News this Week

Science  09 Nov 2007:
Vol. 318, Issue 5852, pp. 896
1. ASTROPHYSICS

Universe's Highest-Energy Particles Traced Back to Other Galaxies

Every so often, a subatomic particle crashes into Earth's atmosphere packing as much energy as a large hailstone. Physicists have struggled for decades to determine where such ultrahigh-energy cosmic rays come from, what they consist of, and how they are accelerated to energies 100 million times greater than particle accelerators have reached. Now answers may be in sight. Ultra-high-energy cosmic rays appear to come from the neighborhoods of certain nearby churning galaxies, physicists working with the gigantic Pierre Auger Observatory in western Argentina report on page 938. The finding marks a first big step toward explaining the mysterious particles, others say.

“This is of the greatest importance,” says Veniamin Berezinsky, a theorist at Gran Sasso National Laboratory in Assergi, Italy. “If it is correct, it solves one part of this puzzle completely.” Alan Watson, an astrophysicist at the University of Leeds in the U.K. and spokesperson for the 300-member Auger collaboration, says, “For myself, it's deeply satisfying to have gotten to the beginning of the end of this whole riddle.”

To trace the cosmic rays, the Auger team employed a detector array larger than Tokyo. When a cosmic ray hits the atmosphere, it triggers an avalanche of particles tens of kilometers long called an extensive air shower. By sampling the shower with detectors on the ground, physicists can determine its direction and, hence, the direction of the cosmic ray. The shower also causes nitrogen molecules in the air to fluoresce, so on moonless nights, special telescopes can see the showers. Because the highest energy cosmic rays arrive at a rate of less than one per square kilometer per century, the Auger team has carpeted 3000 square kilometers of Argentina's Pampa Amarilla with more than 1300 detectors and deployed four batteries of telescopes.

Physicists measure the energy of the highest energy rays in exa-electron volts (EeV). The Auger team finds that rays with energies higher than 57 EeV—of which they see 27—generally come from directions within 3° of “active galactic nuclei” (AGNs) that lie within roughly 250 million light-years of Earth. That's close enough that the particles aren't sapped of their energy by interactions with the afterglow of the big bang, the cosmic microwave background. AGNs are thought to be supermassive black holes at the hearts of galaxies that are slurping up matter and spewing radiation. The cosmic rays do not point precisely to the AGNs; presumably, our galaxy's magnetic field deflects them in transit. Details of the analysis suggest that the cosmic rays are protons.

The results don't prove AGNs are sources of the rays. “Anything else that's distributed on the sky in the same way as AGNs could be the source,” Watson says. For example, galaxies tend to clump, so some other sort of galaxy might be the culprit. James Cronin, a particle physicist at the University of Chicago in Illinois and co-founder of the project, says the key point is that cosmic rays do not arrive in equal numbers from all directions. Such “anisotropy” suggests that researchers are finally on the trail of the rays'origins and gives scientists a new way to view the heavens, Cronin says: “It's the beginning of an astronomy of cosmic rays at the highest energies.”

Others had claimed to trace the origins of the highest energy rays before. In 2001, theorists Peter Tinyakov and Igor Tkachev of the Russian Academy of Sciences in Moscow analyzed data from Japan's Akeno Giant Air Shower Array (AGASA), a 100-square-kilometer array outside Tokyo that took data from 1993 to 2004. The highest energy rays appeared to come from objects called BL Lacs, galaxies with jets of matter and energy shooting out of them directly at Earth, they reported. That claim was disputed by researchers working with Hi-Res, twin batteries of fluorescence telescopes located in Dugway, Utah, that took data from 1997 to 2006. Last year, Hi-Res researchers argued that there was no correlation in either AGASA's data or their own but that there was correlation in their data if they included another subtype of BL Lac.

These claims suffered from a key weakness, says Todor Stanev, a theorist at the University of Delaware, Newark. The sky is full of possible sources, so by selecting the sources in just the right way, researchers can inadvertently manufacture a specious correlation, he says. The only way to guard against that is to test the claimed correlation in a new data set. That's what the Auger team did. First, researchers searched data collected from 1 January 2004 to 26 May 2006 to find a correlation; then they confirmed the correlation by analyzing data collected from 27 May 2006 to 31 August 2007 using the same criteria. “The Auger collaboration has been excellent in setting the rules and not deviating from them for any reason,” Stanev says.

Still, not everyone is convinced that the observation will hold up as Auger collects more data. Even with the check against the second data set, the Auger team estimates that there is a 1-in-1000 chance the correlation with AGNs is a meaningless fluke. Statistically speaking, that makes it a “three-sigma” observation, says Gordon Thomson, an experimenter at Rutgers University in Piscataway, New Jersey, and a member of the Hi-Res team. “There are many three-sigma signals that come and go,” Thomson says. But he adds that Auger's claim “is sufficiently interesting that a postdoc and I are looking at our data” for a correlation with AGNs.

Spurred by their first big result, the Auger team is pushing to build a second array at least three times larger in the Northern Hemisphere, which would enable them to view the entire sky. Researchers hope to submit a proposal within a year, Cronin says. Meanwhile, theorists have a puzzle to solve: Exactly how might an AGN accelerate a proton to such mind-boggling energies?

2. RESEARCH CAREERS

Postdoc Survey Finds Gender Split on Family Issues

1. Yudhijit Bhattacharjee

A new survey of how young biologists view their prospects suggests that the main concern for women is not a hostile climate but insufficient time to juggle the needs of family and career. The study of 1300 postdocs at the National Institutes of Health (NIH) in Bethesda, Maryland, includes a call for more family-friendly policies at U.S. research institutions.

“What these findings are telling us is that universities and funding institutions need to tune the academic system to the needs of women,” says Elisabeth Martinez, lead author and a former NIH postdoc who is now a pharmacology instructor at the University of Texas Southwestern Medical Center in Dallas. Martinez and her nine co-authors are members of the Second Task Force on the Status of NIH Intramural Women Scientists. (The first issued a report in 1992 calling for equity in pay and hiring practices.) The new report recommends that institutions set up part-time positions for principal investigators (PIs), offer grant supplements to hire qualified spouses on separate or related projects, and provide affordable childcare for all researchers.

NIH is one of the places where the system is out of whack. The share of women among its 900 tenured investigators has barely budged in the past decade, from 18% to 19%, and the figure for tenure-track positions has remained at 29%. (By comparison, women received more than 40% of the Ph.D.s in the life sciences awarded in the United States during the same time frame.)

The survey found that more than 70% of the men have their sights set on a PI position compared with only 50% of the women. (The results were published in the 29 October issue of EMBO Reports.) Men were also more confident—by a margin of 59% to 40%—that they would become PIs. One apparent reason for the gender discrepancy is that women appear more willing to make career sacrifices for the sake of their families (see graph). For example, 57% of female postdocs who were married but without children said that having children would influence their career choices compared with only 29% of married men without children. Similarly, 31% of married women expressed a willingness to make concessions to accommodate their spouses' careers versus 21% of the men.

At the same time, male and female postdocs said that they felt equally comfortable in their working environments. “Overt discrimination does not seem to be the issue,” says Martinez. Three other surveys still being analyzed by the task force—of tenure-track and tenured investigators, staff scientists, and tenure-track researchers who had left NIH before getting tenure—say the same thing, reports Joan Schwartz, assistant director for intramural research at NIH and head of the task force.

