« June 2007 | Main | August 2007 »

July 30, 2007

Didn't mean to be unKIND

Reminder: generic comments on evolution not tied to a particular post, unsupported assertions, tirades, philosophical or religious discussions, etc. are welcome in the comments section of the Troll Refuge but not elsewhere. Repeat offenders will be banned. If anyone feels like arguing with a creationist, who claims that evolution can't create new KINDS -- is this an acronym, or is he just shouting? -- I just moved his comments there, along with my response.

Also in the Troll Refuge, Hermione Granger, founder of Save the Trolls, weighs in on the faith vs. skepticism debate.

July 28, 2007

Darwinian Agriculture III

Next week I will be meeting with a publisher to talk about the possibility of writing a book on Darwinian Agriculture to be published in 2009, the 150th anniversary of The Origin of Species. (I apologize to one reader who apparently thought it was a done deal.) Here's a short draft of the first chapter, mostly about sustainable agriculture by ants and termites.

Farmers of 50,000 millennia

“We’ve been farming sustainably for three years?, read the email. I was glad to learn that my friend was farming in ways that he hoped could continue indefinitely, but how could he be sure, after only three years?

It might have been a reasonable assumption, if the farming methods he used were similar to those that other farmers have used successfully for a long time. But how similar is similar enough? And what qualifies as “a long time?? As director of “the world’s youngest 100-year experiment?, I often thought about these questions....

UC Davis’s Long-Term Research on Agricultural Systems (LTRAS) project was comparing the sustainability of ten different farming systems. Because some important soil properties change over decades, rather than years, the experiment was designed to run for 100 years. There was some initial opposition to LTRAS, from people with strong opinions about sustainability, but things calmed down after a few years. Perhaps, like the philosophers picketing Deep Thought (Adams 1979), they realized that LTRAS would stimulate interest in their opinions, whereas The Ultimate Answer might not be available until after they were dead.

When I received the email, LTRAS had been running for nine years and it was still far from clear which of our systems were most sustainable. For example, wheat grain yields (kg/hectare, roughly equal to pounds/acre) were usually similar, whether the wheat received nitrogen as synthetic fertilizer, or from decomposition of vetch disked under in alternate years (Denison et al. 2004). The vetch obtained much of its nitrogen from the atmosphere, thanks to symbiotic rhizobium bacteria in its root nodules. (I will discuss how conflicts of interest have shaped the evolution of rhizobia in a later chapter.) The good yields of wheat rotated with vetch seemed to show that the vetch was meeting 100% of the nitrogen needs of the wheat.

However, results from unfertilized wheat without vetch showed that this might not be true. Even with no external source of nitrogen, we were able to remove large quantities of nitrogen in wheat grain harvested from these unfertilized control plots: more than half as much nitrogen (kgN/hectare) as in the fertilizer or vetch treatments. This nitrogen presumably came mostly from soil organic matter, about 1% of the soil by weight, which initially contained several times the amount of nitrogen applied as fertilizer each year. This soil nitrogen would have been available to wheat in the vetch/wheat rotation also, so the vetch really only needed to make up the difference: less than half of the total. The sustainability of the vetch/wheat rotation will not be known until the soil organic matter is depleted enough that it is no longer a significant source of nitrogen, as indicated by paltry yields in the control plots.

The world’s longest-running experiment reached that point many years ago. When I visited the Rothamsted Experiment Station, in 1993, they were celebrating the 150th birthday of their Broadbalk Wheat experiment. By then, their unfertilized wheat plots were dominated, not by wheat, but by nitrogen-fixing weeds like vetch. (Agricultural scientists define a weed as a plant in the wrong place at the wrong time. Vetch growing with wheat, rather than before it, will compete with wheat for light and water, and even for nitrogen. “Volunteer? wheat growing as a weed in vetch at LTRAS was actually more nitrogen-deficient than unfertilized wheat growing alone.) In Broadbalk plots where weeds were killed, using herbicides, the wheat plants were stunted and produced very little grain. Wheat that had been fertilized for 150 years with either organic manure or inorganic fertilizers had much higher grain yields than the unfertilized control. Both also had higher yield than when the experiment began in 1843 (Johnston 1994).

