Darwinian Time

n a windowless room, three researchers hunker over a waist-high lab table. Dressed in white coats and latex gloves, the investigators, all members of the Washington University School of Medicine in St. Louis, get down to the business at hand: skinning frozen mice.

Each researcher pulls a curled-up, frozen mouse out of a tiny plastic bag. Ann Carson, a postdoc and one of the three investigators, gives the others some brief instruction. First, thaw the frozen mouse in your hand for a couple of seconds, she tells them. "Then use your scalpel and cut from snout to tail." Like peeling oranges, the three remove the fur from the small white, brown, or black creatures. All the mice organs have been removed already through incisions on the underbelly. Once skinned, the mice become lunch.

The researchers have to remove the rodents' fur so that a colony of hundreds of beetles in plastic storage bins under a nearby table can eat away the mouse flesh, leaving behind tiny bones. And beetles don't eat fur.

The researchers, part of James Cheverud's lab team, will then use a computed tomography (CT) scanner on the slight skeletons as part of a project designed to determine how mutations in individual genes affect specific traits, such as forelimb length or skull width. By mapping the gene regions that code for skeletal dimensions, Cheverud (an anthropologist by training) compares minute, individual differences in bone lengths to the individuals' respective genotypes, correlating any mutations to changes in the phenotypes. This approach illustrates the effect that single mutations can have on traits.

The research, now 15 years in the making, comprises one arm of an ongoing investigation into one of the central questions of evolution, still unanswered two centuries after Charles Darwin's birth: If mutation is one of the driving mechanisms behind evolution, how does a single mutation, whether it alters just one or many traits, control how quickly organisms are able to adapt to their environments by adopting the fittest traits? And once an organism has adopted these traits, becoming more complex, do the traits place constraints on further evolution? Is complexity itself a deterrent to evolutionary plasticity?

Slowly, Cheverud's small team fills sample cup after cup with skinned mice, which will be left to dry out for about a week before they're fed to the beetles. All in all, members of Cheverud's lab have skinned between 4,000 and 5,000 mice over the past several years.

February 12th is the 200th anniversary of Darwin's birth. And 127 years after his death researchers are still grappling with some details of the theory of evolution. Darwin only briefly addressed how quickly evolution might occur. In The Origin of Species, he wrote: "Natural selection can act only by taking advantage of slight successive variations; she can never take a leap, but must advance by the shortest and slowest steps." In other words, he believed that organisms gradually adapt to their environments via minute physical adjustments. (Biologists have since found evidence that dramatic adaptations can also occur, for instance in the form of major morphological changes.)

In 2000, University of Rochester researcher Allen Orr followed up on evolutionary biologist Ronald Fisher's theory of complexity to figure out the relative speed with which organisms of varying complexity might be able to adapt and evolve. Complexity correlates with the number of specially-adapted traits—as organisms become more and more complex, they have more such traits. Orr posited that any given mutation could affect any trait in an organism. Therefore, in a complex organism with many specialized traits, any mutation has more chance of disrupting the system than of improving it. By analogy, changing the length of an arbitrary component would harm the function of a hammer drastically less than a microscope. Orr's model therefore predicted that complex organisms evolve more slowly than simpler organisms, a phenomenon he called the "cost of complexity."¹

Of course, estimating the speed of evolution even in simple organisms is anything but simple. Viruses can evolve in a matter of generations, but, as always, the strength of the selective pressure in an environment affects how fast organisms adapt, so individuals of the same species in different environments will evolve at different rates.

The key assumption in Orr's (and Fisher's) analysis is that a single mutation can potentially alter any and all traits in an organism, including the advantageous ones, changes which likely won't become fixed in the population. This concept—that a mutation can affect most traits—is referred to as "universal pleiotropy." But it remains unclear how extensive pleiotropy really is. If mutations affect just a small percentage of traits, rather than the majority, then complexity would act as less of a speed bump for evolutionary change than Orr and Fisher propose. "We need to collect data on how common pleiotropy is, and what percent of characters in an organism are affected by random mutation," says Orr. "We need more of that."

In 2004, Cheverud's lab, one of the few groups collecting empirical evidence related to the speed of evolution, teamed up with evolutionary biologist Günter Wagner, at Yale University, to get a handle on exactly how many traits pleiotropy affects. The answer, they hoped, would help determine whether complexity serves as a brake on the speed at which evolution can occur.

