A group of Swedish scientists said that a high-profile 2005 study on the mechanisms involved when plants bloom was based in part on manipulated data.
The study was published in Science in September 2005 and listed by the U.S. journal among the biggest scientific breakthroughs that year.
The study's authors, from the Swedish University of Agricultural Sciences in Umea, northern Sweden, have sent a letter asking Science to withdraw the study, said Ove Nilsson, who led the research team. The letter was to be published Friday.
The study claimed to shed new light on the molecular processes that control when a plant blooms.
Nilsson said later findings had shown that a professor from China, who was part of team, had manipulated some of the data.
"It's horrible to have to say that we've published erroneous data, but I'm glad that we were the ones to discover it," Nilsson told The Associated Press.
The Chinese professor, Tao Huang, left the Swedish university in August, but his replacement identified problems with some of his research, prompting Nilsson and his team to review his findings.
"We began an investigation and contacted him and we concluded that he had manipulated them to get the patterns he wanted," Nilsson said.
He said Huang initially acknowledged to cheating but later changed his mind.
"He's now saying he hasn't done it," Nilsson said. "He is still convinced it's correct."
Huang did not immediately answer an e-mail seeking comment.