« February 2009 | Main | April 2009 »

March 28, 2009

Facts and theory in Coyne’s “Why evolution is true”

“But if you believe that primates and guinea pigs [both of which have mutated, nonfunctional versions of a gene for making vitamin-C] were specially created, these facts don’t make sense. Why would a creator put a pathway for making vitamin C in all these species and then inactivate it? Wouldn’t it be easier simply to omit the pathway from the beginning? Why would the same inactivating mutation be present in all primates, and a different one in guinea pigs? Why would the sequences of the dead gene exactly mirror the pattern of resemblance predicted from the known ancestry of these species? And why do humans have thousands of pseudogenes [DNA sequences very similar to genes that are functional in other species, but with mutations that make them inactive] in the first place?” – Jerry Coyne, Why Evolution is True

I have been reading and enjoying “Why evolution is true”, by Jerry Coyne. Here are some thoughts on what I’ve read so far.

The preface struck one wrong note, I thought, stating that “evolution is far more than a ‘theory’…evolution is a fact.” This seemed to accept the popular definition of theory as wild speculation. In our 2003 paper, Strong Inference, The Way of Science (published in American Biology Teaher, vol. 65, p. 419), Tom Kinraide and I used a dictionary definition of theory: “The coherent set of hypothetical, conceptual, and pragmatic principles forming the general frame of reference for a field of inquiry.” Consistent with this definition, I would have said that “evolution is a theory (a set of explanatory principles) supported by thousands of facts.” Later in the book, Coyne has a good explanation of the difference between scientific and popular definitions of theory.

He explains evolutionary theory very well and presents a great selection of facts, all of which are explained by evolutionary theory but inconsistent with the religious claim that all species were created separately over a few days, only a few thousand years ago. There are other religious claims disproved by the facts, such as the claim made by fundamentalist Hawaiians that they are directly descended from taro plants – we are distant cousins, however, descended from a common ancestor -- but Coyne apparently sees Genesis literalists as the main threat to science education.

Kinraide and I also used a dictionary definition of fact: “An occurrence, quality, or relation the reality of which is manifest in experience or may be inferred with certainty.” We can infer with certainty that the earth is billions rather than thousands of years old and that humans and chimps are descended from a common ancestor who lived a few million years ago. Creationists might see the latter statement as the core of evolution, in which case evolution is indeed a fact, albeit an inferred fact. What evolutionary theory adds is explanations, answering questions like “how did that ancestral species split in two, and why are humans so much more intelligent than chimps?”

Here are a few of the manifest facts from as much of the book as I’ve read so far, and the inferred facts and/or theory to which they lead:

Manifest Fact: the facing coasts of Africa and South America match in shape, geological formations, and fossils. Satellite measurements show that the two continents are moving apart at about 6 cm per year. Inferred facts: the two continents were once joined and have been moving apart for millions of years. Six thousand years ago, when some fundamentalists think the earth was created, the two continents would have been only 180 meters closer together than they are today. Over longer time periods, the rate of movement might have changed, but a rough estimate is that Africa and South America separated about 100 million years ago (6000 km divided by 6 cm/yr). Contributions to theory: the earth is at least 100 million years old, old enough for significant evolutionary change to have happened. Also, if two related species live in Africa and South America today, ostriches and rheas, say, and their common ancestor lived long enough ago, their ancestors could have walked from one continent to the other without crossing an ocean (but see “The Elephant Bird’s Tale” in Richard Dawkin’s book, The Ancestors Tale for some complications!).

Manifest Facts: changes in coral growth over a day and over a year produce corresponding “growth rings.” Corals alive today have 365 day rings per year ring, but some fossil corals have more day rings per year ring. Astronomers have measured the speed of rotation of the earth and found that it is slowing slightly each year. Inferred facts: days would have been shorter in the past, resulting in more days per year. If we use growth rings to calculate how long ago a certain fossil coral lived, we get the same answer as from radioisotope dating: 380 million years (Nature vol.197, p. 948). The earth is much older than fundamentalists believe. Contribution to theory: radioisotope dating is accurate enough to be useful in dating rocks associated with fossils.

