Questions over Australia's leading honey bee research

Watch a honey bee return to its hive after foraging and you may witness a curious dance. Homecoming bees rapidly move in a figure-eight motion and then shake their abdomens. About 2300 years ago Aristotle witnessed the dance, suggesting it might be workers offloading pollen. In the 1920s, Austrian scientist Karl von Frisch decoded the dance: it was a form of communication.

He dubbed it Tanzsprache, “dance language”, and showed it was a way for bees to signal to hive mates the direction and distance to a food source. But how could these insects, with simple nervous systems, estimate how far away a flower lay? How did they know in which direction to take flight?

In the early 1990s, Mandyam Veerambudi Srinivasan, a researcher at Australian National University, answered that question. In an elegant series of experiments using patterned tunnels, he led work that calibrated the “honeybee odometer”, showing bees measure distance by determining how quickly the world blurs past their eyes during flight.

The work made Srinivasan a star. In 2001 he was elected a fellow of London’s prestigious Royal Society and by 2006 he’d earnt the Prime Minister’s Prize for Science. Since 2001, Srinivasan has been listed as a chief investigator on Australian Research Council (ARC) funding totalling more than $30 million. Today, he is retired and holds an emeritus professorship at the University of Queensland.

A cloud has formed over his seminal studies, however. In May, two researchers from the California Institute of Technology (Caltech), Laura Luebbert and Lior Pachter, uploaded a manuscript to the website arXiv.org, a repository for scientific articles yet to be peer reviewed or published. In it, they claimed that 10 papers underpinning honey bee odometry contained “inconsistencies in results, duplicated figures, indications of data manipulation, and incorrect calculations”. Srinivasan is listed as an author, or co-author, on all of them.

Luebbert and Pachter believe the errors are a potential sign of misconduct. “I don’t know what there is to debate or discuss or defend, unless the author can produce the notebooks and show exactly what happened and explain why this happened, which he already said he cannot,” Pachter says.

Australia does not have a national, independent research integrity body to investigate claims of scientific misconduct. Instead, universities are tasked with investigating researchers within their institutions, presenting potential conflicts of interest. These investigations remain confidential and participants are often required to sign non-disclosure agreements, which makes it difficult to track the extent of research integrity complaints in Australia.

The University of Queensland, where Srinivasan is an emeritus professor, said it forwarded details of the concerns to Australian National University, where the research in question was conducted. A spokesperson for ANU first told The Saturday Paper it had not received any formal complaints about the work, before clarifying it had received “a referral from the University of Queensland”.

Srinivasan is preparing a response to the claims and plans on submitting it to arXiv. In a draft copy of the response, seen by The Saturday Paper, he states the claims “are totally unfounded and the suggestions and allegations by Luebbert and Pachter as to how data manipulation might have been achieved are not only totally puzzling, but deeply concerning”.

Srinivasan remains a chief investigator on one ARC grant, worth $444,293. He says he receives no money from the grant; his major contribution is helping with data analysis and manuscript writing. Matt Garratt, a professor at the UNSW AI Institute and also a chief investigator on the ARC grant, tells The Saturday Paper: “I have the utmost respect for Srini and can only attest in the work we have done together, he has been very rigorous to ensure all results are properly reported and presented.”

Luebbert says she first noted the alleged inconsistencies in the honey bee papers four years ago. In early 2020, as a first-year PhD student, she was trying to work out where her postgraduate studies would take her. The pandemic cruelled her plans but journal clubs – where researchers review papers and give short presentations to lab mates – provided a way to experience many different research fields.

During a stint in a lab studying insect locomotion, she was tasked with reviewing two papers on how honey bees navigate the world. One, published in the prestigious journal Science in 2000, contained a calculation that Luebbert just couldn’t understand. A single number stood out: 17.7. It didn’t match the observations in the paper. “I had absolutely no idea what it meant,” Luebbert says. “I tried to gain an understanding by reproducing their analysis … and I just kept getting a different number.”

In another paper, published in the Journal of Experimental Biology (JEB), she found graphs that were identical. The graphs appeared to have been reused but claimed to show results from separate experiments with different set-ups. Luebbert was both excited and concerned about bringing the findings to the journal club. On April 9, 2020, she gave her presentation.

The findings were met with “little more than a collective shrug” – even a tenured professor appeared uninterested in the potential integrity issues. Dejected, she decided to drop it. “I honestly had kind of given up hope,” she says.

“I don’t know what there is to debate or discuss or defend, unless the author can produce the notebooks and show exactly what happened and explain why this happened, which he already said he cannot.”

Nevertheless, she raised her concerns on PubPeer, a website that allows researchers and the public to comment on peer-reviewed publications after they’ve been published – an online version of journal club. Luebbert placed her commentary on one of the papers in JEB, pointing out the graph duplications.

She turned her focus to her studies but the collective shrug was something she never forgot. When she began to work in Pachter’s lab, the pair eventually took up the cause again, finding what they believed to be further issues in the honey bee work. They published on arXiv in May and then wrote a lengthy blog post detailing some of the allegations they were making on July 2.

Of the more serious claims, Luebbert and Pachter suggest data has been “seemingly manipulated” across six papers. In their manuscript, they highlight the previously mentioned JEB papers from 1996 and 1997, and four others.

In one, published in the journal Biological Cybernetics in 2000, the Caltech duo say data had been reused from the 1996 JEB paper, and they said a 2004 paper reused “problematic data” from the former. Srinivasan again pointed to these errors being typographical.

On June 25, the two papers published in JEB, in 1996 and 1997, received “expressions of concern” – an editorial note to alert readers of potential issues. Srinivasan says he was contacted by the journal “about a year ago” and explained to it how the errors were made. He also says he takes full responsibility for them, and the data underlying the experiments, which are almost 30 years old, no longer exists.

However, he says he never heard directly from Luebbert or Pachter. “They didn’t have the courtesy to even contact me about this for any clarification before they put the thing up on arXiv.” PubPeer is designed to automatically email authors of papers with new comments, but Srinivasan maintains he was never contacted by the website’s publishers.

The expressions-of-concern notices on the JEB papers did not appease Luebbert or Pachter. “We were really not happy with them,” Luebbert says. “They completely dismissed these instances of data duplication in their two papers.”

Craig Franklin, who serves as editor-in-chief of JEB, tells The Saturday Paper the issues were investigated, alongside “an expert working in this field” and that resulted in the expressions of concern. “If additional concerns are raised, we will investigate further,” he says.

The Caltech pair do not believe the expressions of concern go far enough, nor capture the scope of the issues they’ve raised across many different papers. “I feel like given such a major discrepancy, or like a mess up like this … there’s no choice but to retract the paper,” Pachter says.

Other researchers who’ve read the Luebbert and Pachter manuscript say it would be worthwhile for journals to investigate the claims. “All in all there is enough there to look at it more closely, but I am also not immediately convinced that widespread misconduct happened,” says Roger Schürch, an assistant professor of entomology at Virginia Tech.

Srinivasan disagrees. “The results are completely solid, so I don’t see any reason to retract these papers,” he says.