Documenting plagiarism in a bee taxonomy paper
And what happened when I alerted the journal
--
In November 2020, a correction was quietly published for a paper in the Canadian Journal of Arthropod Identification. It corrected a paper originally published back in 2011, “Leafcutter and Mason Bees of the Genus Megachile Latreille (Hymenoptera: Megachilidae) in Canada and Alaska” by Cory Sheffield, Claudia Ratti, Laurence Packer, and Terry Griswold (full transparency, Terry Griswold was my PhD adviser at Utah State University from 2011–2017). The text of the correction reads, in part:
Although not explicitly stated by Sheffield et al. (2011), large parts of the text for the species descriptions used in that work were taken from the various treatments of the genus Megachile by Mitchell (1926, 1927, 1934, 1935a, 1935b; 1936a, 1936b, 1937a, 1937b, 1937c), especially Mitchell (1962), in addition to Sheffield and Westby (2007).
The entire correction is rather long and, in my opinion, not very straightforward. I think a more appropriate wording would be that Sheffield et al. (2011) plagiarized large portions of text from other works, particularly Mitchell’s (1962) “Bees of the Eastern United States”.
Importantly, while the correction does address the overlap in the species descriptions to some degree, it doesn’t address multiple outstanding issues in different parts of the paper.
The discovery
I originally discovered the plagiarism back in July 2020 while identifying some Megachile specimens. I was using Mitchell’s 1962 publication “Bees of the Eastern United States” to key out some specimens. I came across a male of what appeared to be Megachile gemula, but the front legs were yellow, which didn’t match my reference specimens. I wanted to check what the normal color of the front legs was, so I looked up Mitchell’s species description to check. However, Mitchell’s description wasn’t very clear.
I knew that there was also a description of Megachile gemula in Sheffield et al. (2011), so I opened that work and looked up the description. I can distinctly remember my confusion as I read the text. As I read it, I thought I was confused and figured I must have accidentally read the description in Sheffield twice, so I went back over to Mitchell to read the description. That’s when I realized the text was the same in both works. I kept switching back and forth, and to my growing dismay realized it wasn’t just that section…practically the entire description was the exact same.
Below is an illustration of some of the overlap in the description of the Megachile gemula male. Yellow and green correspond to overlapping areas, words highlighted in red are slightly different but have the same meaning:
After finding the initial overlap in the description of Megachile gemula, I went and checked the description of additional species in Mitchell’s Bees of the Eastern US. I found overlap in every description I checked. It was clear that most or all of the species descriptions contained large sections of text that had been copy/pasted with only minimal changes.
Below is an example showing the entire description of the species Megachile addenda. The left column is from Mitchell’s (1962) Bees of the Eastern US, and on the right is Sheffield et al. (2011) Megachile of Canada.
Yellow denotes overlap, red denotes a synonym, and purple and green show overlap that has been moved to a different location.
And here is the description of the female of Megachile addenda. As in the male description, yellow denotes overlap, red denotes a synonym, and purple and green show overlap that has been moved to a different location.
As can be seen from the comparisons, there are large sections of text that have essentially been copy/pasted with minimal changes. There are 18 species that are treated by both Mitchell’s Bees of the Eastern US and Sheffield et al. (2011). As far as I have checked, there appears to be substantial overlap in every one of those descriptions.
Despite the copy/paste overlap, there is no acknowledgment anywhere in the paper that Mitchell’s work was used. Mitchell’s papers are cited in various parts (such as the introduction), but nowhere does it say his work has been reproduced or modified.
Overlap with other works by Mitchell
The Sheffield et al. (2011) publication covers a number of western Megachile species that are not included in Mitchell’s Bees of the Eastern US. However, it turns out that there is still overlap with previous publications by Mitchell from the 1930’s. There is less overlap with those early works, which may be due in part to the shorter and sparser descriptions in Mitchell’s earlier works. However there are still multiple areas of copy/paste level overlap.