At Wagner's suggestion, Jane Kenney-Hunt, a postdoc in Cheverud's lab, began an enormous study to correlate the effect of mutations on physiological traits in mice. She started with two already-established mouse strains that had been inbred for 100 generations for either small or large body size. Then she let the two strains inter-mate for two generations and, once the animals were necropsied and skinned, began measuring their bones. When all was said and done, Kenney-Hunt evaluated 1,000 mice for 70 different traits, including the size of individual mouse vertebrae, measuring each trait three times (a total of 210,000 measurements).

She and other lab members then examined each mouse genome, looking at molecular markers flanking gene regions—called Quantitative Trait Loci, or QTL—for any changes in genes that were statistically associated with variations in phenotype. (See figure)

The group found 102 QTLs that were correlated with specific phenotypic changes in bone size. On average, QTLs affected about seven traits out of the 70 traits they were looking at. And 50% of QTLs affected fewer than 10% of the 70 measured characters. (The remaining QTLs affected more traits, some up to 30.) "The surprise," says Cheverud, "is the fact that a given QTL affected an average of only seven traits;" they were expecting mutations to affect larger groups of at least 15 or 20 at a time. "The findings mean that the basic assumption, through the whole 20th century, that mutations are highly pleiotropic and affect many characters at the same time, is just not true," says Wagner. The resulting paper was published in Nature last March.² Without extensive pleiotropy, a mutation will only affect certain traits, making it less likely to derail the overall fitness of the complex organism that is fine-tuned to its environment.

But this paper is by no means the last word. Using QTL studies, researchers can attribute most traits to a broad region of the genome. But these regions can contain hundreds of genes, making it impossible to tease out the individual effects of individual mutations. Sequencing could provide a picture of what's happening at the gene level, but with a mouse experiment on a scale as large as Cheverud's, the cost and difficulty of such a project would likely be enormous, says Charles Baer, evolutionary geneticist at the University of Florida.

In response, Cheverud's team has continued breeding successive generations of mice used in the Nature paper. By allowing the mice to breed for more than 35 generations—some 10,000 animals when all is said and done—the DNA has been spliced and recombined over and over. With such slicing and dicing, the chance increases that genes which originally got transmitted together in one region are, in later generations, split up and transmitted separately. And indeed, that's what's happened: The team is now able to map traits down to a region about 1 megabase long, containing just five to 10 genes. Although they haven't finished their analysis, preliminary results are indicating that the number of traits affected by a single mutation is even smaller than was published in the Nature paper, in some cases by half.

The results in the original Nature paper are certainly suggestive, says Orr, but not conclusive in light of other empirical studies. Findings in Drosophila from the early 1990's show that two-thirds of all genes are involved in correct development and neural connectivity of the eye.³ So mutations, even in indirectly related genes, can affect a broad range of traits. "It is clear that there's a lot of pleiotropy out there; the open question is how many characters are affected," says Orr. He says he still believes pleiotropy acts as a speed limit on evolution in some capacity, but whether it's a strong limit, or weak one, is still unanswered.

And can Cheverud's experiments really grasp the full effect of mutations on traits, as one would see in nature? For an organism to evolve, a mutation must become fixed in a population and that will only happen if it confers upon the organism some advantage—something that scientists can't measure in a lab. Plus, "what we call a mutation in the lab tends not to be subtle," says Leonid Kruglyak, evolutionary biologist at Princeton University, "versus what evolution sees—it could easily be that the difference in effects of a mutation on something important for survival could be on order of 1% or one-tenth of a percent" in nature—and how can we possibly account for that, even with 35 generations of lab mice?

Even if researchers achieve optimal lab conditions, they likely won't find that pleiotropy is either universal or minimal. The view that genes relate to specific characters and don't affect others is "largely defunct," says University of California, Irvine, evolutionary biologist Michael Rose. Alternatively, some biologists see the organism as totally interconnected, creating strong limits on what can change. "But neither view is true," says Rose. "Pleiotropy is extensive, but not paralyzing." In Rose's mind, evolution is neither infinitely nor negligibly flexible, creating an intermediate state where the ability of an organism, whether simple or complex, to evolve is kept in moderate check by the effect of mutations.

On a brisk day in September at Yale University, Brendan Ogbungafor, a grad student in Paul Turner's lab, fills a dozen mini-vials with RNA virus phi6. He walks across the high-ceilinged lab room to a PCR machine and inserts the vials, cranking the temperature on the machine down to just above freezing, well below the virus' comfort levels, since they are normally cultivated at 25°C. After a few minutes in the PCR he pulls out the vials, walks them back to his lab bench and begins to pour the contents onto an agar plate seeded with bacterial cells. Once the plates have been left to rest for a few days, Ogbungafor will count how many viruses have survived by counting the number of plaques on the plate, and add these data to observations that Turner's group has been making on how well viruses survive extreme temperatures. Eventually, they hope to gather enough information to support their theory that certain genotypes are able to adapt quicker to new environments than other genotypes of the same species.