A simple definition of “sustainable? is “not getting worse over time.? By this definition, few of today’s human interactions with the natural world are sustainable. Many ocean fish species are being harvested faster than they can reproduce. So are many forests. The natural replenishment rate for oil, coal, metal ores is even lower than for fish or forests; depletion of phosphorus ores is of particular concern for agriculture. Unless balanced by decreasing environmental impact per person, which must have limits, human population growth is unsustainable. Annual US federal government spending of $1000 more, per American, than is collected in taxes, may help politicians get reelected, but it is also clearly unsustainable.

What about agriculture? Given that wheat yields at Broadbalk increased over decades, with only inorganic fertilizers and continuous wheat monoculture (growing only one crop at a time in a given field), can we conclude that these practices are sustainable?

Maybe not. The Agdell experiment, also at Rothamsted, was eventually abandoned, because nitrogen fertilizers gradually acidified the soil, leading to serious root disease in turnips, one of four crops grown in rotation there (Johnston 1994). The first yield decreases from this problem were detected only after 40 years. Although nothing similar has been seen in the Broadbalk wheat plots, yet, could there be slow trends underway that will eventually cause serious problems? 150 years is a long time for an experiment, but it is only two human life-spans. If we are concerned about long-term sustainability, we may need data spanning centuries or more.

No formal experiment has lasted longer than those at Rothamsted, but some fields around the world have been farmed for 1000 years or more. Although none of these farms used synthetic pesticides or transgenic crops, prior to the 20th century, can we at least determine the sustainability of past agricultural practices, many of which would meet today’s “organic farming? guidelines?

Archaeologists have done so, in a few cases. Crops have been grown in Mesopotamia (now Iraq) for about 6000 years. 5500 years ago, wheat and barley seem to have been equally important, based on counting impressions made by their seeds in pottery. By about 4000 years ago, however, only barley was being grown. Ancient documents show that yields decreased greatly over the same period (Jacobsen and Adams 1958).

The most likely explanation for these trends is the accumulation of salt from irrigation water. All water applied to fields contains some salt. When the water evaporates, from the soil or through crop leaves (transpiration), the salt remains. Unless the salt is removed, through natural or artificial drainage systems, it will build up in the soil. Some farm fields in California, irrigated for less than 100 years, are already salty enough that only salt-tolerant crops, like barley, can be grown.

A trend of decreasing yields over centuries, as in Iraq, would be hard to detect without written records and a population that could read and believe them. Long-term trends can be obscured by year-to-year variability in weather. Stories of higher yields in past generations might be considered myth rather than fact. Even if a downward yield trend were recognized, causes and solutions might not be obvious. Are there some witches that need to be killed? Would sacrifices to a different god help? Did the king suppress the findings of Mesopotamian scientists? Consider how long it has taken some politicians today to recognize human impacts on climate!

LTRAS and other long-term experiments were designed, in part, as early-warning systems to detect subtle but dangerous long-term trends. Increased soil acidity at Agdell experiment could have been detected long before yields started to decline, if modern methods for measuring acidity had been available in 1880! Archival samples have allowed analysis of such trends after the fact, by methods not invented when the samples were collected. For example, the polymerase chain reaction (PCR) was used to measure minute amounts of fungal DNA from archival wheat samples from Rothamsted, showing how the abundance of different pathogens varied over decades (Bearchell et al. 2005). Archival samples collected at LTRAS should be equally valuable. We already know how to measure salinity, however. Will organic and conventional systems, for example, differ in salt accumulation over decades?

Yield declines like those seen in Iraq are probably not inevitable. The agronomist Franklin King studied farm fields in Asia, where rice and other crops have been grown successfully for up to 4000 years, and described many of their practices in his book, “Farmers of Forty Centuries? (King 1911). This is an impressive span of time, relative to LTRAS or even Rothamsted. The real experts on sustainability, however, are the fungus-gardening ants, which have been refining their agricultural practices for 50 million years (Aanen and Boomsma 2006).

