Scientists say they've found a reason bird flu isn't spreading easily from person to person: The virus concentrates itself too deep in the respiratory tract to be spewed out by coughing and sneezing.

But the virus could change that behavior by genetic mutation, taking a step toward unleashing a worldwide outbreak of lethal flu.

Experts said the new finding doesn't indicate how likely such a pandemic is. The virus may also need other mutations to take off in the human population, they said. Still, the work suggests a particular sign to watch for in new virus samples to help gauge the danger to humans.

The work, reported in Thursday's issue of the journal Nature, comes from University of Wisconsin-Madison virologist Yoshihiro Kawaoka with colleagues in Japan. Similar results, from the Erasmus Medical Center in Rotterdam, the Netherlands, will be published online Thursday by the journal Science.

More than 180 people are known to have been infected with the bird flu virus H5N1. Virtually all are believed to have caught it from infected poultry. But scientists have long warned that the virus, which is prone to mutation, could transform itself into a version that spreads easily from person to person. That germ could touch off a pandemic.

Ordinary flu viruses spread when an infected person coughs or sneezes, blasting out tiny droplets carrying the germ to others. For that to happen, the virus has to be perched in the right places to be ejected by a cough or sneeze. The new work suggests H5N1, by contrast, infects humans too low in the respiratory tract for that to occur.

Both research teams used human tissue removed from various parts of the respiratory tract — the region from the nose to the lung — to study where virus infection occurs.

Scientists already knew that bird flu viruses use a specific kind of docking site to enter cells they infect, while human flu viruses use a different one. Kawaoka's group found the bird virus docking site appears mostly on lung cells, while being rare on cells found in higher areas like the nose and windpipe. Those higher areas were dominated instead by the human-type docking site.

Kawaoka said that for H5N1 to become a pandemic virus, it would have to mutate in a way that lets it attach to the same docking site human viruses use. Other mutations would be needed as well, he said in a statement.

Robert M. Krug of the University of Texas at Austin called Kawaoka's work an important observation, and said that if H5N1 begins to use the human virus docking site "we've got a lot to worry about." It's not clear whether that would be enough to produce a pandemic germ, he said.

James Paulson of the Scripps Research Institute in La Jolla, Calif., stressed that other viral factors may be important in human-to-human transmission. But he said that once the virus has a foothold in a person, regardless of where it is in the respiratory tract, it may mutate to gain the abilities it needs to start spreading among people.