Biologist Sue Rosser of Georgia Institute of Technology in Atlanta agrees that family-friendly policies are key. But she warns against underestimating how gender discrimination affects women, especially at higher rungs of the academic ladder. “It gets complicated pretty quickly,” she says, adding that many female faculty members face isolation and dismissive attitudes throughout their careers.

Schwartz says NIH is already addressing some of the task force's recommendations. For example, NIH used to allow only tenured investigators to have staff scientists. Now, tenure-track researchers can request the same support if they need to work part time for short stints to take care of a child or a family member. NIH also encourages researchers to telecommute if possible, she adds.

But providing affordable childcare is another matter. More than 1000 people are on a waiting list for 350 slots available on and off campus, says Schwartz, and employees receive preference. That makes it tough on postdocs, who are technically trainees. “Building another daycare center on campus is a priority,” she says, “but there's no money at the moment.”

3. NEUROSCIENCE

Spying On New Neurons in the Human Brain

1. Greg Miller

The realization more than a decade ago that the mammalian brain produces new neurons well into adulthood was a revolution in neurobiology. It also raised a slew of questions about the function of these new cells. In recent years, for example, scientists have debated whether newborn neurons aid learning and memory and whether aberrations in adult neurogenesis contribute to disorders such as depression (Science, 17 February 2006, p. 938; 8 August 2003, p. 757).

Now, researchers may have a powerful new tool to help tackle these questions: a noninvasive method for detecting neural stem cells in live animals—including humans. On page 980, a multidisciplinary team led by child neurologist Mirjana Maletic-Savatic of Stony Brook University in New York state describes the technique, which uses magnetic resonance spectroscopy to detect a biomarker for the progenitor cells that give rise to new neurons.

Other researchers are greeting the work with excitement, tempered with a healthy dose of caution. “This has the potential to open the doors to research that could not have been done in humans before,” says Walter Koroshetz, deputy director of the National Institute of Neurological Disorders and Stroke in Bethesda, Maryland. “If it's validated, it's the coolest thing since sliced bread,” says neurobiologist Theo Palmer of Stanford University in Palo Alto, California.

Maletic-Savatic and her colleagues began their search for a biomarker in a series of experiments with cultured mouse cells, using nuclear magnetic resonance (NMR) spectroscopy to compare the chemical makeup of neural progenitor cells from embryonic brain tissue with that of other types of cells, including mature neurons and glial cells. The NMR spectra of the neural progenitor cells had a prominent peak that was not present in the other cells, indicating high levels of a compound specific to neural progenitors. The identity of this compound isn't known, but its position on the NMR spectra, at 1.28 parts per million (ppm; the conventional unit in NMR spectroscopy), suggests that it's a fatty acid, a class of compounds that play a wide variety of roles inside cells, Maletic-Savatic says.

Encouraged by their mouse-tissue experiments, the researchers next looked for the 1.28-ppm biomarker in live rats. For this work, they turned to magnetic resonance spectroscopy, a method that works on the same principles as NMR but can be used with the scanners found in many hospitals and research centers. At first, the spectra from the live rat experiments were messy, and the peak at 1.28 ppm was a tiny blip in a sea of squiggles. But with the help of Stony Brook University engineer Petar Djurić, the researchers developed a mathematical algorithm that separated the 1.28-ppm peak from the noise. Using this approach, the researchers found relatively high levels of the 1.28-ppm biomarker in the hippocampus of adult rats, where previous studies have found neurogenesis, compared to the cerebral cortex, where neurogenesis has not been confirmed. Moreover, when the researchers administered electroconvulsive shock, which increases hippocampal neurogenesis, levels of the biomarker increased as well.

Finally, the team applied the method to 11 healthy human volunteers who sat for a 45-minute scanning procedure. As predicted, the spectra revealed higher levels of the biomarker in the hippocampus than in the cortex. In addition, levels of the biomarker were considerably higher in the youngest subjects (ages 8 to 10) compared with the oldest (ages 30 to 35), consistent with previous animal studies indicating declining neurogenesis with age.

“Only time will tell how strong this method is, [but] it looks very promising,” says Gerd Kempermann, a neuroscientist at the Center for Regenerative Therapies in Dresden, Germany. “The 1.28-ppm peak really seems to represent something related to progenitor cells.” A crucial next step, Kempermann and others say, will be to figure out exactly what the biomarker is and whether it tracks the overall number of progenitor cells or just those that are in the process of proliferating—an important distinction for interpreting experimental findings.

In addition, some researchers are wary that the unconventional algorithm used to pull out the 1.28-ppm peak from the noisy magnetic resonance spectra could yield artifacts by amplifying noise instead. But if the new method does hold up to additional scrutiny, there's no shortage of ideas about how to use it. Maletic-Savatic hopes to investigate whether neural progenitor cells behave abnormally in disorders of brain development such as cerebral palsy and mental retardation. Palmer says that the new method could help explain the cognitive decline some children experience after receiving radiation therapy for cancer: A valid biomarker could test the hypothesis that the impairments result from radiation damage to neural progenitor cells and potentially help evaluate drugs intended to protect progenitor cells.

4. NEUROSCIENCE

Hippocampal Cells Help Rats Think Ahead

1. Karen Heyman*
1. Karen Heyman is a freelance writer in Santa Monica, California.

One view of the hippocampus, based on human brain-imaging studies and other data, is that it is essential for remembering the past, as well as for imagining the future (Science, 19 January, p. 312). The relatively new idea that this part of the brain helps plan the future gets support from a paper in the 7 November issue of The Journal of Neuroscience by A. David Redish and Adam Johnson of the University of Minnesota, Minneapolis. With electrodes implanted into the brains of rats, they've captured, on the scale of single neurons, rodents thinking ahead about their routes in a maze.

“I think the paper's elegant, and it's going to be a classic,” says Randy Buckner of Harvard University. “They really show a compelling example of the rats representing in their brains a series of possible futures and using that to test what they're going to do next.”

In the 1930s, Edward Tolman looked at decision-making by using the T-maze, in which a rodent comes to a “choice point” and must decide whether it's better to take the left or right branch in pursuit of a food reward. “If you look at Tolman's [research], and you think about how that kind of a paradigm might be done in neural terms, you come to the conclusion that the rats probably have a neural mechanism to preplay what might happen next,” says Buckner, who proposed that idea in 2006 with his Harvard colleague Daniel Carroll.

Redish and Johnson were already searching for such a decision-making mechanism by putting electrode “hats” on rats. These arrays of electrodes implanted into the brains of rats can detect the firing of 100 or so individual neurons. The rats can move freely and naturally while researchers record their brain activity. In the new work, Redish and Johnson placed rats with such hats into a modified T-maze and were able to observe cells in the CA3 area of the hippocampus as the rats walked through the maze. As the animals scurried, so-called place cells, neurons that fire in response to position, signaled where in the maze the rats were.

Such observations have been standard in the hippocampal literature for 30 years. But the data have often been confounded by additional firing, which has been dismissed as noise. The problem is that neuronal firing data, or spiking, are often averaged together, which blurs temporal resolution, explains Emery Brown of Harvard and the Massachusetts Institute of Technology. In the computational equivalent of putting another lens on a high-powered microscope, Brown and his colleague Loren Frank, who is now at the University of California, San Francisco, created an algorithm in 1998 that allowed spike ensembles to be seen at the millisecond time scale at which neuronal firing actually occurs. “If you can do the analysis at the resolution at which the spikes come in,” says Brown, “you can move on to other, more important scientific questions.”

