If the U.S. Forest Service knows one thing about the Canada lynx, it knows that it has a problem. The cat is frustratingly hard to find. Sightings are tell-your-grandchildren events, even for the researchers who spend years tracking them with radio collars. The lynx is also cute—in a bloody-whiskers kind of way. Its rarity and elusiveness are stirring as a symbol, though a symbol that only the most persistent or lucky have seen. The lynx spreads itself thin across the landscape, ranging widely over hundreds of square miles. It avoids human activity and development and lives in rough country. Historical trapping records indicate a broad range in the northern U.S. border states, but the cat hasn’t been seen in many of those places for decades, or more. It’s also unusually fecund and, under the right circumstances, might repopulate, so there’s potential. But so far, researchers have learned only enough of the cat’s habits and habitat to be able to identify likely places to look for it. Further research—what might someday lead to lines on maps that describe the range of the lynx and, perhaps, limit human activity in those areas—may push the boundaries of federal agency comfort. In other words, the lynx carries a lot of political baggage: it’s a land manager’s bogey man waiting to slink insolently through a timber sale or a ski area expansion.

The lynx, and the messengers of science who bear its tidings, are a threat to the authority of the Forest Service. So it is that seven field researchers from the Forest Service, the U.S. Fish and Wildlife Service and the Washington Fish and Wildlife Department have been put through the wringer of public and professional scrutiny for thinking critically as scientists. The scrutiny has come after the researchers—acting independently but working as part of the National Interagency Canada Lynx Survey—sent known samples of lynx and bobcat hair into the lab testing the survey’s hair samples for lynx DNA. By incorporating these blind controls into the study, the researchers intended to confirm the lab’s ability to identify lynx hair, but in doing so, they strayed from the Forest Service’s protocol for the study, which did not include controls mixed in with the field data. Blind controls are standard practice in scientific research and are used regularly to make sure field samples haven’t been contaminated and to check the compliance of contract labs. The lab and the study’s leaders, however, were upset by the breach of protocol and ordered an investigation.

Unfortunately for science, the journey into the frontier of public opinion is notoriously rough. When word leaked out that some of the researchers in the lynx detection study had sent unauthorized control samples of lynx hair, the conservative press and several Republican members of Congress made political hay out of what the Washington Times labeled “biofraud.” The spin, cultivated ignorance from the Times and from the congressmen, became an assault on the Endangered Species Act and an attempt to discredit surveys of other rare animals on public lands. It also made clear the divide between science and the use of science.

Science won’t tell us where the lynx is. If scientists or anyone else says that they know where the cat is, they’re selling something other than science. Science should tell us, or work toward telling us, where to expect the lynx to be. It’s a subtle and important difference. The former is a leap of faith: trust that the scientist knows where the lynx is. The latter is predictive, skeptical and honest to the scientific method. It invites challenge that further refines, or discards, its predictive value.

Science is driven by skepticism. It’s a way of thinking that applies basic steps to explaining the questions of the universe. Observe. Develop a simple model that explains what the observation is. Test the model. Confirm or reject the model based on information derived from the test. Repeat. The scientific method breaks down complex phenomena into discrete, testable questions whose answers can be retested and verified. Faith and trust are not part of this equation.

The Canada lynx is listed as threatened under the Endangered Species Act in sixteen northern states. The lynx was listed in March 2000 after repeated observation that the cat is not as common as it once was. Scientists looked, across a broad range of likely habitat, but couldn’t find the cat in the numbers and range that historic trapping records would indicate.

But the surveys were imperfect. The cats live in high elevations away from human development. They travel in deep snow on huge feet—their paws are about the size of a cougar’s but the animals weigh only about twice as much as a typical housecat. The accepted surveys required researchers to go into the forests in the winter and walk a grid in search of lynx tracks. They then had to take a cast of the tracks. The snow needed to be fresh to get good prints, and the terrain could be dangerous, the work arduous. It was expensive. Another technique, using cameras placed in the wild to snap pictures of lynx investigating an attractant, wasn’t as labor-intensive, but expense also limited its application. As of 1995, those were the accepted methods for finding Canada lynx, according to Forest Service guidelines. The Forest Service manual mentions DNA testing but discounts it as a viable survey technique because the accuracy of testing was then still in question and effective survey methods were unexplored.

