The “Voinnet Case”: Correct Corrections?
(July 7th, 2015) Aside from one retraction, eight articles of ETH Zürich plant biologist Olivier Voinnet have been corrected by the journals so far. Large parts of the scientific community, however, are not exactly satisfied with them.
Since the beginning of this year, approximately 40 articles by the highly successful ETH Zürich plant biologist and small RNA specialist Olivier Voinnet were flagged for alleged data irregularities on the publication discussion platform PubPeer. In April, Voinnet’s current and former employers, ETH and the French Centre national de la recherche scientifique (CNRS), finally initiated formal institutional investigations into the “case”. While the investigations are still ongoing, Voinnet and his co-authors have already retracted one of the incriminated publications (Dunoyer et al., Plant Cell 16: 1235-1250, see LT online from April 13th) – and corrected eight others (see list at the bottom of the article).
Those Corrigenda by Voinnet and Co., however, left many colleagues largely unsatisfied.
The correction of the paper Parizotto et al. (Genes Dev. 18(18): 2237-42), for example, did not actually address the data originally criticised on PubPeer but, in fact, corrected a completely different image aberration.
In the case of the PLoS Genetics publication of Ciaudo et al. (Vol. 9(11): e1003791), the authors rebutted the criticism of certain image irregularities by providing the original scan of the underlying western blot. The journal accepted this evidence and added the scan to the corrected version of the article. Unfortunately, however, this newly supplemented image was found by some PubPeer commenters to be only similar, but not the same as the one presented in the original figure. It may be for this or other reasons that PLoS Genetics has now issued an editorial Expression of Concern on the already corrected article, claiming that a final resolution on how to proceed with the paper has not yet been reached.
In other cases, the problematic data was simply replaced with completely new images. Examples are the correction for Brodersen et al. (PNAS, 109 (5): 1778) as well as the one for Ciaudo et al. in PLoS Genetics (Vol(8):e1000620). In both correction notices, Voinnet et al. practically admitted a whole plethora of critical flaws (i.e., duplicated bands) but were still allowed to simply replace the offending figures with completely different ones. In the latter case, for example, the original paper presented an identical band twice on the same Northern Blot scan, which one can hardly believe to be by sheer mistake (see photo).
The perhaps most striking case so far, however, concerns the 2003 first-author paper Voinnet et al. in The Plant Journal (Vol. 33(5): 949-56), which originated from Voinnet’s PhD time in David Baulcombe’s group at The Sainsbury Laboratory in Norwich, UK. In the 2015 correction note, Voinnet again admits some band duplications in an immunoblot figure when stating “that the original Figure 3b in this paper was assembled incorrectly and included image duplications”. Since, however, the original data for the problematic immunoblot were no longer available, co-author Susana Rivas repeated the experiment twelve years later with the agreement of the editors – in order to finally replace the incriminated figure with an utterly new one. The correction notice then closes, “The data from the repeated experiment, presented below together with the original figure legend, lead to the same interpretation and conclusions as in the original paper.” So, the new data posthumously confirm conclusions that were obtained twelve years ago with demonstrably incorrect data?
Fortunately, this seems not to be the last word on this paper. Last week, Lab Times learned from its senior author David Baulcombe that it is apparently to be further corrected or even retracted as a whole. When asked whether a retraction is being considered, Baulcombe replied, “You may be aware that PubPeer has now detected additional anomalies with the Plant J. paper. I am corresponding with the authors about the response to these new revelations and have informed the journal.” And in a second email he added, “Regarding the Plant J. paper, I am waiting for agreement from all authors over the wording of a subsequent notice.” Obviously, not only PubPeer users think there is still something wrong with the paper, inclusive of its first correction.
It’s not only this Plant J. correction but also most of the others that have left many researchers asking why the corresponding articles have not been retracted right away. A question they certainly pose primarily to the journals involved as they feel that in almost all of the eight Corrigenda they have not handled the matter appropriately, given the severity of the uncovered irregularities.
