Paper claiming extra CO2 doesn’t always lower plant nutrients pulled for errors

Authors have retracted a large meta-analysis claiming that rising levels of carbon dioxide don’t always reduce nutrients in plants.

After commenters on PubPeer raised concerns, the authors say they found several unintentional errors in their data that could “significantly change conclusions” of the paper in Plant Ecology, according to the retraction note.

The paper found that the impact of rising CO2 depends on many factors — in some cases, extra amounts of this greenhouse gas could actually increase plant nutrients. Trouble is, some of the papers that cited the now-retracted article came to the opposite conclusion: Increased carbon dioxide levels do decrease plant nutrients.

The retraction note for “CO2 effects on plant nutrient concentration depend on plant functional group and available nitrogen: a meta-analysis” explains some of the specifics of the errors, and says that there was “no evidence of bias:”

This article has been retracted at the request of the authors because of multiple and unintentional data entry errors. After concerns were raised on the forum PubPeer (Duval et al. 2012) related to the open access online data sets compiled by lead author Duval, co- authors Blankinship, Dijkstra, and Hungate compiled a review dataset by consulting the original literature and re-extracting data from all the papers used in the original meta-analysis. In this effort, they found multiple and unintentional errors in the published dataset that could significantly change conclusions made in the paper regarding the generality of elevated CO2 effects on plant nutrient concentrations.

Specific errors were attributable to inadvertently reversing the elevated and ambient CO2 treatments in the original dataset, errors in transcribing numbers from published tables to the original dataset, incorrectly treating element stocks as element concentra- tions, and incorrectly assigning plant parts to proper categories. In others cases, figures were scanned inaccurately. In still others, data were selected arbitrarily in a manner not described in the methods (e.g., one crop variety was selected from a paper describing responses of multiple crop varieties). There was no evidence of bias in these errors. They were unintentional. Because these errors may influence the conclusions made in the paper, and because the published online datasets are demonstrably incorrect, the authors requested that the paper be retracted.

The paper, which has been cited 21 times, according to Thomson Scientific’s Web of Knowledge, received coverage after publication on a website called CO2 Science.

Irakli Loladze, a mathematical biologist at the University of Maryland who has criticized this paper, explained that the errors made it hard to see the true picture of what was going on:

Biological data are usually noisy. In Duval et al. (2012), the noise was amplified by the combination of numerous data errors and a statistically flawed methodology. The noise was amplified to the point where it hid the true signal, the true pattern – rising atmospheric CO2 diminishes concentrations of essential for human nutrition minerals in C3 plants (C3 plants account for > 90% of all plant species.) To the authors credit, they acknowledged the errors and retracted the paper.

In his 2014 eLife paper “Hidden shift of the ionome of plants exposed to elevated CO2 depletes minerals at the base of human nutrition,” Loladze criticizes the retracted study’s statistical methods:

…a close examination of the results of Duval et al. [2012] reveals that every statistically significant increase in mineral concentrations was obtained by bootstrapping a sample of size 2, 4 or 5 – a recipe for generating invalid 95% [confidence intervals].

A Nature letter titled “Increasing CO2 threatens human nutrition” also contradicted the conclusions of the retracted paper, showing reduced levels of some nutrients in plants grown in high-CO2:

C3 grains and legumes have lower concentrations of zinc and iron when grown under field conditions at the elevated atmospheric CO2 concentration predicted for the middle of this century.

We reached out to last author Bruce Hungate, a biologist at Northern Arizona University.

First author Benjamin Duval, a biologist at the University of Wisconsin Madison who compiled the dataset, declined to comment.

Hat tip: Rolf Degen 

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, and sign up on our homepage for an email every time there’s a new post. Click here to review our Comments Policy.