Enter John Weaver of the Wilderness Conservation Society. In 1996, he started work on finding a viable DNA-based lynx survey. DNA held great promise for lynx surveying: if it could be refined with an effective survey technique, researchers could identify lynx with a greater degree of certainty and in more hospitable seasons than tracks in the snow would allow. But first he had to find a way to collect DNA from wild lynx.

Weaver has seen only two lynx in the wild. He once cared for a domesticated lynx, though, which helped him to observe the cat’s behavior. Noticing that the lynx rubbed its cheek against objects like housecats do, he wondered if he could exploit that behavior to collect lynx hair. After some trial and error, he settled on pads with carpet nails in them doctored with attractant as effective hair collectors. The hair, with follicle attached, was a good source of DNA. He then tried the setup two years in a row in Canada and in Montana where he knew there were good lynx populations and was successful in consistently snagging hairs from wild cats. “I felt reasonably confident that lynx would come and rub on these pads,” Weaver says.

Meanwhile he set about finding a lab that could differentiate between lynx hair and all the other hairs he expected to capture on the pads: bobcats and cougars and coyotes. Bears loved them, too. The lab would have to refine a procedure to sort through all the DNA from the hair Weaver’s pads would capture and find the strands that correlated to lynx.

The stakes were high. The lynx, charismatic as it is, and its potential listing as an endangered species had become highly politicized. The research, however, had not caught up to the lynx’s new political status. How to differentiate lynx DNA from among the other DNA pieces that might be rendered from the hair on the mats had not been investigated, though lab techniques had been refined to the point that it was a question that a lab should be able to answer.

After two years of collecting hair in the field and sending it to the Bronx Zoo in New York to be analyzed for lynx DNA, Weaver says word started getting back to him about confidence in those results. Samples from Oregon and Washington had tested positive for lynx, and he’d detailed these findings in an interim report, but the signal strength, the amount of lynx DNA found in each sample, seemed higher than what it should be. In fall 2000, he agreed to submit what was left of the New York samples to another lab in Canada. The results from that lab pointed to possible contamination of the previous samples. When the new lab analyzed the lynx DNA, it found that all of the samples came from the same lynx, even though some of the sample stations were hundreds of miles apart.

“Conceptually, with DNA there should not be uncertainty,” Weaver says.

In the end, the Forest Service and other federal land management agencies discredited Weaver’s study because the lab results were inconsistent. He was unsuccessful in determining the presence or absence of lynx. But in designing the National Lynx Detection Protocol, which researchers took into the field in 1999 as part of the interagency lynx survey, Forest Service researchers adopted much of Weaver’s field sampling method. From a scientific method standpoint, he’d developed a verifiable technique for collecting lynx hair across broad landscapes in more hospitable seasons. He’d also shown that DNA testing in different labs produced mixed results: it had the potential for contamination but could also pinpoint a single lynx’s DNA tainting a number of unlikely samples. Weaver’s study showed that there was still another model to test—whether a lab can accurately find lynx DNA in the soup of other wildlife DNA collected from the mats in the field—before asking the big question: are there lynx here?

Science is cheap, requiring only a fertile and disciplined mind and maybe a chalkboard. Collecting data, however, is expensive. Who owns that data and how they will apply it is determined by who pays. This is where ego and institutional pressure come to bear on the simple and pure scientific method. Drug companies use science to satisfy their business plans. They pay researchers to collect data and to think in the name of developing a new drug. In trying to put the first man on the moon, NASA used science to further a political agenda. The Forest Service, in trying to identify where Canada lynx live in the United States, is trying to advance its multiple-use approach to land management and not be shut down by the Endangered Species Act.