In their new study, Johnson and Redish tweaked the algorithm and used it to reveal that the apparent noise in place-cell firings was actually signal. The additional neuronal firing corresponds to place cells representing the paths forward of where the rat is standing. The firing proceeds in a sequence lasting a few hundred milliseconds. At a choice point, when the rat pauses and looks left, place cells fire that correspond to the left path even though these cells were thought to respond only when an animal was actually at a particular place. When the rat looks right, place cells corresponding to the right path go off, again as though the rat were imagining walking the route. Cognition is faster than behavior, explains Redish: “We can now see, as Adam put it, the rat thinking faster than it walks.”

If rats can visualize their routes, the next research question is, how do they decide which path to choose? There are reward pathways within the brains of rats and humans, says Buckner, who predicts that studies will link those pathways with the rats' decision-making areas. Cheese, anyone?

5. TUBERCULOSIS

Few Mutations Divide Some Drug-Resistant TB Strains

1. Robert Koenig

The first genome analysis of an extensively drug-resistant tuberculosis (XDR-TB) strain has found that only a small number of mutations distinguish it from a less drug-resistant strain and a drug-sensitive one. That means that clarifying the molecular basis of TB drug resistance might not be as difficult as some had expected, say the researchers from the Broad Institute and Harvard School of Public Health (HSPH). Other scientists, although welcoming the new data, say it is too soon to tell.

The analysis, to be posted this week on the Broad Institute's Web site, offers initial results of an ambitious international project that will eventually compare the complete genomes of several dozen TB strains from around the globe. A major goal is to generate enough molecular data to speed the development of better diagnostic techniques and therapies for multidrug-resistant (MDR) TB, which causes an estimated 450,000 new infections a year.

Detected only a few years ago, XDR-TB is even more challenging to treat because it is resistant to second-line as well as first-line drugs. The World Health Organization (WHO) in Geneva, Switzerland, estimates that XDR-TB infects about 27,000 people worldwide, but that number is uncertain because diagnosis is difficult and most cases go unreported. The TB isolate sequenced at Broad came from the largest reported XDR outbreak, in the town of Tugela Ferry in South Africa's KwaZulu-Natal province in 2005–2006. That outbreak killed 52 of the 53 persons it infected, all of whom were also infected with HIV. Since then, another 450 cases of MDR-TB have been reported in Tugela Ferry, and more than half are XDR cases.

Scientists at Broad and Harvard scrutinized draft genomes covering about 95% of that one XDR-TB isolate and two other strains from KwaZulu-Natal. They found that only 33 single-nucleotide polymorphisms (SNPs) separated the XDR strain from a drug-sensitive strain, and 29 SNPs distinguished the drug-sensitive strain from an MDR one. They also identified a small number of other mutations, including genome insertions and deletions. Several mutations were previously known to be associated with drug resistance.

“We were surprised and delighted to find so few differences,” says project leader Megan Murray of HSPH. “This simplifies the task of investigating those mutations to determine which are related to drug resistance.” Scientists also want to identify so-called compensatory mutations that enable TB bacteria to thrive despite drug-resistance mutations that might otherwise weaken them.

Sebastien Gagneux, who studies TB genomic diversity at the National Institute for Medical Research in London, told Science that sequencing the KwaZulu-Natal strains was important but that much more SNP data from a global collection of TB strains is needed to understand the molecular basis of drug resistance.

Clinical advances may take longer. Paul Nunn, who coordinates the drug-resistant TB team at WHO, cautions that improved diagnostic tools based on new sequence data “are probably a few years off,” whereas developing a new generation of drugs would take longer.

The Broad-Harvard study comes a month after a South African group announced the first sequence of the full genome of an XDR-TB strain. But the South African data have not yet been fully assembled or posted on a public Web site. The TB research lab of A. Willem Sturm of the University of KwaZulu-Natal in Durban provided the XDR isolates to both teams. The Harvard group's main South African collaborators, Tommie Victor and Rob Warren of the Centre of Excellence for Biomedical TB Research at the University of Stellenbosch in South Africa, are providing other strains for the wider Broad analysis.

The Broad team used two sequencing technologies, says institute director Eric Lander. The data were generated using Solexa sequencers—a new, massively parallel technology—and also crosschecked on more traditional sequencers. The South African team used a rival new sequencing technology, the Roche GS-FLX. Says Lander: “As the speed and efficiency of microbial sequencing skyrockets, it should be possible to extract enormous amounts of information about population variation in TB and other infectious organisms.”

As part of the wider project, HSPH and Broad Institute scientists, sequencing isolates provided by international labs, are analyzing dozens of “evolved” TB strains that include several XDR and MDR isolates from the same persons or communities. Says Murray: “At the end of the day, we will have a comprehensive list of polymorphisms associated with the acquisition of drug resistance.”

6. SCIENCE IN CHINA

Max Planck's Asian Venture Rethinks Its Agenda

1. Richard Stone

SHANGHAI—After an auspicious start, a unique joint venture of the science academies of China and Germany is undergoing a tough outside review that could lead to significant changes. The Partner Institute for Computational Biology (PICB), founded 2 years ago by the Chinese Academy of Sciences (CAS) and the Max Planck Society (MPG), has won praise for its scientific activities. But reviewers and others are concerned that management troubles are preventing PICB from realizing its full potential.

At a closed session on 23 October, PICB's outside scientific advisory board chewed over a range of options, including recommending greater autonomy for the institute and closing it down after its 5-year contract runs out in 2010, according to sources involved in the discussions. The board is now drafting recommendations that will be presented next month to CAS President Lu Yongxiang and MPG President Peter Gruss, who together will decide PICB's fate. Lu, in a meeting with a senior MPG official last week, “emphasized that closing PICB is not an option,” says CAS program officer Fang Qiang.

The two science powerhouses have been exchanging students and faculty members since the late 1970s. PICB is their most ambitious endeavor yet, and observers say that the institute and its managing director, mathematician Andreas Dress, are making a mark. Dress has demonstrated “a lot of personal dedication and energy,” says Jürgen Jost, director of the Max Planck Institute for Mathematics in the Sciences in Leipzig, Germany.

But PICB's complex management structure has caused problems from the beginning. “Neither MPG nor CAS were under the naive impression that an undertaking like a partner institute will develop straight, smoothly, and without frictions,” says an MPG official. Or as Fang says, “It's like a German baby growing up in a Chinese world.”

