Authors crop estimate that was off by a factor of 60 — or $3 trillion

A paper that tried to estimate the cost of invasive species to farming in Africa has been corrected because the researchers made a pair of errors that dramatically inflated their calculations. 

One mistake led the group, from Switzerland, the United Kingdom, Ghana and Kenya, to overstate the cost to African agricultural of invasive vegetation by roughly $3 trillion — yes, that’s trillion with a T. The second error pared their estimate for crop losses due to a single plant species from $11.4 billion to $4.1 billion. 

Now, we’re not suggesting that the corrections negate the overall importance of the research. But we do wonder how errors of this magnitude weren’t immediately obvious to not only the peer reviewers and editors of the article, but to the researchers themselves. After all, the gross domestic product of the entire continent of Africa was an estimated $2.6 trillion in 2019.  

The journal, CABI Agriculture and Bioscience — the official journal of CABI, a global nonprofit group focused on agriculture and the environment — has both a brief correction and a more detailed notice for the May 2021 paper, “Towards estimating the economic cost of invasive alien species to African crop and livestock production”. The short version reads: 

This paper is being corrected after the identification of two errors. The first error was in the calculation of weeding costs in which the wage costs of weeding per square kilometre were applied instead per hectare. The recalculation estimates the total cost of invasive alien species to the African agricultural sector as $65.58 billion per year, not, as originally stated, $3.66 trillion. This is equivalent to 2.5% of the gross domestic product of all African countries combined, rather than the originally stated 150%. The second error was in the calculation of the crop loss due to Phthorimaea absoluta in which the value was not corrected for the abundance of this species within each country. The recalculation estimates that this species costs the agricultural sector $4.1 billion in crop losses per year, rather than the originally stated $11.4 billion. Further details can be found in the correction text.

René Eschen, a research scientist at CABI, in Delémont, Switzerland, and the corresponding author of the paper, told us: 

We were made aware of the error by readers of the published article, two in particular, who indicated that the estimated cost of weeding was in their opinion unrealistic. One of them asked for some of the data we used for the calculation of the weeding cost (it is available online but requires extracting by a specialist), which we provided and he then repeated our analysis to show that something was amiss. This convinced us that there was an error and we then found the calculation error, which is due to estimating the cost of weeding per km2 instead of per ha that leads to an overestimation of the costs. The reviewers didn’t raise the large estimated value as an issue and, although surprised by the large value of the estimated weeding cost, we weren’t suspicious about our results. Two of the co-authors independently checked all the calculations and we didn’t find an error during this internal, pre-submission review process. 

Eschen said his group opted to correct, rather than call for a retraction of, their paper, and the journal agreed:  

The decision to allow a correction was taken by the editor in chief. The correction, although major in terms of the headline figure, only affects one part of the estimated cost of IAS to Africa that we calculated and it does not change the conclusion or recommendations of the paper. I think it is important to note that neither of the people who wrote to us contested the general calculation of the estimates or the input data. Once the decision was made, we have notified the two people who made us aware that we would correct the paper and they appear to be satisfied with that, as they believe that even the corrected paper is important.  

Nik Grunwald, the editor-in-chief of CABI Agriculture and Bioscience, told us:  

The authors alerted us to the error after they themselves had been made aware of the concerns by a reader. We are committed to addressing post-publication issues and we worked with the authors and the Springer Nature Research Integrity Group to correct the paper as soon as possible. While the process was ongoing, we added an editor’s note to the paper to alert readers that concerns had been raised. 

We asked Grunwald whether the errors should have been picked up prior to publication, and whether retraction might have been a more appropriate step:  

We treat the details of the peer review process as confidential, so I am unable to comment further here. However, we believe this was an honest mistake and that a correction is the appropriate response. We are reviewing our processes to avoid similar issues in the future.

The more detailed notice states:

After publication, it was brought to our attention that the estimated cost of weeding was much higher than can reasonably be expected although the input data and general approach to calculating the estimate were not disputed. We therefore revisited our calculations and found a scaling error that occurred during the application of wages to the harvested area for each of the five crop types. As described in the original article, we calculated the cost of weeding IAS in five crop types based on the harvested area in each African country, calculated from data in the SPAM database, the average abundance of alien species as a fraction of the weed community in African agricultural fields as deducted from the published literature. The average time spent weeding a hectare of each of the five crop types was taken from the published literature, and wages paid in each country for agricultural labour or similar jobs as reported on https://wagecalcul ator.org. We erroneously applied costs per square kilometre instead of the cost of weeding per hectare, resulting in substantially overestimated weeding costs. The calculations of the estimates of yield loss, the loss of livestock-based income and research costs were not affected by this error.

We corrected this error in our calculations and have updated the results, specifically Table 3, which presents to [sic] total estimate, and Table  4, which presents the estimates for each country. Tese corrected Tables are presented here. We have updated the numbers stated in the Abstract and in the Results section of the manuscript text to reflect the corrected results. We also modified the discussion where these estimates are put into context. While the total estimate of IAS cost to African agriculture are now significantly lower, the conclusions and recommendations have not changed. We have also corrected the value for the yield loss due to Phthorimaea absoluta based on the literature and a survey in Table 5. The value in the previous version of the manuscript was wrong due to a transcription error. The new value is corrected for abundance within each country, like the other values in the Table. This species was deleted from the abstract as the most costly species and minor changes to the Results and Discussion were made to align the text with the correct number. A reference made to an article ‘in press’ during publication has also been updated with the article’s final reference detail. We apologise for the errors and any confusion they may have caused. The original article has been updated.

Like Retraction Watch? You can make a one-time tax-deductible contribution or a monthly tax-deductible donation to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].