In 2005, Turner's group allowed single viruses to infect a cell, multiply, and release progeny. In another treatment they let multiple viruses of the same genetic lineage simultaneously infect a single cell. The researchers let both populations replicate for 100 generations, hypothesizing that viruses that infected cells together had an advantage, because if one virus starts coding lousy proteins, it can capitalize on good proteins brought in by coinfecting viruses.

However, when the researchers examined the clones, they saw that viruses that infected a single cell together were less fit than viruses that infected cells one-by-one; specifically, harmful mutations had a more severe effect on viruses that co-infected cells.

To see if certain robust lineages (meaning, viruses that were able to survive deleterious mutations) were better at adapting to environmental changes than others, Turner's group started changing the environmental conditions that the viruses were used to—enter the extreme temperature experiments. In 2007, his group showed that certain individual RNA viruses (about 10% of a population) could survive heat shock treatments. The survivors and clones of non-survivors were bred to about 100 generations, and raised in normal conditions. Then the scientists heat-shocked both populations again. Viruses that had survivor ancestors, called the robust strain, survived at a rate of 80%, whereas less than 10% from the non-surviving strain made it.⁴

These findings suggest that questions of how fast a particular species, whether complex or not, evolves will be complicated by the fact that, within species, some individuals will have an easier time adapting to their changing environment. So, among Cheverud's mouse strains there may be a genetic lineage better suited for making the most of mutations that pop up, and more quickly adapting to their environments—a factor that cannot be accounted for in his experiments. "The ability to adapt to a new environment can differ among genotypes and among species," Turner says. "Evolution itself has the ability to evolve."

Have a comment? E-mail us at mail@the-scientist.com

As a molecular biophysicist always thinking about protein structure and protein folding, Eugene Shakhnovich at Harvard University was frustrated that researchers studied the effect of evolution on protein structure, but little about the effect that protein structure had on evolution. There must be a link, he reasoned, between the stability of proteins and the mutations (a driving force of evolution) that can alter protein structure and function. For instance, can the stability of proteins play a role in evolutionary speed?

First, Shakhnovich considered the impact that mutated proteins have on organisms. "As mutations accumulate, these mutations impact the stability of proteins." You get to a point where enough mutations become lethal, because too many proteins are not folding correctly. But is there a fixed limit on how much change a protein can withstand?

In a series of mathematical models, Shakhnovich determined that more than six mutations per essential part of the genome (i.e., the percentage of the entire genome that codes for essential proteins whose failure leads to the organism's death) per genome replication, and any given mesophilic organism cannot survive (PNAS, 104:16152-7, 2007). "Molecular evolution cannot exceed this rate—if it goes higher then the organism loses all protein stability," he says. Since mutations underlie novel traits that get selected for or against by natural selection, the mutation rate plays directly into the pace of evolution. Is this a speed limit to evolution? And might this act as an alternate "brake" to the speed of evolution, besides the theoretical cost of complexity?

It turns out that RNA viruses, which are known to rapidly mutate, operate quite near this evolutionary speed limit, at about 5.5 mutations per essential part of the genome per genomic replication cycle. Complex organisms, though, operate at a mutation rate 1000-times under the speed limit. Why? Especially when mutations can be advantageous and produce traits that infer fitness?

In further modeling studies and empirical experiments underway with bacteria, Shakhnovich is seeing that, as organisms become well adapted to their environments, individuals with high rates of mutations are selected against. In addition, proteins that get selected for show a higher rate of improving from beneficial mutations, not deleterious ones, suggesting that evolution is selecting for robust proteins. In more complex organisms "we see a lot of complexity in the way proteins interact with each other - proteins that are functional. So organisms keep their evolution rates because they've evolved a solution. It's crazy to ruin it," says Shakhnovich.

As mentioned in the article, laboratory trials cannot account for the relative fitness of new traits. I think lab trials mask an even bigger issue: the "plastic" or paragenetic contribution to evolution made by environmental change. Pleiotropic effects produced in the lab needn't mirror those produced in the wild, where predation pressure, changing of the seasons, etc., all stress populations and individuals in ways that a caged existence can't. Where we see one gene impacting maybe two characters, we may misread a result that depends on stress hormones, dietary factors, and the proximity of other individuals to produce "normal" hormonal responses, which in turn, synergize or mask the effects of mutant genes.