Much of the angst is over who should control PICB's purse strings. The institute has a €1.5 million ($2.2 million) budget, two-thirds of which is supplied by CAS. Spending decisions must be approved by CAS's Shanghai Institutes for Biological Sciences (SIBS), which oversees PICB. SIBS and other institutions with a hand in PICB sometimes do not see eye to eye on priorities. One recent example is Dress's aim to establish a research center on the toponome, the spatial arrangement underlying the functional organization of proteins and protein networks in cells. Dress says he struggled for months to win approval for the center from Germany's science ministry in early 2006. Since then, PICB has been waiting for SIBS, CAS, and MPG to reach an agreement to release the funds. One of PICB's biggest challenges, CAS and MPG officials concur, has been finding leaders. By tradition, Max Planck institutes have several directors. When MPG and CAS advertised the directorship posts in 2004, Dress, who had retired in 2003 from Bielefeld University in Germany, was invited to apply. He was appointed head of the combinatorics and geometry department. Meanwhile, a Sino-German search committee interviewed a dozen Chinese candidates but offered the job to none. Former CAS vice president Chen Zhu, now China's health minister, then recruited geneticist Jin Li, and MPG approved him as the second director. Jin, now vice president of Fudan University in Shanghai, logs only about a week of each month at the institute. But his contributions are substantial, says Dress: “He's an excellent scientist and values the institute.” Candidates for a third directorship and for Dress's job after he retires—possibly after his 3-year contract is up next year—were vetted in Shanghai last month. Management issues aside, “scientifically, the institute has done quite a good job,” says Fang. PICB's scholarship has drawn praise. Highlights include a new computer method for mapping protein networks in cells that was published in Nature Biotechnology in October 2006 and a conference dedicated to the 400th anniversary of the Chinese translation of Euclid's Elements (Science, 2 November, p. 733). Noting such achievements, MPG and CAS officials at last month's meeting forcefully argued against dissolving the institute, sources say. “We are willing to try new things,” says Fang, who notes that Lu last week floated the idea of encouraging PICB to expand the number of junior scientist groups (it currently has two) in order to cultivate future leaders. A trickier issue for Lu and Gruss is whether, and how, a precocious young institute with two masters can be granted more freedom to run its own affairs. 7. EVOLUTIONARY GENOMICS Fruit Fly Blitz Shows the Power of Comparative Genomics 1. Elizabeth Pennisi To the uninitiated, one fruit fly is like any other—all equally pesky and deserving a good swat. But in reality, these insects are quite diverse, with species that differ from each other, genetically speaking, more than a platypus differs from a primate. A consortium of about 250 researchers harnessed this diversity and in a blast of reports this week demonstrated the analytical power of comparative genomics. The group has sequenced 10 genomes from different fruit fly species; combining these with existing DNA data, they have done a 12-way comparison to track the evolution of genes, regulatory regions, entire pathways, and cellular processes. Having these patterns in hand makes it easier to spot similar features in the genomes of other species, including humans, researchers report in more than 40 research papers in the 8 November issue of Nature and in other journals. “This work has really increased the sophistication of what we can learn from comparative sequence analysis,” says genomicist Elliott H. Margulies of the National Human Genome Research Institute (NHGRI) in Bethesda, Maryland. As project co-leader Michael Eisen of Lawrence Berkeley National Laboratory in California points out, the comparison “allows you to map where [genetic] changes occur along the tree, and that allows you to study the process of evolution, not just the product.” Drosophila melanogaster, the lab standard, has powered genetics studies for almost a century. In 2000, its genome became the second animal genome sequenced, but partly because it was one-of-a-kind, it was hard to decipher. Subsequent comparisons between the human and mouse genomes and, in 2003, between the genomes of four yeast species drove home the value of comparative genomics. Immediately, Drosophila enthusiasts appealed to NHGRI to support an even larger multispecies comparison in fruit flies. NHGRI approved a plan to look at 12 Drosophila species that diverged between a half-million and 60 million years ago. Each has its own lifestyle and geographic distribution—and each has a distinct evolutionary relatedness to D. melanogaster. The genomes proved quite varied, ranging in size from 130 million to 364 million bases, with 13,683 to 17,325 genes. The amount of DNA taken up by transposable elements, or repeated regions of DNA, varied by an order of magnitude, the consort ium reports in the 8 November issue of Nature. From this material, the researchers pieced together evolutionary stories. Probing a biological pathway as it changes from one species to the next “lets you see where there's evolutionary pressure driving the change,” says Andrew Clark of Cornell University. His group looked at the impact of pathogens. They examined the repertoire of genes involved in recognizing microbes, signaling an invader's presence, and producing toxins to thwart attacks—245 genes in D. melanogaster alone and 1200 across the 12 species. In many fruit fly species, some families of genes that code for antimicrobial peptides have been expanded, Clark and other researchers found. The expansion makes sense, as a fly with multiple copies of the right genes can produce more toxin and mount a stronger defense, Clark explains. Although the group didn't find quite as many duplicated pathogen-recognition genes, these were the fastest evolving, reflecting the need for ever-changing defenses as microbes constantly come up with ways to evade detection, Clark and colleagues reported online 8 November in Nature Genetics. The evolutionary analyses yielded surprises as well. For example, Clark's team found that a new gene called drosomycin, which codes for an antifungal compound, appears only in D. melanogaster and its close relatives. There are no clues, however, as to how this gene came to be. Another surprise came from the species D. willistoni: It doesn't seem to have genes to make proteins containing selenium—proteins that researchers had thought were common to all animals. Manolis Kellis of the Broad Institute in Cambridge, Massachusetts, led an assessment of how each type of gene or regulatory region changed—or didn't change—from one species to the next, revealing specific evolutionary patterns, or signatures. Kellis and others have incorporated those telltale patterns into software to look for the same patterns in other species to pinpoint each type of DNA. “This allows us to assign function” to some regions “through computation alone,” says Margulies. Based on a common pattern of insertions, deletions, base usage, and substitutions, Kellis and his colleagues detected 192 undiscovered protein-coding genes as well as 150 that do not follow standard rules. Typically, proteincoding genes have a “stop” sequence that signals the end of the gene. But in these 150 cases, protein-coding sequences extended beyond the “stop.” “It's always a little humbling that the assumptions we are taught in school do not apply across all genes,” says Ewan Birney of the EMBL European Bioinformatics Institute in Hinxton, U.K. With these new tools, which are particularly useful for recognizing regulatory DNA, Kellis and his colleagues have pieced together a fruit fly gene regulatory network that incorporates 81 microRNAs and 67 transcription factors. “The methodology and principles are absolutely general, and they are applicable to any genome,” says Kellis. Others say that the model still needs refining to reconcile it with experimental results. But geneticist Rama Singh of McMaster University in Hamilton, Canada, is quite pleased with this beginning. Because many fruit fly and human genes are equivalent, the network “is going to tell us a lot about humans,” he predicts. The analysis bodes well for the utility of bird, marsupial, and reptile sequences in analyzing the human genome. It also argues for sequencing and comparing all the primates, says Birney: “The take-home message is that there are a lot of clear wins from doing this sort of evolutionary genomics.” 