50 million years is a long time, even for those evolutionary biologists who study the history of life. (Other scientists focus on evolutionary changes over much shorter periods of time: sometimes less than one day, for bacteria with generation times of less than an hour.) Humans only invented agriculture about 10 thousand years ago, not long after the first humans arrived in the Americas. About 3 million years ago (MYA), the last series of ice ages began and the isthmus of Panama closed, blocking an important connection between the Atlantic and Pacific. The common ancestor of humans and chimps lived around 6 MYA and the common ancestor of humans, other apes, and monkeys 40 MYA. 50 million years ago, the Earth would have looked quite different from today, even from space. The South Pole was ice free, South America and Africa were as close to each other as to North America and Eurasia, and India was about to collide with Asia, leading to the rise of the Himalayas (Freeman and Herron 1998).

But the antiquity of ant agriculture is not the main reason that it deserves our attention. Not all “ancient wisdom? is equally reliable, as will be explained in later chapters. The important thing is that ant agriculture has been continuously tested and improved over the millennia, by natural selection.

Ant colonies compete against neighboring ant colonies, at least occasionally. Over their 50 million year history, colonies that harvested leaves (to feed their fungi) mostly in the morning must sometimes have competed against colonies that harvested all day. Whether competition involved direct aggression or not, a leaf harvested by one colony was not available to the other. Colonies that “weeded? their fungus gardens more must sometimes have competed, for leaves or other resources, against those that weeded less, and so on. These behaviors were influenced by genes that worker ants inherited from their mother, the queen. As the queens of winning colonies produced more female workers and male drones, the winning genes became more common. Therefore, we can assume, even without understanding the details, that evolution has improved the agricultural practices of ants, by the criteria used by natural selection. The relevance of these criteria to human agriculture is the theme of this book.

What are the agricultural practices of ants, maintained by natural selection?

First, ants practice monoculture. Each ant colony grows only a single, genetically uniform fungal clone, although different colonies may grow different fungi (Mueller et al. 2005). Termites, which apparently evolved fungus-gardening more recently than ants – less than 40 million years ago -- also practice monoculture, with less between-colony diversity. Fungus-gardening ants are found only in the Americas, fungus-gardening termites in Africa. Ants and termites use different fungal species. So they appear to have invented agriculture independently, yet monoculture was apparently favored by natural selection in both cases.

Second, ants use confined feeding, analogous to industrial animal production in the US. Fungi are actually more closely related to animals than to plants. That is, fungi and animals share a more recent common ancestor than either shares with plants. Unlike plants, neither can photosynthesize, so they need to digest other living, or once-living things. Industrial chicken farmers keep hens indoors and bring them grain (seeds). Similarly, ants bring food (mostly leaves) to fungi kept in underground chambers.

Third, ants use toxins to control pests. The most serious pest in ant fungus gardens – maybe I should call them “feedlots? – is another fungus, Escovopsis, which greatly reduces the growth of the fungi the ants use for food. Ants control Escovopsis using toxins (antibiotics) produced by certain bacteria. These bacteria are housed within special structures (“crypts?) on the bodies of the ants themselves and fed by excretions from specialized ant glands (Currie et al. 2006). This is analogous to spraying crops with the BT toxin produced by the bacterium, Bacillus thuringensis, raised in vats. It is less similar to traditional “biological control? methods, where a predator or parasitoid species feeds on, and coevolves with, the pest it controls.

These practices all seem more similar to industrial agriculture than to traditional or organic practices. Before blindly copying the agricultural methods of ants and termites, however, can we first understand why these approaches were favored by natural selection? Some of the reasons may not be relevant to human agriculture. To see why, we need to explore when natural selection does, and does not, further human interests. That is the topic of the next chapter.

Begging: the question

My wife and I have a bird feeder outside our kitchen window. Yesterday I saw an adult male cardinal feeding some of the seed to an immature cardinal not much smaller than he was. I guess it's hard to say "no", but should he have?

