The EoC for “Asia’s glaciers are a regionally important buffer against drought,” published by Hamish Pritchard, a glaciologist at the British Antarctic Survey, came out today. The May 11, 2017 article — which has been cited three times, according to Clarivate Analytics’ Web of Science — considers the contribution of glaciers to water supply in Central Asia and the potential for glacier loss to exacerbate water stress in the region. The paper received limited news coverage when it came out from science sites, including Phys.org.
Pritchard appears to have improperly used a particular data set — an error that was reported to the journal by two outside experts within weeks after the paper was published. Specifically, Pritchard thought certain numbers represented glacier loss due to melt over a given decade, when in fact they represented the average loss per year within that decade.
In other words, according to Hester Jiskoot, a glacier researcher at the University of Lethbridge in Canada, who was not involved with Pritchard’s paper but who reviewed the paper and the EoC for Retraction Watch:
[Pritchard] assumed that there’s ten times less water than there actually is.
Here’s the full notice:
The editors of Nature have become aware that this Article contains an error in the use of mass imbalance data. Specifically, the author used mass imbalance data from table 2 in ref. 32 described as decadal averages (millimetres water equivalent) that are in fact annual values averaged over a decade (millimetres water equivalent per year). The specific numbers presented for the net melt fraction and affected population are therefore incorrect, and probably underestimated. Nature is working with the author to address the error, but in the meantime, readers are cautioned against using the paper’s quantifications. The author agrees with our expression of concern. We thank L. Zhao and J. Moore for bringing the error to our attention on 2 June 2017.
Pritchard told us:
Only one part of a larger calculation is affected so the totals are off by substantially less than a factor of ten.
John Moore, a glacier researcher with appointments at Beijing Normal University and the University of Lapland in Finland who helped alert Nature to the error, said he calculated that glaciers contribute about twice as much water to the region’s rivers as Pritchard originally estimated.
He added that the mistake was something that should have been caught in peer review:
I think anyone familiar with the topic would have realized intuitively that the numbers could not have been as low as interpreted.
Last week, we reported on an EoC for another recent paper in a high-profile journal (Science) regarding a mistake that was quickly caught by an outside expert but had eluded the reviewers. In that instance, the journal posted the EoC within days after publication.
Moore and Zhang alerted the journal to the problem June 2; the EoC appeared months later. Moore told us:
In the same circumstances I think other journals would just publish the error and that would be the end of it, back in June.
We asked Nature why it took months to issue a notice; a spokesperson referred us to a statement it sent when we first inquired about the EoC:
Decisions to publish Editorial EoCs are made following a deliberative process that involves internal and external consultation, with authors and others, in an attempt to understand the issues more deeply. We are committed to treating research responsibly, and we have been working with the author to address the error. We will continue to explore the issues to reach an appropriate resolution.
This process is ongoing and we cannot comment further at this stage.
The abstract for Pritchard’s article states that glaciers in Central Asia are able to provide enough water to sustain “basic needs of 136 million people, or most of the annual municipal and industrial needs of Pakistan, Tajikistan, Turkmenistan, Uzbekistan, and Kyrgyzstan.”
Pritchard said he misread the caption of the data table he pulled his glacier melt figures from, which he took from another paper. He added the mix-up actually strengthens his arguments — meaning, that these regions receive even more water from melting glaciers than he originally reported.
The source of Pritchard’s confusion is “Historically unprecedented global glacier decline in the early 21st century,” published in 2015 in the Journal of Glaciology by the World Glacier Monitoring Service, a consortium based at the University of Zürich. According to Pritchard:
Specifically, in the main results table it described historical glacier losses as decadal averages in millimetres of water equivalent but these were actually decadally averaged annual values (in millimetres water equivalent per year).
Jiskoot told us she didn’t think the problem necessarily lay with the paper or its data table:
[Looking at the table caption], I would assume it’s a decadal average too, but if you read the text it becomes clear its an annual average. You always have to read the text — you’re not publishing a table as a standalone. You cannot assume something from a table unless you read what was meant with that.
Jiskoot is an associate chief editor at the Journal of Glaciology, but said she was not involved in that paper’s publication process. Pritchard is also one of the journal’s scientific editors.
Pritchard told us that he’s been working with Nature on a correction that is currently being reviewed:
The consequence of the correction is to strengthen the study conclusions by increasing the amount of water provided by Asian glaciers (remembering that the title is ‘Asia’s glaciers are a regionally important buffer against drought’). In practical terms, the correction increases the glacial melt fractions that I report by increasing the meltwater volumes, which consequently also increases the number of people that can be sustained by this meltwater. In a few places through the text, these numbers are increased. The results figures are also slightly revised.
Moore told us that he and Zhao had submitted a “Brief Communications Arising” (BCA) to Nature, but that it had not been accepted.
The reason it was rejected, Moore told us, was because Pritchard agreed he’d made a mistake. Moore forwarded us Nature‘s rejection letter, which said:
we must therefore decline to publish your BCA, as there is no disagreement regarding the error. That said, I do understand that you and Hamish Pritchard seem to disagree about the implications of the error for the paper’s overall conclusions. Therefore, once the correction is published, I would invite you to reassess the paper and — if you continue to disagree with the paper’s conclusion — submit a new BCA at that time.
Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up on our homepage for an email every time there’s a new post, or subscribe to our daily digest. Click here to review our Comments Policy. For a sneak peek at what we’re working on, click here.