8. MICHAEL WALKER Seeking Nature's Inner Compass 1. John Bohannon Michael Walker has navigated cultures and courted controversy in his quest to prove that life forms possess a single organ for perceiving magnetic fields—a sensor based on magnetite AUCKLAND, NEW ZEALAND—Michael Walker slips off his shoes and enters an enormous Maori ceremonial room, the University of Auckland's Tänenuiarangi Hall. Padding across long wooden planks, the biologist explains the significance of each of dozens of painted wooden carvings on the walls and pillars. “This is Turi,” he says, pointing out a highly stylized humanoid with a green bird on its shoulder. “He was the skipper of a canoe which sailed to New Zealand from the Cook Islands.” For that unrivaled feat, Turi gained a place among the gods. The colonization of New Zealand about 1000 years ago, the last of the great Polynesian migrations, is a marvel of navigation. The first Maori visitors traversed thousands of kilometers of ocean with nothing but the sun and stars to guide them. How did those pioneers manage to get here? Indeed, how does any creature find its way across featureless expanses such as the Pacific Ocean? The question fascinates Walker and has come to define his career. Competing theories have been proposed to explain the uncanny orienteering of animals as diverse as birds, bees, and fish. A decade ago, Walker's team galvanized the field with the discovery of an organ that may function as an internal compass: a string of magnetic crystals in the nose of trout (Science, 5 February 1999, p. 775). “Walker is definitely a pioneer,” says biophysicist Thorsten Ritz of the University of California, Irvine. Now Walker believes he is on the verge of clinching the case that magnetite is the universal animal compass that scientists have been seeking for a century. But the belief makes Walker a maverick: Most others in the field are convinced that animals have more than one navigation organ. A hybrid mind Walker had to chart his own path to scientific success. His mother is of European ancestry, his father a famous Maori leader, “and I kept those halves of my life separate and parallel,” he says. There was the Walker who donned shirt and tie and keenly devoured his physics textbooks. And there was the Walker whose Maori grandmother taught him mythology and the lunar system for planting and fishing. The result, he says, was “a hybrid mind.” It wasn't until after a Pacific odyssey of his own—a Ph.D. at the University of Hawaii followed by 5 years of postdoctoral research in California—that “the halves finally came together,” Walker says. The fusion was triggered by his mentoring of Maori science students after returning to his alma mater, the University of Auckland, in 1990. Many of his students struggle with cultural dissonance. Walker, 53, is now helping them embrace both cultures (see sidebar on p. 907). Having a hybrid mind can at times produce “discomfort in your own skin,” says Walker, but it has advantages. “It has helped me to be aware that there's always more than one way to do things,” he says. For instance, he points out that New Zealand's first European visitor, the 18th century English explorer James Cook, crossed the ocean with the help of a magnetic compass, tracking his ship's position on a two-dimensional grid. Along the way, Cook met Maori navigators who could island-hop more reliably using polar coordinates based on star-zenith positions and known distances between islands. Walker has sailed the Pacific using both systems. He has put that experience to use in his research on biological navigation. “Animals are similar in that they use different strategies on different scales of time and space,” he says. A bird winging its way back to its nest uses visual cues, from the position of the sun to the lay of the land. But the same bird at sea will die without more guidance. The sun indicates a heading, but over a long haul even the slightest cross-breeze will cause vast displacements. How does the bird know it has been blown off course? The sixth sense In the 1860s, Russian zoologist Alexander von Middendorff first noticed that some birds migrate along fixed paths, later recognized as Earth's magnetic field lines. But it took another century before behavioral biologists demonstrated that animals can detect magnetic fields. Sure enough, attaching a magnet to a pigeon's beak interferes with its return to the roost. For a snack reward, fish can be trained to respond to magnetic fields. And a lobster transplanted from sea bottom to an aquarium orients itself toward home like a compass with claws. In principle, Earth's magnetic field provides enough information for an animal to cross an ocean, by fin or wing, without going astray. From most places on the planet's surface, the magnetic field points north, so knowing which direction you are facing is the easy part. Changes in a field's intensity or angle, which generally grow stronger and steeper toward the poles, reveal latitude. Gauging longitude is trickier. One approach is to note the sun's position as it rises and sets. Because Earth spins on an axis off kilter from its magnetic axis, the sun's intersection with the horizon changes systematically the farther east or west you go. For underwater animals, there is a mesh of north-south magnetic ridges on the ocean floor—created by the spreading crust and the periodic reversal of Earth's magnetic poles—that can be memorized. But a long-standing riddle is whether animals can detect such subtle fluctuations. “For a magnetoreception organ to work, it must be amazingly fine-tuned,” says Ritz. Not only is Earth's magnetic field faint—hundreds of times weaker than that of a child's bar magnet—but for an animal to track its position using field variations, it must detect changes as small as 1% of that signal. Several possible mechanisms emerged in the 1970s. One bona fide contender is a biological compass made of magnetite. Also known as lodestone, magnetite is the most magnetic form of iron oxide and occurs naturally in rocks. Living cells can make their own tiny magnetite crystals. Some mud-loving bacteria make magnetite for orientation, and similar crystals have been found in a multitude of higher organisms. Magnetite may serve other functions, such as locking up excess iron, says Kenneth Lohmann of the University of North Carolina, Chapel Hill, who's an expert on turtle migration. “But it's very hard to imagine these crystals aren't there for magnetic detection.” Enter Walker, who had already made a name with his innovative behavioral studies of fish in controlled magnetic fields. He became wedded to the magnetite model during a postdoc stint in California, where he bounced the idea off biophysicist Joseph Kirschvink of the California Institute of Technology in Pasadena. With his geosciences background, Kirschvink is the physics Yin to Walker's biology Yang. After the duo extracted magnetitelike crystals from a tuna (Science, 18 May 1984, p. 751), Walker was convinced that they were on the trail of the real biological compass. The magnetite model surged with two studies of rainbow trout that Walker's team published in Nature in 1997 and 2000. “We studied the fish as if it were a rock,” he says. Working their way through the head in thin, frozen slices under a magnetic force microscope—a geologist's tool—the researchers found a group of cells in the nose laden with strings of magnetite crystals. Crucially, the cells are wired to the brain via a nerve sensitive to magnetic fields. “This is really the first truly new type of sensory cell to be discovered in a long while,” says Kirschvink, who was not an author of the Nature papers. “If there is ever a Nobel Prize for magnetic field perception, Walker's name will be on it.” And a seventh? Others are not convinced that this is the full story. An alternative mechanism for magnetoreception, known as the radical-pair model, has gained a strong following over the past decade. The idea is that besides using magnetite to detect the push and pull of Earth's magnetic field, animals also keep track of it with a chemical reaction. It was only theoretical until 1998, when a candidate magnetoreceptor called cryptochrome was discovered in the eyes of animals as diverse as fruit flies and mice. When light strikes this protein, it produces two possible intermediate states differing in the configuration of a single electron. Their ratio depends on the orientation of cryptochrome—and hence, the orientation of the organism—relative to the ambient magnetic field. Because cryptochrome is in the retina, Ritz and others have proposed that it feeds magnetic information to the brain through the optic nerves. Birds, for example, may “see” Earth's magnetic field with a few quick turns of the head. “Many pieces of the puzzle that never fit well with the magnetite model have started to make a lot of sense,” says Ritz. One of these is that some animals, particularly birds and amphibians, do not seem to be sensitive to polarity; reversing a field's north and south poles in the lab often has no effect on behavior. “That is a prediction of the radical-pair mechanism,” says Ritz, “because it only detects displacement from the north-south axis, not a flipping of the axis.” A 2004 study in Nature by Ritz and others provided another test. The cryptochrome compass, but not magnetite, should be disrupted by certain frequencies of electromagnetic fields. Birds exposed to these frequencies were disoriented. “That result is very difficult to explain if birds are only using magnetite to orient themselves,” says biologist Henrik Mouritsen of the University of Oldenburg, Germany. Walker doesn't buy it. He labels the reported effect of cryptochrome-targeted radiation on orientation as “noise” rather than evidence of an alternative magnetic sense. The cryptochrome model, he says, is only supported in birds and amphibians, implying that it has been gained and lost repeatedly during vertebrate evolution, whereas magnetite—and, Walker argues, its compass function—has remained constant. The idea that evolution has maintained a second organ for magnetoreception is “absurdity,” he says. Walker's uncompromising position frustrates many colleagues. “He completely misrepresents the field,” grumbles John Phillips, a magnetoreception researcher at Virginia Polytechnic Institute and State University in Blacksburg, “only citing papers that support his view and ignoring the rest.” Phillips argues that the evidence for the radical-pair model is at least as solid as is the evidence for magnetite. “Both systems have been maintained because they do different things,” he says. Cryptochrome is the more reliable compass for determining a heading, Phillips says, whereas magnetite may be used for mapping displacements en route. Mouritsen, who also studies magnetoreception in birds, agrees: “Whether or not the story is parsimonious, the evidence shows that there are two systems for sensing magnetic fields.” Kirschvink, like Walker, dismisses the two-magnetoreceptor hypothesis. “That simply does not make sense,” he says. Walker enjoys the jousting. “This is all vigorous scientific debate,” he says. “It drives the development of theory and experiment, and peer review winnows things out.” The issue may be resolved soon. Mouritsen and Phillips, with cryptochrome-knockout mice in hand, are confident that the detailed behavioral studies will prove the radical-pair model beyond a doubt. Meanwhile, Walker, Kirschvink, and others just landed a$1.4 million grant from the Human Frontier Science Program. The grant will allow them to do ultra-fine structural studies of fish magnetite that could reveal exactly how it works.

