Upstart Phylogenists Slug it Out Over Primate Data

Caviling, carping, and quarreling, is this any way to advance science? New HAVEN, CONN. "For a science that deals with the apparently simple question of who is related to whom, phylogeny has been positively littered with bones of contention. The root problem, explains former Yale University taxonomist Charles Sibley, is that a creatures appearance may not always bean accurate guide to its place in the pantheon of beasts. So until recently, phylogenists were left to cavil over the appropriate measures of relatedness. But then came a high-tech, molecular genetics procedure known as DNA hybridization that precisely measures genetic differences between organisms. Many scientists became ecstatic as Sibley and other scientists developed and refined the technique over the last decade or so. "Perhaps we should The Great Ape Debate Degenerates Into A mount the parapets and shout: ‘The problem of phylogeny has been solved,’ rhapsodized Harvard paleontologist Stephen Jay Gould in an April 1985 column in Natural History. Or do they? The small world of phylogeny is currently rocking because of a blistering, multipronged assault on the methods and conclusions of Sibley, now at San Francisco State, and his longtime associate Jon Ahlquist of Ohio University. The attack is the latest and the most serious challenge in a long-simmering feud over the proper place of molecular methods in the science of classification, a field whose practitioners have traditionally looked to morphology to establish taxonomic relationships. While it appears that Sibley and Ahlquist will weather the storm, the cost has been heavy, "I’m weary, weary, weary, of all this," says Sibley. And well he might be. Many scientists on the sidelines compare the ugly tussle to the free-for-all finale of a mud-wrestling tournament. Even that may be too kind. As one researcher puts it, "This is a story about knives in the night." The attack is led by Yale biologist Jon Marks, anthropologist-biochemist Vincent Sarich of the University of California, Berkeley, and UC, Davis chemist Carl M. Schmid. At professional meetings and seminars, and in papers under review, the three researchers are contending that DNA hybridization as practiced by Sibley and Ahlquist is so flawed that it is incapable of solving taxonomic problems. In addition, the scientists claim that Sibley and Ahlquist’s published conclusions are unsupported by the duo’s raw data. The acrimonious debate revolves around interpretations of highly technical statistics, and it has profound ramifications. At stake is the answer to perhaps the most contentious taxonomic question of all: the relationship between humans, chimpanzees, and gorillas. Until recently, most investigators either grouped chimps with gorillas, or were unable to say whether chimps were more closely related to gorillas or to humans. In 1984, however, Sibley and Ahlquist, who had worked on bird phylogenies using molecular techniques for 10 years, turned their attention to primates. They shook the scientific community to its ancestral roots with their announcement in the Journal of Molecular Evolution (20:2-15, 1984) that chimps are actually closer to us than they are to their knuckle-walking gorilla cousins. That still highly controversial conclusion, along with Sibley’s sometimes brusque, bull-in-a-china-shop dismissal of both critics and traditional taxonomic methods, ruffled more than a few feathers. Many phylogenists felt that the Yale biologist was waving his new-fangled high-tech methods in their faces and telling them their own approaches were inferior and wrong. Since then, admits a scientist who requested anonymity, "Sibley-bashing has become something of a cottage industry." The assault by Marks, Sarich, and Schmid, though, goes beyond mere acrimony to red data, or more precisely, to strange things that allegediy happened to one particular data set. Marks, who admits to substantial philosophical differences with Sibley over the proper approach to taxonomy, claims that he entered the fray for the most innocent of reasons. The self-described "young Turk" believes that chimps are closer to gorillas than to humans— indeed his Ph.D. thesis concluded as much—and he was curious about the numbers that enabled Sibley and Ahlquist to come to the opposite conclusion. Raw data, however, appeared in neither their landmark 1984 paper nor in a subsequent 1987 JME paper (26: 99-121). That the autbors would publish only the statistical manipulations of their original numbers is not unusual. What is unusual is Marks’ contention that he was denied access to the data, when he first requested it in 1986. Both Sibley and Ahlquist say that they never intentionally withheld data, they cite the hassles of moving from Yale to new lab sites—but to the accusers, something smelled fishy. Finally, late last year, Marks was able to obtain, with Sibley’s permission, a small data set from Caltech’s Roy Britten, a DNA hybridization pioneer who also happened to be analyzing the work. When the Marks group performed what they thought was the standard analysis on the 50 pages of numbers, they discovered something extremely unsettling. The results didn’t match what appeared in print approximately 40% of the time, a mismatch that Marks considers "outrageous." The discrepency came about because Sibley and Ahlquist had been applying a correction factor in analyzing some of the data. While the use of correction factors doesn’t necessarily mean had science, says Marks, nowhere in their papers had they mentioned why or how the corrections were actually done. Marks goes on to point out that the much-touted strength of molecular techniques is their supposed objectivity. But in light of what appeared to be capricious and subjective number-crunching, DNA hybridization took on what Marks and company saw as a distinct aura of voodoo genetics. "The difference between science and witchcraft is that, in science, we have data and documentation for the points we make," Marks explains. "We have a situation here where people were willing to accept the conclusions as if they were witchcraft, as if they didn’t require documentation." "Scientists have a responsibility to analyze data in the best possible manner, to make the method of analysis absolutely clear, so that somebody can see whether or not it’s been done properly, and to make the data available," Marks continues. "Sibley and Ahlquist haven’t lived up to their responsibilities." Was the two