Below is an example for the description of the male of Megachile coquilletti. On the left is text from Mitchell (1935) “A revision of the genus Megachile in the Nearctic region. Part II.” Yellow, blue, green, and purple all denote exactly matching text, and red denotes synonyms.
When exactly is it plagiarism in taxonomy?
In taxonomy, there is definitely room for reusing the language of previous workers. It’s common practice to take work from others and modify, update, or remix it. This is particularly common with things like identification keys, where it doesn’t make sense to rewrite something from scratch that only needs a few updates. And in general, for complex characters that are difficult to describe, if someone previously came up with a good way of wording it, it makes sense to reuse it.
However, if previous work is reused, two things should happen. First, credit should be given to the previous work. Second, one should not simply copy and paste entire sections with no or minimal changes. In the latter case, one should simply refer the reader back to the original work rather than copying it wholesale.
In this particular case, I believe it can be clearly defined as plagiarism because it both lacks credit to previous workers and it contains large sections of copy/pasted text with minimal changes. While the various works of Mitchell are cited in other parts of the manuscript, no indication is given that any of that work has been reused or built upon. Any reader or reviewer would instead be given the impression that it is all original work.
Deciding what to do
After discovering the plagiarism, I knew I needed to do something, but wasn’t sure what. I looked around online about the proper practices and protocols, and I came across the COPE guidelines, which clearly lay out the protocols that journals should follow in cases of suspected research misconduct, including plagiarism, and they even include a helpful flowchart. The guidelines are clear and reassuring, and after reading though them, I felt like I had a solid picture of what would happen if I contacted the journal about the issues.
I also discussed the issue with a couple of trusted people who advised me to either not report it or to remain anonymous. It was taken as a given that pursuing this publicly or non-anonymously would have negative repercussions for my career. Ultimately I decided to alert the editor of the journal but request that my identity not be passed onto the authors. Not reporting was something I never seriously considered.
Alerting the journal
I originally alerted the editor in chief of the Canadian Journal of Arthropod Identification, Dr. Heather Proctor, to the issues in July 2020.
According to COPE guidelines, if suspected plagiarism is reported, the journal should first check the degree of copying and then take appropriate action based on those findings, which may involve correcting or retracting the paper. As a result, I sent the editor a couple brief examples of the plagiarism, which I felt was enough to convince them that a full investigation was necessary.
In September 2020, the editor informed me that the issue would be addressed by a correction issued by the journal. At the time, I disagreed with that decision but felt reluctant to push back, in part because I had myself just submitted a manuscript to the Canadian Journal of Arthropod Identification. In fact, the editor informed me of the decision to issue a correction in the same email where she acknowledged the receipt of my manuscript. While I don’t believe the editor necessarily meant to link the two things together, I was uncomfortable enough with the whole situation that I ended up withdrawing that manuscript from consideration and communication between myself and the editor ceased.
The correction… and other issues not addressed by the correction
The correction was subsequently issued on November 27th, 2020. It was only authored by Dr. Sheffield and did not include any of the other three authors of the original paper. The correction acknowledged, in part, that “Although not explicitly stated by Sheffield et al. (2011), large parts of the text for the species descriptions used in that work were taken from the various treatments of the genus Megachile by Mitchell.”
However, the correction conspicuously avoids the word “plagiarism” and it includes a lot of details that, in my opinion, seem like they do more to obfuscate the issue than acknowledge it.
It remains unclear whether the journal performed its own independent investigation to determine the full extent of the issues. I suspect they did not, in part because the correction failed to address other outstanding issues, including overlap in the body-length measurements reported by Mitchell as well as an additional instance of self-plagiarism in the introduction. Further, there are additional minor issues that include uncredited use of other works in the key and species diagnoses.
Additional Issue #1: Reuse of the body length measurements from Mitchell
In addition to the overlap in the text of the descriptions, I noticed that most of the body length measurements reported by Sheffield et al. (2011) overlap with those of Mitchell (1962) as well as some of Mitchell’s works from the 1930s.