Is Walker reaching the end of his odyssey? Perhaps. But whatever he finds in the nose of the fish, says Ritz, “there will be a truly wonderful story.”

9. MICHAEL WALKER

A Home for Maori Science

1. John Bohannon

AUCKLAND, NEW ZEALAND—A meeting of Maori scientists begins like no other. In a sundrenched conference room at the University of Auckland, a couple of dozen researchers filter in, smile, and press foreheads together in a hongi greeting. A flowing speech in the Maori tongue introduces the newcomers. Only then does the language switch to English and the topic to science.

This is Horizons of Insight, New Zealand's National Institute of Research Excellence for Maori Development and Advancement—known here as the Maori Research Centre. Today, graduate students on center grants are giving an update on their work.

The Maori are a success story among postcolonial indigenous peoples, but they still face serious problems. Among New Zealanders, Maori citizens are burdened with a disproportionately high rate of poverty, drug abuse, and violent crime. Maori students are far less likely than those of European ancestry to finish an undergraduate degree, let alone enter graduate school. And among those who do, few choose science. Boosting Maori academic success is a necessary starting point for equality, the center's co-director, Michael Walker (see main text), and others argue. That's the raison d'être of the 5-year-old center, which aims to improve the social and environmental health of Maori communities as well as support Maori Ph.D. students. When the center started, it hoped to boost the number of Maori Ph.D. students from a few dozen enrolled in 2002 to at least 500 in 5 years. They pulled it off ahead of schedule, chalking up the 500th Maori Ph.D. student last year.

One of the center's high achievers, Melanie Cheung, walks to the head of the long table.

Her project on Huntington's disease began traditionally enough, she says. To investigate what might be going wrong in the brains of patients who suffer from Huntington's, she planned to develop a model using lab-grown neurons. Her hypothesis is that faulty mitochondria, the cellular dynamos, are a key part of the puzzle.

Then Maori custom intervened. To grow neurons, she must extract cells from the brains of cadavers. But in Maori culture, she says, “the head and brain are tapu,” a sacred body part that must not be tampered with.

Rather than bagging the project, Cheung turned to her community for advice. She discussed the problem with her family, then with the elders of her tribe, the Ngati Rangitihi. Some implored her not to “mess with tapu,” she says. Others dismissed Huntington's as “a white man's disease.” But the elders “supported the purpose of the work,” she says, and invited Cheung's team, including her Ph.D. adviser, Richard Faull, to take part. The confab lasted an entire day, with a song following every speech. Not only did Cheung allay her tribe's worries about the project, “now they wholeheartedly support it,” she says.

Although the elders could not change the brain's status as tapu, they created rituals that are now part of her daily lab routine. Before and after isolating and culturing neurons, she says a prayer to “acknowledge the person who has passed, the gift that their family has given us.” She also learned a song about creation. “Sometimes I sing, sometimes I don't,” she says. “It depends if there are other researchers around, and if I feel the need. These processes are about keeping me culturally safe.”

What may seem like an extra burden has proved to be a boon. When Cheung started her Ph.D., cases of Huntington's disease were unknown among the Maori. But after engaging her tribe, “people started coming forward,” she says. Four likely cases have been diagnosed, and genetic counselors are working with Maori communities for the first time.

Cheung's efforts at bridging the cultures are ongoing. “In February, we are returning to visit my tribe with our entire research group of 60 or so people and their families,” she says. The tribe is eager for an update on their progress.

Cultural engagement like this is not only a happy outcome of the Maori Research Centre's grants, says Walker: “It's required.” Every project includes an element of community service, he says. “You have to give back.”

10. PLANETARY SCIENCE

Majority Rules in Finding a Path for the Next Mars Rover

1. Richard A. Kerr

A few dozen self-selected planetary scientists are voting their way through a minefield of fiscal and scientific uncertainty in the pursuit of martian life

PASADENA, CALIFORNIA—When it comes to exploring Mars, geologists love rocks. Engineers looking to land the next Mars rover hate them. Throw in rover mission cost overruns, a crucial design failure, a severely limited understanding of Mars, and a rush to launch 2 years from now, and the Mars exploration community has a lot to talk about.

Scientists in a small planetary town hall meeting here are helping sort out where the billion-dollar Mars Science Laboratory (MSL) rover should be landed in NASA's drive to understand life beyond the home planet. After dozens of sometimes boisterous votes by a grab bag of specialists, most of their proposed landing sites lay on the cutting-room floor. “Is there anything like this elsewhere in science?” asked a meeting organizer in an aside. Not likely.

At the end of the MSL landing site selection workshop,* participants had pared their 51 proposed sites to six for further consideration by NASA. It was a painful if democratic process. The same procedure had gone wrong once before, sending the Spirit rover to what looked like an ancient lakebed but turned out to be a barren lava plain. And this time researchers are not just “following the water” on Mars in search of past environments conducive to life. They are also hunting for places that might still harbor organic matter from past or present life. That raised the bar for candidate landing sites, as did the tighter constraints mission engineers had to place on where MSL can land.

In the end, the majority-rules approach fared well. “We've pretty much captured all the types” of landing sites while getting down to just six, says geochemist David Des Marais of NASA's Ames Research Center in Mountain View, California. “A lot of people subverted their interests [in a particular site] to the science. This degree of community participation is one reason the Mars program has been so successful.” In 9 months, planetary scientists will return to see if they can agree on a single landing site that's both safe and exciting.

A new challenge

Five craft, all American, have successfully landed on Mars. The first—two Viking landers that set down on roaring retrorockets in 1976—went in nearly blind on a wing and a prayer. The two latest—the Opportunity and Spirit rovers that rolled onto the martian surface encased in airbags—benefited from years-long analyses of landing hazards and the prospects for good science (Science, 17 January 2003, p. 326). In four workshops, interested planetary scientists proposed 185 landing zones for the Mars Exploration Rover (MER) mission and then pared their list to the safest-looking, most scientifically interesting two on the basis of the latest observations from Mars. More-detailed imaging from orbiting spacecraft, for example, allowed accurate estimates of the abundance of mission-ending boulders too small to be seen from orbit. And better spectroscopy pointed to an enticing site loaded with the mineral hematite that forms only in water.