Where the interagency lynx survey broke down into the simmering scandal of the last few months is in ownership. The survey is a new type of collaboration for the Forest Service, led by codirectors from the research arm of the agency and the National Forest System, the part of the agency funded by logging and multiple use. The Forest Service designed the protocol to be used across the northern states from Vermont to Washington, paid for much of the data collection and planned to use its own lab. So it could legitimately call the shots about how that information was gathered and applied, and it did so without a ripple of dispute leaking into the public consciousness. It’s interesting to note, however, that the Forest Service didn’t completely fund the study in Washington state. That’s where mostly nonÐForest Service researchers from the other funding agencies questioned the protocol and built in their own controls to the study, according to the Forest Service’s investigation report “on improper hair sample submission” to the national survey.

“Many of the individuals interviewed expressed reservations about the conduct of the national interagency Canada lynx survey, particularly about the possibility for contamination during all phases of the process and about the reliability of the DNA analysis method used by the Carnivore Conservations [sic] Genetics Laboratory,” the investigation report states. “In addition, many of the individuals … stated they were concerned that a 1998 survey of lynx in Region 6 conducted by Federal contractor Dr. John Weaver made different preliminary findings than did the first year of the national survey.”

Weaver’s work to develop a functional survey of Canada lynx still had an unanswered question, and the researchers involved in the lynx hair controversy were trying to satisfy their skepticism. Like many field researchers, the first priority is to their work on the resource and not to agency allegiance. “The problem is field folks felt the data was theirs,” one Forest Service wildlife biologist wrote in an email to a group of other Forest Service biologists in December. “We’ve bought into the survey protocol and we want to know the lab results and have confidence its [sic] correct.”

For its part, the lab—ostensibly part of the University of Montana, but using space in a Forest Service building and Forest Service financing and employing Forest Service researchers and technicians—boasted that its work was unassailable. Scott Mills, the Carnivore Conservation Genetics Laboratory’s director, helped develop a technique to consistently identify lynx DNA and published that technique in a peer-reviewed journal. Mills has publicly said that there is no reason to question his lab’s work. They do sound science and, at the lab level, incorporate controls into each sample set they test.

Using documented positive controls is standard practice in research. The wisdom of doing so is akin to looking both ways before crossing even a lightly trafficked street. Researchers include controls—known values—in field-collected data so that there’s a reference point for interpreting the data later. The controls in this case were vials of lynx and bobcat hair, which the lab successfully identified. According to an unidentified researcher quoted in the Forest Service’s investigation report, “The purpose of submitting the blind sample was not to catch the lab doing something wrong; introducing a blind control seemed like a simple way to alleviate concerns about samples being contaminated.”

To these doubts, the Forest Service researchers organizing the survey have offered reassurances. “There seems to be a concern that there is a risk of contamination or error in processing at the lab. That is absolutely not the case. Under the supervision of [names blacked out] has ensured that the samples are handled in a precise manner such that we have eliminated risks of contamination and basically eliminated the potential for error,” the researcher states in the investigation report.

“If these individuals wanted to test the accuracy of the lab, they could have asked that controls of this type be introduced into the survey and we certainly would have considered the request.”

From the lead researchers to the field biologists involved in the Forest Service’s lynx survey, nobody’s talking. The General Accounting Office and the inspector generals of the departments of the Interior and Agriculture are investigating, and the federal employees involved in the interagency lynx survey are under a gag order. When the story started to blow up, the Forest Service called it a personnel issue and under confidentiality rules refused to disclose information about what happened. Mills would not return repeated telephone calls from Forest Magazine. Kevin McKelvey and Gary Hanvey, listed among designers of the lynx protocol, regretted the conflicting stories in the press but referred questions about the lynx survey to the Washington, D.C., communications office of the Forest Service, per agency orders. Several of the field researchers involved in the case also would not return repeated telephone calls. Other federal researchers didn’t want to get swept up in the mess and refused to speak publicly.

