecological Monographs, 77 (2), 203-20 DOI: 10.1890/06-0929

Siepielski, A., & Benkman, C. (2009). Conflicting selection from an antagonist and a mutualist enhances phenotypic variation in a plant. Evolution DOI: 10.1111/j.1558-5646.2009.00867.x


Un-bear-able (ha) predation creates variable natural selection

ResearchBlogging.orgNatural selection is a fact of life. As Steven Jay Gould put it, it’s an “inescapable conclusion” arising from the “undeniable facts” that (1) populations of living things have inheritable variation in many traits; and (2) produce a surplus of offspring. But populations often experience selection from multiple sources, and in conflicting directions. The cover article for this month’s issue of Evolution suggests that bears may be creating ongoing selection in wild salmon populations, but the strength, and outcome, of that selection varies from stream to stream [$-a].

Selective agent at work (Flickr: Dr.DeNo)

Salmon are famously anadromous — they hatch in freshwater streams and swim out to sea, only to return to the stream of their birth to spawn before they die. Male salmon are generally better off if they’re bigger, both to maximize stored energy for the return to their spawning site, and to better compete for mates when they arrive. Natural selection for larger bodies, however, is checked by bears, who preferentially target large, fatty fish. Yet bear predation varies from stream to stream: in narrower streams, where salmon are easier to catch, bears can fill up on big, newly-arrived fish; but in wide streams, bigger fish can more easily evade bears, so bears tend to target older, weaker fish instead.

Continue reading


Evolution-proof insecticide?

ResearchBlogging.orgIn this week’s issue of PLoS Biology, an essay describes the perfect means for controlling malaria-carrying mosquitoes: an “evolution-proof insecticide.” By taking advantage of the life history traits of both mosquitoes and the malaria parasite, Read et al. argue it should be possible to create an insecticide that will cut malaria transmission without selecting for resistance in the mosquitoes.

Malaria remains a major public health problem in much of the world – according the World Health Organization, a child dies of the disease every 30 seconds, and the cost of malaria may cut economic growth by as much as 1.3% in countries with high infection rates. In the absence of a vaccine, the best approach for malaria management is to control the mosquitoes that transmit the malaria parasite. This is usually done with insecticides, but these have a limited useful lifespan, as they create strong selective pressure for mosquito populations to evolve resistance.

Photo by LoreleiRanveig.

As Read et al. point out, it’s not that we need to kill off mosquitoes as such; we just need to stop them from transmitting malaria. If this can be accomplished without strongly reducing the mosquitoes’ fitness, it would reduce or eliminate selection for resistance. Malaria typically needs a long time to incubate inside a mosquito before it becomes transmissible to humans, and, in what Read et al. call “one of the great ironies of malaria,” this incubation time is longer than most mosquitoes live. That is, the mosquitoes who successfully transmit malaria are the small proportion of the population who live long enough to incubate the parasite.

Here’s where evolutionary biology interacts with the life history of malaria parasites in a highly convenient way: an insecticide that selectively targets older mosquitoes will have a smaller impact on the mosquito population’s fitness. This is because most of a female mosquito’s fitness – the total number of offspring she produces – is concentrated in her first one or two egg-laying cycles. Her fitness can increase if she survives to complete more cycles, but it’s pretty rare that she does. From natural selection’s point of view, that first of eggs counts much more than possible future batches, because they’re not very likely.

For that hypothetical female mosquito to transmit malaria, she has to bite an infected human in the course of feeding to fuel one egg-laying cycle, then incubate the malaria parasites for an additional two to six cycles. Therefore, say Read et al., an insecticide that doesn’t harm mosquitoes until they complete their first few egg-laying cycles is the “evolution-proof” solution – the only offspring it “steals” from the affected mosquitoes were pretty improbable anyway, and it prevents the malaria parasites from incubating long enough to successfully infect a new human host.

As it happens, the evolution-proof insecticide might not be a chemical agent, but a biological one. A paper I discussed back in January suggested that infecting malaria-carrying mosquitoes with the parasitic Wolbachia bacterium could control mosquito populations [$-a] by, yes, reducing their total lifespan to something less than the malaria parasite’s incubation time. In short, it looks like the goal of a malaria-free world is not as improbable as it used to be.


McMeniman, C., Lane, R., Cass, B., Fong, A., Sidhu, M., Wang, Y., & O’Neill, S. (2009). Stable introduction of a life-shortening Wolbachia infection into the mosquito Aedes aegypti Science, 323 (5910), 141-144 DOI: 10.1126/science.1165326

Read, A., Lynch, P., & Thomas, M. (2009). How to make evolution-proof insecticides for malaria control PLoS Biology, 7 (4) DOI: 10.1371/journal.pbio.1000058