Paper overestimated risk of COVID-19 to endangered apes
A Springer Nature journal has retracted a 2021 article with dire news for mountain gorillas in Rwanda’s Volcanoes National Park about the prospects of extinction on the spikes of SARS-CoV-2 after finding a fatal error in their model of the outbreak.
The article, “Exploring the potential effect of COVID-19 on an endangered great ape,” appeared in October in Scientific Reports and was written by a group at the University of Southern Denmark and the the Dian Fossey Gorilla Fund, in Atlanta. The Fund has been concerned – for good reason – about the potential of respiratory illnesses to decimate populations of great apes, whose social nature can foster transmission of infections.
According to the abstract of the paper:
The current COVID-19 pandemic has created unmeasurable damages to society at a global level, from the irreplaceable loss of life, to the massive economic losses. In addition, the disease threatens further biodiversity loss. Due to their shared physiology with humans, primates, and particularly great apes, are susceptible to the disease. However, it is still uncertain how their populations would respond in case of infection. Here, we combine stochastic population and epidemiological models to simulate the range of potential effects of COVID-19 on the probability of extinction of mountain gorillas. We find that extinction is sharply driven by increases in the basic reproductive number and that the probability of extinction is greatly exacerbated if the immunity lasts less than 6 months. These results stress the need to limit exposure of the mountain gorilla population, the park personnel and visitors, as well as the potential of vaccination campaigns to extend the immunity duration.
The article received a modest amount of attention, according to Altmetric, which registered at least eight news articles and a handful of tweets – including one from Scientific Reports (we’ll note that we haven’t seen a tweet from the journal about the retraction).
But as the retraction notice explains, the prospect of extinction doesn’t appear to be as bad as the authors reported:
It was brought to our attention that we had made an error in our simulation study of the potential effect of SARS-CoV-2 on a subpopulation of the Virunga mountain gorilla population. Specifically, instead of using human infection fatality rates (IFRs) as a reference for our simulations, we used case fatality rates (CFRs) from studies published early during the pandemic in Wuhan, China, and Italy, adjusted to the longevity of the mountain gorillas. Case fatality rates, particularly those early during the pandemic, have since been found to be considerably higher than the actual infection fatality rates. After repeating our analyses with this updated information and using the adjusted CFRs instead of IFRs, we found that our published results significantly overestimated the chances of extinction of the population should a COVID-19 outbreak occur. Although the revised findings are important for mountain gorilla conservation, they substantially change the conclusions of the study. In light of these important discrepancies, we deem it necessary to retract this study.
Fernando Colchero, of the Department of Mathematics and Computer Science at the University of Southern Denmark, in Odense, forwarded us a letter he’d sent to the journal explaining the reason for the retraction. In addition to laying out the nature of the error, he wrote:
While the revised findings are also important for mountain gorilla conservation, they substantially change the conclusions of this paper, and a simple correction would be insufficient. Therefore, in all consciousness, we, the coauthors of this study, have deemed necessary to retract it to avoid future projects to draw the wrong conclusions from it.
We are carrying out additional analyses with an improved modelling approach and in the hope to incorporate information on gorilla epidemiology, particularly from groups in zoological institutions that had COVID-19 outbreaks. Our intention is to submit a manuscript when these improved analyses have been completed, albeit not before gaging [sic] the opinion of experts in the field.
We can only stress that we have a deep respect for scientific rigor and accountability, and that we are convinced that science can only remain true to itself if we, as scientists, are willing to accept our mistakes and act upon them, irrespective of how late it may seem.
We are grateful to Drs. Sarah H Olson, Peter J Hudson, Chris Walzer, and Alexis Lécu for their timely critique of our study, for pointing out this serious oversight, and for their useful suggestions on how to improve it.
Olson, an epidemiologist at the Wildlife Conservation Society, told us that she had doubts about the study because:
the IFR curve (Fig 2a) didn’t look right as most IFR curves for human SARS-CoV-2 look like a hockey stick. From there I reached out to colleagues to give it a further review.
Olson’s group notified the journal of their concerns about the paper at the same time as they notified the authors of the paper, and tried to post the following comment to the now-retracted paper, without success.
During the retraction process the journal did not offer to publish the critiques. Given the comment section is marginally useful, a post on Retraction Watch would help with our full transparency goal, so we are glad to share with Retraction Watch!
The retraction is the 208th on Covid-19, by our count.