This community-supported selection process reaped considerable rewards. The rovers landed safely on opposite sides of the planet to find landing conditions much as predicted. And Opportunity roved across plenty of hematite, although the mineral had formed in briny groundwater rather than on the postulated lakebed. Spirit, on the other hand, was in for a surprise. For all the analysis of geologic features on the floor of giant Gusev crater, Spirit landed not on the expected ancient lakebed but on a vast lava plain devoid of signs of water. Only by chance did it eventually reach nearby hills that harbored enigmatic water-altered rock.

This time around, NASA is sending the Humvee of rovers to Mars. MSL, which is slated to rove for two Earth years, weighs in at three-quarters of a ton (more than four times the mass of Opportunity or Spirit), boasts nine instruments (each developed specially for this mission, including a rock-zapping laser chemical analyzer), and runs on nuclear power. Even the landing system is brand-new. In a landing scenario variously described by scientists as “exciting” or “scary,” the rover—only one this time—will descend the final 1500 meters hanging by cables from a rocket-festooned descent stage. All this for $1.7 billion. Mars scientists would have happily settled for a couple of more modest rovers like Opportunity and Spirit (at$410 million each), but that isn't the NASA way, they explain; each new mission must be bigger and obviously better. MSL is clearly in the NASA spirit. “This has been a very ambitious task,” MSL project manager Richard Cook of the Jet Propulsion Laboratory (JPL) in Pasadena, California, said at the workshop. Engineers speak of the number of miracles required on a project, he said, and “on MSL, it seems we have more than our share.”

Whittling down

Not all of the required miracles came through. At the project's most recent NASA review, it appeared that MSL would exceed its budget by $50 million to$100 million, NASA's Mars program scientist Michael Meyer told Science. As a result, the project ended up economizing on rover instrumentation, among various cuts, and reducing the number of candidate landing sites it would scrutinize from 12 to five. “We have become painfully aware how complex and time-consuming it is to select and certify a landing site,” said Cook, and consequently expensive. New imaging revealing fields of lander-destroying boulders recently required relocation of the Phoenix landing site.

The heart of the MSL winnowing process was almost 2 days of 15-minute presentations, discussion, and voting in a Pasadena hotel. Anyone with a site to propose (and the money to travel) could make their PowerPoint pitch to the assembled dozen members of the MSL steering committee, other proposers, a handful of invited outside experts, and interested planetary scientists and their grad students. Discussion followed, with one eye on the prospects for doing science and the other on the engineering constraints. Then came the voting: a show of hands for green, yellow, or red on four science-related questions for each site. With considerable good-humored joshing, voters expressed their opinions on each site “as a potential habitat for life, past or present.”

Yellows predominated. Although strictly speaking, the voting was on the science potential, not the rover's physical capabilities, new engineering restrictions were at times factored in. For example, mission engineers had planned on using an exotic dry lubricant to keep the driving and steering systems and the instrument-laden rover arm operating down to −150°C, far below the −50°C limits of the MERs. But in testing, the new dry lubricant failed after just 20% of its required lifetime, MSL mission manager Michael Watkins of JPL said at the workshop. Returning to a wet lubricant made one site impossible and put any sites poleward of 25°S in jeopardy.

The rover's mobility hasn't worked out quite as expected, either. At the first workshop, many scientists proposed “go-to” landing sites. In these, MSL would land somewhere in a smooth, safe, 20-kilometer-diameter landing zone and then drive beyond the landing zone to interesting but unlandably rough geology.

But at this workshop, Watkins reported that on further consideration, project engineers expect MSL's driving to be “MER-like,” something like 200 meters per day rather than the expected 500 meters per day. MSL, it turns out, must charge its batteries with its radioisotope thermoelectric generator and use them to drive, rather than driving directly from the RTG. “We need prime science within 10 kilometers” of the center of the landing zone, said Watkins. That eliminated a few sites and sent proposers scrambling to find interesting geology in their landing zones.

Some attendees wondered out loud whether the geologists might be squinting at their images a bit too hard. This could be “Gusev all over again,” said spectroscopist Steven Ruff of Arizona State University in Tempe. Some geologists were interpreting layered rock as sediments laid down under lake water, he said, without discussing less interesting but quite plausible alternatives, such as blankets of volcanic ash. Such concerns helped eliminate exquisitely layered basins with no sign that water ever flowed there.

Mineralogy rather than geology ended up being the prime criterion. Most attendees took the detection of clays—the product of the watery alteration of igneous rocks—as a sign that water had long been in contact with rock there. That would bode well for life. Clays are also adept at preserving organic matter over the eons. So any site without at least a hint of clay was out of the running.

This was a bit too much for some. Many of the geobiochemists and astrobiologists—the smallest contingent in the crowd—felt that people were leaning on the clays far too much. Having clays is nice, they said, but it's not everything. Clays that formed within the ancient Mars crust, for example, are far less promising signposts of early life than clays in a river delta formed in a crater lake. Astrobiologist Dawn Sumner of the University of California, Davis, pointed to Vernal Crater, which had fared poorly in initial voting because a thin dust layer blocks any spectral signature, so researchers couldn't tell whether clay was present. But Vernal's proposer, planetary scientist Carlton Allen of NASA's Johnson Space Center in Houston, Texas, had drawn attention to what appeared to be lake deposits, lake shorelines, channels, and even hot-spring deposits, the first proposed on Mars. “These things are really important,” said Sumner. “I voted all green” on Vernal. Nonetheless, a revote kept Vernal out of the top 10.

In a final afternoon session, the workshop got down to a final six by demoting four sites to “purgatory” because they held too little promise of traces of life or their landing zones were just too boring. The remaining six push the engineering limits, although those limits remain flexible. The project asked that two of the final five landing zones be demonstrably safe, but at this point in the hazard analysis, none of the current batch is. Three are former crater lakes, the workshop's obvious favorite type of site, even though two of them are so far south that the rover would likely have to hibernate one-third of the time, beginning on landing. And two others are the crustal-clay sites so popular with the geologists but shunned by astrobiologists. Still, “it turned out better than I would have expected,” says Sumner. They ended up with a diverse suite of sites whose riskiness will “poke the engineers to find what we really want.”

• * Second MSL Landing Site Workshop, 23–25 October, sponsored by the NASA-appointed Mars Landing Site Steering Committee and the MSL Project.

11. GENETICS

Who's the Queen? Ask the Genes

1. John Whitfield*
1. John Whitfield, a science writer in London, is author of In the Beat of a Heart: Life, Energy, and the Unity of Nature.

Biologists are finding that in some social insects nature, not nurture, determines whether offspring become workers or royalty

In 1712, the English scholar Joseph Warder dedicated his treatise, The True Amazons: Or, The Monarchy of Bees, to Queen Anne, citing the caste divisions of the hive—the queen built for breeding and the workers tending her and her brood, foraging, and dying to defend their home—as evidence that nature adored royalty. But much of what entomologists have learned since then has made the lives of bees and other social insects seem closer to the American dream: Given the right nurturing—a diet of royal jelly in honeybees, or being reared at a certain temperature in some ants—any female grub in a beehive or in an ant's nest can grow up to be queen.

At least this nurture-over-nature paradigm was the prevailing wisdom, backed by theory that argued that any gene that required a developing insect to become a sterile worker would be committing evolutionary suicide. But a few years ago, social-insect research was rocked by the discovery that in some ant species, workers and queens are determined by their genes—in other words, born, not made. And this was discovered not in some obscure rain forest species but in harvester ants, which are ubiquitous across the southern United States, have been studied for decades, and are even sold online for ant farms.