When word of the unauthorized control samples came to light in the Forest Service and among the rest of the agencies involved in the survey, the researchers involved were “counseled” for not following the protocol, the barest minimum of disciplinary action. That was March 2001. In December, the Washington Times ran a story, “Rare lynx hairs found in forests exposed as hoax,” saying that the biologists involved “planted false evidence of a rare cat species in two national forests” and calling the control samples a “deception.” The Associated Press took the Times piece—almost verbatim—and made it a national story. At the same time, House Resources Committee Chairman James Hansen, a Utah Republican, and Republican Representative Scott McInnis, a vocal Endangered Species Act opponent from Colorado, wrote a letter to Agriculture Secretary Ann Veneman and Interior Secretary Gale Norton calling the biologists’ actions “professional malfeasance of the highest order.” A conservative lynch mob had formed seemingly out of thin air. The Hansen and McInnis camp called for the firing of the researchers and for, possibly, filing criminal charges.

The Washington Times waded into the story with abandon. The controls were “biofraud” that cast doubt on other endangered species surveying—like a grizzly bear survey program in Washington state. At the highest levels of the Forest Service, leaders seemed unfazed about the poorly informed impeachment of agency research. In a question-and-answer session with loggers in Oregon in January, Undersecretary of Agriculture Mark Rey referred to “Hairgate” and the questions it raised about dissatisfaction with how the Forest Service does science. “We take the matter very seriously,” Forest Service spokeswoman Heidi Valetkevich said when asked about the investigation into the survey’s unauthorized control samples. “We realize the extent to which people are questioning our science integrity.”

Internally, the samples were never anything but controls. The investigation reports from the Forest Service are careful to note that. They were documented and the researchers informed their direct superiors that they were submitting the blind samples. The intent was never to do anything more than test the lab, according to the Forest Service investigation report.

Nature, the preeminent British scientific journal, weighed in from afar in January with an editorial about the lynx hair blowup. Calling the control samples a “reasonable procedure,” the editorial calls the Times on its reporting errors. “This lynching is undeserved. The fact that the biologists felt compelled to take the action they did illustrates that the study’s experimental design was deficient—blind controls should have been incorporated from the start,” the journal wrote. “Scientific leaders of the lead participating agencies, the U.S. Fish and Wildlife Service and the Forest Service, must shore up their study methods. Clean data are needed to prevent years of court battles over use of the forests. The stakes—environmental, economic and recreational—are all too high for casual methodology to endure.”

Mills rebutted the editorial a few weeks later, taking “exception to the damaging yet unsubstantiated conclusions” of Nature’s news story and editorial. “The methodology was peer-reviewed, published, explicit, complete and followed by all except those few who intentionally mislabelled samples,” Mills wrote. By not following protocol, Mills argued, the field researchers could not call the unauthorized hair samples blind controls.

This incident could have been handled better. The idea of sending in controls to the lab was water cooler talk among many of the field researchers, and a few chose to go ahead and do it. The control samples (one of them clipped from a stuffed bobcat named Harry) were irregular, the technician who received them noted in the investigation report. The locations reported didn’t correspond to known locations in the survey. Some of what turned out to be controls didn’t conform to the protocol for sending them to the lab. They were in the wrong vials or weren’t bagged properly and the rubbing mats were missing. The technician knew something peculiar was going on, according to the investigation report. Suspicions were confirmed when one of the field researchers, on his last day at work before retirement, told the lab what he’d done. That was almost a year and a half ago.

The backlash hasn’t let up. The lab wanted the researchers punished, and the Forest Service conducted an internal personnel and criminal investigation and then removed them from the survey. The issue simmered until December when it leaked to the Times and spun out of control.

The investigation report quotes two of the survey leaders giving several reasons why sending “totally unauthorized” control samples to the lab was a bad idea. Among other things, answering questions about the controversy ended up wasting their time. The media might misinterpret the story and “lead the public to question the results of the entire survey.” The “integrity” of all the samples coming from those national forests might be called into question. This last statement holds an interesting twist—by questioning the authority of the lab, the fieldwork itself is in question.

It’s hard to imagine that when the survey is completed in 2004 that this incident won’t color the survey’s findings and conclusive proof of the presence or absence of lynx in an area will continue to be elusive.

But that’s the lynx.

“These are very rare animals and it’s impossible to state that they absolutely don’t occur in an area,” Weaver said.