Manifest Facts: modern whales have tiny bones, not connected to anything, about where their rear legs would be, if they had them (visit a natural history museum and see for yourself!). The fossil species Dorudon, dated to 40 million years ago (MYA), resembled whales, but smaller, with recognizable rear legs too small to do much of anything. Fossil species Rodhocetus, from 47 MYA, was even smaller, with rear legs extending backwards like flippers. 50 MYA was Ambulocetus, with legs long and strong enough to walk awkwardly on land or more easily in water, like a hippo. Before that (52 MYA) lived Pakicetus, smaller still and with longer legs. This series of fossils, all from the same area, also differs progressively in the position of the nostrils, with the most recent fossil, Dorudon having them farthest back, a whale-like blow-hole. Inferred facts: whales evolved from smaller mammals that spent a lot of time in water, with much of the change in size and shape occurring over a 10-million year period. Contribution to theory: ten million years (much less than the minimum ages for the earth inferred above) is plenty of time for major evolutionary change. So, for example, we don’t need to doubt molecular and fossil evidence that the common ancestor of humans and chimps lived only about six million years ago.

Manifest facts:
Coyne recommends two evolution blogs, Laelaps, which recently discussed evolution of whales, and This Week in Evolution. He also states that I am a professor at Cornell. I earned my Ph.D. there, in Crop Science, but have been a professor only at UC Davis and here (as an adjunct following my wife) at the University of Minnesota. Inferred fact: Coyne is unusually insightful, but there may be a few minor errors in his book! Contribution to theory: it is difficult to write a full-length book without making a few minor errors. That may be some consolation as I work on Darwinian Agriculture. I missed my April 1 deadline, but Princeton University Press kindly gave me another six months.

March 20, 2009

No butterflies were harmed by this research

With a species using cryptic resemblance [camouflage] for its protection, the very existence of neighbours involves a danger to the individual, since the discovery of one by a predator will be a step in teaching it to recognize the crypsis. With an aposematic [bad-tasting, warning-coloration] species, on the other hand, the existence of neighbours is an asset, since they may well serve to teach an inexperienced predator the warning pattern. -- William Hamilton, 1964
This week's paper describes research that could have been a winning science fair project. "Does colour polymorphism enhance survival of prey populations?", published online by Lena Wennersten and Anders Forsman in Proceedings of the Royal Society, helps answer an interesting evolutionary question, using materials available in many kitchens.

It has been suggested that variability within a species (polymorphism) in color may reduce losses to predators. One reason is that predators may not have enough mental capacity to search for two or more different-looking prey at the same time. According to Robert Gegear, who gave a seminar here recently, this is why bees often focus on one species of flower at a time, passing over species they might visit on other occasions.

If it is true that groups of individuals varying in color are less likely to be eaten (at least completely), that would raise all sorts of interesting questions. Suppose red individuals are more likely to be eaten but distract attention from members of their species that are green? Who wants to volunteer to be red? But this paper is interesting, not for its contributions to theory, but for their method and their results.

They made "artificial pastry prey", cylinders of dough resembling caterpillars, in different colors: red, yellow, brown and green. They placed groups of 12 "prey", perhaps similar to family groups, in woodland trees, distributed over 1-2 square meters, and then recorded how long it took for wild birds to find and eat them. Some groups were all the same color, whereas others had four different colors.

Over the course of the experiment, 79% of the 2976 prey "disappeared or bore evident beak marks." All-green groups survived longest, on average. Mixed (polymorphic) groups were next. Within polymorphic groups, green "prey" survived much longer than the other colors. So there didn't seem to be any individual or group-level advantage to being any color other than green, and no group advantage to having a mixture of colors.

The authors argue against the possibility that groups with only cryptic coloration (brown and green, say as opposed to the red, yellow, brown and green they used) might survive longer than one-color groups chosen from the same range of colors. I would like to see actual data on this, however. If there's a student out there looking for a great science fair project, here's your chance!

March 18, 2009

"I.B.M. Said to Be in Talks to Buy Sun for $7 Billion"

...according to the New York Times. Although private ownership is one way to prevent degradation of the commons, this is going too far, in my opinion.

March 14, 2009

Experimental evolution of an RNA world

How did the first life on Earth arise? We may never know for sure, but can we at least demonstrate one or more mechanisms that could have led to life as we know it? Not yet, but this week’s paper seems like a significant step towards that goal. “Self-sustained replication of an RNA enzyme” was published in Science by Tracey Lincoln and Gerald Joyce.

Most species have protein-based enzymes (running the biochemical reactions needed for growth and reproduction) and DNA-based heredity (passing genetic information to the next generation), with RNA serving various other functions. Under the “RNA-world” hypothesis, however, RNA molecules once served both as enzymes and for heredity. Some viruses use RNA as their hereditary material and some RNA molecules still act as enzymes, with a key role in protein synthesis, for example.

Can we recreate the early RNA world in a laboratory? What is the simplest system that could evolve by natural selection, eventually leading to something that would be universally recognized as alive?