One concrete criticism is that the ETH Zürich and the CNRS, who both are currently conducting formal investigations into the Voinnet papers, were obviously not involved in the editorial processes of the corrections. Seven of the eight corrections concern articles associated with Voinnet’s earlier affiliation with the CNRS. Daniel Salisbury, Deputy Executive Editor at PNAS, explained, “We have been in contact with Dr. Voinnet regarding his PNAS articles but have not communicated directly with CNRS about the matter.” And Rosemary Dickin, Editorial Manager of PLOS Genetics, also stated that their corrections “take into account only those interactions and communications that have occurred between the journal and the author group”.
Similarly, according to Baulcombe, The Sainsbury Laboratory (TSL) in Norwich, UK, where Voinnet produced the data for his above-mentioned first-author paper, was actually “not consulted regarding the Plant J correction”. Cyril Zipfel, head of TSL, stated, “I was not involved or consulted on the correction published in The Plant Journal”. Since six more of the Voinnet publications incriminated at PubPeer originated at TSL, Zipfel had already earlier assured that “The Sainsbury Laboratory is taking these issues extremely seriously”, and that “an investigation is currently ongoing and its outcome will be communicated publicly in due course”.
It seems, therefore, that most of the journals have entirely placed their trust in the authority and integrity of Voinnet himself.
The scientific community, however, appears not to be amused about the strategy the journals have implemented, so far. EMBO director Maria Leptin, for example, reacted in describing the whole Voinnet affair as “very complex” and announced that the EMBO Council had assigned a workgroup, led by its Secretary General, Nobel Prize winner Paul Nurse, to address the issue of further literature corrections.
Also, more and more scientist colleagues have been expressing their general disapproval of the affair.
Detlef Weigel, director at the Max Planck Institute for Developmental Biology in Tübingen, Germany, for example, stated, “Now it is clear that the Voinnet lab has worked against the rules in many places and has thus obtained an unfair advantage at the publishing stage.” Weigel also vented his frustration that no official decisions from the side of CNRS or ETH Zürich have been announced, yet.
And Julia Kehr, Professor for Molecular Plant Genetics at the University of Hamburg, adds, “Certainly, there are profound doubts about many of the results from the Voinnet laboratories and it seems unlikely that so many mistakes and errors have been caused by a single co-worker. I find the whole process rather disconcerting, especially the way of how the whole affair is being kept away from the public.”
Very clear words came from some comments to the case on the well-known blog Retraction Watch. Commenter “Gary”, for instance, just listed the following quotes from the eight corrections, “’paper was assembled incorrectly’, ‘the rRNA images were duplicated without explanation’, ‘due to an error in the figure preparation’, ‘A mistake was made by the authors during the assembly’, ‘Fig are not presented correctly’, ‘some errors were made in figure preparation’,...“ And then asked, “Really? All errors? [...] If I was running an experiment and came up with the above I don’t think I’d put it down to random chance (p<0.01 ?).“
David Baltrus, assistant professor at the University of Arizona, went even further and wrote to Lab Times: “As a young scientist I'm annoyed at the whole issue. [...] A lot of us struggle to get the best looking, publication quality figures. If a control doesn't work, I run the experiment again. If a gel looks bad, I run the gel again. If things aren't quite placed well enough in the gel for publication, I run the gel again. [...] There are a lot of young and old scientists that take pride in our work and avoid short cuts when it comes to generating publication quality data. We spend time doing this to the detriment of other experiments we could be performing. [...] I'm annoyed that I now have to wonder whether I should be taking such short cuts in order to be productive. I won't, but I hate that it's an option for some very successful people.”
Hopefully, it won’t be a viable option any time soon.
Olivier Voinnet’s literature corrections and retraction… so far
- Dunoyer, P. et al., Plant J. 29: 555-67 (2002):
- Voinnet, O. et al., Plant J. 33: 949-56 (2003):
- Parizotto, E.A. et al., Genes & Development 18: 2237-42 (2004):
- Dunoyer P. et al., Nat. Genet. 38: 258-63 (2006):
- Ciaudo C. et al., PLoS Genet. 5(8): e1000620 (2009):
- Brodersen P. et al., Proc. Nat.l Acad. Sci. USA 109: 1778-83 (2012):
- Ciaudo C. et al., PLoS Genet. 9(11): e1003791 (2013):
- Boccara M., et al., PLoS Pathog. 10(1): e1003883 (2014):
- Dunoyer P. et al., Plant Cell 16: 1235-50 (2004):