This week's paper, "The adaptive value of parental responsiveness to nestling begging" by Uri Gordzinski and Arnon Lotem, published online in Proceedings of the Royal Society, may have answered this question.

Their experiments used pairs of sparrow siblings raised by humans. They simulated either a "responsive parent" (giving more food to whichever of the two birds "begged" more), or a "nonresponsive parent" (picking which bird got more food at random).

The weight gain of the baby birds (averaged over the two siblings) was 2.8 g with the simulated responsive parent, but only 2.2 g with nonresponsive (random) feeding. The difference appeared to be due to wasted food: with random feeding, twice as much food that was supplied was not eaten, presumably because it was given to a bird who was not begging because not hungry. The authors suggest that responding to begging also reduces the chance that a nestling will miss being fed often enough to affect its growth, which could happen with random feeding.

So kin selection would favor responsive rather than nonresponsive behavior. Genes for responding to begging are common because baby birds are more likely to survive if their parents respond to begging -- and the babies have a 50:50 chance of inheriting the same genes.

The same issue has an article looking at the effects of grandparents on the reproductive success of their children. In contrast to grandmothers, grandfathers that live longer don't end up with more grandchildren, at least in Finland. This seems like something that might vary among societies. But what if your grandfather is a cardinal?

July 23, 2007

Diversity, stability, productivity, and policing

This week I will discuss two papers, both of which consider possible benefits of biological diversity. In interpreting the data in the experimental paper, on bees, we need to remember that a given set of data can often be consistent with two or more different hypotheses. This point is reinforced in the review article, which discusses the relationship between diversity and stability of ecosystems.

The experimental paper is "Genetic diversity in honey bee colonies enhances productivity and fitness" by Heather Mattila and Thomas Seeley, of Cornell University (Science 317:362). The review article is "Stability and diversity of ecosystems" by Anthony Ives and Stephen Carpenter, of the University of Wisconsin (Science 317:58).

Any hypothesis worthy of the name makes predictions. Testing these predictions may take a long time or lots of money. Edmond Halley's 1716 prediction that a transit of Venus could be used to measure the distance to the sun could not be tested until the next transits of Venus in 1761 and 1769, and required global scientific expeditions. (Mark your calendars for 6 June 2012!) The Rothamsted Experiment Station has been testing the hypothesis that wheat can be grown with only inorganic inputs and without rotating to other crops, since 1843. This may not be long enough to uncover all possible problems with inorganic fertilizers and continuous monoculture, but it's quite a contrast with an acquaintance who wrote that "I've been farming sustainably for three years."

If even one of a hypothesis's predictions turns out to be unambiguously wrong, the hypothesis must be discarded or revised. On the other hand, multiple correct predictions do not prove that a hypothesis is true -- there might be other hypotheses that make the same predictions. Either way, it's useful to consider several hypotheses, if you can.

Tom Kinraide and I discussed these points in an article in American Biology Teacher in 2003. His boss wouldn't let him include his USDA affiliation, because someone at the lab complained that testing hypotheses would undermine his religion. I don't know; maybe it would. But back to this week's papers...

Mattila and Seeley impregnated honey bee queens with the same total amount of sperm from either one or 15 male bees. The latter led to more genetic diversity among worker bees, as in most wild colonies. More diverse colonies usually foraged more actively and stored 39% more food. They also produced more female worker bees and more male drones.

What hypotheses can explain these results? Bees apparently vary genetically in the amount of stimulus they require to start foraging or other tasks. The authors suggested that more diverse colonies might "respond to a broad[er] range of task-specific stimuli", but noted that the treatments differed even when food reserves were so low that "stimuli reflecting these needs could not have been greater."

An alternative hypothesis is that genetic diversity would affect within-colony cooperation. The authors suggested that any such effect should be in the wrong direction, because greater genetic diversity "erodes high levels of relatedness among female offspring, thereby hindering the evolution of altruistic behavior toward kin", consistent with Hamilton's Rule.