1) The system would need to be able to reproduce using materials available in its environment.
2) The system would need to be open-ended, in the sense that a very large number of variants are possible, not just a few.
3) Copying accuracy would need to be good, but not 100%, so that there is some possibility of evolutionary change.
4) The occasional error would need to be propagated in copies, i.e., in successive generations.
5) Different versions (i.e., with or without a particular error) would have to reproduce at different rates.

An RNA molecule that could copy itself or – this is important – any variant of itself that arose through past copying errors, might meet these requirements. Another option would be two or more RNA molecules that have the collective ability to reproduce. This was achieved a few years ago, using a pair of RNA enzymes that copied each other, but the rate of reproduction was low. There are at least three ways that the rate of reproduction could have been improved: a) let the system run for a few years or millennia and see what evolves, b) intelligent design of faster RNA enzymes, or c) directed evolution. Nobody knows how to design faster RNA enzymes from first principles – maybe we aren’t intelligent enough – but the Joyce lab has lots of expertise in creating conditions under which populations of molecules evolve desired characteristics, in ways analogous to natural selection.

They introduced random changes to the RNA enzymes and selected for faster reaction time, discarding those that failed to react within a specified amount of time. After six cycles of mutation and selection, they analyzed the faster enzymes that resulted. All of them had a “G-U wobble pair.” Usually, G pairs with C and A with U (the RNA equivalent of the DNA “letter” T), but G can pair with U. G-U pairs have previously been found to be important to the activity of some naturally occurring RNA enzymes.

Not being inclined to argue with evolution’s wisdom, Lincoln and Joyce “installed” the G-U pair in the two mutually copying enzymes (and in the RNA pieces the enzymes join to copy each other). That worked great: the evolution-inspired enzyme pair multiplied 25-fold in five hours (faster than many bacteria) before running out of materials. Like bacteria, they resume growth if a drop is transferred to a new tube with fresh materials. So this system clearly meets criterion #1 above, although it’s worth noting that the materials required for reproduction would not necessarily have been available to early RNA reproducers on Earth.

What about the other criteria? The RNA molecules are large enough to meet criterion #2, but can variants also reproduce themselves? They showed that some variants can, at least, making 12 different pairs of RNA enzymes that could work together to copy each other. Criterion #3 and 4 were also met: each enzyme pair usually copied itself, but sometimes one partner joined together the wrong pieces, creating a “recombinant” enzyme. Because the choice of which subunits to join depending on pairing between RNA bases, these recombinants made recombinant partners, which then made accurate copies of the recombinants.

Finally, they met criterion #5, demonstrating the differences in reproductive success among different versions of the enzymes, as required for evolution by natural selection. They started with the 12 pairs of mutually copying RNA enzymes and let this “population” evolve. After repeated transfers to tubes with fresh materials, the population was dominated by three pairs of enzymes. Interestingly, all three were recombinants. One of the original 12 pairs was still fastest if only its own subunits were present as materials to join, but a particular recombinant did best in a more chemically diverse environment.

The materials provided by the experimenters were relatively complex: RNA molecules half the size of the RNA enzymes. There was no attempt in this paper to explain how such complex molecules arose. Instead, they are looking forward to investigating stages of evolution slightly more recent than mere self-replication from ready-made materials, such as evolution of an RNA enzyme that would help lipids for a “cell membrane” to enclose a population of reproducing RNA. I look forward to this next paper. This one closes:

“Ultimately, the system should provide open-ended opportunities for discovering novel function, something that probably has not occurred on Earth since the time of the RNA world but presents and increasingly tangible research opportunity.”

March 5, 2009

What I should have said to Richard Dawkins

Richard Dawkins gave a pretty good talk here last night. I have often thought that much political and religious speech (Jindal trying to talk folksy, for example) sends the underlying message "We X are kin, distant cousins or something maybe; They are not; give me your vote, your money, or your sons' lives and I will defeat Them." But I hadn't heard the term "fictive kin" used to describe this illusory relationship.

I also got to join a large group of students and faculty for lunch with Dawkins. What I should have said was "I'm writing a book called Darwinian Agriculture, heavily influenced by The Selfish Gene." Instead, I thanked him for calling my attention (when we met several years ago) to a paper about human-imposed group selection in chickens leading to increased egg production. A theme of my book is that humans can sometimes impose strong enough selection for group-level performance to overcome individual selection that undermines group performance (e.g., selecting for wheat that puts more resources into grain and less into tall stems to shade out competitors), whereas nature rarely if ever does that. But I think he just heard "group selection" and tuned me out. It must be tiring touring like that.