For example, of 18 species treated by both Mitchell (1962) and Sheffield et al. (2011), 15 species have the exact same body length for both males and females. Some of the body lengths are rather unique ranges, such as Megachile pugnata which has a body length of 12–18 mm in the females and 11–13 mm in the males. The odds of all these overlapping by chance seem extremely low.
In addition, many of the western species treated by Sheffield et al. (2011) that were not included in Mitchell (1962) still overlap with Mitchell’s earlier work from the 1930’s. For example, both Sheffield et al. (2011) and Mitchell (1937) report the body length of Megachile mellitarsis as 11–14 mm in the female and 11 mm in the male.
It’s also worth pointing out that the measurements reported in Mitchell (1962) are largely the same as Mitchell’s earlier work from the 1930’s, so many of the measurements are originally from the 1930’s, when much more limited material was available.
Overall, this is an issue because like the descriptions, the body lengths are presented in a way that implies they were measured by the authors, but they were not. In my opinion, this verges into the realm of data fabrication.
Additional Issue #2: Self-plagiarism in the introduction
The second issue not addressed by the correction is that there is word-for-word overlap in the introduction of Sheffield et al. (2011) with a paper previously published by Sheffield and Westby in 2007 in the Journal of Hymenoptera Research. A paragraph on the biology of Megachile is reused, along with a few other sentences in the introduction. I consider this to fit the definition of self-plagiarism.
The overlap in the biology section is shown below, with exact overlap highlighted in yellow:
Additional Minor Issue: Uncredited use of previous work in diagnoses
A number of the species diagnoses are clearly based in part on a 1979 MSc Thesis by Michael Ivanochko on the Megachile of Canada. The diagnoses appear to be paraphrased and have novel parts added rather than copy/pasted.
One example of more extensive overlap is found in the diagnosis of the female of Megachile perihirta. Here, corresponding highlighted colors indicate matching paraphrased sections rather than verbatim overlap:
The practice of adapting diagnoses is also something that is fairly standard in taxonomy, and I wouldn’t consider it a big issue if there was a statement somewhere in the paper that the work was based in part on the previous work by Ivanochko (1979) or if this issue occurred in isolation.
However, while Ivanochko (1979) is cited in the paper, it is only in parts about biology and in a summary of the previous work done on Megachile. Specifically, Sheffield et al. (2011) states in the introduction:
Ivanochko (1979) provided the first comprehensive account of leafcutter bees in Canada, though he omitted the subgenus Chelostomoides (which was later covered by Snelling (1990)), and summarized their biology with respect to alfalfa pollinating potential, but remains unpublished and virtually unknown.
This lack of credit combined with the other issues falls into a broader pattern of reusing work with insufficient credit.
Additional Minor Issue: Uncredited use of previous work in the key
One additional issue is that parts of the identification key in Sheffield et al. (2011) are very clearly modified from other works. For example, the part of the key to the seven species in the subgenus Litomegachile is clearly adapted from Mitchell’s 1935 revision of that subgenus. However, this is not acknowledged in either the original work or the correction. It’s really not clear why, since modifying and reusing keys (typically with attribution) is standard practice in the field.
Here is an example where the wording is paraphrased but fundamentally the same, with no new characters added. Here, corresponding colors indicate matching paraphrased sections rather than verbatim overlap:
A similar example can be seen in the portion of the key in Sheffield et al. (2011) for species of the subgenus Sayapis, which is clearly modified from the 1979 Ivanochko thesis on the Megachile of Canada. Again, the same colors indicate matching paraphrased sections rather than verbatim overlap:
Again, it is standard practice to modify existing keys, but attribution should be given. The authors of this paper would have known it is necessary to give attribution, as this is exactly what the lead author did in one of his earlier papers (Sheffield and Westby 2007), stating:
The key provided is based on those provided elsewhere (Mitchell 1962, Ivanochko 1979) and through examination of specimens collected throughout Canada
Why a similar statement was not included in the Sheffield et al. (2011) paper is unclear.