“Harvester ants are the poster child for ant research,” says Sara Helms Cahan of the University of Vermont in Burlington, one of the co-discoverers of genetic caste determination (GCD). “Imagine what the other 11,999 species might be up to.” Indeed, several more examples of GCD have popped up among other ant species. Researchers such as Helms Cahan are now tackling a raft of questions about this mode of life, trying to unravel these ants' evolutionary history, exactly what genes determine castes in the insects, and what, if any, evolutionary advantage GCD confers.

The trait, it turns out, is not confined to ants. A report on page 985 shows that a species of termite also uses genes to split breeders from workers. “This is probably the first clear-cut example of how caste is determined in termites,” says Nathan Lo of the University of Sydney, Australia, a member of the team behind the discovery. And some believe that many more strange genetic and reproductive systems await discovery in social insects.

Lucky ants

It took some luck to reveal harvester ants' odd genetics. “We found it by sheer accident,” says ecologist Deborah Gordon of Stanford University in Palo Alto, California. The ants' natural history is normal enough: A single queen founds each nest and gives birth to all the colony's workers, sterile females that give harvester ants their name by gathering seeds for food. The queen also produces the next generation of reproductives: males and daughter queens, neither of which work. As in all ants, wasps, and bees, males of harvester ant species develop from unfertilized eggs and so have half as many chromosomes as females. After summer rains, males and prospective queens leave the nest on swarmlike mating flights. Males die after mating; queens found new colonies and can live for more than 20 years.

Five years ago, three teams, including Gordon's, Helms Cahan's, and one led by Jennifer Fewell of Arizona State University in Tempe, independently noticed that in some nests of the red harvester ant (Pogonomyrmex barbatus) and in other nests containing the rough harvester ant (P. rugosus), all the workers carry two different versions of a gene at certain DNA markers, whereas the queens in these colonies have identical pairs of genes at the same markers. This was weird, as genes should be spread about between the two castes at random in a colony in which any female can supposedly become a queen or worker.

Further genetic analysis of workers and queens showed that these ants were in fact neither P. barbatus nor P. rugosus but a hybrid of the two species. More complicated still, the hybrid population consists of two strains. In each nest, the queen and her daughter queens are purebred members of one strain or the other, whereas the workers are a cross between the two strains. The only way this system could work, the various research groups concluded, was if queens used sperm from males of their own strain to make more queens and sperm from the other strain to make workers, with each ant's caste decided not by its upbringing but by the combination of genes it receives from its parents. So queens must mate with males of both strains to raise a functioning colony.

Subsequent work has found that harvester ants with this GCD dominate a strip of desert from western Texas to eastern Arizona. (Other ants in this genus appear to use diet to determine caste.) So far, eight genetically distinct hybrid strains have been found, coexisting in four pairs of dependent strains, suggesting that hybridization has occurred—and GCD evolved—more than once.

Mating between species is often a dead end because it leads to sterile offspring, but two aspects of ant biology allow Pogonomyrmex to get around this. Queens with GCD that have mated with only the other strain do not lose all their reproductive options: They can still produce males from unfertilized eggs. And workers are already condemned not to reproduce, so the sterility of interstrain offspring is moot.

The origin of GCD in ants is almost as mysterious as its advantages. The obvious hypothesis, favored by Helms Cahan, is that the hybrid strains are descended from ancient crossings between the two parental species. The presence of genes from both species in each strain lends weight to this idea, as does the fact that within each pair, one strain's genes are more similar to P. rugosus, whereas the other is closer to P. barbatus (even though the two strains in each pair look alike physically). Helms Cahan has found that colonies with GCD grow more quickly than those of either pure P. barbatus or P. rugosus and that their workers are more aggressive than those of the parent species. “I wouldn't be surprised if they had an ecological advantage,” she says. Hybrid plants often grow well, she points out, something long exploited by plant breeders.

Fewell has a different view of how GCD arose. Based on analyses of DNA from ant mitochondria—an energy-generating structure inside the cell with its own small genome—she believes that the oldest of the eight known strains with GCD is most closely related to P. barbatus. This species, she suggests, evolved an “egoistic” gene that made its bearers more likely to become queens. Such a gene would aid its carriers but would also create selection for worker-producing genes to prevent colony extinction. GCD might be a way to cope with egoistic genes, not something that gives a competitive advantage, she says, and the crossing with P. rugosus might be a later event not directly related to GCD's origin.

So far, biologists have spotted only genetic markers of caste determination; the actual genes behind it are unknown. But finding those genes in harvester ants could reveal those that control the developmental pathways of social insects in general. “Pogonomyrmex is going to be a very good model to look at mechanisms of caste determination,” says Laurent Keller of the University of Lausanne, Switzerland.

Since GCD was discovered in harvester ants, it has been found in a number of other species. The southern fire ant (Solenopsis xyloni) uses sperm from males of a close relative to make workers and sperm of its own species to make queens. There are also genetic differences between the types of workers in leaf-cutting and army ant colonies. These workers come in various shapes and sizes, called subcastes, specialized for jobs such as soldiering, foraging, and nest maintenance. Like harvester ants, queens of these species mate with many males, and William Hughes of the University of Leeds, U.K., and his colleagues have found that in leaf-cutting ants, the offspring of different males are biased toward becoming a particular kind of worker, suggesting that a worker's genetic makeup predisposes her toward joining a particular subcaste. At the most extreme, more than 90% of a male leaf-cutting ant's offspring can develop into one subcaste. Hughes is now looking for genetic differences between queens and workers in leaf-cutting ants.

Hidden complexity

Inspired by the discoveries in ants, Lo and colleagues at Ibaraki University in Japan looked for GCD in termites. Unlike ants and bees, termite societies have both male and female workers, and each colony has a king and a queen. It was thought that pheromones from other nest members controlled what termite larvae developed into, although no one had ever identified such caste-determining chemicals.

Working with the Japanese termite species Reticulitermes speratus, Lo and his colleagues examined this question by depriving colonies of their king and queen, which induces some larvae and normally sterile workers to become fertile creatures called neotenics. Through crossbreeding experiments between male and female neotenics, the team established that a single gene on the X chromosome controls caste in this termite species. The gene comes in two forms, dubbed A and B. Two A's in a female make a queen, whereas males with a B gene become kings. AB females and A males become workers.

When a queen with her two A's mates with a king and his B gene, all their offspring will be sterile workers. Lo believes this helps prevent cheating in the nest. Reticulitermesworkers only become reproductives if the king or queen dies, removing a chemical signal that suppresses worker development. In contrast, in other termite species, all workers have the potential to become reproductives, placing a drain on the colony's resources. “They just sit around being looked after,” says Lo.

Lo suggests that GCD is likely relatively rare in ants, the result of unusual circumstances such as hybridization, but that it may be common among termites because of the king-queen system and its associated genetics. It's still unclear how widespread GCD is in social insects, says ant expert Andrew Bourke of the University of East Anglia in Norwich, U.K. “There's a lot of hidden complexity that it's taking modern molecular techniques to discover,” he says.

Now that insect researchers have spotted some of this complexity, they may start finding more. “People have been a bit blind,” Keller says. “They read something in a textbook, and then when the data doesn't fit that model, they throw out the data.” The discoveries so far are just the beginning, he predicts: “I think that 5 to 10% of ant species will turn out to have very weird modes of reproduction.” What Warder would have made of such royal perversity is anyone's guess.