By focusing just on mutations, researchers will eventually lose sight of the near-real-time interplay of gene and environment. I can conceive of experiments in which mutations are produced at-will in the lab, and run for many generations...while the same mutations, produced in the wild, lead to phenotypes so different as to go unrecognized in cause.

looks no further than the United States of America!

I have to say that this all sounds interesting, but it might be worse than misleading for a number of reasons.
First is that when I mention the principles of traits to people, I always get told about crossovers and complexities that expand and complicate the connections between traits with the suggestion that there is too much complexity to reduce or delimit the domain of a trait. If that were true, natural selection couldn't propel evolution. The domain of a trait generally must be fairly limited or it would be too hard for natural selection to differentiate and effect it.
Another point is that this is going to be a variable characteristic through the genome, determined by evolutionary history. For some genes, variability is good, such as genes that interact with the outside environment. Some genes change much slower because their purpose changes slower, ie. metabolic processes vs morphology. The same thing would apply in the case of inter-relatedness of how a gene in a trait gets expressed. The slower it has needed to evolve, the less granular the parts of the trait would be. The opposite would be true as well. It is going to be dictated by evolution.
Another reason to look at this a bit sideways is that this would also relate to the additive principle of evolution. Traits at the bottom of the stack so to speak, would have broader effects. Traits at the top of the stack, ie. the neo-cortex, would have a more localized effect.
I study an integral gene concept that shows the natural domain or object nature of a trait. It obviates a lot of questions I see in the literature.

As a great admirer of Charles Darwin, I am dismayed to see your cover depicting him as a somewhat crazed old man wearing a birthday hat. I find it demeaning.

Thanks first of all for this most interesting and informative article.
I think one needs to mention that a new field, called systems biology, has evolved that concerns itself with exactly this definition of the degree of interaction between separate components in complex systems. The mathematical models should predict how the whole will respond if one increases or decreases the amount, or activity, of one component (e.g. also possibly as the result of mutation). One of the lessons learned in these type of studies is that there is modularity. There are groups of proteins that interact in groups to carry out a specific function. The effect this module has on the whole system can be dealt with essentially as a single variable. This allows a large part of the machinery required by a living organism to be put in boxes where their individual components are acting locally and with limited pleiotropy. Additionally, in view of high functional redundancy, including the discovery of variable gene copy number, the effect of most single mutations will be negligible, since they will be masked, e.g., for three fifths of all genes in say yeast. It seems that the limitations that would result in the case of universal pleiotropy in ever more complex systems has been offset by functional redundancy (this includes the diploid or polyploid state and consequently also sexuality)and by modularity.

To two other comments:
1. The thought game of "who was the first human" is great! Any students who can answer this on one page deserve a Distinction!

2. Darwin would be more than dismayed to see his name associated with what is known as "Social Darwinism". It has nothing to do with Darwin. Darwin always pointed out that man has his intellect to guide a humanitarian mode of action in relationship to the whole living kingdom and his fellow-men and -women. We do not have an unlimited time to take our role as "husband" of the natural World seriously. We would be well advised to use all our energy to gain enough knowledge to act wisely.

I am much in admiration of Charles Darwin for his contributions to where science was in his day, and for those who have updated and upgraded those contributions admirably so far.

Just wishing to let you know that there are some of us out here who do not believe the last datum has been attained... not yet. And wishing to share with you that the surest trap on the path to learning is bigotry, no matter which side take the bait. If there be a clash between one side of an issue, let neither side claim a monopoly on the right to interpret what is yet unknown to anyone. Any extrapolation from the known to the yet unknown is speculative, even if informed speculation, until the pool of things unknown to science and philosophy is exhausted.

Retired and happy, I find nothing more enjoyable than reading polemical claims on first the one side of the interpretation of Darwin's data, and subsequent data. But I discern that the contest is not over, and hence neither opponent's assertions as to the ultimate meaning of any evidence is resolved.

The enemy of man's learning, as I perceive him to be, is not who is right about the yet unknown, but bigotry in the interim.

Accordingly, I embrace all views that are not bigoted, and trust that the future shall take matters into its own hands, in the end.

Compare my enjoyment to that of watching two valued and admired friends playing a hard fought game of tennis. I relish the best efforts of each, and am amused by the attempts of each to insist upon the other a sole right to determine which of the other's shots goes afoul as seen from that one's end of the court.