Confronting the authors
After alerting the journal and seeing the correction, I set the issue aside for two years. However, the problem came back into the spotlight when I was writing the checklist of the bees of Minnesota, which cites all the taxonomic references for each bee group.
The question arose about what to do with the Sheffield et al. (2011) paper. Should we cite it or not? In the end, we decided to cite it but with note on how it was “problematic.”
Once my manuscript was submitted in February 2023, I emailed three of the authors: Cory Sheffield, Laurence Packer, and Terry Griswold (I could not find contact information for C. Ratti) to inform them that I would be mentioning the issue with their paper and calling their paper “problematic,” and I also took that opportunity to confront them about the plagiarism, correction, and unanswered questions about whether the body-length measurements were also copied from Mitchell. I suggested that a retraction and reissue of the paper with the plagiarized parts cut out would have been a better way to address the issue than a correction.
I also asked if there were any additional instances of plagiarism or unacknowledged overlap in this work or any others.
I received a response from Dr. Sheffield that, in my opinion, downplayed the issue. He presented it as a matter of a lack of proper attribution that was addressed by the correction. In addition, Dr. Sheffield said that many of the body-length measurements were indeed taken from Mitchell, but he considered them part of the descriptions and thus already addressed by the correction. He also said he would discuss my suggestion of retracting and reissuing the article with his coauthors and the editor.
The response by Dr. Sheffield did not address any of the other outstanding issues that were not addressed in the correction and that I did not bring up in the email. At the time I asked whether there were additional issues, I already knew of the self-plagiarism in the introduction. When Dr. Sheffield did not come clean about that issue, I performed a more in-depth check of the paper that revealed the additional issues of the unacknowledged use of previous work in the key and species diagnoses.
I replied to Dr. Sheffield on February 14th and said that I disagreed that the correction addressed the body-length measurements since it specified “the text for the species descriptions” and did not mention the body length measurements despite delving into such minute details in many other aspects. In addition, I said I was disappointed that he did not take responsibility for his actions. I never received a reply to that last email and have not had further communication with the authors.
When I confronted the authors, it was my hope that Dr. Sheffield would acknowledge all of the issues, take responsibility for them, and take steps to fix them. In my opinion, that did not happen.
The way forward
As I have suggested to the authors, I think one potential way forward would be to issue a simultaneous retraction and reissue of the Megachile paper with all the plagiarized parts excised. This way, the research misconduct would be addressed while at the same time preventing confusion about the status of the taxonomic and nomenclatural changes made by the paper.
When I first found the plagiarism in the species descriptions, a retraction and reissue seemed like a good potential option. But now that additional issues with plagiarism and lack of proper credit have continued to build up in essentially every part of the paper, I’m less certain that is a viable way to address this. I still think retracting the original paper is the correct thing to do, but I’m not sure that the problems with the paper can be fixed with a reissue of an updated version.
Ideally I would have preferred to see the authors take responsibility for this issue and take steps to proactively rectify things. However, at this point, given the growing list of issues in the Sheffield et al. (2011) paper, combined with what I see as attempts to downplay and avoid addressing them, I wonder what additional issues could be out there.
Finally, I would also like to see the journal where this paper was published — the Canadian Journal of Arthropod Identification — update it’s policies and website to clearly state the expectations and processes it will follow in cases of suspected research misconduct. I think they did a poor job handling this and I hope that the are able to commit to doing a better job in the future.
Conclusion
In the end, the main thing I feel at this point is disappointment. Disappointment that this happened, and disappointment at the authors and all the other people who don’t see this as a big issue or one that particularly needs to be fixed.
I’m disappointed that I exist in a system where the vast majority of advice I have received on this topic is to not make a big deal out of this because it could negatively impact my career. I’m disappointed by all the people who are reluctant to name this for what it is — plagiarism and research misconduct.
Frankly I’m tired of feeling like I’m the bad guy for trying to address this.
As scientists, trust forms the foundation of our whole enterprise. When mistakes are made or that trust is broken, I believe the answer is to address them head on and transparently. So I’ve written this article as a way to achieve that.
