There's a push and pull between plants and soil when it comes to the amount of carbon dioxide they can sequester in response to greenhouse gas emissions, scientists reported this week, which means current climate models may be overestimating how much of the gas the world's forests will be able to remove from the atmosphere.

The researchers analyzed more than 100 experiments that exposed different habitats to elevated carbon dioxide levels, and found that the amount of carbon stored in soil was inversely tied to plant growth. The reason for this surprising relationship likely has to do with how plants interact with different kinds of symbiotic fungi, the researchers reported March 24 in Nature.

"When plants increase their biomass in a climate-change context, [they do] so at a cost ... which decreases soil carbon accrual and thus limits carbon storage," said César Terrer, a fellow at Lawrence Livermore National Laboratory in California and first author of the paper.

Terrestrial plants and soils remove about 30% of the carbon dioxide released by human activities each year. 

"Increased carbon dioxide leads to faster photosynthesis, the food production by plants, so we expected faster plant growth and more biomass to increase soil organic carbon as extra leaves fell to the forest floor," said Rob Jackson, a professor of Earth system science at Stanford University and last author of the paper. "It didn't, and that was the biggest surprise in our work; where we found additional plant biomass we didn't see extra soil carbon, in most cases."

He, Terrer and their colleagues examined the results from 108 published experiments in which researchers had added carbon dioxide to the air around temperate forests, grasslands, shrublands or agricultural land and monitored the area for several years. They tracked the amount of carbon that was taken up by plants and soil and found that grasslands, croplands and shrublands were better able to accumulate carbon in their soils than forests. 

Overall, the researchers calculated that rising carbon dioxide levels cause grasslands to increase their soil carbon by about 8% and plant biomass by 9%, while forests don't increase their soil carbon but do see a 23% increase in plant biomass. 

The relationship was only pronounced in experiments in which the plants hadn't been fertilized, suggesting that it was related to how plants acquire nutrients.

Forest trees, on one hand, tend to have symbiotic relationships with so-called ectomycorrhizal fungi. 

"These fungi form a sheath around the root tips and then they send their tendrils out into the soil to extract nutrients and bring them back to the tree, and the tree in exchange is providing food and carbon for the fungi," Jackson said. "To get the nitrogen to foster the increased growth of branches and leaves, these ectomycorrhizal fungi are going into the soil organic matter and kind of pulling the nitrogen out of it."

However, this nutrient mining breaks down the organic matter, increasing the rate of decomposition and preventing carbon from building up as much as it would in grasslands.

By contrast, grasses are more likely to team up with arbuscular mycorrhizal fungi, which grow into their roots rather than forming an outer sheath. "Their fungal partners aren't as good at getting nitrogen from organic matter as the forest fungal partners are," Jackson said.

On average, grasses deploy more of their carbon to root growth than trees do. 

"Grasslands are pushing more carbon directly into the soil through their roots, and that carbon is more likely to stay there than in the forest ecosystems," Jackson said.

The findings indicate that as carbon dioxide increases, the relative amount of carbon stored by plants and soils will vary from one ecosystem to another. 

"We don't see both pools increasing in the same place on average," Jackson said. However, the work also emphasizes the potential of grassland soils to absorb some of the excess carbon. 

Soils store more carbon than is contained in all plants and the atmosphere, Jackson says, and this carbon tends to stay put for many years. This makes it vital to understand how soils will respond to increasing greenhouse gas emissions. 

"Our potential to store carbon in the soil is much bigger than it is in trees or plants above ground," Jackson said. "Another reason we need to understand this better is so that our global models that predict plant growth in the future represent what's happening between plants and the soil more accurately."

In the future, he would like to extend the work to tropical ecosystems, which so far have received "shockingly few" experiments of the type Jackson and his colleagues analyzed. 

In the meantime, "We need to put more effort into studying the soils in these experiments compared to the focus most of the time strictly on the trees and the plants," Jackson said. "We need to dig harder and deeper into the soils and what's happening there." 

The study, "A trade-off between plant and soil carbon storage under elevated CO₂," published March 24 in Nature, was authored by C. Terrer, Lawrence Livermore National Laboratory and Stanford University; A. F. A. Pellegrini, Stanford University and University of Cambridge; R. B. Jackson, Stanford University; R. P. Phillips, Indiana University; B. A. Hungate, Northern Arizona University; J. Rosende, Universitat Autònoma de Barcelona; J. Pett-Ridge, Lawrence Livermore National Laboratory; M. E. Craig, Indiana University and Oak Ridge National Laboratory; K. J. van Groenigen, University of Exeter; T. F. Keenan, University of California, Berkeley and Lawrence Berkeley National Laboratory; B. N. Sulman, Oak Ridge National Laboratory; B. D. Stocker, ETH Zürich and Swiss Federal Institute for Forest, Snow and Landscape Research WSL; P. B. Reich, University of Minnesota and Western Sydney University; E. Pendall and Y. Carrillo, Western Sydney University; H. Zhang, University of Oxford; R. D. Evans, Washington State University; J. B. Fisher, California Institute of Technology and University of California, Los Angeles; and K. Van Sundert and Sara Vicca, University of Antwerp.