March 3, 2009

Science Fair Secrets 4: start early, work hard

This is a series (copyright R Ford Denison) on the secrets of winning science fair projects. Click "science fairs" under Categories (at right) for more.

Two high school students who did projects in our lab (Kyra Underbakke and Tiffanie Stone) have now won trips to the International Science Fair. Is there something magic about our lab? Both students had excellent mentoring, from my grad students Will Ratcliff and Ryoko Oono. So that's one Science Fair Secret: look for a smart mentor who's willing to spend some time helping you explore ideas and methods. The evolutionary focus in our lab may have helped the students ask more scientifically interesting questions, but we aren't curing cancer or saving cute endangered species (with the possible exception of organic farmers!).

But the most obvious characteristic these two winners had in common was their willingness to put many (hundreds?) of hours into their projects, starting in spring and working through the summer and winter breaks, spending long hours in the lab when other students were off on vacation.

This contrasts with a student I judged at last year's fair. He had an interesting project, but when I asked about an obvious control he hadn't done, he protested: "but that would have taken five hours!" He didn't win.

Another thing both winners had in common was how they dealt with failure. Neither project gave clear results at first, due to unexpected problems. Unexpected problems are only to be expected, in real scientific research. Rather than giving up, both students worked with their mentors to revise methods and try again. This meant more work, of course, and trying again would have been impossible if they hadn't started their project until a month before the deadline, but the results were worth it.

Starting early and working long hours don't guarantee success, of course. But not starting early and not working long hours will probably guarantee that your project won't be among the best at the science fair.

Mixed infections, for better or worse

If being infected is bad, is being infected by two different pathogens at once even worse? Not necessarily, as this week's paper shows. "Quorum sensing and the social evolution of bacterial virulence" was published in Current Biology by Kendra Rumbaugh and colleagues. Their results contradict an earlier prediction, although not the fundamental evolutionary principle behind that prediction.

The fundamental principle is that a multiply infected host is analogous to grazing land shared by several families, with no overall regulation. Garret Hardin used this example in his 1968 essay, "The Tragedy of the Commons." If the land is owned by only one family, they might limit the number of sheep to what the land can feed sustainably. If ten families share the land, however, each might reason that "one more sheep (ours) won't do much harm, relative to the ten already there, and we'll have another sheep." Together, they add ten more sheep, destroying the grass.

Similarly, if most hosts are infected by only one strain of bacteria, those strains that kill their host to quickly (before spreading to another host) will tend to die out. With mixed infections, there is no benefit to self-restraint, as the other strain may kill the host anyway. Based on this idea, George Williams and Randolph Nesse (in "The Dawn of Darwinian Medicine") predicted that "diseases that result from single infections of a host will be less virulent than those that normally arise from multiple infections from different sources." (The "normally" is what makes this an evolutionary argument. We need not assume that strains act differently when they are part of a mixed infection, but only that their behavior has been shaped by past natural selection. Natural selection is less likely to have favored restraint if most past infections were mixed.)

The prediction that mixed infections will be more severe assumes that "selfish" bacteria cause more damage to the host than those that "cooperate" with other bacteria in that host. This is true if cooperation equals restraint, but there are other kinds of cooperation.
In particular, some bacteria release "virulence factors", molecules that are expensive for an individual bacterial cell to produce but which benefit bacteria collectively (at the expense of the host) by breaking down defenses or releasing nutrients from host tissues. With one strain per host, a strain that doesn’t produce virulence factors may die out. But what if hosts are often infected by two strains at once?

In that case, strains that do not produce virulence factors may outcompete those that do. This is because they can benefit from virulence factors produced by another strain in the same host, without paying the cost of producing virulence factors themselves.

To test this hypothesis, the researchers infected mice with one or two strains of pathogenic bacteria. With one strain at a time, strains that made virulence factors killed the mice faster, as expected. A mixed infection, however, was no more lethal than the no-virulence-factor strain alone. You might expect lethality intermediate between the two strains, but only if you ignore evolution.

The cost savings from not making virulence factors was enough that the low-virulence mutant increased in frequency (e.g., from 1.3% initially to 32.4%) at the expense of the strain making virulence factor. The “cheater” that didn’t make virulence factor had higher fitness when rare, so increased in frequency. Once they became a larger fraction of the population, their relative fitness benefit decreased, because there were fewer bacteria making the virulence factor they need. This sort of frequency-dependent selection is common and not surprising, but it suggests that the 50:50 mixture probably didn’t evolve to 90% low-virulence cheaters. So maybe the fact that the 50:50 mixture was no more lethal than the low-virulence strain alone is surprising after all. Does making half as much virulence factor result in much less than half the virulence?