Hooray for Darwin. And Hooray for any who would attempt to return neo-Darsinist's best shots, to their disaproval. How amusing it is to me, when one complains that the other has no right to declare the outcome before further learning comes to pass. That, it seems to me, is what happens when each seeks to set himself up as both player and umpire, and assert that the other has no right to make the same claim.

We shall see what we shall see. And, if this old f--t must get up and leave before the final score, it will be with a smile, and will not impact the outcome one iota.

Allow me to sign this as,

A happy spectator.

If the environment is relatively static, attributes that allow the organism to out-compete should slowly become more prevalent. For instance the Musk Ox, being heavily-furred and short limbed are both strong adaptations to an arctic environment and genetic changes that allow better heat conservation will improve that traits likelihood to be passed on. It would seem this scenario, would be modeled well by the issues delineated by the article. Organisms with greater genetic information will change more slowly due to the greater potential for any particular mutation to have a negative effect on the organism (our microscope organism). The less complicated organism (our hammer organism) will evolve more rapidly.

When we have large environmental perturbations (Turner?s temperature experiment), then the diversity within the existing population would seem to be the controlling factor; in a natural population the genetic variation to tolerate the heat-shock would have allowed the organism to continue (with my two examples, I seem to have some potential analogy occurring regarding global warming, which as it may, is completely happenstancial, at least consciously). Post cataclysm, the genetic distribution is going to be massively moved to favor the tolerant variety. Whether my example of a microscope organism (Musk Ox, which is interestingly very much not microscopic) has the genetic diversity to survive a large environmental event is obviously open to speculation. However, as in Turner?s example of the bacteria, without the particular genetic strain, it would have not managed its environmental catastrophe.

I do not intend to be arguing that organisms that are more complex per se, can handle rapid environmental events better than less complex, but whether more overall genetic diversity does? And does a more complex organism have the potential, within the species, have more overall genetic diversity thereby giving it an increased potential of survival.

So, in thinking about the analogy of the hammer organism and the microscope organism, changing an internal length of a microscope dimension may be more deleterious than to a hammer, but given the relative simplicity of the hammer organism versus the microscope organism, there are a lot more ways to build a microscope than a hammer. What strikes me is greater potential for genetic variability an advantage for a more complex organism that tends to offset the negative issues as outlined in the article? Of course, any potential advantage of the more complex organism having greater genetic variability must be taken advantage (I see no reason to assume just because an organism may have the potential for a wide range of variation, that it necessarily does). As in Turner?s experiment with the temperature shock, if the bacteria did not have the heat tolerant gene(s) within the population, this would have been an extinction in microcosm. Would a more complex organism, with more potential genetic variability within the population have a greater potential for survival?


Greg

My son (7 years) had been observing that everybody have parents. From there, He asked me how the first human was born and I found myself unable to answer him. It is suppose that at one moment in the past, a female gave birth to a progenie with characteristics somehow more human or with characteristics that make this organismo able to start evolving into a human but this explanation still does not make sense to me. Of course, it was not good enough for my son. I was tempted to use the explanation based on god but I am scientist. I read this article and I can see that we are still far from elucidating the driving force of evolution. About complexity, we must remember that even a yeast is already a complex organism sharing most of our metabolic pathways with humans. There must be something else besides complexity controlling evolution.

The Darwinian view on origin and evolution of life is fundamentally wrong because life more likely originated from different abiotic forms and thus cannot be rooted in any single ancester cell. For more detailed information, please read "A Fundamentally New Perspective on the Origin and Evolution of Life" (LINK or LINK and "Evolution: An Integrated Theory − Criticisms on Darwinism − Fifteen Years Ago"(LINK
Shi V. Liu (SVL@logibio.com, LINK

Defining complexity is a tricky business, but I don't think accumulation of adaptive traits per se is a good metric for gauging increases in complexity. Even dim-witted ninnies like Behe have noted that breaking the lock on your front door will keep out a thief who has a key, but that "adaptive trait" or mutation in your house/door would not constitute an increase in complexity (a gain in functional information if you will), but rather, a loss. Evolution is very good for explaining adaptation and variation, but labeling this as "complexity", I think, is possibly naive.

It seems to be a lot clearer than the biological version.

While I see the value in this study, I'm not sure just how much can be truly determined about the extent of pleiotrophy by focusing on 70 skeletal traits. Considering an organism as complex as a mouse, I would not be surprised if 70 skeletal traits is just a sliver of what phenotypes could be affected. Perhaps traits that are not easily measured, or even traits that are not obviously "skeletal" in nature, are just as likely to be affected, or even more likely.