researchers’ manipulation of the data legitimate? The answer is buried in the arcane details of the molecular technique. The two scientists begin by separating the double strands of DNA from two different species, and labeling one set of the single strands with radioactive iodine. Next, the single strands are cut into appropriate lengths, mixed, and allowed to find partners. The resulting double-stranded molecules (where one strand comes from each species) are then filtered out and heated. As the temperature rises, more and more of the double strands break apart or "melt," a process that can be measured by the changing levels of radioactivity. In general, the more similar two DNA strands are, the more tightly they will bind, and the higher the temperature of melting will be. But the two warring sides disagree sharply over which temperature analysis procedure to use. Sibley and Ahlquist, like most molecular taxonomists, opt for calculating an offshoot of the median melting temperature. Marks camp, on the other hand, insists that the mode melting temperature, the level at which the most strands come apart, is more "biologically meaningful," as Marks puts it. The choice is important because conclusions about who is related to whom can vary dramatically with the analysis technique. Once the calculation is made, the resulting number must then be compared with its counterpart following the melting of two strands of DNA from the same species. The difference between the two numbers is dubbed "delta T." According to the tenets of hybridization lore, small delta Ts indicate close kinship while larger ones suggest more distant relationships. Usually the results are repeatable, explains Jon Ahlquist, but sometimes the radioactive iodine "bashes up" the DNA strands and produces distorted melting curves. Like the Supreme Court’s celebrated “I know it when I see it” definition of pornography, Sibley and Ahlquist’s long experience enables them to know a reasonable run from a bad one, they say. And so they apply correction factors that allow them to use the occasional bad runs. But when Marks, Sarich, and Schimid analyzed Sibley and Alquist’s raw data, they got a significantly different result: Chimps are not closer to humans than to gorillas. In their analysis, the relationships are too close to call. "Given the personalities [of Sibley and Ahlquist], I can understand what was done," says Sarich. "In part, it’s bad judgment [on their part], compounded by a certain degree of actual fudging, compounded by a certain degree of wishful thinking. Fraud is rare in science, but manipulation to fit a mindset is common. It’s classic." To this, Sibley fumes. "I regret the day I ever decided to do anything with the hominoids and get mixed up in all this," he says. "It’s been a complete pain." Sibley and Ahlquist are ornithologists, and they’re much more interested in the relationships between Australian songbirds than in rancorous questions of kinship among primates. "If it were just any three old monkeys or birds, it wouldn’t be a cause cèlébre. It’s a trivial matter, except that it’s us [that is, humans]," says Sibley. "We weren’t clued into the feelings about hominoid relationships. We simply looked at the numbers, and there it was." But that still doesn’t answer the scientific challenge to their methods. Did they in fact go too far in massaging their data? "Perhaps they should never have made the corrections in the first place, it appears to represent the things graduate students do on dark nights when they get confused," comments Roy Britten. But Britten, like most other molecular taxonomists, nevertheless gives Sibley and Ahiquist a clean bill of scientific health. "You have to look at the whole story, not at just a small amount of data." Yale biologist Jeffrey Powell puts it this way: "You always think of molecular biology as objective,”" he says, "but in point of fact, the experimenter often looks at the data and says, ‘Oh, this didn’t work today.’ That’s the value of experience. It’s not truly objective anymore, but that’s the way all science is done." Sibley, for his part, is by turns contrite and defiant. Asked to respond to the critique, he sighs, aloud about declining to comment, and then launches into a spirited defense. "We admit to two mistakes. We should have found some way to publish some of the raw data," he explains, adding that editors are more often interested in seeing the calculated data than reams of numbers. "And we should have explained in print the way we made corrections, which are perfectly logical, appropriate, and correct. Why we didn’t is, ex post facto, kind of mysterious to me." Sibley adds that they’d written about their procedures for an asyet-unpublished book that was supposed to be available two years ago. Both sides are sticking to their guns about the propriety, or lack thereof, of Sibley and Ahlquist’s methods. Marks continues to stump for the chimp / gorilla kinship, while, interestingly, Sarich thinks that the scientists he’s criticizing got the right answer but their methods are so tainted that they don’t permit conclusions. And various scientists in the field who have analyzed the original delta Ts are of two minds whether they unequivocally link us with chimps. But even as Sarich, Marks, and Schmid wait for journals’ responses to their papers (one of which Britten describes with a barnyard epithet, saying that it is "madness beyond belief"), the evidence seems to be tilting against them. Recent work by Yale’s Jeffrey Powell, which was reported on by a colleague, Adalgisa Caccone, at the meeting of the Society for the Study of Evolution on June 7, supports Sibley. The Powell group employs a refined method of DNA hybridization, and their techniques eliminate many of the methodological concerns of the critics. Their findings, corroborate what Sibley and Ahlquist have reported, Powell says. "We’re right on their data. I’m kind of surprised it’s so close." So while it is too early for Harvard’s Steven Jay Gould to return to the parapets, it appears that Sibley, Ahlquist, and the technique of DNA hybridization will survive this latest challenge. "We all get locked into tradition, and what we really need is a dialogue," concludes Ahlquist, playing the role of diplomat. "DNA hybridization is simply one more good tool to help us interpret morphology." Bruce Fellman is a freelance writer based in Connecticut.

Upstart Phylogenists Slug it Out Over Primate Data

Interested in reading more?

Become a Member of