But altruism towards which kin: their sisters or their mother, the queen? In an earlier analysis by Ratnieks (American Naturalist 132:217) noted that unfertilized worker bees can sometimes lay male eggs. He wrote that:

As the number of males that mate with a queen increases, the relatedness (G) between a worker and the male offspring of the queen remains constant (G = 0.25), whereas the mean relatedness between workers and the male offspring of other workers declines from 0.375 to 0.125... Therefore, it is hypothesized that an allele, referred to as a "police allele," that causes workers to favor the production of queen-produced males over worker-produced males (i.e., to police other workers) should spread in eusocial hymenopteran populations in which queens mate with many males. Increased reproductive harmony, in which worker reproduction is reduced, is, therefore (and counterintuitively), a possible outcome of lowered relatedness between workers.

If egg-laying workers forage less, then reduced altruism towards fellow workers (leading to policing) may be a benefit of increased diversity, not a cost.

Additional details on the behavior of the offspring of singly- versus multiply-mated queens could help distinguish between these hypotheses. At this point, it would be premature -- the authors have not done so -- to generalize from this study to any broader relationship between diversity and productivity.

The review article focuses on the relationship between species diversity and the stability of ecosystems, but most of their insights would also apply to other ecosystem-level properties, including productivity. Ives and Carpenter point out that there are many ways to measure stability, and different measures can lead to opposite conclusions. Consistent with my comments above about hypotheses, they say:

it is not sufficient for theory to predict correctly whether the diversity-stability relationship is positive or negative; models could give the right prediction for the wrong reasons. Instead, theoretical models must be judged by their ability to capture the entire dynamics of the empirical system.
They also call for more explicit consideration of mechanisms by which diversity might affect stability, because
if the mechanisms underlying diversity-stability relationships are not identified, it is unclear whether an observed diversity-stability relationship can be generalized to any other system.

For example, research by colleagues here at the University of Minnesota found that, although the number of plant species had no statistically significant effect on soil nitrate levels, nitrate was significantly lower when there were more distinct functional categories of plants (Science 277:1300). Lower nitrate levels in the soil mean less risk of nitrate pollution in rivers and groundwater. Can we generalize from these results to other ecosystems, possibly including agriculture?

Ives and Carpenter might say that it would depend on the whether the mechanism linking functional diversity to lower nitrate applies to other ecosystems. For example, if less diversity meant that roots were inactive (not taking up nitrate) more months of the year, then you might expect a similar pattern in other ecosystems with a similar pattern.

But what I noticed, in visiting the research plots, was that at least some of the single-species plots had lots of bare soil, not shaded by any plants. This contrasts with other ecosystems dominated by a single species (redwood forests, say, or corn fields), which often have fairly complete plant cover. Of course, some differences between ecosystems might be irrelevant to the diversity-nitrate relationship. But I agree with Ives and Carpenter: more rigorous testing of hypotheses that explicitly include mechanisms will lead to faster scientific progress.

July 18, 2007

Low-cost cooperation

In the classic cartoon posted on my office door, little Calvin refuses to take a phone message for his father, saying "people always assume you're some kind of altruist." Two papers in the latest PLoS Biology show that some altruistic behaviors can be found in chimps and rats, as well as humans. Should we be surprised?

W.D. Hamilton explained how altruism towards relatives can evolve. An altruistic behavior will be favored by kin selection if the cost to the altruist is less than the benefit to the recipient, times their relatedness. Brothers have a 50:50 chance of sharing a given gene, so Haldane supposedly joked that he would give his life to save two brothers (or eight cousins). Strictly speaking, this "relatedness" is the extent to which the two are more related to each other than to their usual competitors.

Experimental tests of Hamilton's Rule are difficult. It is often impossible to measure fitness costs and benefits because "cost" and "benefit" are measured as decreased or increased fitness: survival and reproduction. It's usually impossible to tell how much a given behavior affected lifetime reproductive success. This week's papers may seem to bypass that problem, however. Both papers studied interactions between individuals whose (extra) relatedness was zero. This would seem to predict zero altruism, even if the cost was very low.

Nonetheless, chimps were found to pull a chain twice as often when it opened a door an unrelated chimp was trying to open to get food, relative to another door. In a somewhat similar experiment, rats were more likely to pull a lever that gave another rat access to food, if the helping rat had been helped similarly in the recent past. In both experiments, the individuals were unrelated. Given zero relatedness, these results seem inconsistent with Hamilton's Rule.

1) the fitness cost to the altruist from pulling a chain must be very low, both in absolute terms and relative to the fitness benefit to the recipient of getting food, and
2) evolutionary theory doesn't predict that behavior shaped by past evolution will be optimum under novel conditions. "To fly in a straight line, keep the light on your right" was a good rule for moths navigating using the moon, but it makes them circle (and sometimes fly into) candle flames.

Over much of their evolutionary history, chimps and rats were surrounded by relatives. "Help those around you, if you can do so at very little cost" would have been favored by kin selection under those circumstances, especially if those helped tended to reciprocate. Even though the experiments used pairs of unrelated individuals who were prevented from reciprocating, behavior that would have been favored by past natural selection is exactly what we should expect. Altruism involving a significant cost (sharing limited food) or apparent risk (snakes?) would be harder to explain.

Until recently, humans also spent most of their lives surrounded by close kin. We also interacted repeatedly with people who would remember, and might return, a favor. Even today, we seem to have some tendency to treat those we see most often (members of the same church or military unit, for example) as "honorary family." But as more people move away from home and change cities and jobs every few years, will genetic or cultural evolution tend to undermine altruism? Or will Calvin finally grow up some day?

July 17, 2007

Farewell and funding

Rob Knop is leaving academia, essentially because it's so hard to get grants to support his research.

I'm all for increasing research funding, because I think benefits to society exceed the costs, on average. But increased funding might not help individual assistant professors as much as you would think. One problem, at least in the US, is that those who already have grants will usually have more publications and more preliminary data, so they (we) will also tend to be more competitive for any new money. So some professors have multiple postdocs working for them (not me!), generating yet more publications and preliminary data, while others can't get a grant.

Also, universities (which take half of each grant) would (and do) respond to increased research funding by hiring more professors than they need for teaching. This increased hiring would make it easier for new PhDs to get a job, at least in the short run, but it would then increase competition for grants, until the "tenure misery index" comes back into equilibrium.

My impression is that the Canadian system is more egalitarian and maybe less stressful. Anyone know if this is true? If so, how does this affect overall scientific productivity and quality (citations per dollar, say)?

Yes, I will be discussing an evolution paper this week.

July 12, 2007

Rhizobia, pesticides, and peer review

I have some comments on a recent paper that's only tangentially related to evolution. Actually, it's more relevant to science fair projects, the topic of my last post.

One type of science fair project my fellow judges and I are really sick of is "The effect of X on plants", where X is mouthwash, vinegar, cola, etc. The obvious question, which we always ask, is "how often are plants in the field exposed to high concentrations of mouthwash?" Unfortunately, whoever reviewed this paper in Proceedings of the National Academy of Science, claiming that "Pesticides reduce symbiotic efficiency of nitrogen-fixing rhizobia and host plants" apparently failed to ask this question.

The general hypothesis they proposed was plausible enough to be worth testing. Rhizobium bacteria infect the roots of some important crops, including soybeans and alfalfa, and provide them with nitrogen. This is my own area of research. Successful infection involves the exchange of chemical signals between bacteria and roots. It's not inconceivable that pesticides, especially any whose chemical structure is similar to the usual signal, could interfere with signaling. Of course, soil (especially the rhizosphere, right around the root) naturally contains thousands of chemicals released from roots and produced by microbes. So signal exchange between bacteria and plants has been subject to millions of years of selection for resistance to interference by extraneous chemicals. Still, maybe some pesticides are different enough from anything natural that crop plants haven't yet evolved such resistance, especially since crop evolution is been partly under our control since before the invention of synthetic pesticides. So, testing some chemicals commonly used on alfalfa or soybeans, to see their effects on the benefits plants get from rhizobia, is a reasonable idea.

But the key "science fair question" is which chemicals and at what concentrations? The authors don't seem to have given a lot of thought to this. Neither, apparently, did the reviewers who recommended that the paper be accepted for publication in this once-prestigious journal.

The chemical that reduced alfalfa growth the most was pentachlorophenol, a wood preservative. This is nasty stuff. I remember when our cat fell into an open bucket of it; he recovered, but was pretty sick for a while. Its use has since been restricted, and rightly so. But it has never been applied to agricultural fields.

Another chemical on the list, DDT, has been banned in the US for decades. It's much more toxic to insects (until they evolve resistance!) than to humans, but it breaks down very slowly in the environment, and can be a serious threat to birds. Limited, carefully targeted use for control of malaria-spreading mosquitoes may save lives, but use on crops should be, and is, prohibited in most countries.

Methyl parathion is the only chemical on the list that is applied to crops. It is more toxic to humans, so it's mostly restricted to use on crops that aren't eaten, like cotton. I haven't heard of it being used on alfalfa.

The authors don't present any evidence that these chemicals are commonly found in alfalfa fields, at the concentrations used in their experiments. They mention that "long-term agricultural studies have shown SNF [symbiotic nitrogen fixation] is markedly lower in legume crops that are treated with N fertilizer and pesticides." However, that is known to be true for nitrogen fertilizer alone (Denison and Harter, 1995, for example), so such studies provide no evidence of a separate pesticide effect.

If I had judged this paper as a grade-school science fair project, I would have wondered why they didn't include other chemicals that are found only at trace concentrations in agricultural soils, such as mouthwash and vinegar. Maybe the "peer reviewers" at PNAS made them take the data on those chemicals out of the published manuscript? Also puzzling is their apparent failure to test any of the pesticides that are commonly used on alfalfa, at realistic concentrations, of course.

July 9, 2007

Gifted education and science fairs

I have linked to Terence Tao's blog for some time, because a surprising number of people come here from there. But, although I am more comfortable with math than some biologists, I don't have any idea what most of his posts are about. On the other hand, his career advice seems good and widely applicable. A recent post on gifted education seems like good advice both for parents and for any gifted students who might be reading this.

I've always been annoyed by the competitive aspect of science fairs. I worry that students who do a really good project are going to feel like they wasted their time if they don't win a particular prize. In Oregon, where I went to high school, there was an annual scientific meeting for high school students. It was considered an honor to have your talk or poster accepted for presentation, but it was an honor within the reach of any reasonably smart student who worked for it. It was great talking to other students about their projects, without worrying about winning or losing. Sure, a student thinking about grad school needs to know that competition for research faculty positions and grants is intense, but why kill the joy of science at an early age?

Selection beats design, again

This week's paper is "HIV-1 proviral DNA excision using an evolved recombinase" by Indrani Sarkar and others, published in Science (vol.316, p.1912). This paper is yet another example showing that selection (natural or artificial) can outperform design.

To illustrate the point, let me start with a well-known example from plant breeding. Suppose you wanted to make broccoli, starting with its ancestor, wild kale? You could cross them, identify which genetic differences are most responsible for the large edible inflorescence, and transfer those genes to the wild kale. But what if broccoli didn't exist?

We know something about which genes are involved in flowering, branching, etc., so we could mess with those genes and see what happened. But I don't think all of the genetic engineers in the world, working together, could design a genome that would result in broccoli. Not on their first try, anyway.

And yet, we know that broccoli was developed by humans, without using any molecular methods at all. They did this growing a bunch of kale plants and planting seed only from those that were slightly more broccoli-like than the rest. They presumably selected for bigger inflorescences, without necessarily expecting that the final product would have one more than 6" across. In an old gardening book, I once saw a picture of an intermediate, called "brokale" or something like that, with several 1" inflorescences.

The authors of this week's paper faced a similar problem. HIV, the virus that causes AIDS, is getting easier to control, but it's almost impossible to get rid of. Like other retroviruses, it copies itself into the DNA of host cells. They wanted to make an enzyme that would cut the HIV provirus out of human DNA, starting with an existing enzyme that does the same thing with certain other viruses. The problem was, they didn't know which amino acids in the enzyme to change, to make it cut out HIV rather than its current target virus.

So, they did what people are increasingly doing when a problem is too tough for anyone to design a solution: they evolved a solution. Their approach to directed evolution was ingenious and fairly complex. They identified a sequence in HIV that had some similarity to that recognized by their starting enzyme. They modified this target sequence until the "ancestral" enzyme occasionally recognized it, then tested various mutant versions of the enzyme to find those that worked better. They recombined those that worked best, mutated again, and so on. All in all, it took 126 cycles of selection to get an enzyme that worked well. A less-sophisticated approach, analogous to breeding broccoli from wild kale without knowing the DNA sequence of either, might also have worked, but it would have taken much longer.

Some version of directed evolution is often the best approach, whenever you can't design an answer to a problem, but 1) you can compare several possible answers and say which is better, and 2) you can generate a bunch of new answers that are somehow similar to your best current answers. My personal favorites are evolving an enzyme made out of DNA, rather than protein or RNA (Science 286:2441) and this "story problem" I used in my Crop Ecology class:

A certain weed produces 100,000 seeds. The number of weeds in the field is the same from year to year. What is the chance that a given seed will grow into an adult plant and produce seeds of its own?

If you don't see the answer immediately, guess a number at random, then see what effect it would have on changes in the weed's population over years. If weed population would decrease, your number is too big, so decrease it by a random amount, and so on. Of course, more sophisticated versions of this algorithm will find the answer more quickly.

July 6, 2007


John Dennehy, the Evilutionary Biologist, has tried to live up to his name by "tagging" me and seven others. He doesn't seem that evil to me, but I like his blog. I'm supposed to post eight "random" facts about myself...

How am I supposed to select facts at random? Sure, I could select at random from some collection of factoids, as this web page does for a fellow Twin Cities blogger. But I would need to ensure that the collection is itself random. This is, of course, impossible.

The proposed algorithm (each person tagged tagging eight others) assumes an infinite supply of blogs. This is, of course, impossible.

The number of people tagged needs to decrease in each generation, or we quickly run out of untagged blogs. (This has apparently happened already.) Suppose we start with 10 people, have each of them tag 9, and so on. The total number of blogs tagged would be 10 factorial or 3628800. This is exactly the number of seconds in 42 days. A "bizarrely improbable coincidence"? That's what they said about the Babel fish.

Here are some assorted nonrandom facts, some more factual than others:

1) My parents were devout Pastafarians. That's why they named me R!

2) RR R!RR RR RRR R !!!! ! !R! !RR !R !!! !R R!R! !R! R!RR !RR! R RRR RR! !R! !R !RR! !!!! ! !R!

3) I have earned the following science merit badges (besides the obvious ones):

4) My wife and I lived apart for many years, rather than give up science, but we're together now. We live within walking distance of campus and, so far, I've had good grad students and money to support them from NSF. Who could ask for more?

5) When I asked students in my Crop Ecology class to comment (anonymously) about bias, two said I was biased against organic farming, two said I was biased in favor of organic farming, and one said I was biased in favor of the legume family. Guilty.

6) My father was a pioneer in studying the forest canopy, among other things. My mother is a teacher turned AIDS activist, whereas my sister is an AIDS activist turned teacher. My brothers are musically talented organic farming and software geniuses. My wife and I have a nice mix of nieces and nephews.

7) I have mild prosopagnosia. I can talk to someone for an hour and then not be able to spot him in a crowd. I can usually recognize people I know well, though.

8) The old farmhouse where I lived as a grad student (at Cornell) was formerly rented to Roger Payne. We pulled a scrap of newspaper out of the wall that read "Senator X says the government is prolonging the war unnecessarily, but President Lincoln denies this..."